Anal. Chem. 2007, 79, 7360-7366
Microfabricated Two-Dimensional Electrophoresis Device for Differential Protein Expression Profiling Charles A. Emrich,‡ Igor L. Medintz,§ Wai K. Chu,† and Richard A. Mathies*,†
Department of Chemistry and Biophysics Graduate Group, University of California, Berkeley, California 94720, and Center for Bio/Molecular Science and Engineering, U.S. Naval Research Laboratory, Washington, DC 20375
A microfluidic separation system is developed to perform two-dimensional differential gel electrophoretic (DIGE) separations of complex, cellular protein mixtures produced by induced protein expression in E. coli. The micro-DIGE analyzer is a two-layer borosilicate glass microdevice consisting of a single 3.75 cm long channel for isoelectric focusing, which is sampled in parallel by 20 channels effecting a second-dimension separation by native electrophoresis. The connection between the orthogonal separation systems is accomplished by smaller channels comprising a microfluidic interface (MFI) that prevents media leakage between the two dimensions and enables facile loading of discontinuous gel systems in each dimension. Proteins are covalently labeled with Cy2 and Cy3 DIGE and detected simultaneously with a rotary confocal fluorescence scanner. Reproducible two-dimensional separations of both purified proteins and complex protein mixtures are performed with minimal run-to-run variation by including 7 M urea in the second-dimension separation matrix. The capabilities of the micro-DIGE analyzer are demonstrated by following the induced expression of maltose binding protein in E. coli. Although the absence of sodium dodecyl sulfate (SDS) in the second-dimension sizing separation limits the orthogonality and peak capacity of the separation, this analyzer is a significant first step toward the reproducible two-dimensional analysis of complex protein samples in microfabricated devices. Furthermore, the microchannel interface structures developed here will facilitate other multidimensional separations in microdevices. Miniaturized analysis systems have made possible huge advances in speed and parallelism for the analysis of DNA,1 but similar payoffs for protein analysis have been slow to arrive because of the unique challenges inherent to protein separations.2 Unlike DNA, which is soluble, rigid, and strongly charged, proteins encompass a daunting range of solubilities, sizes, native charges, and importantly, are present over a huge range of concentrations (4-6 orders of magnitude in S. cerevisiae cells, * To whom correspondence should be addressed. Phone: 510-642-4192. Fax: 510-642-3599. E-mail:
[email protected]. † Department of Chemistry, University of California. ‡ Biophysics Graduate Group, University of California. § U.S. Naval Research Laboratory. (1) Dittrich, P. S.; Tachikawa, K.; Manz, A. Anal. Chem. 2006, 78, 3887-3907. (2) Freire, S. L. S.; Wheeler, A. R. Lab Chip 2006, 6, 1415-1423.
7360 Analytical Chemistry, Vol. 79, No. 19, October 1, 2007
and up to 12 logs for plasma).3 It has been estimated that in a yeast cell, half of the protein mass is the product of only 100 genes,3 meaning that most proteins and their isoforms are underrepresented. Without a PCR-like technique to amplify scarce samples, protein analysis techniques have to be highly efficient and have a wide dynamic range, making the miniaturization of protein analysis systems inevitable. Current techniques for probing the proteome all employ some type of multidimensional separation, often with prefractionation,4 so a clear picture of the proteins and their isoforms can be drawn. The oldest and still-benchmark standard, two-dimensional electrophoresis (2DE),5 uses two sequential gel-based systems to separate proteins first by their isoelectric points (pI) via isoelectric focusing (IEF) and then according to mass by denaturing electrophoresis. 2DE is often followed by mass spectrometry, which can unambiguously identify proteins and their posttranslational modifications. Mass spectrometry (MS) performs the same back-end role for multidimensional liquid chromatography separations of proteins, such as the strong-cation-exchange/ reversed-phase separations that are the mainstay of shotgun proteomics.6,7 Protein microarrays offer an alternative to multidimensional separations but, because of inherent complexities of construction,8 have largely been relegated to probing class-specific protein binding interactions. Understanding how protein expression levels change in response to stimuli, environment, or cell type requires quantitation of those protein levels and can be accomplished by isotope-tagging strategies for MS9 and by 2D (two-dimensional) differential gel electrophoresis (DIGE) which uses fluorescent multiplexing of pooled samples.10 Microfabricated analysis systems offer a low-volume alternative to macroscale techniques, enabling fast analysis of nucleic acids,11 (3) Herbert, B. R.; Harry, J. L.; Packer, N. H.; Gooley, A. A.; Pedersen, S. K.; Williams, K. L. Trends Biotechnol. 