Nanostructured Electrochemical Biosensors: Towards Point of Care Diagnostics

Transcription

1 Nanostructured Electrochemical Biosensors: Towards Point of Care Diagnostics by Brian Lam A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Chemistry University of Toronto Copyright by Brian Lam 2013

2 Nanostructured Electrochemical Biosensors: Towards Point of Care Diagnostics Abstract Brian Lam Doctor of Philosophy Department of Chemistry University of Toronto 2013 An important research area in medicine is molecular diagnostics of cancers and infectious diseases, which can be diagnosed, managed and treated more effectively with genetic information. We have developed an integrated sample to answer bacterial detection platform combining a simple, universal bacterial lysis approach and sensitive nanomaterial electrochemical biosensors. Lysis is rapid and effective at releasing intercellular nucleic acid targets. The platform was directly challenged with unpurified lysates and successful at determining the presence of clinically relevant concentrations within 30min from sample to answer. Another important aspect of biosensor development is the development of cheap and efficient methods for manufacturing nanostructured microelectrodes. Previously, we have used costly silicon wafers for fabrication. Here we explored alternate inexpensive materials for fabrication including printed circuit boards, plastics and glass. We show that plain borosilicate glass is effective for templated bottom-up fabrication, with comparable performance to expensive silicon based nanostructured microelectrodes. ii

3 Current state-of-the-art readout of many biomarkers is hampered by serially addressing arrays of low cost biosensors, without the use of high cost active electronics. Here we have developed a new concept, solution-based electrochemical circuits, which makes highly multiplexed sensing feasible on the surface of low-cost, glass chips. This method utilizes the idea that physical separation of liquid on an insulator can result in electrochemical isolation. Using this we can reduce the number of outputs to 2 n, where n would be the number of serially connected sensors. We use urinary tract infections as a model system and prove that we can accurately detect species and antimicrobial resistance in multiplexed formats at clinically relevant concentrations. iii

4 Acknowledgments First I would like to thank Professor Shana O. Kelley for giving me the opportunity to work in her group and giving me freedom to pursue new ideas. You have always been encouraging with invaluable suggestions and compliments, in times where sometimes I fail to see the success of my work. Also you have shown me what it takes to become an exemplary principle investigator. I would like to thank Professor Edward H. Sargent for advice and challenging me to be a better scientist and engineer. You have provided me with countless helpful suggestions and many new and exciting ideas have arisen from our discussions. I would like to thank my committee members Professor Gilbert Walker for the invaluable advice and support you have provided me before and throughout my graduate studies, and Professor Aaron Wheeler for helpful suggestions and support. I would like to thank Dr. Jagotamoy Das for important contributions with who I collaborated closely with on many aspects of this work. I would like to thank Dr. Zhichao Fang, Elizaveta Vasilyeva, Dr. Leyla Solyemani, Dr. Ludovic Live, Andrew Sage, Richard D. Holmes for their important contributions to this work. I would like to thank all past and current members of the Kelley group including, Dr. Mark Pereira for his suggestions and advice with bacteria, Hooman Zamani for his cheerful advice and support. I would like to thank Justin Besant, Dr. Reza Mohamadi, Mario Moscovici, Rida Mourtada, Sean Guo, Alexander Zaragoza, Barbara Alexander, Gabriela Kranac for help and support in and outside the lab. I would like to thank my mother Dorothy, sister Karen and father Lawrence for their support love and support throughout my life and graduate career. Finally I would like to thank Liza for carrying that centrifuge and educating me that bacteria do not grow properly in a conical tube. Thank you for your love and support in my life and graduate studies. iv

5 Table of Contents Contents Acknowledgments... iv Table of Contents... v List of Figures... viii List of Abbreviations... xi 1 Medical Diagnostics Past, Present and Future A Brief History of Medical Diagnosis Age of Molecular Diagnostics Medical Diagnosis of Cancers and Infectious Disease Point-of-Care Biosensor Technology Scope of Thesis References Electrochemical Biosensor Methodology and Background Electrochemical Biosensors Electrochemical Techniques Glucose Biosensors Nanostructured Microelectrode Electrochemical Biosensors Platform Development Goals References Polymerase Chain Reaction-Free, Sample-to-Answer Bacterial Detection in 30 Minutes with Integrated Cell Lysis Abstract Introduction Methods and Materials v

6 3.4 Results and Discussion Conclusions References Optimized templates for bottom-up growth of high-performance integrated biomolecular detectors Abstract Introduction Materials and Methods Results and Discussion Baseline performance of sensors fabricated on silicon Testing of printed circuit board as a substrate for sensor deposition Testing of plastic as a substrate for sensor deposition Testing of glass as a substrate for sensor deposition Validation of clinically-relevant sensitivity and specificity using glass chips Conclusions References Solution-based circuits enable rapid and multiplexed pathogen detection Abstract Introduction Methods and Materials Results Overview of approach Characterization of the SCC Detection of urinary tract infection pathogens Multiplexed detection of urinary tract infection pathogens Discussion vi

7 5.6 References Conclusions and Future Directions Thesis Findings Future Work Publications and Other Contributions Supplementary Information Microchannel Electrical Lysis Conclusions Description of Solution Circuit Chip Description of SCC platform Interference evaluation References ` vii

8 List of Figures Figure 1.1 First Medical X-ray Figure 1.2 Structure of DNA... 3 Figure 2.1 Three electrode electrochemical cell Figure 2.2 Cyclic Voltammetry Figure 2.3 Differential Pulse Voltammetry Figure 2.4 Chronoamperometry Figure 2.5 Freestyle glucose meter and sensor strips Figure 2.6 Nanowire polycarbonate template fabrication Figure NME fabrication process Figure 2.8 Electrode Morphologies Figure Electrocatalytic nucleic acid assay Figure 2.10 NME sensitivity and specificity Figure Bacterial detection sensors Figure Integrated sensing system Figure Characterization of electrically lysed bacterial solutions Figure Direct bacterial detection in unpurified lysates Figure 4.1- Silicon-based NME characterization Figure PCB-based NME characterization Figure Plastic-supported NME characterization viii

9 Figure Glass-supported NME characterization Figure Glass-based NME assay validation Figure The solution circuit chip Figure Electrochemical validation of the SCC Figure Validation of pathogen and antibiotic-resistance probes Figure Multiplexed pathogen and antibiotic-resistance testing on an SCC Figure 8.1 PDMS microchannel electrical lysis Figure 8.2 Microchannel schematic Figure Experimental lysis setup Figure Electrical lysis of Ecoli and Ssap Figure 8.5 Bacterial concentration versus electrical lysis Figure 8.6 Flow rate versus electrical lysis Figure 8.7 Applied voltage versus electrical lysis Figure 8.8 Optical images of lysed versus unlysed bacteria Figure 8.9 Propidium iodide uptake versus microchannel lysis Figure 8.10 Flow cytometry propidium iodide uptake Figure 8.11 RT-PCR confirmation of mrna target release on lysed samples Figure Experimental workflow electrochemical detection from sample to answer Figure Sensor footprint versus accumulation time Figure Electrochemical detection of bacterial lysates for different size sensors ix

10 Figure 8.15 Chronic mylegenous leukemia Figure Microchannel lysis of K562 cells Figure 8.17 Representation of 5x5 array of an SCC chip Figure 8.18 Cross section schematic of single NME in SCC chip Figure 8.19 SCC chip connected in traditional electrochemical setup Figure 8.20 Physical liquid channel separation orthogonal to common WEs Figure 8.21 Illustration of SCC method Figure 8.22 SCC interference from sequential addition of Ru/Ferri Figure 8.23 Hypothesis of interference phenomena in SCC chips Figure Interference theory experiment Figure 8.25 Elimination of interference x

11 List of Abbreviations CA Chronoamperometry CE Counter electrode cfu Colony forming units CML Chronic myelogenous leukemia CV Cyclic voltammetry DNA Deoxyribonucleic acid DPV Differential pulse voltammetry FAD Flavin group FEP Fluorinated ethylene polymer GOx Glucose oxidase IPA Isopropyl alcohol MB Methylene blue mrna Messenger ribonucleic acid NME Nanostructured microelectrode PCB Printed circuit board PCR Polymerase chain reaction PDMS polydimethylsiloxane PI Propidium iodide PNA Peptide nucleic acid RE Reference electrode RNA Ribonucleic acid rpoβ RNA polymerase gene RT-PCR Reverse transcriptase polymerase chain reaction SCC Solution circuit chip SEM Scanning electron microscopy UTI Urinary tract infection WE Working electrode β-lac - β-lactamase xi

12 1 1 Medical Diagnostics Past, Present and Future Medical diagnostic technology development is one of most important fields of applied science since it directly affects the overall health of the general population. Throughout history, clinicians and scientists have endeavored to develop more effective technologies to rapid and accurate diagnosis of medical aliments [1]. 1.1 A Brief History of Medical Diagnosis One of the first instances of medical diagnostic device was the stethoscope, which has become the most widely used diagnostic device by clinicians to date. Prior to its inception in 1816 by Rene Theophile-Hyacinthe Laennec, since the age of Hippocrates in 400 BC, clinicians placed their ears directly onto a patient s chest to investigate cardiovascular health [2]. Laennec also used this method until one day while listening to a patient s chest he recalled that he could hear a pin scraping at one end of a plank while his ear was placed at the other end. He quickly rolled up a stack of parchment and placed it to the patient s chest and was surprised that he could hear the heart more clearly than if he placed his ear to their chest. Laennec went about manufacturing the first stethoscope which was essentially a simple wooden monaural tube and has been developed over the past two centuries into a binaural flexible device with integrated low pass filtration among other numerous advances. The stethoscope provided valuable information on the internal workings of the human body while still remaining totally non-invasive. This was the first instance of non-lethal device for exploring internal anatomy and has shaped the face of the medical profession. It is responsible for the, discovery of new diseases, the creation of criteria for accurate and rapid diagnosis and the development of many new treatments. Also it has become the most recognizable symbol of the medical profession [3] where patients trust doctors more when they are wearing a stethoscope. The next most important medical tool was the discovery and use of X-Rays by Wilhelm Conrad Röntgen in 1895 [4]. Röntgen at the time was testing vacuum tubes, provided by Hertz and Tesla amongst others, as electrical charges discharged through them. Serendipitously, although the tube was covered by cardboard, a neighboring barium platinocyanide screen emitted a faint florescence a few feet away. He surmised that a new type of radiation was responsible, which he

13 2 named X-Rays. A few weeks after his discovery he took the first medically related X-Ray of his wife s hand (Figure 1.1) to which she exclaimed I have seen my death! Deservingly Röntgen won the first Nobel prize in physics in 1901 for his discovery which has affected countless lives and paved the way for new medical discoveries and diagnostic techniques. Figure 1.1 First Medical X-ray. Röntgen s wife s hand. [4] Another very important discovery that had a profound impact on the future of medical diagnostics was the discovery of the structure of deoxyribonucleic acid (DNA) by James Watson and Francis Crick in 1953 [5]. DNA was first isolated by Friedrich Miescher in 1869 in the pus of used surgical bandages. He observed a microscopic substance in the nucleus of cells which he named nuclein. Albrecht Kossel was the first to isolate the five primary nucleobases in 1878, followed by Phoebus Levene who discovered the phosphate, sugar and base unit in Interestingly enough the structure of DNA was discovered utilizing the diffraction of X-rays, and the analysis of a single diffraction image, Photo 51 (Figure 1.2A) taken by Rosalind Franklin and Raymond Gosling in By chance, Watson was shown Photo 51 by Franklin and it

14 3 convinced him that the structure of DNA must be made up of two intertwining chains in a paired helix resembling a spiral staircase. Working closely together, Watson and Crick developed a stick and ball model of the structure of DNA where the sides of the staircase was made of alternating sugar deoxyribose and phosphate molecules, and the stairs consisted of paired bases. (Figure 1.2B) Their model was published in Nature in 1953 and essentially gave birth to the field of molecular biology, and consequently, in 1962 they received the Nobel Prize in Physiology/Medicine for one of the most important discoveries of humankind. Figure 1.2 Structure of DNA (A) Photo 51: X-ray diffraction image of DNA (B) Watson & Crick ball and stick representation of the double helix structure of DNA [5] 1.2 Age of Molecular Diagnostics After the discovery of the structure of DNA, much work went into studying its role in all life on earth. The discovery gave birth to the new age of molecular diagnostics and much effort was expended to understand DNA s role in the function of the cell and disease. A host of new techniques were developed to study the role of nucleic acids and read its sequences. One of the first sequencing techniques developed was by Fredrick Sanger and coworkers in 1977 [6]. It is a chain termination method relying on terminating DNA templates with dideoxynucleotides that

15 4 are fluorescently or radioactively labeled followed by size separation with electrophoresis to read sequences in order. Following the Sanger method, one of the most important techniques developed was polymerase chain reaction (PCR), which was developed by Kary Mullins in 1983 [7]. PCR relies on heat cycling to denature double stranded DNA, followed by primers which specify a region of amplification for a heat stable DNA polymerase which enzymatically assembles DNA from the base nucleotides. Heat cycling exponentially multiples the amplified region designated by the primers. Many current nucleic acid diagnostic methods rely on PCR to essentially amplify trace amounts of nucleic acids present for analysis. 1.3 Medical Diagnosis of Cancers and Infectious Disease Efficient and rapid genetic analysis of diseases is an important goal, since it can be used to provide more accurate diagnosis and effective treatment then traditional techniques [8]. Cancer has become prominent emerging disease over the past half century and is one of the top causes of death in the world [9]. Cancer is a broad class of various diseases with one commonality, unregulated growth of cells causing malignant tumors that assail other parts of the body causing them to malfunction and eventually fail [10]. Another commonality is the fact that growth of cells is regulated by genes, and mutations in these genes cause malignant growth. Traditional methods for diagnosis of cancer are from symptoms and/or through screening techniques followed by a biopsy sample analyzed by a pathologist and other medical tests such as X-rays, CT scans, endoscopies and blood tests [11]. Cancers are generally classified by the type of cell the tumor resembles. However, diagnosis is generally made only when the tumor is large enough to cause symptoms or be seen with radiological techniques. Early detection of cancers would be of great benefit to the patient, however current screening methods are not sensitive or efficient enough for early detection of cancers [12]. Also genetic level analysis of cancers would provide valuable information on the prognosis and effective treatments for cancers [12], however they are not widely used. Hence there is a need for sensitive, rapid and inexpensive molecular diagnostic methods for use in early detection and disease state monitoring of cancer at the genetic level [13]. Infectious diseases are an important area for development of an efficient and rapid molecular diagnostic test. During approximately 32 million people died due to infectious diseases [14] excluding HIV/AIDS related deaths, which is roughly ~0.5% of the world

16 5 population or approximately 1 in 200 individuals. Infectious diseases can be caused by many organisms, including viruses, parasites and pathogenic bacteria. The most common diagnostic for infectious disease is a traditional symptomatic approach, however many infectious diseases share common symptoms. For infections caused by pathogenic bacteria a symptomatic approach will not provide any information on antimicrobial resistance in the initial assessment of the patient. This hampers effective treatment and amplifies the emerging problem of antibiotic resistance in pathogenic bacteria due to misuse of antibiotics [15]. Another common diagnostic tool is phenotypic testing [16] which is based on microbial culture, and utilizes a growth media to amplify and visualize the presence of an infectious disease. However, this method is not applicable to all infectious diseases, because not all pathogens can be cultured. Phenotypic testing can be used to determine speciation of pathogenic bacteria since colonies grown can posses different visual characteristics, however this is not a concrete method. Also it can be used to determine antibiotic susceptibility, where various antibiotics are inserted into the growth media, and susceptibility is determined by local growth. Phenotypic testing can take long periods of time, anywhere from a day to a month for certain pathogenic bacteria, which limits the effectiveness of phenotypic testing as a diagnostic tool. Clinicians must provide effective treatment at the point of care within a rapid time frame. With current clinical methods they must make an educated guess, based upon a symptomatic approach without evidence of antimicrobial resistance or susceptibility. This leads to ineffective and/or inaccurate treatments being prescribed for infectious diseases, and the misuse of antibiotics contributing to the increasing antibiotic resistance of infectious diseases [15]. In theory, all infectious disease could be diagnosed by molecular diagnostic methods, however this is not the case since current molecular diagnostic methods are time consuming, require trained technicians and are expensive. Due to the varied nature of infectious diseases a point of care diagnostic test that can rapidly and inexpensively determine speciation and antibiotic resistance from numerous candidates will be a valuable asset in stemming infectious disease related deaths [17]. 1.4 Point-of-Care Biosensor Technology Biosensors are a relatively new class of medical diagnostic tool. Biosensors are essentially made up of three main components, first is a biological sample; which can include cell cultures, human (blood, urine, saliva and tissue), animal, food and environmental samples. Second, is the

17 6 transduction of a specific recognition element of target analyte within the sample of interest. The transducer method is based on optical, mass or electrochemical measurements. Third, is a readout method for the transduced signal which can either be optical and/or electronic [18]. Point of care biosensors have additional required criteria, including ease of use, rapid sample to answer timeframes and low cost. The main advantage of point of care optical based biosensors is the direct visual interpretation of results. However optical methods can suffer from the following drawbacks poor sensitivity, need for optically transparent samples/materials and expensive auxiliary equipment.. Examples of notable commercially available optically based point of care biosensors are simple lateral flow assays such as point of care pregnancy tests [20]. The main advantages of point of care electrochemical based biosensors are high levels of sensitivity, low cost instrumentation and ease of miniaturization. Some disadvantages of electrochemical based biosensors are requirement of additional instrumentation for readout, results of readout are not as easily interpreted and methods are not as well characterized or as developed as with optical systems. The most notable commercially available electrochemical biosensors are point of care glucose sensors [21]. Major issues for future development of point of care biosensors which will need to address speed, ease of use, cost, and multiplexing. 1.5 Scope of Thesis The scope of this work is to develop new and exciting methods for application to point of care diagnostics. First, sample handling of biological samples in infectious disease and cancer is important. We utilize electrical lysis in microfludic chambers and lyse mammalian [22] and bacterial cells [23] and detect the presence of nucleic acids electrochemically. Our goal in chapter 3 was to develop new simplified parallel-plate lysis chambers that can operate at lower voltages, and directly integrate them to our nanostructured microelectrode (NME) platform. With this new platform, we perform polymerase chain reaction (PCR) free electrochemical detection of raw bacterial lysates in buffer and urine [24]. In chapter 4 our goal was to investigate new low cost materials which are comparable to silicon for fabrication of NMEs. High material cost of silicon can be prohibitive to point of care diagnostic devices. We investigated several materials as alternative inexpensive substrates for

18 7 growth of NMEs for use in point of care electrochemical biosensors. We evaluated printed circuit boards (PCB), plastics and glass. With PCBs we found copper problematic as a base metal for growth of NME structures, since it would spontaneously etch when exposed to our traditional plating solutions. We were able to resolve these material issues, however we could not obtain comparable sensitivity to silicon counterparts. Plastics did not have base metal issues, however we found that plastics were too flexible to obtain reasonable lithographic resolution. This resulted in poor sensitivity compared to silicon based NMEs. Glass proved to be rigid enough for good lithographic resolution and had comparable sensitivity to silicon based sensors [25]. In chapter 5 our goal was to develop a new technique for inexpensive and efficient multiplexing that does not rely on resource heavy active electronics. A current goal for electrochemical biosensors is the ever increasing multiplexing of multiple targets. Serial connections are the most obvious method to increase multiplexing, however electrical output connections become ever increasingly difficult to perform, since n sensors requires n individual output connections. Parallel connections are more efficient reducing the minimum number of output connections to 2 n. However, traditional methods require active switching electronics on the surface of the sensor chip which drastically increases cost and complexity of the device. We developed a new solution based method that takes advantage of physical separation of liquid on the surface of electrochemical biosensors to create equivalent parallel connections to active electronic methods, which we call solution based electrochemical circuits (SCC). We show effective isolation of sensors connected in parallel and show that they are equivalent to serially connected sensors. In addition we accurately detected the species and antibiotic resistance of pathogenic bacteria in multiplexed bacterial lysates [26]. `

19 8 1.6 References 1. Berger, B. D. A brief history of medical diagnosis and the birth of the clinical laboratory. Medical Laboratory Observer (1999). 2. Blaufox, M. D. An ear to the chest: An illustrated history of the evolution of the stethoscope. 149p (Parthenon Pub. Group: London: 2002). 3. Rehman, S. U., Nietert, P. J., Cope, D. W. & Kilpatrick, A. O. What to wear today? Effect of doctor s attire on the trust and confidence of patients. The American Journal of Medicine 118, (2005). 4. Röntgen, W. On a new kind of rays. Science 3, (1896). 5. Watson, J. D. & Crick, F. H. C. Molecular Structure of Nucleic Acids. Nature 171, (1953). 6. Sanger, F., Nicklen, S. & Coulson, A. R. DNA sequencing with chain-terminating. Proceedings of the National Acadamey of Sciences 74, (1977). 7. Saiki, R. K., Scharf, S., Faloona, F., Mullis, K. B., Horn, G. T., Erlich, H. A., & Arnheim, N. Enzymatic amplification of beta-globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 230, (1985). 8. Grody, W. W. Molecular diagnostics: Techniques and applications for the clinical laboratory. 484p (Elsevier/Academic Press: 2010). 9. Siegel, R., Naishadham, D. & Jemal, A. Cancer Statistics, CA: A Cancer Journal for Clinicians 62, (2012). 10. Weinberg, R. A. The Biology of Cancer. (Garland Science: New York, 2007). 11. Nakamura, R. M. Cancer Diagnostics: Current and Future Trends. (Humana Press: Totowa, N.J, 2004). 12.Jorde, L. B., Carey, J.C., M. J. B. Medical Genetics. (Mosby/Elsevier: Philadelphia, 2010).

