Abstract
Background: Developing rapid, high-throughput assays for detecting and characterizing protein–protein interactions is a great challenge in the postgenomic era. We have developed a new method that allows parallel analysis of multiple analytes in biological fluids and is suitable for biological and medical studies.
Methods: This technology for studying peptide–antibody interactions is based on polypyrrole-peptide chips and surface plasmon resonance imaging (SPRi). We generated a chip bearing a large panel of peptide probes by successive electro-directed copolymerizations of pyrrole–peptide conjugates on a gold surface.
Results: We provide evidence that (a) the signal produced by antibody binding is highly specific; (b) the detected signal specifically reflects the antibody concentration of the tested solution in a dose-dependent manner; (c) this technique is appropriate for analyzing complex media such as undiluted sera, a novelty with respect to previous techniques; and (d) correlation between classic ELISA results and the SPRi signal is good (P = 0.008). We also validated this system in a medical model by detecting anti-hepatitis C antibodies in patient-derived sera.
Conclusion: Because of its characteristics (easy preparation of the peptide chip; high-throughput, label-free, real-time detection; high specificity; and low background), this technology is suitable for screening biological samples and for large-scale studies.
The advent of high-throughput genomic technologies such as DNA/RNA chips (1)(2) prompted us to study proteins in a similar manner. Proteins constitute the essential machinery in cellular life; almost all cellular responses are triggered by protein–protein interactions, such as receptor–ligand and enzyme–substrate binding. Assays capable of specifically analyzing recognition and binding events are therefore important in efforts to understand cellular mechanisms and to design pharmacologic targets. These events cannot be studied by use of DNA/RNA chips because RNA concentrations do not always correlate with protein production and/or activity (3). Moreover, the use of systems biology as a new approach to the study of disease demands devices suitable for parallel measurements of multiple molecular interactions to integrate the whole of these data into network models (4)(5). Protein arrays have emerged recently (6) after the report of Ekins and Chu (7), but these arrays remain difficult to construct because of the complexity and heterogeneity of proteins and the necessity to preserve their conformational folding after covalent binding to a surface. Peptide chips are an attractive alternative to protein arrays because the lower complexity of the immobilized probes makes them easier to produce. Peptide chips also reduce nonspecific binding and allow the identification of peptide motifs involved in protein–protein interactions.
Microarray technology has a crucial stake in detecting molecule–molecule interactions. The use of peptide chips in conjunction with indirect detection by use of labeled probes (6)(8)(9) allows qualitative analyses of samples but fails to provide quantitative data such as kinetic and thermodynamic characteristics. Optical detection by surface plasmon resonance (SPR) 1 is an accurate one-step method for the real-time direct measurement of ligand binding without previous purification or labeling. This method has been used to analyze antibody–antigen interactions (10)(11)(12), serum antibodies against pathogens (13)(14), or specific disease markers (15)(16). However, this approach does not allow for parallel analysis on a single chip bearing an array of proteins or peptides (protein or peptide chip). To be compatible with an array format, SPR imaging (SPRi) has been developed with different biological models, including DNA and proteins (17)(18)(19)(20)(21), but these devices have never been shown to be applicable to screening of biological samples. The aim of this study was to combine the advantages of peptide chips and direct label-free detection achieved by SPRi.
Materials and Methods
materials and reagents
Prisms bearing a gold surface were prepared as described previously (17) and provided by Genoptics (Orsay, France). Activated pyrrole was obtained according to the protocol described by Livache et al. (22). One peptide derived from ovalbumin (Ova 273–288) and 11 peptides from hepatitis C virus (HCV) structural proteins (core, C 1–21, C 20–40, and C 131–150; envelope 1, E1 311–330; envelope 2, E2 528–546) and nonstructural proteins (NS2, 958–977; NS3, NS3 1248–1265; NS4, NS4/1 1688–1708; NS4/2 1921–1940; and NS5, NS5/1 2294–2313 and NS5/2 2944–2959) were synthesized with a maleimide-modified NH2 terminus by Neosystem (Strasbourg, France). Full sequences can be found on the ExPASy server (http://ca.expasy.org/) under the accession number P26664 (accessed June 2005). Peroxidase-conjugated goat anti-human IgG was purchased from Jackson ImmunoResearch Laboratories, bovine serum albumin (BSA) was from Sigma, and succinimidyl acetyl thiopropionate was from Pierce. The SPR imager was from Genoptics. Rabbit immune sera against the C 20–40 and C 131–150 peptides were prepared by Neosystem. Human sera from healthy donors (n = 5) and HCV-infected patients (n = 10) were kindly provided by the Etablissement Français du Sang (Grenoble, France) and by the Département d’Hépato-Gastroentérologie (Centre Hospitalo Universitaire, Grenoble, France), respectively. Sera were stored at −20 °C.
