Determination of Biophysical Parameters of Polypeptide Retro

ACS eBooks; C&EN Global Enterprise ..... Biochemical and Biophysical Research Communications 2007 364, 608-613 ... Hooi Hong Keah , Natalie Allen , Ro...
0 downloads 0 Views 114KB Size
Download PDF
Anal. Chem. 2000, 72, 1964-1972

Determination of Biophysical Parameters of Polypeptide Retro-Inverso Isomers and Their Analogues by Capillary Electrophoresis M. T. W. Hearn,*,† H. H. Keah,† R. I. Boysen,† I. Messana,‡.§ F. Misiti,‡ D. V. Rossetti,‡ B. Giardina,‡,§ and M. Castagnola§,|

Center for Bioprocess Technology, Department of Biochemistry and Molecular Biology, Monash University, Wellington Road, Clayton, Australia 3168, Istituto di Chimica e Chimica Clinica, Facolta` di Medicina e Chirurgia, Universita` Cattolica e Centro di Studio per la Chimica dei Recettori e delle Molecole Biologicamente Attive, Roma, Italy, and Dipartimento di Biochimica e Fisiologia Umana, Universita di Cagliari, Cagliari, Italy

The relationship between the electrophoretic mobility, µobs, Stokes radius, rs, ionization state, and solution conformation of the all L-r-polypeptide, 1, the corresponding retro-all D-r-polypeptide, 2, and several truncated analogues, 3-5, has been investigated under low pH buffer conditions by high-performance capillary zonal electrophoresis (HPCZE) with coated capillaries. The results confirm that, under these conditions, the all L-rpolypeptide, 1, and its retro-inverso isomer, 2, exhibit nonidentical electrophoretic mobilities and thus different Stokes radii. At higher pH values, i.e., pH 5.0, the electrophoretic behavior of this retro-inverso isomer pair, however, converges. These results indicate that variations in the dipole characteristics of the polypeptide main chain and subtle differences introduced by the spatial constraints of the L-r-Pro f D-r-Pro residue replacement lead to differences in the Stokes radii and electrophoretic mobilities of these polypeptides. Since the observed electrophoretic mobilities, µobs, reflect the mean of the mobilities of each charge species participating according to their Stokes radius or their intrinsic charge and mole fraction abundances, the results confirm that polypeptide retro-inverso isomers with unmodified amino and carboxy termini are resolvable. This outcome was achieved despite their notional topographical and conformational similarities as assessed from high-field proton nuclear magnetic resonance (1H NMR) spectroscopy and circular dichroism (CD) spectroscopy.

used to identify a viable peptidomimetic lead compound. Various areas of research have benefited from this approach, including the development of structural mimics related to peptide hormones, peptide neurotransmitters, antimicrobial peptides and peptidesubunit vaccines.1-3 A theme, common to all of these developments, has been the motivation to enhance the affinity of the ligand for its appropriate biological acceptor and to improve the biological stability against proteolytic degradation. Of particular currency has been the incorporation of D-R-amino acids into a target sequence,4-8 thereby permitting the chirality of selected R-amino acid residues to be used as a useful parameter to dictate activity and to limit proteolytic degradation. One important attribute of exploiting the stereochemical possibilities of chiral centers in a target peptide sequence is that the options are not limited just to single L-R f D-R replacements. Rather, stereochemical opportunities can be used to evaluate the biophysical, topological, and functional consequences not only of polypeptide diastereomers (i.e., isomers containing the same distribution of chiral centers but with one L-R-amino acid residue substituted by the corresponding D-R-amino acid residue resulting in polypeptide isomers that are nonsuperimposable mirror images) but also polypeptides that are related in their structural properties as retro pairs, (i.e., sequences that contain only all L-R- or, alternatively, only all D-R- amino acid residues but with the reading frame of the two sequences reversed) or as inverso pairs (i.e., chirality of the individual amino acid residues inverted but with the reading frame of the sequences in the same direction). From a stereochemical as well as a design perspective, probably the most fascinating group of structural isomers involves

In recent years, incorporation of non-L-R-amino acids into the sequences of bioactive polypeptides has become a relatively routine strategy in structure-function studies. Subsequent modification of key amino acid side chain functionalities can then be

(1) Wiley, R. A.; Rich, D. H. Med. Res. Rev. 1993, 13, 327-384. (2) Fauchere, J. L.; Thurieau, C. Adv. Drug Res.1992, 22, 127-159. (3) Giannis, A.; Kolter, T. Angew. Chem., Int. Ed. Engl. 1993, 32, 1244-1267. (4) Wade, D.; Boman, A.; Wahlin, B.; Drain, C.; Andreu, D.; Boman, H. G.; Merrifield, R. B. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 4761-4765. (5) Thompson, P. E.; Ede, N. J.; Lim, N.; Ng, F. M.; Rae, I. D.; Hearn, M. T. W. Drug Des. Discovery 1995, 13, 55-72. (6) Fisher, P. J.; Prendergast, F. G.; Ehrhardt, M. R.; Urbauer, J. L.; Wand, A. J.; Sedarous, S. S.; McCormick, D. J.; Buckley, P. J. Nature (London) 1994, 368, 651-653. (7) Guichard, G.; Benkirane, N.; Zeder-Lutz, G.; van Regenmortel, M.; Briand, J.-P.; Muller, S. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 9765-9769. (8) Bobke, V.; Sasidhar, Y. U.; Durani, S. Int. J. Pept. Protein Res. 1994, 43, 209-218.

* Corresponding author: (Fax)Int + 61 + 3 + 9905 5882, (e-mail) [email protected]. † Monash University. ‡ Universita ` Cattolica. § Centro di Studio per la Chimica dei Recettori e delle Molecole Biologicamente. | Universita di Cagliari.