20 9 13. Offit, K. Personalized medicine: new genomics, old lessons. Human genetics 130, 3 14 (2011). 14. World Health Organization. World Health Statistics (2013). 15. Levy, S. B. & Marshall, B. Antibacterial resistance worldwide: causes, challenges and responses. Nature Medicine 10, S122 S129 (2004). 16. Bochner, B. R. Global phenotypic characterization of bacteria. FEMS Microbiology reviews 33, (2009). 17. Niemz, A., Ferguson, T. M. & Boyle, D. S. Point-of-care nucleic acid testing for infectious diseases. Trends in biotechnology 29, (2011). 18. Jonathan M. Cooper, A. E. G. C. Biosensors : a practical approach. 251p (Oxford University Press: New York, 2004). 19. Issadore, D. I. & Westervielt, R.M. Point-of-care diagnostics on a chip. (Springer: New York, 2013). 20. Williams, L. Home and Point-of-Care Pregnancy Tests : A Review of the Technology. Epidemiology 13, (2013). 21. Newman, J. D. & Turner, A. P. F. Home blood glucose biosensors: a commercial perspective. Biosensors & bioelectronics 20, (2005). 22. Vasilyeva, E., Lam, B., Fang, Z., Minden, M. D., Sargent, E. H., & Kelley, S. O. Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mrna capture. Angewandte Chemie (International ed. in English) 50, (2011). 23. Soleymani, L., Fang, Z., Lam, B., Bin, X., Vasilyeva, E., Ross, A. J., Sargent, E. H., & Kelley, S. O. Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS nano 5, (2011).

21 Lam, B., Fang, Z., Sargent, E. H. & Kelley, S. O. Polymerase Chain Reaction-Free, Sampleto-Answer Bacterial Detection in 30 Minutes with Integrated Cell Lysis. Analytical Chemistry 84, (2012). 25. Lam, B., Holmes, R. D., Das, J., Poudineh, M., Sargent, E. H., & Kelley, S. O. Optimized Templates for Bottom-Up Growth of High-Performance Integrated Biomolecular Detectors. Lab on a Chip 13, (2013) 26. Lam, B., Das, J., Holmes, R. D., Live, L., Sage, A., Sargent, E. H., & Kelley, S. O. Solutionbased circuits enable rapid and multiplexed pathogen detection. Nature Communications 4, 1 8 (2013).

22 11 2 Electrochemical Biosensor Methodology and Background Various types of biosensors are currently being developed and many are commercially available. Generally there are only two readout methods employed, optical or electrical. Biosensor developments contained herein are electrochemical based and therefore a brief overview of electrochemical techniques [1] and electrochemical biosensors follows. 2.1 Electrochemical Biosensors The field of electrochemistry was born in 1791 by Luigi Galvani when he established a connection between chemical reactions and electricity with his famous frog leg experiment. Galvani was a biologist and was one day performing an experiment on frog legs, and accidently touched an exposed nerve with a statically charged metal scalpel. He observed electrical discharges along with a kicking motion of the frog legs and realized there was a connection between electricity and life [2]. Sparked by Galvani s discoveries Alessandro Volta in 1800 invented the voltaic pile, the first electrical battery, one of the most important inventions in electrochemistry. One of the first electrochemical sensing methods measured ph utilizing a glass electrode which was developed in the early 1900s. The first commercial ph meter was developed by Arnold Beckman in This was followed by the development of the first electrochemical biosensor in 1962 by Leland Clark with the first glucose oxidase enzyme electrode [3]. The first commercial glucose meters were available by the 1970s and have become the gold standard of the biosensor field [4]. These developments have made electrochemical biosensors one of the most important fields in applied science. The main advantages of electrochemical biosensors are ease of miniaturization, low cost instrumentation, robustness, good detection limits, small sample volumes, and ability to work in turbid optically absorbing samples. The potential low cost of electrochemical biosensors combined with ease of miniaturization is the definitive advantage when used for point of care biosensors. The main drawbacks are direct visual observation of detection is usually not possible and multiplexing is less viable compared to optical methods [5]. Electrochemical based biosensors are invaluable medical diagnostic tools and are a capable method for detection of medically related analytes [6]. Many electrochemical biosensor

23 12 techniques have been developed to detect nucleic acids [7-12], proteins [13-16], and small molecules [17-19]. Studies have shown that electrochemical methods are robust and can accurately detect biomarkers in complex unpurified heterogeneous biological samples [18]. Electrochemical biosensors have been applied to many cancer [13, 20, 21] and infectious disease [22-24] biomarkers which have illustrated the utility of electrochemical biosensors for future medical diagnostic applications. Electrochemical techniques are the foundation of electrochemical biosensors [1], and a short review of common electrochemical techniques follows Electrochemical Techniques Electrochemical biosensors can be classified into three main types; potentiometric, impedimetric and amperometric. Potentiometric devices measure the charge/potential collected at a given sensor surface with respect to a reference when no current flows and provides information on ion activity. Impedimetric devices measure impedance change; generally change of resistance or capacitance, at the sensor surface. All devices developed in this work are amperometric which measure current generated at the sensor surface, usually in response to an applied potential. In general most amperometric electrochemical biosensors are three electrode systems comprised of a working (WE), auxiliary/counter (CE) and reference electrode (RE) (Figure 2.1). The WE can be considered the most important electrode because electrochemical reactions of interest occur on the surface. In terms of electrochemical biosensing these reactions can be the direct reduction/oxidation of a biological analyte of interest or indicative of a biomolecular recognition event such as the complementary binding of DNA or antibody/antigen binding events. The WE is fragile, since in many applications the surface is coated with a sensitive biological probe or enzyme which can be easily fouled or stripped if handled incorrectly.

24 13 Figure 2.1 Three electrode electrochemical cell The purpose of the CE is to complete the circuit to measure current and handle any variance potential that could damage the sensitive surface of the WE. The CE is generally made of an inert and strong material, which is usually platinum or carbon. The reference electrode is used to provide a stable potential reference point for which potential is applied against to drive redox reactions of interest at the surface of the WE. The most common reference electrode is silver/silver chloride (Ag/AgCl), since it is easy to construct, inexpensive and non-toxic. The potential of the electrode depends on the activity of chloride ions which is held constant by a saturated KCl or NaCl solution and is physically separated from the sample using a porous bridge while remaining in electrical contact. The potentiostat applies and measures signals to and between the electrodes. In most electrochemical techniques it applies a voltage to the working electrode with respect to the reference electrode, and measures current flowing from the working electrode. We will briefly overview the electrochemical techniques herein which include cyclic voltammetry (CV), differential pulse voltammetry (DPV), chronoamperometry (CA). Cyclic voltammetry is one of the most common used techniques, and involves the application of a linear sweep voltage to the WE with respect to the RE, as depicted in Figure 2.2A. In this example, the electro active species is Fe(CN) 3-6. The resultant current is measured and plotted versus the applied voltage as depicted in Figure 2.2B. At point (a) the potential is not sufficient to reduce Fe(CN) 3-6, the potential is forward scanned (negatively increased in this case), at (b) the potential is now sufficient to begin reducing Fe(CN) e- Fe(CN) 4-6 generating a small cathodic current. This cathodic current increases rapidly between (b-d) quickly consuming

25 14 Fe(CN) 3-6 in the diffusive region near the WE. The cathodic current then decreases (d-f) where all the Fe(CN) 3-6 is consumed in the diffusive region and current will be limited by transport of new Fe(CN) 3-6 into the diffusive region. The reverse scan is now performed (f-g) in this region the potential is still sufficient to reduce Fe(CN) 3-6. At point (h) the potential of the WE has decreased so that it can now oxidize the Fe(CN) 4-6 that was generated in the forward scan. The anodic current quickly increases between (h-j) quickly consuming and depleting Fe(CN) 4-6 in the diffusive region near the WE. The anodic current then decreases between (j-f) consuming the remaining Fe(CN) 4-6 in the diffusive layer. Figure 2.2 Cyclic Voltammetry. Demonstration of typical CV measurement utilizing K 3 Fe(CN) 6 [25] Cyclic voltammetry can provide many insights on the analytes present in the sample and help characterize the WE in the system of interest. Information about the concentration of electroactive moieties can be obtained from peak current and width when compared to reference samples. Cyclic voltammetry can also be used to determine the electroactive area of the WE and the ease of electron transport to the electroactive moieties,, allowing the effectiveness of different WEs to be compared. In addition, CV provides information on the electrochemical

26 15 reversibility of the reaction of interest, where irreversibility is easily determined through the disappearance of the reverse scan peak. Differential pulse voltammetry is a similar technique to CV, where a small pulse is superimposed onto a single linear forward scan, with no reverse scan as shown in Figure 2.3A. The current is measured just before each pulse (i a ) and just before the end (i b ) of each pulse and the resultant difference in current (i b -i a ) is plotted versus the linear sweep voltage. An example of DPV for the reduction of Ru(NH 3 ) 3+ 6 /Ru(NH 3 ) 2+ 6 is shown in red (Figure 2.3B) and the corresponding CV is shown in blue. DPV can be loosely thought of as the differential of the forward scan of the corresponding CV. Again the peak current will be proportional to the concentration of the electroactive moieties of interest, but we can no longer determine if the system is reversible. The main advantage of using DPV is the elimination of background charging currents, resulting in generally higher sensitivity and better peak resolutions then CV. Figure 2.3 Differential Pulse Voltammetry. Example of applied and measured DPV signals. In chronoamperometry (CA), the applied signals are similar, as depicted in Figure 2.4A. Initially the potential is held at a value for which no redox reaction occurs (E 1 ) then the potential is stepped to a value (E 2 ) at which the oxidation/reduction of the analyte of interest may occur. In CA the resultant current versus time is measured, an example is plotted in Figure 2.4B for short time periods, and can be modelled using the Cotrell equation,

27 16 (1) where n is the number of electrons, F is Faraday s constant, C is concentration, D is the diffusion constant and t is time. This technique can be used measure the electroactive area of your electrode and the diffusion constant of your electroactive species. Another common use of CA is the real-time monitoring of systems, where the potential is held at the redox potential of a electroactive moiety until steady state value is reached. Any addition or generation of that electroacitive moiety will cause a proportional response in current, where an increasing value would represent oxidation and decreasing reduction (Figure 2.4B). Subsequent addition or generation of that moiety will produce further response and a staircase like signal. Another common use of CA is to electroplate surfaces, which is commonly done by immersing surface in an metal salt plating solution, holding the surface above the reduction potential reducing the metal onto the surface. Figure 2.4 Chronoamperometry. (A) Typical applied signal (B) measured current Chronoamperometry is the most common electrochemical technique utilized since the results are easy to interpret and the technique requires minimal instrumentation to implement. CA is utilized by the most successful electrochemical biosensor, the blood glucose meter.

28 Glucose Biosensors Regardless of many advances in biosensor technologies, glucose biosensors still dominate the commercial biosensor market, accounting for approximately 85% of an estimated 5 billion dollar market [4]. One could consider the electrochemical glucose sensor the original biosensor, and it was first conceived by Clark and Lyons in 1962 [3]. Other technologies to measure glucose were developed during that time such as the Ames Reflectance Meter invented by Anton Clemens in 1971 [4]. The reflectance meter was not as readily adopted due to the fact that it was large and heavy ~1kg, expensive and required a prescription. The electrochemical glucose sensor has dominated the market since they have offered acceptable sensitivity & reproducibility, and can be manufactured in high volumes at low cost. The most common enzyme utilized for electrochemical glucose detection is glucose oxidase (GOx). This enzyme changes the redox state of glucose and produces products that can be detected electrochemically. The mechanism is generally as follows, (1) Glucose + O 2 gluconic acid + H 2 O 2 (1) O 2 + 4H + + 4e - 2H 2 O (2) H 2 O 2 O 2 + 2H + + 2e - (3) Glucose oxidation is performed by GOx and molecular oxygen producing gluconolactone and hydrogen peroxide. Originally Clark & Lyons monitored the consumption of O 2 over an oxygen electrode through a dialysis membrane (2), where a negative potential was applied to a platinum cathode. Further developments relied on the direct oxidation of hydrogen peroxide formation (3). First generation devices relied on the use of flavin group (FAD) as an oxygen co-substrate, which is reduced by glucose to FADH 2 (4). This is followed by the subsequent oxidation by O 2 cycling back to the oxidized form and hydrogen peroxide (5). GOx(FAD) + glucose GOx(FADH 2 ) + gluconolactone (4) GOx(FADH 2 ) + O 2 GOx(FAD) + H 2 O 2 (5)

29 18 The peroxide formation was measured directly utilizing (3) and simple CA. Further developments of glucose biosensors eliminated the need for molecular oxygen and introduced mediators that facilitate the transfer of electrons between GOx and the WE surface [16]. Most modern glucose meters utilize these developments combined with cheap screen printing techniques to create cheap disposable glucose sensing strips to be used as consumables. Combining consumable test strips with miniaturized cheap amperometric meters, which are usually given away for free, the modern glucose meter (Figure 2.5) can rapidly and accurately determine glucose concentration from a small droplet of blood. Figure 2.5 Freestyle glucose meter and sensor strips Glucose biosensors have become the gold standard of the biosensor field. However, the glucose sensor platform is not suited to molecular diagnostics, since it is an enzymatic approach with moderate sensitivity. Development of new electrochemical techniques and platforms with high sensitivities for point of care molecular diagnostics is an important goal. Nanostructuring is an effective method to improve sensitivities by enhancing surface properties of sensing WEs Nanostructured Microelectrode Electrochemical Biosensors A important goal in electrochemical biosensor development is to miniaturize the size of the WE to achieve better signal to noise ratios and lower detection limits [27]. Nanostructuring the surface of WEs can increase sensitivity due to increasing surface to volume ratios, and more favorable biorecognition element orientation with respect to binding target analytes [28].

30 19 However the extent to which the WE can be miniaturized reaches a limit. If the WE is too small such that the interactions between the surface of the WE and the desired target analytes are minimized, at low concentrations detection cannot occur within a reasonable time frame because of the lack of collisions/interactions with the WE surface. A past goal within the Kelley laboratories was to increase electrochemical assay sensitivities by utilizing nanostructured working electrodes. Initial work toward this goal, in the Kelley lab was to use templated gold nanowires as nanostructured microelectrodes [29]. The nanowires were fabricated by electroless deposition of gold onto a polycarbonate membrane (Figure 2.6) which was followed by reactive ion etching to expose the gold nanowires. Figure 2.6 Nanowire polycarbonate template fabrication. [29] A complementary electrocatalytic nucleic acid detection method was developed around the same period [30]. The Kelley group was able achieve good detection limits with this platform, down to 1pM of synthetic oligonucleotides. The major issue was the difficult fabrication process for nanowires which would be difficult to scale up for manufacturing. The Kelley lab wanted to develop a more manufacturable streamlined platform, which could rely on traditional lithographic techniques for fabrication.

31 20 One of the first efforts toward this goal was made by Yang et al [31]. The foundation of this platform was to utilize a rigid microfabricated template made from silicon for directed bottom up electroplating of NME. The fabrication method is outlined in Figure 2.7. Figure NME fabrication process (A) silicon base substrate passivated with SiO 2 followed by Au electrode structure and top SiO 2 passivation layer (B) small apertures etched into top passivating layer (C) nanostructured microelectrode electroplated within small apertures Starting with a base silicon wafer that is first passivated by a thick layer of thermal oxide (~2µm), standard gold electrode leads are patterned utilizing standard metal lift-off techniques. This is followed by the deposition of a top oxide passivating layer utilizing plasma enhanced chemical vapor deposition. Small apertures are etched above the ends of the gold electrode leads and a metal salt is electroplated into the apertures to fabricate NMEs. Numerous structures can be fabricated utilizing this method (Figure 2.8), by simply varying plating conditions, aperture size and/or plating solutions.

32 21 Figure 2.8 Electrode Morphologies. (A C) Many types of gold morphologies can be fabricated utilizing different plating potentials and (D) hierarchal structures can be fabricated by subsequent plating using different metals and/or plating conditions such as palladium We combined NMEs with the electrocatalyic nucleic acid detection method [30] outlined in Figure 2.9. Bare NMEs are first functionalized with target specific peptide nucleic acid (PNA) or deoxyribonucleic acid (DNA) probe molecule. PNA probes bind to complementary strands with higher affinity and selectivity than their DNA analogue mainly due to the uncharged backbone of PNA[32]. The electrocatalytic reporter pair consists of Ru(NH 3 ) 3+ 6 and Fe(CN) 3-6. Ru(NH 3 ) 3+ 6 is electrostatically attracted to any nucleic acids present on the surface of the NME. Hence, Ru(NH 3 ) 3+ 6 will be strongly attracted to the NME surface if the complementary nucleic acid target is bound to the PNA probe. Using DPV, the surface of the NME is scanned over a specified potential window before (Figure 2.9B) and after sample incubation (Figure 2.9C). Reduction of Ru(NH 3 ) 3+ 6 to Ru(NH 3 ) 2+ 6 occurs near the surface when we hit the reduction potential. Fe(CN) 3-6 present at higher concentration is reduced to Fe(CN) 4-6 by oxidizing Ru(NH 3 ) 2+ 6 back to Ru(NH 3 ) 3+ 6 near the surface, generating an electrocatalytic current. Typical positive and negative DPV are shown at the end of Figure 2.9B & D.

33 22 Figure Electrocatalytic nucleic acid assay. (A) functionalization of NME with probe molecule followed by (B) background scan in electrocatalyic solution. (C) After hybridization with target sample and subsequent (D) after hybridization scan. Shown are typical negative and positive scan results. In the Kelley laboratories we have shown the NME platform to be highly sensitive [33] with detection limits for synthetic oligonucleotides down to 10 am. In addition by tuning the degree of nanostructing on the surface of the sensors, we can tune their dynamic sensing range (Figure 2.10A) over six orders of magnitude; from 10 am for a finely nanostructured sensor, 10 fm for a moderately nanostructured sensor and 100fM for an unstructured WE surface. Tuning of surface nanostructure is performed by simply adjusting plating parameters and solution properties. We have also shown that the NME platform is highly selective and able to determine specific prostate cancer gene fusions (Figure 2.10B) [34] in different prostate cancer cell lines and patient samples.

