Protein Post-Translational Modification Analyses Using On-Chip

Jan 30, 2013 - Department of Bioengineering, University of California, 342 Stanley ...... (12) Colton, I. J.; Anderson, J. R.; Gao, J. M.; Chapman, R...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

Protein Post-Translational Modification Analyses Using On-Chip Immunoprobed Isoelectric Focusing Samuel Q. Tia,† Katharine Brown,‡ Danica Chen,‡ and Amy E. Herr*,†,§ †

The UC Berkeley−UCSF Graduate Program in Bioengineering, University of California, Berkeley, California 94720, United States Department of Nutritional Science and Toxicology, University of California, Berkeley, California 94720, United States § Department of Bioengineering, University of California, 342 Stanley Hall, Berkeley, California 94720, United States ‡

S Supporting Information *

ABSTRACT: Post-translational modifications play a critical role in regulating protein function. Increasingly, determination of protein identity, estimation of abundance, and characterization of posttranslational modifications are required for analysis of protein-mediated cell signaling networks. As such, we report an integrated and rapid multispectral immunoprobed isoelectric focusing technique for identifying specific proteins bearing post-translational modifications. Immunoprobed isoelectric focusing is composed of isoelectric focusing in a large pore-size polyacrylamide gel to determine protein pI followed by immobilization of pI-resolved proteins. Proteins are immobilized via covalent attachment to a channel-filling benzophenonefunctionalized polyacrylamide gel via brief UV exposure (photoblot), followed by multispectral antibody-based detection. The assay correlates observed post-translational modifications to pI shifts relative to the unmodified protein of interest. During the electrokinetically driven antibody probing stage, we observed nonuniform electrophoretic probe mobility along the channel axis. The spatially varying mobility is attributed to nonuniform charge arising from covalent attachment of ampholytes to the benzophenone-functionalized gel matrix during the photoblotting step. Using the multistep microfluidic assay, phosphorylated and acetylated forms of heat shock protein 27 and superoxide dismutase 2 were detected, respectively. The assay reported protein isoforms in immune-purified sample and raw cell lysate in 2 hours with sample volume requirements of 2 μL. This new assay is well-matched to systems biology frameworks for study of protein post-translational modifications.

W

tions.10,11 For example, the substitution of lysine’s primary amine with an acetyl group (acetylation) results in a more negative charge conferred upon the entire molecule.12 Thus, an acetylation event may be identified if the pI of the target protein is known and if a corresponding charge shift can be detected. That said, a charge shift alone may not be enough to conclusively determine the presence of a PTM. For example, in pan-specific protein staining of two-dimensional polyacrylamide gel electrophoresis for a complex sample, the abundance of background proteins may obscure the observation of a charge shift for a single protein spot.13 The immunoprobed IEF assay presented here addresses challenges to current PTM analysis by combining the separative power of IEF with the specificity imparted by antibodymediated target identification. Antibodies specific for PTM residues have been validated and are commercially available.14,15 Immunoprobed IEF takes advantage of a unique microscale format to enable charge-based protein separations within 20 min followed by rapid and efficient antibody probing steps integrated in a chip-based assay, as recently introduced by our group.16 In contrast to capillary-based immunoprobed IEF

ithin recent decades, post-translational diversity has emerged as a field of critical importance toward the understanding of proteomic regulatory and signaling mechanisms.1,2 The diversity of the human genome can only account for a fraction of total proteomic complexity, and so the role of post-translational modifications (PTMs) must be considered to elucidate the full range of protein function and relationships.3 PTMs are diverse, with a recent survey indicating over 80 000 PTMs characterized experimentally.4 However, continued understanding of PTMs will be dependent upon the availability of reliable and quantitative tools for measurement. Standard techniques for PTM analysis can be resource intensive and may have difficulty accommodating complex or volume-limited samples. Results from mass spectrometry (MS) require sophisticated computer algorithms for interpretation,5 and signal-to-noise limitations of the detection mechanism can pose challenges to analysis of low-abundance proteins.6,7 MS is often used in conjunction with antibody-based techniques which provide an enrichment step and further confidence in identifying a particular target prior to MS analysis.8 However, antibody-based techniques can require large amounts of starting material and sample prefractionation steps are often necessary.9 Isoelectric focusing (IEF) is often used to separate and identify target proteins based upon their characteristic isoelectric point (pI) while enabling characterization of subsequent modifica© 2013 American Chemical Society

