Microchemical Journal 65 Ž2000. 189]195 Compositional analysis of small peptides using capillary electrophoresis and Ruž bpy/ 33q-based chemiluminescence detection Howard P. Hendrickson1, Paul Anderson2 , Xin Wang, Zidia Pittman, Donald R. BobbittU Department of Chemistry and Biochemistry, Uni¨ ersity of Arkansas, Fayette¨ ille, AR 72701, USA Received 22 November 1999; received in revised form 25 May 2000; accepted 26 May 2000 Abstract RuŽbpy. 33q-based chemiluminescence detection of underivatized amino acids was coupled to capillary electrophoresis and evaluated as a separation and detection methodology for the compositional analysis of peptides and proteins. Two tripeptides were studied: glycine]phenylalanine]alanine, and valine]proline]leucine, and the method was demonstrated to be quantitative and reproducible at the 10-mg and 5-mg peptide levels, with a limit of detection ŽLOD. of 2.5 pmol for gly]phe]ala and 80 fmol for val]pro]leu ŽSrNs 2.. RuŽbpy. 33q-based chemiluminescence detection coupled to CE offers a sensitive and time-efficient method for the determination of the amino acid composition of peptides or proteins without derivatization. The detection limits for most amino acids were demonstrated to be at, or below the level achievable using ninhydrin or phenyl isothiocyanate derivatization. Amino acids were identified by their migration time through the capillary and their relative luminescence. The relative luminescent response obtained upon reaction of RuŽbpy. 33q with amino acids was dependent upon the amino acid R-group at the a-carbon, and the relative response may be useful in broadly classifying a number of amino acids. Q 2000 Elsevier Science B.V. All rights reserved. Keywords: RuŽbpy. 33q-based chemiluminescence detection; Capillary electrophoresis; Peptides U Corresponding author. Tel.: q501-575-6861; fax: q501-575-4049. E-mail address: dbobbitt@comp.uark.edu ŽD.R. Bobbitt.. 1 Present Address: Department of PharmacologyrToxicology, University of Arkansas School for Medical Sciences, Little Rock, AR 72205, USA. 2 Present Address: Femtometrics, Irvine, CA 92614, USA. 0026-265Xr00r$ - see front matter Q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 6 - 2 6 5 X Ž 0 0 . 0 0 0 5 2 - 7 190 H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 1. Introduction Information about the amino acid composition of a peptide or protein is critical to gaining insight into the relationship between peptide structure and function. The determination of the amino acid composition of a peptide or protein can reveal information about its size, the presence of labile residues, and the molar ratio of amino acids present w1x. Traditionally, for compositional analysis, peptides and proteins are first hydrolyzed, the amino acids are then derivatized with, for example, ninhydrin w2x, phenyl isothiocyanate, or dansyl chloride Ž1-dimethylaminonaphthalene5-sulfonyl chloride. to enhance detection. The derivatives are then separated by reversed-phase chromatography w3x and detected spectrophotometrically at 540 nm for ninhydrin derivatives, 280 nm for isothiocyanate derivatives, or by the resulting fluorescence for the derivatives of dansyl chloride. Ultimately, the molar absorptivity or quantum efficiency of the derivatizing agent, the efficiency of the derivatizing reaction, the quality of the separation mode, and the stability of the amino acid under hydrolysis conditions restrict the detection limit w4x. Recent reports have demonstrated improved limits of detection and separation efficiencies by first derivatizing the amino acid with a fluorescent label and then separating the mixture by capillary electrophoresis. Jansson, Roeraade, and Laurell have labeled amino acids with naphthalene-2,3-dicarboxaldehyde and reported limits of detection of 900 zmol w5x. Thermo-optical absorption w6x, chemiluminescence w7x, and electrochemical detection w8x have also been used to detect derivatized amino