Molecularly Engineered Charge-Conversion of Proteins for Sensitive

Oct 12, 2010 - Richard , J. P. Biochemistry 1991, 30, 4581– 4585. [ACS Full Text ACS Full Text ], [CAS]. 28. Kinetic parameters for the elimination ...
0 downloads 0 Views 490KB Size
Anal. Chem. 2010, 82, 8946–8953

Molecularly Engineered Charge-Conversion of Proteins for Sensitive Biosensing Tatsuro Goda†,§ and Yuji Miyahara*,†,‡,§ Biomaterials Center and International Center for Material and Nanoarchitectonics (MANA), National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan To gain better signals in potentiometric biosensing of protein, site-selective chemical modification of amino acid residues was employed by exogenous acylation and glycation reactions of primary amines and the guanidinium group, converting them from cationic into anionic or neutral. The mass shift corresponding to the lysine and arginine adducts was confirmed only at the surfaceexposed solvent-accessible residues. The site-selectivity of the charge-conversions resulted in maintained structural integrity and bioactivity of the proteins. The estimated negative charge density of bovine serum albumin (BSA) under physiological pH increased by 5-fold as a result of the formation of stable succinic lysine. Real-time measurement of protein adsorption onto the 1-undecanethiol self-assembled monolayer (SAM) on gold was detected using an extended gate-field effect transistor (FET). The potential shifts by the adsorption was 3-fold higher in succinylated BSA than origianl BSA, whereas more significant amplification of the signal (11-fold) was observed by the modifications of lysozyme. Furthermore, in situ modification of amino acids during the potentiometry was achieved for the tightly adsorbed lysozyme onto the SAM. In summary, site-selective charge-conversion provides a new type for the molecular “label” for biosensing with preserved conformational integrity of the protein. Semiconductor-based potentiometric biosensors using field effect transistors (FETs) have been attracting broad interest as a fast, real-time, label-free, highly sensitive, simple, and inexpensive method for analysis of DNA recognition events including hybridization, extension, ligation, and single-nucleotide polymorphism (SNP) genotyping on a solid surface.1-8 FET can directly transform variations in the gate-electrolyte surface potential, which * To whom correspondence should be addressed. E-mail: miyahara.bsr@ tmd.ac.jp. † Biomaterials Center. ‡ International Center for Material and Nanoarchitectonics (MANA). § Current address: Institute of Biomaterials and Bioengineering, Tokyo Medical and Dental University, 2-3-10 Kanda-Surugadai, Chiyoda, Tokyo 1010062, Japan. (1) Sakata, T.; Kamahori, M.; Miyahara, Y. Mater. Sci. Eng., C 2004, 24, 827– 832. (2) Kamahori, M.; Ishige, Y.; Shimoda, M. Biosens. Bioelectron. 2008, 23, 1046– 1054. (3) Lin, C. H.; Hung, C. H.; Hsiao, C. Y.; Lin, H. C.; Ko, F. H.; Yang, Y. S. Biosens. Bioelectron. 2009, 24, 3019–3024. (4) Sakata, T.; Miyahara, Y. ChemBioChem 2005, 6, 703–710. (5) Sakata, T.; Miyahara, Y. Biosens. Bioelectron. 2007, 22, 1311–1316. (6) Sakata, T.; Miyahara, Y. Angew. Chem., Int. Ed. 2006, 45, 2225–2228.

8946

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

arise from the binding of DNA molecules to the solid surface. A remarkable sensitivity of the genetic FET results from the abundant negative charges in nucleotides (-308 Da/charge) owing to their phosphate-deoxyribose backbone structure. Recently, even a wider range of biomacromolecular events has been investigated as new targets of detection using the FET.9,10 For example, extensive research is attempting to exploit the FET for label-free detection of protein.11-13 In the context of charge detection, proteins represent more challenging targets in the following two aspects; first, in contrast to the case of DNA, not all constituent moieties (amino acids) are charged where only five types of amino acids are primarily responsible for the electrical properties (i.e., arginine, lysine, and histidine are cationic, and aspartic acid and glutamic acid are anionic), and second, proteins are polyampholytes forming zwitterions in a certain range of pH, thus incurring insufficient magnitude of the change in charge density. Indeed, the estimated charge density of human serum albumin is -5733 Da/charge at pH 7.4. Because of the substantially lower charge density of protein compared with DNA, the detection limit of typical FET-based protein biosensors is in the order of subnanomolar.13,14 To address this issue, we attempted to “tag” additional negative charges to proteins by employing site-selective chemical modifications of positive amino acid side chains using conventional acylation and glycation reactions. Succinic anhydride (SA) reacts specifically with the ε-amino groups of lysine and the R-amino group of N-terminus, converting the residues from cationic to anionic.15 Therefore, the succinylation of each amino group leads to alternation of the net charge of a protein by up to two units. Alternatively, 2,3-butanedione (BD) reacts solely with arginine, converting from the cationic to neutral group.16-18 On the other (7) Uslu, F.; Ingebrandt, S.; Mayer, D.; Bo¨cker-Meffert, S.; Odenthal, M.; Offenha¨usser, A. Biosens. Bioelectron. 2004, 19, 1723–1731. (8) Gentil, C.; Philippin, G.; Bockelmann, U. Phys. Rev. E 2007, 75, 011926. (9) Bergveld, P. Sens. Actuators, B: Chem. 2003, 88, 1–20. (10) Ratilainen, T.; Holme´n, A.; Tuite, E.; Haaima, G.; Christensen, L.; Nielsen, P. E.; Norde´n, B. Biochemistry 1998, 37, 12331–12342. (11) Kharitonov, A. B.; Zayats, M.; Lichtenstein, A.; Katz, E.; Willner, I. Sens. Actuators, B: Chem. 2000, 70, 222–231. (12) Sakata, T.; Ihara, M.; Makino, I.; Miyahara, Y.; Ueda, H. Anal. Chem. 2009, 81, 7532–7537. (13) Maehashi, K.; Katsura, T.; Kerman, K.; Takamura, Y.; Matsumoto, K.; Tamiya, E. Anal. Chem. 2007, 79, 782–787. (14) Kim, K. S.; Lee, H. S.; Yang, J. A.; Jo, M. H.; Hahn, S. K. Nanotechnology 2009, 20, 235501. (15) Gounaris, A. D.; Perlmann, G. E. J. Biol. Chem. 1967, 242, 2739–2745. (16) Yankeelov, J. A. Biochemistry 1970, 9, 2433–2439. (17) Riordan, J. F.; McElvant, K. D.; Borders, C. L. Science 1977, 195, 884– 885. 10.1021/ac1018233  2010 American Chemical Society Published on Web 10/12/2010

