The interplay between biomaterial degradation and tissue properties : relevance for in situ cardiovascular tissue engineering Brugmans, M.C.P.

Transcription

1 The interplay between biomaterial degradation and tissue properties : relevance for in situ cardiovascular tissue engineering Brugmans, M.C.P. Published: 01/01/2015 Document Version Publisher s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. The final author version and the galley proof are versions of the publication after peer review. The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 23. Aug. 2018

2 The interplay between biomaterial degradation and tissue properties Relevance for in situ cardiovascular tissue engineering Marieke Brugmans

3 A catalogue record is available from the Eindhoven University of Technology Library ISBN: Copyright 2015 by M.C.P. Brugmans All rights reserved. No part of this book may be reproduced, stored in a database or retrieval system, or published, in any form or in any way, electronically, mechanically, by print, photo print, microfilm or any other means without prior written permission by the author. Printed by Ipskamp Drukkers B.V., Enschede, the Netherlands. The research and printing of this thesis was supported by: Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully acknowledged. This work was supported by a grant from the Dutch government to the Netherlands Institute for Regenerative Medicine (NIRM, grant No. FES0908).

4 The interplay between biomaterial degradation and tissue properties Relevance for in situ cardiovascular tissue engineering PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus, prof.dr.ir. F.P.T. Baaijens, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op woensdag 10 juni 2015 om uur door Maria Cornelia Philomena Brugmans geboren te Veghel

5 Dit proefschrift is goedgekeurd door de promotoren en de samenstelling van de promotiecommissie is als volgt: Voorzitter: prof. dr. P.A.J. Hilbers 1 e promotor: prof.dr.ir. F.P.T. Baaijens 2e promotor: prof.dr. C.V.C. Bouten copromotor: dr. A. Driessen-Mol leden: dr. J. Kluin (UvA) dr. P. Habibovic (UM) dr.rer.nat. C. Ottmann adviseur: dr. P.Y.W. Dankers

6 Contents Summary III Chapter 1: General introduction Human cardiovascular tissues Heart valves The heart valve leaflets Blood vessels 1.2 Cardiovascular diseases and current treatments 1.3 Cardiovascular tissue engineering approaches and challenges 1.4 Biomaterials Natural biomaterials Synthetic biomaterials 1.5 In vivo resorption of biomaterials Resorption pathways Variation in resorption of biomaterials 1.6 The host response to biomaterials The phases of the natural healing response Macrophage phenotypes 1.7 Rationale and outline Chapter 2: Polycaprolactone scaffold and reduced in vitro cell culture: Beneficial effect on compaction and improved valvular tissue formation Abstract 2.2 Introduction 2.3 Materials and Methods 2.4 Results 2.5 Discussion 2.6 Conclusion Chapter 3: Superior tissue evolution in slow-degrading scaffolds for valvular tissue engineering Abstract 3.2 Introduction 3.3 Materials and Methods 3.4 Results 3.5 Discussion 3.6 Conclusion I

7 Chapter 4: Hydrolytic and oxidative degradation of electrospun supramolecular biomaterials: In vitro degradation pathways Abstract 4.2 Introduction 4.3 Materials and Methods 4.4 Results 4.5 Discussion 4.6 Conclusion Chapter 5: Advanced electrospun scaffold degradation by inflammatory macrophages in comparison with healing macrophages Abstract 5.2 Introduction 5.3 Materials and Methods 5.4 Results 5.5 Discussion 5.6 Conclusion Chapter 6: General discussion Main findings of the thesis 6.2 Towards the most promising tissue engineering approach and scaffold material 6.3 Study limitations and the future of in-situ cardiovascular tissue engineering 6.4 Conclusion References 107 Nederlandse samenvatting 127 Dankwoord 129 Curriculum vitae 131 List of publications 133 II

8 Summary The interplay between biomaterial degradation and tissue properties: Relevance for in situ cardiovascular tissue engineering Various tissue engineering (TE) approaches are currently under investigation to create cardiovascular tissue replacements. The most promising strategy is the in-situ TE approach, in which off-the-shelf available synthetic electrospun scaffolds are used to replace diseased vessels or heart valves. After implantation, a host inflammatory response is activated, leading to the infiltration of macrophages, which play a key role in both scaffold degradation and tissue formation. As a result, a living tissue that is able to remodel and adapt to the environmental changes is obtained in-situ. It is crucial to select the optimal scaffold material to ensure mechanical integrity immediately after implantation, which starts degrading as soon as sufficient tissue is formed to take over the native function. The aim of the research described in this thesis was to examine the interplay between scaffold degradation rates and the amount and composition of the formed tissue within the scaffold. Furthermore, degradation characteristics of scaffolds manufactured from different supramolecular biomaterials, were investigated. By imbalance between scaffold degradation and tissue formation, the mechanical integrity cannot be ensured and compaction and retraction of in-vitro TE heart valves occurs, causing regurgitation in-vivo. We studied whether compaction could be reduced by the use of slow-degrading polycaprolactone (PCL) instead of fast-degrading poly-4- hydroxybutyrate coated polyglycolic acid (PGA-P4HB) electrospun scaffolds and/or the use of a lower cell passage number to enhance tissue formation. The use of slowdegrading materials improved resistance to retraction of TE valvular leaflets and reduced compaction of TE rectangular scaffold strips. In addition, tissue formation, stiffness, and strength increased with decreasing cell passage number, but did not affect compaction of the engineered tissues. Thereafter, the effect of scaffold degradation rate on the amount and composition of tissue, the mechanical integrity, and the tissue to scaffold ratio were investigated. Slowand fast-degrading scaffolds were seeded with vascular cells or kept unseeded. We hypothesized that the cells in fast-degrading scaffolds would compensate for the rapid loss of mechanical integrity by increased tissue production. Increasing amounts of tissue with time were shown in both scaffold groups, which was indeed more pronounced for PGA-P4HB-based tissues during the first two weeks of culture. Ultimately, PCL-based tissues resulted in the highest amount of tissue after 6 weeks. In addition, we described a method to correct for the amount of remaining scaffold weight, in order to allow a fair comparison between in-vitro engineered tissues grown on scaffolds with a different III

9 Summary degradation rate and in-vitro engineered tissues and native tissues. By implementation of this correction, extracellular matrix values similar to values of native pulmonary heart valves were found. The amounts of collagen crosslinks were still below native values in all engineered tissues, but did display a continuing increase during culture. In-vivo, degradation of scaffold materials can be accomplished by the (enzymatic accelerated) hydrolytic and/or the oxidative pathway. To investigate both pathways, separately and in an accelerated fashion, in-vitro degradation assays were designed. For in-situ TE of cardiovascular tissues, the supramolecular materials PCL-2-ureido-[1H]- pyrimidin-4-one (PCL-UPy) and PCL-bisurea (PCL-BU) are used, due to their combination of strength and elastic properties. Degradation characteristics and susceptibility to the hydrolytic or the oxidative degradation pathway of these materials were investigated and compared with those of conventional PCL. Depending on the morphological and chemical composition of the materials, conventional and supramolecular PCL-based scaffolds responded differently to both degradation pathways. Conventional PCL is more prone to hydrolytic enzymatic degradation as compared to the supramolecular materials, while the opposite was shown when degraded by the oxidative pathway. We demonstrated the ability of tuning degradation characteristics by mix-and-match PCL backbones with supramolecular moieties. This allows screening and selecting the optimal biomaterial for pre-clinical studies targeted to different clinical applications. Macrophages are known to play an important role in the degradation of the implant, however the contribution of macrophage phenotype to scaffold degradation was still unclear. The inflammatory phenotype is known to secrete both reactive oxygen species (ROS) and enzymes involved in scaffold degradation. However, degradation might also be accomplished by the healing phenotype, as also these secrete enzymes involved in scaffold degradation. The correlation between macrophage phenotype and degradation of electrospun scaffolds was investigated in this thesis. We elucidated that the macrophage phenotype affected the contribution to scaffold degradation, consolidating that inflammatory macrophages indeed accelerated degradation. In addition, the electrospun PCL induced macrophage polarization towards the healing phenotype, which is a beneficial feature for in-situ TE. In conclusion, the choice of scaffold material is of high importance to maintain mechanical integrity. Results in this thesis emphasize that a slow-degrading material is favored over a fast-degrading material, as mechanical integrity will be maintained for a longer period, which is important for in-situ TE purposes. Furthermore, tissue seemed better organized when cultured on slow-degrading scaffold materials and therefore is less prone to compaction. In addition, this thesis demonstrated that degradation characteristics can be tailored, which is essential as different degradation characteristics are desired for various applications. IV

10 1 General introduction C 1

11 Chapter 1 Cardiovascular in situ tissue engineering using synthetic materials is a promising approach for overcoming the limitations of current available treatments for the repair or replacement of damaged or diseased cardiovascular tissues. Off-the-shelf, highly porous scaffolds, in the shape of the desired construct, act as templates for replacing the diseased tissues with healthy tissues. They should provide temporary mechanical strength, while endogenous cells are attracted to the implanted scaffold and produce new tissue. At the same time new tissue is formed, the implanted synthetic scaffold slowly resorbs and is ultimately removed from the body, leaving behind functional, viable tissue that is able to adapt to environmental changes. In order to maintain good mechanical integrity, immediately after implantation until the newly formed tissue takes over this role, the balance between tissue formation and bioresorption of the scaffold is of high importance. As a consequence, bioresorption of the implanted scaffold plays a crucial role in the final outcome of the engineered construct. Tunability of scaffold resorption in vivo is desired, as different resorption rates are needed for various applications. Furthermore, amount and quality of newly formed tissue might be influenced when growing in slow- or fastresorbing materials. In vitro, no resorption by the body occurs and therefore the break down of scaffolds in this thesis is referred to as degradation. The aim of this thesis is to elucidate in vitro degradation characteristics of scaffolds manufactured from different synthetic biomaterials, and the effect of degradation on tissue formation. With the use of this knowledge, bioresorbable scaffolds can be created with appropriate resorption characteristics for use as cardiovascular tissue replacements. 1.1 Human cardiovascular tissues Heart valves The heart is a muscular organ that regulates blood flow throughout the body in order to transport oxygen and nutrients to tissues, and remove metabolic waste from the tissues. Oxygen-poor blood returns back into the right ventricle, via the right atrium. When the right ventricle contracts, this blood is pumped through the pulmonary artery into the lungs, where it becomes oxygenated again. The left ventricle receives this oxygen-rich blood from the lungs via the left atrium. Upon contraction of the left ventricle, blood is pumped into the aorta and distributed throughout the whole body. To ensure unidirectional blood flow, the heart is provided with four valves: the tricuspid valve, the mitral valve, the pulmonary valve, and the aortic valve (Figure 1.1). The tricuspid and mitral valves (atrioventricular valves) are situated between the atria and the ventricles, and prevent blood from flowing back from the ventricles into the atria. The pulmonary and aortic valves (semilunar valves) are situated between the right ventricle and the pulmonary artery, and the left ventricle and the aorta, respectively. These valves prevent blood from flowing back from the pulmonary artery and aorta into the ventricles. Valves open and close approximately times each day and about 3.7 billion times in a 2

12 General introduction lifetime, subjecting the thin and flexible leaflets to loads and deformations with every heartbeat. 1 Figure 1.1 Schematic transverse sections of the human heart and its four valves. Cross section of the heart, anterior view (A). Direction of blood flow is indicated with arrows. Cross section of the heart, top view, showing the opened (B) and closed (C) position of the pulmonary and aortic valves to allow blood flow from the ventricles into the pulmonary artery and aorta (adapted from zoominmedicine.com) The heart valve leaflets The pulmonary and aortic valves are referred to as semilunar valves due to the half-moon shape of their three thin leaflets. The leaflets are connected to a fibrous, ring shaped thickening of the arterial wall, called the annulus. Leaflets are composed of cells, embedded in an extracellular matrix (ECM). The cross section in Figure 1.2 shows that leaflets have a layered architecture, which comprise three distinct layers; the fibrosa, the spongiosa and the ventricularis. These layers can be identified in both the aortic and pulmonary heart valve leaflets, however, a more pronounced fibrosa layer can be found in the leaflets of the aortic heart valve. Figure 1.2 Cross section of one of the leaflets of a heart valve (left). Schematic overview of the composition of a leaflet, consisting of three distinct layers, each comprising valvular interstitial cells (VICs) and ECM components (right) (adapted from Vessely, 1998 and Schoen, 2013). 3

13 Chapter 1 Each layer has a specific composition and organization of the ECM. The fibrosa, which is located at the arterial side of the leaflet, consists mainly of a dense collagen network and provides mechanical strength to the tissue. The spongiosa, situated between the fibrosa and the ventricularis, consists mainly of proteoglycans and water-binding glycosaminoglycans (GAGs) to absorb shocks on the leaflet. The ventricularis, at the ventricular side, is rich in elastin fibers, which ensure flexibility and restores the contracted configuration of the leaflets [1, 2]. Two types of cells are present within the leaflets; valvular interstitial cells (VICs) and valvular endothelial cells (VECs). VECs form a single endothelial layer, covering the whole leaflet surface to prevent direct contact of the ECM with blood, and thereby provide a non-thrombogenic layer. VICs are the most abundant cellular component of the heart valves and are found throughout the leaflets. In healthy adult heart valves, these cells reside in a fibroblast-like quiescent state, but they can differentiate into myofibroblasts-like cells and mediate ECM synthesis and remodeling [3-7] Blood vessels There are three major types of blood vessels; the arteries, the veins and the capillaries (Figure 1.3). In general, arteries carry oxygen- and nutrient-rich blood away from the heart, after which the actual exchange of oxygen and nutrients between blood and tissues takes place in the capillaries. Oxygen-poor blood is collected in the veins, and is carried back to the heart. Capillaries consist of only a single layer of endothelial cells to enable optimal gas and nutrient exchange. Figure 1.3 Schematic overview of the human circulatory system. The further away vessels are from the heart, the smaller they become. In the smallest vessels, the capillaries, nutrient and oxygen exchange takes place (From Martini, Frederich H.; Timmons Michael J.; Tallitsch, Robert B.; Human anatomy, 7th Edition, Reprinted by permission of Pearson Education, New York). 4

14 General introduction Within the arteries and veins three different layers of tissue can be distinguished; the tunica adventitia, the tunica media and the tunica intima. These layers differ in thickness, depending on the function of the vessel. The tunica adventitia, which is the outer layer, consists of loosely woven collagen fibers and may contain nutrient capillaries in the larger vessels. In the middle layer, the tunica media, smooth muscle cells and elastin can be found, which regulate and assist in vasodilatation or vasoconstriction. The tunica intima, situated at the lumen, is in direct contact with blood and consists of a single endothelial layer and some elastic fibers [8] Cardiovascular diseases and current treatments Cardiovascular diseases (CVD) remain the leading cause of death worldwide, among both men and women, resulting in almost half of all deaths in Europe and one third of all deaths in the United States [9, 10]. Among CVD, coronary artery disease is the most frequent and is often treated with bypass grafting [9, 10]. Each year, surgeons perform approximately coronary bypass surgeries worldwide [11]. Furthermore, vascular grafts are needed in diabetic patients, end-stage renal disease and pediatric heart operations. Autologous small-diameter arteries and veins are the preferred replacement grafts [12] despite 50% of the grafts occluding within 10 years [13]. However, it is estimated that these tissues are not available in 30% of all patients, due to either inherent disease or harvest during previous operations [14]. In these cases, non-degradable synthetic grafts can be used (Figure 1.5), such as expanded polytetrafluoroethylene (eptfe, i.e. GORE- TEX ) or polyethylene terephthalate (PET, i.e. Dacron ). These materials are widely used in the clinic for over 50 years and have shown to be successful for medium to large diameter vascular graft applications. However, data on small-diameter grafts (<6mm) is still very poor. Results showed that these synthetic grafts are prone to thrombus formation and intimal hyperplasia, leading to occlusion of the graft [12, 15, 16]. Therefore, they have lower patency rates compared to autologous grafts, with patency rates of 24-44% for PTFE compared to 70% for saphenous veins after 5 years in peripheral applications [17]. This shows the obvious need of small-diameter vascular grafts that resemble autologous grafts. Heart valve disease (HVD) can occur in any single valve, or a combination of several valves. Diseases related to the aortic and mitral valves are most common and result in the highest mortality rate, because of its important hemodynamic positions [9, 18]. HVD can lead to stenosis (narrowing of the valve opening), or regurgitation (leakage of the valve) (Figure 1.4). These pathologies can be caused by a congenital abnormality (e.g. 2 leaflets instead of 3), calcification or by damage to the valve due to rheumatic fever or endocarditis [18-21]. 5

15 Chapter 1 Figure 1.4 Schematic cross sections of a healthy heart during systole (A) and diastole (C). During systole, the left ventricle contracts and opens the aortic valve, allowing blood flowing into the aorta. In case of a stenotic valve (B), blood flow is obstructed, resulting in thickening of the left ventricle. During diastole, the left ventricle relaxes and fills with blood. The aortic valve is closed, to prevent backflow from the aorta. Regurgitation, due to incomplete closed leaflets (D), results in an enlarged heart cavity and a thickened left ventricle (adapted from Nishimura 2002 with permission from Wolters Kluwer Health). The most common treatment of end-stage disease is replacement of the valve. Worldwide, approximately heart valve replacements are performed each year, and this number is expected to increase up to by 2050, due to aging of the population and the increased ability to diagnose valvular heart disease [9, 22]. Current available heart valve replacements are either mechanical or bioprosthetic (Figure 1.5), each having their own benefits and disadvantages [23-26]. Different mechanical valves have been designed; ball-and-cage valves, mono-leaflet valves and bi-leaflet valves, which are made of for example carbon, Teflon or titanium [25]. Mechanical valves can last a lifetime, with a valve replacement rate <2% over 25 years [27], and are readily available. However, they are prone to thrombus formation due to non-physiological flow profiles that result in blood cell damage [28]. As a consequence, life-long anti-coagulation therapy is required, which results in increased risk of spontaneous bleeding in those patients. Bioprosthetic valves can be harvested from a human (homograft) or from an animal (xenograft). Homografts are closest to natural valves and can be derived from a donor (allograft) or from patients themselves (autograft). Donor valves are sterilized using antibiotic and anti-fungi solutions and stored by cryopreservation or fixated. However, there is limited availability of this type of valves. Xenografts, made of glutaraldehyde fixed porcine or bovine material are often used instead. These valves do not require anticoagulation therapy, but are prone to tissue degeneration and calcification, with reoperation rates of 20% after 10 years and 30% at 15 years, which limits their durability [25, 29-31]. Furthermore, the risk of transmission of animal diseases to human and immunogenic reactions is increased with this type of valve replacement [32, 33]. An alternative is to decellularize these tissues, and thereby decrease the immunological response without limiting the remodeling capacity of the implants [15, 34]. This results in native-like tissue replacements, which can be implanted as such [35], or can be re-seeded with autologous cells prior to implantation [36-38]. 6

16 General introduction 1 Figure 1.5 Examples of existing heart valve and vascular replacements. Starr-Edwards prostheses (ball-cage) (A), Medtronic open pivot TM mechanical valve (B), Edwards SAPIEN 3 Transcatheter Heart Valve (C), Medtronic Hancock II bioprostheses (D), Medtronic Melody transcatheter pulmonary valve (E), Medtronic Contegra pulmonary valved conduit (F), GORE-TEX vascular grafts (G), Dacron vascular grafts (H). Images A and C are reproduced with permission of Edwards Lifesciences LLC, Irvine, CA. Images B, D, E, and F are reproduced with permission of Medtronic, Inc., a subsidiary of Medtronic plc. Image G courtesy of W. L. Gore & Associates, Inc. Although the current available cardiovascular tissue replacements significantly improve quality of life and life expectancy, a shortcoming is that they are not able to adapt to changing physiological demands, as they consist of non-living materials. The development of a living tissue that can adapt is of utmost importance to further improve quality of life and life expectancy of patients with cardiovascular diseases. Growth potential has also been assumed as a desired property of a living heart valve replacement to prevent re-operations in pediatric patients. As current treatments do not accommodate for growth, oversized replacements are often used in pediatric patients to prevent early outgrowth of the replacement. However, research on failed replacements in children have shown that not outgrowth of the replacement, but contracture and stenotic valves are the most important failure modes [39-41]. This indicates that preventing the most common failure modes should have priority over growth potential. 1.3 Cardiovascular tissue engineering approaches and challenges Different cardiovascular tissue engineering approaches are used within the field of regenerative medicine. These include the classical in vitro tissue engineering approach, with or without decellularization of the created tissues afterwards, the in vivo tissue 7

17 Chapter 1 engineering approach, and the promising in situ tissue engineering approach (Figure 1.6). They all aim to create a living cardiovascular substitute that is able to adapt after implantation. The classical approach is the in vitro tissue engineering approach, in which autologous cells, to prevent immune responses, are expanded in vitro. After cell seeding, the bioresorbable scaffold construct is often subjected to stimuli, which mimic physiological pressures and/or flows in a bioreactor to enhance tissue formation [42-45]. Different cell sources have been examined [4] including vascular-derived cells [46], which are also used in our lab [47-49], neonatal cell sources [50-52], mesenchymal stem cells [53, 54], adiposederived cells [55], and endothelial progenitor cells [56, 57]. Also, different materials are used to create scaffolds, which include the natural polymers e.g. fibrin and collagen [15], and the synthetic polymers e.g. PCL [58]. Weinberg and Bell produced the first tissue engineered vascular graft (TEVG) based on collagen and vascular-derived cells, using this in vitro approach [59]. However, it was found that this graft was mechanically unstable and not suitable for implantation. Encouraging progression was made within this field, as shown in both in vitro and in vivo studies on vascular grafts, with high patency rates up to 13 months [60-65]. However, to date no living small-diameter vascular graft is made that remains patent during a life-time. The in vitro tissue engineering approach has also shown to be promising for clinical applications. Engineered tubes based on a bioresorbable scaffold material, seeded with autologous vascular cells, have been successfully implanted into humans to reconstruct the pulmonary artery [46, 66]. Proof of concept of an in vitro tissue engineered heart valve (TEHV) was demonstrated in 1995 by Shinoka et al. [67], where a single autologous tissue engineered leaflet was implanted in a sheep. The next step was to develop functional three-leaflet tissue engineered valves. This was first reported by Hoerstrup, Sodian and Stock. They showed functionality of TEHV at the pulmonary position for up to 24 weeks [68-70]. More recent studies also showed promising in vivo results of TEHV with functional leaflets in sheep for up to eight months [53]. Nevertheless, the main challenge in all recently performed pre-clinical studies is retraction of the heart valve leaflets leading to regurgitation [43, 71-73]. The balance between the contractile tissue-producing cells and the mechanical integrity of the remaining scaffold is very important in order to prevent this retraction of the leaflets [47, 74]. Therefore, researchers decellularize the tissue-engineered constructs after in vitro culture in order to remove the contraction forces exerted by these cells. Furthermore, this decellularization protocol is used to create off-the-shelf available tissue replacements [43, 73, 75-77]. While decellularization of native tissues has demonstrated various rates of repopulation after implantation, decellularization of in vitro cultured constructs has shown faster host cell repopulation [73]. This is probably due to the lack of elastin barriers and a less mature collagen structure [78-80]. In vivo tissue engineering can be defined as a process where the peritoneal cavity or subcutaneous space is used to generate an autologous graft by taking advance of the 8

18 General introduction immune response to foreign materials. This in vivo formed tissue can subsequently be removed and used as a vascular graft [62, 81-85]. The main challenge remains to maintain all vascular grafts patent after implantation. 1 Figure 1.6 Overview of the different tissue engineering approaches. In the in vitro tissue engineering technique, cell-seeded scaffolds are placed into bioreactors, to mature the tissue before implantation (A.1, middle and right photos made by Bart van Overbeeke) or are decellularized before implantation (A.2, reprinted from Dijkman 2012 with permission from Elsevier). Scaffold in the shape of a blood vessel or a heart valve is implanted directly in the in situ tissue engineering approach (B). In vivo tissue engineering makes use of an e.g. silicon rod which is implanted into the peritoneal cavity. After some time this rod is explanted and the tissue formed around this rod is used for replacement of the diseased tissue (C, reprinted from Yamanami 2013 with permission from Springer). Although tissues with properties towards autologous grafts can be created with the in vitro and in vivo tissue engineering approaches, it takes weeks to months to produce these implants. Together with regulatory issues for transportation and storage of tissue engineered implants, this approach is very expensive and time-consuming. Furthermore, it could result in products with batch-to-batch variation in tissue quality due to variation in performance of biological material. To circumvent these disadvantages, a trend towards in situ tissue engineering is seen in academic research and industry [86, 87]. Within this approach, a synthetic bioresorbable scaffold is either implanted cell-free [88-92], or pre-seeded with autologous cells prior to implantation [72, 93-96]. The scaffold, in the shape of the desired replacement, should maintain mechanical functionality immediately after implantation, while endogenous cells are attracted to the implanted material and produce new tissue. While neo-tissue is 9

