Continuous-flow microfluidic printing of proteins for array-based applications including surface plasmon resonance imaging

Size: px

Start display at page:

Download "Continuous-flow microfluidic printing of proteins for array-based applications including surface plasmon resonance imaging"

Doreen Johnson
6 years ago
Views:

1 Available online at Analytical Biochemistry 373 (2008) ANALYTICAL BIOCHEMISTRY Continuous-flow microfluidic printing of proteins for array-based applications including surface plasmon resonance imaging Sriram Natarajan a, Phini S. Katsamba b, Adam Miles c, Josh Eckman d, Giuseppe A. Papalia b, Rebecca L. Rich b, Bruce K. Gale d, David G. Myszka b, * a Department of Chemical Engineering, University of Utah, Salt Lake City, UT 84132, USA b Center for Biomolecular Interaction Analysis, University of Utah, Salt Lake City, UT 84132, USA c Department of Materials Science and Engineering, University of Utah, Salt Lake City, UT 84132, USA d Department of Mechanical Engineering, University of Utah, Salt Lake City, UT 84132, USA Received 18 July 2007 Available online 19 August 2007 Abstract Arraying proteins is often more challenging than creating oligonucleotide arrays. Protein concentration and purity can severely limit the capacity of spots created by traditional pin and ink jet printing techniques. To improve protein printing methods, we have developed a three-dimensional microfluidic system to deposit protein samples within discrete spots (250-lm squares) on a target surface. Our current technology produces a 48-spot array within a cm target area. A chief advantage of this method is that samples may be introduced in continuous flow, which makes it possible to expose each spot to a larger volume of sample than would be possible with standard printing methods. Using Biacore TM Flexchip (Biacore AB) surface plasmon resonance array-based biosensor as a chip reader, we demonstrate that the microfluidic printer is capable of spotting proteins that are dilute (<0.1 lg/ml) and contain high concentrations of contaminating protein (>10,000-fold molar excess). We also show that the spots created by the microfluidic printer are more uniform and have better-defined borders than what can be achieved with pin printing. The ability to readily print proteins using continuous flow will help expand the application of protein arrays. Ó 2007 Elsevier Inc. All rights reserved. Keywords: Biacore; Protein protein interaction; Affinity; Kinetics Pin printing is a standard method of producing oligonucleotide arrays for expression profiling [1 3]. However, a challenge with pin printing is that the resulting spot depends on many factors, including pin tip size, slit size (for quill pins), time of contact with substrate, and solution and substrate physical properties [4]. Substrate variables such as surface energy, pin geometry, atmospheric humidity, temperature, and pin velocity also have roles in the spotting outcome. Spot morphology and uniformity are further affected by the mechanism of drop drying on a surface. The coffee ring and comet tailing phenomena are two commonly observed artifacts of pin spotting that reduce spot quality. Pin spotting becomes even more challenging when it comes to printing protein solutions, which vary in viscosity, surface tension, and density [5]. An even more fundamental problem with printing proteins is that many proteins of interest are not available at high enough concentrations or are not pure enough to allow for direct printing. Together, these factors have limited the application of protein arrays, particularly in the area of surface plasmon resonance (SPR) 1 -based imaging systems. * Corresponding author. Fax: address: dmyszka@cores.utah.edu (D.G. Myszka). 1 Abbreviations used: PBS, phosphate-buffered saline; SPR, surface plasmon resonance; PDMS, polydimethylsiloxane; CFM, continuous-flow microspotter; BSA, bovine serum albumin /$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi: /j.ab