acids. Their limits of detection refer strictly to the separationrdetection process, as the actual amount of amino acid required to react with the derivatizing agent is in the mmol-to-mmol range. In addition, the separation of derivatized amino acids by a high efficiency separation mode, for example, CE, is not straightforward, since a surfactant must be added to the running buffer for neutral derivatives, or the capillary must be chemically modified to modify the EOF and enhance the selectivity of the separation. The enhanced selectivity is necessary due to the fact that the derivatizing moiety added to enhance detection usually dominates the physiochemical properties of the derivatives, thereby masking subtle differences that could be exploited to facilitate separation. For example, in CE, since migration through the capillary is based in part on the mass-to-charge ratio, the large molecular weight of the derivatizing moiety relative to the amino acid usually dominates the electrophoretic properties of the derivative. In addition, in many cases these compounds have been found to non-specifically adsorb to the fused silica capillary walls, causing peak broadening. Thus, from both a separation and detection perspective, derivatization procedures add complexity to both the qualitative and quantitative aspects of peptide and protein compositional analysis. CL detection in CE has been shown to be ideally suited to the challenging volume Žnl. and detection limit requirements characteristic of this high efficiency separation technique. Various CL reagents and schemes have been demonstrated for detection in CE including: luminol w9]12x; peroxyoxalate w13]15x; and acridium-based CL protocols w16,17x. Tris Ž2,29. bipyridyl rutheniumŽIII. has recently been shown to be an important chemiluminescent reagent for the detection of amines and amino acids. The active RuŽbpy. 33q reagent can be prepared electrochemically in a high yield from its stable 2 q state, while the reactive 3 q state provides a high quantum yield for emission. Various interfaces between the RuŽbpy. 33q-based CL detection approach and CE have been described w18,19x. Brune and Bobbitt have shown that RuŽbpy. 33q-based chemiluminescent detection of underivatized amines and amino acids is feasible and that the chemiluminescent intensity is dependent on the electron-withdrawing character of the R-group attached to the a-carbon w20x. This suggests that the amino acid-dependent CL response may be used to aid in the identification of a given amino acid w21x. He, Cox and Danielson w22x, and Jackson w23x have further shown that amino acids with a secondary amine functionality produce more luminescence than amino acids with a primary amine. H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 In this paper, the tripeptides gly]phe]ala and val]pro]leu were hydrolyzed using a commercially available instrument; the resulting free amino acids were separated by capillary electrophoresis using an unmodified capillary and a sodium borate running buffer, and then detected via RuŽbpy. 33q-based chemiluminescence. These peptides were chosen to demonstrate the flexibility of the method in detecting amino acids of different chemical functionality. The amino acids were identified based on their CE migration time and their relative luminescence. The quantitative capabilities of the technique was evaluated and compared to a commercially available instrument which utilizes ninhydrin derivatization for detection. 2. Experimental section Reagents. All amino acids and peptides were purchased from Sigma ŽSt. Louis MO, USA. and used as received. Boric acid, sodium borate, trisŽ2,29-bipyridyl. rutheniumŽII. chloride hexahydrate, and a 76-mm platinum Ž99.99% pure. wire were purchased from Aldrich ŽMilwaukee WI, USA.. Acetonitrile ŽHPLC-grade. and sodium hydroxide were obtained from Fisher Scientific ŽPlano TX, USA.. All solutions were prepared using Milli-Q water ŽSybron-Barnstead, Boston MA, USA.. Capillary electrophoresis. A 75-mm i.d.