19 Chapter 1 formed, the scaffold slowly degrades and is ultimately removed from the body, leaving behind living tissue that is able to grow and adapt to changing physiological demands. From nearly 30 years ago until now, encouraging results of in vivo studies using cell-free vascular grafts in rats, dogs and the canine model have been reported [88, 90, 91, ]. Further, promising data of clinical trials based on in situ tissue engineering of large diameter TEVG, is reported. Large diameter bioresorbable vascular grafts, pre-seeded with autologous bone marrow mononuclear cells before implantation into pediatric patients, demonstrated growth capacity, while no graft-related mortality or graft failures were observed during a mean follow-up of 5.8 years [46, 94, 95, 102]. The unguided in situ tissue engineering process in pristine scaffolds, where no cells, proteins, or other biologicals are added to the scaffold before implantation, is here referred to as endogenous tissue growth (ETG). Recently, the company Xeltis implanted cell-free, bioresorbable vascular conduits into five pediatric patients. These conduits were designed to enable ETG and resulted, to this date, in successful tissue replacements [92]. In situ tissue engineering of heart valves also showed good progress during the last years. Pulmonary valves, based on a bioresorbable material and pre-seeded with autologous bone marrow cells, were implanted into non-human primates, and demonstrated a confluent layer of endothelial cells after 4 weeks and proper valvular functionality up to 4 weeks [72]. In a recent study performed by the Dutch BioMedical Materials program ivalve, cell-free heart valve constructs, based on a bioresorbable scaffold, were implanted at the pulmonary position in an ovine model. After 12 months, functional heart valve leaflets were demonstrated with good tissue formation [103]. In conclusion, several tissue engineering approaches have demonstrated promising results, although each approach still has challenges to overcome. The in situ tissue engineering approach is especially appealing and promising. This device-based approach is based on faster, easier, and cheaper production of off-the-shelf available grafts and encounters less regulatory hurdles, compared to cell-based approaches. Of particular interest is the ETG approach, where a bare scaffold is used without any additives. 1.4 Biomaterials Selection of the right biomaterial is important within tissue engineering, as mechanical integrity should maintain until neo-formed tissue can take over this role. This is mainly important for in situ tissue engineering, as a bare scaffold is implanted, which should provide sufficient mechanical strength by itself, immediately after implantation. The ideal biomaterial for cardiovascular tissue engineering should also be highly porous with an interconnected pore network to allow for cell infiltration and tissue in-growth, nutrient supply, and removal of metabolic waste products. Furthermore, it should be biocompatible, bioresorbable, reproducible, and contain mechanical properties that are consistent with the anatomical site of implantation to prevent compliance mismatch. 10

20 General introduction Different types of biomaterials, either natural or synthetic, are described within this section Natural biomaterials 1 Apart from native matrices that are decellularized before implantation, scaffolds could also be made from natural materials. These include fibrin, elastin, hyaluronan, silk fibroin and collagen [15, 77, 104, 105], which show good biocompatibility in terms of chronic inflammation and toxicity, and closely mimic the natural ECM of tissues. A disadvantage of these materials is the high batch-to-batch variations and researchers have limited control, although progression is made, over the material properties, which often results in lack of mechanical performance [15, 106, 107] Synthetic biomaterials Bioresorbable synthetic biomaterials are widely used for cardiovascular tissue engineering purposes. They are cheap to fabricate, readily available and researchers have better control over critical properties, such as the resorption rate or mechanical properties compared to the natural biomaterials. Among them are PCL, polyglycolic acid (PGA) and polylactic acid (PLA), which are used in medical devices that are already approved by the Food and Drug Administration (FDA) or have European Conformity (CE) mark registration [58, 108]. Recent studies showed promising results for cardiovascular applications with these and other polymers, like polyglycerolsebacate (PGS) and polyurethanes [73, 91, 109, 110]. Each polymer has different characteristics in terms of mechanical properties or resorption and might be suitable for different applications. PCL has been shown to be an interesting candidate for TEVG [111], however, due to its limited fatigue resistance, this material is less suitable for TEHV, as in TEHV the materials are exposed to demanding mechanical loads. A unique and new set of synthetic materials are the supramolecular polymers, which are formed by arrays of directed, non-covalent interactions, such as hydrogen bonds, between the polymer chains (Figure 1.7) [112, 113]. These supramolecular polymers can form complex 3D-structures by self-assembly. Material properties such as mechanical properties and/or resorption rate can be modified easily by combining or changing ratios of the same building blocks, providing a broad variety of biomaterial properties. As the monomeric units in supramolecular materials possess relatively low molecular weights, they can easily be dissolved and processed. Examples of the supramolecular biomaterials are the PCL-based materials modified with 2-ureido-[1H]-pyrimidin-4-one (UPy) [ ] or bis-urea (BU) [119] units. These exhibit strong and elastic properties and therefore might be more suitable for cardiovascular applications like heart valves, when compared to some of the conventional polymers, e.g. PCL. 11

21 Chapter 1 Figure 1.7 Schematic overview of an example of a supramolecular PCL-based material with ureidopyrimidinone moieties (grey blocks). Polymer chains are held together via hydrogen bonds (dotted lines). 1.5 In vivo resorption of biomaterials Within the field of tissue engineering, the balance between tissue formation and scaffold resorption, which is different in every application, is of high importance. Bioresorbable scaffolds should provide mechanical strength until sufficient mature neo-tissue is formed to take over this function Resorption pathways In vivo, implanted scaffolds can be degraded by different pathways that may operate at the same time and that even may affect each other (Figure 1.8). These are the hydrolytic and the oxidative resorption pathways. During hydrolysis, chemical bonds (mostly esters) of the polymer chain are cleaved by the reaction of water molecules, forming shorter polymer chains and finally oligomers or monomers that can be cleared from the body [120, 121]. Enzymes, like esterases, which are present in the blood or are secreted by macrophages and other activated cells after implantation of the scaffold, are known to accelerate this process [122]. For example, lipases are known to accelerate PCL resorption [ ]. The oxidative resorption pathway is mediated by reactive oxygen species (ROS) that are secreted by inflammatory cells, like macrophages, neutrophils and giants cells, that are recruited to the scaffold fibers [124, 126]. These ROS include hydrogen peroxide (H2O2), nitric oxide (NO), hydroxyl radical ( OH) and superoxide (O2 - ) and are responsible for chain scission of the polymers [127, 128]. Previous studies have investigated that oxidation of polymers is often initiated by abstraction of a hydrogen atom by radicals, resulting in chain scission of the polymer [127]. 12

22 General introduction Resorption can arise by two different mechanisms; surface erosion or bulk erosion [122]. In surface erosion, the exterior layer of the material is affected, while the core remains intact until the surrounding layer has been resorbed. This typically results in mass loss, thinning of the material and a stable molecular weight of the inner part of the material. Bulk erosion occurs throughout the whole material simultaneously, resulting in decreased molecular weight and mass loss throughout the material. 1 Figure 1.8 Schematic overview of in-vivo resorption pathways. After implantation, cells attach to the scaffold fibers and secrete both enzymes and ROS. This results in resorption of the scaffold fibers via the enzymatic accelerated hydrolytic pathway, and/or the oxidative pathway. Depending on the chemical composition and the morphology of the biomaterial, one of these pathways plays a more dominant role. Here, material A is affected by enzymatic hydrolysis, resulting in thinning of the fibers (surface erosion), while fibers are unaffected by the oxidative pathway. Material B is unaffected by the enzymatic pathway, but demonstrates broken fibers (bulk erosion) due to the ROS products generated in the oxidative pathway. [Drawing courtesy from Anthal Smits] Variation in resorption of biomaterials Resorption properties of widely used materials such as polyesters, polyethers and polyurethanes have been examined extensively. The mechanism and rate of material resorption depend on environmental factors, such as temperature, ph and mechanical stress [122]. Furthermore, the chemical composition and morphology of the polymers have an influence on the resorption rate [106]. In general, it is shown that polymers containing ester bonds react with water molecules and undergo hydrolysis. PGA is a hydrophilic material, in which water molecules can enter easily, resulting in fast hydrolytic resorption [129, 130]. The polyester PCL is a more hydrophobic material and results in slower hydrolytic resorption compared to PGA [129]. Solutions are able to be in contact with a larger surface area of the material, often resulting in faster resorption, when a porous scaffold is created compared to a solid film. Other polymers, including polyurethanes, were found to be more susceptible to the oxidative resorption pathway [127, 131]. 13

23 Chapter The host response to biomaterials Physiological wound healing is a response to injury, induced by the implantation of a biomaterial. This healing response involves complex, well-regulated processes, which include the four overlapping phases of haemostasis, inflammation, proliferation and remodeling (Figure 1.9). The entire healing response is mediated by cytokines and growth factors, which are secreted by different cell types involved in this host response The phases of the natural healing response Phase 1: Hemostasis (seconds to minutes) After the first interaction of the biomaterial with blood, proteins from the blood and interstitial fluid adsorb to the biomaterial, dependent on the biomaterial surface properties [126, 132, 133]. These proteins serve as binding sites for leukocytes [134]. Platelets also adhere to the biomaterial and secrete chemoattractants for immune cells that are involved in the second inflammatory phase. Phase 2: Acute inflammation (minutes to days) During the early phase of acute inflammation, the most prominent cell type that migrates from the blood toward the implanted biomaterial are neutrophils. After hours, these neutrophils undergo apoptosis and are phagocytosed by resident tissue macrophages. Monocytes enter the site of implantation and differentiate into macrophages. Macrophages function as phagocytic cells that clear wound debris and cell remnants, or foreign material. Phase 3: Proliferation (days to weeks) After 3 to 5 days, fibroblasts, which are recruited by macrophages, enter the site of implantation and start to deposit ECM proteins like fibronectin, collagen and proteoglycans. Furthermore, new blood vessels are generated by endothelial cells within the newly formed tissue during this regeneration phase. Phase 4: Remodeling (weeks to years) During the remodeling phase, there is clearance of macrophages. The final outcome of tissue regeneration or scar formation is dependent on the duration of the chronic inflammatory phase. In case of an optimal healing process, the scaffold is completely degraded and phagocytosed by the cells, while ECM is synthesized, matured, and remodeled simultaneously. In case of a prolonged healing response, fibrous scar tissue will be formed. It is believed that foreign body giant cells play an important role in this prolonged healing response, as they continuously activate fibroblasts, resulting in excessive deposition of ECM components [132]. This often results in encapsulation of the 14

24 General introduction (remaining) scaffold by avascular fibrous connective tissue instead of complete resorption of the scaffold and full replacement by native-like tissue [135]. 1 Figure 1.9 The four phases of wound healing, which include hemostasis, inflammation, proliferation and remodeling. Different cell types are involved in each phase, which undergo apoptosis when they fulfilled their function. Figure adapted from Enoch and Leaper Macrophage phenotypes Upon migration into affected or inflamed tissue, monocytes differentiate into macrophages. Dependent on micro-environmental signaling factors, macrophages can polarize into a heterogeneous population with different markers and functions. Different classes of macrophages, based on these markers and functions, have been identified and are believed to play an important role in the balance and final outcome of tissue regeneration or scar formation. The classically activated, pro-inflammatory macrophages are referred to as the M1 type. These are activated by pro-inflammatory signals, such as interferon gamma (IFN-ɣ) and lipopolysaccharide (LPS), and secrete pro-inflammatory cytokines and ROS. M2 type macrophages are the alternatively activated, antiinflammatory cell type, involved in immunoregulation and wound-healing [132]. These cells are activated by molecular cues such as IL-4 and IL-13. Others have described more subsets of the M2 macrophages, called M2a, M2b and M2c. M2a and M2b are both associated with wound healing and immunoregulatory functions, while M2c is involved in suppression of the immune response [ ]. Although the classes of macrophage phenotypes are defined, it is well known that these classes are the extremes of a continuum, and the macrophage phenotype is plastic and can change due to microenvironmental factors [136]. In an optimal healing process, the macrophages should undergo a phenotypic change from the M1 type during the inflammation phase towards the M2 type during the regenerative phase. Furthermore, several studies suggest that the 15

25 Chapter 1 scaffold surface, fiber diameter and pore size might also influence the macrophage phenotype [ ]. This indicates that scaffold fiber morphology and composition should be carefully selected to promote optimal healing responses. 1.7 Rationale and Outline One of the main challenges of tissue engineering is to control the balance between tissue formation and in vivo scaffold resorption. In order to design a scaffold with appropriate resorption properties for cardiovascular in situ tissue engineering applications, the aim of this thesis is to elucidate in vitro degradation characteristics of scaffolds manufactured from several (supramolecular) biomaterials and the effect of degradation rates on tissue formation and composition. A disturbed balance between tissue formation and scaffold degradation, where scaffolds degrade too fast in combination with traction forces exerted by the cells, resulted in compaction and retraction of in vitro tissue engineered heart valves, causing regurgitation in vivo when these valves were implanted [43, 71-73]. Therefore, we used in vitro tissue engineering in chapter 2, to study whether the use of 1) slow- (PCL) instead of a fastdegrading (PGA-P4HB) electrospun scaffold meshes and 2) a lower cell passage number to enhance tissue formation, has beneficial results on compaction. Furthermore, tissues were engineered using both ovine and human cells, to determine the effect of interspecies differences on tissue development. In our search for the appropriate scaffold material for cardiovascular applications, we investigated and discussed in chapter 3 how tissue development and composition changed during 6 weeks of in vitro culture, when cells were cultured on slow- (PCL) and fast-degrading (PGA-P4HB) electrospun scaffolds. Furthermore, the values of ECM components and collagen crosslinks were measured in the tissue engineered constructs and compared to values found in native human heart valves. To take another step forward in the world-wide availability of cardiovascular grafts, in situ tissue engineering seems to be a very promising alternative, as this approach results in a reduction in costs, production time, and regulatory issues related to tissue culture, compared to the classical in vitro tissue engineering approach. This also induces other demands, e.g. prolonged mechanical integrity, on the scaffold material, as the grafts are implanted as bare scaffolds, without any pre-cultured tissue. Supramolecular materials like PCL-UPy or PCL-BU are promising for in situ tissue engineering as these materials comprise strong and elastic properties, which are desired properties to replace loadbearing tissues like heart valves. To better understand the degradation characteristics of various electrospun scaffolds, accelerated in vitro degradation assays were designed in chapter 4. With the use of these assays, the degradation characteristics of different 16

26 General introduction electrospun supramolecular materials were explored and compared to a conventional material. Macrophages are known to play an essential role in the resorption of the implanted scaffold meshes. To illustrate whether there is a correlation between the inflammatory (M1) or healing (M2) macrophage phenotypes and degradation of electrospun meshes, in vitro culture systems were used in chapter 5. In addition, we investigated the preferred polarization phenotype of macrophages when cultured onto PCL meshes with a fiber diameter of 10 µm. 1 Finally, the main findings of the thesis are summarized and discussed in chapter 6. This includes a discussion on remaining challenges and the required future (research) steps towards safe clinical application of cardiovascular tissue replacements. 17

27 Chapter 1 18

28 Poly-ε-caprolactone scaffold and reduced in vitro cell culture: 2 Beneficial effect on compaction and improved valvular tissue formation M. Brugmans A. Driessen-Mol M. Rubbens M. Cox F. Baaijens J Tissue Eng Regen Med

29 Chapter Abstract Tissue-engineered heart valves (TEHV), based on PGA scaffold coated with poly-4- hydroxybutyrate (P4HB), have shown promising in vivo results in terms of tissue formation. However, a major drawback of these TEHV is compaction and retraction of the leaflets causing regurgitation. To overcome this problem the aim of this study was to investigate 1) the use of the slow degrading PCL scaffold for prolonged mechanical integrity and 2) the use of lower passage cells for enhanced tissue formation. Passage 3, 5 and 7 (p3, p5 and p7) human and ovine vascular-derived cells were seeded onto both PGA-P4HB and PCL scaffold strips. After 4 weeks of culture, compaction, tissue formation, mechanical properties and cell phenotypes were compared. TEHV were cultured to observe retraction of the leaflets in the native like geometry. After culture, tissues based on PGA-P4HB scaffold showed 50-60% compaction, while PCLbased tissues showed compaction of 0-10%. Tissue formation, stiffness and strength were increased with decreasing passage number, however, this did not influence compaction. Ovine PCL based tissues did render less strong tissues compared to PGA-P4HB based tissues. No differences in cell phenotype between the scaffold materials, species or cell passage numbers were observed. This study shows that PCL scaffolds may serve as alternative scaffold material for human TEHV with minimal compaction and without compromising on tissue composition and properties, while further optimization of ovine TEHV is needed. Reducing cell expansion time will result in faster generation of TEHV, providing a more rapid treatment to patients. 20

30 PCL scaffold and reduced in vitro cell culture 2.2 Introduction With an increasing number and aging of the world population, valvular heart disease is an expanding health problem. Approximately heart valve replacements are performed annually worldwide and this number is estimated to increase up to by 2050 [22]. Bioprosthetic and mechanical heart valves, which are successfully used for decades, improve quality of life and life prolongation for most patients [146, 147]. However, these valves have some restrictions as they consist of non-living and nonautologous materials. Therefore, they are not able to grow, adapt or remodel to changing physiological environments, resulting in decreased durability [22]. Furthermore, bioprosthetic valves are susceptible to calcification, while mechanical valves require lifelong anticoagulation therapy to prevent thrombo-embolism [22, 146]. To overcome these problems, researchers are studying the possibility of creating tissue engineered heart valves (TEHV) [146]. Patients own cells are incorporated, resulting in valves of autologous living tissue that are able to grow, remodel and adapt to the changing environment after implantation[146]. Our approach to create such TEHV is to isolate patients cells from the vena saphena magna, expand them in vitro up to the desired amount of cells and subsequently seed them onto a bioresorbable synthetic scaffold in the shape of a heart valve. After a culture period in a bioreactor of 4 weeks, where the valves are exposed to mechanical stimuli in order to stimulate tissue formation, the valves are able to withstand systemic pressures in in vitro tests [148], aiming ultimately at implanting them into patients. 2 Different types of synthetic scaffolds are used for cardiovascular tissue engineering applications. In particular, PGA scaffold, coated with P4HB, or combined with another scaffold material, showed to be a promising candidate in terms of tissue formation, as demonstrated in vascular graft studies [42, 60] and in vivo TEHV studies [68, 71, 72, 149]. Hoerstrup et al. demonstrated in an ovine model that after 20 weeks in vivo, the valves yielded an organized, layered structure with many architectural features and ECM characteristics that are present in native valves. In vivo, PGA and P4HB are resorbed completely within 4 and 8 weeks respectively [68]. The downside of using this rapid resorbing PGA scaffold is compaction (flattening of the leaflets) and retraction (shrinkage of the leaflets), causing regurgitation [71, 72, 150]. This is a result of traction forces exerted by the cells, likely in combination with an imbalance of the newly formed tissue and loss of mechanical integrity of the scaffold due to degradation [74, 148, 151]. Rabkin- Aikawa demonstrated TEHV containing αsma positive cells during in vitro culture, while after 20 weeks in vivo, there was a strong decrease of αsma positive cells [6]. As αsma is related to traction forces of the cells [152], we assume that after 20 weeks, these traction forces will be decreased in vivo. Therefore, a scaffold with proper mechanical integrity during in vitro culture and the first months after implantation is desired to withstand the cell traction forces during this phase. The use of a slower degrading scaffold material such 21

31 Chapter 2 as PCL may represent a promising alternative, as TEHV can be produced that are mechanically reliable for months, thereby offering sufficient mechanical integrity to prevent tissue compaction and retraction [153]. As PCL can be processed by electrospinning, it is possible to create complex geometries and mold the scaffold directly into the desired 3D shape of a heart valve [153]. This direct 3D molding is not feasible for PGA scaffolds, which are only available in sheets. Another benefit of PCL is the possibility to create thin leaflets with a thickness of 300 µm, while the PGA meshes are produced with a thickness of 1000 µm. As PGA-P4HB scaffolds are more rapidly degrading, the cells might be exposed to larger magnitudes of mechanical loading as compared to the cells in PCL scaffolds, which might on their turn be partly protected from loads by the long-term presence of the scaffold. As the stress level exerted on the vascular cells is known to change phenotype of the cells towards activated myofibroblasts[154], tissue formation capacity of cells in the two scaffold types might differ, along with different phenotypes [ ]. Therefore, it is important to compare cell phenotype, tissue formation capacity and compaction in tissues based on both scaffold types when considering the use of PCL as a scaffold material to produce TEHV. Based on the above, we hypothesize that the cells in PGA-P4HB might have a more activated phenotype accompanied by increased tissue formation capacity as compared to cells in PCL scaffolds. Another alternative to tackle compaction and retraction of TEHV might be by using cells of a low passage number. Aging cells, due to in vitro expansion, lose their potential to proliferate [159, 160]. Currently in our lab, cells are expanded up to passage 6-7 to ensure enough cells for seeding multiple TEHV [148]. Whether the amount of tissue formation or cell phenotype in 3D cultures is influenced by the use of cells of a low passage number is still unclear as to the best of our knowledge, previous work on the effect of cell aging by expansion has been performed on 2D cultures only. Therefore, the role of cell aging in 3D tissue formation capacity needs to be further investigated. We hypothesize that cells of a low passage number (passage 3) are more productive, resulting in more tissue formation and of a higher quality, compared to cells of a high passage number (passage 7). This improved tissue formation capacity on its turn may result in less compaction and retraction, as it is influencing the balance between matrix quality and the mechanical integrity of the scaffold towards increased matrix quality. We assume that the increased matrix formation will increase the resistance to the traction forces exerted by the cells. An additional benefit of using cells of a lower passage number is the reduction in cell expansion time, which will result in faster generation of TEHVs and, thereby, providing a more rapid treatment to patients. To summarize, the aim of this study is to evaluate alternative approaches to overcome the compaction and retraction of TEHV as observed with the use of rapid degrading PGA- P4HB scaffolds, without compromising on tissue composition and properties. The alternative approaches that are being studied here are 1) the use of a slow degrading PCL 22

32 PCL scaffold and reduced in vitro cell culture scaffold for prolonged mechanical integrity and 2) the use of lower passage vascular cells for enhanced tissue formation. Compaction, tissue formation, cell phenotype and mechanical properties of engineered tissues based on passage 3, 5 and 7 vascular cells in both PCL and PGA-P4HB scaffolds are compared. TEHV aim to be designed for humans, but since the ovine model is commonly used to show proof of principle, both human and ovine cells were used. 2.3 Materials and methods Cell culture 2 Human vascular-derived cells were harvested from segments of a vena saphena magna from a 60 years old patient that underwent bypass surgery, and was obtained according to the Dutch guidelines for secondary used materials. Ovine vascular-derived cells were obtained from the vena jugularis of adult sheep of approximately 2 years old (n=2, Swifter). The cells were isolated via the outgrowth method. In short, endothelial cells of the vessels were removed by incubation with a collagenase solution. Remaining endothelial cells were removed from the lumen side using a cell scraper. After removal of the endothelial cells, the vessels were minced into small pieces of approximately 1 mm 2 and the fragments were plated into 6 wells-plates. The outgrowing cells were expanded using standard culture methods in a humidified atmosphere containing 5% CO2 at 37 C, and passaged at % confluency. Plating densities were *10 3 per cm 2 for human and *10 4 per cm 2 for ovine cells, based on differences in cell size. Isolation and expansion medium consisted of advanced Dulbecco s Modified Eagle Medium (DMEM; Invitrogen, Breda, Netherlands), supplemented with 1% GlutaMax (Invitrogen), 1% Penicillin/Streptomycin (P/S, Lonza, Basel, Switzerland), and 10% Fetal Bovine Serum (FBS, Greiner Bio one, Frickenhausen, Germany) for human cells or 10% Lamb Serum (Invitrogen) for ovine cells. During culture, cells of all passage numbers grew in the characteristic hill and valley morphology, indicating smooth muscle cells Scaffold preparation and sterilization Rectangular strips (25x5 mm) were cut out of PGA meshes (PGA, Cellon, Bascharage, Luxemburg) and conventionally electrospun PCL meshes, with a thickness of 1000 μm and 300 μm, respectively. As heart valves contain a more complex geometry compared to strips, which might result in differences in terms of compaction, trileaflet heart valve scaffolds were fabricated using scaffold meshes of the same thickness. PGA scaffolds were additionally coated with poly-4-hydroxybutyrate (P4HB, received via a collaboration with Prof. Hoerstrup of the University Hospital Zurich) to provide structural integrity to the 23