Received: December 3, 2012 Accepted: January 30, 2013 Published: January 30, 2013 2882

dx.doi.org/10.1021/ac3035053 | Anal. Chem. 2013, 85, 2882−2890

Analytical Chemistry

Article

systems,11 the planar microfluidic PTM assay detailed here relies on only electrophoretic transport, eliminating the need for bulk fluid control hardware (i.e., pumps, valves). As compared to surface-functionalized capillary strategies, the integration of photoactive polyacrylamide gels filling the microchannels allows for two major performance advances.16 First, the three-dimensional (3D) reactive gels offer 102−103 more benzophenone capture sites per unit volume than photoactive surface capture strategies. Second, distribution of photoactive sites throughout the channel cross section leads to favorable pseudohomogeneous reaction kinetics during blotting and probing steps. Multispectral fluorescence detectionin some cases using a standard microarray scannerallows simultaneous detection of colocalized antibody probes. We demonstrate immunoprobed IEF for analysis of phosphorylated and acetylated proteins. Immunoprobed IEF enables direct assay of complex samples coupled with an intuitive visualization and analysis of results, leading to quantitative and specific measurements performed within a single workflow.

BPMAC) is incorporated to the backbone of the polyacrylamide gel.16 This monomer allows for covalent immobilization of protein to the benzophenone polyacrylamide gel (BP−PA) when UV light excites carbonyl groups of the BPMAC into an electrophilic triplet state.11 Three parallel channels connecting two fluid reservoirs yield triplicate measurements of each sample. Details of the assay have been previously reported;16 briefly, the sample is diluted between 4× and 20× in a loading buffer spiked with a cocktail of UV-fluorescent pI ladder markers. Target analytes are unlabeled, so a low concentration (10−60 nM) of green fluorescent protein (GFP) is included as a marker to ensure proper sample loading, subsequent IEF, and efficient protein immobilization. The sample mixture is titrated to pH 9.9 using sodium hydroxide, and approximately 2 μL is applied at the sample reservoir. Diluted sample is electrophoretically introduced to the device for 5 min under an applied electric field of E = 200 V/cm. The electric field is controlled using a custom programmable high-voltage power supply with platinum electrode tips placed in liquid filled reservoirs. After sample loading, channel wells are washed via gentle aspiration and fluid exchange, and then loading buffers are replaced with catholyte and anolyte solutions at each terminus reservoir, respectively. Proteins are separated through IEF, with programmed stepwise increases in electric field strength over the course of 17 min. The electric field is initiated at 50 V/cm for 4 min, followed by applications of 100 and 200 V/cm for 5 min each, eventually set to a maximum value of 300 V/cm for 3 min. When all proteins have reached a focused equilibrium position, the linearly distributed GFP and pI markers are imaged to calibrate pI locations along the separation axis. After IEF completion, the separation is exposed to an 11 s high-intensity UV pulse to initiate protein immobilization to the BP−PA gel. A Hamamatsu LC5 light source is directed through a Lumatec 380 liquid light guide, generating a UV power of 160 mW/cm2 at 70% intensity (UV513AB meter; General Tools). Following UV exposure and protein immobilization, electrophoretic washout of unbound proteins is necessary before antibody probing (washout: 20 min at 150 V/cm followed by 15 min at 200 V/cm). Estimates of the total protein capture efficiency are made by comparing fluorescence intensity of focused bands after IEF to fluorescence intensity obtained after immobilization and washout. To probe the immobilized pIordered bands, a cocktail of multispectral probes against the specified antigen and the PTM of interest are introduced (5 min at 200 V/cm followed by 20 min at 100 V/cm). Following removal of unbound probes (20 min at 150 V/cm followed by 15 min at 200 V/cm), subsequent imaging of fluorescently probed protein bands provides a quantitative image allowing determination of pI values where spectral signals are colocalized, thus indicating the presence of a PTM target. Quantitation of protein bands is performed by examining the peak area under the curve (AUC) when fluorescence intensity is plotted as a function of axial position (see Supporting Information Figure S-5). For PTM proteins, multiple fluorescence spectra can be analyzed by comparing the ratio of AUC for colocalized bands (normalized to the reference intensity for each antibody probe). The immunoprobed IEF assay is complete within 120 min. The initial concentration of each protein is correlated to its resultant probe signal using a dose−response calibration. The binding stoichiometry of a fluorescent antibody probe to its