preparation of pyrrole–peptide conjugates
Peptides (2 mmol/L) were conjugated with SH-activated pyrrole at a 1:10 molecular ratio in phosphate-buffered saline (PBS; 137 mmol/L NaCl, 3 mmol/L KCl, 1.5 mmol/L KH2PO4, 8 mmol/L Na2HPO4, pH 7.2) for 2 h at room temperature. At the end of the process, ∼20% of the input peptides were conjugated to pyrrole, as estimated by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Because it is still difficult to separate peptide–pyrrole conjugates from the free peptide, this reaction mixture was used for electrochemical polymerization without further purification.
polymerization of pyrrole–peptide conjugates on a gold layer
Briefly, the electrochemical polymerization (electrospotting) process was carried out in sodium phosphate buffer (50 mmol/L; pH 6.8) containing 20 mmol/L free pyrrole (from a 1 mol/L solution in acetonitrile), 0.2 mmol/L pyrrolated peptide, and 100 mL/L glycerol. The chip was a prism (Genoptics) coated with a 50-nm-thick layer of gold. For electrochemical polymerization, an electrical pulse (2.4 V for 200 ms) was generated between the working electrode (gold layer) and the counter electrode (located in the tip filled with the pyrrole/pyrrole conjugate mixture) (17). The distance of the counter electrode from the gold surface was 4 mm, and the tip was rinsed with water between each electrochemical polymerization. Successive addressing of the tip filled with different peptides was carried out by an x/y/z robot from Microcontrole. After all peptide spots were synthesized, the slides were rinsed with PBS and stored at 4 °C.
antibody binding assay
All reactions were carried out in PBS at room temperature in a 10-μL Teflon cell connected to a peristaltic pump. The flow rate of solutions within the cell was 35 or 47 μL/min. The chip was saturated with PBS containing 10 g/L BSA, after which serum was injected and the sensor surface was rinsed with PBS to remove unbound molecules. The chip was then regenerated with 0.1 mol/L glycine (pH 2) for 1 min and washed with PBS before a new injection. Competition assays were carried out in 2 steps: in the first step, diluted immune serum (1:200) was injected into the Teflon cell and the chip was washed with PBS for 5 min; in the second step, relevant (competition) or irrelevant (control) free peptide (8, 15, and 24 μmol/L) diluted in PBS was injected.
SPRi interaction monitoring
The optical setup has been detailed elsewhere (17). Briefly, light was shone on the reverse side of the chip, which was constructed of a high optical index glass prism coated with thin layers of chromium and gold (5 and 45 nm, respectively). Changes in light reflectivity attributable to antibodies interacting with the immobilized peptides were recorded by a 12-bit charge-coupled device (CCD) camera. Sequential images were recorded at 1.5-s intervals, and binding kinetics were monitored by dedicated software (Genoptics).
preparation of bsa–peptide conjugates for elisa control
BSA was activated by use of succinimidyl acetyl thiopropionate according to the manufacturer’s recommendations and was further purified by gel filtration through Sephadex G50-fine columns (Amersham Biosciences). Maleimide-modified peptides were mixed with SH-activated BSA (10:1 molar ratio) in PBS for 2 h at room temperature and stored at −20 °C.
elisa
Plates were coated with BSA–peptide conjugates (equivalent to 0.5 mmol/L BSA). The ELISA was performed as described by Villiers et al. (23), and bound antibodies were revealed with peroxidase-conjugated goat anti-human IgG. Results represent the absorbance measured for a 1:1000 serum dilution.