=92-cm capillary ŽPolymicro Technologies, Inc., Phoenix AZ, USA. was used for all separations with a running buffer consisting of 15 mM sodium borate, pH 9.0, with 10% Žvrv. acetonitrile, unless otherwise noted. Acetonitrile was added to the buffer to dissolve the more hydrophobic amino acids. This buffer was filtered through a 0.45-mm nylon filter ŽSigma. and then allowed to equilibrate with room air for at least 2 h. An applied voltage of 20 kV was employed for all separations producing a run current of 5 mA. Samples were injected hydrodynamically by placing the analyte solution 8 cm above the cathode for 10 s. The 191 injection volume, V Žcm3 . was estimated using Eq. Ž1.: Vs r gD Hd 4 p t . 128hL Ž1. In Eq. Ž1., D H is the height difference between the anode and cathode ends of the capillary Žcm.; g is the gravitational constant Ž980 cmrs 2 .; r is the density of the liquid Ž0.99707 grcm3 .; d is the capillary diameter Žcm.; t is the time Žs.; h is the buffer viscosity Ž0.8904= 10y2 gsrcm.; and L is the capillary length Žcm.. Using values for the parameters which are characteristic of the experiments described in this study, the injection volume was estimated to be 11 nl. RuŽbpy. 33q was generated from RuŽbpy. 32q at a 76-mm Žo.d.. platinum working electrode operated at 1.1 V vs. an AgrAgCl reference using a Princeton Applied Research Potentiostat ŽPrinceton, NJ, USA.. The AgrAgCl reference was prepared by anodizing chloride onto a 1-mm Žo.d.. silver wire in 1.2 M hydrochloric acid at 1.5 V vs. a sodium-saturated calomel electrode ŽSSCE.. A stainless steel spatula served as the auxiliary electrode and as a reservoir for 125 ml of the 20-mM RuŽbpy. 3 Cl 2 reagent, placed at the capillaryr electrochemical cell interface. RuŽbpy. 3 Cl 2 was prepared in 15 mM boric acid. A schematic diagram of the electrode arrangement and light detection apparatus is given in Fig. 1. The electrophoretic current was separated from the electrochemical current by an on-column fracture which was covered with a Nafion tube ŽPerma Pure, Inc., Toms River, NJ, USA., as described by O’Shea et al. w24x. A section of polyimide coating was removed from the distal end of the capillary to allow light to pass. Chemiluminescent emission was detected by a Hamamatsu R928 photomultiplier tube ŽPMT. ŽNew Jersey, USA. operated at a bias of y800 V. The signal from the PMT was fed into a Keithley ŽCleveland, OH, USA. model 485 picoammeter. The output of the PMT was filtered with a 0.15-s passive filter and then digitized using the ArD converter of a Stanford Research Systems ŽSunnyvale, CA, USA. SR575 lockin amplifier, and smoothed using a nine-point averaging routine. 192 H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 Table 1 Detection limits for selected amino acids determined via RuŽbpy. 33q-based chemiluminescencerCE Fig. 1. Schematic of the experimental arrangement for RuŽbpy. 33q-based chemiluminescence detection in CE. C s capillary showing electrical discontinuity Žjoint.; RE s reference electrode; AE s auxiliary electrode; WE s working electrode; R s Ru Ž bpy . 32 q reservoir; and PM T s photomultiplier tube. Peptide digestion. A solution of each peptide was prepared in 50:50 Žvrv. waterracetonitrile. A 200-ml aliquot of this concentrated peptide was digested in constantly boiling hydrochloric acid ŽPierce, Rockford, IL, USA. under N2 at 1158C for 24 h, using a Waters Pico-Tag AA analyzer. Sequanal TM -grade triethylamine ŽPierce. was used to neutralize the digested peptide, which was then dried in vacuo and diluted with 100 ml of running buffer prior to analysis. Amino acid LOD Žpmol; srn s 2.a Pro Leu Val Glu Asp Phe Ala Ser Arg Gly 9.9= 10y3 8.2= 10y2 8.2= 10y2 8.3= 10y2 0.11 0.12 0.25 0.78 0.89 2.5 a Signal-to-noise ratio calculated from peak height and peak]peak noise sampled over 1.0 min. amino acids shown in Table 1. For example, when wglyxrmM was plotted against peak height, a straight line resulted, with a least-squares fit of y s 3.8 = 10y4 x q 0.03 Ž r 2 s 0.998.. At wglyx greater than 1 mM, the response leveled off, 3. Results and discussion Table 1 shows the limits of detection ŽLOD. for ten amino acids which were detected by RuŽbpy. 