33 Chapter 2 mesh. The outer 3-4 mm of both PGA and PCL scaffold strips were attached onto stainless steel rings (RVS Paleis, Geleen, Netherlands) using 15% polyurethane-tetrahydrofuran (PU, Desmopan) glue, leaving an 18*5 mm area for cell seeding. The solvent was allowed to evaporate overnight in a vacuum oven. PCL scaffolds were sterilized by gamma irradiation (Isotron, Ede, Netherlands). PGA-P4HB scaffold sterilization was achieved by immersion in 70% sterile ethanol for 30 minutes. To facilitate cell attachment, the scaffolds were incubated overnight with tissue engineered (TE) medium, consisting of expansion medium supplemented with 0.25 mg/ml L-ascorbic acid 2-phosphate (Sigma). Lamb serum (0.1%) and FBS (10%) was added to ovine and human TE medium, respectively Cell seeding and tissue culture Passage 3, 5 and 7 (referred to as p3, p5 and p7) cells were seeded onto both PGA and PCL scaffolds (n=6 per passage and scaffold for each cell type), with a seeding density of 20 million cells per cm 3 using fibrin as a cell carrier [161]. In short, cells were suspended in TE medium containing thrombin (10 U/ml, Sigma). This cell suspension was mixed with an equal volume of TE medium containing fibrinogen (10 mg/ml, Sigma) and dripped onto one side of the scaffolds before polymerization of the gel was accomplished. As control strips, three PGA and PCL scaffolds were seeded with fibrin only. After seeding, the scaffolds were placed in an incubator at 37 C for 30 minutes, to allow polymerization of the fibrin gel. Thereafter, 6 ml of TE medium was added to each scaffold. The strips were cultured for 4 weeks and TE medium was changed twice a week. For the heart valve cultures, passage 7 cells were used and seeded according to similar protocols as for the strips. After seeding, the valves were placed in a bioreactor system and cultured for 4 weeks [43] Compaction Compaction was assessed from upper view photographs of the strips that were taken once a week. The valves were photographed after 4 weeks only. Compaction of the strips was defined as the reduction of width, compared to the width at the start of culture. Photographs were analyzed using the program Image J (version 1.43u) Biochemical assays For the quantification of tissue formation after 4 weeks of culture, TE strips were lyophilized after mechanical testing (n=4-5 per group) and used for biochemical assays. The total amount of DNA was determined as an indicator of cell number, the amount of hydroxyproline as an indicator for collagen content, and the amount of sulfated 24

34 PCL scaffold and reduced in vitro cell culture glycosaminoglycans (sgag). Measurements were averaged per group. Lyophilized tissue samples were weighted and digested in papain solution (100 mm phosphate buffer (ph=6.5), 5 mm L-cysteine, 5 mm ethylene-di-amine-tetra-acetic acid (EDTA), and μg papain per ml, all from Sigma) at 60 C for 16 hours. After centrifuging the samples, the digest supernatant was collected and used for the DNA, sgag and collagen assays. The amount of DNA in the TE strips was determined using the Hoechst dye method [162] and a standard curve prepared of calf thymus DNA (Sigma). Using the assumption that all cells contain 6.5 pg of DNA [163], the amount of cells per TE construct was calculated. sgag content was determined with a modification of the protocol described by Farndale et al. [164]. In short, 40 μl of diluted sample was pipetted into a 96-well plate in duplo followed by addition of 150 μl di-methyl-methylene blue per well. Absorbance was measured at 540 and 595 nm and extracted from each other. Subsequently, the amount of sgags in the TE strips was determined from a reference curve prepared from shark cartilage chondroitin sulfate (Sigma). Collagen content was determined by an assay as described by Huszar et al [165], and a standard curve was prepared from trans-4- hydroxyproline (Sigma) Mechanical testing After 4 weeks of culture, the mechanical properties of the TE strips (n=4-5 per group) were assessed by uniaxial tensile tests in longitudinal direction with a uniaxial tensile stage (Kammrath &Weis, Dortmund, Germany) equipped with a 20N load cell. Mechanical test data was averaged per group. Thickness of the strips was determined from representative histology sections. Samples were measured at three spots and mean thickness was used. Standard deviation of the measurements ranged %. Stress-strain curves were obtained and as a measure for tissue strength, the ultimate tensile strength (UTS) was defined as the peak stress value. The elasticity modulus (E-modulus) was determined as the slope of the linear (end) part of the curve, as a measure for tissue stiffness Histology To analyze tissue formation qualitatively, TE strips were processed for histology (n=1 per group). Representative tissue samples were embedded in tissue freezing medium (Tissue Tek, Sakura, Torrance, USA) and cryosections of 10 μm were cut. The sections were formalin-fixed and studied by Masson Trichrome (MT) staining (MTC kit, Sigma, Venlo, Netherlands) for collagen deposition and by Picrosirius Red (PR) staining to assess the maturity of the collagen matrix [166]. The MT staining was analyzed using light microscopy and the PR staining by polarized light microscopy (Axio Observer, Zeiss, Göttingen, Germany). In this study, maturity of the collagen fibers was assessed by amount and density of the collagen fibers visible with polarized light microscopy. Mature fibers with a high density are colored orange/red, while immature or less dense fibers are green. 25

35 Chapter 2 Cell phenotype within the TE strips was analyzed by immunofluorescence. After acetone fixation for 10 minutes, the sections were incubated with 5% bovine serum albumin (BSA) in PBS for 30 minutes at room temperature. After blocking, the sections were incubated with a primary antibody overnight at 4 C. Antibodies used were mouse anti α-smooth muscle actin (αsma) to stain smooth muscle cells and myofibroblasts (a2547, clone 1A4, Sigma, 1:400), mouse anti-smoothelin to stain contractile smooth muscle cells (Clone R4A, kindly provided by GJ van Eys from the University Maastricht, 1:4) or rabbit anti S100A4, which belongs to the S100 superfamily of cytoplasmic calcium-binding proteins, to stain fibroblasts and myofibroblasts (ab27957, Abcam, 1:200). After primary antibody incubation, the sections were washed with PBS and incubated with Alexa 488 labeled secondary antibodies (Sigma and Molecular probes, 1:300) to visualize the specific stainings and DAPI (Sigma, 1:500) to stain all cell nuclei for 30 minutes at room temperature. After staining, sections were mounted with Mowiol 4-88 (Calbiochem, San Diego, USA) and visualized by fluorescent microscopy (Axiovert 200M, Zeiss, Göttingen, Germany) Statistical analyses All data are presented as mean ± standard error of the mean. Data of all experiments were normalized to human passage 3 PGA-P4HB strips in each experiment to be able to compare experiments and perform statistical analyses. Pearson correlation coefficients were calculated to determine correlations between tissue parameters and cell passage numbers for both species and scaffold groups. Unpaired t-tests were used to compare the tissue properties between the scaffold materials within one cell passage and species, and to compare the tissue properties between species, within the same scaffold material and cell passage number. Statistics were performed using GRAPHPAD Prism (version 5.04) and differences were considered significant for p-values < Results Compaction after 4 weeks The remaining width of the strips of all groups after 4 weeks of culture is shown in Figure 2.1 A. A remaining width of strips of 100% is the initial width of the strips and represents no compaction. The tissues based on PCL scaffold, and PCL and PGA-P4HB control strips, showed compaction of 0-10%. The tissues based on PGA-P4HB scaffold resulted in significant more compaction of around 50% after 4 weeks (p<0.001). In ovine strips, no significant correlation between passage number and both types of scaffold was found. A negative correlation was found between human cell passage numbers and PGA-P4HB strips (p<0.01), while there was a positive correlation between the human cell passage numbers and PCL strips (p<0.05). This indicates that passage 26

36 PCL scaffold and reduced in vitro cell culture number and species did not consistently influence compaction. TEHV based on PGA-P4HB scaffolds show severe compaction and retraction of the leaflets after 4 weeks culture in both species, while no compaction or retraction was observed in the PCL based valves (Figure 2.1 B-E), confirming the results as found in the engineered strips. 2 Figure 2.1 Compaction of strips after 4 weeks of culturing. Initial width of strips was set at 100% (dotted line) (A). PGA-P4HB showed around 50% compaction of the strips, while the use of PCL strips demonstrated reduced compaction as the final reduction in width was 0-10% only. ** indicates the difference between the scaffold materials with a p-value<0.001, while # and ## denote significant differences of p<0.05 and p<0.001 compared to human tissues respectively. Negative or positive Pearson r correlations between the cell passage numbers are presented by arrows combined with their p-values. Species and cell passage number did not consistently influence compaction of the TE strips. Top view photos of a human PGA-P4HB (B), human PCL (C), ovine PGA-P4HB (D) and ovine PCL (E) TEHV after 4 weeks of culture. Valves based on PGA-P4HB scaffold resulted in severe retraction of the leaflets after 4 weeks, while PCL valves did not show this. These results were consistent for both human and ovine cells Biochemical assays Normalized collagen and sgag per DNA of all groups are presented in Figure 2.2. Significant negative correlations between cell passage numbers and collagen amount per DNA, were found in both human and ovine tissues of both scaffold materials (p<0.001), demonstrating that increasing passage number, resulted in decreased collagen per DNA. Low amount of collagen per DNA was detectable in ovine PCL p7 strips. In general ovine tissue strips demonstrated an increased amount of collagen when compared to human (p<0.001). Collagen content per DNA of both human and ovine p7 cells seeded on PCL scaffolds was decreased, compared to human and ovine cells that were seeded on PGA- P4HB scaffolds (p<0.05 for human cells and p<0.001 for ovine cells). Although we showed that collagen and sgag per DNA was increased with decreasing passage number, no differences in compaction of the tissues could be observed. Biochemical parameters are related, as observed by correlation matrices, showing that collagen per DNA was increased when sgag per DNA was increased. Overall, the amount 27

37 Chapter 2 of sgag per DNA decreased with increasing cell passage (p<0.05 for ovine PGA-P4HB strips and p<0.001 for human PCL strips) although this effect was less pronounced as seen for to collagen per DNA. Except ovine p7 PCL strips, ovine cells resulted in a higher amount of sgag per DNA compared to human cells (p<0.05 for ovine p3 PCL strips and p<0.001 for all other ovine strips). No consistent differences in sgag content by the cells were observed due to different scaffold materials. Figure 2.2 Collagen per DNA (A) and sgag per DNA (B). # and ## denote significant differences compared to human tissues, while * and ** denotes significances of differences between scaffold materials with p<0.05 and p< Pearson r correlations between the cell passage numbers are presented by arrows combined with their p-values. Collagen per DNA is decreasing with increasing passage number in both human and ovine tissues and both scaffold materials (A). sgag per DNA show the same trends although less distinct (B). Scaffold does not influence the amount of formed collagen and sgag, while per DNA, more collagen and sgag are formed within ovine tissues compared to human tissues Mechanical testing In Figure 2.3A and 2.3B, averaged stress strain curves of the human and ovine p3 strips, which are representative for the other passage numbers, and the PGA-P4HB and PCL control strips are presented. Bare PCL strips were able to bear higher stresses compared to bare PGA-P4HB strips, which is due to the differences in degradation time of both scaffold materials. The PGA-P4HB cultured tissues of both human and ovine cells showed typical non-linear mechanical behavior representing tissue behavior. When PCL scaffold was used, human tissues showed linear mechanical behavior, while the ovine tissues were following the curve of the control PCL strips. Thus, PCL scaffolds are still influencing the mechanical properties of the engineered tissues after 4 weeks of culture, while PGA-P4HB scaffolds do not. 28

38 PCL scaffold and reduced in vitro cell culture 2 Figure 2.3 Mechanical data of engineered strips. Averaged stress strain curves of human (A) and ovine (B) p3 strips are given as mean ± SEM. PGA-P4HB based tissues demonstrate non-linear curves in both human and ovine strips, representing tissue behavior. The stress strain curve of human PCL strips is linear, while ovine strips follow the curve of the control scaffolds. Control PCL scaffolds are still influencing mechanical properties after 4 weeks of culture, while PGA-P4HB scaffolds are not. Tissue stiffness (C) and strength (D) are increasing with decreasing passage number. # and ## denote significant differences compared to human tissues, while * and ** denotes significances of differences between scaffold materials with p<0.05 and p< Pearson r correlations between the cell passage numbers are presented by arrows combined with their p-values. In human samples, highest values are obtained in PCL strips, while in ovine this is observed in PGA-P4HB scaffold strips. With a decrease of cell passage numbers, the parameters stiffness and strength were increasing in both species and scaffold materials, as significant negative correlations were observed between increasing cell passage numbers and both the stiffness (p<0.05 for human PGA-P4HB strips and p<0.001 for human PCL and ovine PGA-P4HB strips) and strength (p<0.05 for human PGA-P4HB and ovine PCL strips and p<0.001 for human PCL and ovine PGA-P4HB strips), in human and ovine tissues based on both scaffold materials (Figure 2.3C and 2.3D). In human tissue samples, stiffness was higher in PCL samples compared to PGA-P4HB samples (p<0.05 in p3 and p7 tissues), while in ovine tissue samples a higher stiffness was obtained in tissues based on PGA-P4HB scaffolds compared to PCL scaffolds (p<0.05). Furthermore, tissue strength was increased in human PCL samples of all passage numbers and ovine PCL samples of passage 5 and 7, compared to PGA-P4HB tissue samples (p<0.05) which probably is due to the influence of the PCL scaffold that is not yet degraded. When PCL scaffold was used, the values of the 29

39 Chapter 2 mechanical properties of the ovine tissues were equally or just slightly increased compared to the PCL control strips, while the values of the human tissues were higher compared to the control strips (data not shown). This indicated that the newly formed tissues by ovine cells were not of the same quality as their human equivalents, as the added value of tissue to the mechanical properties of the ovine strips was relatively low. Correlation matrices demonstrated that mechanical parameters are related to each other, resulting in increased tissue strength when tissue stiffness obtained higher values, while mechanical parameters were not related to matrix properties of the tissues Histology Histology of the TE strips revealed cellular tissue with dense surface layers. Picrosirius Red and Masson Trichrome stainings (Figures 2.4 and 2.5), showed collagen fibers in strips of all groups after 4 weeks of culture. A higher amount of red fibers was seen in most tissues with cells of a low passage number (Figure 2.4). This indicated that tissues based on a low cell passage number resulted in more mature collagen fiber formation. Histology furthermore indicated that the total amount of collagen fibers was decreasing with increasing passage numbers in both PGA-P4HB and PCL strips (Figure 2.5). Ovine PGA- P4HB tissues showed a higher amount of collagen compared to the human tissues. However, ovine PCL based tissues showed little amount of collagen compared to human PCL based tissues. The total amount of collagen was higher in PGA-P4HB strips compared to PCL strips, which can be explained by triple the amount of cells seeded onto the PGA- P4HB strips compared to PCL strips, due to differences in thickness of the scaffold materials. Immunofluorescent stainings indicated no differences in cell phenotype with cell passage number, scaffold material or species, as tissues of all groups contained cells that were αsma and S100A4 positive, and smoothelin negative (Figure 2.6), indicative for synthetic myofibroblasts. Cells in all strips were distributed homogenously throughout the strips as shown by cell nuclear staining (DAPI). 30

40 PCL scaffold and reduced in vitro cell culture P3 P5 P7 Ovine PCL Ovine PGA-P4HB Human PCL Human PGA-P4HB 2 Figure 2.4 Picrosirius Red stained sections of human PGA-P4HB (A-C), PCL (D-F), ovine PGA-P4HB (G-I) and PCL (J-L) visualized by polarized light microscopy. Maturity of collagen fibers is visualized as green (immature) and orange/red (mature). Most red fibers are visualized in tissues based on cells with a low passage number, indicating that maturity of collagen fibers after 4 weeks of culture is decreasing with increasing passage number. The white scale bars represent 200 µm. Vacuoles within the scaffolds are cutting artifacts due to scaffolds remnants, and grey parts in the PGA-P4HB groups are P4HB remnants. 31

41 Ovine PCL Ovine PGA-P4HB Human PCL Human PGA-P4HB Chapter 2 P3 P5 P7 Figure 2.5 Masson Trichrome staining of human PGA-P4HB (A-C), PCL (D-F), ovine PGA-P4HB (G-I) and PCL (J-L) sections. The blue scale bars represent 200 µm. Collagen is shown in blue and red represents cytoplasm and muscle tissue. Vacuoles within the scaffolds are cutting artifacts due to scaffolds remnants. The total amount of collagen fibers seem to decrease with increasing passage number in both scaffold materials. Ovine PGA-P4HB strips show more collagen compared to human strips, while in PCL strips most collagen is visualized in human samples. Figure 2.6 Representative photos of immunofluorescent stainings of the αsma (A), S100A4 (B) and the Smoothelin (C) cell markers, with the white scale bars representing 200 µm. In green the protein of interest is colored, in blue DAPI is visible to stain cell nuclei. All stained tissues contain cells that were αsma and S100A4 positive and smoothelin negative. This indicates that passage number, scaffold material and species are not influencing cell phenotypes. Vacuoles within the scaffolds are cutting artifacts due to scaffolds remnants. 32

42 PCL scaffold and reduced in vitro cell culture 2.5 Discussion Compaction and retraction of heart valve leaflets in vitro, resulting in regurgitation in vivo, is a common problem in TEHV that are based on rapid degrading PGA-P4HB scaffolds. Therefore, alternative approaches to overcome compaction and retraction of TEHV are needed to meet in vivo demands. This study has focused on the effect of two alternative approaches: 1) the use of a slow degrading PCL scaffold and 2) the use of lower passage vascular cells. Compaction, tissue formation, cell phenotype and mechanical properties of both human and ovine tissues were investigated Differences due to vascular cell expansion times In this study, we demonstrated that reduced in vitro expansion time of vascular cells resulted in improved tissue amount as sgag per DNA, collagen per DNA, tissue strength and stiffness were increased with decreasing passage number. A comparison of the net amounts of collagen and sgag could not be made, as different amounts of cells were seeded, due to differences in scaffold thickness. Therefore, collagen and sgag were normalized to DNA. A 2D study of ovine jugular vein derived cells showed that sgag content was highest in low passage cells [167]. Although cells in 2D may act differently compared to cells in 3D, our data also indicated that cells with an increasing passage number became less synthetic, as collagen and sgag content was decreased by cells of a higher passage number. Some in vitro studies showed that the vascular contractile smooth muscle cell marker smoothelin, disappeared within a few days of in vitro expansion, and cells differentiated into synthetic, tissue producing cells [168], while others observed this only after the 9-11 th passage [169, 170]. All our human and ovine cells have been differentiated into the synthetic phenotype, as no change of phenotype could be observed in this study due to cell passage number, and all tissue sections showed αsma and S100A4 positive, and smoothelin negative cells indicating activated, synthetic myofibroblasts. Cell phenotype of our samples and amount of tissue were not related as no change in cell phenotype could be observed, while it was shown that the amount of tissue increased with decreasing passage numbers. The tissue stiffness of strips was obtained from the linear end part of the stress strain curves and represents the end stiffness of our tissues. Increase of tissue stiffness, was seen in strips based on a decreased cell passage number. The increase in end stiffness of our tissues resulted in stronger tissues, although, native leaflets still do show much higher values of stiffness compared to our tissues; 15.6 ± 6.4 MPa in the circumferential direction and 2.0 ± 1.5 MPa in the radial direction [171]. Native valves are also more flexible compared to our engineered strips when comparing the physiological relevant stiffness. The opening and closing functions of the heart valves are controlled by pressure differences. As the native valves are more flexible compared to their engineered counterparts, a lower pressure is needed for opening the valves. 33

43 Chapter 2 Histology of the PGA-P4HB samples confirmed the biochemical results of collagen content per DNA, as higher amounts of collagen were observed in the ovine PGA-P4HB tissues compared to the human tissues. This increased amount of collagen in ovine PGA-P4HB based tissues is not only explained by increased synthetic ovine cells, but also by a higher proliferation rate of these cells compared to human cells when seeded on PGA-P4HB scaffolds (proliferation data not shown). However, ovine PCL based tissues show little collagen in the histology slides compared to human PCL based tissues, while the biochemical data showed an increased amount of collagen per DNA in ovine tissue compared to human. This can be explained by the proliferation rate of ovine and human cells in PCL scaffolds. As human cells showed a higher proliferation rate when seeded onto PCL scaffolds (data not shown) and, therefore, an increased amount of total DNA per strip in PCL scaffolds compared to ovine cells, the amount of collagen per DNA is lower in human, while the total amount of collagen per strip might be higher due to the presence of more collagen producing cells. More research is needed to investigate why differences in proliferation rates of ovine and human cells are present when different types of scaffolds are used. Mechanical results also correlated with the histological findings. Strips that showed more, and increased maturity of collagen fibers, also resulted in an increased tissue stiffness and strength. This is in line with previous findings, where a dominant role for collagen maturity by cross-linking of the collagen over collagen content was found with respect to mechanical properties of the tissues [171]. Remarkable is that ovine p7 PCL strips resulted in only few cells present after 4 weeks. Collagen content of these cells was also low resulting in weak strips as observed in the tensile tests. We hypothesize that this might be due to the combination of several factors. One might be the use of a low amount of serum (0.1% in ovine 3D medium). This could have resulted in non-synthetic and non-dividing cells. In combination with the high passage number, which also showed to result in less activated or synthetic cells, this could have been the reason for the low amount of cells present after 4 weeks and reduced amount of collagen. Furthermore, the use of PCL scaffold is likely to have influenced the amount of collagen, as ovine p7 cells seeded on PGA-P4HB scaffolds, did show higher amounts of collagen. We hypothesized that the use of PCL scaffold with ovine cells, resulted in non-synthetic cells, as the mechanical integrity of this scaffold was present for a longer time span, resulting in no urgent need for the cells to create tissue. However, culturing TEHV with ovine p7 cells did result in proper tissue formation. This might be explained by different culture protocols of engineered strips and TEHV. TEHV undergo mechanical loading in a bioreactor during culture, while strips are cultured statically. Furthermore, interspecies differences might have played a role, as cells of a different sheep were used to culture the TEHV. 34

44 PCL scaffold and reduced in vitro cell culture Concerns might rise about the clinical applicability of using cells with a low passage number, mainly in children, as a relatively large number of cells need to be obtained. However, in the case of children fewer cells are needed to be able to produce a TEHV compared to adults, as the annulus of the pulmonary valve in children is mm, while this is around 25 mm in adults. The size of the leaflets in young patients is also smaller. Furthermore, when PCL based TEHV are produced instead of PGA-P4HB based TEHV, fewer cells are needed due to differences in scaffold thickness. To produce a PCL based TEHV scaffold for adults, 20 x 10 6 cells are needed, while this would be 2-10 x 10 6 cells is case of children. These amounts of cells can be obtained by the outgrowth method as the saphenous vein segments need to be a centimeter only. In conclusion, cells from a lower passage number demonstrated to increase the amount of tissue formation and tissue strength, without influencing cell phenotype. Despite the improved tissue formation, compaction of the tissues was not influenced by a lower cell passage number PGA-P4HB versus PCL scaffold In this study, we demonstrated that human and ovine tissues cultured for 4 weeks using PCL scaffold strips showed almost no compaction (0-10%), while PGA-P4HB based tissues showed compaction up to 50%. Furthermore, we showed that PGA-P4HB based TEHV resulted in severe retraction of the leaflets in both species, while this was not seen in the PCL based TEHV. This proves that PCL is a promising scaffold material to reduce compaction and retraction in TEHV. Dijkman et al described another approach to prevent compaction and retraction of PGA-P4HB based TEHV [43]. Trileaflet heart valve of PGA- P4HB scaffolds were seeded with ovine myofibroblasts and subsequently decellularized to prevent retraction. Decellularization represented to be a powerful tool to reduce tissue retraction, as it was shown that cell-induced retraction accounted for 85% of total tissue retraction. Residual matrix stresses are known to still account for 15% of the total retraction [74]. These residual matrix stresses minimized the coaptation area in the study of Dijkman et al. and it has to be elucidated in future studies whether this will influence in vivo valve behaviour. We believe that by using a slow degrading scaffold, retraction can be even more effectively reduced by resisting residual matrix stresses, while maintaining tissue viability. Results of the mechanical tests demonstrated that in PCL strips the mechanical properties were not only determined by the formed tissue, but also by the remaining PCL scaffold, as it was not yet degraded. PGA-P4HB is known to start to degrade after 2 weeks, and, therefore, was not influencing the mechanical properties of the tissues. As amounts of sgag and collagen per DNA were not influenced by the scaffold materials, the increased tissue strength of the human PCL strips compared to the PGA-P4HB strips are likely due to the remaining PCL scaffold. Ovine strips did not show the same results, which might be due to the low amount of DNA and, therefore, a lower amount of total tissue, in ovine PCL 35

