Nonaqueous Capillary Zone Electrophoresis of Synthetic Organic

Sep 9, 2003 - Laboratoire Organisation Moléculaire, EÄvolution et Matériaux Fluorés, UMR CNRS 5073, Université de Montpellier II, case courrier 1...
0 downloads 0 Views 122KB Size
Anal. Chem. 2003, 75, 5554-5560

Nonaqueous Capillary Zone Electrophoresis of Synthetic Organic Polypeptides Herve´ Cottet,* Willy Vayaboury, Daniel Kirby, Olivia Giani, Jacques Taillades, and Franc¸ ois Schue´

Laboratoire Organisation Mole´ culaire, EÄ volution et Mate´ riaux Fluore´ s, UMR CNRS 5073, Universite´ de Montpellier II, case courrier 17, Place Euge` ne Bataillon, 34095 Montpellier Cedex 5, France

Poly(NE-trifluoroacetyl-L-lysine) was used as a model solute to investigate the potential of nonaqueous capillary electrophoresis (NACE) for the characterization of synthetic organic polymers. The information obtained by NACE was compared to that derived from size exclusion chromatography (SEC) experiments, and the two techniques were found to be complimentary for polymer characterization. On one hand, NACE permitted (i) the separation of oligomers according to their molar mass and (ii) the separation of the polymers according to the nature of the end groups. On the other hand, SEC experiments were used for the characterization of the molar mass distribution for higher molar masses. Due to the tendency of the solutes (polypeptides) to adsorb onto the fusedsilica capillary wall, careful attention was paid to the rinsing procedure of the capillary between runs in order to keep the capillary surface clean. For that purpose, the use of electrophoretic desorption under denaturating conditions was very effective. Optimization of the separation was performed by studying (i) the influence of the proportion of methanol in a methanol/acetonitrile mixture and (ii) the influence of acetic acid concentration in the background electrolyte. Highly resolved separation of the oligomers (up to a degree of polymerization n of ∼50) was obtained by adding trifluoroacetic acid to the electrolyte. Important information concerning the polymer conformations could be obtained from the mobility data. Two different plots relating the effective mobility data to the degree of polymerization were proposed for monitoring the changes in polymer conformations as a function of the number of monomers. Nonaqueous capillary electrophoresis is now recognized as a capillary zone electrophoresis (CZE) mode that can offer significant advantages over the aqueous CZE mode for many applications.1 The increased solubility of hydrophobic compounds in nonaqueous media consequently opens the possibility to extend the field of applications for CZE.1,2 Moreover, the use of nonaqueous electrolytes can induce an important change in selectivity in comparison with aqueous media.1,2 Numerous applications for the separation of charged or uncharged molecules by nonaqueous * Corresponding author. Tel: +33-4-6714-3427. fax: +33-4-6763-1046. Email: [email protected]. (1) Riekkola, M.-L. Electrophoresis 2002, 23, 3865. (2) Steiner, F.; Hassel, M. Electrophoresis 2000, 21, 3994.