33q-based chemiluminescence following their separation by capillary electrophoresis. It is important to note that these differences in the LOD for the amino acids shown here are not due to differences in separation efficiencies, but in the relative amount of luminescence produced for each amino acid. A typical electropherogram is shown in Fig. 2 for the separation of 20 pmol of glycine, alanine, and phenylalanine. A pH of 9.0 was chosen for the separation buffer to optimize both the separation and detection of the amino acids shown here w21x. The response was linear over at least two orders of magnitude for the Fig. 2. Electropherogram of alanine ŽA., glycine ŽG., and phenylalanine ŽF.. Twenty pmol of each amino acid were hydrodynamically injected onto a 75-mm i.d.=360-mm o.d. fused silica capillary. Separation was achieved at 20 kV with a running buffer consisting of 15 mM sodium borate, pH 9.0, with 10% Žvrv. acetonitrile. Compounds were detected using RuŽbpy. 33q-chemiluminescent detection as described in the text. H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 Fig. 3. Electropherogram of 10 mg gly]phe]ala hydrolysate. The peptide was hydrolyzed in hydrochloric acid for 24 h under nitrogen, neutralized with triethylamine, dried in vacuo, diluted with 100 ml of running buffer and hydrodynamically injected onto a 75-mm capillary, as described in the text. presumably due to an insufficient amount of RuŽbpy. 33q available for reaction with the amino acid. Similar quantitative results were observed for the other amino acids evaluated and listed in Table 1. In Fig. 2, each peak represents the injection of 20 pmol of amino acid, demonstrating the efficacy of this technique for the identification 193 of amino acids by both migration time and relative luminescence. The LOD for the analysis of these three amino acids was 2.5 pmol, based on the response for glycine, the poorest luminescing amino acid studied here. Bidlingmeyer, Cohen, and Tarvin reported an LOD of 1 pmol under optimal reaction conditions for the detection of PTH-derivatized amino acids w3x. One hundred pmol of protein were required to accurately determine the amino acid composition of a peptide. An improvement in the SrN can be expected if light is sampled at a point where luminescence from the amino acid]RuŽbpy. 33q reaction is maximized relative to the background luminescence. Jackson has demonstrated that when the analyte emission zone and optics are matched, improved detection limits and a greater dynamic range are possible w21x. Under the separation and optical conditions shown here, 30]50 000 theoretical plates were typical, as shown in Fig. 2. The analysis of acid hydrolyzed gly]phe]ala using RuŽbpy. 33q-chemiluminescence detection and capillary electrophoresis separation is shown in Fig. 3. The large peak at approximately 6 min was triethylamine, which was used to neutralize unreacted acid following the hydrolysis procedure. The relative mole ratio of each amino acid in the peptide, as determined by this method, is shown in Table 2. At both the 10-mg and 5-mg level of hydrolyzed peptide, 100% " 15% of the peptide was recovered. Approximately 5% per- Table 2 Amino acid composition of gly]phe]ala, and val]pro]leu as determined by capillary electrophoresis and RuŽbpy. 33q-based chemiluminescent detection Peptide mass Žmg. Amino acid Relative peak heighta ŽmV. Relative moles Gly]phe]ala Ž10.0. Ala Gly Phe Ala Gly Phe Leu Pro Val 0.301" 0.048 0.022" 0.004 1.09" 0.16 0.273" 0.006 0.018" 0.001 1.19 " 0.1 0.034" 0.005 1.0" 0.05 0.101" 0.005 1.0" 0.2 0.9" 0.2 1.1" 0.2 0.9" 0.1 0.8" 0.1 1.2" 0.1 1.1" 0.1 0.9" 0.1 1.1" 0.1 Gly]phe]ala Ž5.0. Val]pro]leu Ž6.6. a Values; mean " S.D. relative to phenylalanine or proline, with N s 3]4. Amino acid detection using RuŽbpy. 