hand, methylglyoxal (MG) can react with both lysine and arginine to impair their cationic charges.19-22 If a sufficient fraction of lysine and/or arginine are modified, the isoelectric point (pI) of the protein should decrease significantly. Thus, we attempted to amplify the signal of the electrical potential in regards to adsorption of the modified proteins onto a self-assembled monolayer (SAM) on the gold electrode extended to the gate of the FET.23 At the same time, there were potential risks for damaging the structural integrity of proteins upon modification of their charges so that changes in conformational integrity and biological activities were carefully investigated in this study. MATERIALS AND METHODS Materials. Bovine serum albumin (BSA), lysozyme (LZ), fluorescamine, and 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) were purchased from Sigma (St. Louis, MO, USA). 1-Undecanethiol [HS-(CH2)10CH3] (C11), methylglyoxal (MG) solution (40%), trichloroacetic acid, and 4-nitrophenylacetate were purchased from Aldrich. Succinic anhydride (SA), 2,3-butanedione (BD), guanidine hydrochloride, dithiothreitol (DTT), and iodoacetamide were purchased from Wako Pure Chemicals (Tokyo, Japan). Phenylboronic acid (PBA) was purchased from TCI (Tokyo, Japan). Dulbecco’s phosphate buffered saline (DPBS) and Tris-HC1 were purchased from Invitrogen (Carlsbad, CA, USA). Deionized ultrapure Milli-Q water (18.2 MΩ cm-1, Millipore, Bedford, MA, USA) was used throughout the experiments. Protein concentration in aqueous solution was determined spectrophotometrically by the UV absorbance at 280 nm. Molar absorptivity (εM) of 49 915 and 37 970 M-1cm-1 for BSA and LZ, respectively, was used for the calculations.24 Charge-Conversion of Proteins. Alkylation of ε-amine of lysine with SA was performed as follows; 10 mg of BSA or LZ was dissolved in 5.0 mL of 0.5 M carbonate buffer (pH 9.7). The solution was stirred at 0 °C for 30 min and was added to 14.4 mg or 61.6 mg of SA for BSA or LZ, respectively. After stirring on ice for 2 h and being kept at 4 °C overnight, the mixture was purified with a Microcon YM-10 centrifugal filter device (14 000g, 30 min, 4 °C; Millipore) three times with pure water. The final product (suc-BSA or suc-LZ) was obtained as a white powder after lyophilization. Glycation of arginine was performed with dicarbonyl reagents; 10 mg of LZ was dissolved in 5.0 mL of 0.5 M ammonium acetate buffer (pH 9.5). To the solution was added 63 mg of MG for mg-LZ and 30.2 mg of BD and 42.7 mg of PBA for but-LZ. Heat-denatured BSA and LZ were prepared by incubating with 4 M guanidine hydrochloride in 15 mM DPBS at 85 °C for 30 min. DTNB Assay. The number of free cysteine in proteins was determined by Ellman’s analysis.25 A 20 mM DTNB solution was prepared in 100 mM Tris-HCl (pH 8.0) buffer with 0.5 mM ethylenediaminetetraacetic acid (EDTA) just before the measure(18) Leithner, A.; Lindner, W. Anal. Chem. 2005, 77, 4481–4488. (19) Ahmed, M. U.; Brinkmann Frye, E.; Degenhardt, T. P.; Thorpe, S. R.; Baynes, J. W. Biochem. J. 1997, 324, 565–570. (20) Westwood, M. E.; McLellan, A. C.; Thornalley, P. J. J. Biol. Chem. 1994, 269, 32293–32298. (21) Lo, T. W. C.; Westwood, M. E.; McLellan, A. C.; Selwood, T.; Thornalley, P. J. J. Biol. Chem. 1994, 269, 32299–32305. (22) Gao, Y.; Wang, Y. Biochemistry 2006, 45, 15654–15660. (23) Goda, T.; Miyahara, Y. Anal. Chem. 2010, 82, 1803–1810. (24) Pace, C. N.; Vajdos, F.; Fee, L.; Grimsley, G.; Gray, T. Protein Sci. 1995, 4, 2411–2423. (25) Ellman, G. L. Arch. Biochem. Biophys. 1958, 74, 443–450.