34 23 Figure 2.10 NME sensitivity and specificity. (A) Tuning dynamic range of NME sensors by varying levels of surface nanotexture; highly nanostructured (blue) moderately nanostructured (red) smooth (black) [33] (B) selective detection of prostate cancer fusions in cell lines and patient samples [34] 2.2 Platform Development Goals At the time I joined the Kelley laboratory, our goal was to drive towards point-of-care devices utilizing the NME platform. This would require the streamlined analysis of real world samples, therefore sample processing and integration became one of the major goals. The chip-based approach senses intercellular molecules (mrna targets) within mammalian or bacterial cells and therefore, an effective method to lyse and release these intercellular contents was needed. We first utilized a simple microchannel electrical lysis method for release of intercellular mrna targets, and integrated this approach with our NME platform for rapid detection of CML and pathogenic bacteria (Supplementary Information 8.1). In chapter 3 our first goal was to develop a simplified parallel plate lysis method which could operate at lower voltages and could be easily integrated with our NME platform. In chapter 4 our second goal was to reduce the cost of fabricated devices and explore new material paradigms since silicon is a rather expensive substrate. We explored new more manufacturable materials for bottom up template fabrication of NMEs. We tried various materials and settled on glass as an optimal substrate.

35 24 In chapter 5 our third goal was to develop highly multiplexed electrochemical biosensors, which is an important goal of diagnostics so that large panels of biomarkers can be analyzed. An issue for increased multiplexing in electrochemical systems is if n sensors need to be addressed, n electrical output contacts are needed, which becomes increasingly unmanageable for high levels of multiplexing. To reduce output contacts using existing approaches requires active switching electronics on each chip, driving up cost and complexity. We developed a passive solution based electrochemical method to replace resource heavy active electronics.

36 References 1. Bard, A. J. Electrochemical methods: Fundamentals and applications. (Wiley: New York, 2001). 2. Ostwald, W. Electrochemistry: History and theory. (Amerind Pub. Co.: New Delhi, 1980). 3. Clark, L. C. & Lyons., C. Electrode systems for continuous monitoring in cardiovascular surgery. Annals of the New York Academy of Sciences 102, (1962). 4. Newman, J. D. & Turner, A. P. F. Home blood glucose biosensors: a commercial perspective. Biosensors & bioelectronics 20, (2005). 5. Zhang, X., Ju, H. & Wang, J. Electrochemical sensors, biosensors, and their biomedical applications. (Academic Press: Amsterdam, 2008). 6. Ronkainen, N. J., Halsall, H. B. & Heineman, W. R. Electrochemical biosensors. Chemical Society Reviews 39, (2010). 7. Palec, E. & Bartos, M. Electrochemistry of Nucleic Acids. Chemical Reviews, 112 (6), (2012). 8. Palec ek, E. Fifty Years of Nucleic Acid Electrochemistry. Electroanalysis 21, (2009). 9. Wang, J. Electrochemical nucleic acid biosensors. Analytica Chimica Acta 469, (2002). 10. Drummond, T. G., Hill, M. G. & Barton, J. K. Electrochemical DNA sensors. Nat. Biotechnol. 21, (2003). 11. Venkatesan, B. M. & Bashir, R. Nanopore sensors for nucleic acid analysis. Nature nanotechnology 6, (2011). 12. Wanunu, M., Dadosh, T., Ray, V., Jin, J., McReynolds, L. & Drndic, M.. Rapid electronic detection of probe-specific micrornas using thin nanopore sensors. Nature nanotechnology 5, (2010).

37 Zheng, G., Patolsky, F., Cui, Y., Wang, W. U. & Lieber, C. M. Multiplexed electrical detection of cancer markers with nanowire sensor arrays. Nat. Biotechnol. 23, (2005). 14. Das, J. & Kelley, S. O. Protein detection using arrayed microsensor chips: tuning sensor footprint to achieve ultrasensitive readout of CA-125 in serum and whole blood. Anal. Chem. 83, (2011). 15. Malhotra, R., Patel, V., Vaqué, J. P., Gutkind, J. S. & Rusling, J. F. Ultrasensitive electrochemical immunosensor for oral cancer biomarker IL-6 using carbon nanotube forest electrodes and multilabel amplification. Analytical chemistry 82, (2010). 16. Tang, D., Yuan, R. & Chai, Y. Ultrasensitive electrochemical immunosensor for clinical immunoassay using thionine-doped magnetic gold nanospheres as labels and horseradish peroxidase as enhancer. Anal. Chem. 80, (2008). 17. Zuo, X., Xiao, Y. & Plaxco, K. W. High specificity, electrochemical sandwich assays based on single aptamer sequences and suitable for the direct detection of small-molecule targets in blood and other complex matrices. Journal of the American Chemical Society 131, (2009). 18. Swensen, J. S. Continuous, real-time monitoring of cocaine in undiluted blood serum via a microfluidic, electrochemical aptamer-based sensor. Journal of the American Chemical Society 131, (2009). 19. Liu, H., Xiang, Y., Lu, Y. & Crooks, R. M. Aptamer-based origami paper analytical device for electrochemical detection of adenosine. Angew. Chem. Int. Ed. 51, (2012). 20. Feng, L., Chen, Y., Ren, J. & Qu, X. A graphene functionalized electrochemical aptasensor for selective label-free detection of cancer cells. Biomaterials 32, (2011).

38 Malhotra, R., Patel, V., Vaqué, J. P., Gutkind, J. S. & Rusling, J. F. Ultrasensitive electrochemical immunosensor for oral cancer biomarker IL-6 using carbon nanotube forest electrodes and multilabel amplification. Anal. Chem. 82, (2010). 22. Mannoor, M. S., Zhang, S., Link, A. J. & McAlpine, M. C. Electrical detection of pathogenic bacteria via immobilized antimicrobial peptides. Proceedings of the National Academy of Sciences of the United States of America 107, (2010). 23. Liao, J.C., Mastali, M., Li, Y., Gau, V., Suchard, M.A., Babbitt, J., Gornbein, J., Landaw, E.M., McCabe, E.R., Churchill, B.M., Haake, D.A..Development of an advanced electrochemical DNA biosensor for bacterial pathogen detection. The Journal of molecular diagnostics 9, (2007). 24. Mannoor, M. S., Zhang, S., Link, A. J. & McAlpine, M. C. Electrical detection of pathogenic bacteria via immobilized antimicrobial peptides. Proc. Natl Acad. Sci. USA 107, (2010). 25. Kissinger, P. T., Lafayette, W. & Heineman, W. R. Cyclic voltammetry. Journal of Chemical Education 60, (1983). 26. Wang, J. Electrochemical glucose biosensors. Chemical reviews 108, (2008). 27. Watkins, J. J., Zhang, B. & White, H. S. Chemistry at the Nanometer Scale Electrochemistry at Nanometer-Scaled Electrodes. Journal of Chemical Education 82, 712 (2005). 28. Bin, X., Sargent, E. H. & Kelley, S. O. Nanostructuring of Sensors Determines the Efficiency of Biomolecular Capture. Analytical Chemistry 82, (2010). 29. Gasparac, R. et al. Ultrasensitive electrocatalytic DNA detection at two- and threedimensional nanoelectrodes. Journal of the American Chemical Society 126, (2004). 30. Lapierre, M. a, O Keefe, M., Taft, B. J. & Kelley, S. O. Electrocatalytic detection of pathogenic DNA sequences and antibiotic resistance markers. Analytical chemistry 75, (2003).

39 Yang, H.; Hui, A.; Pampalakis, G.; Soleymani, L.; Liu, F.-F.; Sargent, E. H.; Kelley, S. O. Direct, electronic microrna detection for the rapid determination of differential expression profiles. Angew. Chem. Int. Ed. 48, (2009). 32. Nielsen, P. E. Peptide Nucleic Acid. A Molecule with Two Identities. Accounts of Chemical Research 32, (1999). 33. Soleymani, L., Fang, Z., Sargent, E. H. & Kelley, S. O. Programming the detection limits of biosensors through controlled nanostructuring. Nature nanotechnology 4, (2009). 34. Fang, Z., Soleymani, L., Pampalakis, G., Yoshimoto, M., Squire, J.A., Sargent, E.H., Kelley, S.O. Direct Profiling of Cancer Biomarkers in Tumor Tissue Using a Multiplexed Nanostructured Microelectrode Integrated Circuit. ACS nano 3, (2009).

40 29 3 Polymerase Chain Reaction-Free, Sample-to- Answer Bacterial Detection in 30 Minutes with Integrated Cell Lysis Our goal was to perform efficient and simple processing for direct detection of cancer and pathogenic bacteria with our nanostructured microelectrode platform. Prior to my first manuscript we utilized a microfluidic electrical lysis platform for lysis of bacteria [22] and cancer cells [23] additional details of the microfluidic approach can be found in the supplementary information. We followed this work by developing a simple parallel plate electrical lysis chamber and directly integrated it with our NME detection platform. Disclosure of work within this manuscript; B.L., E.H.S and S.O.K developed parallel plate lysis chambers, B.L. fabricated, tested and characterized parallel plate lysis chambers, B.L. and Z.F. performed bacterial lysate detection. B.L., E.H.S. and S.O.K. wrote the manuscript. Lam, B., Fang, Z., Sargent, E. H. & Kelley, S. O. Polymerase Chain Reaction-Free, Sample-to- Answer Bacterial Detection in 30 Minutes with Integrated Cell Lysis. Analytical Chemistry 84, (2012). 3.1 Abstract An important goal for improved diagnosis and management of infectious disease is the development of rapid and accurate technologies for the decentralized detection of bacterial pathogens. Most current clinical methods that identify bacterial strains require time-consuming culture of the sample or procedures involving the polymerase chain reaction [1-3]. Neither of these approaches has enabled testing at the point-of-need because of the requirement for skilled technicians and laboratory facilities. Here, we demonstrate the performance of an effective, integrated platform for the rapid detection of bacteria that combines a universal bacterial lysis approach and a sensitive nanostructured electrochemical biosensor. The lysis is rapid, is effective at releasing intercellular RNA from bacterial samples, and can be performed in a simple, costeffective device integrated with an analysis chip. The platform was directly challenged with these unpurified lysates in buffer and urine. We successfully detected the presence of bacteria with high sensitivity and specificity and achieved a sample-to-answer turnaround time of 30 min. We have met the clinically relevant detection limit of 1 cfu/μl, indicating that uncultured

41 30 samples can be analyzed. This advance will greatly reduce time to successful detection from days to minutes. 3.2 Introduction The effective management of infectious disease caused by bacterial pathogens is a major problem in clinical medicine that is hampered by the lack of rapid diagnostic methods [1-4] Approaches currently used for correct diagnosis of infectious bacterial strains include phenotypic testing and assays that rely on the polymerase chain reaction (PCR) [3,5,6]. Many methods require a time-consuming culture step that takes days to weeks depending on the strain of bacteria, and phenotypic testing to confirm antibiotic resistance can double the diagnosis time. To speed analysis, PCR may also be performed on cultured samples or in some cases uncultured samples; however, this approach typically requires stringent purification of nucleic acids. The delays in the availability of diagnostic information limits the effectiveness of treatment. Hence, there is need for a rapid platform that can classify bacterial species. A great deal of effort has gone into the development of point-of-need methods to meet the challenge of rapid bacterial identification;[7-10] most of the methods developed rely on PCR and face inherent limitations because of the requirement for enzymatic components and thermal control. In addition, methods based on surface plasmon resonance,[11-13] quartz crystal microbalance,[14,15] and fluorescence[16,17] have been reported with good detection limits. However, many of these are immunological[11,12,14,16] and are ineffective at providing genetic-level information required for strain typing. Furthermore, these methods can require labeled markers[13,15] and additional optical[11-13, 16, 17] and/or fluid handling systems,[7-10, 12, 13, 15] which adds to their complexity, cost, and lack of applicability to point-of-care testing. Work in our laboratories has focused on developing an electrochemical strategy that combines ultrasensitive detection, straightforward sample processing, and inexpensive components that can be integrated into a cost-effective, user-friendly device. Our detection platform combines an electrochemical reporter system and nanostructured microelectrodes (NMEs) (Figure 3.1A,B) to detect specific nucleic acid sequences that hybridize to probe molecules immobilized on the sensors. We have previously shown that the NME platform is highly sensitive, with a tunable degree of sensitivity,[18, 19] and highly selective.[20, 21] Moreover, it is multiplexed and scalable, with straightforward photolithography used for fabrication that is highly versatile.

42 31 Figure Bacterial detection sensors. (A) The NME platform consisting of Si chip with patterned Au working, reference, and auxiliary electrodes. The working electrode surface is passivated with SiO 2, and 5 μm apertures are etched at the tip of each electrode. NMEs are electroplated within each aperture, with a typical size of 100 μm. (B) Electrochemical detection scheme for nucleic acids utilizing Ru 3+ and Fe 3+ electrocatalytic reporter pair. While prior efforts to exploit this platform for RNA detection showed that very high levels of performance could be achieved both with bacterial and mammalian targets,[22, 23] integrated sample processing, an essential feature for a point-of-care diagnostic device, had not yet been addressed. We therefore explored a processing approach that would be complementary to our electronic readout strategy: electrical cell lysis. Alternative methods that could be used for this purpose include chemical, physical,[24, 25] and thermal lysis methods.[26] However, the addition of chemical agents, complicated device geometries, or thermal elements are undesirable given that they can introduce interfering agents or increase the complexity of the device. Electrical lysis of bacterial cells has been well studied in the past.[27-30] The major drawback is that high electric field requirements, greater than 10 kv/cm, are required to lyse bacterial cells, which has limited its use for inline sensing. Prior work in our laboratories utilized microfludic

43 32 lysis chambers[31, 32] which take advantage of geometrical field effects to lower applied voltages. However, voltage requirements were still high (1000 V). In addition, this and other systems[33] require fluidics that can only analyze small sample volumes and increase processing times. To address these issues, we hypothesized that, by assembling a chamber composed of two conductive gold electrodes with a very thin spacer (500 μm), we could lyse bacteria introduced to the electrodes with an applied potential. If this type of sample processing module was coupled with a NME chip (Figure 3.2A), it could be used to achieve rapid sample-to-answer bacterial detection with minimal intervention by the user (Figure 3.2B). The workflow used here involved: (1) a solution being introduced into the chamber with a syringe, (2) lysis being induced with an applied field, (3) the sample being moved to the chip with an injection of air, (4) mixing with reporter groups, and (5) readout. This workflow can be completed in less than 30 min and permits bacterial identification and classification.

44 33 Figure Integrated sensing system. (A) Schematic of cartridge integrating lysis chamber, NME chip, and connector to analyzer. (B) Overview of detection scheme; injection, lysis, delivery, and readout in 30 min. (C) Typical differential pulse voltammograms of positive (left) and negative (right) samples where the dotted line is the background and the solid line is the readout.

45 Methods and Materials Chip Fabrication Detection chips were fabricated using 6 in. thin silicon wafers passivated with a thick thermally grown silicon oxide layer. First, a positive photoresist was patterned to the desired electrical contact and lead structure using standard photolithographic methods. Subsequently, a 500 nm gold layer was deposited using electron-beam assisted gold evaporation, and a standard lift off process was used to expose the desired contact and lead structure. Next, a second layer of 500 nm silicon dioxide was deposited to passivate the lead structure using chemical vapor deposition. Finally, 5 μm apertures were etched into the second passivating silicon dioxide layer, exposing the gold layer at the end of each lead structure. Nanostructured Microelectrode (NME) Fabrication Chips were cleaned by sonicating in acetone for 1 min and rinsing with isopropanol and deionized water for 30 s. NMEs were electroplated using a standard 3 electrode system featuring a Ag/AgCl reference, platinum auxiliary electrode and the 5 μm gold aperture as the working electrode. An electroplating solution of 20 mm HAuCl 4 in 0.5 M HCl was used. The substructures of NMEs were plated by holding each electrode at 0 mv for 250 s. Finally, a nanostructured overlayer was plated by holding the electrode at 700 mv for 10 s. Synthesis and Purification of Peptide Nucleic Acid Probes PNA probes were synthesized in house using a Protein Technologies Prelude peptide synthesizer. The following probe sequences specific to the rpoβ mrna were utilized for detection of unpurified lysates: NH 2 -Cys-Gly-Asp-ATC TGC TCT GTG GTG TAG TT-Asp- CONH 2 (E. coli) and NH 2 -Cys-Gly-Asp-AAG TAA GAC ATT GAT GCA AT-Asp-CONH 2 (S. saprophyticus). All probes were stringently purified by reverse phase high performance liquid chromatography. Probe sequences were quantified by measuring absorbance at 260 nm, and excitation coefficients were obtained from Modification of NMEs with PNA probes

46 35 A solution of 1 μm purified thiolated PNA probe in 25 mm NaCl was deposited onto the surface of an NME chip in a dark humidity chamber overnight at room temperature. A dam constructed from adhesive silicone spacers was used to deposit two different probes on each NME chip. Bacterial Samples Escherichia coli was obtained from Invitrogen ( ). Staphylococcus saprophyticus, methicillin-resistant Staphylococcus aureus, and methicillin-susceptible Staphylococcus aureuswas obtained from ATCC (ATCC 15305, BAA-1720, 29213). All strains were grown in the appropriate growth media in an incubating shaker at 37 C. After growth to the desired population, the growth media was replaced with 1 PBS. Lysis Chamber Fabrication and Operation Lysis chambers were fabricated using adhesive silicone hybridization spacers ( mm) obtained from Grace Biolabs and gold coated slides (25 25 mm) obtained from EMF Corporation. Chambers were constructed by first cutting a narrow channel 1 mm wide into the spacer, which was then sandwiched between two gold slides. To lyse the bacterial samples, a 200 μl suspension was loaded into the chamber using a syringe and 100 V, 10 ms DC pulses were applied to the sample at a frequency of 1 Hz for 20 s. Hybridization Protocol and Electrochemical Measurements Electrochemical measurements were made using a PalmSens EmStat embedded potentiostat. After modification of NMEs with PNA probes, a background signal was scanned in electrocatalyic buffer containing 10 μm Ru(NH 3 ) 3+ 6 and 1 mm Fe(CN) 3 6 in 0.1 PBS. Immediately after lysis, NMEs were incubated with unpurified lysates for 20 min at 37 C. After hybridization, chips were washed twice in 0.1 PBS. We subsequently scanned the hybridization signal after incubation in the same electrocatalytic buffer. Reverse-Transcriptase Polymerase Chain Reaction Primer sequences specific to a 185 bp region of the E. coli rpoβ mrna were synthesized. A Qiagen one-step RT-PCR kit (210210) was used to perform RT-PCR on lysates. After lysis, samples were centrifuged at rpm to remove intact bacterial cells that would generate a

47 36 positive signal. RT-PCR was then performed on the supernatant. The products were visualized using agarose gel electrophoresis and ethidium bromide fluorescent stain. Flow Cytometry Flow cytometry measurements were made utilizing a BD FACS Canto Instrument. After lysis, samples were incubated in propidium iodide in the dark at room temperature at a concentration of 25 μg/ml for 30 min before injection into the flow cytometer. Counts versus fluorescence intensity measurements were made in the red channel of the flow cytometer. 3.4 Results and Discussion To validate our lysis approach, we examined samples of two model organisms, Escherichia coli(ec) and Staphylococcus saprophyticus (SS). Bacterial samples suspended in buffered solution were introduced into the lysis chamber and lysed with varying voltages (0 100 V) and pulse widths (0 10 ms). A small amount of bubbling was observed, but the escape of these bubbles could be controlled by minimizing the width of the exit port on the lysis chamber. To assess lysis efficiency, we first looked at cell viability after lysis by monitoring growth on agar plates. With applied voltages as low as 2 V, all of the bacteria in processed samples were killed (data not shown). This loss in viability, however, cannot be used as an unequivocal test for cell lysis, as it does not indicate whether the bacterial cell walls were compromised prior to cell death. Therefore, to confirm that the applied electrical fields did cause irreversible cell rupture, we analyzed propidium iodide (PI) uptake using flow cytometry. PI fluoresces only when intercalated with DNA and does not cross uncompromised cell walls and, therefore, can be used as an indicator of cell lysis. After incubation in PI, samples were analyzed using flow cytometry and histograms of counts versus fluorescence intensity were plotted versus different pulse widths (Figure 3.3A) and applied voltages (Figure 3.3B).