2 142 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) To address the problem of protein printing, we engineered a three-dimensional microfluidic array-based system that allows one to spot proteins in a defined area while under continuous flow. This system uses microfluidic channel networks embedded in polydimethylsiloxane (PDMS) combined with a unique liquid-handling interface to flow solutions over a substrate area. To achieve individual spot addressing, each spot interface is linked to fluidic ports that allow fluid to circulate across the spotted area. At the printhead face (where the microchannel contacts the substrate), the protein may be adsorbed to the substrate directly or captured by a recognition molecule. Fig. 1A illustrates the design of a continuous-flow microspotter (CFM) system, which consists of 48 sample and 48 collection wells laid out in a 96-well plate format for easy sample transfer and recovery. Sample solutions are cycled between the wells through a series of microchannels. These microchannels converge in a print head to produce a two-dimensional 4 12 array, as shown at the bottom of Fig. 1B. Each array spot is sealed on the substrate by the bulk PDMS surrounding it. This seal confines the solution to only the area of the individual spots. During operation, the CFM is mounted on a platform with a machined polycarbonate manifold placed over it. An integrated circuit board is used to control the direction and speed of the samples as they are pumped through the CFM by applying either a positive or a negative pressure. To demonstrate the benefits of the CFM system, we compared the results of spotting both purified and crude protein samples using the pin spotting and the continuous-flow approaches. We demonstrate that the continuous-flow microspotter produces spots that are significantly more uniform and have better-defined edges than those obtained by pin spotting. The continuous-flow system is also able to produce reliable spots at 100-fold lower protein concentrations than pin spotting and can capture samples from a crude solution in which the background protein milieu is 10,000 times more concentrated than the target protein. These improvements in protein spotting should expand the application of both fluorescent- and SPR-based protein arrays. Materials and methods CFM fabrication The CFM was cast from PDMS as described previously [6]. The fluid channels are 100 lm wide and the open end of each cavity in the print head is a 250 by 250 lm square. During operation, the CFM is mounted on a platform with a polycarbonate manifold placed over it. Using this manifold, the instrument is able to pump all the channels at once by applying either a positive or a negative pressure. Each flow loop in the CFM is connected to two wells, one for sample loading (outside the manifold) and one for sample collection (inside the manifold). By alternating negative and positive pressures, samples can be cycled back and forth across the substrate between the two connected wells. The 48 sample and recovery wells have a maximum volume of 75 ll each. Pretreating CFM printers The CFM channels and printing face were treated with a pluronic F108NF prill surfactant (BASF) to prevent protein adsorption on the PDMS channel surface. The channels were coated by filling them with 0.5 mg/ml pluronic surfactant and allowed to sit overnight. The pluronic surfactant was then flushed out of the channels with deionized water and the print head was immersed in 50 mg/ml pluronic surfactant for 1 h. The channels and print head were again rinsed with deionized water to remove any residual surfactant solution before printing. CFM and pin printing Protein printing studies were done on Flexchip substrates, which are 1 3 inch gold-coated slides obtained from Biacore AB. These chips have an mm area for SPR analysis. For each experimental run, the CFM print-head face was brought into contact with the substrate and then compressed another 500 lm to form a tight seal with the substrate. Approximately 70 ll of solution was loaded into each well and the manifold was placed over Fig. 1. (A) 48-spot CFM with integrated wells identical to those on a 96-well plate. (B) Print-head face that creates a seal around each spot (4 12 array) once pressed onto the substrate.