ment. A 1.0 mg mL-1 protein solution was prepared in 15 mM DPBS. The 50 µL of DTNB solution was added to the 1.0 mL of the protein solution, and the resulting solution was incubated for 30 min at 30.0 °C. Then, absorbance at 412 nm was recorded using a NanoDrop 1000 spectrophotometer (Thermo Scientific). The spectrophotometer was zeroed with the DTNB solution (50 µL) added to the 1.0 mL of 15 mM DPBS. Circular Dichroism Measurements. Circular dichroism (CD) measurements were performed on a J-725 spectropolarimeter (Jasco) with a 0.1 cm path length quartz cell with temperature controller. Proteins were dissolved in 20 mM Tris-HCl buffer (pH 8.0) at 0.1 mg mL-1. CD spectra were taken in a wavelength range of 195-250 nm with a nitrogen gas flow of 12 L min-1, and each spectrum was the average of four scans. The results were expressed as mean residue ellipticity ([θ], deg cm2 dmol-1) which can be obtained using the equation, [θ] ) θ/(10nlC), where θ is the CD (mdeg) obtained from the spectra, n is the number of amino acid residues of the protein, l is the path length of the cell (0.1 cm), and C is the mol fraction of the protein. Then, percent helicity of the protein can be calculated from the [θ]208nm according to the following equation: % helix ) (-[θ]208nm - 4000)/(33 000 - 4000) × 100. Autofluorescence Measurements of the Proteins. Information regarding the tertiary structure of the proteins was attained by intrinsic fluorescence emission. The fluorescence measurements of these proteins were performed on a FP-6500 fluorescence spectrometer (Jasco). Proteins were dissolved in 15 mM DPBS (pH 7.4) at 0.5 mg mL-1. The emission spectra were recorded in the wavelength range of 300-500 nm upon excitation at 295 ± 5 nm. The choice of the excitation wavelength of 295 nm was to avoid the contribution from tyrosine. Fabrication of Extended Gate-FET Biosensor. The source and drain of a commercially available FET (2SK241, Toshiba, Japan) were connected to a real-time FET analyzer (Optogenesis, Saitama, Japan), and the gate was extended to a sputtered polycrystalline gold on a separate chip of synthetic quartz substrate (Daico MFG., Kyoto, Japan) using titanium as an adhesion layer. Surface roughness of the electrode was measured using a surface profiler (Ra ) 1.8 nm, Dektak 6M, Ulvac, Kanagawa, Japan). To form a reaction chamber (200 µL in volume) on the gold electrode, a glass tube (5 mm in inner diameter) was immobilized on the electrode with a thermosetting insulating epoxy resin (ZC-203, Nippon Pelnox, Tokyo, Japan). The rest of the chip was completely covered with the resin. The effective surface area was 0.164 cm2 (roughness factor ∼ 1), which was determined by the redox currents of ferrocyanide/ferricyanide by cyclic voltammetry. All the values in this paper were referenced to the effective surface area. The glass chamber was cleaned before use with piranha solution (H2O2/H2SO4 ) 30/70 vol/vol). (Extreme caution must be exercised when using piranha etch. An explosionproof hood should be used.) The C11 SAM was formed on the clean gold electrode by immersing in 2 mM C11 in ethanol under nitrogen atmosphere for 18 h at room temperature. The surface was then rinsed thoroughly with ethanol. Detection of Protein Adsorption. Real-time changes in the gate potential were recorded at a source current (IS) and VG of 1800 µA and 0 V (versus Ag/AgCl), respectively. The sourcedrain voltage (VD) was set at 1 V. Various concentrations of a Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

8947

Figure 1. (a) Schematic representation showing the preparation of the charge-conversional proteins by site-selective side-chain modification. (b) Reaction of lysine residues with SA, arginine with MG, or arginine with BD in the presence of PBA.