Results
peptide chip preparation and SPRi analysis
To prepare the peptide chips, we conjugated peptides N-terminally modified with maleimide to thiol-activated pyrrole and immobilized the conjugate on a gold surface, as described in the Materials and Methods. A spacer was inserted between the pyrrole moiety and the peptide to favor epitope accessibility to the antibody (Fig. 1A⇓ ). The electrochemically directed polymerization of pyrrole and peptide–pyrrole conjugates, initially developed for DNA arrays (17), scatters a polypyrrole film on the surface of the working electrode. Different peptides were immobilized on spatially defined areas of the gold surface of the chip through successive polymerizations with different peptides (Fig. 1A⇓ ). The resulting chip hosted 40 different spots on a 0.5-cm2 area. The typical diameter of a spot ranged between 300 and 500 μm (internal diameter of the tip used for the electrochemical polymerization). The electrochemical charge needed for electrocopolymerization for all spots was recorded to ensure the reproducibility of the process. For SPRi measurements, the functionalized gold layer was deposited on a prism; a polarized light illuminated the whole surface, and the beam reflected by each point was analyzed individually, allowing simultaneous monitoring of each peptide spot (Fig. 1B⇓ ). A CCD camera converts the changes in reflectivity into gray-level variations that are directly related to the mass(es) of the compound(s) bound to the surface (see the movie in the Data Supplement that accompanies this article at http://www.clinchem.org/content/vol52/issue2/ ).
General scheme of the experimental setup.
Peptide chip preparation and SPRi device. (A), electrocopolymerization of pyrrole and pyrrole–peptide conjugates is carried out on a prism surface coated with a thin gold layer. M, maleimide. (B), the sensor chip is illuminated by a p-polarized LED beam, and the reflected light is recorded by a CCD camera. As shown in the sensorgrams (C), the CCD camera monitors the whole sensor chip surface, allowing simultaneous analysis of binding events on all spots (26 in this case). All curves show a typical profile with an association phase during sample injection, followed by a dissociation phase on washing. Regeneration of the chip allows for further sample measurements.
The example presented in Fig. 1C⇑ shows typical curves (sensorgrams) obtained when the reflectivity of each of the 26 spots analyzed simultaneously on this chip is plotted against time: injection of the ligand induces an initial increase in reflectivity; nonspecific interactions are then removed by washing, and the chip surface is regenerated with a glycine-based solution. Thus, the initial signal is recovered.
specificity of antibody binding to peptide chips and SPRi detection
Three peptides (C 20–40, C 131–150, and Ova 273–288) were spotted on the surface of a chip, and different rabbit immune sera were successively injected. The chip surface was regenerated after each assay. As shown in Fig. 2A⇓ , injection of an anti-C 20–40 immune serum led to an initial increase (association) and then to the stabilization (equilibrium) of the reflectivity signal corresponding to the spot bearing the C 20–40 peptide, whereas low signals were observed with the irrelevant C 131–150 and Ova 273–288 peptides as well as with the negative control (polypyrrole alone). The difference at the plateau (ΔR) between the sensorgrams generated by the positive assay and the control (irrelevant peptide) corresponds to the specific signal attributable to antibody capture. In contrast, injection of nonimmune serum under the same conditions led to a low and similar reflectivity from all of the spots, which corresponded to nonspecific binding (background value). Subsequent injection of the anti-C 131–150 immune serum confirmed the specificity of the method; only the C 131–150 spot emitted a strong reflectivity signal. Moreover, this third injection demonstrated that the peptide chip was still efficient after regeneration. Thus, consecutive injections (n = 4) of the same serum led to a reproducible signal (within 0.03%).
Specificity of antibody binding on peptide chips.
(A), direct assays. Rabbit anti-C 20–40 and anti-C 131–150 immune sera were successively injected on a peptide chip bearing peptides C 20–40 (green line), C 131–150 (red line), Ova 273–288 (gray line), or polypyrrole alone (blue line). Rabbit nonimmune serum was injected as control. All sera were used at the same dilution (1:100), and a regeneration step was performed between 2 successive injections. (B and C), competition assays. Rabbit anti-C 131–150 immune serum was injected first (1); after a 5-min wash, the C 20–40 peptide (B) or the C 131–150 peptide (C) was injected (2); the chip was then regenerated (3).
The specificity of the signal observed by SPRi was further assessed by competition assays: anti-C 131–150 antiserum was injected first and produced an increased reflectivity signal corresponding to the C 131–150 spot. Free C 20–40 (Fig. 2B⇑ ) or C 131–150 (Fig. 2C⇑ ) peptides, used as competitors, were subsequently injected. In the former case, the signal from all of the spots was slightly enhanced, likely because of an optical index shift, and then stabilized with time (Fig. 2B⇑ ). By contrast, injection of free C 131–150 peptide persistently and specifically decreased the reflectivity signal from the corresponding peptide spot (Fig. 2C⇑ ). This latter result indicates that the bound antibodies are displaced from the peptide chip by the relevant free peptide through selective competition and in a dose-dependent manner (see Fig. 1 in the online Data Supplement).