45 Chapter 2 strips. Mechanical properties of the ovine PCL strips were mainly influenced by the remaining scaffold and not by the formed tissue, while in PGA-P4HB strips the measured mechanical properties represented tissue only. Furthermore, ovine tissues based on PCL scaffold did not influence mechanical properties as much as compared to human PCL tissues, as tissue strength and stiffness values were equally or just slightly increased compared to the PCL control strips. This indicated that the newly formed tissues based on ovine cells were not of the same quality as their human equivalents. Differences in scaffold thickness could possibly have resulted in differences in tissue formation, due to variation in nutrient and oxygen levels within the tissues. This is mainly seen in ovine strips as human strips show more homogeneously distributed tissue. Our ovine strips possess a denser layer of collagen and cells on the surface in both scaffold material. However, cells were not only present at the surface layer, but also distributed throughout the center of both scaffold materials. Not only the cells at the surface layer, but also the cells in the center produced collagen and expressed the synthetic smooth muscle cell markers, as visualized by histology. Furthermore, biochemical assays demonstrated no influence of the scaffold materials on the collagen and sgag formation per DNA, and differences in mechanical properties of the tissues are most likely due to PCL scaffold remnants instead of differences between material thicknesses. Directly after seeding, the high porosity of the scaffold strips allowed oxygen and nutrient supply to the cells that were situated on the scaffold fibers in the middle part of the strip. When tissue was produced, porosity decreased and oxygen and nutrient supply might have been decreased resulting in the formation of surface layers. Native human heart valve leaflets are avascular as they are thin enough to receive nutrients and oxygen through diffusion and hemodynamic convection [18]. As PCL scaffolds are 300 µm, we do not expect problems when placing PCL TEHV in vivo. TEHV based on PGA-P4HB did show increased thickness in the ovine model [149], which might lead to reduced oxygen and nutrient supply to the cells in the center. This problem might be less pronounced in human as these tissues are also compacting in the vertical direction, and therefore decreasing in thickness. In conclusion, the use of PCL scaffold seems to be an alternative scaffold material for the culture of human TEHV to reduce compaction, while further optimization is needed when ovine cells are used to ensure proper tissue formation Interspecies differences Tissue properties were different between species. In our study, ovine cells presented to be more synthetic compared to human cells as they contained more sgag and collagen per DNA, while a study by van Geemen et al, demonstrated the opposite effect [48]. Van Geemen showed that human passage 7 cells contained double the amount of sgag per 36

46 PCL scaffold and reduced in vitro cell culture DNA (4.8 ± 0.8 µg/µg DNA in ovine and 8.2 ± 1.4 µg/µg DNA in human cells) and five times the amount of collagen per DNA (1.1 ± 0.3 µg/µg DNA in ovine and 5.9 ± 2.5 µg/µg DNA in human cells) compared to ovine passage 7 cells. Tissues based on passage 7 cells in our experiments obtained values for sgag per DNA of 6.5 ± 0.2 µg/µg ovine DNA and 5.5 ± 0.3 µg/µg human DNA. Collagen per DNA was 3.2 ± 0.1 µg/µg DNA, and 3.7 ± 0.3 µg/µg DNA, in ovine and human respectively. This suggests that ovine cells in our study were more synthetic or less proliferative, which might be due to the amount of serum used in the culture medium. Van Geemen used 2.5% of lamb serum, while in this study 0.1% serum was used only, as an in vitro TEHV study by Dijkman demonstrated more homogeneous tissue formation throughout the wall and leaflets when 0.1% lamb serum was used [172]. A review by Mol et al described that the outcome of ovine TEHV was dramatically different from their human equivalents when using the same culture conditions, and lower amounts of serum resulted in tissue outcome comparable to human [173]. This shows the difficulties in the translation step from animal studies towards the clinic and vice versa. Furthermore, previous studies showed that not only interspecies, but also intraspecies variations of tissue properties are large [48, 171, 174]. Within this study we investigated the tissue properties of the strips seeded with cells of one sheep and one patient only. While it would be preferred to have more data on several human and ovine cell sources, we assume that within species the effects of e.g. cell passage number are comparable. Furthermore, the first goal of this study was to compare different types of scaffold to prevent compaction. This was investigated on cells of two species (human and ovine) and different cell passage numbers of those species. While two species and cell passage numbers were used and differences in terms of tissue production were observed between these species and cells passage numbers, the outcome of compaction was similar in all research groups. This indicates that the influence of the scaffold type is larger as compared to the influence of the tissue production of several cell sources, in terms of compaction. 2 A limitation of our study is that the ovine cells originated from a young, healthy sheep, while the human vascular derived cells were obtained from an older person that underwent bypass surgery. This might have influenced the outcome of the tissue properties as not only cell passage number, but also patient age may have an effect on the cell functioning, doubling time and ability of tissue production in different cell types [160, 175, 176]. In conclusion, differences in absolute values between ovine and human samples were seen within this experiment, although the general effects of reducing cell passage numbers and the use of PCL scaffold on compaction and the amount of tissue formation were comparable. 37

47 Chapter Conclusion This study showed that PCL scaffolds may serve as alternative scaffold material for human TEHV with minimal compaction and without compromising on tissue composition and properties, while further optimization of ovine TEHV based on PCL scaffold is needed to not only ensure reduced compaction but also strong tissues of a high quality. Cells from lower passages demonstrated to improve tissue formation, without influencing compaction and cell phenotype. In addition, reducing cell expansion will result in faster generation of TEHV, providing a more rapid treatment to patients. Acknowledgements This work was supported by a grant from the Dutch government to the Netherlands Institute for Regenerative Medicine (NIRM, grant No. FES0908). The authors wish to thank Tom Lavrijsen, Leonie Grootzwagers and Anita van de Loo for their help with the mechanical tests and culturing the TEHV. Furthermore Marc Simonet is acknowledged for the production of PCL scaffolds. The smoothelin antibody was kindly provided by Dr. GJ Van Eys, department of molecular genetics, cardiovascular research institute Maastricht, University Maastricht. 38

48 Superior tissue evolution in slowdegrading scaffolds for valvular tissue engineering 3 M. Brugmans R. Soekhradj-Soechit D. van Geemen M. Cox C. Bouten F. Baaijens A. Driessen-Mol Submitted 39

49 Chapter Abstract Synthetic polymers are widely used to fabricate porous scaffolds for the regeneration of cardiovascular tissues. To ensure mechanical integrity after implantation, a balance between the rate of scaffold resorption and tissue formation is of high importance. In vivo, a higher rate of tissue formation is expected in fast-resorbing materials compared to slow-resorbing materials, as a result of highly synthetic cells, which aim to compensate for the fast loss of mechanical integrity of the scaffold by deposition of newly formed collagen fibers. Here, we studied the effect of fast- (PGA-P4HB) and slow-degrading (PCL) synthetic scaffolds on tissue amount, composition, and mechanical characteristics in time in vitro, and compared these engineered values with values for native human heart valves. Electrospun porous PGA-P4HB and PCL scaffolds were either kept unseeded in culture or were seeded with human vascular-derived cells. Tissue formation, ECM composition, remaining scaffold weight, tissue to scaffold weight ratio, and mechanical properties were analyzed weekly up to 6 weeks. Unseeded PCL scaffolds remained stable in weight during the 6-week culture, while PGA-P4HB scaffolds degraded rapidly. When seeded with cells, both scaffold types demonstrated increasing amounts of tissue with time, which was more pronounced for PGA-P4HB-based tissues during the first two weeks due to highly synthetic cells, however PCL-based tissues resulted in the highest amount of tissue after 6 weeks. This study is the first to provide insight into the tissue to scaffold weight ratio, therewith allowing for a fair comparison between engineered tissues cultured on scaffolds with different degradation rates, as well as to native heart valve tissues. Although the absolute amount of ECM components differed between the engineered tissues, the ratio between ECM components was similar after 6 weeks. PCL-based tissues maintained their 3D shape during culture, while the deformed PGA-P4HB-based tissues showed appositional growth with culture time. After 6 weeks, PCL-based engineered tissues showed amounts of cells, collagen, and glycosaminoglycans that were comparable to human native heart valve leaflets, while engineered values were lower in the PGA-P4HBbased tissues. Although increasing in time, the amounts of collagen crosslinks were still below native values in all engineered tissues. In conclusion, this study indicates that slowdegrading scaffold materials are favored over fast-degrading materials in order to create organized ECM-rich tissues in vitro, which keep their 3D structure before implantation. 40

50 Superior tissue evolution in slow-degrading scaffolds 3.2 Introduction Bioresorbable synthetic polymers are used extensively in the field of cardiovascular tissue engineering to fabricate three-dimensional porous scaffolds, aiming for the regeneration of different types of tissues, such as heart valves and blood vessels [72, 177]. The classical tissue engineering paradigm to develop tissue replacements is the in vitro tissue engineering approach, where cells are seeded into synthetic bioresorbable porous scaffolds. During subsequent culture, tissue will be produced by the cells and after culture the construct can be implanted as a living autologous replacement. Alternatively, the engineered tissue can be decellularized after culture, to create allogenic off-the-shelf replacements that are rapidly repopulated to function as living replacement [43, 73, 75, 178]. As we focus on cardiovascular applications, we use human primary vascular-derived cells to grow tissue in fast-degrading PGA-P4H-based scaffolds, or slower-degrading PCL-based scaffolds. Both scaffold types have previously shown excellent results in terms of biocompatibility, processing ability, and cell infiltration [43, 47, 58, 75, 179, 180]. In vivo, scaffolds made of these materials will be fully resorbed by the body, ultimately resulting in a living implant that is able to adapt and remodel. Different research groups have studied the resorption of these scaffolds, mainly in vivo. It appeared that the site of implantation, presence of enzymes, molecular weight of the material, and scaffold porosity all affect resorption rates in vivo [181]. Reported complete resorption times of PGA vary from 1.5 months [68] to 4-6 months [182]. For electrospun PCL scaffolds, resorption takes much longer and this type of scaffold is reported to be completely resorbed in vivo after at least 2 years [183]. Obviously, scaffolds with slow- and fastresorption rates will contribute to the mechanical integrity of the tissue differently with time, both in vivo and in vitro. We hypothesize that cells in fast-resorbing scaffold materials will comprehend increased rates of tissue production, in order to compensate the loss of mechanical integrity by formation of collagen fibers, compared to cells in slowresorbing scaffold materials where mechanical integrity is maintained for a longer period of time. How tissue composition is changing during in vitro culture, and how this affects mechanical integrity due to different degradation properties of the scaffold materials, was never fully assessed. In an attempt to balance scaffold degradation, tissue stability, and mechanical integrity for in vitro cardiovascular tissue engineering, we determined the weight ratio between scaffold and tissue weekly, during a 6-week culture period, using both a slow- and a fast-degrading scaffold. In addition, we analyzed absolute and relative amounts of ECM and mechanical properties of the constructs with time. 3 Previously, attempts were undertaken to compare tissue composition of engineered constructs, cultured on different types of bioresorbable scaffolds, to native cardiovascular tissues [68, 80, 93, 184]. Results from conventional tissue composition assays express tissue composition relative to the dry weight of the sample. Often the weight of the 41

51 Chapter 3 remaining scaffold is integrated into the weight that is used to compare tissue composition of engineered tissues with native tissues [182, ]. This results in an overall overestimation of the actual formed tissue weight and, therewith, an underestimation of the amount of ECM components per mg formed tissue. To make a fair comparison between engineered and native tissues, the obtained data should be corrected for the remaining scaffold weight. During ageing, cardiovascular tissues are known to change in terms of ECM composition. We compared our engineered cardiovascular tissues to native valvular human data of several age groups as determined by van Geemen et al.[188], to assess similarity of the engineered tissues to native tissues. 3.3 Materials and methods Cell culture Human vascular-derived cells were harvested from segments of a Vena Saphena Magna from a patient that underwent bypass surgery, and was obtained according to the Dutch guidelines for secondary used materials. Cells were obtained using the outgrowth method and cultured using standard culture methods in a humidified atmosphere containing 5% CO2 at 37 C, as described previously [47]. Isolation and expansion medium consisted of advanced Dulbecco s Modified Eagle Medium (DMEM; Invitrogen, Breda, Netherlands), supplemented with 1% GlutaMax (Gibco), 1% Penicillin/Streptomycin (P/S, Lonza, Basel, Switzerland), and 10% Fetal Bovine Serum (Greiner Bio-one, Alphen a/d Rijn, Netherlands) Scaffold preparation and sterilization Rectangular strips (25x5 mm) were cut out of PGA meshes (PGA; specific gravity, 70 mg/cm3; Cellon, Bascharage, Luxemburg) and conventionally electrospun PCL meshes, both with a thickness of 1 mm and comparable fiber diameter. PGA scaffolds were additionally coated with poly-4-hydroxybutyrate (P4HB, received via a collaboration with Prof. Hoerstrup of the University Hospital Zurich) to provide structural integrity to the mesh. The outer 3-4 mm of both PGA and PCL scaffold strips were attached onto stainless steel rings (RVS Paleis, Geleen, Netherlands) using 15% polyurethane-tetrahydrofuran (PU, DSM, Geleen, Netherlands) glue, leaving an 18*5 mm area for cell seeding. The solvent was allowed to evaporate overnight in a vacuum oven. The rings with the scaffold strips were placed in 6-well plates and sterilization was achieved by immersion in an antibiotic/anti fungi solution, consisting of 10% Penicillin/Streptomycin (P/S; Lonza) and 50 µg/ml Fungin in sterile Phosphate Buffered Saline (PBS) (Sigma, Venlo, Netherlands) for 30 minutes on a shaker at 37 C. Subsequently, the antibiotics/anti fungi solution was removed and 70% ethanol was added for 15 minutes. The ethanol step was repeated and, thereafter, the strips were washed twice in PBS. To facilitate cell attachment, the scaffolds 42

52 Superior tissue evolution in slow-degrading scaffolds were incubated overnight with tissue engineering (TE) medium, consisting of expansion medium supplemented with 0.25 mg/ml L-ascorbic acid 2-phosphate (Sigma) Experimental design Scaffold strips of both materials (n=58) were either kept unseeded in culture (n=4 per week) or were seeded with cells (n=4-5 per week). Passage 7 cells were used and seeded onto both PGA-P4HB and PCL scaffolds, with a seeding density of 2.0 x 10 6 per cm 3 using fibrin as a cell carrier [161]. In short, cells were suspended in TE medium containing thrombin (10 U/ml, Sigma). This cell suspension was mixed with an equal volume of TE medium containing fibrinogen (10 mg/ml, Sigma) and dripped onto one side of the scaffolds. After seeding, the constructs were placed in an incubator at 37 C for 30 minutes, to allow polymerization of the fibrin gel. Thereafter, 6 ml of TE medium was added to each scaffold. The constructs were cultured for up to 6 weeks and TE medium was changed twice a week. After 1, 2, 3, 4, 5, and 6 weeks, seeded strips (n=4-5 per week) were sacrificed. One strip was used for histology and the remaining strips were used for mechanical testing followed by biochemical assays. At week 0, 1, 2, 3, 4, 5, and 6, unseeded strips (n=4 per week) were sacrificed. These strips were used for mechanical testing only Biochemical assays For the quantification of tissue formation during culture, engineered constructs were lyophilized after mechanical testing (n=3-4 per group) and used for biochemical assays. The total amount of DNA was determined as an indicator of cell number, the amount of hydroxyproline (hyp) as an indicator for collagen content, and the amount of sulfated glycosaminoglycans (sgag) was measured. Measurements were averaged per group. Lyophilized constructs were weighed and digested in papain solution (100 mm phosphate buffer (ph=6.5), 5 mm L-cysteine, 5 mm ethylene-di-amine-tetra-acetic acid (EDTA), and μg papain per ml, all from Sigma) at 50 C for 16 hours. To compare DNA, sgag and collagen within engineered constructs in time and with values found in native tissue, the weight of the engineered tissues without scaffold needs to be calculated. To obtain these values, the weight of the remaining scaffold of the unseeded strips of equal time points was subtracted from the weight of the seeded strips. The digest supernatant was collected and used for the DNA, sgag and collagen assays. The amount of DNA in the constructs was determined using the Hoechst dye method [162] and a standard curve prepared of calf thymus DNA (Sigma). As described before, sgag content was determined with a modification of the protocol described by Farndale et al.[47, 164]. Collagen content was determined by an assay as described by Huszar et al. [165], and a standard curve was prepared from trans-4-hydroxyproline (Sigma). The number of mature collagen hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP) crosslinks, as a measure of tissue maturity were measured in the digests of the constructs using high performance liquid 43

53 Chapter 3 chromatography as described previously [ ]. The number of HP and LP crosslinks were expressed per triple helix (TH), and the ratio of HP and LP crosslinks was determined Mechanical testing After 0 (only for unseeded group), 1, 2, 3, 4, 5, and 6 weeks of culture, the mechanical properties of the engineered constructs (n=3-4 per group) were assessed by uniaxial tensile tests in longitudinal direction of the constructs, using a BioTester 5000 (CellScale, Canada). The samples were stretched to 5, 10, and 15% strain for 5 times to precondition the samples. Mechanical test data was averaged per group. Sample thickness and width were measured with an electronic caliper. The Young s modulus was determined at a strain of 15% Histology To analyze tissue formation qualitatively, constructs were processed for histology (n=1 per group). Representative samples were fixed with 3.7% formaldehyde (Merck) and embedded in paraffin. Tissue sections of 10 μm were cut and studied by Masson Trichrome (MT) staining (MTC kit, Sigma, Venlo, Netherlands) for collagen deposition. The stainings were analyzed using light microscopy (Axio Observer, Zeiss, Germany) Statistical analyses Statistics were performed using GRAPHPAD Prism (version 5.04) and differences were considered significant for p-values <0.05. All data were presented as mean ± standard error of the mean (SEM). Regression analyses were performed to determine changes in tissue weight, scaffold weight, amount of ECM components, stiffness of the samples, and crosslinks over time. In case of a significant in- or decrease, the percentual increase or decrease was calculated using the predicted model equation. Also plateau and slope of the different curves were compared using regression analyses. One-way ANOVA, followed by a Tukey s multiple comparison post-hoc test, was used to compare TE composition with native tissues. 3.4 Results Scaffold to tissue ratio changes over time Dry weight of PCL scaffold material remained constant during culture time, while dry weight of PGA-P4HB scaffolds indicated mass loss starting after week 1, with a decrease of 93% compared to the initial values after 6 weeks of culture (p<0.05, Figure 3.1A). A contribution in weight due to tissue formation was observed in both scaffold types, as weight of tissues cultured in both PCL- and PGA-P4HB-based scaffolds increased during culture. When comparing the ratio between tissue weight and remaining scaffold weight, 44

54 Superior tissue evolution in slow-degrading scaffolds a percentual decrease in scaffold contribution and a percentual increase in tissue weight was observed in both the PGA-P4HB-based (Figure 3.1B) and PCL-based (Figure 3.1C) scaffold groups (p<0.05). After seeding, mainly scaffold weight was contributing to the total weight of the constructs, as no tissue was formed yet. Although tissue was formed within the PGA-P4HB constructs, a decrease in total weight was observed during culture, which was due to the fast degradation, and thus mass loss, of PGA-P4HB scaffolds. In the PGA-P4HB-based constructs a change was observed after roughly 2 weeks, as after this time point mainly tissue weight contributed to the total weight of the constructs. PGA- P4HB scaffolds were completely degraded after 6 weeks, with only tissue weight contributing to the total weight of the constructs. This change was not observed in the PCL-based constructs, as PCL scaffolds did not degrade as fast as PGA-P4HB scaffolds and primarily contributed to the total weight. 3 Figure 3.1 Dry weights of scaffolds and tissues during culture (A). Weights of the scaffolds and tissues are given as mean ± SEM. PCL scaffold remains stable during culture, while PGA-P4HB started to degrade after 1 week already. Newly formed tissue is contributing to the total weight of the strips. Ratio of the fastdegrading PGA-P4HB (B) and the slow-degrading PCL (C) scaffold to tissue during culture, given in percentages. Weight of tissue and scaffold are given as mean percentage, as a section of the total weight of the samples. Total weight of the whole samples are set at 100% Tissue evolution in slow- and fast-degrading scaffold Total amounts of DNA, sgag and collagen per construct increased during culture in both the PGA-P4HB (Figure 3.2A) and the PCL (Figure 3.2B) groups (all p<0.01). When comparing the sgag formation between the scaffold groups (Figure 3.2C), production rates were similar, however the total amount formed of the PGA-P4HB-based tissues were significantly lower compared to their PCL counterparts (p<0.01). Furthermore, although 45

55 Chapter 3 collagen production was increased in the PGA-P4HB groups compared to the PCL groups during the first two weeks of culture, the total amount formed of the PGA-P4HB group was lower compared to the PCL group after 6 weeks (p<0.01, Figure 3.2D). Total ECM values in the PCL-based constructs after 6 weeks of culture were higher compared to the PGA-P4HB-based constructs, which were 192±3 μg and 166±12 μg for the PCL and PGA- P4HB constructs, respectively. Although lower total amounts were observed for the PGA- P4HB-based tissues, cells in the PGA-P4HB-based tissues seemed to be more synthetic during the first two weeks of culture compared to cells in PCL-based scaffolds, with increased total amounts of ECM when corrected for the amount of DNA. Synthetic activity, in terms of sgag per DNA and collagen per DNA, was decreasing with time for cells in the PGA-P4HB scaffolds (p<0.01), while this was increasing with time for the cells in the PCL scaffolds (p<0.01) (Figure 3.2E). Figure 3.2 Combined results of DNA, sgag and collagen per strip during culture on PGA-P4HB (A) and PCL (B) scaffolds. During culture, the total amount of ECM increased, which was more pronounced for PCL-based tissues. PGA-P4HB-based constructs demonstrated lower plateau levels of amount of sgag (C) and collagen (D) compared to PCL-based constructs. sgag and collagen production per DNA (E) of PGA-P4HB-based tissues were increased during the first weeks, and became comparable to PCL-based tissues after 3 weeks. All results are given as mean ± SEM. 46

56 Superior tissue evolution in slow-degrading scaffolds Both the HP and LP crosslinks per triple helix increased during cultured time in tissues cultured on both the PCL and PGA-P4HB scaffolds (p<0.01, Figure 3.3A), with increased production rates of HP per triple helix compared to LP per triple helix in both scaffold groups (p<0.01). When compared between the scaffold groups, faster production of LP per triple helix was observed in the PGA-P4HB-based tissues (p<0.05) compared to the PCL-based tissues, while no difference in the production rates of HP per triple helix was found between the scaffold groups. This resulted in a higher HP/LP ratio in the PCL-based group (Figure 3.3B). 3 Figure 3.3 Collagen crosslinks in both scaffold groups given as HP/triple helix and LP/triple helix (A). Crosslinks within tissues grown on both type of scaffolds increased with culture time, while HP/triple helix increased with a higher rate compared to LP/triple helix. HP/LP ratio (B) was increased in PCL-based tissues compared to PGA-P4HB-based tissues Engineered tissues versus native heart valves After 6 weeks of culture, the ECM amount per mg tissue (Figure 3.4A) and ratio (Figure 3.4B) of engineered tissues were compared to human aortic heart valves. DNA per mg tissue was comparable between engineered tissue based on both scaffold types, and native values of different age groups. sgag per mg tissue is decreasing during ageing of human, while the amount of collagen per mg tissue is increasing (p<0.05). All engineered tissues resulted in sgag values comparable to native adolescent and adult values, while PGA-P4HB-based tissues demonstrated lower sgag values compared to native values in children (p<0.05). Collagen values of PCL-based tissues were not significant different from native values, while PGA-P4HB-based tissues resulted in lower values (p<0.05 compared to children, and p<0.001 compared to adolescents and adults). Although the amounts of ECM differed between the engineered tissue groups, their ECM ratios were comparable. These ratios were also similar to ratios found in children and adolescents. When compared to adult tissues, the percentual portion of collagen differed between adults and both engineered groups (p<0.05), while the percentage of sgag was only significantly different from the PCL-based tissues (p<0.05). Although the amount of newly formed collagen in PCL-based engineered tissues after 6 weeks is similar to values measured in native, the HP collagen crosslinks of both PCL and PGA-based tissues do not reach native values during culture (data not shown). After 6 weeks of culture, HP crosslinks of PCL- and PGA-P4HB-based tissues were 0.63±0.04 and 0.52±0.03 HP/triple helix, respectively, while values observed in children, adolescents and adults were 2.0±0.1, 2.0±0.03 and 2.6±0.1 47