3 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) the collection wells. An electronically controlled pump with a custom control system and interface program controlled the applied pressure, i.e., the flow rate. A high initial negative pressure was applied to the manifold to overcome surface tension and fill the channels with solution and then the pressure was adjusted to provide the desired deposition flow rates. The samples were cycled through the CFM spotter for 15 min at a flow rate of 60 ll/min. The flow direction was cycled back and forth through the duration of each run until the total run time had completed, at which time a positive pressure emptied the channels and the solution was pipetted from the wells. Finally, a wash solution of phosphate-buffered saline (PBS) was loaded into each well and flushed through the channels at 10 ll/min for 5 min to clear any residual protein solution. Protein samples were pin printed using a QArray Mini array system (Genetix). To minimize spot variation, a single solid pin was used to print each sample. The sample ink and stamp times were set at 100 and 300 ms, respectively. Between samples, the pin was washed for 10 s each with a solution of 0.05% SDS followed by water before being dried for 10 s with a stream of compressed air. Protein A printing Pin and CFM printers were used to deposit Protein A (Pierce Chemical Co.) on the same plain gold Flexchip slide. Twelve different concentrations (250 to 0.12 lg/ml) of Protein A in PBS were printed. Each concentration of Protein A was printed four times to assess reproducibility. Antilysozyme antibody printing Mouse antilysozyme mab (BGN/0696/2B10, AbD Serotec) was printed onto a prototype dextran-coated Flexchip slide, which contained an amine-coupled anti-mouse RAM. The antilysozyme antibody was printed at four concentrations (10, 2, 0.4, and 0.08 lg/ml) in eight PBS solutions containing bovine serum album (BSA) in a threefold dilution series from 13 lg/ml to 10 mg/ml. Flexchip imaging Once the pin and CFM printing steps were complete, the flow cell gasket and cover window were assembled over the printed region and the slide was docked into the Flexchip instrument. The chip was filled with running buffer (PBS supplemented with 0.005% Tween 20 and 0.1 mg/ml BSA). The chip was then imaged using the Flexchip optics and the regions of interest (protein spots and interstitial reference positions) were defined within the software. Analytes were injected at a flow rate of 300 ll/min for 8 min, followed by a 8-min dissociation phase in which the surface was washed with running buffer. Bovine IgG (Sigma) was injected over the Protein A slide at a concentration of 100 nm. Lysozyme (AbD Serotec) was injected over the antilysozyme slide at a concentration of 20 nm. Binding experiments were performed at 25 C. The Flexchip response data were exported from the instrument software and analyzed using Scrubber 2 (Biologic Software Pty. Ltd., Australia). Results Spot morphology To illustrate some of the benefits of the CFM printing approach, we compared the results of printing Protein A onto a gold surface using pin spotting and the CFM system. Twelve different concentrations of Protein A, from a high concentration of 250 lg/ml down to a low concentration of 0.12 lg/ml, were printed using both techniques. Once assembled into a flow cell, the chip surface was wetted and imaged within the Biacore Flexchip system (Fig. 2). The first observation is that the pin-printed spots (Fig. 2A) show poor spot morphologies compared to the CFMprinted spots (Fig. 2B). The area within the spots are clearly nonuniform when protein was pin printed at concentrations less than 3 lg/ml (Fig. 2C) and the spots show significant spreading at concentrations greater than 10 lg/ ml (Fig. 2D). In contrast, the spots created with the CFM system appear uniform throughout the spot area (even at low protein concentrations; Fig. 2E) and show no evidence for spreading even at the highest protein concentrations that were printed (see for example Fig. 2F). Spot capacity Next we compared how well the different Protein A spots could capture bovine IgG. A mixture of bovine gamma globulins was injected over the Protein A spots at a concentration of 100 nm for 8 min and the dissociation phase was monitored for 8 min. Figs. 3A and 3B shows an overlay of the binding responses from the pin- and CFM-printed Protein A spots, respectively. The shapes of the binding responses are similar in the pin- and CFMspotted samples. This indicates that the two printing methods do not change the kinetic binding rates for the Protein A/IgG reaction. A gradient in the magnitude of the IgG binding response is observed in both data sets, which corresponds to the different levels of Protein A that have been spotted onto the chip surface. We used the response value at the end of the association phase to quantitate the amount of IgG bound at each spot. Fig. 4 shows the IgG response values plotted verses the printing concentration of Protein A. The pin-printed samples show no significant binding response for the IgG until the concentration of Protein A reaches 10 lg/ml. In contrast, we were able to detect the binding of Protein A at even the lowest concentration of Protein A that was spotted (0.12 lg/ml). These results demonstrate that the minimum concentration of Protein A required for CFM printing is at least 100 times less than that required for pin printing. Also, the observed signals from CFM-printed

144 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) 141 146 Fig. 2. Protein A chip images.

48 (C) and 31 lg/ml (D) Protein A. Selected CFM-printed spots of 0.12 (E) and 250 lg/ml (F) Protein A. Fig. 3. Flexchip sensorgrams for bovine IgG (100 nm) injected over the pin- (A) and CFM- (B) printed Protein A surfaces.