33q-based chemiluminescence following separation by capillary electrophoresis; hydrodynamic injection at a height of 8 cm for 10 s. 194 H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 cent of this uncertainty could be attributed to the reproducibility of the hydrodynamic injection method used here. Uncertainty in the mole ratio for each amino acid was much less, since values were obtained from peak height ratios. In order to show the ability of this method to work under conditions of poor chromatographic resolution, val]pro]leu hydrolysates were separated using a 100-mm i.d. capillary instead of the 75-mm i.d. capillary used to obtain the electropherograms in Figs. 2 and 3. An electropherogram of a val] pro]leu hydrolysate under these conditions is shown in Fig. 4. The initial increase in luminescence observed at the beginning of the electropherogram is more evident in Fig. 4 than in previous electropherograms, because of the scaling factor. This initial increase occurred when the high voltage was turned on at the start of a separation, and more hydroxide was available in the running buffer to react with the RuŽbpy. 33q, resulting in increased background luminescence. Brune and Bobbitt have shown that RuŽbpy. 33q chemiluminesces in the presence of hydroxide w20x. This background luminescence decreases with time as the RuŽbpy. 32q surrounding the Ptgenerating electrode is depleted, until a steadystate luminescence is reached within approximately 2 min. In a separate experiment, repeated injections of the mixture were made to illustrate the reproducibility of the method. From these injections, peak heights for the three amino acids in the tripeptide exhibited a relative standard deviation ŽR.S.D.. of 4]5%, which is within the R.S.D. expected for the hydrodynamic injection method. It is also clear that the resolution between valine and leucine was much less than 1.0 as a result of the larger capillary; however, the CErCL response is still adequate for identification and quantitation. The relative molar ratio of each amino acid in val]pro]leu as determined by this method is shown in Table 2, and the experimental results agree with that predicted from the known composition of the tripeptide. 4. Conclusions A method for the amino acid compositional Fig. 4. Electropherogram of 6.6 mg val]pro]leu acid hydrolysate. The peptide was hydrolyzed in hydrochloric acid for 24 h under nitrogen, neutralized with triethylamine, dried in vacuo, diluted with 400 ml of running buffer, diluted 10-fold in running buffer, and hydrodynamically injected onto a 100-mm i.d. capillary, as described in the text. analysis of peptides has been demonstrated using RuŽbpy. 33q-based chemiluminescence detection coupled to capillary electrophoresis. Underivatized amino acids were detected post-capillary after separation on an unmodified fused silica capillary using a sodium borate running buffer. Since the amino acids studied here were underivatized, they were easily separated using a running buffer without a surfactant. The absence of a surfactant in the running buffer allows higher separation potentials to be used, increasing the efficiency of the separation and the simplicity of the buffer, and making the method more versatile, robust and amenable to the clinical laboratory. Waldron and Dovichi have noted that, in the presence of sodium dodecyl sulfate, bubble formation in the capillary limits the magnitude of the separation potential that can be used w6x. In cases where the selectivity of the separation is poor and when the relative chemiluminescence efficiency differs for the two overlapping analytes, the detection method shown here can be used to H.P. Hendrickson et al. r Microchemical Journal 65 (2000) 189]195 identify the amino acid based on its relative luminescence and its migration time. It has previously been demonstrated that the noise limiting this detection protocol is PMTlimited and not effected by noise at the RuŽbpy. 