48 37 Figure Characterization of electrically lysed bacterial solutions. (A) Flow cytometry histograms of propidium iodide uptake versus pulse width (100 V, 1 Hz, 20 s) collected with E. coli. Increasing the pulse duration decreases the number of unlysed cells. (B) Flow cytometry measurements of S. saprophyticus lysed at different voltages (10 ms, 1 Hz, 20 s) showing effective lysis down to 5 V. (C) RT-PCR measurements on E. coli. The PCR targeted a 185 bp region within the rpob mrna. Pulse durations were 1 ms, 5 ms, and 10 ms. A negative (no applied potential) and positive control (isopropanol-based lysis) were also run. Relative PCR efficiency was established by comparison with a postive control sample that was lysed with isopropanol (lane next to DNA ladder). (D) Flow cytometry measurements from lysis of E. coli, S. saprophyticus, MRSA, and MSSA. The red trace represents the unlysed control, and blue is the lysed sample (100 V, 10 ms, 1 Hz, 20 s). Interestingly, when the lysis of SS was monitored, it was observed that larger voltages were required to trigger PI uptake relative to those needed to cause cell death as observed on a culture plate. This indicates that cellular death alone is not proof of cellular lysis. Another interesting observation that emerged from these studies was that lower voltages caused PI uptake in EC

49 38 relative to SS, indicating that voltages must be tailored for gram-negative bacteria versus grampositive bacteria and that tailored pulse structures could potentially be used for selective lysis. We also explored the feasibility of lysing Staphylococcus aureus (SA) and methicillin-resistant SA and observed successful cell rupture for these organisms (Figure 3.3C). These results verify that the approach is generally applicable to bacterial organisms. To confirm that the PI uptake monitored in the experiments described above corresponded to the release of nucleic acids, we used the reverse-transcriptase polymerase chain reaction (RT-PCR) to confirm the release of intercellular RNA targets from EC. Shown in Figure 3.3D are RT-PCR measurements that were performed on the supernatant of lysed and unlysed samples in Figure3.3B. A positive control lysed with isopropanol was used for the calculation of relative PCR efficiency. The 180 bp RT-PCR products were visualized on an agarose gel where the correct primer specific products were verified. The intensity of the product bands observed was directly proportional to the pulse width and correlated well with the amount of PI uptake observed. To validate that this sample processing approach could be used with our NME detection platform, we directly challenged our sensors with unpurified lysates generated using our lysis chamber. Our NME detectors are fabricated on the surface of silicon wafers using traditional photolithographic methods[19] (Figure 3.1A). This NME sensor chip includes 20 working electrodes and on-board auxiliary and reference electrodes. Gold NMEs are electroplated into 5 μm apertures at the surface of each working electrode using a gold salt plating solution. The size and morphology of structures can be controlled through varying applied voltage and plating solution as previously described.[19] NME structures used for this study were 100 μm in diameter. Detection of nucleic acids was achieved using an electrocatalytic method developed in our laboratory.[34, 35] The method is depicted in Figure 3.1B, where the bare NMEs are first functionalized with target specific peptide nucleic acid (PNA) probe molecule. PNA probes, possessing a neutral backbone, bind to complementary strands with higher affinity and selectivity than their DNA analogues.[36] The electrocatalytic reporter pair consists of Ru(NH 3 ) 3+ 6 and Fe(CN) 3 6. Ru(NH 3 ) 3+ 6 is electrostatically attracted to anionic nucleic acids that accumulate on the surface of the NME. Ru(NH 3 ) 3+ 6 is therefore accumulated at the NME surface

50 39 if the complementary nucleic acid target is bound to the PNA probe. Using differential pulse voltammetry (DPV), the surface of the NME is scanned over a specified potential window before and after hybridization. Reduction of Ru(NH 3 ) 3+ 6 to Ru(NH 3 ) 2+ 6 occurs near the surface when the Ru(III) reduction potential is reached. Fe(CN) 3 6 then oxidizes Ru(NH 3 ) 2+ 6 back to Ru(NH 3 ) 3+ 6, generating an electrocatalytic current. Typical positive and negative DPV are shown in Figure 3.2C. The use of this method with unpurified lysates of EC and SS generated using electrical lysis is demonstrated in Figure 3.4A,B. NME sensors were modified with probes corresponding to the sequence of the RNA polymerase β mrna found in EC or SS. The lysed bacteria were introduced, and electrochemical signals were obtained within 30 min. Typical background subtracted DPV measurements obtained at NME sensors are shown in Figure 3.4C. Evaluation of limits of detection (Figure 3.4B) verify that this approach is successful with as few as 1 bacterial cell per microliter, a concentration that corresponds to the levels of bacteria found in many types of clinical samples. A limited dynamic range was explored in this study, but if analysis of a larger range of concentrations was desired, prior work on the use of sensor nanostructuring[19] and size[20] could be leveraged to widen dynamic range.

51 40 Figure Direct bacterial detection in unpurified lysates (A) Representative backgroundsubtracted electrochemical differential pulse voltammograms used for study of sensitivity and specificity. The data shown was collected with the SS probe directly challenged with the corresponding unpurified lysates. (B) Background-subtracted peak currents of sensors challenged with unpurified lysates demonstrating sensitivity and specificity. Values shown represent averages of >6 trials; coefficient of variation was <20%. (C) Direct detection of E. coli and S. saprophyticus in urine samples. Sensors were challenged directly with unpurified lysates of spiked urine samples for 30 min prior to electrochemical analysis. A control probe was used in each trial to assess background signals. A current value was collected for each trial and was plotted individually. (D) Real-time analysis of a 100 cfu/ul E.coli lysate spiked with the electrocatalytic reporter groups. A differential pulse voltammogram was measured at each time point for both complementary and noncomplementary sensors, and peak currents were plotted as

52 41 a function of time. (Bacterial lysis and sample preparation: Brian Lam, Electrochemical measurements: Zhichao Fang) To validate applicability of our integrated platform to samples resembling those relevant for clinical analysis, we challenged it with samples of urine spiked with both EC and SS. This analysis simulates real-world urinary tract infections where the relevant threshold is 100 cfu/μl.[37] After crude urine samples were lysed, they were directly applied to NME sensors specific to EC and SS. Successful detection of both EC and SS was achieved even in the presence of this complex biological background (Figure 3.4C). The adaptation of this approach to real-time detection was investigated, with signals being collected during hybridization of NME sensors with an unpurified lysate. This analysis was done in one-pot, with reporter groups present during hybridization. Specific detection of EC could be achieved using this approach, with very rapid readout achieved within minutes at a concentration of this pathogen that corresponds to its levels in samples collected from patients with a urinary tract infection (Figure 3.4D). This indicates that a positive result could be obtained from this type of sample within 2 min, a significant improvement over the culture-based methods typically employed for this type of analysis. 3.5 Conclusions The advances reported here demonstrate the first PCR-free, chip-based sensing system to provide sample-to-answer sensing of bacterial pathogens at clinically relevant levels. A simple lysis chamber was used to trigger electrical rupture of bacteria, and these crude lysates, generated in buffer or urine, were directly analyzed with an ultrasensitive microchip featuring nanostructured microsensors. Real-time analysis is also enabled by the robust sensors that are resistant to fouling by cellular contents.

53 References 1. Greenwood, D. Antimicrobial Susceptibility Tests and Their Clinical Relevance. Journal of Infectious Disease 144, (1981). 2. Optimal DNA Isolation Method for Detection of Bacteria in Clinical Specimens by Broad- Range PCR. 40, (2002). 3. Tang, Y. W. & Stratton, C. W. Advanced techniques in diagnostic microbiology. (Springer: New York: 2006). 4. Levy, S. B. & Marshall, B. Antibacterial resistance worldwide: causes, challenges and responses. Nature medicine 10, S122 9 (2004). 5. Institute., C. and L. S. CLSI document M02-A10. (Clinical and Laboratories Standards Institute: Pennsylvania: 2009). 6. Institute, C. and L. S. CLSI document M100-S21. (Clinical and Laboratories Standards Institute: Pennsylvania: 2011). 7. Miniature on-chip detection of unpurified methicillin-resistant Staphylococcus aureus (MRSA) DNA using real-time PCR. Journal of biotechnology 146, 93 9 (2010). 8. Self-contained, fully integrated biochip for sample preparation, polymerase chain reaction amplification, and DNA microarray detection. Analytical chemistry 76, (2004). 9. Shen, F. et al. Nanoliter Multiplex PCR Arrays on a SlipChip. Lab on a Chip 82, (2010). 10. A DNA biochip for on-the-spot multiplexed pathogen identification. Nucleic acids research 34, e118 (2006). 11. Surface plasmon resonance immunosensor for the detection of Salmonella typhimurium. Biosensors & bioelectronics 19, (2004). 12. A mixed self-assembled monolayer-based surface plasmon immunosensor for detection of E. coli O157:H7. Biosensors & bioelectronics 21, (2006).

54 Colloidal Au-Enhanced Surface Plasmon Resonance for Ultrasensitive Detection of DNA Hybridization (2000). 14. Su, X.-L. & Li, Y. A self-assembled monolayer-based piezoelectric immunosensor for rapid detection of Escherichia coli O157:H7. Biosensors and Bioelectronics 19, (2004). 15. Mao, X., Yang, L., Su, X.-L. & Li, Y. A nanoparticle amplification based quartz crystal microbalance DNA sensor for detection of Escherichia coli O157:H7. Biosensors & bioelectronics 21, (2006). 16. Ko, S. & Grant, S. a A novel FRET-based optical fiber biosensor for rapid detection of Salmonella typhimurium. Biosensors & bioelectronics 21, (2006). 17. Edgar, R. et al. High-sensitivity bacterial detection using biotin-tagged phage and quantumdot nanocomplexes. Proceedings of the National Academy of Sciences of the United States of America 103, (2006). 18. Bin, X., Sargent, E. H. & Kelley, S. O. Letters to Analytical Chemistry Nanostructuring of Sensors Determines the Efficiency of Biomolecular Capture. 82, (2010). 19. Soleymani, L., Fang, Z., Sargent, E. H. & Kelley, S. O. Programming the detection limits of biosensors through controlled nanostructuring. Nature nanotechnology 4, (2009). 20. Bin, X., Sargent, E. H. & Kelley, S. O. Nanostructuring of sensors determines the efficiency of biomolecular capture. Anal. Chem. 82, (2010). 21. Yang, H.; Hui, A.; Pampalakis, G.; Soleymani, L.; Liu, F.-F.; Sargent, E. H.; Kelley, S. O. Direct, electronic microrna detection for the rapid determination of differential expression profiles. Angew. Chem. Int. Ed. 48, (2009). 22. Soleymani, L. et al. Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS nano 5, (2011).

55 Vasilyeva, E. et al. Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mrna capture. Angewandtbe Chemie (International ed. in English) 50, (2011). 24. Huh, Y. S. et al. Microfluidic cell disruption system employing a magnetically actuated diaphragm. Electrophoresis 28, (2007). 25. Di Carlo, D., Jeong, K.-H. & Lee, L. P. Reagentless mechanical cell lysis by nanoscale barbs in microchannels for sample preparation. Lab on a chip 3, (2003). 26. Privorotskaya, N. et al. Rapid thermal lysis of cells using silicon-diamond microcantilever heaters. Lab on a chip 10, (2010). 27. Grahl, T. & Märkl, H. Killing of microorganisms by pulsed electric fields. Applied microbiology and biotechnology 45, (1996). 28. Hülsheger, H., Potel, J. & Niemann, E. G. Electric field effects on bacteria and yeast cells. Radiation and environmental biophysics 22, (1983). 29. Sale, A. J. H.. H. Effects of high electric field on microorganisms I. Killing of bacteria and yeasts. Biochim. Biophys. Acta 148, (1967). 30. Hamilton, W. A.; Sale, A. J. H. Effects of high electric fields on microorganisms: II. Mechanism of action of lethal effect. Biochim. Biophys. Acta 48, (1967). 31. Wang, H.-Y., Bhunia, A. K. & Lu, C. A microfluidic flow-through device for high throughput electrical lysis of bacterial cells based on continuous dc voltage. Biosensors & bioelectronics 22, (2006). 32. Bao, N. & Lu, C. A microfluidic device for physical trapping and electrical lysis of bacterial cells. Applied Physics Letters 92, (2008). 33. Lee, S. & Tai, Y. A micro cell lysis device. Sensors and Actuators A 73, (1999).

56 Gasparac, R. et al. Ultrasensitive electrocatalytic DNA detection at two- and threedimensional nanoelectrodes. Journal of the American Chemical Society 126, (2004). 35. Lapierre, M. A., O Keefe, M., Taft, B. J. & Kelley, S. O. Electrocatalytic detection of pathogenic DNA sequences and antibiotic resistance markers. Anal. Chem. 75, (2003). 36. Nielsen, P. E. Peptide Nucleic Acid. A Molecule with Two Identities. Accounts of Chemical Research 32, (1999). 37. Aguirre-Avalos, G., Zavala-Silva, M. L., Díaz-Nava, a, Amaya-Tapia, G. & Aguilar- Benavides, S. Asymptomatic bacteriuria and inflammatory response to urinary tract infection of elderly ambulatory women in nursing homes. Archives of medical research 30, (1999).

57 46 4 Optimized templates for bottom-up growth of high-performance integrated biomolecular detectors Another important aspect of biosensors is consideration of the materials and techniques utilized in fabrication. Since the materials a detection platform is restricted by can determine if the platform will rise or fall on the open market. The goal of my second manuscript was to address these material issues by testing various material platforms including printed circuit boards, plastics and glass, as suitable material platforms for the growth of nanostructured microelectrodes. Disclosure of work within this manuscript; B.L. designed and manufactured alternate materials test platforms, B.L. and R.D.H. tested and characterized different platforms, B.L. and J.D. performed electrochemical detection assay on platforms, M. P. designed and tested various size templates on glass. B.L., R.D.H., J.D., E.H.S. and S.O.K. wrote the manuscript with contributions from all of the other authors. Lam, B., Holmes, R. D., Das, J., Poudineh, M., Sargent, E. H., & Kelley, S. O. Optimized Templates for Bottom-Up Growth of High-Performance Integrated Biomolecular Detectors. Lab on a Chip 13, (2013) 4.1 Abstract Electrochemical deposition of metals represents an important approach in the bottom-up fabrication of nanostructures and microstructures. We have used this approach to generate highperformance chip-based biosensors using silicon as a platform for the generation of sensor arrays. Here, we explore the applicability of different materials to support the electrodeposition and identify the parameters that are essential for robust sensor growth. We show that inexpensive materials can be used as templates for electrodeposition, and demonstrate that these low-cost sensors exhibit clinically-relevant levels of sensitivity and specificity. In particular, we prove herein that the glass-based sensors successfully detect E. coli in urine, when present at the 100 cfu μl 1 levels found typically in samples of patients with urinary tract infections.

58 Introduction Chip-based biosensors are an important class of tools for integrated biomolecular detection devices, and enable specific identification of clinically-relevant biomarkers [1-8]. Sensors that read out protein, nucleic acid or small molecular biomarkers can be used to classify disease states or monitor progression. A variety of readout methods have been used in conjunction with chipbased sensors, including colorimetric [7,9], fluorescence [3,10], electronic [8,11] and electrochemical approaches [4, 12-17]. A key limitation of chip-based sensors relates to efficient capture of molecular analytes [18-20]. Generating a detectable signal requires a collision between a molecule of interest and a sensor that is able to capture and detect the molecule. Unfortunately, collisional frequencies achieved with planar sensors can be low. We recently reported a solution to this limitation, one that relies on the fabrication of microscale three-dimensional electrodeposited structures sensors that enable efficient capture and readout of slow-moving mrna molecules [21]. Electrodeposition is one of the few approaches available that permits the production of large (>50 micron) structures that are three-dimensional, and has the added advantage of producing tunable surface morphology at the nanoscale [22-23]. We have used these features of nanostructured microelectrodes (NMEs) to control dynamic range, and have also shown that a combination of structural features on the nano- and microscale are essential for attaining clinically-relevant levels of sensitivity [21,22]. Templated electrodeposition is widely used in the generation of nanostructured surfaces, nanowires, and nanoparticles, and its tunability has been exploited to tailor the sizes and shapes of these nanomaterials [24-29]. Less work has been done, however, on the growth of threedimensional microscale structures that are produced using this technique. The properties of the base substrate, the material used for templating, and the size and shape of the aperture used as a template could all have an effect on the resultant structure and its performance. Our previous work leveraged electrodeposition to produce nanostructured microelectrodes-structures composed of gold or palladium that were coated with a nanostructured layer of palladium. In prior studies, we explored how nanostructuring and sensor size affect sensor performance, but did not investigate the tolerance of the approach to different chip structures and substrates [16, 21-23]. Here, we investigate a variety of materials including glass, plastic and printed circuit board as substrates and evaluate their performance. We identify the critical features that a

59 48 material must have to support the electrodeposition of microscale structures. Using low-costsubstrates identified during this study, we demonstrate that high levels of sensitivity and specificity are maintained by the sensors, and we explore the relationship between template properties and sensor morphology. 4.3 Materials and Methods Silicon chip fabrication Chips were fabricated by Advanced Micro Sensors (Shrewsbury, MA) on the surface of 6 diameter 300 μm thick prime silicon wafers obtained from Silicon Valley Microelectronics (Santa Clara, CA). Wafers were first passivated with a 1 μm layer of thermal SiO 2 to isolate the sensing electrodes. Using positive photoresist a lift-off layer was patterned on the surface of the oxide. A metal layer consisting of 5 nm Ti followed by 50 nm Au was evaporated onto the surface, followed by lift-off to create the sensing electrode layer. A SiO 2 passivating layer was deposited on the surface using Plasma Enhanced Chemical Vapor Deposition. Using positive photoresist an aperture etch mask was patterned on the surface and 5 μm apertures were etched into the SiO 2 layer using hydrofluoric acid. Fabrication of nanostructured microelectrodes (NMEs) on the surface was done by electroplating in a solution of 20 mm HAuCl 4 and 0.5 M HCl at a constant potential of 0 mv for 90 s. A finely nanostructured Pd coating was electroplated by using a solution of 5 mm PdCl 2 and 0.5 M HClO 4 at a constant potential of 250 mv for 10 s. Printed circuit board chip and nme fabrication PCB chips were fabricated by Omega Circuits (Toronto, ON) on the surface of standard FR-4 fiberglass board. Boards were pre-coated with copper foil. A shadowmask is first laminated onto the surface of the board and patterned in similar fashion to positive photolithography, but achieving less fine resolution (e.g. typical critical dimensions on the μm lengthscale). Boards are immersed in a copper etchant to create the sensing electrode layer. Soldermask is applied to the surface of the board, and acts as an aperture layer similar to the silicon chips. The soldermask is patterned in a similar fashion to negative photoresist and apertures 40 μm in diameter were created in the aperture layer. Individual PCB chips are machined from the board using an auto router. Corrosion of the Cu layer immediately occurs in above plating solution. An

60 49 alternative plating method utilizes a Ni protection layer from electroplating in 200 mm NiSO 4 and 0.5 M HBO 3 at 850 mv for 1 h followed by an Au electrode layer from electroplating in 20 mm HAuCl 4 and 0.5 M NaOH at 400 mv for 60 s. A second alternative method utilized a Au sulfite protection layer by electroplating in a gold sulfite solution obtained from Transene Inc (Danvers, MA) at 500 mv for 1 h followed by an Au electrode layer from electroplating in 20 mm HAuCl 4 and 0.5 M HCl at 0 mv for 90 s. Plastic chip and nme fabrication Plastic chips were fabricated by MiniFAB (Scoresby, AUS) on the surface of fluorinated eythlene polymer (FEP) film. Sensing electrodes were printed on the surface of the film using a colloidal gold ink. Chips were made on flexible FEP film which is not suitable for spin coating, hence we first adhered them to glass to act as a rigid substrate. We spun a negative photoresist SU (3000 rpm, 30 s) to create the aperture layer, which is patterned using contact lithography to create apertures 40 μm in diameter. Better resolution was not achievable because the substrate is not perfectly rigid and the contact of the mask to the surface is not optimal. Fabrication of NMEs on the surface was done by electroplating in a solution of 20 mm HAuCl 4 and 0.5 M HCl at a constant potential of 0 mv for 90 s. Glass chip fabrication Glass chips were fabricated in-house utilizing substrates obtained from Telic Company (Valencia, CA) that were pre-coated with 5 nm Cr 50 nm Au and AZ1600 positive photoresist. Sensing electrodes were patterned using standard contact lithography and etched using Au and Cr wet etchants followed by removal of the positive photoresist etchant mask. We spin-coated a negative photoresist SU (5000 rpm, 30 s) to create the aperture layer, and is patterned using contact lithography to create apertures 5 μm in diameter. Shipley 1811 positive photoresist was spun on the surface and patterned to create the plasma etch well layer. Chips were diced in house using a standard glass cutter. Fabrication of NMEs on the surface was done by electroplating in a solution of 20 mm HAuCl 4 and 0.5 M HCl at a constant potential of 0 mv for 90 s. A fine nanostructured Pd coating was electroplated by using a solution of 5 mm PdCl 2 and 0.5 M HClO 4 at a constant potential of 250 mv for 10 s. Synthesis and purification of peptide nucleic acid