4 144 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) Fig. 2. Protein A chip images. Protein A was printed in a twofold dilution series from 0.12 to 250 lg/ml (left to right in image) using pin (A) and CFM (B) spotting. Selected pin-printed spots of 0.48 (C) and 31 lg/ml (D) Protein A. Selected CFM-printed spots of 0.12 (E) and 250 lg/ml (F) Protein A. Fig. 3. Flexchip sensorgrams for bovine IgG (100 nm) injected over the pin- (A) and CFM- (B) printed Protein A surfaces. spots show a linear increase in response with the log increase in Protein A concentration until >100 lg/ml, where the capacity of the CFM-printed spots appears to level off, suggesting that the surface has reached saturation. The pin-printed spots above 100 lg/ml appear to lose capacity, which is likely the result of Protein A spreading into the reference spot regions that are interstitial to the reaction spots (this spreading is clearly visible in Fig. 2A).

5 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) Fig. 4. Observed maximum response of bovine IgG verses the printing concentration of Protein A. Circles and squares represent response data from the CFM- and pin-printed spots, respectively. The error bars represent the experimental standard deviation determined from the replicate spots. Capturing from crude samples To compare the performance of pin and CFM printing in a capturing assay, we tested the ability of each method to extract a mouse antilysozyme monoclonal antibody from an impure solution using an anti-mouse surface. The antilysozyme mab was printed at four concentrations (10, 2, 0.4, and 0.08 lg/ml). To simulate crude samples, BSA was added to the mab samples from a concentration of 13 lg/ml up to 10 mg/ml. Lysozyme (20 nm) was then injected over the pin and the CFM-printed antilysozyme chip was docked within the Flexchip instrument. Fig. 5 shows an example of the binding responses observed for lysozyme from the four concentrations of antilysozyme when spotted without BSA in the sample. The responses from the pin-printed antilysozyme samples are very low (Fig. 5A). In contrast, lysozyme binding to the four different antilysozyme concentrations spotted by CFM printing (Fig. 5B) is easily observed even at the lowest spotting concentration (0.08 lg/ml). Fig. 6 shows the maximum response observed for the different antilysozyme and BSA conditions produced from CFM printing. The presence of very high concentrations of BSA during the printing step appears to have only a small effect on reducing the capture level of the antilysozyme mab. This illustrates that it is possible to capture mabs both when they are at low concentrations (<0.1 lg/ml) and in the presence of high amounts of a contaminating protein such as BSA (10 mg/ml). It appears that low concentrations of BSA actually enhance the capturing levels of mab compared to spots when no BSA was present. In this case, BSA is likely acting as a carrier protein, helping to deliver a more consistent concentration of the mab through the CFM printer. Discussion Protein arrays are gaining interest for both research and diagnostic applications. However, the standard methods for printing oligonucleotide arrays do not translate well for proteins. Pin and ink jet printing often lead to problems of protein denaturation during spotting or as samples are dried down on the substrate surface. Also, these printing methods transfer only a limited volume of sample, which means that the protein itself needs to be relatively concentrated and pure to obtain efficient deposition. And if the concentration is too high, dried sample that is not absorbed to the surface may spread once the spot is rewetted. Fig. 5. Lysozyme/antilysozyme interaction. Lysozyme (20 nm) was flowed over an anti-mouse Flexchip surface containing antilysozyme mab printed at 10, 2, 0.4, and 0.08 lg/ml using pin (A) or CFM (B) printing. Red lines represent a global fit of the response data to a 1:1 interaction model.