2001, 19, S3-S9. (4) Righetti, P. G.; Castagna, A.; Herbert, B.; Reymond, F.; Rossier, J. S. Proteomics 2003, 3, 1397-1407. (5) O’Farrell, P. H. J. Biol. Chem. 1975, 250, 4007-4021. (6) Swanson, S. K.; Washburn, M. P. Drug Discovery Today 2005, 10, 719725. (7) Washburn, M. P.; Wolters, D.; Yates, J. R. Nat. Biotechnol. 2001, 19, 242247. (8) Kung, L. A.; Snyder, M. Nat. Rev. Mol. Cell Biol. 2006, 7, 617-622. (9) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994-999. (10) Unlu, M.; Morgan, M. E.; Minden, J. S. Electrophoresis 1997, 18, 20712077. (11) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 1134811352. 10.1021/ac0711485 CCC: $37.00
© 2007 American Chemical Society Published on Web 09/07/2007
tight integration of sample processing and analysis steps,12 and the creation of high-density arrays of channels for massively parallel analysis.13 On-chip protein analysis should also benefit greatly from these advances, but with most published studies demonstrating multidimensional separations only of purified proteins,14-16 the technology is still in its infancy. Microfluidic designs for complex protein analyses generally use a single channel for the first-dimension separation and then either use serial or parallel sampling of the first-dimension channel to further resolve analytes. Simple serial second-dimension sampling has been demonstrated for single-cell 2DE using conventional capillaries coupled end-to-end17 but has only proven effective in chip format for separations of peptide mixtures18 and high-concentration mixtures of purified proteins.19 Parallel sampling uses an array of second-dimension channels to analyze the separated species from the first-dimension making it a good match for steady-state techniques like IEF and leveraging the ability of microfabrication to easily create high-density channel arrays. Building on an early prototype design,20 parallel-sampling 2DE systems have been demonstrated but never for the kinds of complex protein mixtures such as cell lysates that would enable on-chip proteomics. The peak capacity for any 2D separation is the product of the peak capacities of each individual dimension, provided that the dimensions are orthogonal. In conventional 2DE systems, this means careful choice of separation parameters is required, and in a microfluidic system it also means that the contents of the two dimensions should not mix. This has been accomplished by physically isolating and then assembling the dimensions,21,22 by using valves,16 and most often by directly connecting the dimensions.14,15 Here we present a new approach for performing 2D separations where the first- and second-dimension channels are connected with passive valve structures consisting of much shallower, narrower channels that form a microfluidic interface (MFI). The MFI acts as a barrier to fluid flow and diffusion-driven transport as a result of its much reduced channel cross section. The unique attribute of this format is that it allows bubble-free loading of different gel systems in each dimension in a straightforward process that maximizes the flexibility of each separation dimension. The reproducibility of 2DE separations, found to be extremely sensitive to the second-dimension electrophoresis conditions, was maximized by the inclusion of 7 M urea in run buffers and by decreasing the second-dimension separation field strength. (12) Blazej, R. G.; Kumaresan, P.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 7240-7245. (13) Emrich, C. A.; Tian, H. J.; Medintz, I. L.; Mathies, R. A. Anal. Chem. 2002, 74, 5076-5083. (14) Herr, A. E.; Molho, J. I.; Drouvalakis, K. A.; Mikkelsen, J. C.; Utz, P. J.; Santiago, J. G.; Kenny, T. W. Anal. Chem. 2003, 75, 1180-1187. (15) Li, Y.; Buch, J. S.; Rosenberger, F.; DeVoe, D. L.; Lee, C. S. Anal. Chem. 2004, 76, 742-748. (16) Wang, Y. C.; Choi, M. N.; Han, J. Y. Anal. Chem. 2004, 76, 4426-4431. (17) Harwood, M. M.; Christians, E. S.; Fazal, M. A.; Dovichi, N. J. J. Chromatogr., A 2006, 1130, 190-194. (18) Rocklin, R. D.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 52445249. (19) Shadpour, H.; Soper, S. A. Anal. Chem. 2006, 78, 3519-3527. (20) Becker, H.; Lowack, K.; Manz, A. J. Micromech. Microeng. 1998, 8, 24-28. (21) Chen, X. X.; Wu, H. K.; Mao, C. D.; Whitesides, G. M. Anal. Chem. 2002, 74, 1772-1778. (22) Usui, K.; Hiratsuka, A.; Shiseki, K.; Maruo, Y.; Matsushima, T.; Takahashi, K.; Unuma, Y.; Sakairi, K.; Namatame, I.; Ogawa, Y.; Yokoyama, K. Electrophoresis 2006, 27, 3635-3642.