1964 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

10.1021/ac990369a CCC: $19.00

© 2000 American Chemical Society Published on Web 03/28/2000

Table 1. Amino Acid Sequences, Stokes Radii (Å), and pK Values of the Different Polypeptides polypeptide

sequence

Stokes radius (Å) rs (at charge)

Stokes radius (Å) rs (at charge)

pK (C-terminus)

pK (side chain)

1 2 3 4 5

NH2-DDALYDDKNWDRAPQRCYYQ-COOH NH2-QYYCRQPARDWNKDDYLADD-COOH NH2-DDALYDDKNWDRAPQRCYQ-COOH NH2-DDALYDDKNWDRAPQ-COOH NH2-DDKNWDRAPQRCYYQ-COOH

23.0 ( 0.5 (+4) 20.0 ( 0.5 (+4) 22.0 ( 0.5 (+4) 19.0 ( 0.4 (+3) 19.5 ( 0.4 (+4)

30.0 ( 0.8 (-2) 24.0 ( 0.7 (-2) 25.0 ( 0.7 (-2) 19.0 ( 0.6 (-3) 18.0 ( 1.0 (0)

3.4 ( 0.2 3.4 ( 0.2 3.1 ( 0.2 3.2 ( 0.2 3.2 ( 0.2

3.8 ( 0.2 4.3 ( 0.3 3.9 ( 0.3 3.9 ( 0.2 4.1( 0.4

polypeptide pairs that exhibit a retro-inverso isomer relationship (i.e., one sequence contains only all L-R-amino acid residues while the second sequence contains only all D-R-amino acid residues but with the reading frame of the sequences reversed). With linear polypeptides, a retro-inverso isomer can bear a close but not identical structural complementarity and topological similarity to the parent polypeptide if amino and carboxyl end group modifications are not included. The origins of these topological approaches to the design of bioactive polypeptides date back to the pioneering studies of Shemyakin et al.9-11 with enantio-anniatin A and B or retro-enantio-gramicidin S. Subsequently, other investigators, including Goodman et al.12-15 and Volger et al.,16 have studied retro-enantio and retro-inverso isomers of different antibiotic and hormonal peptides. The further development of retro-enantio and retro-inverso peptides has also attracted attention in various fields of polypeptide-based therapeutics, including retro-inverso analogues of the major antigenic site of foot and mouth disease virus (FMDV) VP1 protein17 and the all D-R isomers of human immunodeficiency virus type 1 (HIV-1) aspartylprotease.18 Despite the current interest in the biological properties of retro-inverso or retro-enantio isomer pairs, particularly as part of peptide-subunit vaccines,17,19,20 very little information is available on their biophysical properties or solution conformations. An important question previously unresolved is whether retroinverso isomer pairs, synthesized without regard to amino and carboxyl end group modifications, can be separated by capillary (zone) electrophoretic (CZE) methods and other high-resolution separation techniques. These polypeptide isomers contain the same amino acid residues distributed in an identical relative sequential context and can undergo similar side chain ionization processes in solution. Moreover, they have the same molecular weight, and the peptide backbone is theoretically anticipated from stereochemical arguments to adopt the same folded state and (9) Schemyakin, M. M.; Ovchinnikov, Yu. A.; Ivanov, V. T.; Evstranov, A. V. Nature (London) 1967, 213, 412-414. (10) Schemyakin, M. M.; Ovchinnikov, Yu. A.; Ivanov, V. T.; Ryabova, S. Experientia 1967, 23, 326-334. (11) Schemyakin, M. M.; Ovchinnikov, Yu. A.; Ivanov, V. T. Angew. Chem., Int. Ed. Engl. 1969, 8, 492-498. (12) Goodman, M.; Chorev, M. Acc. Chem. Res. 1979, 12, 1-7. (13) Chorev, M.; Goodman, M. Acc. Chem. Res. 1993, 26, 266-273. (14) Chorev, M.; Goodman, M. Trends Biotechnol. 1995, 13, 438-445. (15) Chorev, M.; Goodman, M. Trends Biotechnol. 1996, 14, 42-44. (16) Vogler, T.; Kurth, R.; Norley, S. J. Immunol. 1994, 153, 1895-1904. (17) Muller, S.; Guichard, G.; Benkirane, N.; Brown, F.; van Regenmortel, M. H. V.; Briand, J.-P. Pept. Res. 1995, 8, 138-144. (18) Milton, R. C. deL.; Milton, S. C. F.; Kent, S. B. H. Science (Washington, D.C.) 1992, 256, 1445-1448. (19) Guichard, G.; Benkirane, N.; Zeder-Lutz, G.; van Regenmortel, M. H. V.; Briand, J.-P.; Muller, S. Pept. Res. 1995, 8, 138-144. (20) Keah, H. H.; Kecorius, E.; Hearn, M. T. W. Pept. Res. 1998, 51, 2-11.

topology in solution. In addition to elucidating key physicochemical questions on their solution behavior, the availability of rapid, high-sensitivity procedures to characterize the proton dissociation constants of such polypeptide retro-inverso isomers, as well as their charge states and Stokes radius characteristics under different solution conditions, would greatly facilitate structurefunction analysis and design improvements with these molecules. The present investigations address this issue, through analysis by a sensitive CZE approach of the pH dependencies of the electrophoretic mobilities and polypeptide dissociation constants of two 20-mer polypeptides 1 and 2 (one letter amino acid sequences shown below) and several analogues (Table 1). It can be noted that in polypeptide 2 the positions of the carbonyl and amido groups associated with each of the amide bonds have been reversed relative to the situation applying to polypeptide 1. In addition, the orientation of the side chain groups at each R-carbon has been inverted from the L-R to the D-R stereochemistry. The resulting polypeptide 2 should, in principle, exhibit a topological resemblance to that produced by the all L-Ramino acid polypeptide 1 in terms of the conformational array of the amino acid side chains along the polypeptide backbone. However, it must be noted that an intrinsic structural nonequivalence exists between the parent all L-R- polypeptide, 1, and the retro all D-R-polypeptide, 2. In particular, the N-terminal L-R-aspartic acid (Asp1) in 1 carries a free R-amino group and the C-terminal L-R-glutamine (Gln20) a free carboxyl group, while in 2 the N-terminal D-R-glutamine (Gln1) and the C-terminal D-R-aspartic acid (Asp20) carry a free amino and a free carboxyl group, respectively. To determine whether these subtle differences in otherwise topologically identical polypeptides can be detected under different solvational and ionization conditions, the corresponding dependencies of the electrophoretic mobilities of the polypeptides 1 and 2 and several analogues, 3-5, on pH were examined. In addition, conformational features of these polypeptides were assessed from high-field proton NMR and circular dichroism (CD) spectroscopic data. Previous studies have shown21-23 that dissociation constant measurements by CZE procedures correlate well with microtitration measurements, with the advantage that the contribution from Stokes radius effects and other conformational parameters24 to the electrophoretic migration can also be determined. The strategy (21) Castagnola, M.; Rossetti, D. V.; Cassiano, L.; Misiti, F.; Pennacchietti, L.; Giardina, B.; Messana, I. Electrophoresis 1996, 17, 1925-1930. (22) Bekkers, J. L.; Everaerts, P. M.; Ackermann, M. T. J Chromatogr. 1991, 537, 407-428. (23) Gao, J.; Mrksich, M.; Gomez, F. A.; Whitesites, G. M. Anal. Chem. 1995, 67, 3093-3100. (24) Sitarim, B.; Keah, H. H.; Hearn, M. T. W. J. Chromatogr. A 1999.

Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

1965

followed in the present investigation was thus intended to resolve hitherto unexplored questions about polypeptide retro-inverso isomer pairs that related to the extent of perturbation of the backbone and side chain conformations that is mediated by (i) inversion of the chirality at the R-carbon atom of each amino acid residue, (ii) the reversal of the sequence, and (iii) the contribution from the end-group effects. In view of the predicted propensity for D-Pro residues to induce type II′ β-turns,25,26 these studies also permitted the conformational effects of the L-R-Pro f D-R-Pro replacement to be further assessed. MATERIAL AND METHODS Chemicals and Reagents. All common chemical substances were purchased from Sigma Chemical Co. (St. Louis, MO) or E. Merck (Darmstadt, Germany). N,N-dimethylformamide (DMF), trifluoroacetic acid, piperidine, HOBt, HBTU, p-alkoxybenzyl alcohol resin, and all the L-R- and D-R-Fmoc-protected amino acids were obtained from Auspep (Melbourne, Australia). Unless otherwise stated, all the reagents and solvents were of analytical grade. Phenol was obtained from Merck Aust. Ltd. (Kilsyth, Vic, Australia). Diisopropylethylamine, ethanedithiol, acetic anhydride, thioanisole, and trifluoromethanesulfonic acid were obtained from Aldrich Chemical Co. (Milwaukee, WI). Diisopropylcarbodiimide was obtained from Sigma Chemical Co. (St. Louis, MO). The Fmoc-Glu(Rink amide mBHA)-OtBu resin was obtained from Novabiochem (Sydney, Australia). Synthesis, Purification and Characterization of Polypeptides 1-5. (A) Synthesis. Preparation of polypeptides 1-5 was achieved using standard Fmoc-based solid-phase peptide synthetic methods either manually or with a PS3 Automated Peptide Synthesizer (Rainin Instrument Co., Woburn, MA). In brief, polypeptide 1 was synthesized using the Fmoc-Glu (Rink amide mBHA)-OtBu resin (0.34 mmol/g, 600 mg) with the standard Fmoc synthesis protocol and HBTU/HOBt activation procedure.27 The crude polypeptide was cleaved from the resin with a onestep TFA protocol using phenol (0.75 g), ethanedithiol (0.25 mL), thioanisole (0.50 mL), deionized water (0.50 mL), and TFA (10 mL). The cleavage mixture was stirred for 2 h before filtering, with the filtrate concentrated and the crude product precipitated with cold ether. The ether solution was kept at 263 K overnight before being filtered, the crude product dissolved in 50% aqueous acetonitrile and lyophilised. The analogues, 3-5, were prepared in an identical manner. Polypeptide 2 was synthesized using the p-alkoxybenzyl alcohol resin (0.90 meq/g, 620 mg). The first (Cterminal) amino acid residue was attached by adapting the method of Green and Bradley28 using DIC/HOBt activation. All subsequent amino acid residues were coupled using a standard Fmoc synthesis protocol and HBTU/HOBt activation method. Cleavage and side-chain deprotection were achieved as described above for 1, except that the reaction mixture was stirred for 3 h at 20 °C. (B) Purification. HPLC procedures were performed with a Waters Assoc. (Milford, MA) liquid chromatography system that (25) Brady, L.; Dodson, G. Nature (London) 1994, 368, 692-693. (26) Jameson, B. A.; McDonnell, J. M.; Marini, J. C.; Korngold, R. Nature (London) 1994, 368, 744-746. (27) Thompson, P. E.; Keah, H. H.; Gomme, P. T.; Stanton, P. G.; Hearn, M. T. W. Int. J. Pept. Protein Res. 1995, 46, 174-180. (28) Green, J.; Bradley, K. Tetrahedron 1993, 49, 4141-4146.