57 Chapter 3 HP/triple helix, respectively. LP/triple helix also showed to increase during culture time and directed towards values measured in children (0.2±0.06 LP/triple helix) at the end of culture. HP/LP ratio of engineered tissues was similar with native values for adolescents and adults, however, differed significantly with the ratio found in children (p<0.001) (Figure 3.4C). During aging, the ratio drops rapidly from 14.4 in children, towards 5.7 and 4.8 in adolescents and adults, respectively, as a result from a fast increase in LP/triple helix in adolescents. Figure 3.4 Comparison between amount of ECM (A) and ECM ratio (B) per mg formed engineered tissue with native data. Results are given as mean ± SEM. Tissues based on PCL scaffolds showed comparable amounts of ECM compared to native human aortic valve values, while amounts found in PGA-P4HB-based tissues were lower compared to their native counterparts. ECM ratio was similar in all engineered tissues, and comparable to ratios found in children and adolescents, while they differed compared to the ratio observed in adults. The HP/LP ratio (C) of the engineered tissues was comparable to the ratio observed in aortic valves of adolescents and adults, while HP/LP ratio was lower compared to children. #, * and ^ represent significant differences of sgag, collagen and HP/LP ratio, respectively. Single or double symbols indicate p<0.05 and p<

58 Superior tissue evolution in slow-degrading scaffolds Mechanical characteristics of formed tissues based on fast- or slow-degrading scaffold Due to fast loss of mechanical integrity of the PGA-P4HB scaffold strips, mechanical tests on the unseeded PGA-P4HB samples over time could not be performed. This indicates that the observed mechanical properties in the seeded PGA-P4HB constructs are solely determined by the tissue only. Contribution of tissue formation to the mechanical properties was observed in both seeded PGA-P4HB and PCL samples, as samples became stiffer with culture time (p<0.01 for PCL and p<0.05 for PGA-P4HB, Figure 3.5), while the Young s modulus remained constant in the unseeded PCL scaffold strips. E-modulus at 15% strain [MPa] Culture time [weeks] PGA-P4HB seeded PCL unseeded PCL seeded 3 Figure 3.5 Young s modulus of seeded and unseeded scaffold strips during culture, given as mean ± SEM. Scaffold in PCL-based constructs is still contributing to the mechanical properties, while for the PGA-P4HBbased constructs mechanical properties are determined by tissue only. Newly formed tissue showed an additional effect on the Young s-modulus, as demonstrated with an increased stiffness in the seeded samples compared to the unseeded samples Histological visualization of engineered tissues in time Histology of the constructs revealed cellular tissues with dense surface layers, which was more pronounced in PGA-P4HB-based tissues. Masson Trichrome stainings showed collagen fibers throughout the strips of all groups during culture. Collagen is less homogeneously distributed in the PGA-P4HB-based strips (Figure 3.6A-F) compared to the PCL-based strips (Figure 3.6H-M). Furthermore, PCL-based tissues resulted in interstitial growth of tissue, while appositional growth was observed in the PGA-P4HB-based tissues, where a thick layer of tissue was formed around the scaffold. In addition, PGA-P4HB constructs showed compaction (decreased scaffold width) during culture, with significant differences compared to the original width (p<0.01), while the width of PCL constructs remained stable during culture (Figure 3.6G). 49

59 Chapter 3 Figure 3.6 Masson Trichrome staining of PGA-P4HB (week 1-6 representing by A-F) and PCL (week 1-6 representing by H-M) sections. The black scale bars represent 600 μm. Collagen is shown in blue, and red represents cytoplasm and muscle tissue. Vacuoles within the PCL sections are due to scaffolds remnants, which are dissolved during the dehydration step. PGA-P4HB sections do still show scaffold remnants (uncolored parts). Collagen is more homogeneously distributed in the PCL strips compared to the PGA-P4HB strips. Thickness of the strips (G) remains stable for PCL strips, while PGA-P4HB strips showed compaction. 3.5 Discussion A balance between the rate of scaffold degradation and tissue formation is crucial for maintaining mechanical integrity of the replaced tissues. We estimated the influence of slow- versus fast-degrading scaffolds on the amount and composition of engineered cardiovascular tissues, and mechanical integrity during culture. In addition, we compared these values of the engineered tissues to values found in native human heart valves leaflets In vitro evolution of tissue formation The unseeded PCL scaffold strips did not degrade in vitro, in terms of weight, while the unseeded PGA-P4HB scaffold started to loose mass already after 1 week. This resulted, 50

60 Superior tissue evolution in slow-degrading scaffolds together with the contribution in weight of the tissues, in different scaffold to tissue ratios during culture, with tissue weight being the main contributing factor in PGA-P4HB constructs, while for PCL constructs both tissue and scaffold weight contributed to the total weight. Our results on scaffold degradation are comparable with findings by Klouda [153], where a mass loss of 0.9% and 11% for PCL and PGA-P4HB scaffolds, respectively, was found after 15 days of static incubation. However, we observed a more severe mass loss of PGA-P4HB scaffold, as it decreased by 33% after 14 days. This might be due to the fact that in the study of Klouda samples were incubated with PBS, while our samples were incubated in culture medium containing FBS. As certain enzymes present in serum are known for degrading scaffolds [120, 123, 129, ], this might have led to accelerated degradation of the PGA-P4HB scaffold strips as compared to the study of Klouda. Although increased amounts of ECM components were shown in both scaffold groups with time, differences in tissue composition, when cultured using fast- or slow-degrading scaffolds were observed. A first observation was that cells in the PGA-P4HB-based tissues seemed to be highly synthetic as sgag and collagen per DNA were higher compared to PCL-based tissues. However, this was only observed during the first two weeks of culture, where after the cells became less synthetic. At the end of culture, higher total amounts of sgag and collagen for PCL-based tissues were observed. We hypothesize that this difference in tissue evolution is due to the fast degradation of the PGA-P4HB scaffolds, resulting in highly synthetic cells during the first weeks to compensate for the loss of mechanical integrity by newly formed collagen fibers. Compaction of the PGA-P4HB scaffolds resulted in a smaller surface area and less volume for the cells within these tissues to lay down their ECM, compared to PCL-based tissues, which might have resulted in a higher amount of ECM after 6 weeks of culture in the latter. 3 It is well described that degradation of PGA scaffolds with or without P4HB can alter the ph of the environment, due to their acid degradation products [181, ]. A low ph possibly affects the viability, proliferation or tissue synthesis of the cells. Higgins et al. showed that the amount of porcine smooth muscle cells decreased and cells dedifferentiated, due to PGA degradation products [195]. However, in their study media was collected after 7 days only, while within our study media was changed twice a week to prevent building up of degradation products and thus an acidic environment. We, therefore, assume that released degradation products into the culture medium did not have a profound effect on the viability, proliferation, and tissue synthesis within our experiments. All tissues demonstrated a continuous increase in LP and HP crosslinks during culture. However, the ratio between these crosslink types differed between tissues. The HP/LP ratio was lower for PGA-P4HB-based tissues compared to PCL-based tissues. Wassen et al.[198] described that a lower HP/LP ratio caused by a relative high amount of LP/triple helix, as observed in our PGA-P4HB samples, is seen in mineralized tissues only. This might 51

61 Chapter 3 assume that tissues cultured in PGA-P4HB scaffolds are more prone to mineralization compared to their PCL counterparts. They also hypothesize that mineralization of collagen fibrils is promoted by specific orientation of the molecules within these fibrils, which might be different between the PCL and PGA-P4HB-based tissues due to a potential higher degree of tissue remodeling of the PGA-P4HB-based tissues as a result of faster degradation of PGA-P4HB scaffolds Comparison between engineered and native tissues In literature, different methods are described to compare ECM components and amounts of tissue, between engineered tissues, or to their native counterparts. These include a non-invasive monitoring system to correlate biomarkers present in culture medium with the synthesized tissue [187] and a method where ECM components are expressed as mg per cm 3 tissue [80]. However, these are suboptimal methods as the first one does not include the total amount of tissue formed, and, in the second method, remaining scaffold can contribute to the dimensions and, therefore, possibly influences the outcome, especially when the scaffold is not degraded yet. To allow for accurate insight into tissue evolution during culture and a fair comparison between engineered and native tissues, only tissue weight without the contribution of remaining scaffold should be used. Our study is the first that provides these insights as we corrected for the presence of remaining scaffold. This correction is of importance when comparing properties of tissues that were cultured using scaffolds with different degradation rates and when comparing engineered tissues that were grown on slow-degrading scaffolds, which is still (partly) present, with native tissues. A limitation of this method is that we do not account for the effect of cells and tissue on scaffold degradation. The presence of cells can result in accelerated degradation, as cells might release enzymes that stimulate this degradation. Furthermore, in vivo macrophages will migrate to the scaffold materials and start to degrade the materials, which is not the case in our in vitro set-up. Despite this limitation, this new method comes nearest to the actual values, compared to all other studies performed until today. Figure 3.7 provides an overview of the generated amount of ECM and mechanical properties, of PCL and PGA-P4HB-based tissues after 6 weeks of culture, compared to values of pulmonary valves of children, which is the first target of tissue engineered valves. PCL-based tissues show ECM values which are most similar to native values found in children, while PGA-P4HB-based tissues showed a somewhat lower amount of ECM. Similar stiffness values were observed in both PCL and PGA-P4HB-based tissues, while the values of the latter is only determined by the newly formed tissue and not by remaining scaffold, as observed for PCL. Stiffness of engineered samples are higher compared to native values, however, we do not expect difficulties in opening or closing of the leaflets after implantation, as PGA-P4HB valves with a similar stiffness were successfully implanted before [68]. 52

62 Superior tissue evolution in slow-degrading scaffolds Percentage DNA GAG Collagen Young's-modulus DNA GAG Collagen Young's-modulus PGA-P4HB PCL Pulmonary child Figure 3.7 Comparison of native values of a child s pulmonary valves with engineered tissues after 6 weeks of both PCL and PGA-P4HB-based tissues. Values of pulmonary valves of children are set at 100% (horizontal dashed line). Values of PCL and PGA-P4HB-based tissues are given as percentage compared to native values of children. PCL shows values that are similar or towards to native values in terms of ECM, while PGA-P4HB shows lower values. Engineered tissues are stiffer compared to their native counterparts. Although PGA- P4HB scaffold does not influence the mechanical properties of the tissue after 6 weeks, stiffness is similar to PCL-based tissues which are still partly influenced by remaining scaffold (marked area in the bar) Conclusion In conclusion, tissues based on slow-degrading materials, which maintained weight and mechanical integrity during culture, demonstrated tissues which preserved their 3D shape. Tissues based on fast-degrading material, which quickly demonstrated mass loss and loss of mechanical integrity, resulted in compaction during culture and different tissue to scaffold ratios. Although cells in PGA-P4HB constructs produced tissue at a higher rate during the first weeks of culture compared to cells in PCL constructs, the amount of tissue after 6 weeks was higher in the latter. ECM ratios were comparable between the scaffold groups and also between engineered and native human values. This study demonstrates the importance of using slow-degrading scaffolds in order to create constructs with stable mechanical integrity, which maintain their configuration upon implantation. Further longterm research is needed to investigate properties of PCL-based tissues when this scaffold material is completely degraded. 53

63 Chapter 3 Acknowledgements This work was supported by a grant from the Dutch government to the Netherlands Institute for Regenerative Medicine (NIRM, grant No. FES0908). This research also forms part of the Project P1.01 ivalve of the research program of the BioMedical Materials institute, co-funded by the Dutch Ministry of Economic Affairs. The financial contribution of the Nederlandse Hartstichting is gratefully acknowledged. The authors gratefully thank Marc Simonet for electrospinning of the PCL scaffolds. 54

64 Hydrolytic and oxidative degradation of electrospun supramolecular biomaterials: 4 In vitro degradation pathways M. Brugmans S. Sontjens M. Cox A. Nandakumar A. Bosman T. Mes H. Janssen C. Bouten F. Baaijens A. Driessen-Mol Submitted 55

65 Chapter Abstract The emerging field of in situ tissue engineering of load bearing tissues places high demands on the scaffolds, as these scaffolds should provide mechanical stability immediately upon implantation. A new class of synthetic biomaterials are the supramolecular polymers, which contain non-covalent interactions between the polymer chains, and can form complex 3D structures by self assembly. Here, we aimed to map the degradation characteristics of promising (supramolecular) materials, as well as their susceptibility to degradation. The selected biomaterials were all PCL, either unmodified or with supramolecular (either 2-ureido-[1H]-pyrimidin-4-one or bis-urea units) hydrogen bonding moieties incorporated into the backbone. As these materials contain elastomeric properties, they are suitable for cardiovascular applications. Electrospun scaffold strips of these materials were incubated with solutions containing enzymes that catalyze hydrolysis, or solutions containing oxidative species. At several time points, chemical, morphological, and mechanical properties were investigated. It was demonstrated that conventional and supramolecular PCL-based polymers respond differently to enzymeaccelerated hydrolytic or oxidative degradation, depending on the morphological and chemical composition of the material. Conventional PCL is more prone to hydrolytic enzymatic degradation as compared to the supramolecular materials, while the opposite was shown when degraded by an oxidative pathway. Given this knowledge regarding degradation characteristics of different (supramolecular) materials, we are able to tailor degradation characteristics by combining different PCL backbones with additional supramolecular moieties. This toolbox can be employed to screen, limit, and select biomaterials for pre-clinical in vivo studies targeted to different clinical applications. 56

66 In vitro degradation pathways 4.2 Introduction Tissue engineering aims to restore tissue structure and function of diseased or damaged tissues by implantation of specifically designed bioresorbable materials, with or without the addition of cells [ ]. Conventional tissue engineering aims to collect autologous cells from patients, which are utilized for the in vitro generation of new tissues, and are often cultured in bioreactors for several weeks before implantation. A new and promising approach is in situ tissue engineering, in which in vitro culture is omitted and the patient s body is used as a bioreactor [86, ]. New tissue will be regenerated directly in the body by host cells after implantation of, for example, a bioresorbable electrospun polymeric scaffold. This makes the overall procedure less demanding in terms of costs, time, and regulatory challenges, and creates off-the-shelf availability. In situ tissue engineering of load-bearing tissues places high demands on the bioresorbable scaffolds, as these scaffolds should be able to provide mechanical stability immediately upon implantation, and for a prolonged period thereafter, until sufficient mature neo-tissue is formed by recruited cells to take over the mechanical function of the scaffold. Various synthetic bioresorbable polymers are used for tissue engineering applications, and these polymers include aliphatic polyesters (e.g. polylactic acid (PLA), PGA and PCL), as well as various polyurethanes [129, 181, 205, 206]. A new set of synthetic materials are the supramolecular polymers, which are formed by arrays of directed, noncovalent interactions between the building blocks, and can form complex 3D-structures by self assembly [112]. Material properties such as mechanics and resorption rate, which are critical for the success of in situ tissue engineering can be modified by combining or changing ratios of the same building blocks. This potentially allows for a variety of polymers with varying properties to be synthesized in a relatively short time span, thereby accelerating the development process. Monomeric units of the supramolecular polymers possess a relatively low molecular weight, resulting in beneficial processing properties, e.g. easy dissolution in organic solvents. Furthermore, supramolecular polymers may show self-healing properties [113, 207, 208], can easily be made bioactive [119, 209], and allow for a more controlled way of synthesis, which can result in complex molecular structures [112]. Because of these features, these materials pose excellent candidates for use in in situ tissue engineering. Particularly, we are interested in biomaterials that either have 2- ureido-[1h]-pyrimidin-4-one (UPy) [ ] or bis-urea (BU) [119] motifs incorporated into their molecular structure, as these contain elastomeric properties, which makes them suitable for cardiovascular applications. 4 To enable the formation of a completely autologous tissue, the scaffold should degrade at the right pace during neo-tissue formation, leaving behind a living implant that is able to remodel and grow. In vivo, degradation of implanted scaffold materials can be accomplished via different pathways that operate at the same time, and that even may affect each other [120, 121, 124, 128]. A well-known pathway is hydrolytic degradation, 57

67 Chapter 4 where chemical bonds of the polymer chains are cleaved by reaction with water molecules, forming oligomers and ultimately generating small molecules that can be cleared from the body [120, 121]. Previous studies have reported that several enzymes, like proteases and esterases, which are present in human serum or are expressed by macrophages and other activated cells that are in contact with the scaffold, are known to catalytically accelerate this process [120, , 193]. Another well-described pathway is oxidative degradation, which is mediated by ROS that are secreted by macrophages, neutrophils and giant cells that are in contact with the scaffold [120, 126]. Previous studies have investigated that oxidation of polymers is often initiated by abstraction of a hydrogen atom by radicals, resulting in chain scission and/or crosslinking of the polymer [210, 211]. Mapping the degradation characteristics of promising (supramolecular) materials for use in in situ tissue engineering approaches, as well as their susceptibility for certain degradation pathways, paves the way for screening and selection of materials for various clinical implantation sites. The degradation properties of widely used and well-known materials such as polyesters, polycarbonates and polyurethanes have been examined extensively, both in vitro and in vivo [125, 180, 206, ]. In general, results of these studies show that polymers containing ester or anhydride linkages react with water molecules and undergo hydrolysis [58, 81, 121, 180]. The water molecules can access those chemical species more easily, and thus increase the hydrolytic activity, when the polymer is amorphous or contains aliphatic structures [121, 217]. Other polymers, including polyethers and polyurethanes, were found to be more susceptible to the oxidative pathway, as these materials contain α- methylene groups adjacent to ether or urethane groups, which are more prone to the formation of carbon centered radicals by abstraction of a hydrogen atom [121, 127, 128, 131, 211, 218]. Just a few studies reported on the degradation characteristics of various polymers (PCL, polycarbonates, or polyurethane) modified with UPy or BU units. These were performed by incubating the materials in phosphate buffer saline (PBS) or solutions of various lipases at 37 C [117, 209, 219, 220]. These studies showed that the rates of enzymatic degradation can span a wide range, from less than 1% degradation after 1 month [220] to 90% after only 15 days [209], depending on the types of lipase and polymers used. No hydrolytic degradation, in terms of weight loss, of the UPy containing materials was observed for 126 days when samples were incubated with PBS [117], and a decrease in weight of only 2% after 120 days was observed for BU-containing materials [219]. Although these studies gave some insight into the degradation properties of bioresorbable materials, the major part of these studies were performed on films or disks which are quite dense, while degradation rate of electrospun scaffolds, that are more porous and have higher surface to volume ratio, can be different. Studying the degradation properties of electrospun meshes is, from a clinical point of view, more relevant as these are more likely to be implanted as a tissue replacement, rather than a compact, solid construct. 58

68 In vitro degradation pathways Furthermore, most research has focused on a single degradation pathway, while it is of importance to assess either the enzyme-accelerated hydrolytic and the oxidative degradation pathways, since in vivo both pathways may be operative and consequently, both may affect the implanted scaffold. Here, an in vitro study was designed to investigate both degradation pathways in an accelerated fashion and was used to assess the degradation of several promising supramolecular biomaterials for in situ tissue engineering. We have chosen three previously reported supramolecular biomaterials, in which PCL backbones are combined with either UPy hydrogen bonding groups (materials PCL2000-UPy and PCL800-UPy) [221] or BU hydrogen bonding groups (PCL2000-BU) [119]. High molecular weight PCL, a material frequently used for tissue engineering scaffolds, was added as a benchmark. All materials were electrospun and the resulting scaffold meshes were either exposed to enzymes that catalyze hydrolysis or to oxidative conditions. Degradation was monitored over time by examining the remaining scaffold with respect to weight, molecular weight, fiber diameter, and mechanical properties. Statistical analyses were performed to analyze changes in properties over time of all polymers with the various treatments, as well as to investigate their susceptibility to degradation and its mechanism (surface or bulk erosion). 4.3 Materials and methods Materials 4 All reagents, chemicals, materials, and organic solvents were obtained from commercial sources and were used without further purification, unless otherwise noted. The PCLbased supramolecular biomaterials PCL2000-UPy, PCL800-UPy and PCL2000-BU were synthesized as previously described from PCL diol building blocks of molecular weights 800 or 2000 [119, 221]. These PCL2000-diol and PCL800-diol building blocks are prepared by initiation from diethylene glycol, so they contain one ether bond in their structure. Conventional PCL (Purasorb PC 12, IV=1.24 dl/g) was purchased at Purac Biochem, Gorinchem, the Netherlands. Thermal characterization of these materials was performed by differential scanning calorimetry (DSC) on a Perkin Elmer Pyris 1 or on a TA Instruments Q2000. Reported data are from the melt, so after the sample has been in the isotropic state, and were determined in the second heating run at a heating rate of 10 C/min. The glass transition temperature (Tg) is reported as the inflection point, while the melting transition (Tm) is reported as the peak of the transition Scaffold preparation Scaffolds were fabricated in a climate-controlled electrospinning cabinet (IME Technologies, Geldrop, The Netherlands) using the conventional electrospinning method as described before [222]. Rectangular strips (25 (l) x 5 (w) x 0.44 (t) mm) were punched 59

69 Chapter 4 out of the electrospun scaffold meshes. Initial weight (W0) and thickness of all individual strips were measured using a digital balance (Mettler Toledo, XS105, Greifensee, Switzerland)) and a digital thickness gauge (Mitutoyo, SGM, Groningen, The Netherlands). Prior to incubation for degradation, the meshes were centrifuged at 4500 rpm in purified water for 5 minutes to remove air bubbles Accelerated in vitro degradation Strips (n=60 per material) were incubated at 37 C in 1.5 ml enzyme solution, referred to as enzymatic degradation, or in a 4 ml oxidative degradation solution each. The enzyme solution consisted of 100 U/mL lipase from Thermomyces lanuginosus (L0777, Sigma- Aldrich) in PBS or 10 U/mL cholesterol esterase from bovine pancreas (C-3766, Sigma- Aldrich) in PBS. These enzymes, which are present in serum and are secreted by activated macrophages, are known to cleave ester and urethane bonds to a higher extent as compared to other secreted enzymes [215, 223, 224]. The oxidative solution comprised of 20% hydrogen peroxide (Sigma-Aldrich) and 0.1 M cobalt(ii) chloride (Sigma-Aldrich) in purified water (ph of this solution is 4.5). Hydrogen peroxide and cobalt(ii) chloride undergo a Haber-Weiss reaction, creating reactive hydroxyl radicals [211]. Incubation times of the scaffolds in lipase, cholesterol esterase, or oxidative solutions were up to 56, 96 and 400 hours, respectively. Based on literature [131, 225], solutions were changed every 3-4 days to maintain enzymatic activity and a constant concentration of radicals Scaffold characterization Analyses of the (remaining) scaffolds were performed at 5 time points for the enzymatic groups and 7 time points for the oxidative group (n=4 per group per time point). Mass loss, molecular weight, fiber diameter, and mechanical properties were determined Mass loss Scaffold strips were removed from the degradation solution, washed three times with purified water, dried under vacuum at 37 C for 16 hours and weighed (Mettler Toledo, XS105, Greifensee, Switzerland), to assess weight loss due to scaffold degradation. Mass loss of the scaffolds (n=4 per group per time point) was determined using the equation: W1/W0 100%, where W0 is the initial scaffold weight and W1 indicates remaining scaffold weight Scanning electron microscopy (SEM) Scaffold fiber morphology and average fiber diameters were assessed and determined by scanning electron microscope (SEM), (Phenomworld, Eindhoven, The Netherlands) of one sample per group per time point. Average fiber diameters were determined by 20 60

70 In vitro degradation pathways individual measurements performed on four SEM images per scaffold strip, using Phenomworld software (Fibermetric, Phenom pro suite version 2.0) Mechanical properties To study the effect of degradation on the mechanical properties of the scaffolds, uniaxial tensile tests in longitudinal direction of the strips (n=3 per group per time point) were performed. Due to a loss of mechanical integrity over time, associated with degradation, it was not possible to perform tensile tests on all PCL-BU and PCL-UPy strips of the latest oxidation time points. Sample thickness and width were measured with an electronic caliper. Stress-strain curves were obtained (Mecmesin multitest-i) at an elongation rate of 100% per minute and the mechanical test data was averaged per group per time point. The elasticity modulus (Young s-modulus) was determined as the slope of the initial linear part of the curve, as a measure for stiffness. As a measure for strength, the ultimate tensile strength (UTS) was defined as the peak stress value, while strain at break is a measure for the maximal elongation of the samples until break Molecular weight (GPC) After tensile testing, one strip per group per time point of each material was taken and dissolved in dimethylformamide ((DMF), Sigma) in order to determine the mass averaged molecular weight (Mw) of the samples by gel permeation chromatography (GPC) analysis. GPC was performed on a Varian/Polymer Laboratories PL-GPC 50, using DMF with 10 mmol/l lithium bromide as eluent and maintaining the temperature of the equipment at 50 C. The relative or apparent molecular weights (Mw) were determined with respect to polyethylene glycol standards. Samples were measured in duplicate and the Mw was averaged from this duplicate measurement Statistical analyses All data are presented as mean ± standard deviation. Statistics were performed using GRAPHPAD Prism (version 5) and differences were considered significant for p-values < Changes over time Regression analyses were performed to determine changes in weight, Mw, fiber diameter, Young s-modulus, UTS, and strain at break over time. Both a one-phase decay model (assuming the rate at which changes occur is proportional to the amount that is left) and a linear model (assuming a constant rate) were used to fit the data. In case of a significant increase or decrease with p<0.05 or p<0.01, the percentual in- or decrease was calculated from the predicted model equation and classified as non-relevant (0-10%), small (10-61