61 50 In-house synthesis of peptide nucleic acid (PNA) probes was carried out using a Protein Technologies Prelude peptide synthesizer. The following probe sequences specific to mrna targets were utilized for detection: NH 2 -Cys-Gly-Asp-ATC TGC TCT GTG GTG TAG TT-Asp- CONH 2 (E. coli), NH 2 -Cys-Gly-Asp-CCC GGG GAT TTC ACA TCC AAC TT-Asp-CONH 2 (P. aergugin.), NH 2 -Cys-Gly-Asp-CGA CAC CCG AAA GCG CC TTT-Asp-CONH 2 (E. faecalis.) and NH 2 -Cys-Gly-Asp-CCA CAC ATC TTA TCA CCA AC-Asp-CONH 2 (S. aureus). All probes were stringently purified by reverse phase high performance liquid chromatography. Probe sequences were quantified by measuring absorbance at 260 nm with a NanoDrop and excitation coefficients were calculated from Bacterial samples and lysis Escherichia coli was acquired from Invitrogen ( ). E. coli was grown in LB-Broth medium in an incubating shaker at 37 C. After growth to the desired population the growth media was replaced with 1 PBS. Total RNA was extracted utilizing an Invitrogen Purelink Total RNA Extraction Kit ( ) and quantified with the NanoDrop. Lysis of bacteria was performed utilizing a Claremont BioSolutions OmniLyse rapid cell lysis kit. Human urine samples were obtained from Bioreclamation (Westbury, NY) and were spiked with E. coli prior to lysis. G. Electrochemical measurements Electrochemical measurements were performed using a BASi EC Epsilon potentiostat in a standard 3-electrode configuration with a Ag/AgCl reference and Pt counter electrode. Acid etch scans were performed in 50 mm H 2 SO 4 in H 2 O. Electrocatalytic solutions contained 10μM Ru(NH 3 ) 3+ 6 and 4 mm Fe(CN) 3-6 in a 0.1 PBS buffer solution. Electrocatalytic solutions were purged with N 2 gas for 5 min prior to electrochemical scans. Differential pulse voltammetry (DPV) was utilized to scan before and after hybridization signals. Functionalization and hybridization protocol Electrodes were functionalized with 100 nm of the designated probe and 900 nm mercaptohexanol for 30 min at room temperature. Chips were washed 2 5min with 0.1 PBS buffer after probe deposition and sample hybridization. After washing DPV measurements were

62 51 performed following probe deposition and sample hybridization in the above electrocatalytic solution. Chips were hybridized with synthetic DNA, E. colitotal RNA, E. coli lysate or E. coli urine lysate samples for 30 min at 37 C. 4.4 Results and Discussion Baseline performance of sensors fabricated on silicon Silicon is a widely used material for photolithographic patterning and the development of highperformance devices. The ability to print multiple chips on a silicon wafer that can then be segmented into individual devices allows highly parallelized fabrication, and the ability to access this material in a form that is very flat at the nanoscale allows the generation of very intricate circuits. Our devices are generated with a set of gold leads first being adhered and patterned onto silicon, with a passivation layer of silicon oxide or nitride then being introduced as a dielectric (Figure 4.1). Apertures with diameters of 5 microns are then introduced on the tips of the leads, and it is in these openings that the electrodeposition of gold is catalyzed by an applied potential. The plated metal fills the aperture, and then forms needles that grow anisotropically. The resultant structures can be characterized by SEM and optical microscopy. Electrochemical analysis can also be useful to analyze the integrity of the gold plated within the microstructure. Scanning a sensor in 50 mm H 2 SO 4 produces a characteristic cyclic voltammogram that features the production of gold oxide at +1.2 V and its removal at +0.8 V vs. Ag/AgCl (Figure 4.1D)

63 52 Figure 4.1- Silicon-based NME characterization. (A) Silicon-based NME sensor chip. (B&C) SEM and optical image of aperture before (B) and after (C) HAuCl 4 plating. (A) Inset cross section schematic from bottom to top; Si wafer base, thermal SiO 2, Ti Au electrode, SiO 2 insulating layer with aperture and electroplated NME. (D) Acid scan of typical Si NME in 50 mm H 2 SO 4. (E) Electrochemical nucleic acid detection scheme PNA probe deposition and pre-scan (left), sample hybridized and post-scan (right). (F) Electrochemical current changes observed when sensors coated with the same probe were incubated with 1 fm complementary (+)

64 53 and 100 nm non-complementary ( ) DNA sequences. (F Silicon performance evaluated by Jagotamoy Das) Nanostructured microelectrodes produced using these chips have been studied previously [21-23], and shown to be effective in the specific detection of nucleic acids at femtomolar levels, even when present in unpurified crude lysates. Nucleic acid probes that are designed to be complementary of a gene of interest are designed, synthesized, and attached to the sensors via a thiolate linker similar as described previously [21,30]. The reporter system used for readout leverages the electrostatic attraction between anionic nucleic acids analytes and a cationic electron acceptor Ru(NH 3 ) 3+ 6 to generate an electrochemical response [31]. This response is amplified with the inclusion of Fe(CN) 3 6, a more easily reduced anionic electron acceptor that efficiently reoxidizes Ru(II) and allows it to be available for further redox cycles (Figure 4.1E). This reporter system allows ultrasensitive detection of nucleic acids without the need for enzymatic amplification. Figure 4.1F shows representative data obtained when these sensors are exposed to solutions containing either 1 fm of a complementary sequence, or 100 nm of a noncomplementary sequence. A large positive current change is observed with the complement, while a negative current change is observed with a non-complement. The latter effect is expected to arise because non-specifically bound probe is washed away during hybridization and lowers the overall background signal Testing of printed circuit board as a substrate for sensor deposition While impressive performance and femtomolar detection limits were achieved with electrodeposited gold sensors fabricated on silicon, it was unclear whether such a refined material was required. We hypothesize that the impressive performance of Si based sensors is dependent only on the morphology and composition of the electrodeposited sensor itself and not on the base substrate. Given the need to keep materials costs at a minimum for eventual clinical use of the sensor system, an exploration of other substrates was merited. Printed circuit boards, which can be rapidly fabricated without the need for expensive masks, were an excellent candidate for testing. Electrode arrays of recessed apertures could be straightforwardly and rapidly produced, and with an inherent cost much lower than silicon. Printed circuit board based sensor chips were made that featured apertures created in the soldermask layer that could be used to template NME growth. In order to prevent corrosion of

65 54 Cu during electroplating (Figure 4.2C), it was necessary to produce an initial protection layer of NiSO 4 (Figure 4.2D) or AuSO 3 (Figure 4.2F). To create extruded electrodes for electrochemical sensing the protection layer was followed by electroplating Au(OH) 2 (Figure 4.2E) or HAuCl 4 (Figure 4.2G). Electrodes fabricated in this fashion exhibit increased growth at the aperture edges as compared with the recessed centre. This is a direct result of a loss in aperture resolution (40 μm), compared to Si (5 μm) and the thickness of soldermask layer (50 μm) compared the Si passivating layer (<1 μm). The recessed center of the aperture would experiences much slower diffusion of the plating reagents, and this would worsen as the edges of the aperture are plated. Figure PCB-based NME characterization. (A) Image of PCB NME sensor chip. (B) SEM and optical image of aperture before plating. Immersing in HAuCl 4 causes corrosion of copper layer (C). To prevent corrosion, PCBs were plated first with NiSO 4 (D) or with AuSO 3 (F) and subsequently plated with HAuCl 4 shown respectively in (E) and (G). Cross section schematic (H) from bottom to top; FR-4 PCB fiberglass base, Cu electrode layer,

66 55 soldermask insulating layer with aperture filled by plating with (D) or (F), electroplated NME. (I) Acid scan of typical PCB NME from (G) in 50 mm H 2 SO 4. Electrochemical analysis was performed to interrogate the integrity of the gold plated in PCB NMEs (Figure 4.2I). The characteristic gold oxidative peak at 1.2 V and reductive peak at 0.8 V vs. Ag/AgCl is observed similar to Si. However, an auxiliary peak appears at 0.3 V which is likely due to Cu impurities from the underlying Cu layer given the ability of this element to migrate through Au. The performance of electrochemical sensors generated with PCB as a base was poor and inconsistent when compared to Si NMEs. When the DNA detection experiment described above was used to benchmark performance, large background currents were observed leading to poor signal to noise, and the Cu impurities in the Au NMEs made the quantitation of signals due to DNA binding difficult to monitor (data not shown) Testing of plastic as a substrate for sensor deposition Given the incompatibility of metal-containing materials for NME fabrication, we focused on more inert substrates that might be better suited for producing robust structures. We tested fluorinated ethylene polymer (FEP) a common flexible printed circuit material as a substrate for NME growth. A printed Au ink was used to generate the Au leads, and an insulating SU-8 aperture layer was added using traditional cleanroom techniques. The flexibility of FEP posed a challenge, as it was difficult to process reliably, since both spin coating SU-8 and contact mask lithography is difficult and unreliable, and produces apertures with poor resolution that can't be made smaller than 40 μm. Electroplating of the FEP-based NMEs (Figure 4.3C) was straightforward and used the same plating method as with Si NMEs. The structures obtained again featured recessed sections in the center of the aperture, indicating that this effect does arise because of the large size of the aperture. Electrochemical analysis was again performed to interrogate the integrity of the gold electroplated in FEP-based NMEs (Figure 4.3E). The cyclic voltammograms of these structures did not contain any peaks except those characteristic of gold. However, performance of electrochemical DNA sensing with the FEP-based NMEs was poor (100 nm limit of detection) and inconsistent as compared with Si NMEs. The inability to generate a structure that features

67 56 the same verticality of the silicon-based NMEs likely affects sensitivity, and limits the participation of the entire structure in productive collisions with target molecules. Figure Plastic-supported NME characterization. (A) Image of plastic NME sensor chip. (B&C) SEM and optical image of aperture before (B) and after plating (C) in HAuCl 4. (D) Cross-section schematic from bottom to top; plastic base, Au electrode layer, SU-8 insulating layer with aperture, electroplated NME. (E) Acid scan of typical plastic-supported NME in 50 mm H 2 SO 4.

68 Testing of glass as a substrate for sensor deposition We also tested borosilicate glass as potential substrate to evaluate whether it possessed better features for NME growth. The rigidity and inertness of this material makes it a good substitute for silicon, but it is much more cost-effective alternative. Glass-based NME structures were fabricated on the surface of plain borosilicate glass (Figure 4.4). Glass slides were coated with an Au gold layer and positive photoresist, and SU-8 and contact mask lithography were used to generate aperture patterns. The aperture sizes generated were comparable to those obtained with silicon (Figure 4.4B) and were highly reproducible. Figure Glass-supported NME characterization. (A) Image of a glass NME sensor chip. (B&C) SEM and optical image of aperture before (B) and after plating (C) in HAuCl 4. (D) Cross section schematic from bottom to top; glass base, Cr Au electrode layer, SU-8 insulating layer with aperture, electroplated NME. (E) Acid scan of typical glass NME in 50 mm H 2 SO 4. (F, G, H) Electrodeposition within 100, 25, and 5 micron square apertures, respectively. (I, J, K) Electrodeposition within 100, 25, and 5 micron circular apertures, respectively. (L) Electrodeposition in an aperture 100 microns by 5 microns. (F G aperture variations performed by Mahla Poudineh)

69 58 Electroplating of glass-based NMEs (Figure 4.4C) was performed using the same protocol needed to produce Si-based NMEs, and structures were produced that exhibited similar sizes and morphologies. Electrochemistry was again used to investigate the integrity of the Au electroplated in glass NMEs (Figure 4.4E), and the expected scans were obtained reflecting pure Au NMEs. We also investigated the role of aperture size and shape on the morphology of glass NME structures (Figure 4.4F 4.4L). We observed that structures with larger aperture size 25 μm or greater exhibit edge effects and recessed interiors, regardless if the aperture is square or circular (Figure 4.4F, 4.4G, 4.4I, 4.4J). As aperture size is decreased to 5 μm (Figure 4.4H, 4.4K), edge effects are no longer an issue and structures protrude in a more uniform fashion without apparent recessed areas. Restriction to 5 μm in one-dimension only (Figure 4.4L) is sufficient to eliminate the edge effect, where electrodes protrude in a uniform fashion along the lateral direction Validation of clinically-relevant sensitivity and specificity using glass chips Given that glass appeared to support the growth of structures that were physically and electrochemically indistinguishable from those made on silicon, we evaluated the performance of the NMEs when challenged with synthetic oligonucleotides, and crude E. coli lysates in buffer and urine. These experiments were performed with different types of sensors that target different bacterial organisms: E. coli, E. faecalis, S. aureus, or P. aeruginosa. In any given trial, E. coli sensors were tested alongside two types of non-target sensors in order to assess specificity (Figure 4.5). Excellent sensitivity and specificity was observed, indicating that the glass-based sensors are comparable to those originally generated on silicon.

70 59 Figure Glass-based NME assay validation. All chips were coated with specific pathogen probes and challenged with (A) total E. coli RNA extract (1 ng μl 1 ) (B) E. coli lysate (100 cfu μl 1 ) and (C) urine samples spiked with 100 cfu μl 1 E. coli and subsequently lysed. (Probe molecules and bacterial lysis: Davis Holmes, Electrochemical measurements: Jagotamoy Das) 4.5 Conclusions Here we have investigated various materials that possess desirable manufacturing qualities including low-cost and quick design to prototype cycles. Printed circuit board is one of the lowest cost materials, with well-established quick and cheap manufacturability. However, issues arose with Cu compatibility and large background currents due to loss in aperture resolution, which led to poor assay performance. A plastic was also tested that appeared to be an ideal low

71 60 cost material, but due to the flexible nature of the material lithographic processing is difficult and produces apertures with low resolution, resulting in large background currents and poor electrochemical assay performance. Our investigation of standard borosilicate glass as a substrate for NME growth revealed that this is the best substrate for NME growth. We found that glass has sufficient rigidity and flatness for lithographic techniques required for small aperture sizes, yet it is widely available and low-cost. We found that aperture size is the main factor in eliminating predominate edge growth effects, which would cause recessed electrodes that are undesirable for electrochemical sensing. These results are generally applicable to the electrodeposition of any type of microscale structured template for growth.

72 References 1. Washburn, A. L. & Bailey, R. C. Photonics-on-a-chip: recent advances in integrated waveguides as enabling detection elements for real-world, lab-on-a-chip biosensing applications. The Analyst 136, (2011). 2. Balasubramanian, K. Challenges in the use of 1D nanostructures for on-chip biosensing and diagnostics: a review. Biosensors & bioelectronics 26, (2010). 3. Walt, D. R. Chemistry. Miniature analytical methods for medical diagnostics. Science (New York, N.Y.) 308, (2005). 4. Cederquist, K. B. & Kelley, S. O. Nanostructured biomolecular detectors: pushing performance at the nanoscale. Current opinion in chemical biology 16, (2012). 5. Rosi, N. L. & Mirkin, C. a Nanostructures in biodiagnostics. Chemical reviews 105, (2005). 6. Hianik, T. & Wang, J. Electrochemical Aptasensors - Recent Achievements and Perspectives. Electroanalysis 21, (2009). 7. Giljohann, D. a et al. Gold nanoparticles for biology and medicine. Angewandte Chemie (International ed. in English) 49, (2010). 8. Sapsford, K. E., Francis, J., Sun, S., Kostov, Y. & Rasooly, A. Miniaturized 96-well ELISA chips for staphylococcal enterotoxin B detection using portable colorimetric detector. Analytical and bioanalytical chemistry 394, (2009). 9. Park, C. H. et al. A Direct, Multiplex Biosensor Platform for Pathogen Detection Based on Cross-linked Polydiacetylene (PDA) Supramolecules. Advanced Functional Materials 19, (2009). 10. Duan, X. et al. Quantification of the affinities and kinetics of protein interactions using silicon nanowire biosensors. Nature nanotechnology 7, (2012). 11. Alligrant, T. M., Nettleton, E. G. & Crooks, R. M. Electrochemical detection of individual DNA hybridization events. Lab on a chip 13, (2013).

73 Slinker, J. D., Muren, N. B., Gorodetsky, A. A. & Barton, J. K. Multiplexed DNAmodified electrodes. J. Am. Chem. Soc. 132, (2010). 13. Xiao, Y., Lubin, A. a, Baker, B. R., Plaxco, K. W. & Heeger, A. J. Single-step electronic detection of femtomolar DNA by target-induced strand displacement in an electrodebound duplex. Proceedings of the National Academy of Sciences of the United States of America 103, (2006). 14. Malhotra, R., Patel, V., Vaqué, J. P., Gutkind, J. S. & Rusling, J. F. Ultrasensitive electrochemical immunosensor for oral cancer biomarker IL-6 using carbon nanotube forest electrodes and multilabel amplification. Anal. Chem. 82, (2010). 15. Das, J. & Kelley, S. O. Protein detection using arrayed microsensor chips: tuning sensor footprint to achieve ultrasensitive readout of CA-125 in serum and whole blood. Analytical chemistry 83, (2011). 16. Das, J. et al. An ultrasensitive universal detector based on neutralizer displacement. 4, (2012). 17. Squires, T. M., Messinger, R. J. & Manalis, S. R. Making it stick: convection, reaction and diffusion in surface-based biosensors. Nature biotechnology 26, (2008). 18. Sheehan, P. E. & Whitman, L. J. Detection limits for nanoscale biosensors. Nano letters 5, (2005). 19. Nair, P. R. & Alam, M. a Screening-limited response of nanobiosensors. Nano letters 8, (2008). 20. Soleymani, L. et al. Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS nano 5, (2011). 21. Soleymani, L., Fang, Z., Sargent, E. H. & Kelley, S. O. Programming the detection limits of biosensors through controlled nanostructuring. Nature nanotechnology 4, (2009). 22. Bin, X., Sargent, E. H. & Kelley, S. O. Nanostructuring of sensors determines the efficiency of biomolecular capture. Anal. Chem. 82, (2010).

74 Plowman, B. J., Bhargava, S. K. & O Mullane, A. P. Electrochemical fabrication of metallic nanostructured electrodes for electroanalytical applications. Analyst 136, (2011). 24. S. E. Brunker, K. B. C. and C. D. K. Metallic barcodes for multiplexed bioassays. Nanomedicine 2, (2007). 25. Bera, D., Kuiry, S. C. & Seal, S. Synthesis of nanostructured materials using templateassisted electrodeposition. JOM 56, (2004). 26. Bera, D., Kuiry, S. C., Patil, S. & Seal, S. Palladium nanoparticle arrays using templateassisted electrodeposition. Applied Physics Letters 82, 3089 (2003). 27. Fasol, G. & Runge, K. Selective electrodeposition of nanometer scale magnetic wires. Applied Physics Letters 70, 2467 (1997). 28. Wang, B. D. et al. Electrodeposition of Metallic Nanowire Thin Films Using Mesoporous Silica Templates. Advanced Materials (2003). 29. Vasilyeva, E. et al. Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mrna capture. Angewandtbe Chemie (International ed. in English) 50, (2011). 30. Lapierre, M. a, O Keefe, M., Taft, B. J. & Kelley, S. O. Electrocatalytic detection of pathogenic DNA sequences and antibiotic resistance markers. Analytical chemistry 75, (2003).