1966

Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

consisted of either (i) two Model 501 solvent delivery pumps, a U6K universal injector, a WISP Model 701B autosampler, and an automated gradient controller, or (ii) a Model 600 solvent delivery pump, a Rheodyne injector, a WISP Model 712 autosampler, and an automated gradient controller. Detection of eluted polypeptides and other synthesis products was achieved using a Waters Model 441 UV Absorbance detector, coupled to a Perkin-Elmer LCI-100 Laboratory Computing Integrator or a Waters Model 486 variable wavelength UV detector connected to Waters Millenium software control (version 2.1). Preparative purification of the crude polypeptides was performed with a TSK-ODS-120T C-18 column (300 × 21.5 mm i. d., particle size 10 µm), protected with a guard column packed with the same sorbent, using gradient elution procedures from 0.1% TFA in water (buffer A) to 0.1% TFA in water/ acetonitrile (40/60 v/v) (buffer B) over 60 min at a flow rate of 6 mL/min with detection at 254 nm. Fractions were collected using a Pharmacia (Frac-100) fraction collector and characterized by analytical RP-HPLC with a TSK C-18 column (150 × 4.6 mm, particle size 5 µm) with the samples eluted by a linear 25-min gradient of buffer A to 85% buffer B at a flow rate of 1 mL/min with detection at 214 nm. (C) Characterization. Purified polypeptides were characterized by amino acid analyses performed using the phenylisothiocyanate (PITC) derivatization procedure.29,30 Fractions containing the polypeptide products were dried and hydrolyzed using constant boiling point 6 M HCl (0.4 mL/mg) at 383 °K for 24 h under reduced pressure. The purified polypeptides were further characterized by electrospray ionization mass spectroscopy (ESI-MS) with a PerkinElmer-Sciex Mass Spectrometer model PE Sciex API /// using a scan range between 200 and 2400 amu. Between 5 and 50 µL of sample was injected via a ‘PE ISS 200” auto injector. The solvent used for the sample injection was 60% acetonitrile/0.1% acetic acid (HPLC grade). The 2D NMR spectra were acquired31 in 40% trifluoroethanol (TFE)/H2O at pH 2.3, a concentration of 2 mM, and a temperature of 287 °K. The 1H NMR experiments were run on a Bruker AMX spectrometer operating at 500 MHz with spectral widths of 6000 Hz, respectively. The 2D 1H NMR experiments undertaken on each of the samples included DQF-COSY;32 NOESY33 with mixing times of 200, 300, and 450 ms for the TFE/H2O samples; and TOCSY34,35 using the MLEV spin-lock sequence with mixing times of 40 and 120 ms. The 2D 1H NMR data were typically acquired using 2K data points in F2 (with 4K for DQF-COSY spectra), 400512 increments in F1 (with 800 for DQF-COSY spectra). In each case, 32 transients (96 for NOESY) were collected, and relaxation delays of 1.5-2.0 s were used. Water suppression was achieved using continuous low power presaturation during the relaxation delay and during the mixing time in NOESY spectra for the (29) Cohen, S. A.; Strydom, D. S. Anal. Biochem. 1988, 174, 1-16. (30) Stanton, P. G.; Robertson, D. M.; Burgon, P. G.; Schau-White, B.; Hearn, M. T. W. Endocrinology 1992, 130, 2820-2832. (31) Higgins, K. A.; Bicknell, W.; Keah, H. H.; Hearn, M. T. W. Pept. Res. 1997, 50, 421-435. (32) Rance, M.; Sorensen, O. W.; Bodenhausen, G.; Wagner, G.; Ernst, R. R.; Wuthrich, K. Biochem. Biophys. Res Commun. 1983, 117, 479-485. (33) Kumar, A.; Ernst, R. R.; Wuthrich, K. Biochem. Biophys Res. Commun. 1980, 95, 1-6. (34) Braunsschweiler, L.; Ernst, R. R. J. Magn. Reson. 1983, 53, 521-528. (35) Davis, D. G.; Bax, A. J. Magn. Reson. 1985, 64, 533-535.

aqueous solution and TFE/H2O samples. All spectral data were processed using the UXNMR (Bruker) on an SGI Indigo2 workstation. A polynomial baseline correction algorithm was applied to each FID prior to transformation. Window functions were iteratively optimized for each 2D spectrum but in general the data were multiplied by a 60° shifted sine bell function in each dimension. Zero filling in the F1 dimension resulted in a data matrix of 1K × 1K (2K × 2K for DQF-COSY spectra) for each 2D spectrum. Chemical shifts were referenced to internal DSS added to the samples after the 2D spectra were recorded. Amide temperature coefficients were measured over temperature ranges of 287-305 K.36,37 Measurement of spin-spin coupling constants 3JHNR was made using a cross-section parallel to F2 from the DQF-COSY spectra. The F2 cross section was zero filled to 8K or 16K prior to measurement. A second method used was based on the procedure described by Ludvigsen et al.38 and involved the use of cross sections from both the DQF-COSY and NOESY spectra. Both methods were used to determine coupling constants for the polypeptides in the TFE/H2O systems, with the coupling-constant values from the two methods deviating by less than 0.5 Hz. Circular dichroism (CD) spectra were acquired on a J-700 spectropolarimeter (Jasco, Japan) at 288 K. Data sampling was carried out every 0.2 nm using a bandwidth of 1.0 nm and response time of 1 s. The scan speed was 10 nm/min, and each spectrum was averaged from 4 individual scans. CD measurements were performed from 190 to 250 nm in a 200-µL cell with a path length of 0.02 cm and polypeptide concentration of 390 µM in the following buffers: (a) 40% TFE-80 mM sodium phosphate, pH 2.3; (b) 80 mM sodium phosphate, pH 2.6; and (c) 20% acetonitrile-0.09% TFA (pH 2.1) at 288 °K. All data were handled by an evaluation program J-700 (Jasco, Japan). Baseline-compensated and normalized CD spectra (obtained after subtraction of a corresponding blank) were obtained39 after performing noise reduction using fast Fourier transformation (FFT) procedures. Capillary Electrophoresis. The CE apparatus employed was a P/ACE 2100 instrument (Beckman, Palo Alto, CA) equipped with the Beckman Gold Series 711 system for automated apparatus control and data acquisition. The fused-silica capillaries, purchased from Beckman, were coated40 directly on the electrophoresis apparatus and stabilized as described previously.41 Under these stabilized coating conditions, no appreciable electroendosmosis occurred at low pH values, as evident from the migration behavior of the marker, acetanilide. The capillary dimensions were 56.5 cm (50 cm at the detection window) × 75 µm. A set of phosphate, formate, and acetate solutions, prepared according to the method of Gluck and colleagues42 was used for the electrophoretic buffers, (36) Jardetzky, O.; Roberts, G. K. C. NMR in Molecular Biology; Academic Press: New York, 1981; p 166. (37) Wuthrich, K. NMR of Proteins and Nucleic Acids John Wiley & Sons: New York, 1986; pp 1-276. (38) Ludvigsen, S.; Anderson, K. V.; Poulsen, F. M. J. Mol. Biol. 1991, 271, 731-736. (39) Boysen, R. I.; Wang, Y.; Keah, H. H.; Hearn, M. T. W. 1999 J. Biophys. Chem. 77, 79-97. (40) Hjerten, S. J. Chromatogr. 1985, 347, 191-198. (41) Castagnola, M.; Cassiano, L.; Messana, I.; Paci, M.; Rossetti, D. V.; Giardina, B. J. Chromatogr. 1996, 735, 271-281. (42) Gluck, S. J.; Steele, K. P.; Benko ¨, M. K. J. Chromatogr., A 1996, 745, 117125.