SPRi quantification of antibody binding on peptide chips
To evaluate the usefulness of this method for analyzing peptide–antibody interactions, we assessed whether the SPRi signal was indeed related to the amount of antibodies present in the input solution. Serial dilutions of anti-C 131–150 antiserum were successively injected into the Teflon cell, and the chip surface was regenerated after each assay. As shown in Fig. 3A⇓ , only background reflectivity was detected from the C 20–40 spot, whereas a dose-dependent signal was emitted from the relevant C 131–150 spot. More specifically, the measured ΔR correlated linearly with the dilution factor (Fig. 3B⇓ ), until a plateau was reached on saturation of the system. The value obtained at the plateau (8% of reflectivity) corresponded to a monolayer of antibody (24) equivalent to 1 pmol/mm2.
SPRi analysis and quantification of antibodies bound to peptide chips.
(A), sensorgrams obtained when rabbit anti-C 131–150 immune sera serially diluted in PBS were successively injected on a peptide chip on which C 20–40 (gray line) or C 131–150 (black line) was immobilized, with a regeneration step between 2 successive injections. (B), ΔR plotted against the dilution factor of the immune serum in buffer ( ) and in nonimmune serum (▪).
One important application of this method is the detection of antibodies (or other molecules) present, even at low concentrations, in a complex medium such as serum. Because this technique was optimized by use of sera diluted in an appropriate buffer, we tested its efficacy in more stringent conditions. To that end, we diluted the anti-C 131–150 immune serum in whole nonimmune serum, mimicking a weak antibody concentration in a complex medium. Sera were successively injected on the chip under the same conditions used for the previous series of experiments (see Fig. 2 in the online Data Supplement). As shown in Fig. 3B⇑ , the curves featured similar profiles regardless of whether the screened antibodies were diluted in nonimmune serum or in buffer, confirming that the method could indeed be used for the analysis of a complex medium. The shift between the 2 curves was likely attributable to the blocking effect of the serum used to dilute the samples.
analysis of patient-derived sera by use of peptide chips and SPRi detection
Patients infected with HCV raise a broad antibody response that persists long after the disease has resolved. We used peptide chips and SPRi detection to screen sera from 10 patients who had recovered from HCV infection, as assessed by undetectable viral RNA, for anti-C 20–40 and anti-C 131–150 antibodies. As shown in Table 1⇓ , 3 sera induced a significant enhancement in reflectivity (ΔR >0.3) from the C 131–150 spot and 8 sera induced an increased reflectivity signal from the C 20–40 spot, revealing the presence of specific antibodies against these 2 peptides in patient-derived sera. On the other hand, the 5 sera from healthy donors did not produce a significant change in reflectivity.
SPRi analysis of human sera by use of a chip bearing immobilized peptide C 20–40 or C 131–150.
To further demonstrate the relevance of our technique, we compared these results with those obtained by an ELISA. Correlation between the 2 methods was good (Fig. 4⇓ ): sera giving negative or weakly positive ELISA results also generated a weak SPR signal and sera generating high ELISA values also gave high SPRi signals. We confirmed this correlation by use of the two-tailed Spearman test (P = 0.008).
Correlation between graphs plotting anti-C 20–40 antibodies in sera from patients quantified by SPRi (ΔR, difference in reflectivity between positive and negative peptide spots) and by ELISA (A, absorbance).
The statistical analysis was performed with the two-tailed Spearman test.
Because microarrays are designed for assaying multiple analytes, we broadened our technique to a wider panel of epitopes. We implanted 11 HCV peptides on a chip. These epitopes were derived from structural (C 1–21, C 20–40, C 131–150, E1 311–330, and E2 528–546) and nonstructural (NS2 958–977, NS3 1248–1265, NS4/1 1688–1708, NS4/2 1921–1940, NS5/1 2294–2313, and NS5/2 2944–2959) proteins. Ova 273–288 was used as negative control. Eleven sera from HCV-positive patients and from healthy donors were successively injected on this chip, with a regeneration step between each new injection. The reflectivity changes from each spot (n = 12) were recorded simultaneously (Fig. 5A⇓ ), with each curve accounting for the reactivity of each serum against each peptide. As shown in Fig. 5B⇓ , the specific SPRi signal (ΔR) varied according to the patient and the peptide, revealing donor-related differences in the anti-HCV antibody responses (specificity, amount, and affinity). It is noteworthy that antibodies against some peptides were never (NS5/1) or rarely (E2) detected in the sera, suggesting that these epitopes induce only a weak humoral response.
SPRi analysis of sera from 10 HCV patients (P1–P10) and from 1 healthy donor (H1).