75 64 5 Solution-based circuits enable rapid and multiplexed pathogen detection An important goal for biosensors is the multiplexed analysis of samples with a single device. For many applications highly multiplexed analysis would provide valuable utility providing accurate analysis on a panel of relevant biomarkers. An issue with highly multiplexed electrochemical sensors is the need for electrical connections to each sensor element. In a serial connection approach this requires n electrical output connections for n number of sensors. In a parallel connection approach the number of electrical output connections can be reduced to 2 n, however this requires the on chip integration of active switching electronics. The goal of my third manuscript was to develop a passive switching method that is equivalent to resource heavy active switching electronics. A more in depth description of the method and illustration of the crosstalk issue is given in the supplementary information. Disclosure of work within this manuscript; B.L., J.D., R.D.H., L.L., A.S., E.H.S. and S.O.K. developed the concepts described and designed the experiments; B.L. designed and fabricated multiplexed chips, B.L. performed electrochemical crosstalk experiments; R.D.H. developed probes for analysis, R.D.H., L.L. and A.S. validated probes, B.L. and J.D. performed multiplexed lysate experiments, B.L., J.D., E.H.S. and S.O.K. wrote the manuscript with contributions from all of the other authors. Lam, B., Das, J., Holmes, R. D., Live, L., Sage, A., Sargent, E. H., & Kelley, S. O. Solution- Based Circuits Enable Multiplexed, Rapid Pathogen Detection. Nature Communications 4, 2001 (2013) 5.1 Abstract Electronic readout of markers of disease provides compelling simplicity, sensitivity and specificity in the detection of small panels of biomarkers in clinical samples; however, the most important emerging tests for disease, such as infectious disease speciation and antibioticresistance profiling, will need to interrogate samples for many dozens of biomarkers. Electronic readout of large panels of markers has been hampered by the difficulty of addressing large arrays of electrode-based sensors on inexpensive platforms. Here we report a new concept solutionbased circuits formed on chip that makes highly multiplexed electrochemical sensing feasible

76 65 on passive chips. The solution-based circuits switch the information-carrying signal readout channels and eliminate all measurable crosstalk from adjacent, biomolecule-specific microsensors. We build chips that feature this advance and prove that they analyse unpurified samples successfully, and accurately classify pathogens at clinically relevant concentrations. We also show that signature molecules can be accurately read 2 minutes after sample introduction. 5.2 Introduction Electronic readout of the presence of specific biological molecules in solution represents a powerful means to detect disease-related markers [1-4]. In particular, highly sensitive and specific methods have been developed to detect nucleic acids [5-13], proteins [14-18] and small molecules [19-22], using electrochemical readout, and it has been shown that the robustness of electrochemistry allows accurate detection to be done in the presence of heterogeneous, unpurified samples [20, 23-25]. Numerous studies documenting the application of electrochemical sensing to cancer [17, 23, 26, 27] and infectious disease [25, 28] markers have illustrated the promise of this strategy for future clinical diagnostics, and the cost-effectiveness of the simple instrumentation needed for analysis further enhances the appeal of electrochemical detection for further development [1]. Arrays of serially addressed biosensors can be fabricated to work in conjunction with electrochemical reporter systems, enabling multiplexing and detection of several analytes simultaneously [29]. However, the need for independently addressed electrical contacts corresponding to each sensor, as well as reference and counter electrodes, requires that highly multiplexed arrays employ an active multiplexing strategy. The additional complexity of integrated active electronics explains why, before this work, electrochemical biodetection reports describe studies that employ very low levels of multiplexing [5-29]. Here we create distinct columns of connected biosensors, perpendicular to which we array distinct rows of electrochemical solution. Using this two-dimensional array of electrodes, we programme transient, solution-based circuits that permit individual analysis of sensors using shared contacts. This approach allows a much higher level of multiplexing than was attainable previously using a small set of contacts. We apply this advance towards the detection of panels of pathogenic bacteria and antibiotic-resistance markers.

77 Methods and Materials Multiplexed chip fabrication Multiplexed chips were fabricated in house on plain glass substrates obtained from Telic Company (Valencia, CA). The substrates are precoated with 5 nm Cr followed by 50 nm Au and positive photoresist (AZ1600). The common WEs were imaged using standard contact lithography and etched using Au and Cr wet etchants, followed by removal of the positive photoresist. A negative photoresist SU was spin coated at 5,000 r.p.m. for 30 s and the aperture layer was imaged into the surface. A positive photoresist S1811 was spin coated on the surface and imaged to act as a lift-off layer. The AUX/REF electrodes (5 nm Cr 50 nm Au) were fabricated onto the surface using standard e-beam evaporation followed by lift-off by sonication in acetone. The probe well layer was created using SU spin coated at 3,000 r.p.m. for 30 s. A thicker liquid channel layer was fabricated with SU at 2,000 r.p.m. for 30 s. A plasma etch mask layer with openings defined for the probe wells was patterned on the surface using SPR 7.0 at 2,000 r.p.m. for 30 s. Chips were diced in house using a standard straight edge scriber. Before electroplating, chips were O 2 plasma etched in a Samco-RIE-1C reactive ion etcher at 15 W for 15 s to create a hydrophilic probe well area. The core structure of the nanostructured microelectrodes was fabricated by electroplating chips in a solution of 50 mm HAuCl 4 and 0.5 M HCl at 0 mv for 100 s. The nanostructured surface of the nanostructured microelectrodes was generated by electrodeposition of a Pd coating in a solution of 5 mm PdCl 2 and 0.5 HClO 4 at 250 mv for 10 s. Electrochemical chip characterization Ferricyanide was used to evaluate the electrochemical characteristics of the multiplexed chip. Standard cyclic voltammetry in 2 mm Fe(CN) 3 6 in 0.1 PBS scanned from 50 mv to 300 mv at 100 mv s 1 versus on-chip Au AUX/REF was performed on the surface of chips of similar layout that have WEs that are individually addressed versus the multiplexed chips with solution circuits. In addition, ferricyanide crosstalk was evaluated by filling a single channel with 2 mm Fe(CN) 3 6 in 0.1 PBS and adjacent channels with 0.1 PBS. Differential pulse voltammetry from 50 mv to 300 mv was performed in the ferricyanide-containing channel and adjacent channels without ferricyanide.

78 67 Spectroscopic chip characterization The methylene blue crosstalk experiment was performed on the surface of SCCs by first loading each liquid channel with a solution of 100 μm MB and 50 mm NaCl. Chips were placed in a humidity chamber where the middle lane WEs and AUX/REF was held under bias for 1 h at 650 mv, which is a sufficient potential for reduction of MB. The MB in each channel was then diluted by a factor of 10 and the corresponding absorbance was measured with a ultraviolet visible spectrophotometer. Electrochemical assay crosstalk evaluation Probe wells on the surface of multiplexed chips were used to functionalize with target-specific PNA probes. Each well was filled with 1.5 μl of complementary or non-complementary probe solution initially heated to 60 C for 5 min, containing 100 nm probe and 900 nm mercaptohexanol, and was incubated directly on chip for 30 min at room temperature. Chips were washed with 0.1 PBS twice for 5 min after probe deposition, and an initial DPV background scan in electrocatalytic solution (10 μm Ru(NH3) 3+ 6 and 4 mm Fe(CN) 3 6 in a 0.1 PBS) was performed. Chips were then hybridized with 10 nm complementary target in 1 PBS for 30 min at 37 C. Chips were washed with 0.1 PBS twice for 5 min after target hybridization and differential pulse voltammetry hybridization scan in electrocatalytic solution was performed. Identification of pathogen-specific probes A list of potential probes based on the rpoβ gene sequences of the bacteria under study was first generated. All sequences were obtained from the NCBI Nucleotide Database. For a given bacterial sequence, a BLAST search was performed to identify the rpoβ sequence of the most similar bacterial species that could potentially cross-hybridize. This sequence was retrieved and aligned with the targeted bacterial sequence using CLUSTALW2. A computer script was used to identify regions of greatest variability, as they can be used to best differentiate the target species from non-target species. Potential rpoβ probes that could cross-hybridize with non-target molecules in patient samples were eliminated first. Other probe characteristics, such as secondary structure melting temperature, were also analysed to ensure optimal specificity. If

79 68 suitable probes could not be identified targeting the rpoβ mrna, then the 16S rrna (Morganella morganii, Pseudomonas aeruginosa, Enterobacteria family and universal probe) or the 28S rrna (Candida albicans) was used as a target instead. Synthesis of probes Synthesis of PNA probes was performed in house using a Protein Technologies Prelude peptide synthesizer. The following pathogen probe sequences (NH2-Cys-Gly-Asp SEQUENCE Asp- CONH2) that are specific to mrna targets were synthesized: E. coli, 5 -ATC-TGC-TCT-GTG- GTG-TAG-TT-3 ; Proteus mirabilis, 5 -AAG-CGA-GCT-AAC-ACA-TCT-AA-3 ; S. saprophyticus, 5 -AAG-TAA-GAC-ATT-GAT-GCA-AT-3 ; S. aureus, 5 -CCA-CAC-ATC- TTA-TCA-CCA-AC-3 ; Klebsiella pneumoniae, 5 -GTT-TAG-CCA-CGG-CAG-TAA-CA-3 ; M. morganii, 5 -CGC-TTT-GGT-CCG-AAG-ACA-TTA-T-3 ; P. aeruginosa, 5 -CCC-GGG- GAT-TTC-ACA-TCC-AAC-TT-3 ; K. oxytoca, 5 -CCA-GTA-GAT-TCG-TCA-ACA-TA-3 ; Serratia marescens, 5 -TGC-GAG-TAA-CGT-CAA-TTG-ATG-A-3 ; Enterococcus faecalis, 5 - CGA-CAC-CCG-AAA-GCG-CCT-TT-3 ; Acinetobacter baumannii, 5 -CGT-CAA-GTC-AGC- ACG-TAA-TG-3 ; Streptococcus pyogenes, 5 -TCT-TGA-CGA-CGG-ATT-TCC-AC-3 ; Streptococcus agalactiae, 5 -GTT-CAG-TAA-CTA-CAG-CAT-AA-3 ; Staphylococcus epidermidis, 5 -AAA-TAA-CTC-ATT-GAG-GCA-AC-3 ; Enterobacter cloacae, 5 -TCA-ACG- TAA-TCT-TTC-GCG-GC-3 ; Streptococcus pneumoniae, 5 -GTT-ACG-ACG-CGA-TCT-GGA- TC-3 ; Providencia stuartii, 5 -GCC-AAG-TGC-CAA-TTC-ACC-TAG-3 ; C. albicans, 5 -GCT- ATA-ACA-CAC-AGC-AGA-AG-3 ; Chlamydia trachomatis, 5 -TGC-ATT-TGC-CGT-CAA- CTG-3 ; Enterobacteriaceae, 5 -ACT-TTA-TGA-GGT-CCG-CTT-GCT-CT-3 ; and Universal bacteria probe, 5 -GGT-TAC-CTT-GTT-ACG-ACT-T-3. After synthesis, all probes were stringently purified using reverse-phase HPLC. Excitation coefficients were calculated from and concentrations of probe molecules were determined by measuring absorbance at 260 nm with a NanoDrop spectrophotometer. Preparation of bacterial samples and lysis The following bacterial strains were used in this study: K. pneumoniae ATCC 27799, E. coli K12 ATCC 33876, E. coli Invitrogen , S. saprophyticus ATCC 15305, P. aeruginosa PAO1, E. faecalis ATCC 29212, S. aureus, C. trachomatis and C. albicans. All bacteria were grown in the appropriate growth media and conditions. Approximate quantification of bacteria

80 69 was performed by measuring the optical density at 600 nm with an Agilent 8453 ultraviolet visible spectrometer. After the desired population was reached, the growth media was replaced with 1 PBS. Lysis of bacteria was performed utilizing the Claremont BioSolutions OmniLyse rapid cell lysis kit. Probe validation Probe solutions were initially heated to 60 C for 5 min before deposition, and contained 100 nm probe and 900 nm mercaptohexanol. Deposition was carried out for 30 min at room temperature. Chips were washed with 0.1 PBS twice for 5 min after probe deposition and chips were then challenged with a non-complementary oligomer at 100 nm in 1 PBS for 30 min. Chips were washed with 0.1 PBS twice for 5 min, and then an initial non-complementary DPV background scan was performed in electrocatalytic solution (10 Ru(NH 3 ) 3+ 6 and 4 mm Fe(CN) 3 6 in a 0.1 PBS). Chips were then hybridized with 1 nm complementary target in 1 PBS for 30 min at 37 C. Chips were washed with 0.1 PBS twice for 5 min after target hybridization and a DPV in electrocatalytic solution was then performed. A similar protocol was followed for lysate testing. Chips were hybridized with lysates in 1 PBS for 30 min at 37 C. Chips were washed with 0.1 PBS twice for 5 min after target hybridization and DPV hybridization scan in electrocatalytic solution was performed. Multiplexed bacterial lysate experiments Multiplexed chips for bacterial lysate testing were functionalized with nine probes using nine probe wells. Probe solutions were initially heated to 60 C for 5 min, and contained 100 nm probe and 900 nm mercaptohexanol that was incubated directly on chip for 30 min at room temperature. Chips were washed with 0.1 PBS twice for 5 min after probe deposition and an initial DPV background scan was collected in electrocatalytic solution. Multiplexed chips were then hybridized with individual and mixed lysates at 100 cells per μl in 1 PBS for 30 min at 37 C in a 200-μl volume.

81 Results Overview of approach The solution circuit chip (SCC) is depicted in Figure 5.1. It consisted of 100 working electrodes (WEs) with 30 off-chip contacts. It included 20 common WEs and 5 counter/reference (CE/RE) electrode pairs, with 25 probe wells to facilitate manual probe deposition and 5 separate liquid channels (Figure 5.1a). We defined templates for the sensing regions of the WEs by opening 5 μm apertures (Figure 5.1c) in the top passivation layer. By restricting growth within these apertures, we then grew micron-sized tree-like electrodes via electrodeposition, resulting in nanostructured microelectrodes that protrude from the surface and reach into solution. It was previously shown that the micron-sized scale of the protruding electrodes increases the crosssection for interaction with analyte molecules [30], whereas the nanostructuring maximizes sensitivity by enhancing hybridization efficiency between tethered probe and the analyte in solution [31]. Here these structures are being used to enable the creation of layered, insulated electrode arrays (Figure 5.1d).

82 71 Figure The solution circuit chip. (a) An SCC featuring 5 liquid channels containing 20 sensors each. (b) An SCC featuring common WEs, and counter and reference electrode pairs (CE and RE) that can be activated in sets to form solution-based circuits (red). (c) Optical image of

83 72 single probe well with 4 WEs. Inset: a scanning electron microscopy(sem) image of a nanostructured microelectrode. Scale bar, 50 μm. (d) Cross-section looking down liquid channel of a sensor on a SCC: glass substrate (light grey), common WE (yellow), SU-8 passivation/aperture layer (dark grey), CE/RE (red), SU-8 probe wells (green) and SU-8 liquid channel barrier (blue). (e) Sensor-to-sensor comparison of SEM images and acid stripping scans for 20 sensors. Although morphological differences may exist, the surface areas of the sensors are highly consistent, as evidenced by the low levels of s.e. (blue error bars) observed for scans of the 20 sensors conducted in acidic solution. Gold oxide is formed at 1.05 V 1.3 V and reduced at 0.85 V. (f) Electrochemical nucleic acid assay scheme. PNA probes are immobilized on microsensors, and in the presence of a complementary target the electrostatic charge on the sensors is increased. This charge change is read out in the presence of Ru(NH 3 ) 3+ 6 and Fe(CN) 3 6, a mixture that yields currents that report on the amount of nucleic acid bound to the sensor. By patterning channels on the SCC, we created separate liquid compartments (Figure 5.1a). WEs are multiplexed on common leads such that they are physically isolated because of the air/water interface in liquid channels separating them. Reference and counter electrodes are routed along the liquid channels. The electrical isolation of the reference and counter layers from the WEs within the SCC, and the ability to bring the electrodes into contact at specific positions using the liquid channels, are the essential elements that allow solution-based circuits to be formed. The contacts made through the conductive solution allow transient circuits to be formed and allow individual contacts to address many sensors in series. The patterned microsensors are functionalized with peptide nucleic acid (PNA) probes specific to regions of targeted pathogens (Figure 5.1f). Electrodes are exposed to samples of interest and binding occurs if a target nucleic acid is present. To detect positive target binding we use an electrocatalytic reporter pair [32] comprising Ru(NH 3 ) 3+ 6 and Fe(CN) 3 6. Ru(NH 3 ) 3+ 6 is electrostatically attracted to the phosphate backbone of nucleic acids bound near the surface of electrodes by probe molecules and is reduced to Ru(NH 3 ) 2+ 6 when the electrode is biased at the reduction potential. The Fe(CN) 3 6 present in solution auto-oxidizes Ru(NH 3 ) 2+ 6 back to Ru(NH 3 ) 3+ 6, which allows for multiple turnovers of Ru(NH 3 ) 3+ 6 and generates an electrocatalytic current. The difference between pre-hybridization and post-hybridization currents are used as a metric to determine positive target binding.

84 Characterization of the SCC The SCC was fabricated using a series of simple lithographic steps followed by the electrodeposition of three-dimensional microsensors. The electrodeposition process produces sensors with somewhat variable nanoscopic morphologies, but as it is programmed to deposit the same number of gold atoms in each structure, it produces sensors with surface areas that vary by less than 10% (Figure 5.1e). We employed electrochemical analysis to determine whether SCCs provided the necessary level of electrochemical isolation. We also devised a spectroscopic approach to confirm the isolation analysis. Cyclic voltammetry of ferrocyanide was used to evaluate electrochemical characteristics of SCCs versus standard serially connected chips (Figure 5.2a). We observed that SCCs have nearly identical electrochemical signals to serially wired chips. It is noteworthy that grounding of all other unbiased counter, reference and working electrodes is required to eliminate crosstalk, as evidenced by the differential pulse voltammetry scans of ferrocyanide with and without the grounding of these electrodes (Figure 5.2b). To further investigate whether signals from adjacent liquid channels were picked up by individual sensors, we added ferrocyanide adjacent to channels and monitored the effect on the signals obtained (Figure 5.2c). It was observed that when ferrocyanide is added to an adjacent channel, there is minimal perturbation of the signal of ferrocyanide within the channel of interest. The effects of repeated scanning were also explored and were found to be minimal (data not shown).

85 74 Figure Electrochemical validation of the SCC. (a) Cyclic voltammograms collected with standard chips with individually addressable electrodes and SCC electrodes in 2 mm ferrocyanide. (b) Differential pulse voltammetry of biased WE and CE+RE solution circuit in 2 mm ferrocyanide when unbiased CE, RE and WEs are ungrounded (red) and grounded (green) versus standard serially connected WE (dotted gray). (c) Differential pulse voltammetry of 2 mm ferrocyanide within liquid channel (green) and on adjacent liquid channels (red) of interest. (d) Evaluation of cross-talk by monitoring MB electrochemical bleaching. Middle channel was activated. (e) MB absorbance measurements for middle and adjacent lanes of SCC. (f) Analysis of signals obtained when sensors that are positive or (g) negative for a target sequence are surrounded by sensors yielding the opposite result. (f g analysis performed by Jagotamoy Das)

86 75 A non-electrochemical strategy was also pursued to verify independently that the solution circuits were formed as desired, and to study any evolving leakage over longer time periods. To investigate electrochemical isolation using a spectroscopic approach we utilized methylene blue (MB), which can be electrochemically reduced to a colourless form (Figure. 5.2d). MB was loaded into all liquid channels, and WEs within the middle channel of the chip were held at the reduction potential of MB for 1 h. Visual evidence of the reduction of MB was observed exclusively in the selected channel, providing a qualitative measure of electrical isolation. Measurements of the absorbance (Figure 5.2e) of each channel quantitatively confirmed that the significant loss in MB absorbance is detected in the middle lane only. To investigate whether signals from adjacent channels would interfere with our electrochemical nucleic acid assay, we investigated different orientations of sensors that would yield positive and negative electrochemical responses in the presence of a target DNA sequence. A small area of sensors was functionalized with a PNA probe sequence that would not bind a specific target DNA sequence, and this area was surrounded by sensors functionalized with a PNA probe that would bind the DNA target (Figure 5.2f). On this chip, a large positive signal change was obtained with the positive sensors, and a small negative signal change was observed on the much smaller number of negative sensors. This indicates that with the nanoampere levels of current generated during sequence analysis, crosstalk between sensors does not influence the results obtained. When trials were conducted that reversed the positions of the positive and negative sensors, the expected reversed results were obtained (Figure 5.2g). These results indicate that the solution-based circuits are indeed suitable for multiplexed sequence detection Detection of urinary tract infection pathogens A compelling application for electrochemical analysis is the identification and classification of pathogens. Infectious disease can be difficult to diagnose accurately, because symptoms caused by different pathogens can be quite similar. However, it is important that a causative organism is identified correctly before a treatment is selected. We therefore elected to adapt the multiplexing capabilities of our chips to look for many different types of pathogens, and to also probe for antibiotic resistance. A set of PNA probes were designed, synthesized and tested, and validated for sensitivity and specificity (Figure 5.3). It is these probes that make individual sensors specific for pathogens, as they are targeted to unique sequences within the genomes of the organisms.