characterized by constant ionic strength in the range of pH 2.25.5. The synthetic polypeptides were dissolved in water (conductivity < 18 µmho) at a concentration of 2.0 mg/mL. Electrophoretic separations were carried out at a constant voltage of 25 kV at 298 K. The anode electrode was coincident with the capillary injection terminal, with injections performed by the pressure fill method for 1 or 2 s, corresponding to an injected volume of 5 or 10 nL. Polypeptides were detected at 200 nm. All experiments were replicated at least twice. Electrophoretic mobilities were analyzed according to the equations described in the Results and Discussion section. Stokes radii and pK values were derived by parametric best-fit minimization procedures based on Marquard’s algorithm. Results are reported with a error function corresponding to two standard deviations. RESULTS AND DISCUSSION Theoretical Considerations. The dissociation of a generic polypeptide (Xcharge) of n amino acid residues with j basic ionogenic groups (N-terminus; histidine, lysine, arginine) and i acid ionogenic groups (C-terminus; aspartic and glutamic acid; cysteine, tyrosine), can be described by:

Xj 9 8 Xj-1 9 8...9 8 X-i K K K 1

2

(1)

j+i

In common with all protonic dissociations, this reaction pathway assumes that all k species of the same charge (k ) j + i + 1) are equivalent and takes advantage of the assumed independence of the macroscopic dissociation constants K1, K2, ... Kj+i. Moreover, this reaction pathway assumes that the protonexchange kinetics as well as conformational interconversions are fast compared to the electrophoretic migration times. Because polypeptides fall into the class of compounds that have variable charge as a function of the pH of the surrounding solvent, changes in the proton activity of the buffer can lead39,43 to variation in the Stokes radius, rs, of the polypeptide. The observed electrophoretic mobility (µobs) of a polypeptide as a function of pH can thus be considered as weighted mean mobility of the different protonated species such that

µobs ) k)-i

µjχj + µj-1χj-1 + µj-2χj-2 + ... + µ-iχ-i )

∑µ χ

k k

(2)

k)j

where µj, µj-1, µj-2, ... µ-i are the electrophoretic mobilities of the individually ionized species and χj, χj-1, χj-2, ... χ-i are the molar fractions of the different species. Because the Stokes radius, rs, of the polypeptide will not be constant as a function of the buffer pH, eq 2 can be written

µobs )

q

k)-i

k

k)j

s,k

∑r 6πη

χk

(3)

where χk is the molar fraction of any species, j; q is the electron charge (1.60 10-19 coulomb); η is the solution viscosity (water: (43) Sahota, R. S.; Khaledi, M. G. Anal.Chem. 1994, 66, 2374-2381.

Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

1967

8.95 × 10-4 Ns/m2 at 298 K); rs,k is the Stokes radius (i.e., the radius of the sphere equivalent to hydrated polypeptide) of any species j; and k is the dimensionless value of the species integer charge. The molar fraction χ of any ionized species of the polypeptide can be expressed as a function of proton activity and dissociation constants, as follows

χj-n ) [χj-n] [[χj] + [χj-1] + [χj-2] + ... + [χ-i]]

)

βn/[H+]n (4) P

where n ) 0, 1, 2, ..., j + 1, and P is the binding polynomial expression that links the respective dissociation constants to the hydrogen-ion concentration and is given by

P ) β0 + β1/[H+] + β2/[H+]2 + ... + βj+1/[H+]j+1 (5) and βn is given by

β0 ) 1; β1 ) K1; β2 ) K1K2; ...; βj+1 ) K1K2K3 ... Kj+1 (6) For the polypeptides 1-5 investigated in this study, several assumptions were made: (1) The electroosmotic flow was negligible, i.e., µeo ≈ 0. This was confirmed with the coated capillaries using the appropriate markers at all investigated pH values. (2) Over the pH range studied (pH 2.0-5.5), the relevant dissociation constants are those associated with the C-terminus and aspartic and glutamic acid side chains. Thereby, the dissociation constants of other ionogenic groups have been neglected. (3) When more than one aspartic (and/or glutamic acid) side chain was present in the polypeptide sequence, then their intrinsic constants were assumed to be equal, but the macroscopic constants accounted for the statistic. Thus, in the case of the polypeptide 1 which contains 5 aspartic acid residues, from the data fit, pKi,(Asp)(intrinsic) ) 3.9 and hence Ki )1.26 × 10-4. The progressive macroscopic constants Kj were related to the intrinsic constant by the expression Kj ) [(n - j + 1)/j] Ki. Thus, the five constants of 1 were derived from the data fit to have the values K1 ) 6.29 × 10-4; K2 ) 2.52 × 10-4; K3 ) 1.26 × 10-4; K4 ) 6.29 × 10-5; K5 ) 2.52 × 10-5. (4) Only two Stokes radii were set free during the data fit: the Stokes radius of the fully protonated form (i.e., +4 in 1) and the one corresponding to the polypeptide deriving from the dissociation of all the acidic groups (C-terminus, D, E) (i.e., -2 in 1). The Stokes radii of the intermediate polypeptide forms were assumed to be progressively changing from the +j to the -i form. As a consequence, the intermediate Stokes radii did not have any real physical significance, but they were used as a mathematical aid to perform a fit on mobility data. (5) With the above premises, the algorithmic fitting procedures contain four free parameters: the Stokes radius of the initial (fully protonated) form, the Stokes radius of the fully dissociated form, the dissociation constant of the C-terminus, and the intrinsic dissociation constant of the side chains. It can be noted that polypeptide 5 has the fully dissociated form with a total charge 1968 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

Figure 1. CD spectra of the polypeptides, 1 and 2, measured under two different R-helix inducing solvent conditions: (a) 20% acetonitrilewater-0.09% TFA (pH 2.1) at 288 K and (b) 40% trifluoroethanolwater-80 mM sodium phosphate (pH 2.3) at 288 K. Other conditions are described in the Experimental Procedures.