33q-generating electrode w21x. Therefore, the use of a cooled PMT with a correspondingly smaller dark current will allow one to improve the limit of detection substantially. This study also made no attempt to spatially distinguish the light collected due to analyte from background luminescence and further improvements in the SrN might also be expected if the signal from the analyte is collected more efficiently using a narrow-field lens mounted directly to the PMT. As configured, however, RuŽbpy. 33q-based chemiluminescence detection of amino acids separated by capillary electrophoresis was demonstrated to have a limit of detection of 2.5 pmol of gly]phe]ala in an analysis time of less than 12 min. The limit of detection reported here was based on the limit of detection for glycine, which was the least luminescent amino acid studied. The corresponding limit of detection for val]pro]leu was lower at 80 fmol, with an analysis time of less than 25 min. The method reported here, although not fully optimized, is less complex than existing methods requiring derivatization and has clear advantages in terms of speed, convenience, and a demonstrated limit of detection. Capillary electrophoresis is an extremely efficient sample utilization separation mode, allowing multiple determinations with sub-microliter quantities of samples resulting in an increase in experimental precision. Recently, sampling methods have been developed for CE which require nanoliter quantities of sample compared with the 20-to 100-ml quantities, which are typical with HPLC methods, and which provide similar injection-to-injection precision as fixed loop HPLC injectors. Acknowledgements This work was supported in part by the Natio- 195 nal Science Foundation through grant CHE9619557. References w1x W.A. Schroeder, in: H.O. Halvorson, H.L. Roman, E. Bell ŽEds.., The Primary Structure of Proteins, Harper & Row, New York, 1968, p. 47. w2x R.J. West, Chem. Educ. 42 Ž1965. 386]390. w3x B.A. Bidlingmeyer, S.A. Cohen, T.L. Tarvin, J. Chromatogr. 336 Ž1984. 93]104. w4x B.J. Smith ŽEd.. Protein Sequencing Protocols, in: Methods in Molecular Biology, vol. 64, Chaps 19]20, Humana Press, Totowa, NJ, 1997. w5x M. Jansson, J. Roeraade, F. Laurell, Anal. Chem. 65 Ž1993. 2766]2769. w6x K.C. Waldron, N.J. Dovichi, Anal. Chem. 64 Ž1992. 1396]1399. w7x N. Wu, C.W. Huie, J. Chromatogr. 634 Ž1993. 309]315. w8x T.J. O’Shea, P.L. Weber, B.P. Bammel, C.E. Lunte, S.M. Lunte, M.R. Smyth, J. Chromatgr. 608 Ž1992. 189]195. w9x J.Y. Zhao, J. Labbe, N.J. Dovichi, J. Microcolumn Sep. 5 Ž1993. 331]339. w10x R. Dadoo, A.G. Seto, L.A. Colon, R.N. Zare, Anal. Chem. 66 Ž1994. 303]306. w11x M. Hashimoto, K. Tsukagoshi, R. Nakajima, K. Kondo, J. Chromatogr. 832 Ž1999. 191]202. w12x B. Huang, J.J. Li, L. Zhang, J.K. Cheng, Anal. Chem. 68 Ž1996. 2366]2369. w13x T. Hara, J. Yokogi, S. Okamura, S. Kato, J. Chromatogr. A 652 Ž1993. 361]367. w14x L.L. Shultz, S. Shippy, T.A. Nieman, J.V. Sweedler, J. Microcolumn Sep. 10 Ž1998. 329]337. w15x K. Tsukagoshi, Y. Okumura, R. Nakajima, J. Chromatogr. A 813 Ž1998. 402]407. w16x M.A. Ruberto, M.L. Grayeski, Anal. Chem. 64 Ž1992. 2758]2762. w17x M.A. Ruberto, M.L. Grayeski, J. Microcolumn Sep. 6 Ž1994. 545]550. w18x J.A. Dickson, M.M. Ferris, R.E. Milofsky, J. High Resol. Chromatogr. 20 Ž1997. 643]646. w19x X. Wang, D.R. Bobbitt, Anal. Chim. Acta 383 Ž1999. 213]219. w20x S.N. Brune, D.R. Bobbitt, Anal. Chem. 64 Ž1992. 166]170. w21x W.A. Jackson, D.R. Bobbitt, Anal. Chim. Acta 285 Ž1994. 309]320. w22x L. He, K.A. Cox, N.D. Danielson, Anal. Lett. 23 Ž1990. 195]199. w23x W.A. Jackson, Ph.D., Dissertation, University of Arkansas, Fayetteville, AR, Ž1994.. w24x T.J. O’Shea, R.D. Greenhagen, S.M. Lunte et al., J. Chromatogr. 593 Ž1992. 305]312.