87 76 The probe set developed covered 90% of the common urinary tract pathogens 33 and many of the major types of drug resistance that are encountered in the clinic 34. Figure Validation of pathogen and antibiotic-resistance probes. (a) Backgroundcorrected post-current values for pathogen probe set. Sensors were challenged with 1 nm of a synthetic DNA complement and the signal generated was compared with that obtained with a 100-nM solution of a non-complementary DNA sequence to obtain the background-corrected value. (b) Background-corrected post-current values for antibiotic-resistance probe set. Sensors were challenged with 1 nm of a synthetic DNA complement, and these signals were compared with those obtained with a 100-nM solution of a non-complementary DNA sequence to obtain the background-corrected value. (c) Pathogen probes validated versus bacterial lysates (EC, E. coli; SS, S. saprophyticus; SA, S. aureus;. CT,C. trachomatis; MM, M. morganii; PA, P. aeruginosa; PM, P. mirabilis; KO, K. oxytoca; EF, E. faecalis; SP, S. pneumonia; AB, A. baumannii; Sag, S. agalactiae; SE, S. epidermis; CA, C. albicans; KP, K. pneumonia; SP, S. pyogenes; Spy, S. pyogenes; SM, S. marescens; ECl, E. cloacae; EN, enterobactergenus; UN,

88 77 universal bacteria probe). Error bars reflect s.d. collected from data sets obtained with >3 independent chips. (d) Representative differential pulse voltammograms (DPVs) obtained for a control sample (Ctrl), a lysate containing 1 cfu μl 1, and a lysate containing 100 cfu μl 1.(Probe design and bacterial lysis: Davis Holmes, Probe validation: Ludovic Live & Andrew Sage, Lysate measurements: Jagotamoy Das) We used our multiplexed chips to achieve the parallelized assessment of 30 probes for pathogens and antibiotic-resistance markers (Figure 5.3a). The pathogen probes were targeted against either the RNA polymerase β mrna (rpoβ), or a ribosomal RNA, and the antibiotic-resistance probes were targeted against known sequences correlated with drug deactivation. To screen sequences for specificity, we compared the response obtained from a solution containing a 1-nM concentration of a complementary target with the response from a solution containing a 100-nM concentration of a non-complementary target. The background-subtracted current generated was then analysed, and for the large majority of the probes that were tested the current obtained was greater than baseline by three standard deviations. However, it is noteworthy that the amount of current generated with each probe varies for the different probes. Nonetheless, the magnitude of current generated for each probe was highly reproducible and hence, despite this effect, the approach can be used for specific pathogen detection. The sensitivity of sensors modified with these probes was then tested against a panel of pathogens amenable to culture (Figure 5.3c). Unpurified bacterial lysates containing 1 cfu μl 1 and 100 cfu μl 1 were incubated with the sensors, and electrochemical signals were compared with those obtained when the same sensor type was exposed to 100 cfu μl 1 of nontarget bacteria (Escherichia coli for each trial, except for those testing the sensitivity of the E. coli probes, where Staphylococcus saprophyticus was used). The signals obtained for each pathogen differed, which likely reflects idiosyncratic nucleic acid structures, but in each case excellent sensitivity and specificity were obtained using samples that have undergone minimal processing. In each case, the signals obtained with solutions containing 1 cfu μl 1 were greater by a factor of at least three standard deviations relative to background signals, indicating that the limits of detection reside at or below this concentration. The detection limit obtained here of 1 cfu μl 1 is clinically significant. Many applications in infectious disease testing including testing of swab samples for health-care-associated

89 78 infections or sexually transmitted infections, or the testing of urine for infectious pathogens yield samples that contain concentrations higher than 1 cfu μl 1. This detection and speed combination also approaches the diffusional limits for large molecules in static solutions Multiplexed detection of urinary tract infection pathogens The SCC was then put to the ultimate test: the analysis of samples containing clinically relevant concentrations of pathogens for panels of markers. SCCs were functionalized simultaneously using nine different probes (Figure 5.4a) and were challenged with bacterial lysates at 100 cfu μl 1. Chips were first challenged with lysates of E. coli, the most common urinary tract infection-causing pathogen (Figure 5.4b). The response of sensors modified with the E. coli probe targeted against the RNA polymerase gene (rpoβ) was significant, whereas no other probes showed significant response to the lysate. We also challenged chips prepared with the same sensors and probes with a form of antibiotic-resistant E. coli that contains the β- lactamase (β-lac) gene (Figure 5.4c). With this sample, only EC and β-lactamase sensors exhibited a significant response, indicating that the SCC can classify pathogens and detect antibiotic resistance simultaneously. SCCs were also challenged with lysates of Staphylococcus aureus to confirm successful detection of Gram-positive pathogens (Figure 5.4d). Only electrodes functionalized with SA probe showed a significant electrochemical response. In addition, we challenged chips with a mixture of S. aureus and antibiotic-resistant E. coli (+β-lac) to evaluate the performance of chips brought into contact with several analytes producing a positive response (Figure 5.4e). Only electrodes functionalized with matching probes exhibited significant electrochemical responses to the mixed lysed sample. These results illustrate that the multiplexing provided by solution-based circuits enables the parallelized detection of multiple analytes at clinically relevant levels.

90 79 Figure Multiplexed pathogen and antibiotic-resistance testing on an SCC. (a) Arrangement of sensors on a 100-plexed SCC. (b) Response of SCC challenged with 100 cfu μl 1 E. coli. ( β-lac). (c) Response of SCC challenged with 100 cfu μl 1 E. coli. (+β-lac). (d) Response of SCC challenged with 100 cfu μl 1 S. aureus. (e) Response of SCC challenged

91 80 with 100 cfu μl 1 S. aureus+e. coli (+β-lac) bacterial lysate. (f) Response of SCC at 2 min and 5 min challenged with E. coli lysates at 100 cfu μl 1. (Bacterial lysis: Brian Lam, Electrochemical measurements: Jagotamoy Das) As a final test of the utility of the SCC, we investigated whether very short hybridization times could be used to enable very rapid bacterial detection (Figure 5.4f). We investigated the evolution of signals that could be detected on chip with a 2- and 5-min hybridization, and determined that a pathogen-specific response could be obtained even with the shortest time studied. These results indicate that SCCs can be used for highly multiplexed pathogen detection, and can also be used to deliver the very rapid results needed to make diagnostic information clinically actionable. 5.5 Discussion Using the simple idea that solution-based circuits could be harnessed on chip to enhance the level of multiplexing that could be achieved on a passive chip, we created a new method for electrochemical multiplexing. Compared with complicated and expensive active electronics on the surface of silicon, this approach has significant advantages for the development of low-cost diagnostic tools. A simple six-step fabrication approach, where most of the steps were performed using transparency masks and low-grade glass was used as a substrate, is advantageous over more involved protocols used to create active silicon electronics. The approach reported herein is scalable in that it could be deployed to produce chips with higher levels of multiplexing than shown here. Given the current capabilities of photolithography and technologies that can be used to deliver probe molecules to high-density arrays, thousands of sensors could be addressed using the solution circuit strategy. The development of strategies to maintain the large number of liquid channels would represent the greatest challenge in realizing this level of multiplexing. The sensor technology used here, which relies on the electrodeposition of microscale, threedimensional sensors, is a powerful tool in the analysis of complex samples. It has been shown to perform well in the presence of blood [23] and urine [25], and can also discriminate single base changes in sequence [35]. As demonstrated here, it is compatible with unpurified bacterial lysates and, therefore, is easily integrated for sample-to-answer testing. These features provide a significant advantage over gold standard molecular testing methods, such as the PCR, which typically requires sample clean-up and the use of costly enzymatic reagents. It is also compatible

92 81 with samples ranging from microlitres to millilitres, which broadens the potential clinical utility. Here we have demonstrated that in addition to possessing these features, it can also be used with a highly multiplexed format if the solution circuit approach is used to contact the sensors. The solution-based circuit chip, as it yields rapid and accurate information on the identity and antibiotic resistance of pathogens, represents a significant advance in the field of biomolecular sensing.

93 References 1. Ronkainen, N. J. Halsall, H. B. & Heineman, W. R. Electrochemical biosensors. Chem. Soc. Rev. 39, (2010). 2. Drummond, T. G. Hill, M. G. & Barton, J. K. Electrochemical DNA sensors. Nat. Biotechnol. 21, (2003). 3. Song, S. et al. Functional nanoprobes for ultrasensitive detection of biomolecules. Chem. Soc. Rev. 39, (2010). 4. Lin, P. & Yan, F. Organic thin-film transistors for chemical and biological sensing. Adv. Mater.24, (2012). 5. Soleymani, L. Fang, Z. Sargent, E. H. & Kelley, S. O. Programming the detection limits of biosensors through controlled nanostructuring. Nat. Nanotechnol. 4, (2009). 6. Alligrant, T. M. Nettleton, E. G. & Crooks, R. M. Electrochemical detection of individual DNA hybridization events. Lab. Chip. 13, (2013). 7. Sorgenfrei, S. et al. Label-free single-molecule detection of DNA-hybridization kinetics with a carbon nanotube field-effect transistor. Nat. Nanotechnol. 6, (2011). 8. Slinker, J. D. Muren, N. B. Gorodetsky, A. A. & Barton, J. K. Multiplexed DNA-modified electrodes. J. Am. Chem. Soc. 132, (2010). 9. Fan, C. Plaxco, K. W. & Heeger, A. J. Electrochemical interrogation of conformational changes as a reagentless method for the sequence-specific detection of DNA. Proc. Natl Acad. Sci. USA 100, (2003). 10. Patolsky, F. Lichtenstein, A. & Willner, I. Detection of single-base DNA mutations by enzyme-amplified electronic transduction. Nat. Biotechnol. 19, (2001). 11. Pheeney, C. G. Guerra, L. F. & Barton, J. K. DNA sensing by electrocatalysis with hemoglobin. Proc. Natl Acad. Sci. USA 109, (2012).

94 Wanunu, M. et al. Rapid electronic detection of probe-specific micrornas using thin nanopore sensors. Nat. Nano 5, (2010). 13. Khan, H. U. et al. In situ, label-free DNA detection using organic transistor sensors. Adv. Mater. 22, (2010). 14. Das, J. & Kelley, S. O. Protein detection using arrayed microsensor chips: tuning sensor footprint to achieve ultrasensitive readout of CA-125 in serum and whole blood. Anal. Chem.83, (2011). 15. Duan, X. et al. Quantification of the affinities and kinetics of protein interactions using silicon nanowire biosensors. Nat. Nano 7, (2012). 16. Tang, D. Yuan, R. & Chai, Y. Ultrasensitive electrochemical immunosensor for clinical immunoassay using thionine-doped magnetic gold nanospheres as labels and horseradish peroxidase as enhancer. Anal. Chem. 80, (2008). 17. Malhotra, R. Patel, V. Vaqué, J. P. Gutkind, J. S. & Rusling, J. F. Ultrasensitive electrochemical immunosensor for oral cancer biomarker IL-6 using carbon nanotube forest electrodes and multilabel amplification. Anal. Chem. 82, (2010). 18. Xiang, Y. Xie, M. Bash, R. Chen, J. J. L. & Wang, J. Ultrasensitive label-free aptamer-based electronic detection. Angew. Chem. Int. Ed. 46, (2007). 19. Das, J. et al. An ultrasensitive universal detector based on neutralizer displacement. Nat. Chem. 4, (2012). 20. Zuo, X. Xiao, Y. & Plaxco, K. W. High specificity, electrochemical sandwich assays based on single aptamer sequences and suitable for the direct detection of small-molecule targets in blood and other complex matrices. J. Am. Chem. Soc. 131, (2009). 21. Liu, H. Xiang, Y. Lu, Y. & Crooks, R. M. Aptamer-based origami paper analytical device for electrochemical detection of adenosine. Angew. Chem. Int. Ed. 51, (2012).

95 Kuang, Z. Kim, S. N. Crookes-Goodson, W. J. Farmer, B. L. & Naik, R. R. Biomimetic chemosensor: designing peptide recognition elements for surface functionalization of carbon nanotube field effect transistors. ACS Nano 4, (2010). 23. Vasilyeva, E. et al. Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mrna capture. Angew. Chem. Int. Ed. 50, (2011). 24. Swensen, J. S. et al. Continuous, real-time monitoring of cocaine in undiluted blood serum via a microfluidic, electrochemical aptamer-based sensor. J. Am. Chem. Soc. 131, (2009). 25. Lam, B. Fang, Z. Sargent, E. H. & Kelley, S. O. Polymerase chain reaction-free, sample-toanswer bacterial detection in 30 min with integrated cell lysis. Anal. Chem. 84, (2012). 26. Mani, V. Chikkaveeraiah, B. V. Patel, V. Gutkind, J. S. & Rusling, J. F. Ultrasensitive immunosensor for cancer multienzyme-particle amplification. ACS Nano 3, (2009). 27. Lerner, M. B. et al. Hybrids of a genetically engineered antibody and a carbon nanotube transistor for detection of prostate cancer biomarkers. ACS Nano 6, (2012). 28. Mannoor, M. S. Zhang, S. Link, A. J. & McAlpine, M. C. Electrical detection of pathogenic bacteria via immobilized antimicrobial peptides. Proc. Natl Acad. Sci. USA 107, (2010). 29. Zheng, G. Patolsky, F. Cui, Y. Wang, W. U. & Lieber, C. M. Multiplexed electrical detection of cancer markers with nanowire sensor arrays. Nat. Biotechnol. 23, (2005). 30. Soleymani, L. et al. Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS Nano 5, (2011). 31. Bin, X. Sargent, E. H. & Kelley, S. O. Nanostructuring of sensors determines the efficiency of biomolecular capture. Anal. Chem. 82, (2010).

96 Lapierre, M. A. O Keefe, M. Taft, B. J. & Kelley, S. O. Electrocatalytic detection of pathogenic DNA sequences and antibiotic resistance markers. Anal. Chem. 75, (2003). 33. Foxman, B. The epidemiology of urinary tract infection. Nat. Rev. Urol. 7, (2010). 34. Levy, S. B. & Marshall, B. Antibacterial resistance worldwide: causes, challenges and responses. Nat. Med. 10, S122 S129 (2004). 35. Yang, H. et al. Direct, electronic microrna detection for the rapid determination of differential expression profiles. Angew. Chem. Int. Ed. 48, (2009).

97 86 6 Conclusions and Future Directions 6.1 Thesis Findings The goal of this work was to develop new effective methods for nanomaterial point-of-care electrochemical biosensors. Sample handling and process is an important aspect of any point of care diagnostic technique and must be simple, rapid and effective. High applied voltages, external pump requirements and difficult direct intergration with our NME platform were distinct drawbacks of microchannel electrical lysis (supplementary information). In chapter 3 we developed simple parallel plate lysis chambers that could operate at lower voltages, did not need external pumps and could be directly coupled to our NME chips. We showed these chambers are effective at lysing pathogenic bacteria, releasing intercellular genetic targets. We were able to detect intercellular genetic markers of bacteria and cancer cells by combining our parallel plate lysis platform with our NME detection platform. Remarkably we were able to perform detection at clinically relevant concentrations in crude lysates in buffer and urine without any purification. In chapter 4 we evaluated various materials that possessed desirable manufacturing qualities, such as low cost and quick prototype turnaround times. PCB is one of the cheapest materials with simple and rapid prototyping cycles. However, we ran into many material compatibility issues since copper is the only widely available base metal material and other materials would be costly. Impurities in PCB NME and large background produced poor sensitivities when PCB NMEs were used in our electrocatalyic nucleic acid assay. We also tried FEP, a common plastic used in the electronics industry for manufacturing FPCs. Plastics are cheap materials, however we needed to post process manufactured devices to fabricate a passivating aperture layer for template electrodeposition of NMEs. Flexibility of FEP was an issue and lead to poor aperture resolutions when compared to silicon devices along with poor assay sensitivities due to high background currents. Net we evaluated plain borosilicate glass since it is a cheap and rigid material. We fabricated similar chips to our silicon based devices with similar aperture resolutions and NMEs fabricated on glass were of comparable morphology and electrochemical character to silicon based NMEs. We tested our electrocatalyic nucleic acid assay on glass and found comparable sensitivities to silicon. In addition we challenged our glass platform with bacterial lysates and found we could detect clinically relevant concentrations similar to the silicon platform.

98 87 An important goal for future medical diagnostic technologies is the highly multiplexed analysis of multiple biomarkers. For electrochemical biosensors the most common multiplexing method is the implementation of active switching electronics. However, active electronics are complex and costly to develop and manufacture limiting their practicality for use in consumable biosensors. In chapter 5 we developed a simple passive method for electrochemical multiplexing using simple physical separation of liquid to create SCCs. We characterize this method and show that interference between channels is not significant when non-biased NMEs are held at a fixed potential such as ground. We demonstrate the utility of this method in the successful multiplexed analysis of bacterial lysates at clinically relevant concentrations. 6.2 Future Work We showed that parallel plate lysis chambers were rapid, effective and could be directly coupled to our NME detection platform. Further development of parallel lysis chambers should include indepth analysis of the mechanism of lysis since it is unclear if bacteria are lysed electrochemically and/or electrically. In future developments we intend to design electrical lysis areas directly on chip in cost effective and efficient manner. To aid in the ease of use in the system a method spontaneous fluid flow should be developed to reduce operator involvement in detection of infectious diseases and/or cancers. A mobile integrated measurement system should be developed since testing in developing countries will become an important goal for our platform. From a materials and manufacturing standpoint it would be invaluable to develop nonlithographic methods for fabrication of glass based NME chips which will be instrumental in reducing costs. Fabrication of NMEs on flexible materials is an interesting avenue to pursue, which could used in biosensors directly applied to the skin or used in other applications where rigid materials would not be suitable. The liquid channel multiplexing platform is still in prototyping stages and better liquid channel methods should be investigated. Currently manual filling of liquid channels is not optimal and spontaneous filling methods should be investigated. Another issue with the liquid channel system is the time required to scan an entire chip, since we utilize a single potentiostat and scan each working electrode individually. For high levels of multiplexing this scanning time would become prohibitive, for example if it takes 2 sec per scan, and there are 1000 electrodes it would take

99 88 over 30 min to scan all sensors on a single chip. Therefore it would be invaluable to develop a method that could scan working electrodes in parallel. A possible method is to scan working electrodes in parallel where a single working electrode within each liquid channel is scanned simultaneously with different potentiostats. This would reduce scan times by the root factor of the number of electrodes which would increase the viability of this platform to highly multiplexed systems in point of care settings.