single protein solution were prepared in 15 mM DPBS solution. The SAM-formed gold gate was equilibrated in 15 mM DPBS solution for 2 h immediately prior to use. Protein adsorption was conducted by incubating the SAMs in the protein solution at 37 °C. Weakly bound proteins were removed by rinsing in 15 mM DPBS solution at 37 °C. Data were collected from at least three independent measurements. Faradic Impedance Measurement. Electrochemical impedance spectroscopy was performed using a standard three-electrode cell (Autolab PGSTAT 302) in 15 mM DPBS with a redox marker of 5 mM ferrocyanide/ferricyanide. The counter electrode was a platinum wire, and the reference electrode was Ag/AgCl. The working electrode was the SAM-formed gold. Impedance measurements were made in a wide frequency range from 1 Hz to 10 kHz with 10 points per decade, DC bias voltage of 0.2 V (versus Ag/AgCl), and AC voltage amplitude of 50 mV. Bundled frequency response analyzer was used to model the data by equivalent electrical circuit of R(Q[RW]). RESULTS AND DISCUSSION Charge-Conversion of the Proteins. The site-selective reactions of SA convert ε-amino groups of L-lysine and R-amino group of N-terminus into an acidic group (Figure 1). Direct evidence of the side-chain modification was confirmed using liquid chromatography with tandem mass spectrometry (LC-MS/MS) after the enzymatic digestion of the samples. The obtained mass shift of +100.0 Da was correlated to the succinic lysine for the modified BSA (suc-BSA), whereas no mass shift was observed for the original BSA. Detailed data and product-ion spectra of the peptide fragments are shown in the Supporting Information (Figures S1-S14). The succinic lysine was detected at 27 out of 60 sites at the sequence coverage of 68% (Figure S15, Supporting Information). The lysine adduct was also confirmed for LZ modified with SA (suc-LZ), and the mass shift of +100.0 Da was detected at two out of six sites at the sequence coverage of 67%. When the abundant arginines in LZ (12 out of 147 residues, 8.2%) were considered, the charge-conversion of the guanidinium 8948

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

group was employed by MG. The MG is a highly reactive ketoaldehyde that can be produced endogenously during aminecatalyzed sugar fragmentation reactions in the body, decomposition of triose phosphate intermediates in glycolysis, and the metabolism of acetol.26-28 In addition, the MG is responsible for nonenzymatic post-translational modification by generating the advanced glycation end-products (AGEs) which could cause enzymatic activity inhibition,29 transcriptional activation,30 apoptosis,31,32 and tissue injury.33 In this study, the MG converts cationic guanidinium group into neutral that alters the net charge by one unit (Figure 1). MG can also react with ε-amino groups of lysine to form N-(carboxyethyl) lysine.19 From the LC-MS/MS analysis, mass shifts of +52.0 and +54.0 Da of arginine residue and +72.0 Da of lysine residue were observed. Previously, the major MGarginine adducts have been reported to be 5-hydro-5-methylimidazolone (MG-H1)34,35 and argpyrimidine,36 whereas these mass shifts imply the formation of imidazolones (MG-H1 and its derivatives) and N-(carboxyethyl) lysine, respectively (Figure 1). The results are consistent with the previous observation that the major MG-arginine adduct is MG-H1 rather than argpyrimidine.22 The mass shift of arginine and lysine residues by MG was confirmed at 5 sites out of 12 and 5 sites out of 6, respectively (Figure S15, Supporting Information). The arginine-specific tagging reactions using BD and phenylboronic acid (PBA) in LZ were based on modifications with R-dicarbonyls to produce a diol group and subsequently employs a cyclization step of the moiety with arylboronic acid (but-LZ).18 The conversion of the guanidinium group accompanied by mass shifts of +58.0, +158.0, +160.0, and +174.1 Da indicates the formation of four major types of arginine adducts (Figure 1). The mass shifts were confirmed at 5 sites out of 12 (Figure S15, Supporting Information). The phenylboronic acid bestows a negative charge depending on the surrounding pH. However, the relatively high pKa of 8.8 for PBA prevents the ionization at the neutral pH 7.4. Of note, we tried to substitute PBA into 4-carboxy-3-fluorophenylboronic acid that can ionize at physiological pH,37 but the increased hydrophobicity of this PBAderivative impairs the solubility of the modified LZ in water. Generally, the modification site is related to the solvent accessibility of the side chain.38 To gain quantitative understanding for the solvent accessibility of side chains and the modification (26) (27) (28) (29) (30)

(31) (32) (33) (34) (35) (36) (37) (38)