5554 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

capillary electrophoresis (NACE) were reported and reviewed.1-4 In contrast to this work, almost all of the previous applications of CZE concerning the separation of synthetic polymers were performed in aqueous media5 either in free solution6,7 or in the presence of an entangled polymer solution.8,9 Indeed, considering the high dielectric constant of water, the majority of polymers bearing ionizable groups can be easily charged in aqueous electrolytes. However, the separation of non-water-soluble polymers by CZE is also of high interest since they constitute a significant portion of the synthetic polymer market. In this respect, solvent compatibility in terms of polymer solubility and polymer ionization is crucial. Methanol and acetonitrile are very good candidates for NACE since they offer good transparency for detection by UV absorbance and high compatibility with mass spectrometry. In NACE, formamide and N-methylformamide are also commonly used solvents.10,11 The advantageous properties of these solvents are their high dielectric constant, high solubilizing power, and low conductivity.10 However, these solvents possess a high UV cutoff, which may be overcome either by indirect absorbance detection12 or by on-line coupling with mass spectrometry.11 There are very few examples in the literature of the characterization of polymers by NACE. Nonionic polyethers were separated in methanolic electrolytes by complexation with cations.13 Mengerink et al.14 reported the separation of polyamide-6 oligomers by CZE in a hexafluoro-2-propanol/25 mM H3PO4 (65: 35 v/v) electrolyte, but this application was not based on a true nonaqueous electrolyte since 35% of it was constituted of water. The separation of anionic surfactants according to the alkyl chain length by NACE was reported by different groups,12,15-16 and (3) Bowser, M. T.; Kranack, A. R.; Chen, D. D. Y. Trends Anal. Chem. 1998, 17, 424. (4) Riekkola, M.-L.; Jussila, M.; Porras, S. P.; Valko, I. E. J. Chromatogr., A 2000, 892, 155. (5) Kok, W. Th.; Stol, R.; Tijssen, R. Anal. Chem. 2000, 72, 468A. (6) Hoagland, D. A.; Smisek, D. L.; Chen, D. Y. Electrophoresis 1996, 17, 1151. (7) Gao, J. Y.; Dubin, P. L.; Sato, T.; Morishima, Y. J. Chromatogr., A 1997, 766, 233. (8) Poli, J. B.; Schure, M. R. Anal. Chem. 1992, 64, 896. (9) Cottet, H.; Gareil, P. J. Chromatogr., A 1997, 772, 369. (10) Jansson, M.; Roeraade, J. Chromatographia 1995, 40, 163. (11) Geiser, L.; Cherkaoui, S.; Veuthey, J.-L. J. Chromatogr., A 2002, 979, 389. (12) Grob, M.; Steiner, F. J. Sep. Sci. 2002, 25, 615. (13) Okada, T. J. Chromatogr., A 1995, 695, 309. (14) Mengerink, Y.; van der Wal, S.; Claessens, H. A.; Cramers, C. A. J. Chromatogr., A 2000, 871, 259. (15) Drange, E.; Lundanes, E. J. Chromatogr., A 1997, 771, 301. (16) Grob, M.; Steiner, F. Electrophoresis 2002, 23, 1921. 10.1021/ac034526o CCC: $25.00