71 Chapter 4 25%), moderate (25-100% for an increase and 25-50% for a decrease), and severe (>100% for an increase and % for a decrease) Susceptibility to degradation and its mechanism The susceptibility for both enzymatic and oxidative degradation was determined via correlation analyses of all measured parameters. Significant correlations were classified as weak (p<0.05), average (p<0.01), and strong (p<0.001). Susceptibility for degradation was calculated as the number of significant correlations (with more weight to the average and strong correlations as compared to the weak correlations) divided by the maximum number of possible correlations and expressed as a percentage. Susceptibility was classified as not susceptible (<20%), susceptible (20-60%), or highly susceptible (>60%). To obtain insight into the mechanism of degradation, correlations were either attributed to surface erosion or to bulk erosion. Correlations that were considered to attribute to surface erosion were correlations between mass loss and fiber diameter, between mechanical properties, between mechanical properties and fiber diameter, and between mass loss and mechanical properties. Correlations that were considered to attribute to bulk erosion were correlations between Mw and mass loss, mechanical properties, or fiber diameter and inverse correlations between parameters. The susceptibility to either enzymatic or oxidative degradation was subsequently determined as described above with similar classifications for susceptibility. 4.4 Results Material properties The studied supramolecular biomaterials PCL2000-UPy, PCL800-UPy and PCL2000-BU are in fact thermoplastic elastomers with PCL soft blocks and hard blocks composed of interacting and phase separated hydrogen bonding units (Figure 4.1). Figure 4.1 Schematic overview of the materials examined in this degradation study, with bis-urea (BU) (A), and ureidopyrimidinone (UPy) (B) based supramolecular biomaterials. 62

72 In vitro degradation pathways PCL is a semi-crystalline polyester (Tg = 64 C, Tm = 52 C), while the PCL2000-BU thermoplastic elastomer shows a first melting transition (Tm1) of the semi-crystalline PCL soft block at a lower temperature and a second melting transition (Tm2) of the BU hard block at a higher temperature (Tg = 54 C, Tm1 = 27 C, Tm2 = 98 C) [119]. Both PCL800 and PCL2000 UPy are also thermoplastic elastomers (PCL800-UPy: Tg = -39 C, Tm1 = 65 C, Tm2 = 91 C; PCL2000-UPy: Tg = -58 C, Tm1 = 53 C, Tm2 = 116 C) In vitro degradation as monitored by scaffold mass loss and Mw Enzymatic degradation (Figures 4.2A-D) of conventional PCL scaffolds resulted in moderate (44%, p<0.01) to severe (92%, p<0.01) mass loss by lipase and cholesterol esterase treatment, respectively, while Mw remained constant over time. For the supramolecular materials, only the PCL2000-BU was affected by enzymatic degradation with moderate weight loss by both lipase (30%, p<0.01) and cholesterol esterase (22%, p<0.01) treatment (Figures 2A,C). Mw of PCL2000-BU did not change with lipase treatment, while a small decrease in Mw (14%, p<0.05) was observed during cholesterol esterase treatment. The PCL-UPy materials did not show changes in weight and Mw over time due to enzymatic degradation. Oxidative degradation (Figures 4.2E,F) did not affect mass and Mw of conventional PCL scaffolds, while all supramolecular materials were affected. Both PCL-UPy materials showed moderate mass loss (42% and 27%, p<0.01 for PCL800-UPy and PCL2000-UPy, respectively) and a severe reduction in Mw (71% and 83%, p<0.01 for PCL800-UPy and PCL2000-UPy, respectively). The PCL2000-BU also demonstrated moderate mass loss (35%, p<0.01) and a severe reduction in Mw (94%, p<0.01) due to oxidative degradation. 4 63

73 Chapter 4 Figure 4.2 Influence of enzymatic (A-D) and oxidative degradation (E,F) on mass loss (A,C,E) and Mw (B,D,F) of conventional and supramolecular PCL-based scaffold strips. Significant and relevant changes over time are indicated by lines between data points. Conventional PCL was mainly affected by enzymatic degradation with moderate to severe mass loss, but with stable molecular weight. The supramolecular materials were mostly affected by oxidative degradation, with mass loss as well as decreases in molecular weight In vitro degradation as monitored by scaffold fiber diameter and morphology Enzymatic degradation (Figures 4.3A,B) of conventional PCL scaffolds resulted in small to severe fiber diameter reduction, depending on the enzyme used (18% and 62%, p<0.01 for cholesterol esterase and lipase treatment, respectively). Enzymatic degradation did not affect the fiber diameter of both PCL-UPy materials, but resulted in a moderate reduction in fiber diameter in PCL2000-BU scaffolds after both lipase (31%, p<0.05) and cholesterol esterase (25%, p<0.01) treatment. 64

74 In vitro degradation pathways Oxidative degradation (Figures 4.3C) did not affect the fiber diameter of the conventional PCL scaffolds, but resulted in moderate reduction in fiber diameter for the PCL-UPy materials (45% and 49%, p<0.01 for PCL800-UPy and PCL2000-UPy, respectively). PCL2000-BU showed a small reduction in fiber diameter after oxidative treatment (10%, p<0.01). Figure 4.3 Influence of enzymatic (A,B) and oxidative degradation (C) on the fiber diameter of conventional and supramolecular PCL-based scaffold strips. Significant and relevant changes over time are indicated by lines between data points. The fiber diameter of conventional PCL scaffolds was affected only by enzymatic degradation, while the supramolecular materials showed mainly reduced fiber diameters with oxidative degradation. 4 SEM images of scaffold strips before and after enzymatic and oxidative degradation treatment confirmed these changes in fiber diameter (Figure 4.4). They further demonstrate that the surface of the conventional PCL fibers is clearly affected by degradation, while the fiber surface of the supramolecular materials seemed less affected as compared to the conventional PCL, though more fragmented fibers were observed in the supramolecular scaffold groups. 65

75 Chapter 4 Figure 4.4 SEM images with different magnifications of PCL-based scaffold strips before (A-D) and after enzymatic (E-L) and oxidative (M-P) degradation. Conventional PCL is mainly affected by enzymatic degradation, resulting in thinner and clearly affected fibers, while the supramolecular materials were mainly affected by oxidative degradation with thinner fibers. The fiber surface of the supramolecular materials seemed less affected as compared to the conventional PCL, though more fragmented fibers were observed. The white dots on the conventional PCL scaffold after oxidative degradation are presumably cobalt chloride remnants. White scale bars represent 20 micrometer Changes in mechanical properties during in vitro degradation Enzymatic degradation (Figure 4.5A, C, E) resulted in overall weakening of conventional PCL scaffolds with severe reductions in Young s modulus (96%, p<0.01 and 57%, p<0.05 for lipase and cholesterol esterase, respectively), UTS (96%, p<0.05 and 51%, p<0.01 for lipase and cholesterol esterase, respectively), and strain at break (80% and 66%, p<0.05 for lipase and cholesterol esterase, respectively). The PCL-UPy materials did not demonstrate weakening, but changed mechanical properties with small to moderate increases in modulus, depending on the PCL soft segment length, for both lipase (13% and 44%, p<0.01 for PCL2000-UPy and PCL800-UPy, respectively) and cholesterol esterase treatment (19% and 99%, p<0.05 for PCL2000-UPy and PCL800-UPy, respectively). PCL2000- UPy further showed a moderate reduction in strain at break with lipase treatment (27%, p<0.05), indicating a change towards a more brittle material. PCL2000-BU showed a combination of weakening and a change toward a more brittle material with moderate 66

76 In vitro degradation pathways reductions in UTS (40%, p<0.01) and strain at break (39%, p<0.01) by cholesterol esterase treatment, and a severe increase in modulus (56%, p<0.01) after lipase treatment. Oxidative treatment (Figures 4.5B, D, F) only affected strain at break of conventional PCL scaffolds with a moderate decrease (25%, p<0.05), while modulus and UTS remained unaffected. The PCL-UPy materials showed a combination of weakening and a change toward a more brittle material with severe reductions in UTS (96%, p<0.05 and 87%, p<0.01 for PCL800-UPy and PCL2000-UPy, respectively) and strain at break (100% and 96%, p<0.01 for PCL800-UPy and PCL2000-UPy, respectively), and a severe increase in Young s modulus (>300%, p<0.01) after lipase treatment. Similar weakening and changes toward a more brittle material were observed for PCL2000-BU with severe reductions in UTS (80%, p<0.05) and strain at break (99%, p<0.05), accompanied by a severe increase in Young s modulus (>1000%, p<0.01) after oxidative treatment. 4 Figure 4.5 Influence of enzymatic (A,C,E) and oxidative (B,D,F) degradation on the Young s modulus (A,B), UTS (C,D), and strain at break (E,F) of PCL-based scaffold strips. The results for cholesterol esterase treatment are not shown, but are comparable to those of the lipase treatment. Significant and relevant changes over time are indicated by lines between data points. The mechanical properties of conventional PCL were mainly affected by enzymatic degradation and represented by overall weakening. The mechanical properties of the supramolecular materials were affected by enzymatic degradation, but to a larger extent by oxidative degradation. Here, a change to a more brittle material was evident, accompanied by an overall weakening of the material. 67

77 Chapter Susceptibility to degradation and its mechanisms Correlation analyses revealed that the conventional PCL scaffolds were susceptible to enzymatic degradation, with the degree of susceptibility depending on the enzyme used. Susceptibility was higher for lipase (62%) as compared to cholesterol esterase (36%) and surface erosion seemed the dominant degradation mechanism (77% and 33% for lipase and cholesterol esterase treatment, respectively). Conventional PCL scaffolds were not susceptible (13%) to oxidative degradation. The PCL-UPy materials were not susceptible to enzymatic degradation, neither lipase (2% for both PCL800-UPy and PCL2000-UPy) nor cholesterol esterase (7% for both PCL800-UPy and PCL2000-UPy). The susceptibility for oxidative degradation was dependent on the PCL soft segment length, with no susceptibility for PCL800-UPy (13%) to susceptible for PCL2000-UPy (40%). Both surface erosion (23%) and bulk erosion (33%) seemed involved. PCL2000-BU was susceptible to enzymatic degradation, though only to cholesterol esterase (16% and 31% for lipase and cholesterol esterase, respectively), and oxidative degradation (24%). Surface erosion seemed the dominant mechanism in degradation of PCL2000-BU (33% and 23% for enzymatic and oxidative degradation, respectively). 4.5 Discussion Electrospun bioresorbable scaffold meshes represent promising candidates for use in in situ tissue engineering to replace diseased or damaged tissue parts. While providing initial mechanical stability upon implantation, host cells are recruited over time for neo-tissue formation, taking over the mechanical function of the scaffold. Supramolecular polymers represent interesting candidates to replace soft and elastic dynamically-loaded tissues. To allow for the development of a stable fully autologous implant that can grow and remodel in the patient, the scaffold should degrade in pace with neo-tissue formation. Here, we aimed to map the degradation characteristics of promising (supramolecular) materials, as well as their susceptibility to certain degradation pathways, for use in in situ tissue engineering approaches. An in vitro test was designed to investigate the degradation of electrospun biomaterial scaffolds either by enzyme-accelerated hydrolysis or by oxidation. In addition to changes in fiber morphology of the meshes, changes in mass of the scaffold, and changes in molecular weight, this in vitro study also monitored and assessed changes in the mechanical properties of electrospun scaffolds over time. The investigated scaffolds were prepared from PCL-based supramolecular biomaterials and conventional PCL served as a benchmark. Figure 4.6 provides a summary of the results obtained in this study indicating the changes over time with both enzymatic and oxidative degradation as well as their susceptibility for each polymer group. 68

78 In vitro degradation pathways Figure 4.6 Schematic summary of the obtained results indicating the changes over time in mass, Mw, fiber diameter, and mechanical properties by either enzymatic or oxidative degradation over time for conventional and supramolecular PCL-based scaffolds. Further, the susceptibility of each polymer group to enzymatic as well as oxidative degradation is represented by a color scale, with red indicating a high susceptibility and green referring to the material being not susceptible to degradation. Illustrations by Anthal Smits 4 The conventional PCL scaffolds were rapidly degraded by enzymatic hydrolysis, using lipase or cholesterol esterase, as evidenced by mass loss, changes in fiber morphology, and overall weakening, while molecular weight remained unaffected. These results are consistent with findings by others, although different types and concentrations of the enzymes resulted in slower or faster degradation of the PCL [125, 193, ]. Polymer degradation by enzymes can be either surface erosion or bulk degradation, depending on the accessibility of the interior of the polymer to the enzyme. Surface erosion was identified here as the dominant degradation mechanism with clear effects to the fiber surface, thus apparently, the ability of the enzymes to infiltrate the hydrophobic semicrystalline PCL is limited (or the activity of the enzyme becomes compromised upon infiltration) [120, 130]. In contrast to the conventional PCL meshes, the supramolecular UPy- and BU-based PCL demonstrated to be less prone to hydrolyze enzymatically with no or minimal changes in mass, molecular weight and fiber diameter. The PCL-UPy materials were classified as not 69

79 Chapter 4 susceptible to enzymatic degradation, though they demonstrated an increased Young s modulus, accompanied by a reduction in strain at break, which indicates a change to a more brittle material. This change may be caused by annealing of the material at 37 C, resulting in a material with an increased crystallinity of the PCL phase, and thus a more brittle material. The PCL-BU was classified as susceptible to cholesterol esterase and not to lipase, though also with a change to a more brittle material. Clearly, the introduction of the BU or UPy hard blocks in the PCL backbone has a marked stabilizing effect on the enzymatic degradation rate, despite increasing the overall polarity of the biomaterial by introduction of the polar BU or UPy groups. Presumably, the different morphology of the materials as compared to conventional PCL is causing the changes in hydrolytic enzymatic degradation behavior. For PCL-BU, it is known that phase separation of the PCL soft block and the BU hard block is on the nanometer scale (ca. 10 nm scale) [119], implying that the partly amorphous PCL soft block in PCL-BU may be less accessible as compared to the more sizable amorphous PCL phases in conventional semi-crystalline PCL. Moreover, the molecular dynamics of the segmented PCL chains may be compromised, first as these chains are relatively short, and second as they are kept into position at both ends by the immobile UPy or BU hard blocks. According to the above factors, we propose that the PCL chains in the supramolecular biomaterials are less accessible to enzymes, and therefore causing the lowered enzymatic degradation susceptibility. Among the supramolecular materials, the PCL-BU was more susceptible to enzymatic degradation as compared to PCL-UPy, though similar PCL soft segment length were used in the backbone of PCL2000- UPy and PCL2000-BU. Apparently, the ester bonds in the BU-based material are more accessible and/or prone to hydrolysis as compared to those in the UPy containing material. Both materials have phase separated soft and hard blocks and the exact manner in which this phase separation takes place may influence and determine their degradation. However, the exact differences in morphology, e.g. level of phase separation, mobility of the PCL chains, and the level of crystallinity of the PCL soft phase, between PCL-BU and PCL-UPy are not known and should be further investigated. Oxidative degradation gave the opposite result as that observed for the enzymeaccelerated degradation. Conventional PCL scaffolds were not susceptible to oxidative conditions, with stable mass, molecular weight, fiber diameter, Young s modulus, UTS, and only a small decreased strain at break. The presence of merely an aqueous solution without enzymes was clearly not enough to hydrolyze conventional PCL. Conventional PCL only has ester groups in its structure, and apparently these ester groups are not significantly degraded by the offered oxidative cobalt (II) solution, despite the fact that the amorphous phase in semi-crystalline PCL must be accessible to the presented small oxidative cobalt (II) and H2O2 (derived) species. The supramolecular materials on the other hand did show susceptibility to oxidative degradation, with decreases in mass, molecular weight, fiber diameter and a combination of weakening of the materials with a change to a more brittle material, with some fragmentation of fibers. We primarily attribute the 70

80 In vitro degradation pathways augmented oxidative degradation of the supramolecular PCL-based biomaterials to their chemical differences with PCL (and not to morphological features), whereby these differences are represented by the presence of ureidopyrimidone for PCL-UPy and the presence of urea groups for PCL-BU. Moreover, all supramolecular PCL are based on PCLdiols initiated from diethylene glycol, hence they comprise a single ether group in every PCL soft block, which also might result in an increased sensitivity to oxidative degradation. Remarkably, the PCL soft segment length influenced the susceptibility to oxidative degradation, with the PCL800 soft block providing more resistance to oxidative degradation. Oxidative degradation of the supramolecular materials was classified as surface erosion, though for the PCL-UPy bulk erosion was also noted, indicating diffusion of the small oxidative cobalt and H2O2 (derived) species into these materials. It has to be noted that degradation is a dynamic process, as the mechanical, morphological and chemical properties of the polymers change during degradation, and all can affect surface and bulk mobility, accessibility by enzymes, and the diffusion of small molecules such as water, oxidative species and degradation products. Here, we have investigated degradation by enzymatic hydrolysis and oxidation separately. Future studies should include a combination of degradation pathways to assess their combined effects. Macrophages play an essential role in the degradation of polymeric scaffold meshes in in situ tissue engineering as an inflammatory response provides the basis for neo-tissue formation. These macrophages secrete both enzymes as well as oxidative species, therewith triggering both degradation pathways. We are currently investigating whether macrophage phenotype, claimed essential in tissue outcome [230], is correlated to trigger degradation in either of the pathways, enabling assessment of the desired degradation characteristics of electrospun bioresorbable meshes. 4 Depending on the application, either fast or slow resorption by the body is desired. When scaffold resorption is too slow it can result in stress shielding of the growing tissue, thereby impeding the regeneration process [231] or leading to undesirable outcomes. When the resorption process is too fast, the mechanical integrity of the implant is not sufficient, as the neo-tissue is not sufficiently developed yet to bear the full mechanical force required [232], leading to failure of the implant. Furthermore, the site of implantation might influence the resorption rate of a biomaterial. Mechanical forces, like compression, fatigue and shear stress, or external factors like ph might affect the resorption rate of the implanted material [121]. This again demonstrates the need to tailor the properties of bioresorbable polymers specifically to the intended application. 71

81 Chapter Conclusion In this study, we demonstrated that conventional and supramolecular PCL-based polymers respond differently to in vitro enzyme-accelerated hydrolytic or oxidative degradation pathways, depending on the morphological and chemical composition of the material. Conventional PCL is more prone to hydrolytic enzymatic degradation as compared to the supramolecular materials, while the opposite is shown for these materials when degraded by an oxidative pathway. Given this knowledge on degradation characteristics of different (supramolecular) materials, we are able to tailor degradation characteristics by combining different PCL backbones with additional supramolecular moieties. This toolbox can be employed to screen, limit and select biomaterials that are going to be used for pre-clinical in vivo studies for different clinical applications. Acknowledgements This research forms part of the ivalve project of the research program of the BioMedical Materials institute, co-funded by the Dutch Ministry of Economic Affairs, Agriculture and Innovation. The financial contribution of the Nederlandse Hartstichting is gratefully acknowledged. Part of the work by Marieke Brugmans was supported by a grant from the Dutch government to the Netherlands Institute for Regenerative Medicine (NIRM, grant No. FES0908). The authors would like to thank Roel Lalieu and Nanayaa Bates from Xeltis for producing the scaffold sheets. Furthermore, Leonie Grootzwagers and Anita van de Loo from Xeltis are acknowledged for performing the tensile tests and taking the SEM images. 72

82 Advanced electrospun scaffold degradation by inflammatory macrophages in comparison with healing macrophages 5 M. Brugmans M. Cox C. Bouten F. Baaijens A. Driessen-Mol In preparation 73

83 Chapter Abstract Implantation of a synthetic scaffold made from a bioresorbable material will cause an inflammatory response in which macrophages are claimed to play an essential role. Macrophages show a continuum of functional properties alternating between the proinflammatory (M1) and the tissue-healing (M2) phenotypes. The contribution of macrophage phenotype to biomaterial resorption remains unclear and should be further elucidated to provide more insight into the immune response to implanted biomaterials, which is of particular relevance for in-situ tissue engineering approaches. In this study, 2D and 3D in-vitro cultures were used to investigate the contribution of macrophage polarization to degradation of electrospun biodegradable scaffolds. In addition, we monitored the phenotypical change of unpolarized macrophages during time as an indication of the initial macrophage response to the electrospun meshes. Monocytes of the human cell line THP-1 were differentiated towards macrophages, seeded into 6-wells plates (2D) or onto rectangular electrospun PCL strips (3D), and polarized towards inflammatory (by LPS/IFNɣ) or healing macrophages (by IL4/IL13), or kept unpolarized. In 2D cultures, sample groups were sacrificed after 1, 3, 6, 8, and 10 days and cells were counted. Furthermore, cell phenotype was assessed from cell morphology via imaging before sacrifice. The 3D samples were sacrificed after 2 days, 1 week, and 4 weeks. Samples were assessed with respect to DNA content, microstructure (SEM, with and without cells), esterase activity, and gene expression (qpcr). Different cell morphologies were observed between the polarized groups, whereas DNA amount decreased with time for all phenotypes in both 2D and 3D cultures, albeit more prominent in the LPS/IFNɣ polarized cells. Unpolarized cells demonstrated similar gene expression levels compared to the healing phenotype. Scaffold degradation was observed in all phenotype groups, but was most pronounced by the LPS/IFNɣ polarized cells. These findings were confirmed by esterase activity and gene expression analysis. In conclusion, macrophage phenotype affects the rate of electrospun scaffold degradation, with inflammatory macrophages accelerating degradation. 74

84 Contribution of macrophage phenotype to scaffold degradation 5.2 Introduction The use of scaffold materials composed of bioresorbable synthetic polymers is a promising approach to replace diseased tissues in patients, as these biomaterials are supposed to be resorbed while new tissue is formed simultaneously. Implantation of a biomaterial evokes an inflammatory response, in which macrophages play an essential role. These cells provide the basis for neo-tissue formation [230], as well as degradation and removal of the implanted polymeric scaffolds [126, 126, 233]. Macrophages show a continuum of functional properties alternating, dependent on micro-environmental factors, between the pro-inflammatory (M1) and the tissue-healing (M2) phenotypes [136, 234]. M1 macrophages are driven by pro-inflammatory signals, such as interferon gamma (IFN-ɣ) and lipopolysaccharide (LPS) and secrete pro-inflammatory cytokines and ROS, while M2 macrophages are driven by interleukins (IL), such as IL4/IL13, and are involved in wound healing and anti-inflammatory processes in favor of ECM formation [139, 235]. Although several studies focused on the suppression of the inflammatory response to improve the biocompatibility of an implanted material [ ], this inflammatory response, in particular the balance between the phenotype of macrophages, is believed to play an important role in the final outcome of tissue regeneration (balance towards M2 phenotype) or chronic inflammation and scar formation (balance towards M1 phenotype) [135, 136, 140, 235, 240]. In a functional healing process, this balance is desired to be mainly a M2 phenotype during the regenerative phase of wound healing. Single macrophages are able to phagocytose small foreign body particles (<10 um) [138], while larger particles ( um) are beyond the phagocytic capacity of a single macrophage and are phagocytized by fused macrophages, which form multinucleated foreign body giant cells (FBGC). In case materials with sizes in the millimeter range are implanted even the FBGC are not capable to engulf these large bulk materials and, therefore, undergo frustrated phagocytosis, whereby ROS and enzymes are released in an attempt to degrade the scaffold material. In a previous study, we used in vitro degradation models to demonstrate the effect of enzymes or ROS products on different (supramolecular) PCL-based scaffold materials (Brugmans et.al, 2015, submitted). It was demonstrated that both enzymes and/or ROS products are able to degrade scaffold materials, depending on the type of biomaterial. Pro-inflammatory M1 macrophages are known to secrete both enzymes and ROS products [126, 139, 233], suggesting that M1 macrophages play an important role in scaffold degradation in vivo. Different studies have shown that the interleukins IL-4 and IL-13 result in formation of FBGC [126, 241], and are also known to polarize macrophages to the anti-inflammatory M2 phenotype [242]. FBGC formation by IL-4 and IL-13 suggests that M2 macrophages are also involved in scaffold degradation. Furthermore, differences in gene expression between M1 and M2 polarized macrophages were studied. Results showed that both the M1 and M2 phenotypes secrete enzymes, such as lipase A cholesterol esterase, which are known to be able to degrade 5 75