6 146 Continuous-flow microfluidic printing of proteins / S. Natarajan et al. / Anal. Biochem. 373 (2008) Response (RU) μg/ml 2 μg/ml 0.4 μg/ml 0.08 μg/ml BSA concentration (mg/ml) Fig. 6. Observed maximum response of lysozyme binding to antilysozme spots created by CFM printing. Antilysozyme was printed at concentrations of 10, 2, 0.4, and 0.08 lg/ml. Each antilysozyme concentration was printed in the presence of BSA from 13 lg/ml to 10 mg/ml and in buffer with no BSA. Lysozyme was injected over the antilysozyme spots at a concentration of 20 nm. The response values for each condition were determined at the end of the lysozyme association phase. To improve the procedure of protein printing, we have developed a microfluidic device that allows one to flow a sample over a defined area. Continuous flow has the advantages that it can deliver larger volumes of sample and the sample solution may be recirculated to produce high-density spots. After the sample has been deposited, the spot can be washed with a buffer solution to remove unwanted bulk material, which further minimizes spot spreading. The CFM printer also eliminates many of the other limitations associated with pin spotting and inkjet printing by isolating each array spot in its own microenvironment. Using Biacore s Flexchip SPR imaging system, we demonstrated the benefits of the CFM printing compared to pin printing of proteins. In one series of experiments, we showed that it was possible to adsorb a purified protein onto a gold surface at a concentration that was 100-fold lower than what was required with pin printing. This is made possible by the ability to deliver a larger volume of sample over the spot for a longer period of time. The CFM-printed spots were also highly uniform in morphology (even at low protein concentrations), whereas pin printing often led to the coffee-ring effect as the printed sample dried on the target. The CFM-printed spots also showed no evidence of spot spreading or comet tailing, as is often observed for pin-printed samples. This is because the CFM-printed spots could be washed with buffer to remove any unadsorbed protein while the CFM print head is still in place. Using a lysozyme/antibody system, we demonstrated that it was possible to specifically deposit an antibody onto a capturing surface at antibody concentrations less than 1 lg/ml and with a background protein concentration of 10 mg/ml. This illustrates that the CFM printing method could be a valid approach to capturing crude mabs for screening studies. In our analysis, we found that the background protein (BSA) had a minimal effect on our ability to capture significant levels of the mab of interest. In fact, it appeared that low concentrations of BSA improved the levels of mab that we were able to print, especially at low mab concentrations. It is likely that the BSA serves as a carrier protein to minimize the nonspecific adsorption of the mab onto the microfluidic channels, thereby delivering a higher concentration of mab to the target surface. In its current configuration, our CFM print head can deliver samples with volumes of 10 to 75 ll, but wells with larger volumes could be easily developed. The device used in this report contained 48 individually addressed spots. We are developing a 96-spot system that will print all spots within in a 1-cm area. Future designs with 384 spots will likely require a larger printing area to accommodate the increase in the number of the fluidic channels. Another advantage of the CFM printer is that it may be possible to deliver a series of reagents to the substrate to chemically modify spots on the surface. In this case the chemical reactions may be run in parallel, which should further improve the consistency of the results compared to pin printing. Acknowledgement This work was supported by funding from the National Science Foundation (EF to D.G.M.) and by the State of Utah Center of Excellence Program. References [1] J. DeRisi, L. Penland, P.O. Brown, M.L. Bittner, P.S. Meltzer, M. Ray, Y. Chen, A. Su, J.M. Trent, Nat Genet 14 (1996) [2] M. Chee, R. Yang, E. Hubbell, A. Berno, X.C. Huang, D. Stern, J. Winkler, D.J. Lockhart, M.S. Morris, S.P. Fodor, Science 274 (1996) [3] J.L. DeRisi, V.R. Iyer, P.O. Brown, Science 278 (1997) [4] M.O. Reese, R.M. van Dam, A. Scherer, S.R. Quake, Microfabricated fountain pens for high-density DNA arrays, Genome Research 13 (2003) [5] T. Martinsky, Protein Microarray Manufacturing, PharmaGenomics (2004) [6] David A. Chang-yen, David Myszka, Bruce K. Gale, J Micromech Syst 15 (2006)

Label-free interaction analysis in realtime using surface plasmon resonance

GE Healthcare Technology Note 23 Biacore systems Label-free interaction analysis in realtime using surface plasmon resonance Providing quantitative data on: report point Specificity sensorgram To what