© 2003 American Chemical Society Published on Web 09/09/2003

although surfactants are not really considered as oligomers, this illustrates the interest of using nonaqueous electrolytes to increase the solute solubility and the selectivity of the separation. In this paper, poly(N-trifluoroacetyl-L-lysine) was used as a model family to study the potential of NACE for the characterization of synthetic polymers. It belongs to the family of synthetic polypeptides that are biomaterials of high interest considering their biocompatibility, their potential to mimic proteins, and their large range of applications such as drug delivery and tissue engineering.17 EXPERIMENTAL SECTION Chemicals. Methanol, acetonitrile, and dimethylformamide (DMF) were purchased from Carlo Erba (Val de Reuil, France). Glacial acetic acid (99%) and trifluoroacetic acid were from Acros (Noisy-le-Grand, France). Ammonium acetate and sodium hydroxide were from Prolabo (Paris, France). Sodium dihydrogenophosphate was from Aldrich (Saint-Quentin Fallavier, France). N-Hexylamine and mesityl oxide were from Avocado (La Tour du Pin, France). Sodium dodecyl sulfate (SDS) was from Merck (Darmstadt, Germany). Deionized water was further purified with a Milli-Q-system from Millipore (Molsheim, France). Polymers. Poly(N-trifluoroacetyl-L-lysine) was synthesized by ring-opening polymerization of N-trifluoroacetyl-L-lysine N-carboxyanhydride (Lys NCA) in DMF.18 The polymerization was initiated by n-hexylamine as described in detail elsewhere for a different monomer.19 Succinctly, DMF was distilled on a 4-Å molecular sieve under vacuum before the polymerization. After the synthesis of NR-carbamoyl-N-trifluoroacetyl-L-lysine (Lys NCAA) with potassium cyanate in aqueous medium,20 the Lys NCA was prepared by nitrosation of Lys NCAA using the NO/O2 gas mixture.21 In a Schlenk flask fitted with a stir bar and a silicon septum, 2.144 g of Lys NCA was dissolved under nitrogen in 50 mL of distilled DMF. Then, 105 µL of n-hexylamine corresponding to a molar ratio of monomer to initiator of 10 was added and the reaction mixture was stirred at room temperature. After 48 h, the solvent was evaporated under vacuum at 100 °C until obtaining a solid. The polymer was used without further purification. The number-average molecular mass measured by 1H NMR was of 2300 g/mol which indicated a ratio monomer to initiator of 10. For SEC analysis in aqueous eluent, the poly(N-trifluoroacetylL-lysine) was deprotected by reaction in a methanol/water mixture (60/40 v/v) containing 1 M piperidine for 48 h at room temperature. After deprotection, the solvent was evaporated under vacuum at 50 °C. Capillary Electrophoresis Instrumentation. Capillary electrophoresis (CE) was carried out either with an Agilent Technologies CE system (Waldbronn, Germany) equipped with diode array detector or a PACE MDQ Beckman Coulter (Fullerton, CA) apparatus. All separations were performed in the positive polarity mode. Sample volumes of ∼4 nL were introduced hydrodynamically. The temperature of the capillary cassettes was maintained (17) Deming, T. J. Adv. Drug Delivery Rev. 2002, 54, 1145. (18) Sela, M.; Arnon, R.; Jacobson I. Biopolymers 1963, 1, 517. (19) Nylund, R. E.; Miller, W. G. Biopolymers 1964, 2, 131. (20) Taillades, J.; Boiteau, L.; Beuzelin, I.; Lagrille, O.; Biron, J.-P.; Vayaboury, W.; Vandenabeele-Trambouze, O.; Giani, O.; Commeyras, A. J. Chem. Soc., Perkin Trans. II 2001, 1247. (21) Collet, H.; Bied, C.; Mion, L.; Taillades, J.; Commeyras, A. Tetrahedron 1996, 37, 9043.