85 Chapter 5 scaffold materials that contain ester bonds [223, 243]. Taken together, the contribution of macrophage phenotype in scaffold degradation remains unclear and should be further elucidated to provide more insight into the desired immune response to implanted biomaterials toward functional healing. Here, we investigated if and how macrophage phenotype, claimed essential in tissue outcome, contributed to the degradation of electrospun bioresorbable meshes. For this purpose, we made use of in vitro models which included cultures of macrophages, polarized towards the inflammatory or healing phenotype on two-dimensional substrates, or on electrospun scaffolds (3D). In addition, unpolarized macrophage were cultured to investigate the contribution of the scaffold meshes on the phenotypical change of these macrophages, as an indication for the initial response of macrophages to the scaffold meshes. 5.3 Materials and methods Scaffold preparation and sterilization Conventional PCL (Purasorb PC 12, IV=1.24 dl/g) was purchased at Purac Biochem, Gorinchem, the Netherlands. Scaffolds were fabricated in a climate-controlled electrospinning cabinet (IME Technology, Eindhoven, The Netherlands) using the conventional electrospinning method, as described before [222]. Initial fiber morphology was determined by scanning electron microscopy (SEM) (Phenomworld, Eindhoven, The Netherlands). Rectangular strips (12.5 (l) x 5 (w) x 0.30 (t) mm) were punched out of electrospun scaffold meshes. Strips were sterilized by gamma irradiation before use (Synergy health, Ede, The Netherlands) Cell culture and seeding Cell culture The human monocytic cell line THP-1 was purchased from Cell Lines Service (CLS, Eppelheim, Germany) and cultured according the suppliers recommendation. The cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium with L-Glutamine and 25 mm HEPES (Invitrogen, Breda, The Netherlands), supplemented with 10% Fetal Bovine Serum ((FBS), Greiner Bio one, Frickenhausen, Germany), 1% Penicillin/Streptomycin (P/S, Lonza, Basel, Switzerland), and 0.05mM 2-mercaptoethanol (Sigma Aldrich) in a humidified atmosphere containing 5% CO2 at 37 C. Cell densities were maintained between *10 6 cells per ml medium. Medium was changed 3 times per week. 76

86 Contribution of macrophage phenotype to scaffold degradation Seeding of monocytes and transformation into macrophages The experimental design of our study is shown in Figure 5.1. Scaffold strips (n=77) were placed in 50 ml tubes containing 10 ml sterile PBS (Fisher, Landsmeer, The Netherlands). These tubes were centrifuged for 10 minutes at 3500 rpm to increase hydrophilicity of the scaffolds. Upon seeding, the scaffolds (n=72) were placed into 2 ml vials containing 1.5 ml culture medium, 3.0*10 6 THP-1 monocytes, and 50 ng/ml of phorbol 12-myristate 13- acetate ((PMA), Sigma Aldrich) to transform the monocytes into macrophages. Scaffold strips that were kept unseeded (n=5) were also placed in 2 ml vials containing the same medium, but without cells. The vials were rotated for 16 hours in a humidified atmosphere containing 5% CO2 at 37 C to allow cells adhere to the scaffold strips. After seeding, each scaffold strip was placed into a well of a 12-wells plates containing 1.5 ml culture medium and 50 ng/ml PMA for another 24 hours. In 2D cultures, 1.25*10 6 cells were plated into 6- wells together with 2 ml culture medium and 2 µl PMA for 48 hours. Figure 5.1 Experimental set-up of the 2D and 3D cultures. Monocytes were transformed into adhering macrophages by PMA. To obtain different cell phenotypes, different cytokines were added to the cell culture medium to allow for macrophage polarization, or cells were kept unpolarized. At day 10 and 20, scaffold strips were replenished to maintain a constant cell population on the fibers. 2D samples were sacrificed after 1, 3, 6, 8, and 10 days, and the 3D samples were sacrificed after 2, 7 and 28 days Polarization and culture of macrophages Cells in 2D cultures and cells on seeded scaffold strips were polarized into the inflammatory or healing type of macrophages referred to as M1 or M2, respectively, or kept unpolarized, referred to as M0 (n=24 per macrophage type). M1 polarization medium consisted of 100 ng/ml Lipopolysaccharide S ((LPS), Sigma Aldrich) and 20 ng/ml Interferon gamma ((IFN-ɣ, Peprotech, London, UK) in culture medium. M2 polarization medium consisted of 20 ng/ml of Interleukin-4 (IL-4) and IL-13 (Peprotech, London, UK) in culture medium. M0 cells were cultured using culture medium without additional cytokines. Polarization and medium change was performed 3 times per week. Medium was stored at -80 C until further use. 2D cell cultures were kept in culture for 1, 3, 6, 8, and 10 days (n=2 per macrophage phenotype per time point), and photos of the cell morphology were taken before sacrifice, using the moticam 2500 camera (Motic AE2000 Trino, Wetzlar, Germany). Seeded scaffold strips were cultured for 2 days, 1 week and 4 weeks, (n=8 per macrophage phenotype per time point). Unseeded scaffold strips were 5 77

87 Chapter 5 kept in culture for 2 days (n=1), 1 week (n=2) and 4 weeks (n=2). To maintain cells on the scaffolds throughout the culture period and as macrophages are known to migrate out of the scaffold or to undergo apoptosis, the samples of the 4 week culture groups were replenished with cells. After 10 and 20 days, 1.0*10 6 and 1.5*10 6 cells, respectively, were added to these scaffold strips, using the rotating seeding method, as described above. After culture, the strips for SEM analysis (n=2 per phenotype per time point) were fixated in formalin for 24 hours, which was replaced with sterile PBS afterwards and stored at 4 C until analysis. Samples for DNA (n=3 per phenotype per time point) and gene expression analysis (n=3 per phenotype per time point) were washed in sterile PBS and stored at - 80 C until further use Medium analyses An esterase assay was performed to determine the esterase activity in culture medium samples of 3D cultures (n=3 per time point and macrophage type) as a measure for the amount of secreted enzymes by the macrophages. Esterase activity can be measured with the use of p-nitrophenyl butyrate (pnpb, Sigma-Aldrich) as a substrate. When incubated with esterase, the substrate is hydrolyzed and the yellow p-nitrophenyl (pnp) is released, which can be measured by UV absorbance at 405 nm. 150 µl of medium sample was added to a well of a 96-well plate, together with 10 µl pnpb solution (3.55µl pnpb in 5.0 ml acetonitrile (Sigma-Aldrich)) and 140 µl Tris-Buffer. The plate was incubated at 37 C for 30 minutes, at 405 nm, using a microplate reader (Multiskan GO, Thermo Scientific) and readings were performed every 5 minutes. Cholesterol esterase (C3766, Sigma-Aldrich) was used as a positive control, and all samples were measured in duplicate. One unit of esterase activity was defined as 1 nmol pnp released from pnpb per minute at 37 C DNA assay The total amount of DNA was determined as an indicator of cell number during culture. This was used to normalize the outcome of the esterase activity per amount of cells. Scaffold strips were lyophilized after culture. Lyophilized tissue samples were digested in 300 µl papain solution (100 mm phosphate buffer (ph=6.5), 5 mm L-cysteine, 5 mm ethylene-di-amine-tetra-acetic acid (EDTA), and 140 μg papain per ml, all from Sigma) at 50 C for 18 hours. After centrifuging the samples, the digest supernatant was collected and used for the DNA assay. The amount of DNA in the samples was determined using the Hoechst dye method [162] and a standard curve prepared of calf thymus DNA (Sigma Aldrich). Using the assumption that all cells contain 6.5 pg of DNA [163], the amount of cells per scaffold was calculated. In 3D cultures, DNA amounts were calculated per mg dry weight scaffold. For 2D cultures, cell number was measured using the Tali Image-based cytometer (Invitrogen, Breda, The Netherlands). Measurements were averaged per group. 78

88 Contribution of macrophage phenotype to scaffold degradation Morphology of scaffold fibers Scaffold fibers were visualized by scanning electron microscope (SEM) to analyze the morphology of the fibers. Formalin-fixed samples were stored in sterile PBS until use. Samples were dehydrated in a graded ethanol series, starting from 50% to 100% in 5 to 20% increments. The ethanol was then allowed to evaporate, and samples were visualized by SEM (Phenomworld, Eindhoven, The Netherlands). After SEM analysis, samples were treated with 4.6% natrium hypochlorite for 15 minutes at room temperature, to remove all cells. Samples were washed twice in water and the same samples were analyzed by SEM again, in order to visualize parts of the scaffold fibers that were covered by the cells during the first SEM analysis Gene expression analysis The cultured scaffolds, stored at -80 C, were disrupted with a microdismembrator (Sartorius, Goettingen, Germany) 3 times for 30 seconds at 3000 rpm, using RNA-se free metal beads (diameter 3 mm, Sartorius, Goettingen, Germany) in Nalgene cryovials (Sigma-Aldrich) to homogenize the samples. Cells of all samples were lysed using RLT buffer with β-mercaptoethanol (Sigma-Aldrich). RNA was isolated with Qiagen RNeasy kit (Qiagen, Venlo, The Netherlands) according to the manufacturer s protocol, including a DNA-se incubation step. RNA quantity and purity were determined with a spectrophotometer (NanoDrop, ND-1000, Isogen Life Science, IJsselstein, The Netherlands). Subsequently, cdna was synthesized starting from 200 ng RNA in a 25 µl reaction volume consisting of 20 ng/ml random primers (Promega, Madison, WI), 5 mm dntps, 5x first strand buffer, 0.1M DTT, 200U/µl M-MLV Reverse Transcriptase (RT) (Invitrogen, Breda, The Netherlands) and double autoclaved water (ddh2o). cdna synthesis was performed in a Thermal Cycler (C1000 Touch, Bio-Rad) by subjecting the samples to a temperature cycle of 72 C for 6 minutes, 37 C for 5 minutes (with subsequent addition of M-MLV), 37 C for 60 minutes, and 95 C for 5 minutes. Absence of genomic contamination was checked using PCR (icycler iq, Bio-Rad, Hercules, US) with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers. 5 Gene expression levels of genes expressed by inflammatory-type macrophages (TNF-α, CCR7, MCP-1 and IL-23) and healing-type macrophages (TGF-β1, VEGFA, MMP9, IL-10, CD163, and MRC-1) were measured, and GAPDH was selected as a reference gene. The primer sequences are listed in Table 1. Gene expression levels were determined by adding 20mM primer mix to the cdna templates, together with SYBR Green Supermix (Bio-Rad) and ddh2o. All samples were analyzed in duplicates. The real-time PCR reaction was carried out for 3 minutes at 95 C, 40x (20 seconds at 95 C, 20 seconds at 60 C, 30 seconds at 72 C), 1 minute at 65 C, followed by a melting curve analysis (icycler MyiQ, Bio-Rad, Hercules, US). Ct values were normalized to the GAPDH reference gene and the expression levels were calculated using the formula 2 ΔCt. 79

89 Chapter 5 Table 1 List of primer sequences used in gene expression analysis. Primer Symbol Accession number Tumor necrosis factor alfa TNF-α NM_ Interleukin 23 IL-23 NM_ Chemokine (C-C motif) receptor 7 CCR7 NM_ Monocyte chemoattractant protein 1 MCP-1 NM_ CD163 molecule CD163 NM_ Transforming growth factor, β1 TGF-β1 NM_ Vascular endothelial growth factor A VEGFA NM_ Interleukin 10 IL-10 NM_ Matrix metalloproteinase 9 MMP-9 NM_ Mannose receptor, C type 1 MRC-1 NM_ Primer Sequence ( 5-3) FW: GAGGCCAAGCCCTGGTATG RV: CGGGCCGATTGATCTCAGC FW: AGCTTCATGCCTCCCTACTG RV: CTGCTGAGTCTCCCAGTGGT FW: AAGCCTGGTTCCTCCCTATC RV: ATGGTCTTGAGCCTCTTGAAATA FW: CAGCCAGATGCAATCAATGCC RV: TGGAATCCTGAACCCACTTCT FW: CACTATGAAGAAGCCAAAATTACCT RV: AGAGAGAAGTCCGAATCACAGA FW: GCAACAATTCCTGGCGATACCTC RV: AGTTCTTCTCCGTGGAGCTGAAG FW: GCAGAATCATCACGAAGTGG RV: GCATGGTGATGTTGGACTCC FW: GACTTTAAGGGTTACCTGGGTTG RV: TCACATGCGCCTTGATGTCTG FW: TGGGGGGCAACTCGGC RV: GGAATGATCTAAGCCCAG FW: TGGGTTCCTCTCTGGTTTCC RV: CAACATTTCTGAACAATCCTATCCA Statistical analyses Statistics were performed using GRAPHPAD Prism (version 5.04) and differences were considered significant for p-values <0.05. All data are presented as mean ± standard error of the mean. Regression analyses were performed to determine changes in amount of DNA over time. In case of a significant in- or decrease with p<0.05 or p<0.01, the increase or decrease was calculated from the predicted model equation and expressed as percentage. In 3D cultures, we assumed a one-sided population, as only a decrease in amount of DNA was expected. Statistical differences in qpcr data were analyzed with oneway ANOVA followed by a Tukey s multiple comparison post-hoc test. The LPS/IFNɣ polarized samples after 1 week and 4 weeks of culture were not included in statistical analysis of gene expression, due to the limited amount of RNA extracted from these samples. 80

90 Contribution of macrophage phenotype to scaffold degradation 5.4 Results Morphology and number of polarized macrophages in 2D cultures Representative pictures of THP-1 monocytes in suspension and the adherent macrophages, which were cultured in 5.2D and either left unpolarized or polarized with LPS/IFNɣ or IL4/IL13, are shown in Figure 5.2A. Cell diameter increased when monocytes adhered to the culture flasks. When cells were polarized with LPS/IFNɣ they became elongated, while polarization with IL4/IL13 showed rounded cells with a larger cell diameter compared to the unpolarized cells. DNA amount in the macrophage cultures decreased with time for all phenotypes (Figure 5.2B), however more prominent in the LPS/IFNɣ polarized cells, as a higher plateau level for IL4/IL13 polarized cells compared to LPS/IFNɣ polarized cells was found (p<0.001). After 10 days of culture, only 11% and 38% (both p<0.01) of the initial amount of the cells polarized with LPS/IFNɣ or IL4/IL13, respectively, remained in the culture wells. 5 Figure 5.2 Representative pictures of THP-1 monocytes and activated macrophages (A). Macrophages were kept unpolarized, or polarized towards an M1 or M2 phenotype by addition of the LPS/IFNɣ cytokines, or IL4/IL13, respectively. Unpolarized cells demonstrated a rounded morphology. Bigger rounded cells were observed when polarized with IL4/IL13, while cells became mainly elongated by addition of LPS/IFNɣ. The amount of DNA was reduced in all groups, however more prominent in the LPS/IFNɣ treated cells (B). 81

91 Chapter Number of polarized macrophages on PCL scaffolds Due to replenishment of the scaffold strips after 10 and 20 days, the amount of DNA during culture in the unpolarized and IL4/IL13 polarized samples remained constant (Figure 5.3). However, decreased amounts of DNA were observed of macrophages on PCL scaffolds when polarized with LPS/IFNɣ to 0.5% of the initial levels after 4 weeks (p<0.05). Figure 5.3 The amount of DNA during culture on PCL scaffold fibers. As a result on re-seeding, the amount of cells on the fibers remained stable in the unpolarized and IL4/IL13 polarized groups. A significant decrease in amount of cells was observed in LPS/IFNɣ polarized cells. Stars indicate time points of macrophage replenishment Morphology of polarized macrophages on PCL scaffolds Two days after seeding, large populations of cells were observed on the scaffold, in each group (Figure 5.4A, D, G). LPS/IFNɣ polarized cells appeared to be more spiky and showed a rougher surface compared to the unpolarized and IL4/IL13 polarized cells, which showed a rounded morphology. After 1 week, many LPS/IFNɣ polarized cells lost viability as we observed many cell remnants on the scaffold fibers. Furthermore, the remaining cells were smaller (Figure 5.4E) compared to the unpolarized (Figure 5.4B) and IL4/IL13 (Figure 5.4H) polarized macrophages. Both elongated and rounded cells were observed on the scaffold fibers of the unpolarized and IL4/IL13 polarized groups. After 4 weeks, less cells were found in the LPS/IFNɣ polarized samples, while many cells, single or in groups were found in the unpolarized and IL4/IL13 polarized sample groups (Figure 5.4C, F, I). 82

92 Contribution of macrophage phenotype to scaffold degradation Figure 5.4 Representative SEM images of cells cultured on scaffold meshes for 2 days, 1 week, and 4 weeks, which were unpolarized (A-C), polarized with LPS/IFNɣ (D-F), or polarized with IL4/IL13 (G-I). Decreasing amounts of cells were detected in time. No clear differences in cell morphology could be observed between the unpolarized and IL4/IL13 polarized cells, while the LPS/IFNɣ polarized cells showed to have a spiky appearance (D) with shrinkage of these cells after 1 week (E). White scale bar represents 50 µm Degradation of PCL scaffolds by polarized macrophages Figures 5.5A-I show representative SEM images of scaffold fibers after removal of the cells. After 2 days, unpolarized cells did not appear to degrade the scaffold fibers, as no visual damage was observed in SEM images. Scaffold fibers of the IL4/IL13 polarized samples showed minor damage in few spots, while this was more pronounced for the LPS/IFNɣ polarized samples. Scaffolds containing the LPS/IFNɣ polarized cells showed the highest degree of degradation of the scaffold fibers after 1 week of culture. Surface erosion of the fibers was clearly visible at various spots in the scaffold strips, while this was observed at only a few spots in the unpolarized and IL4/IL13 polarized groups. At 4 weeks, scaffolds showed similar results compared to the scaffolds after 1 week of culture. 83

93 Chapter 5 Up to 1 week of culture, no damage in terms of broken fibers, surface erosion or cracks in the fibers were observed in the scaffold cultured without cells, while after 4 weeks, minor surface erosion could be observed at some places in the scaffolds (Figure 5.5 J-L). Figure 5.5 Representative SEM images of scaffold fibers after removal of cells. Photos were taken after 2 days, 1 week, and 4 weeks of culture with unpolarized cells (A-C), cells that were polarized with LPS/IFNɣ (D-F), polarized with IL4/IL13 (G-I), or scaffolds that were kept in culture without the presence of cells (J-L). All groups showed scaffold degradation, although this was most pronounced in the LPS/IFNɣ polarized group. When no cells were added to the scaffold, only minor damage was observed, which was due to hydrolytic degradation. White particles are natrium hypochlorite residues. White scale bar represents 30 µm. 84

94 Contribution of macrophage phenotype to scaffold degradation Medium analysis of 2D and 3D cultures of polarized macrophages In order to determine the levels of secreted enzymes by the different macrophage phenotypes, esterase activity assays were performed on culture medium samples of 3D cultures. All phenotypes showed to be able to release esterases. When normalized to DNA amount (Figure 5.6), enzyme activity in LPS/IFNɣ polarized cells was higher compared to unpolarized and IL4/IL13 polarized samples when cells were cultured in 3D for 1 week (P<0.05) and 4 weeks (p<0.001). Furthermore, esterase activity per DNA increased after 4 weeks of culture compared to 1 week in the LPS/IFNɣ polarized groups (p<0.001). Figure 5.6 Esterase activity of the cells, corrected for the amount of DNA present in the scaffold meshes during culture. LPS/IFNɣ treated cells resulted to have increased esterase activity per cell compared to the other groups Gene expression analysis in 3D cultures Expression of macrophage phenotypic and immune response genes was analyzed to investigate whether a different set of markers was expressed when cells were cultured in PCL scaffolds in unpolarized state or polarized towards the M1 or M2 phenotype (Figure 5.7). TNF-α levels were comparable between the groups after 2 days. After 1 week, the expression levels of the unpolarized and IL4/IL13 polarized groups decreased compared to the LPS/IFNɣ polarized group at 2 days (p<0.05). CCR7 and MCP-1 were significant increased in the LPS/IFNɣ polarized groups after 2 days, compared to the unpolarized and IL4/IL13 polarized groups (p<0.001). In time, CCR7 levels decreased in both the unpolarized and IL4/IL13 polarized groups, with p<0.05 after 1 week for both groups, and p<0.05 after 4 weeks in the unpolarized group. Expression levels of CD163 after 1 and 4 weeks increased in the unpolarized samples compared to the LPS/IFNɣ polarized samples (p<0.01 and p<0.05, respectively). Furthermore, CD163 expression levels of the unpolarized cells were increased compared to the levels found in the IL4/IL13 polarized cells, after 1 week (p<0.05). MMP9 expression increased in time for the unpolarized group, with higher levels found after 4 weeks, compared to 2 days and 1 week (p<0.05). At this time point, MMP9 levels were also higher in comparison with the LPS/IFNɣ 5 85

95 Chapter 5 polarized group (p<0.001) and the IL4/IL13 polarized samples after 4 weeks (p<0.01). Apart from expression levels of CD163 and MMP9 no other differences in expression levels were observed between the unpolarized and IL4/IL13 polarized samples. No differences in expression levels of IL-23, TGF-β, VEGF A, IL-10, and MRC-1 were observed between the groups (data not shown). Figure 5.7 Gene expression analysis of cells cultured on PCL scaffolds for 2 days (all groups), 1 and 4 weeks (unpolarized and IL4/IL13 polarized cells only). CCR7 and MCP-1 levels were increased in LPS/IFNɣ polarized cells compared to other groups. Besides minor differences between expression levels of CD163 and MMP9 after 1 and 4 weeks, respectively, no differences in expression levels were observed between the unpolarized and IL4/IL13 polarized samples. No differences in expression levels of IL-23, TGF-β, VEGF A, IL- 10, and MRC-1 were observed between the groups (data not shown). # denotes significant differences compared to 1 week within the same phenotype with p<0.05, ^ denotes significant differences compared to 4 weeks within the same phenotype with p<0.05, while *, ** and *** denote significances of differences compared to the M1 phenotype after 2 days with p<0.05 and p<0.01, p<

96 Contribution of macrophage phenotype to scaffold degradation 5.5 Discussion Macrophages are claimed to play an essential role in tissue outcome and show a continuum of functional properties alternating between the pro-inflammatory (M1) and the tissue-healing (M2) phenotypes. In order to elucidate the effect of macrophage phenotype on scaffold degradation 2D and 3D in vitro studies were performed. During 2D cultures, we observed differences in cell morphology between the groups, which is indicative for different phenotypes. Mainly elongated cells were observed throughout the whole culture time when macrophages were polarized using LPS/IFNɣ and mainly rounded cells in case of IL-4 and IL-13 treatment. This was in line with findings by others [242, 244], while McWhorther et. al. [245] found the opposite morphology during polarization of macrophages after 1 day. This might be due to the use of another cell source and species by McWhorther, as they cultured mouse bone marrow derived macrophages instead of human cells. Apparently, macrophages are very sensitive cells and do not only react to cytokines present in their environment, but also to other substances, such as the PMA concentration that influenced cell shape [246]. Furthermore, it has been described by others that macrophage morphology is also surface-dependent [247, 248]. This indicates that morphology may not be a reliable indicator of macrophage phenotype and phenotype-specific markers should be used in addition to distinguish between cell phenotypes. Also, to make a fair comparison between 2D and 3D cultures, 2D substrates should be made of the same biomaterial that is used in 3D cultures. This was not performed within this study, as this is beyond the scope of our study. Several studies demonstrated that macrophages were not only polarized by secreted cytokines, as also the scaffold surface, cell shape, fiber diameter, pore size, and strain are reported to influence the macrophage phenotype [ , 249]. This indicates that scaffold morphology and composition can be used to promote an optimal healing response. Garg et. al. showed that a high fiber diameter (>3 µm) together with a high porosity (>80%) results in a transition of macrophages towards the M2 phenotype [144]. Also in our study, the results of qpcr data indicated a shift from the unpolarized macrophages towards the M2 phenotype when exposed to the PCL scaffold, as apart from expression levels of the CD163 and MMP9 genes, which were increased at some time points for the unpolarized samples, no other differences in expression levels were observed between the unpolarized and the IL4/IL13 polarized samples. This phenotypic shift towards the M2 type is likely due to the scaffold morphology and composition. M2 macrophages can be further classified into different sub-phenotypes [136, 233]. The differences in MMP9 and CD163 levels between the unpolarized cells and the IL4/IL13 polarized cells might be indicative for a shift of the unpolarized cells towards another subphenotype of the M2 macrophages as compared to the sub-phenotype of the IL4/IL13 polarized cells. IL4/IL13 are believed to polarize the macrophages towards an M2a phenotype, and MMP9 secretion is mainly observed in the M2b phenotype [233]. 5 87