Figure 1. Design of the micro-DIGE protein analyzer. (A) The overall design comprises an arced, 3.75 cm long horizontal channel for firstdimension isoelectric focusing (IEF) that is punctuated with 20 6.8 cm long vertical channels through which focused proteins are separated in the second dimension by native gel electrophoresis. (B) The separation channels of the two dimensions are joined by much smaller channels forming a microfluidic interface (MFI) that fluidically decouples the contents of the two separation dimensions, seen more clearly in electron micrographs (C and D). The MFI channels are 25 µm wide, 400 µm long, and are etched 4 µm deep.
We demonstrate the utility of this system by performing differential 2DE to follow the induced expression of maltose binding protein (MBP) in an engineered strain of E. coli. MBP is a soluble, periplasmic protein common to bacteria that has proven useful as a prototype biosensor23 for both in vitro and in vivo studies. MBP is a member of a large family of bacterial periplasmic binding proteins, all of which share a similar two-domain structure that undergoes large structural rearrangements upon ligand binding. The binding site for MBP is well understood, very specific (approximately micromolar KD), and can be recombinantly tailored to target a variety of ligands. The ability to monitor its expression in cells with microfluidic DIGE is thus of both academic and practical value. METHODS Chip Design and Rationale. The design of the micro-DIGE analyzer in Figure 1 combines a single, arced channel for firstdimension IEF that is punctuated by 20 longer channels for second-dimension native gel electrophoresis. The arc of the firstdimension channel and near-convergence of the lower seconddimension channels allows detection with the Berkeley rotary confocal fluorescence scanner. The length of the IEF channel is 3.75 cm, and the second-dimension channels punctuate it at 650 µm intervals (Figure 1B). The second-dimension channels have an effective separation length of 4.4 cm and an overall length of 6.9 cm. Both the IEF and lower second-dimension separation (23) Medintz, I. L.; Deschamps, J. R. Curr. Opin. Biotechnol. 2006, 17, 17-27.
Analytical Chemistry, Vol. 79, No. 19, October 1, 2007
7361
channels are 200 µm wide after etching, while the shorter, seconddimension separation channels above the IEF channel are narrowed to 100 µm which increases their fluidic resistance and facilitates uniform gel loading. The overall length of every seconddimension channel was made uniform to equalize their fluidic and electrical resistance. Successful 2D separations require a minimum of reagent mixing between the dimensions24 because the presence of highly charged ionic species can lead to high background currents and gradient shifting during IEF.25 This attribute is accomplished in the micro-DIGE analyzer by connecting the 30 µm deep channels of the first and second dimensions with much smaller channels that serve as an MFI. Each MFI channel is 4 µm deep, 25 µm wide, and 400 µm long with a cross-sectional area 65× smaller than the separation channels (Figure 1, parts C and D). Because fluidic and electrical resistance scale, respectively, as the inverse fourth power and square of channel radius, the MFI channels act as an effective obstacle to fluid flow while allowing electrical conduction. This is crucial to both the loading of discreet media in both dimensions and to keeping the constituents of those media separate over time. Microfabrication. Device fabrication closely followed previously published procedures with the addition of a second-step etch to form the MFI channels. Borosilicate glass was chosen as a substrate instead of polymer because of its rigid nature, robust bonding performance, optical clarity, and the large body of knowledge of electrophoretic processes in glass devices. Briefly, 100 mm diameter borosilicate glass wafers (1.1 mm thick, Borofloat, Schott) were coated with 2000 Å of silicon by low-pressure chemical vapor deposition (LPCVD). The separation channels were patterned photolithographically, and the pattern was transferred to the Si by SF6 reactive ion etching, thus creating a hard mask suitable for glass etching. Separation channels were etched 30 µm deep using HF. The photoresist from the first step was then stripped (PRS-3000, JT Baker), and the wafers were dehydrated at 120 °C and hexamethyldisilizane (HMDS) vapor primed to promote adhesion of photoresist to the bare glass left after etching. Thick-film photoresist (∼10 µm, P4620, AZ Electronic Materials) was used to pattern the MFI channels, which were etched to a depth of 4 µm in 5:1 buffered oxide etch (5:1 NH3F/HF, JT Baker) at a rate of ∼2 µm/h. The narrow MFI channels sometimes did not wet in the etch bath due to topographic enhancement of the surface hydrophobicity.26 Inclusion of a commercial wetting agent (Novec 4200, 3M) eliminated the etch rate variation and channel roughness that resulted from nonwetted channels, particularly in prototype designs where channel widths were