Hayashi, T.; Mase, S.; Namiki, M. Agric. Biol. Chem. 1986, 50, 1959–1964. Hayashi, T.; Namiki, M. Agric. Biol. Chem. 1986, 50, 1965–1970. Richard, J. P. Biochemistry 1991, 30, 4581–4585. Murata-Kamiyama, N.; Kamiya, H. Nucleic Acids Res. 2001, 29, 3433–3438. Yao, D.; Taguchi, T.; Matsunuma, T.; Pestell, R.; Edelstein, D.; Giardino, I.; Suske, G.; Ahmed, N.; Thornalley, P. J.; Sarthy, V. P.; Hammes, H. P.; Brownlee, M. Cell 2006, 124, 275–286. Okado, A.; Kawasaki, Y.; Hasuike, Y.; Takahashi, M.; Teshima, T.; Fujii, J.; Taniguchi, N. Biochem. Biophys. Res. Commun. 1996, 225, 219–224. Maeta, K.; Izawa, S.; Okazaki, S.; Kuge, S.; Inoue, Y. Mol. Cell. Biol. 2004, 24, 8753–8764. Thornalley, P. J.; Edwards, L. G.; Kang, Y.; Wyatt, C.; Davies, N.; Ladan, M. J.; Double, J. Biochem. Pharmacol. 1996, 51, 1365–1372. Ahmed, N.; Argirov, O. K.; Minhas, H. S.; Cordeiro, C. A. A.; Thornalley, P. J. Biochem. J. 2002, 364, 1–14. Thornalley, P. J.; Battah, S.; Ahmed, N.; Karachalias, N.; Agalou, S.; BabaeiJadidi, R.; Dawnay, A. Biochem. J. 2003, 375, 581–592. Oya, T.; Hattori, N.; Mizuno, Y.; Miyata, S.; Maeda, S.; Osawa, T.; Uchida, K. J. Biol. Chem. 1999, 274, 18492–18502. Alexeev, V. L.; Das, S.; Finegold, D. N.; Asher, S. A. Clin. Chem. 2004, 50, 2353–2360. Sharp, J. S.; Becker, J. M.; Hettich, R. L. Anal. Chem. 2004, 76, 672–683.

Table 1. Estimated Electrical Properties of the Modified Proteins

sample

number of estimated charge modified molecular estimated charge density, Da residues weight, Da pIa at pH 7.4 per charge

BSA suc-BSA LZ suc-LZ mg-LZ but-LZ

0 27 (K) 0 3 (K) 5 (K), 5 (R) 6 (R)

69293 71993 16238 16438 17009 16856

6.2 4.6 9.1 8.0 6.0 8.1

-14.0 -67.9 +8.2 +2.2 -1.8 +2.2

-4950 -1060 +1980 +7472 -9449 +7661

a The calculation assumes all amino acid residues have pKa values that are equivalent to the isolated residues.

sites, the solvent accessibility of LZ was calculated from the NMR structure (PDB ID: 1E8L) using GETAREA software (http:// curie.utmb.edu/getarea.html). Assignments of the solvent accessibilities classify that the following modification sites in LZ, R32, R39, R132, R143, R146, K19, K31, K114, K115, and K134, were all assigned to be surface-exposed solvent-accessible amino acids (Figure S16, Supporting Information). The percentage of exposed primary amine determined from spectrophotometric measurements indicates that about 27% and 52% of primary amine of lysine were converted for suc-BSA and suc-LZ, respectively (Figures S17 and S18, Supporting Information). These values are roughly consistent with the results by LCMS/MS analysis while no remarkable decrease in percent exposed amine was observed for mg-LZ and but-LZ, probably because the major modification site for MG and BD was arginine. The significant fluorescent intensity for but-LZ suggests contribution of aromatic group of PBA in the arginine derivative. Similarly, the percentage of free arginine by the fluorometric method indicates that about 26% and 30% of free arginine were converted for mg-LZ and but-LZ, respectively, and most of the free arginine was maintained for both intact and succinylated proteins. The approximate pI value and the total amount of charge at pH 7.4 were calculated on the basis of LC-MS/MS data using a protein calculator program (http://www.scripps.edu/∼cdputnam/ protcalc.html). In this calculation, the resulting succinic lysine was assumed to have similar pKa values of aspartate or glutamate and the arginine derivatives modified by MG or BD have similar electrochemical properties to nonpolar amino acid residues. Of note, the calculation assumes all amino acid residues have pKa values that are equivalent to the isolated residues so that the theoretical pI values (Table 1) were slightly different from experimental ones as previously reported (i.e., BSA: pI 4.9; LZ: pI 10.7). The estimation had suggested that pI values of BSA and LZ drop from 6.2 to 4.6 and from 9.1 to 6.0, respectively, upon modification to suc-BSA and mg-LZ. As a result, after each modification, the total anionic charge density of BSA (-4950 Da/ charge) was increased by 5-fold (-1060 Da/charge), whereas the positive net charge of LZ (+1980 Da/charge) was significantly decreased to an extent converting the net charge of the protein from positive to slightly negative (-9449 Da/charge) at pH 7.4. Structural Integrity and Enzymatic Activities of the Proteins. Alteration of electrical properties of the proteins can potentially impinge on conformational denaturation and dysfunction of bioactivities. The circular dichroism (CD) spectrum of pristine BSA showed two negative minima at 208 and 222 nm

Figure 2. (a) CD spectra of BSA and suc-BSA in 20 mM Tris-HCl (pH 8.0) at 25 and 37 °C. (b) CD spectra of LZ, suc-LZ, mg-LZ, and but-LZ in 20 mM Tris-HCl (pH 8.0) at 25 and 37 °C.