MATERIALS AND METHODS Assay Operation. The purely electrophoretic, multistage immunoprobed IEF assay is composed of (Figure 1A) sample

Figure 1. Microfluidic immunoprobed IEF for PTM analyses. (A) An aliquot of 2−3 μL of sample is introduced on-chip. Protein bands are focused, immobilized, and probed with a multiantibody “cocktail” for specific protein modifications. Targeted PTMs are detected via multispectral fluorescence imaging. (B) Single glass device housing three parallel microfluidic channels (dye-filled for illustration).

loading, sample focusing, immobilization, washout, and fluorescent antibody probing/readout. A core enabling component is the photoactive sieving matrix, a multifunctional polyacrylamide gel housed in a straight glass microchannel (70 μm width × 20 μm depth × 10.5 mm length, Figure 1B). To enable integration of IEF with antibody probing in a single microchannel, a benzophenone methacrylamide monomer (N[3-[4-benzoylphenyl]formamido]propyl] methacrylamide or 2883

dx.doi.org/10.1021/ac3035053 | Anal. Chem. 2013, 85, 2882−2890

Analytical Chemistry

Article

monoclonal anti-cytochrome c (α-CytC, ab13575), goat polyclonal anti-GFP (α-GFP, ab6660), and rabbit polyclonal anti-phosphoserine (α-PhoS, ab9332) were purchased from AbCam. Antibody cross-reactivity was investigated using combinatorial “dot blot” screening of pairs of antibody probes and potential targets. Dot blot results indicated no detectable signal when α-AceK was used to probe cytochrome c, nor when α-CytC was used to probe acetylated lysine groups (Supporting Information Figure S-1). No fluorescence emission crosstalk was observed for the two fluorescent labels used for multispectral imaging (Supporting Information Figure S-2). Immobilized pH condition gels were produced by titrating Immobiline species (GE Healthcare) to the appropriate pH, then adding the Immobilines to the polyacrylamide gel precursor solution (20 mM Immobiline, 4%T gel). All polyacrylamide gels were made through dilutions starting from a 30%T, 2.6%C stock solution (37.5:1 acrylamide/ bisacrylamide ratio, Sigma-Aldrich). Novex pH 3−10 slab gel IEF supplies (Life Technologies) were run according to manufacturer’s instructions. Colored IEF ladder markers were obtained from SERVA Electrophoresis (Heidelberg, Germany). AlexaFluor 488 and 568 antibody labeling kits and SNARF dyes were purchased from Life Technologies. All PTM specific antibodies were labeled with AlexaFluor 568 according to manufacturer’s instructions (α-PhoS*, α-AceK*), while all other antibodies (α-Target*) were labeled with AlexaFluor 488 unless stated otherwise. Device Fabrication. Photomasks to generate the glass channel network were designed in-house, and devices were fabricated via standard photolithography, wet-etch, and glass bonding techniques at a microfluidic chip foundry (Caliper Life Sciences, Hopkinton, MA). Prior to use, glass channels were flushed with sodium hydroxide and treated with silane to facilitate adhesion of polyacrylamide to the channel surface. Once the gel precursor solution was prepared, ammonium persulfate (APS) and tetramethylethylenediamine (TEMED) chemical initiators were mixed into the precursor at 0.08% w/v and 0.08% v/v, respectively. The precursor was then degassed for 60 s before being wicked through the microchannels. Catalytic activity of the APS and TEMED resulted in complete polymerization of the gel matrix within 10 min. Data Collection and Analysis. Fluorescence imaging was conducted on an Olympus IX71 inverted microscope equipped with an X-cite eXacte fluorescence light source (Lumen Dynamics, Ontario, Canada). Images were acquired through a 10× objective (Olympus UPlanFL, N.A. 0.3) and iXon3 885 electron multiplying CCD camera (Andor Technology, Belfast, U.K.). Fluorescence band-pass filters employed in this work include the XF02-2 (Omega Optical, Brattleboro, VT) as well as 49011 and 49008 filter sets (Chroma, Bellows Falls, VT) for UV, green, and red wavelength imaging, respectively.