equal to zero. Consequently, the mobility of this form is zero, and the determination of its Stokes radius by CZE was not possible. However, analysis of the curve shape of the plot of µobs versus pH for this polypeptide suggests a slight Stokes radius decrease of the intermediate positive forms. Synthesis and Spectroscopic Properties of the Polypeptide Retro-Inverso Isomers. The polypeptides 1-5 were prepared by standard solid-phase methods based on Fmoc chemistry and, following purification by preparative reversed-phase HPLC, were obtained in ∼35% yield. Amino acid analysis and electrospray ionization mass spectrometry (ESI-MS) confirmed the compositions and molecular masses of the synthetic polypeptides (data not shown). Circular dichroism (CD) studies with these polypeptides in different percentages of the TFE solvent up to 70% TFE (v/v), indicated that the R-helical content reached a maximal value at ∼40% TFE-80 mM sodium phosphate (pH 2.3). In addition, the CD spectra of the polypeptides at different concentrations in 40% aqueous TFE confirmed an absence of aggregation39,44 up to 2-3 mM. It is interesting to observe from Figure 1 that the far ultraviolet CD spectra for polypeptides 1 and 2 in 40% TFE-water-80 mM sodium phosphate (pH 2.3) and the comparable R-helix inducing solvent 20% acetonitrile-water0.09% trifluoroacetic acid (pH 2.1) are essentially symmetrical mirror images, exhibiting the characteristic Cotton effect relationship expected for an all L-R and retro-all D-R-peptide isomer pair. These observations are consistent with other findings6,31,45-47 on the influence of TFE on induction of R-helical secondary structures of polypeptides involving L-R f D-R substitutions. Proton resonance assignments for the synthetic polypeptides at concentrations of 2 mM were achieved using well-established 2D NMR techniques.37 DQF-COSY and TOCSY spectra were used31 to identify the amino acid spin systems for each polypeptide while residue connectivities were established using 2D NOESY (44) Boysen, R. I.; Hearn, M. T. W. unpublished results, 1999.

spectroscopy recorded at 288 K, allowing observation of sequential and medium-range NOEs. Representative of these results are the CR-NH regions and the sequential connectivities of the NOESY spectra shown in Figure 2 parts (a) and (b) for polypeptides 1 and 2 obtained at 500 MHz in the 40% d3-TFE/H2O solvent system. As a result of the large number of AMX spin systems and, in particular, the presence of 5 aspartic acid residues in polypeptides 1 and 2, the corresponding Cβ-NH sequential assignments were also employed to confirm some CR-NH assignments. Analysis of the proton chemical shift data, the identified CR-NH and NHNH cross-peaks, and the medium range NOEs for polypeptide 1 revealed that the spectral data for the medium range i,i+3 and i,i+4 cross-peaks corresponded to the sequence region from Asp6 to Ala.13 In the same region, a series of strong NH-NH sequential cross-peaks was also identified. The observed long-range NOEs indicate that the Asp11 and Tyr18 residues of polypeptide 1 come into relatively close contact (e5 Å). Similarly, a series of strong NH-NH cross-peaks corresponding to the central sequence region of polypeptide 2 were apparent from Ala8 to Tyr16 as well as a group of medium range i,i+3 and i,i+4 cross-peaks corresponding to this polypeptide region. Detailed examination31 of the amide exchange coefficients of polypeptides has confirmed that the NH protons of Lys8, Trp10, Asp11, Arg,12 and Ala13 of polypeptide 1 are in slow exchange with the solvent while polypeptide 2 shows a series of amide exchange coefficients consistent with slow amide proton exchange of the residues Asp10, Trp11, Lys,13 and Asp14 with the solvent. These data correspond well with the observed spin-spin coupling constants 3JHNR, for these polypeptides, which were larger at the C- and N-termini, and smaller in the central region of the polypeptides. In the case of polypeptides 1 and 3, the magnitude of these amide exchange coefficients (-3.0 to -5.0 ppb/K) and 3J 48,49 for the HNR values (4-6 Hz) fell within the ranges expected presence of a considerable population of R-helical structures extending over the Asp6 to Ala13 sequence region. The small values of the amide proton temperature coefficients for the residues Lys8, Trp10, Asp11, Arg,12 and Ala13 were consistent with H-bond formation expected for this type of secondary structure. Larger amide exchange coefficients (i.e., -6.0 to -10.0 ppb/K) and 3JHNR values (10-13 Hz) were observed for both the C and N terminal regions, consistent with these regions being more accessible to the solvent. Similarly, for polypeptide 2, shortened amide exchange rates (-4.0 to -6.0 ppb/K) were observed for residues in the central sequence region with the corresponding observation of 3JHNR coupling constants in the 6-8 Hz range extending through the region from Ala8 to Tyr.16 These data, taken together with the series of strong NH-NH cross-peaks and medium range i,i+3 and i,i+4 NOE cross-peaks, indicate that the conformation of this region of the polypeptide 2 assumes a significant R-helical population under these conditions. The longer exchange rates, larger coupling constants, stronger CR-NH cross-peaks for the (45) Benkirane, N.; Guichard, G.; Van Regenmortel, M. H. V.; Briad, J. P.; Muller, S. J. Biol. Chem. 1995, 270, 11921-11926. (46) Yamazaki, T.; Mierke, D. F.; Said-Nejad, O. E.; Felder, E. R.; Goodman, M. Int. J. Pept. Protein Res. 1992, 39, 161-181. (47) Guichard, G., Muller, S.; van Regenmortel, M.; Briand, J. P.; Mascagni, P.; Giralt, E. Trends Biotechnol. 1996, 14, 44-45. (48) Hinds, M. G.; Welsh, J. H.; Brennand, D. M.; Turner, D. L.; Robinson, J. A. J. Med. Chem. 1991, 34, 1777-1789. (49) Pardi, A.; Billeter, M.; Wuthrich, K. J. Mol. Biol. 1984, 180, 741-751.

Figure 2. The 500 MHz CR-NH region of the NOESY (200 ms) spectrum of polypeptide 1 (a) and 2 (b) in the 40% d3-trifluoroethanolwater, pH2.3, solvent system, showing the sequential connectivities.