(Figure 2), which is characteristics of an R-helical structure of protein. Interestingly enough, no remarkable changes in CD spectra were observed for suc-BSA at 25 °C. Although the peaks corresponding to R-helix were slightly lower in the suc-BSA at 37 °C than the pristine BSA, the percent R-helix presented a very limited influence on a decrease in helicity by the side-chain modifications. Trends in changes in CD spectra by the modification were very similar in between BSA and LZ. Compared with BSA, however, changes in the helicity were slightly higher for LZ than BSA. Among LZ modifications employed in this study, succinylation of lysine had the most influence on unfolding of R-helical structure. Ellman’s assay revealed retention of the number of free cysteine.25 The assay revealed that pristine BSA contained 0.5 of free cysteine, as is commonly observed for adventitiously oxidized commercial batches of BSA. While, there was no substantial increase in the number of free cysteine for suc-BSA, denatured Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

8949

Figure 4. (a) Circuit diagram for FET measurement. (b) Picture of fabricated gold electrodes as an extended gate. (c) Calculated potential distribution and a schematic of protein adsorption at the gate interface.

Figure 3. Fluorescence emission spectra of the tryptophan residues of 0.5 mg mL-1 proteins in 15 mM DPBS. (a) Native, heat-denatured, and suc-BSA. (b) Native, heat-denatured, suc-, mg-, and butlysozyme. The triangles represent peak positions of each spectrum. Excitation: 295 ( 5 nm; 25 °C.

BSA, and the LZ derivatives (Table S1, Supporting Information). For LZ, the results were in good agreement with the fact that natural forms of LZ do not contain free thiols. Hence, the sidechain modifications do not cleave disulfide linkages although MG has been previously reported to produce hemithioacetal from cysteine.20,21 The tertiary conformation of the proteins were assessed by observing the intrinsic fluorescence emission by tryptophan residues.39 BSA possesses two tryptophan residues; one (W134) is located on the surface of the molecule in domain I, whereas the other (W213) is on the bottom of hydrophobic pocket in domain II. Thus, on observing the fluorescence emission of tryptophan in suc-BSA, information about tertiary conformation of the protein can be obtained. Commonly, the blue shift of the emission maximum reflects transferring tryptophan residues into a more hydrophobic environment, and the red shift indicates a more exposed tryptophan to the solvent. The choice of the excitation wavelength of 295 nm enabled to avoid the contribution from tyrosine. The suc-BSA resulted in a slight blue shift (3 nm) of the emission maximum, whereas heat-denatured BSA induced a 9 nm red shift (Figure 3). For suc-LZ and mg-LZ, peak shifts of emission maximum were very limited. The but-LZ resulted in a 3 nm blue shift, while heat-denatured LZ was a 7 nm red shift. From these results, no significant changes occurred in the spatial relationship between the tryptophan residues by the side-chain modifications. It could be seen from the emission spectra that the fluorescence was quenched dramatically by the side-chain modifications, but the underlying mechanism of the quenching is unknown. Hydrolytic activity of BSA and enzymatic activity of LZ were evaluated. Generally, most proteins contain a number of functional groups which are capable of reacting with aryl esters such as p-nitrophenyl acetate to release p-nitrophenol. The esterase-like (39) Shang, L.; Wang, Y.; Jiang, J.; Dong, S. Langmuir 2007, 23, 2714–2721.

8950

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

activity of BSA consists of a single site acetylation of the protein and the catalytic hydrolysis of the ester at two separate sites on the molecule, requiring the conformational integrity of the protein.40 Noteworthy is that the hydrolytic activity of BSA increased by 8-fold by the succinylation (Figure S19, Supporting Information), whereas significant loss of hydrolytic activity (1.5%) was observed for heat-denatured BSA. The substantial increase in bioactivity of suc-BSA might be both due to the effect of succinylation of lysine and due to minor spatial rearrangement of the tertiary structure which account for the catalytic activity. Therefore, the maintained esterase-like activity evidence the folded conformation of suc-BSA. The lytic activity of the LZ was evaluated by measuring decomposition velocity of ML cells. The suc-LZ and mg-LZ demonstrated 94 ± 15% and 87 ± 2% retention of activity compared to pristine LZ, respectively, whereas the activity was reduced to 41 ± 5% for but-LZ (Figure S20, Supporting Information). Heat-denatured LZ retained 89 ± 6% of its activity. These results indicate that the side-chain modifications have minor influences on the bioactivities. Signal Amplification by Charge-Conversion of Proteins in the FET Measurements. To verify the signal amplification by charge-conversion of the proteins, we demonstrated nonspecific protein adsorption onto the C11 SAM using the extended gateFET. Since the electrical double layer works as an electrical capacitor, attachment of charged proteins on the solid surface produces changes in interfacial electrical potential (∆φ). The home-built FET analyzer can directly readout ∆φ induced by protein adsorption onto the SAM as changes in the output potential of the electrical circuit (∆VG, Figure 4). The Debye length is the characteristic distance required for screening the surplus charge by the mobile counterions in a specific ionic strength solution.41 From this principle, a biomolecule placed outside the electrical double layer is screened and cannot be detected. Here, a dilute DPBS solution (15 mM) was used to ensure the characteristic length of 2.5 nm which is relatively comparable to dimensions of the proteins. Of note, the configuration of the extended gate-FET is advantageous not only in diversity of choices in shape and material of the electrode but also in stability of the signal due to separation of the FET chip from the measurement environment.23 Furthermore, because the fabrication process of extended gateFET is amenable to that of semiconductor devices, the sensor is easily and inexpensively miniaturized on the micrometer scale and (40) Tyson Tildon, J.; Ogilvie, J. W. J. Biol. Chem. 1972, 247, 1265–1271. (41) Wang, J.; Bard, A. J. J. Phys. Chem. B 2001, 105, 5217–5222.