target antigen can be calculated if the initial concentration of the antigen is known. In this case, we assume that all proteins exhibit comparable binding to the BP−PA matrix as measured by the GFP surrogate. Fluorescence signal from the antibody probe is calibrated to a known local probe concentration, and this concentration is normalized to the estimated number of antigenic sites to yield binding stoichiometry. The ratio of red (PTM) to green (target) fluorescence can be used to measure the relative degree of modification for a particular target of interest and provides an additional piece of information to supplement the pI data.17 The ratio between the two probes may be mapped along the length of the IEF channel to correlate spectral signals with modified target protein. Direct red to green intensity comparison assumes that both probes exhibit comparable fluorescence characteristics, similar binding affinities, and no competition for the same antigenic binding sites. Thus, probe calibration and cross-reactivity experiments should be performed for each target prior to multiprobe comparison and analysis. Reagents and Samples. The BP−PA gel consists of BPMAC (synthesized in-house) at 5 mM within a 4%T polyacrylamide gel matrix. Other components of the gel buffer include CHAPS at 3% w/v, nondetergent sulfobetaine (NDSB) 256 at 200 mM, and 10% sorbitol. Using a pH 3−10 Pharmalyte ampholyte mixture (GE Healthcare, Uppsala, Sweden) within the BP−PA and loading buffer system resulted in a linear gradient ranging from pH 3.5 to 9. Other ampholyte mixtures have also been demonstrated with the BP−PA system to specify gradients with different boundary conditions and varying pI resolution. The antibody probe and washout buffer consists of 3% CHAPS, 15 mM glycine, 200 mM NDSB 256, and 10% sorbitol at pH 9.9. Catholyte buffer consists of 20 mM arginine and 20 mM lysine at pH 10.1, while the anolyte consists of 70 mM phosphoric acid at pH 1.9 (Bio-Rad, Hercules, CA). Fluorescent pI markers ranging from pH 4.0 to 9.0 (89827, 90699, and similar) were purchased from Sigma-Aldrich (St. Louis, MO). Recombinant GFP (632373) was obtained from Clontech (Mountain View, CA). Cytochrome c peptide (ab38265), heat shock protein 27 (ab48740), phosphorylated heat shock protein 27 (ab113185), and human recombinant superoxide dismutase 2 (SOD2) (ab82656) were obtained from Abcam (Cambridge, MA). Acetylated bovine serum albumin (BSA) was purchased from Life Technologies (Grand Island, NY), and an acetylated cytochrome c ladder was obtained from Affymetrix (Cleveland, OH). FLAG-tagged SOD2 was expressed in a HEK293T cell line. Cultured cells were counted with a hemocytometer and then lysed in a cell lysis buffer containing protease inhibitors. Immunopurification was achieved by incubating lysate with FLAG antibodies for 2 h followed by extensive washing and free FLAG peptide elution. Another portion of the transfected 293T cells remained in raw lysate form. A cell lysate (sc-24816) enriched for phosphorylation through immobilized metal ion affinity chromatography from an NIH/3T3 cell line was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal anti-SOD2 (α-SOD2, sc-133254), goat polyclonal anti-heat shock protein 27 (α-HSP27, sc-1049), and rabbit polyclonal anti-phospho heat shock protein 27 (αpHSP27, sc-101700) were also purchased from Santa Cruz Biotechnology. Mouse monoclonal anti-acetylated lysine (αAceK, 623402) was obtained from Biolegend (San Diego, CA). Mouse monoclonal anti-BSA (α-BSA, ab3781), mouse



RESULTS AND DISCUSSION Multiplexed Immunoprobing To Assess Protein Identity and PTMs. Probing Photopatterned Proteins with a Cocktail of Antibodies. To assess the performance of antibody probing, we initially omitted the IEF stage and directly photopatterned bands of target protein on the BP−PA gel (Figure 2, parts A and B). We employed a mask-based UV exposure after electrophoretically introducing target proteins to the BP−PA gel filled channel. Channel sections exposed to UV light see covalent attachment of protein to gel, while masked regions lack protein immobilization. Two protein samples were 2884