N- and C-terminal regions of polypeptide 2, with no evidence of medium-range NOEs, are consistent with these regions assuming extended or random structures in solution. On the basis of the spectroscopic data, it can been concluded that the retro-inverso pair, 1 and 2 in 80 mM sodium phosphate, pH 2.3, and 40% d3-TFE/H2O, 80 mM sodium phosphate, pH 2.3, solvent systems exhibit similar conformational features. The predominant conformation observed under the 40% d3-TFE/H2O, 80 mM sodium phosphate, pH 2.3, conditions incorporates a central R-helical region that extends over the same series of amino acid residues in each polypeptide; namely the sequences YDDKNWDRA in 1 and ARDWNKDDY in 2, while in the fully aqueous phosphate buffer, a more extended conformation prevails. No evidence of cis/trans isomerism was observed at temperatures as low as 278 K with the NMR spectral behavior consistent with a trans configuration around the Pro residue. The results provide evidence of conformational averaging, due to the limited number of observed medium-range NOEs. Evidence from the amide exchange rates, coupling constants, and observed NOEs indicates that for each of the polypeptides the C- and N-terminii appear to have extended/random structures in solution. Finally, longer range NOEs were observed in 40% d3-TFE/H2O, 80 mM sodium phosphate, pH 2.3, for 1 between Asp11 and Tyr,18 while in the case of 2, similar longer range NOEs were found between the Cys4 and Arg,9 and Arg5 to Asp10 residues, suggesting that with both polypeptides a partial type II′ β-turn conformation may exist centered around the proline residue. Capillary Electrophoresis Observations. Figure 3 represents the plots of the experimentally observed electrophoretic mobilities, µobs, over the pH range 2.0 e pH e 6.0 for the polypeptides 1-5 with the theoretical mobilities, µtheory, computed by utilizing parametric fitting procedures based on Marquardt’s algorithm. It is apparent from the results shown in these Figures that polypeptides 1-5 have distinctly different electromobilities at low pH values, while the µtheory and µobs values of polypeptides 1 and 2 converge to a common value near pH 5.0 and invert at values of pH > 5.2. These studies thus show that polypeptides 1 and 2 can be resolved by CZE using simple phosphate buffer, pH 2.3, conditions but that these two polypeptides coeluted when the buffer pH was near to pH 5.0. Below pH 5.0, polypeptide 2 showed higher electromobility in agreement with the calculated smaller Stokes radius and slightly smaller molecular dimensions. Interestingly, the C-terminus of polypeptide 2 exhibits the same acidity Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

1969

Figure 3. Plots of the experimentally observed electromobilities, µobs, for polypeptides 1-2 (Panel (a)) and 3-5 (Panel (b)) compared with the theoretical mobilities, µtheory, over the pH range 2.0 e pH e 6.0. The theoretical curves were obtained by fitting the experimental data to eqs 1-6 and computed by utilizing parametric fitting procedures based on Marquardt’s algorithm. Other experimental details are given in the Materials and Methods section.

as polypeptide 1 with a calculated pK value of 3.4 (Table 1). The same acidity of the C-terminal residues (L-R-Gln for 1 and D-RAsp for 2) of these polypeptides would result in an equivalent abundance of the zwitterionic species. The differences in mobility of these polypeptides at pH values near to their pK values is indicative of differences in both their Stokes radii and the relative abundances of the molar fractions of each charged species. Clearly, with polypeptide 2, the smaller Stokes radius suggests that the polypeptide is capable of forming a more highly stabilized intramolecular ion-pair species between the C-terminal carboxylate residue and an acceptor group provided by an accessible amino acid side chain, i.e., possibly Arg9 or K.13 With polypeptide 1, on the other hand, greater solvent interaction with the ionized C-terminal carboxylate appears to have occurred. As discussed above, such behavior was reflected in the solution conformations of these polypeptides as assessed by 2D NMR techniques. The end-group effect with retro-inverso polypeptide pairs has been suggested11-14,45-47 by several investigators as a potential source of the variations detected in the biological activities of all L-R-amino acid polypeptides and their retro-inverso isomers. For example, interactions between the side-chain and main-chain atoms may be affected due to swapping of the carbonyl and amide groups in the polypeptide chain, and this structural change potentially may disrupt some H bonds as well as ion-pair bridges, giving rise to differences in solvent accessibility of the polypeptide surface. Similarly, the small differences in the µtheory and µobs values for polypeptides 2 and 3 can be equated to the expected electrophoretic behavior arising from pK value variations due to the sequence deletion of Tyr.19 However, compared with polypeptides 1 and 2, the pK values of the carboxy-terminal residues of polypeptides 3-5 were in all cases lower. Despite the differences in molecular size, over the entire pH range the electrophoretic mobilities of polypeptides 1 and 3 were essentially identical, while at low pH values, i.e., pH < 3.0, the electrophoretic mobility of 1970 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

Figure 4. The pH dependency of the molar fraction, χ, of the different charged forms of polypeptides 1 (a) and 2 (b) obtained from the application of dissociation constants listed in Table 1 and computed according to eqs 4-6, respectively.

polypeptides 2 and 5 were also similar. The experimental data with the truncated analogues further indicates that comparative biophysical data on the polypeptide dissociation constants can be determined by these HPCZE methods, with information related to variations in the Stokes radius as a function of proton equilibrium readily determined with picomolar quantities of these synthetic polypeptides. Since the observed electrophoretic mobility is the mean of the mobilities of each charged species multiplied by the corresponding molar fraction, χ, these difference in µobs imply that differences in the values of χ also exist between the different polypeptides. Representative of this effect, the pH dependency of the molar fraction, χ, of the different charged forms of polypeptides 1 and 2 obtained from the application of dissociation constants listed in Table 1 are illustrated in Figure 4 parts (a) and (b). Accordingly, in Figure 5 parts (a) and (b) are shown the intrinsic electrophoretic mobility contributions to the total mobility, µtotal, of each charged form of the polypeptide 1 and 2, respectively. Although the general shapes of these plots are similar, it is apparent from the results shown in Figure 5(a) and (b) that the species carrying a +4 charge makes a larger contribution to the µtotal of polypeptide 1 than occurs with polypeptide 2. Clearly, a simultaneous reduction in Stokes radius as well as a decrease in the extent of ionization can lead to the situation manifested by polypeptide 3, which exhibits the same electrophoretic mobilities as polypeptide 1 at pH values < 4.0 or polypeptides 2 and 5 at pH values < 3.5. This paradoxical situation, whereby two structurally dissimilar polypeptides can have identical electrophoretic mobilities as a