Figure 5. Time course of potential shift with adsorption/rinse processes for stepwise increases in BSA or suc-BSA concentrations: pH 7.4; 37 °C.

can be integrated into microfluidics and microelectronics. The packing density of the C11 SAM was 3.0 chains nm-2 determined by cyclic voltammogram, and the surface was hydrophobic enough to facilitate protein adsorption (Figure S21, Supporting Information). When the electrode was in contact with the BSA solution, negative shifts of the potential were instantly observed and increased with a stepwise increase in the bulk concentration in the order of magnitude from 10-4 to 100 mg mL-1 (Figure 5). The potential shifts were caused by the contribution of both weakly and tightly adsorbed proteins, whereas those after rinsing the electrode with the buffer can be assigned as the contribution of tightly adsorbed ones. Generally, tightly bound proteins via hydrophobic association, electrostatic interactions, or hydrogen bonding to the substrate undergo slow conformational changes and finally become irreversible physisorption. It is obvious from the results that the potential shifts were larger for suc-BSA than pristine BSA. The potential shifts as a function of bulk BSA concentration were summarized in Figure 6a. The signals corresponding to the adsorption of BSA and suc-BSA in the bulk solution were commonly detected at g10-2 mg mL-1, whereas the degree of potential shift was 1.5- and 2.9-fold higher in sucBSA at the concentration of 10-1 and 100 mg mL-1, respectively. The potential shifts that were measured after rinsing also increased by 3.5-fold for suc-BSA (Figure 6b). The ratios of the potential shifts between the pristine and modified proteins were roughly consistent with those of the estimated charge densities (Table 1). Although trends in the potential shifts were quite similar in between BSA and LZ, the amplifications in the potential shifts were more significant in LZ. The potential shifts for pristine LZ were very low throughout the concentration range, whereas substantial potential shifts were detected for the LZ derivatives at g10-3 mg mL-1 (Figure 6c). The potential shift increased by 7.9-, 3.3-, and 11-fold in the bulk solution of suc-LZ, mg-LZ, and but-LZ at 100 mg mL-1 compared with that in the bulk solution of pristine LZ. Changes in the potential shifts obtained after rinsing were slightly suppressed (Figure 6d). The potential shift was 1.1-, 2.2-, and 3.5-fold higher for suc-LZ, mg-LZ, and but-LZ, respectively, than for pristine LZ after rinsing. The negative potential shifts by

Figure 6. Detection of electrical signals caused by adsorption of pristine or charge-converted proteins onto the C11 SAM-immobilized gold electrode. (a, b) Changes in the potential shift as a function of bulk BSA concentration (a) and potential shift measured in the bulk BSA solution of 1.0 mg and after rinsing with DPBS for 30 min (b). (c, d) Changes in the potential shift as a function of bulk LZ concentration (c) and potential shift measured in the bulk LZ solution of 1.0 mg and after rinsing with DPBS for 30 min (d).

the adsorption of cationic LZ at pH 7.4 were seemingly inconsistent. We speculate that the substantial increase in the interfacial capacitance during the protein adsorption contribute to the negative potential shift.23 The potential shift can be described as: ∆φ ) Qafter/Cafter - Qbefore/Cbefore, where Q and C denote interfacial electrical charges and capacitances before and after protein adsorption, respectively. The above equation clearly denotes that increased capacitance during the protein adsorption yields a negative potential shift. Indeed, we revealed a significant increase in the interfacial capacitance on the SAMcoated electrode with adsorbed proteins using electrochemical impedance spectroscopy (Figure 7). The Nyquist plots indicate that a significant decrease in the charge transfer resistance of ferrocyanide/ferricyanide was accompanied by the increased interfacial capacitance. This signifies that the redox probes are more accessible to the gold electrode across the C11 SAM after the protein adsorption. The observations are explained by the partial penetration of protein molecules together with water of hydration into the hydrophobic SAM, thus increasing the overall permittivity of the capacitor, thereby increasing the interfacial capacitance. In fact, the interfacial capacitance decreased contrary when the protein adsorption was performed on the gold electrode without C11 SAM (Figure 7). The decreased capacitance is due to the replacement of water molecules into proteins that yields overall decreased permittivity.42 Hence, the positive potential shift due to the adsorption of cationic LZ onto the electrode is possibly canceled or even overwhelmed by the capacitive changes. When the capacitive effect is taken into account, charge-conversion of proteins into the negative is preferable for obtaining amplified signals in the FET measurements. The increased negative (42) Smiechowski, M. F.; Lvovich, V. F.; Roy, S.; Fleischman, A.; Fissell, W. H.; Riga, A. T. Biosens. Bioelectron. 2006, 22, 670–677.