dx.doi.org/10.1021/ac3035053 | Anal. Chem. 2013, 85, 2882−2890

Analytical Chemistry

Article

A similar analysis was conducted on the positive control (Figure 2B). Acetylated BSA was photopatterned in the microchannel with a capture efficiency of 1.2% ± 0.1 (n = 3, stock solution at 5.6 μM). The immobilized BSA was simultaneously probed with α-BSA* and α-AceK*. After electrophoretic antibody washout, fluorescence signal from both probes was detected and colocalized at the site of the immobilized BSA. α-BSA* signal was detected at S/N = 130 ± 22.9 (n = 3), while fluorescence signal from α-AceK* was measured with a S/N of 20.5 ± 4.1. In comparison, analysis of nonacetylated BSA reported no signal (S/N = 2.1 ± 0.3) when this BSA was probed with α-AceK*, suggesting negligible nonspecific binding by α-AceK* (Supporting Information Figure S-3). The positive control assay result suggests that antibody probing through a cocktail of characterized antibodies is an efficient and effective approach to correlating protein identity with pI location and PTMs. Integrating Multiplexed Immunoprobing with IEF. Having demonstrated the antibody cocktail probing strategy for concurrent probing of protein identity and PTMs, we next explored the specificity of multiplexed probing after sample analysis by IEF. We measured the stoichiometry of α-BSA* binding to focused BSA in the presence and absence of αAceK* (Supporting Information Figure S-3). We first focused the positive control nonacetylated BSA using IEF, and bands were photoblotted onto the BP−PA gel. After electrophoretic “wash out” of the IEF pH gradient, immobilized BSA protein zones were probed with α-BSA*. Imaging of the α-BSA* probe reported a single protein band centered at pI 4.7 ± 0.1, a value that agrees with previous reports of BSA.19 For α-BSA* probing of BSA with no α-AceK*, we observed a 0.71:1.0 stoichiometry (antibody/antigen concentration, Supporting Information Figure S-3). When probing photopatterned BSA with the two-component antibody cocktail (α-BSA*, α-AceK*), we observed a 0.68:1.0 stoichiometry for α-BSA* (Figure 2B). The agreement between the estimated binding stoichiometries with and without the second antibody probe suggests that α-AceK* and α-BSA* do not compete appreciably for the same BSA epitopes under the conditions studied here. The fact that the binding ratio does not change notably for a focused band versus a selectively photopatterned protein zone demonstrates that the antibody’s binding affinity for the target is not diminished by IEF. The analysis suggests that a multispectral probing strategy allows simultaneous detection of a protein and its modified forms after IEF. Photoblotting of pH Gradient on BP−PA Gel Impacts Subsequent Immunoprobe Electromigration. Photoblotting to the BP−PA gel will immobilize both focused protein zones and background ampholytes.16 Time course imaging of immunoprobe loading into the separation channelafter IEF and electrophoretic pH gradient “wash out”yielded a location-dependent electrophoretic mobility for the antibody probe (Figure 3A). To illustrate the impact of probe introduction polarity (direction) (relative to pH gradient) on the final probe concentration along the IEF separation axis, we loaded fluorescently labeled antibodies into photoblotted BP− PA gels (Figure 3A, 25 min loading at 200 V/cm). A higher and more uniform probe concentration was achieved throughout the channel by loading probe from the basic end of the IEF pH gradient (catholyte pH = 10.1, anolyte pH = 1.9). High probe concentrations are favorable for sensitive detection of immobilized protein target. When saturation of target by

Figure 2. Dual-antibody probing specifically identifies modified residues on target proteins. (A) Cytochrome c demonstrates binding specificity for anti-cyt c, but not against antibodies to AceK residues. (B) Acetylated BSA exhibits binding specificity for both anti-BSA and anti-AceK.

studied: a protein sample with no expected acetylation (cytochrome c as a negative control, Figure 2A) and a protein sample with expected lysine acetylation (acetylated BSA as a positive control, Figure 2B). We probed the fluorescently labeled negative control sample for both identity and acetylated lysine (Figure 2A). Comparison of fluorescence before UV exposure versus fluorescence after UV exposure and electrophoretic washout yielded an immobilization efficiency of 1.5% ± 0.2 (n = 3, with a local concentration of 110 nM). We attribute the observed capture efficiency after IEF to interference between buffer constituents (surfactants, ampholytes) and the BP−PA.16 We have previously observed higher capture efficiencies (>85%) when using the same gel formulation for photoimmobilization after sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), suggesting that photoblotting of denatured proteins enhances capture efficiency.18 As with all heterogeneous immunoassays, optimal probing performance is achieved when the local probe concentration exceeds the concentration of immobilized antigen and the disassociation constant (Kd) of the antigen−antibody probe pair. Given a typical antigen−antibody Kd of 1 nM (kon = 1 × 106 s−1 ≫ koff = 1 × 10−3 M−1 s−1), we introduce all antibody probes at a concentration of 500 nM. Thus, we expect the binding reaction kon

[antigen][probe] → [antigen−probe]

to reach a saturated equilibrium [antigen−probe] ≫ [antigen] before the end of the probe loading process (t < 100 s). A brief incubation period (