Figure 5. Intrinsic electrophoretic mobility contributions to the total mobility, µtotal, of each charged form of the polypeptides 1 (a) and 2 (b), respectively, computed according to the derived values of the molar fraction, χ, of the different charged forms and eqs 4-6, respectively.

function of pH, was theoretically anticipated previously21 for polypeptides. In this context, it is interesting to note that all the L-R-polypeptides 1, and 3-5 have a C-terminal Gln residue, but have different C-terminal pK values. Moreover, the analogue 4 represents the N-terminal fragment of 1 while 5 corresponds to the C-terminal fragment. The observation that polypeptides 1 and 4 diverge significantly in their electrophoretic mobilities over the range pH 2.0-5.5 provides further evidence that the Stokes radius term as well as the relative ionization status of the C-terminus and specific side chains contribute to this behavior. The 2D NMR and CD measurements indicated that these polypeptides can adopt preferred secondary structures under helixinducing solvent conditions. For example, in 40% trifluoro-ethanol (TFE)/H2O at pH 2.3, both polypeptides 1 and 2 exhibit a significant R-helical content as determined from the NOEs and amide exchange data, a finding confirmed also by the CD measurements. Accordingly, attempts were made to measure the electrophoretic mobilities of these polypeptides in the presence of similar helix-inducing organic solvent-water combinations and comparable buffer systems at different pH values to ascertain the extent of changes in the ionization state of the amino, carboxyl, and side-chain groups or the magnitude of the variation in the respective Stokes radii compared with those measured under fully aqueous conditions. However, under these aquo-organic solvent conditions (e.g., with buffers containing 40% TFE (v/v)), very long electromigration times were observed to be associated with temperature instability, precluding reliable measurement of the corresponding dissociation constants or Stokes radii of these

polypeptide retro-inverso isomers by these HPCZE techniques. Ongoing investigations are exploring the origin of this phenomenon with polypeptides in helix-inducing buffer systems, and these results will be reported subsequently. On the basis of the above data, these retro-inverso isomers 1 and 2 and their analogues, 3-5, can be classified as Class 111 polypeptides in terms of their electrophoretic behavior where the slopes of the µobs versus pH curves are a function of at least two Stokes radii terms corresponding to different protonation states. Since the 2D NMR and CD data indicate that the polypeptides 1 and 2 have similar solution conformations, with no cis-trans isomerism evident at 278 °K for either 1 or 2, it can be concluded that the differences in the electrophoretic migration behavior are a direct consequence of the subtle differences in the end-group and side-chain ionizations. At pH 2.3, the N-terminal amino group of the L-R-Asp and the D-R-Gln residues in 1 and 2, respectively, will be largely protonated. The ionization dependencies of the C-terminal L-R-Gln in 1 and 3-5 and D-R-Asp in 2, respectively, will thus represent one of the dominant contributions. Surprisingly, the results indicated that the C-terminal residues of 1 and 2 (LR-Gln and D-R-Asp, respectively) have the same calculated pK value of 3.4. However, the slopes of the µobs versus pH plots of this retro-inverso pair clearly indicate that subtle solvational and polarization effects occurred with these polypeptides, leading to differences in their electrophoretic migrations at low pH (i.e., ∼pH 2.3) conditions. The ionization characteristics associated with the Asp and Glu side chains will provide important contributions to the electrophoretic migration behavior in this regard. Since the 2D NMR and CD spectroscopic measurements indicated that both polypeptides 1 and 2 can adopt a partial type II′ β-turn conformation flanking the R-helical motif which encompasses the central portions of both molecules, the intriguing possibility arises that the CZE procedures have permitted recognition of the β-turn enhancement effect predicted25,26 for D-R-Pro residues compared with the L-R-Pro residues. As such, the CZE technique appears to be more sensitive than the 2D NMR and CD methods for detecting these fine variations in the topological nonequivalence of polypeptide retro-inverso isomers, possibly due to the difference in time averaging that occurs in the data acquisition and the fact that the CZE procedure provides a dynamic rather than a static observation. Recent studies50,51 with polypeptides have also indicated that similar considerations may apply to the orthogonal counterpart of CZE, namely capillary electrochromatography (CEC). Clearly, further work is required to ascertain the significance of these conclusions. However, it can be noted that molecular dynamics studies with the retro-inverso analogue of the B-domain of protein A also show46 a lack of topological equivalence to the parent all L-R-polypeptide. Collectively, the findings from the present and this earlier study suggest that the energy balance between a righthanded helical retro-inverso polypeptide analogue and its lefthanded all L-R-isomer, determining whether identical topological forms are achieved on folding, will be very sensitive to the amino acid sequence and the surrounding solvent environment. Whatever the molecular origin of these subtle differences is, the experimental results nevertheless demonstrate that the highest resolution of these retro-inverso isomers can be achieved at pH (50) Walhagen, K.; Unger, K. K.; Olsson, A. M.; Hearn, M. T. W. J. Chromatogr., A 1999. (51) Walhagen, K.; Unger, K. K.; Hearn, M. T. W. J. Chromatogr., A 2000.

Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

1971

values at or above their respective pK values. At these higher pH values, proton loss from the C-terminal carboxyl group will lead to an increase in the Stokes radius provided that intramolecular ion-pair formation does not concomitantly occur. When an increase in Stokes radius does occur under these conditions, then interaction with the solvent becomes more feasible, leading to a slower electrophoretic migration at lower pH values (i.e., at pH 2.3), as apparently has occurred in the case of polypeptide 1 compared with that of 2.

1972

Analytical Chemistry, Vol. 72, No. 9, May 1, 2000

ACKNOWLEDGMENT These investigations were supported by the Australian Research Council and the Italian Consiglio Nazionale delle Ricerche. Assistance from Dr K. Higgins in the acquisition of the NMR data is also acknowledged. Received for review April 13, 1999. Accepted August 17, 1999. AC990369A