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

8951

Figure 7. (a) Interfacial capacitances of the gold electrode with and without the C11 SAM and the protein adsorptions. (b) Nyquist plots of the C11 SAM-coated gold electrode before and after the protein adsorptions. Inset: Nyquist plot on a large scale. The measurements were performed in 15 mM DPBS with 5 mM ferrocyanide/ferricyanide.

potential shift by the mg-LZ and but-LZ can also be explained by both the capacitive effect and decreased positive charge of the protein. In Situ Charge Conversion of Adsorbed LZ. Either increased negative charge density of each protein or increased amount of adsorption attributes to these amplified potential shifts. However, the increased negative charges of the modified proteins unlikely promote adsorption against the electrode that has a negative surface potential (-40 mV).40 In addition, maintained structural integrity of suc-BSA suggests no significant changes in adsorption behavior. Therefore, we conjecture that the amplified potential shifts mainly originate in the increased net negative charge of BSA by the succinylation. To support our hypothesis that the increased negative potential shifts are mainly caused by the readout of increased negative charges of each protein, we performed in situ side-chain modification of the LZ physisorbed onto the SAM (Figure 8). Changes in the potential on the SAM without adsorbed LZ served as a control. The potential shift increased by -13 mV with an increase in the MG concentrations from 5.5 fM to 5.5 µM, whereas no significant increase in the potential shift was observed at the concentration range in the absence of LZ. The data were statistically significant (p < 0.1) at the MG concentration of 5.5 pM. Because LZ is irreversibly physisorbed on the SAM, these potential shifts do not contain changes in the amount of adsorbed protein nor any contributions of capacitive changes caused by the adsorption. Therefore, we attribute the negative potential shifts to the loss of positive charges of adsorbed LZ by the glycation. Unexpected potential shifts at the highest MG concentration (5.5 mM) were commonly 8952

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

Figure 8. (a) Real-time changes in the potential by in situ sidechain modification of LZ adsorbed onto the SAM by MG: 1, injection of 2 mg mL-1 LZ; 2, rinsing with DPBS; 3-7, injection of MG (5.5 fM, 5.5 pM, 5.5 nM, 5.5 uM, and 5.5 mM, respectively); 8, rinsing with DPBS. (b) Potential shifts as a function of MG concentration on the SAM-coated gold electrode surface with and without LZ attachment.

observed on the SAM surface independently of the LZ adsorption. Since MG is a reducing agent,43 we infer that MG could cleave the gold-alkanethiol bonds and that interfacial capacitance was increased by the replacement of hydrophobic C11 with water molecules on the electrode. This explanation is in good agreement with the observation that the increased capacitance results in the negative potential shift in the FET measurement. Our observations suggest a new method for in situ real-time detection of post-translational modifications of proteins using FET. Post-translational modifications are covalent processing by proteolytic cleavage or by addition of a modifying group to one or more amino acids that can modulate the activity state, localization, turnover, and interactions with other proteins.44 For example, phosphorylation, which changes protein charge and the mass of +80 Da, is closely related to activation/inactivation of enzyme activity, modulation of molecular interactions, and signaling. Mass spectrometry (MS) in combination with stable isotope coding is commonly used to identify proteins.45 While, MS requires an expensive apparatus, pretreatments of the samples, and chemical reagents of extremely high-purity. Hence, the establishment of real-time in situ detection of post-translational modifications will be a large impact. As shown, the extended gate-FET biosensor successfully detected changes in total charge of protein caused by nonenzymatic post-translational modifications without pro(43) Yim, H. S.; Kang, S. O.; Hah, Y. C.; Chock, P. B.; Yim, M. B. J. Biol. Chem. 1995, 270, 28228–28233. (44) Mann, M.; Jensen, O. N. Nat. Biotechnol. 2003, 21, 255–261. (45) Julka, S.; Regnier, F. J. Proteome Res. 2004, 3, 350–363.

teolytic pretreatments. Applications to this end, mechanistic studies, are currently under investigation. CONCLUSIONS We reported the stable, yet not permanent, chemical modifications of protein as a labeling technique for sensitive potentiometry; the modifications rely on succinylation and glycation of the cationic amino acid residues with the anhydride and dicarbonyl compounds, respectively, that could convert the total charge of the proteins. Due to the site-selectivity of the reactions to the surfaceexposed cationic residues, the protein derivatives preserve their folded conformations. Potentiometric signal in the detection of protein adsorption in physiological pH is amplified by the increased charge density of the modified proteins. The detection of in situ charge-conversion is possible for the tightly bound proteins to the solid surface. The molecularly engineered charge conversion technique will likely find wide use in potentiometric

biosensors and electrochemical detection of post-translational modifications of protein. ACKNOWLEDGMENT This research was supported in part by JST, CREST. Preparation of gold electrodes was supported by the Nano-Integration Facility (NIMS). LC-MS/MS and CD measurements were supported by Bio-Organic Materials Facility (NIMS). We thank Dr. H. Kobayashi (NIMS) for offering the spectrofluorometer and Dr. T Aoyagi (NIMS) for offering the spectrophotometer. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs. org. Received for review July 13, 2010. Accepted September 23, 2010. AC1018233

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

8953