97 Chapter 5 Within our study, it seems that addition of cytokines overruled the effect of high fiber diameter and porosity (10 µm and 80% respectively) which guided the unpolarized cells towards the M2 phenotype, as after each burst of LPS/IFNɣ cytokines the macrophages seemed to remain their M1 phenotype, and did not transdifferentiate towards the M2 phenotype, based on the amount of DNA over time and the qpcr data after 2 days. The LPS/IFNɣ polarized cells showed a reduced viability in both 2D and 3D cultures compared to unpolarized cells and IL4/IL13 polarized cells. This might be due to the observed increased levels of TNF-α expression in the LPS/IFNɣ polarized cells, which is known to induce cell death [250, 251]. This was in line with findings by others that showed increased levels of TNF-α production in LPS/IFNɣ polarized cells [242, 252]. In a review by Italiani et. al. [253] it was described that M1 cells are end-stage killer cells, which die during the inflammatory response, probably due to its own nitric oxide (NO) production [254]. Predominantly M2 type of macrophages were observed around implants in several studies [95, 255]. However, it needs to be further clarified whether this is a transition from the M1 phenotype towards the M2 phenotype within the same cells, within the cell population, or whether a selective death of M1 macrophages due to apoptosis results in a relative increase of cells with the M2 phenotype. SEM images of scaffold fibers after removal of the cells showed more local damage of the scaffold fibers in the LPS/IFNɣ polarized cells group compared to the unpolarized cells and IL4/IL13 polarized cell groups. This was also expected, as LPS/IFNɣ polarized cells, are known to produce high levels of enzymes and ROS that contribute to scaffold degradation [136]. Esterase assays confirmed that the amount of secreted esterase per cell was increased in the LPS/IFNɣ polarized cells compared to the unpolarized cells and IL4/IL13 polarized cells. 5.6 Conclusion We were able to generate distinctive macrophage phenotypes in both 2D and 3D cultures by the addition of cytokines, and observed that the life-span of LPS/IFNɣ polarized cells was shorter compared to IL4/IL13 stimulated cells. In 2D, cell morphology was different between the phenotypes, while this was less pronounced in the 3D cultures. Using qpcr we were able to distinguish between the cell phenotypes present on the scaffold fibers. Unpolarized macrophages in PCL scaffolds expressed similar genes as compared to the IL4/IL13 polarized cells, which is indicative for a transition towards a M2 phenotype, and is probably induced by the electrospun mesh, which is a beneficial feature for in situ tissue engineering. We observed that all macrophage phenotypes were able to secrete esterases, which are known to degrade the PCL scaffold fibers. All phenotypes indeed contributed to local scaffold degradation, although this was far more pronounced in the LPS/IFNɣ polarized cells compared the unpolarized and IL4/IL13 polarized cells. Given this 88

98 Contribution of macrophage phenotype to scaffold degradation knowledge, we suggest a correlation between macrophage phenotype and scaffold degradation, with inflammatory macrophages accelerating degradation. Acknowledgements This work was supported by a grant from the Dutch government to the Netherlands Institute for Regenerative Medicine (NIRM, grant No. FES0908). The authors gratefully thank Roel Lalieu from Xeltis for electrospinning of the PCL scaffolds. Virginia Ballotta is acknowledged for providing some of the primers. 5 89

99 Chapter 5 90

100 General discussion 6

101 Chapter 6 Worldwide, cardiovascular diseases are the major cause of death, resulting in a need for bypass surgeries and heart valve replacements each year [11, 22, 256]. With the ageing population and improved diagnostic methods, this number will only increase. The currently available replacements use non-living materials, and are as a consequence unable to remodel and adapt to changes in their environment. Tissue engineering attempts to overcome these shortcomings, and aims at implantation of bioresorbable grafts that will fully integrate with the host tissue, leaving behind a living tissue that is able to adapt and remodel. Various tissue engineering approaches have been investigated in the past with the intention of effective regeneration of diseased cardiovascular tissues. In situ tissue engineering offers several advantages over the classical in vitro tissue engineering approach, as this approach aims to employ off-theshelf products, reducing the costs and time to produce replacements, and comprises less regulatory issues. The main challenge is to find the proper biomaterial that can be used to create matrices that maintain good mechanical integrity, immediately after implantation and start resorbing when sufficient neo-tissue has been formed that can take over this role. As different properties are desired in various applications, it is of high importance to be able to tune the mechanical properties and resorption characteristics of biomaterials. The focus of this thesis was to investigate degradation characteristics of electrospun scaffolds, manufactured from different supramolecular biomaterials. Furthermore, the interplay between scaffold degradation rate and the amount and composition of neo-tissue was examined. 6.1 Main findings of the thesis Slow-degrading scaffold material reduces compaction and retraction A balance between tissue formation and scaffold degradation is believed to be important to maintain a functional replacement with proper mechanical integrity. Slow tissue synthesis or too fast resorption of the implant might disturb this delicate balance, and has shown to result in compaction and retraction of in vitro tissue engineered heart valves, causing regurgitation in vivo [71, 72, 257]. Traction forces exerted by smooth muscle cells, the cells that lay down a new layer of tissue, are believed to enhance this process [74, 151]. During tissue remodeling, the amount of these cells will decrease, which therefore probably also results in lower traction forces. To bridge the phase where tissue is not mature enough to withstand the forces exerted by the cells and constant loading of the implant, a scaffold with sufficient mechanical integrity during the first months after implantation is desired. Therefore, we studied in vitro whether the use of slow- (PCL) instead of a fast-degrading (PGA-P4HB) electrospun scaffold meshes, and a lower cell passage number to enhance tissue formation, reduces compaction (chapter 2). Reported time for complete in vivo bioresorption of PGA-P4HB varies from 4 (PGA) to 8 (P4HB) weeks [68] to 4-6 months for PGA-P4HB, with 50% loss of mechanical properties within 2 92

102 General discussion weeks [258]. Complete bioresorption of electrospun PCL scaffolds is reported to be at least 1-2 years [58, 183]. We demonstrated that the use of a slow-degrading material resulted in improved resistance to retraction of tissue engineered valvular leaflets and reduced compaction of strips compared to fast-degrading material. Tissue formation, stiffness and strength increased with decreasing passage number, however, this did not influence compaction. Furthermore, tissue constructs were engineered using both ovine and human cells, to determine the effect of interspecies differences on tissue development. Although variations between the actual amount of ECM components were found between the species, the effects e.g. on compaction were comparable. Overall, in terms of compaction, the influence of the scaffold type seemed larger than the influence of the tissue production of several cell sources Organized tissues which maintain their 3D shape when cultured onto slowdegrading scaffold materials contain similar ECM values compared to native pulmonary heart valves Scaffolds with slow- and fast-degradation rates will contribute to the mechanical integrity of the implant differently with time. We hypothesized that cells on fast-degrading scaffold material will produce increased amounts of tissue compared to cells on slow-degrading materials, to compensate for the loss in mechanical integrity. Using in vitro tissue engineering, we studied tissue evolution, in terms of ECM composition and mechanical properties of the constructs in time, of vascular cells cultured on slow- (PCL) or fastdegrading (PGA-P4HB) electrospun scaffolds (chapter 3). It was shown that tissues cultured on slow-degrading scaffolds contained organized tissue formation maintaining their 3D shape during culture, while the tissues cultured on fast-degrading scaffold materials demonstrated appositional growth and compaction during culture. This again demonstrated that slow-degrading scaffold material is favored over fast degrading scaffolds to ensure stable mechanical integrity during the initial phase after implantation. During the first two weeks of culture, when the PGA-P4HB scaffold was clearly degrading, the cells cultured onto these scaffold meshes were more synthetic in agreement with our hypothesis. However, this synthetic phenotype was only a temporary feature, as after 6 weeks lower amounts of sgag and collagen were measured in the PGA-P4HB based tissues when compared to the PCL-based tissues. Compaction of the PGA-P4HB scaffolds resulted in a smaller surface area, and less volume for the cells within these tissues to lay down their ECM, compared to PCL-based tissues, which might have resulted in this higher amount of ECM after 6 weeks of culture in the latter. To make a fair comparison between tissue composition of the in vitro engineered tissues grown on scaffolds with a different degradation rate, and between in vitro engineered tissues and native tissues, we described a method to correct for the amount of remaining scaffold weight. Implementing this correction, we found ECM values that were similar to, or towards values of native pulmonary heart valves. Although collagen crosslink values were increasing with in vitro 6 93

103 Chapter 6 culture time, the measured values were still lower in engineered tissues compared to their native counterparts PCL and PCL-based supramolecular polymers exhibit different degradation characteristics As in vitro culture of cardiovascular substitutes is expensive and time-consuming, in situ tissue engineering would be a good alternative. This would also allow for off-the-shelf availability of implants with reduced production time and less regulatory issues that are related to tissue culture. Unpublished data by our group demonstrated that the slowdegrading PCL material resulted in plastic deformation when it was cyclically loaded. Consequently, this material is not suitable for in situ tissue engineering of heart valves, as dynamic loading of the leaflets places high demands on the scaffolds immediately after implantation. Supramolecular biomaterials contain hydrogen bonding motifs, like UPy or BU incorporated into their molecular structure, resulting in suitable materials for in situ heart valve tissue engineering. In vivo, resorption of the implanted material can be accomplished via two main pathways, which includes the enzymatically accelerated hydrolytic pathway and the oxidative pathway. To investigate both pathways, separately and in an accelerated fashion, in vitro degradation assays were designed. With the use of these models, degradation characteristics of several promising supramolecular materials were explored and compared to the conventional PCL material (chapter 4). We demonstrated that, depending on the morphological and chemical composition of the materials, conventional and supramolecular PCL-based electrospun meshes responded differently to both pathways. The enzymatic accelerated hydrolytic pathway mainly affected conventional PCL scaffolds, while supramolecular materials were not (PCL-UPy) or only mildly (PCL-BU) affected, which was enzyme dependent. PCL material was not susceptible to oxidative degradation. The supramolecular PCL-UPy materials were, dependent on the PCL soft segment length and the supramolecular moiety coupled to the PCL backbone, not susceptible (PCL800-UPy) or susceptible (PCL2000-UPy and PCL-BU) to oxidative degradation. When materials were treated with the enzymatic solution, surface degradation seemed to be the dominant degradation mechanism for PCL and PCL-BU. When exposed to oxidative solutions, surface erosion was observed for PCL-BU, while both surface erosion and bulk erosion was seen in the PCL-UPy materials. This insight into the degradation characteristics of PCL-based (supramolecular) materials allows us to tailor degradation characteristics. Different combinations of polymer backbones modified with supramolecular moieties can be created which result in various polymer properties such as mechanical properties and/or resorption rate. This mix-and- 94

104 General discussion match toolbox can be utilized to screen and select the relevant biomaterial for pre-clinical in vivo studies targeted to different clinical applications Electrospun PCL scaffolds guide tissue regeneration and are mainly degraded by inflammatory macrophages During the immune response, which starts immediately after implantation of a scaffold material, macrophages are known to play an important role in the resorption of the implant. Macrophages possess plastic functional properties and represent a continuum in which they can alternate, dependent on micro-environmental factors, between the proinflammatory (M1) and the healing (M2) macrophages [136, 234]. Polarization of macrophages towards the healing-type induced by the scaffold material is desired to improve final tissue outcome. To illustrate whether there is a correlation between the inflammatory or the healing macrophage phenotypes and the degradation of electrospun meshes, in vitro tests were developed and used (chapter 5). In 2D culture, a clear morphological difference was observed when different cytokines were added to the macrophages in order to polarize them towards the inflammatory or healing phenotype, which was less pronounced in 3D cultures. Cells of the inflammatory phenotype had a shorter life-span compared to cells of the healing phenotype or the untreated macrophages, in both 2D and 3D cultures. During 3D cultures of different macrophage phenotypes, we visually observed enhanced degradation of the scaffold fibers by the inflammatory phenotype. No difference, in terms of scaffold degradation, was observed between the healing phenotype and the unpolarized macrophages, which implies that the latter polarized towards the healing phenotype in 3D cultures. This was confirmed by gene expression studies, where the gene expression profile of the unpolarized macrophages was most similar to the gene expression profile of the healing macrophages. In summary, the results presented in this thesis suggest that the choice of a scaffold material is of high importance to maintain a good balance between scaffold degradation and tissue formation, and therewith maintaining mechanical integrity. A slow-degrading material is favored over a fast-degrading material, as mechanical integrity will be maintained for a longer period, which is mainly important in in situ tissue engineering purposes where a bare scaffold is implanted. Furthermore, neo-tissue seemed to be better organized when cultured on slow-degrading scaffold materials and less prone to compaction. Macrophages are known to play an essential role in scaffold degradation. Although all macrophage phenotypes seemed to be able to degrade scaffold material in vitro, we visually observed a higher degree of scaffold degradation by the inflammatory phenotype compared to the healing phenotype. Untreated macrophages that were cultured into the scaffold, polarized into the healing phenotype, indicating that the used material is guiding tissue regeneration rather than repair. 6 95

105 Chapter Towards the most promising tissue engineering approach and scaffold material In vitro tissue engineering versus in situ tissue engineering In order to replace diseased cardiovascular tissues, different tissue engineering approaches are explored, as described in the introduction of this thesis. Although promising results are obtained within every approach, they also all have their challenges that require further optimization before safe translation to the clinic. The classical in vitro tissue engineering approach requires months before an engineered construct is suitable for implantation. Decellularization of the in vitro tissue engineered constructs improves the classical approach, in terms of readily available, off-the-shelf products. This decellularization approach also provides tissues replacements for those patients who do not have tissues available for cell isolation, or are in such a critical situation that a waiting time of a few months would induce high mortality risks. Despite this improvement, the decellularization approach has to overcome more challenges before it will result in reliable implants, as compaction and retraction of decellularized in vitro tissue engineered heart valves is still a hurdle to overcome [43, 71-73]. Results of this thesis show that the use of slow-degrading scaffold materials probably will result in further improvements. Furthermore, changes in geometry could help in preventing compaction and retraction. In situ tissue engineering using synthetic materials is a promising approach. It demonstrates off-the-shelf availability, while production-time and -costs and regulatory issues are significantly reduced compared to classical in vitro tissue engineering methods. The in situ tissue engineering approach relies on the regenerative capacity of patients. A challenge that needs to be addressed is the search for the appropriate scaffold material in relation to the specific application. An immune response needs to be triggered and controlled after implantation to ensure migration of different cell types towards the implanted material. These include macrophages, which are involved in resorption of the scaffold material and regulation of the healing response, and cells that generate the neotissue. Several groups aim at active cell capture of specific cell-types involved in tissue formation. Therefore, they create bioactive scaffold fibers, by incorporating peptides to the material, or coat the fibers with growth factors or antibodies, which are released after implantation in order to recruit specific cells [259, 260]. When certain cells are actively captured in the scaffold material, the natural immune response is influenced and might even be disturbed. Whether the bioactive scaffold materials have a beneficial or negative effect on the final tissue outcome needs to be further elucidated. In situ tissue engineering without the addition of peptides, cytokines or cells to the scaffolds relies on the properties of the material together with a natural healing response and endogenous tissue growth (ETG) by the body itself. This is beneficial, because the absence of bioactive moieties reduces regulatory requirements and simplifies 96

106 General discussion the production process as only synthetic biomaterials are involved. Promising pre-clinical [90, 91] and clinical data [92] were gathered using this approach Choosing the appropriate scaffold material Electrospun scaffold materials aimed to replace cardiovascular tissues by the in situ approach need to fulfill specific requirements [261]. They should be biocompatible, to prevent a severe inflammatory reaction that might impair the healing cascade, or even cause rejection by the body. Another requirement is controlled resorption of the material, with non-toxic resorption products that are removed from the body without side-effects. Dependent on the application, higher or lower mechanical forces will be applied to the material. It is of high importance that the scaffold is able to withstand these forces and ensures mechanical integrity immediately after implantation, until neo-tissue is able to take over this function. The supramolecular materials PCL-BU and PCL-UPy are promising materials to be used in in situ tissue engineering, as they contain strong and elastic properties that can withstand the repetitive loading forces that are exerted immediately after implantation, when the scaffold is implanted as a heart valve replacement. Cells behave differently when growing on a stiff or elastic material. For example, PCL is known to be stiffer compared to PCL-BU. Therefore, the PCL-BU scaffolds display more stretch in vivo compared to PCL scaffolds when implanted at the same anatomical location. Scaffold architecture also influences tissue outcome and additional requirements can be added to promote optimal healing responses. First, scaffolds require an interconnected pore structure with a high porosity to allow for cell infiltration and tissue formation, including vascularization [262, 263]. In addition, it has been shown that fiber diameter not only influences porosity and, thereby, cell infiltration, but also influences the polarization of macrophages during the inflammatory response. A healing phenotype was observed in scaffolds with a high fiber diameter (>3 µm) together with a high porosity (>80%) [144, 145]. This was also observed within our experiments in chapter 5. Macrophages preferentially polarized towards the healing type when cultured on PCL scaffolds with a fiber diameter of 10 µm and a porosity of 80%. Furthermore, it is demonstrated that alignment of the scaffold fibers stimulates the cells to increase collagen synthesis [264, 265]. Taken together, we aim for a scaffold that allows cell infiltration, modulates the immune response, and supports tissue formation that is able to remodel and gradually takes over the mechanical functionality of the scaffold. Ideally, the scaffold should be completely resorbed, to prevent ongoing inflammatory responses against the foreign body material. Finally, easy surgical handling is an additional benefit for clinicians who are implanting these scaffolds. 6 97

107 Chapter Focus on in vivo resorption characteristics Balance between scaffold resorption and tissue formation An important aspect to consider when choosing a biomaterial is its resorption characteristic. In case of too slow resorption of the material stress shielding of the neotissue might occur, thereby impeding the regeneration process [231]. Furthermore, slow resorption might lead to other undesirable outcomes, like a prolonged inflammatory response. When the resorption process is too fast the mechanical integrity of the implant is not maintained, as the neo-tissue is not sufficiently developed yet to bear the full mechanical force required [232], leading to failure of the implant. The use of fastresorbing materials might be one of the reasons that contributed to heart valve leaflet retraction and compaction observed in several in vivo studies [43, 71-73]. This is a result of traction forces exerted by αsma positive cells, likely in combination with an imbalance of the newly formed tissue and fast loss of mechanical integrity of the scaffold due to resorption [74, 148, 151]. As αsma is related to traction forces of the cells [152], and αsma positive cells were demonstrated to decrease again in vivo [6], these traction forces will also be decreased. Therefore, a slow-resorbing scaffold with sufficient mechanical integrity during the first phase after implantation is desired to withstand the cell traction forces during this period. Furthermore, we demonstrated in chapter 3 of this thesis that in vitro less organized tissue was formed, that lost its original shape when cultured onto fast-degrading scaffold materials. Findings by Hasizume et. al. [266] also demonstrated that implantation of a slow-resorbing scaffold patch to treat chronic ischemic cardiomyopathy in rats resulted in beneficial results in terms of cardiac function and histology compared to faster resorbing patches. In addition, de Jonge et al. observed during an in vitro study that after 2 weeks of culture, the newly formed collagen fibers were not dense enough yet to withstand the traction forces of the cells and resulted in collagen reorientation [49]. This suggests that a slow-resorbing scaffold material should be chosen, that allows the newly formed collagen fibers to mature first, and therewith are able to withstand the loads that are applied on the constructs, when aiming at maintaining collagen orientation Size and anatomical location The size of the grafts might also influence the material selection. Both cells from the bloodstream and the adjacent tissue site will infiltrate into the graft, followed by tissue formation. Complete cell infiltration and corresponding tissue formation throughout the graft is expected to occur faster in short replacements e.g. short interpositional vascular grafts, where both transmural and transanastomotic ingrowth contribute to fast cell infiltration throughout the replacement, compared to long replacements like reconstructions of aortic aneurysms where a larger area needs to be infiltrated. This would suggest the use of a slow-resorbing material when long tissues need to be replaced. In addition, the anatomical location of the implant might influence both the cell 98

108 General discussion infiltration and the resorption rate of a biomaterial. Mechanical forces, like compression, fatigue and shear stress, or other factors like flow, ph and the presence of enzymes might affect the resorption rate of the implanted material [121]. With the use of slow-resorbing materials, the question might arise what will happen with the neo-tissue when the scaffold mesh is completely resorbed by the body. We hypothesize that the neo-tissue has prolonged time to mature when growing on slowresorbing scaffolds and, thereby, is able to withstand the loading forces applied on the constructs. Although preliminary results [103] showed decreased mechanical contribution of the implanted heart valve scaffolds in the ovine model after 6 months, longer-term in vivo experiments are needed to demonstrate the fate of the tissue after complete loss of mechanical integrity of the scaffolds. Long-term animal experiments up to when the implant is completely resorbed, should be performed to investigate the final outcome The interplay between direct cell contact and resorption of the scaffold fibers It is suggested that direct contact between the cells and the scaffold fibers is needed in order to degrade the scaffold material [131, 267]. We also found indications that are in line with this hypothesis as pilot in vitro studies on the interplay between macrophages and scaffold fibers showed surface erosion of the scaffold fibers by the macrophages, clearly visible when cells were removed from the scaffold fibers (Figure 6.1). No surface erosion or cracks were observed in fibers of scaffold meshes that were cultured in the same medium, and thereby are in contact with the same amounts of enzymes and ROS products, however separated from the macrophages with a porous membrane to avoid direct cell contact. This indicates that direct cell contact is needed, probably to very locally create a high concentration of enzymes and/or ROS products to degrade the scaffold material. 6 99

109 Chapter 6 Figure 6.1 Schematic figure of a transwell assay experiment (A). In the upper compartment, cells are in direct contact with the scaffold material. The secreted enzymes and oxygen radicals can migrate through a porous membrane towards the lower compartment, where they can reach the scaffold material that is not in direct contact with the cells. SEM images were taken from the scaffold in the upper compartment with cells (B), and after removal of the cells (C). Small holes and surface erosion (black arrows) were observed in the scaffolds that were in direct contact with the cells. The damage was observed underneath the cells and in the near environment (<5 µm) where cells adhered to the scaffold fibers. No holes or surface erosion due to cells could be observed in the scaffold that was cultured in the lower compartment (D), although some general hydrolytic degradation (white arrow) was observed which is common after 4 weeks of culture in an aqueous solution. White scale bars represent 20 µm. Furthermore, in vivo pilot studies using the ovine model [103] demonstrated mainly scaffold resorption at those sites where macrophages infiltrated into the porous scaffolds and tissue was formed (Figure 6.2). Infiltration of cells and tissue formation in the valves was observed to start from the wall, and with time cell infiltration further towards the leaflet tips was observed. This is in line with previous findings where implantation of decellularized in vitro tissue engineered heart valves demonstrated fastest repopulation with highest densities in the tissue wall compared to the heart valve leaflets [73]. The observation that tissue formation always precedes resorption, even though local differences are observed in rates of regeneration is an important safety aspect related to in situ implantation of scaffold materials. 100

110 General discussion Figure 6.2 SEM images of a 6-month valvular explant [103]. Cell infiltration and corresponding tissue formation started from the wall and belly region of the leaflets (C). With time cells infiltrated in the center of the leaflets (B) and further towards the leaflet tips (A). After removal of the tissue, we observed that at those places where no or few cells were infiltrated and therewith no or little tissue was formed, scaffold fibers were only minimally affected (D). When many cells were infiltrated, which was correlated to a large amount of tissue formation, scaffold fibers were severely affected by the infiltrated cells (E). Black and white scale bars represent 1 mm and 20 µm, respectively. Taken together, this would advocate use of a slow-resorbing material to ensure sufficient time for the neo-tissue to form, remodel and mature before the mechanical integrity of the scaffold material is completely lost. As the desired resorption properties are depending on the application, the need to tailor the properties of bioresorbable polymers specifically to the intended application is necessary. 6.3 Study limitations and the future of in situ cardiovascular tissue engineering Benefits of in vitro models In order to understand tissue development on implanted scaffold meshes and how this is affected by scaffold resorption, in vitro model systems providing a 3D environment are very useful. With these in vitro models, we can obtain insight in the interactions between cells and scaffold materials and tissue formation. Furthermore, the necessity of animal 101