The sera were injected on a chip bearing 11 HCV peptides and Ova 273–288 as a control, with a regeneration step between 2 successive injections. (A), sensorgrams obtained after injection of the 11 sera (diluted 1:4 in PBS). (B), ΔR for each serum.
Discussion
Serum samples are very difficult to screen by classic SPR detection (10)(25) unless they are highly diluted. We report here a novel technology that enables SPRi-based analysis of antibody–peptide interactions in a microarray format featuring very low background noise. Polymerization of pyrrole and pyrrole–peptide conjugates makes possible electro-directed immobilization of a panel of different peptides in spatially indexed microspots on a gold-coated prism. The resulting peptide chip can be used for direct and real-time monitoring of ligand–peptide interactions through SPRi. This is a novelty with respect to previously described peptide microarrays because they usually involved an indirect visualization step based on fluorescence (26)(27)(28) or radioactivity (6)(29).
The peptide chips we describe here are easy to generate, are probably suitable for most peptides, and can be regenerated several times (for at least 30 successive assays) without loss of efficiency. We provide evidence that our method enables detection of highly specific antibody–peptide interactions and that the output SPRi signal is directly correlated to antibody concentrations in the tested solutions. Moreover, because of its weak background noise, this technique is suitable for analyzing undiluted sera, a characteristic of great relevance for future clinical applications. To that aim we compared our method with a classic ELISA. Using both methods, we analyzed the presence of antibodies against 2 HCV core protein–derived peptides (C 20–40 and C 131–150) in sera from 10 HCV-seropositive patients. Specific anti-C 131–150 antibodies were detected in only 3 of 10 patients, which is not surprising because this peptide is a T-cell rather than a B-cell epitope (30)(31). By contrast, the majority of the patients had antibodies against C 20–40, which is known to be an immunodominant B epitope responsible for eliciting a strong humoral response (32). Altogether, the results obtained with our peptide chip faithfully reflected those of the ELISA. Because ELISA is a multistep method leading to amplification of the signal, it allows the use of more highly diluted sera than does SPRi (1:1000 vs 1:4). However, additional injection of a secondary label-free antibody could improve the sensitivity of SPRi method. Indeed, SPR signal is related to the mass of bound molecules on the surface, which increases on addition of a second ligand (33). Finally, relevant information (e.g., avidity and apparent Koff) may be extracted from the raw data of kinetic curves; we are currently studying this use. Moreover, there are no data available from previously published results on the kinetics of the antibody–peptide interactions because we are the first group, to our knowledge, that has analyzed the binding of anti-C 20–40 and anti-C 131–150 immunoglobulins present in sera to the corresponding peptides.
SPRi measurements provide simultaneous analysis of multiple spots, each bearing different peptide probes. In this study we used a chip carrying 12 different peptides to analyze the antibody response of HCV-seropositive patients. This method enabled us to sketch an antibody profile for each of the patients. We highlighted differences in both the specificity and the strength of the antibody response elicited on HCV infection, which is a step forward in studying the variability of the anti-HCV humoral response. Moreover, antibody specificities are thought to be a signature of the outcome of infection, i.e., chronicity, cirrhosis, hepatocarcinoma, or recovery. Early identification of the specific antibodies present in sera from patients would therefore facilitate early prognosis and optimization of the therapeutic strategy.
In conclusion, we have developed a method combining peptide chips and SPRi detection for high-throughput, and multianalyte screening of peptide–ligand interactions in label-free, real-time conditions. Moreover, the high specificity of this technique makes it applicable to the study and screening of complex biological samples such as undiluted serum, for which currently used methods are not suitable. These characteristics make this technology a powerful tool for clinical applications such as diagnosis, for pharmacologic studies, and for basic research such as enzyme–substrate, antigen–antibody, or protein–cofactor interaction analyses.
Acknowledgments
We are extremely grateful to Bernard Dublet from the Institut de Biologie Structurale (Grenoble, France) for mass spectrometry analyses. This work was supported by the Agence Nationale de Recherches sur le Sida (ANRS). Boutheina Cherif was supported by the “Ministère de l’Enseignement Supérieur de Tunisie” and the “CEA, Direction des Relations Internationales”.
Footnotes
1 P, patient; H, healthy donor.
↵1 Nonstandard abbreviations: SPRi, surface plasmon resonance imaging; HCV, hepatitis C virus; BSA, bovine serum albumin; PBS, phosphate-buffered saline; and CCD, charge-coupled device.
- © 2006 The American Association for Clinical Chemistry