100 89 7 Publications and Other Contributions 1. Lam, B., Das, J., Holmes, R. D., Live, L., Sage, A., Sargent, E. H., & Kelley, S. O. Solution-Based Circuits Enable Multiplexed, Rapid Pathogen Detection. Nature Communications 4, 2001 (2013) 2. Lam, B., Holmes, R. D., Das, J., Poudineh, M., Sargent, E. H., & Kelley, S. O. Optimized Templates for Bottom-Up Growth of High-Performance Integrated Biomolecular Detectors. Lab on a Chip 13, (2013) 3. Kelley, S.O., Lam, B. & Sargent, E. H. Electrochemical Multiplexer. U.S. Patent 61/ (2012) 4. Lam, B., Fang, Z., Sargent, E. H. & Kelley, S. O. Polymerase Chain Reaction-Free, Sample-to-Answer Bacterial Detection in 30 Minutes with Integrated Cell Lysis. Analytical Chemistry 84, (2012). 5. Vasilyeva, E., Lam, B., Fang, Z., Minden, M. D., Sargent, E. H., & Kelley, S. O. Direct genetic analysis of ten cancer cells: tuning sensor structure and molecular probe design for efficient mrna capture. Angewandte Chemie (International ed. in English) 50, (2011). 6. Soleymani, L. Fang, Z., Lam, B., Bin, X., Vasilyeva, E., Ross, A. J., Sargent, E. H., & Kelley, S. O. Hierarchical nanotextured microelectrodes overcome the molecular transport barrier to achieve rapid, direct bacterial detection. ACS nano 5, (2011).

101 90 8 Supplementary Information 8.1 Microchannel Electrical Lysis Our goal at the beginning of my studies was to perform efficient and simple processing for direct detection of cancer and pathogenic bacteria with our nanostructured microelectrode platform. We first utilized a microchannel electrical lysis platform for lysis of bacteria and cancer cells. We were able to accurately detect the presence of pathogenic bacteria [1] and cancer [2] in unpurified lysates samples Platform Description Traditional electrical lysis methods required extremely high voltages greater than 10kV [3-4] to generate high enough electrical fields required for electrical lysis of bacteria. Wang, et al [5] developed a microchannel lysis method which reduces required voltages by roughly an order of magnitude to ~1000V. We first utilized a microchannel lysis method developed in their group (Figure 8.1) [5]. The microchannels are made using traditional soft lithographic techniques, utilizing the casting of polydimethylsiloxane (PDMS) over physical master molds which were microfabricated in house. The PDMS microchannels consist of two reservoirs connected by a channel consisting of narrow section in the middle and wide sections connecting the narrow section to the reservoirs (Figure 8.1A). Electrical contact to the microchannel is made by piecing the PDMS with Pt wire, such that the wires make contact with the interior of the reservoirs (Figure 8.1B).

102 91 Figure 8.1 PDMS microchannel electrical lysis. (A) microchannels with two reservoirs connected by microfluidic channel (b) image of actual microchannels, electrical connections are made with Pt wires piercing the PDMS making contact with the reservoirs (C) microscope image of channels This method utilizes a geometrical field advantage to reduce applied voltages (Figure 8.2). A short derivation of the field advantage is given, where the electric fields are given by (1). Using a voltage divider argument the voltage drop over each region is given by (2). Substituting into (1) we get the following expressions for the electrical fields (3). The ratio of the fields is given by (4). Therefore the ratio of the wide to narrow channels widths is equal to field advantage gained within the narrow region of the channels. Hence for channels used in this study the field advantage in the narrow section in comparison to the wide sections is roughly a factor of 6.

103 92 Figure 8.2 Microchannel schematic. Layout and dimensions for derivation of geometric field advantage.

104 93 Figure Experimental lysis setup. (A) Syringe pump (left) microchannel (center) and voltage supply (right) (B) syringe pump sample to microchannel using thin tubing, pipette tip used as sample collector and electrical connection to channels with Pt wire The experimental setup for electrical microchannel lysis is shown in Figure 8.3. The setup consists of a syringe pump and voltage power supply. The sample to be lysed is loaded into a syringe and placed within the syringe pump. The syringe is connected to the microchannel using thin tubing that is inserted into the inlet reservoir of the microchannel. Electrical connection is made to the channel with Pt wire connected to the voltage supply. As sample flows through the microchannel at a given rate, a potential is applied to the channel with the voltage supply. The lysed samples are collected with a pipette tip at the outlet for direct analysis Microchannel lysis of bacteria Bacterial samples used in these studies were Escherichia Coli (Ecoli) a standard gram negative strain and Staph Saprophyticus (Ssap) a gram positive strain, which is in general more difficult to lyse due to a thicker cell wall. Bacterial growth is performed in the appropriate media and a serial dilution method and ultraviolet-visible absorbance is used to determine concentrations of bacteria in the units of colony forming units (cfu). Initially to determine if lysis was occurring,

105 94 we performed a simple viability test. If cells do not grow on growth media plates they are dead which we used this as an initial indicator for lysis. Our first tests with were under the following conditions, utilizing a concentration of 10 8 cfu/ml Ecoli and Ssap, we flowed these samples through microchannels at a rate of 15µL/min and applied 500V to the channel. We show that Ecoli and Ssap bacteria are non-viable after this treatment (Figure 8.4) Figure Electrical lysis of Ecoli and Ssap at 15µL/min flow rate with 500V applied voltage across the microchannel To further study electrical lysis of bacterial cells in microchannels we investigated how concentration, flow rate and applied voltage would affect the lysis efficiency. Initial tests of electrical lysis were at saturated bacterial populations of ~10 8 cfu/ml. As is expected lysis of Ecoli utilizing the same conditions as above also occurs completely for lower concentrations (Figure 8.5)

106 95 Figure 8.5 Bacterial concentration versus electrical lysis. To investigate the effect of flow rate we varied the flow from 2µL/min up to 45µL/min while keeping voltage fixed at 500V (Figure 8.6). We found that flow rate has no observable influence on lysis efficiency, all bacteria lyse equally well at all flow rates. We could not go to higher flow rates since channels would burst at flow rates above 50µL/min. Figure 8.6 Flow rate versus electrical lysis Next we investigated the effect of varying voltage on lysis efficiency at a fixed flow rate of 10µL/min (Figure 8.7). We found that lysis efficiency drops off around 100V for this fixed flow rate, hence we can reduce the applied voltage to around 200V to obtain full lysis of bacteria.

107 96 Figure 8.7 Applied voltage versus electrical lysis We knew that we were killing bacteria with microchannel electrical lysis, from the simple bacterial growth tests. However, we are interested in the release of intercellular mrna that we target with our NME nucleic acid assay. Observing unlysed and lysed samples of bacteria under a microscope (Figure 8.8), it is difficult to optically determine the lysed sample or that cell walls have been breached and intercellular contents have diffused out from the cell wall. The cell looks like it remains intact and no distinct visual changes can be observed between the unlysed and lysed samples. Hence we needed an auxiliary method to determine intercellular mrna targets were being released. Figure 8.8 Optical images of lysed versus unlysed bacteria A common method to determine if the cell has been breached is to observe uptake of propidium iodide (PI) an intercalating and fluorescent molecule which intercalates with DNA. The uptake of PI only occurs if cell walls have been breached. We incubated unlysed, isopropyl alcohol (IPA) lysed and microchannel electrically lysed samples with PI at 25 µg/ml for 1 hr and observed them under the microscope (Figure 8.9). We can observe for the unlysed sample that

108 97 there is no uptake of PI. For the positive control, IPA lysed, we can observe strong emission of fluorescence from the cytoplasm of bacteria, indicating that the cell wall has been breached. For the microchannel lysed samples we can also observe good uptake of PI again indicating that the cell wall has been breached. Figure 8.9 Propidium iodide uptake versus microchannel lysis A more quantitative method to determine the degree of PI uptake is to use flow cytometry. Flow cytometry is essentially a highly quantitative high throughput method to count and simultaneously measure optical properties of individual cells. We utilize this method to count individual bacterial cells and determine the degree of PI uptake in the sample. Measurements are plotted as a histogram versus fluorescence intensity. Shown in Figure 3.10 are samples analyzed by flow cytometry and confirm no PI uptake for unlysed samples and good PI uptake for microchannel and IPA lysed bacterial samples.

109 98 Figure 8.10 Flow cytometry propidium iodide uptake Although PI uptake is a good indicator to determine if bacterial cell walls have been breached it does not technically indicate if intercellular mrna we are interested has been released. PI uptake indicates that DNA has not been released from the cytoplasm hence the question remains if mrna has been released. To confirm that mrna has been released from the cell we utilize reverse transcriptase polymerase chain reaction (RT-PCR). Briefly RT-PCR can essentially amplify a specific region of RNA within a given sample which can be visualized using standard gel electrophoresis. We performed RT-PCR on supernatants of unlysed and lysed samples (Figure 8.11). Centrifuging samples and performing RT-PCR on supernatants is necessary since cells are lysed in the thermal cycles of RT-PCR and will produce a positive signal regardless of microchannel electrical lysis. With RT-PCR we have confirmed good release of targeted mrna with microchannel electrical lysis (Figure 8.11) with no apparent difference from increased lysis voltages (Figure 3.14A & B) or unlysed uncentrifuged samples (Figure 3.14D).

110 99 Figure 8.11 RT-PCR confirmation of mrna target release on lysed samples Detection of Bacterial Lysates After confirming release of mrna targets we integrated the microchannel electrical lysis method with our NME detection platform. The process workflow is shown in Figure First we scan a pre-hybridization background signal in electrocatalytic buffer. Followed by microchannel electrical lysis of bacterial samples at 500V and 15µL/min. We collected ~50µL of lysates and directly hybridized these unpurified lysates on chip for 30min, followed by two 5min buffer washing steps. Finally we scanned the post-hybridization signal in electrocatalytic buffer and compared post and pre hybridization signals to determine the change in current (ΔI). The change in current with respect to negative controls determines if successful detection of bacteria has occurred. The entire assay is rapid enough to be performed from sample to answer within a 1hr timeframe. Figure Experimental workflow for electrochemical detection from sample to answer Electrochemical measurements of bacterial lysates were performed by Zhichao Fang [1]. Utilizing the microchannel lysis platform Fang showed that we could accurately determine the

111 100 species of bacteria from raw unpurified lysates at clinically relevant concentrations with our NME platform. An important issue for nanomaterial sensors is the size of the sensor footprint has to be sufficiently large to interact/capture large targets like mrna within a reasonable timeframe (Figure 8.13). Using a simple diffusion model towards a hemispherical sensor the accumulation time based upon the diffusion of the molecule can be calculated [1]. For small nucleic acid molecules, ~20 base pairs in length at 2fM, a clinically relevant concentration, accumulation times are reasonable down to 1µm sized sensors. However for larger molecules, ~4000 base pairs which diffuse much slower, accumulation times are only reasonable for sensors with footprints larger than 100µm [1]. Figure Sensor footprint versus accumulation time for 20-mer and 4000-mer nucleic acid molecule at 2fM. [1] To test our accumulation times we fabricated NME with of three different sizes (10, 30, and 100µm) shown in Figure 3.17 to show that 100µm sensors are necessary to detect Ecoli lysates at clinically relevant concentrations of 1.5cfu/µL which corresponds to approximately 2fM. Following the detection protocol outlined in Figure 8.12 with a hybridization time of 30min we found that 100µm footprint sensors were the only sensors capable of detecting 1.5cfu/µL of Ecoli lysate (Figure 8.14C). Smaller footprint sensors (30 and 10µm) are too small to interact with large mrna target molecules, and hence unsuccessful at 1.5cfu/µL, but capable of detecting samples at 100 fold higher concentration with the same hybridization time.

112 101 Figure Electrochemical detection of bacterial lysates for different footprint sensors. (A-B) SEM images of small to large NME sensors utilized (C) measurement results for low and high bacterial loading [1] (Bacterial lysis and sample preparation: Brian Lam, SEM analysis: Leyla Soleymani, Electrochemical measurements: Leyla Soleymani and Zhichao Fang) Microchannel electrical lysis of leukemia cells In parallel to the bacterial detection project we were to developing a detection method for chronic myelogenous leukemia (CML), a specific cancer of white blood cells. CML is characterized by a specific gene fusion between chromosome 9 and 22 in the ABL and BCR regions and is named the Philadelphia chromosome (Figure 8.15). This specific BCR:ABL gene fusion was our target molecule in our NME nucleic acid assay. Again for real samples we needed a simple streamlined method to release this intercellular target. To prove the multiple utility of the microchannel electrical lysis platform we proceeded to lyse CML positive cells for analysis with our NME detection platform.

113 102 Figure 8.15 Chronic mylegenous leukemia. Characterized by a specific gene fusion between BCR:ABL. Initial testing was with the K562 cell line which is positive for CML and the BCR:ABL gene fusion. Initial lysis testing with K562 cells is shown in Figure Samples of K562 cells containing 10 5 cells/ml in PBS buffer were loaded into syringes and placed in the syringe pump as in Figure 3.4. Control samples were flowed through the microchannel at a rate of 20µL/min with 0V applied to the channel (Figure 8.16A) where no observable lysis occurs. When a potential of 500V is applied, we observe total lysis of cells (Figure 8.16B) with no intact cells and only cell debris visible in the collected lysate. Following the same experimental workflow as shown in Figure 8.12 we analyzed K562 lysates using our NME platform. Electrochemical detection of K562 lysates in buffer is shown in Figure 8.16C. We show that we can detect down to 10 cells and up to 1000 cells in a sample volume of 30µL versus negative control probes. We were also successful in using this method to lyse samples of blood spiked with K562 and positive CML leukocyte patient samples, and accurately determine the presence of the BCR:ABL gene fusion with our NME detection platform [2]

114 103 Figure Microchannel lysis of K562 cells a positive BCR:ABL gene fusion cell line. (A) K562 cells through at 0V and (B) 500V. (C) Electrochemical detection of K562 lysates (Elizaveta Vasilyeva) with NME platform [2] (Electrochemical measurements: Elizaveta Vasilyeva) Conclusions Sample handling and process is an important aspect of any point of care diagnostic technique and must be simple, rapid and effective. We showed that the microchannel lysis platform was a versatile, rapid and effective platform for lysis of bacteria cells and mammalian cells. We were able to detect intercellular genetic markers of bacteria and cancer cells by combining microchannel lysis with our NME detection platform. Remarkably we were able to perform detection in crude lysates without any purification.

115 Description of Solution Circuit Chip An important goal for biosensors is the multiplexed analysis of samples with a single device. For many applications highly multiplexed analysis would provide valuable utility providing accurate analysis on a panel of relevant biomarkers. We developed a solution circuit chip (SCC) which relies on physical separation of liquid to create electrochemical isolation. A description of the SCC platform and further characterization of interference is given Description of SCC platform The core of the SCC method utilizes an array format addressing for working, counter and reference electrodes. An illustration of this format is shown in Figure 8.17 with a 5x5 SCC array. The array consists of 5 common WEs with 5 NMEs addressed in parallel on each common WE. Orthogonal to the common working electrodes are 5 RE and CE pairs deposited on the surface of the aperture passivating layer. A cross section for the configuration near a single NME looking down the RE/CE pairs is shown in Figure This illustrates the underlying common WE, passivating SU-8 aperture layer, NME and surface RE/CE pairs. Figure 8.17 Representation of 5x5 array of an SCC chip

116 105 Figure 8.18 Cross section schematic of single NME in SCC chip If the common WEs were connected in a traditional electrochemical setup where the entire SCC chip was immersed in an electrochemical solution, represented in Figure 8.19, the connected NMEs would act as a single WE. Hence if the common WE was placed under bias we would be interrogating the 5 connected NMEs at the same time without the ability to address individual NME on the same common WE. Figure 8.19 SCC chip connected in traditional electrochemical setup where all NMEs are immersed in the same electrochemical solution. The core of our SCC technique takes advantage of the simple fact the physical separation of liquid can result in electrochemical isolation. Therefore, the concept is to physically separate the electrochemical solution into defined liquid channels orthogonal to the common WEs, which is illustrated in Figure These liquid channels are defined along the RE/CE surface pair electrodes. With this configuration we now have electrochemical isolation between NMEs fabricated on the same common WE.

117 106 Figure 8.20 Physical liquid channel separation orthogonal to common WEs. An illustration of the addressing approach is given in Figure To address a given NME within the SCC chip we bias the desired liquid channel RE/CE pair and the corresponding WE of interest. For example if we bias the first RE/CE row and second common WE, we will only interrogate the NME located on the first row and second column (Figure 8.21A). Switching the array to another NME can easily be done by biasing the appropriate RE/CE row and common WE. For instance if we would like to switch to interrogate the NME located on the third row and fourth column (Figure 8.21B) we would simply bias the third RE/CE pair and fourth common WE. Figure 8.21 Illustration of SCC method (A) interrogation of single NME on first row second column (B) interrogation of NME on third row fourth column

118 Interference evaluation It is important to evaluate of electrochemical interference between liquid channels since interference could affect the outcome. We performed similar sequential addition of electrochemical reagents (Ru/Ferri) to wells while measuring the same NMEs after each sequential addition. Initially Ru/Ferri is loaded into the first liquid channel (Figure 8.22A), followed by a DPV scans on NMEs within the first channel. Second, Ru/Ferri is added to the second liquid channel (Figure 8.22B), followed by a DPV scans on NMEs within the first channel. This sequence is repeated until all liquid channels have been filled with Ru/Ferri. The resultant DPVs are plotted in Figure 8.22C. We observe significant crosstalk from the sequential addition of Ru/Ferri. We surmised that the increased level of interference may be due to the increased number of liquid channels and NMEs compared to the initial parallel test chips. Figure 8.22 SCC interference from sequential addition of Ru/Ferri (A) addition of Ru/Ferri to first channel (B) subsequent addition of Ru/Ferri (C) DPV scans of same NME after sequential additions of Ru/Ferri Our hypothesis was that the NMEs that we were not interested in on the common WE under bias were shorting through adjacent NMEs back to the liquid channel of interest. To illustrate our hypothesis a schematic of our theory is shown in Figure For this example we are trying to

119 108 select the NME in the third row and second column highlighted by a red star. Therefore we place the second column WE and third row RE/CE pair under bias. The second column WE under bias has multiple NMEs connected which we theorize are shorted through the solution, with a solution resistance R s, to adjacent neighboring NMEs in liquid channels not under bias. In turn these neighboring NMEs are shorted through the common WE to the NME within the liquid channel of interest (green arrows). Hence we surmise neighboring NMEs in liquid channels not under bias act as pseudo RE/CE electrodes which results in interference from NMEs we are not interested in. Figure 8.23 Hypothesis of interference phenomena in SCC chips. NMEs we are not interested in on common WE under bias short circuit through solution (Rs) to adjacent NMEs which act as pseudo RE/CE since adjacent NMEs are shorted to an NME within liquid channel under bias To test our interference theory we performed sequential addition of Ru/Ferri combined with a sequential plating of adjacent NMEs outlined in Figure A first sequential addition experiment was performed with only a single common WE electroplated with NMEs. DPV measurements were made on the single NME (red star) for each sequential addition of Ru/Ferri and plotted in Figure 8.24B. There is no significant interference when only a single common WE

120 109 is plated with NMEs, which suggests that adjacent NMEs are responsible for measured interference. To further confirm this finding we plate a second row of NMEs and perform the same sequential addition experiment where DPV measurements made on the same NME are plotted in Figure 8.24C. After plating a second WE we now observe interference arising in scanned DPV signals. The trend of increasing interference continues as a third (Figure 8.24D) and fourth (Figure 8.24E) WE are plated and DPV measurements are made. This confirms that interference is due to adjacent NMEs that are not under bias as our theory predicts. Figure Interference theory experiment. (A) Subsequent addition experiment perform after subsequent plating of additional WE (B E) DPV measurements for subsequent addition experiments after plating 1-4 WEs. Following these findings we explored if it was possible to eliminate the significant interference arising from adjacent NMEs. One such theory was if an electrode was held at a fixed potential, such as ground, where no electrochemical reactions occur, no interfering signals could travel through the electrode. To test this theory we performed the same sequential addition experiment comparing non-grounded versus grounded adjacent NMEs (Figure 8.25). All significant interference is eliminated when adjacent NMEs are grounded (Figure 8.25C). This confirms that holding adjacent NMEs at fixed potentials eliminate any significant channel to channel interference. Here we showed that crosstalk between channels can be eliminated by holding all other WEs, that are not of interest at a fixed ground potential.

121 110 Figure 8.25 Elimination of interference (A) Schematic of sequential addition experiment with grounded adjacent NMEs. Sequential addition experiment with (A) non-grounded and (B) grounded adjacent NMEs.