constant at 25 °C. Data were collected at 200 nm or as otherwise indicated. Separation capillaries prepared from bare silica tubing were purchased from Composite Metal Services (Worcester, U.K.). Capillary dimensions were 48 cm (39.8 cm to the detector) × 50 µm i.d. for the Agilent instrument and 30 cm (20 cm to the detector) × 50 µm i.d. for the Beckman Coulter instrument. New capillaries were conditioned by performing the following washes: 1 M NaOH for 20 min, 0.1 M NaOH for 15 min, and water for 2 min. The polymer samples were prepared at 5 g/L in a methanol/ electrolyte (50:50 v/v) mixture. If necessary, 0.05% (v/v) mesityl oxide was added to the sample as a neutral marker. The prepunchers and electrodes were cleaned two times a week to remove crystalline deposits (SDS, urea). Rinsing Procedure Using Electrophoretic Desorption. To keep the capillary surface as clean as possible by removing any adsorbed polymers, we employed a method based on electrophoretic desorption in the presence of SDS micelles and urea. The denaturating electrolyte consisted of 60 mM SDS, 5 M urea in a 25 mM phosphate buffer, pH 7.0. Before each injection, the capillary was submitted to the following procedure: (i) flush with water for 1 min, (ii) flush with the denaturating electrolyte for 2 min, (iii) electrophoretic desorption by applying 15 kV between vials containing the denaturating electrolyte for 15 min, (iv) flush with denaturating electrolyte for 2 min, (v) flush with water for 1 min, (vi) flush with the NACE electrolyte for 3 min, (vii) equilibration of the capillary in the presence of the NACE electrolyte by applying 15 kV for 5 min, and (viii) flush with the NACE electrolyte for 1 min. Step viii was found to improve the baseline stability. Size Exclusion Chromatography (SEC) Instrumentation. The SEC analysis of the poly(N-TFA-L-lysine) was performed using a Jasco (Tokyo, Japan) PU-1580 pump and a Jasco RI-1530 refractive index detector. A set of three Tosohaas (Stuttgart, Germany) TSKGel G4000, G3000, and G2000 columns, each of 30 cm × 7.8 mm, coupled in a series with a guard column, were used. The temperature of the columns was set at 30 °C. The flow rate was set at 0.7 mL/min using dimethylformamide (DMF) as a mobile phase, and the injection volume was 20 µL. The calibration was performed using polystyrene standards of numberaverage molar masses ranging between 103 and 7 × 105 g/mol. Polymer samples were prepared at 5 g/L in DMF. Size Exclusion Chromatography Coupled with Multiangle Laser Light Scattering Detection (SEC-MALLS) Instrumentation. SEC-MALLS analysis of the poly(L-lysine) resulting from the deprotection of the poly(N-TFA-L-lysine) was performed using a Waters HPLC 515 pump (Saint-Quentin en Yvelines, France), with a flow rate of 0.5 mL/min using 1 M acetate buffer (pH 4.65) as mobile phase. The eluent was filtrated through 0.45- and 0.1-µm hydrophilic membranes (Milex SV, Millipore, Molsheim, France). The samples were injected through a 100-µL loop and eluted on a Waters Ultrahydrogel linear column of 30 cm × 7.8 mm. The temperature of the column and of the detectors was set at 35 °C. The polymer samples were prepared at 11 g/L in the eluent and were filtered through a 0.45-µm hydrophilic membrane before injection. A MALLS detector (Dawn DSP, Wyatt Technology Co, Santa Barbara, CA) coupled with a refractive index detector (Optilab, Wyatt Technology Co, Santa Barbara, CA) as concentration detector was used to obtain on-line determination of the Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5555

Scheme 1. Chemical Reactions Involved in the Propagation Step (a) and in the Termination Reaction b of the NCA Polymerization (the Initiator Is a Primary Amine)

absolute molar mass for each elution fraction of ∼0.01 mL. The refractive index increment of the polymer sample in the eluent was determined at 35 °C using the refractive index detector. RESULTS AND DISCUSSION The reaction in Scheme 1 depicts the propagation step for the ring-opening polymerization of N-trifluoroacetyl-L-lysine N-carboxyanhydride (Lys NCA) when the initiation step is activated by a primary amine. In the ideal case, this polymerization should be a living polymerization. However, and as described elsewhere under similar conditions,17,22 different side reactions can interfere with the propagation step. In this particular case, one of the most important side reactions is depicted by route b in Scheme 1. This side reaction can be considered as a termination reaction for the polymerization since the end group of the resulting polymer is no longer able to react with the monomer. The resulting polymer is described as a dead polymer in contrast to the aforementioned living polymer. It must be noted that the living polymer bears an amine end group while the dead one bears a carboxylic end group, and both polymers also bear a neutral end group (hexyl) originating from the initiation step. Considering their differences in functionalities, it was possible to separate the two polymers by NACE. Figure 1 shows the separation obtained in a MeOH/ACN (87.5: 12.5 v/v) mixture containing 1 M acetic acid and 20 mM ammonium acetate as a supporting electrolyte. In these electrophoretic conditions, the living polymers were found to be fully protonated (cationic) since the addition in the electrolyte of acetic acid up to 2 M did not induce any increase in the electrophoretic mobilities. The full protonation of the living polymers is also in accordance with the pKa data of the literature. Indeed, the pKa of acetic acid in pure MeOH is 9.7.1 Thus, if one neglects the influence of the 12.5% ACN, the pH in the electrolyte used in Figure 1 should be ∼8. The pKa of the amine end group of the polymer can be estimated at about 9-10 units in methanol (a pKa shift of 1-2 units from the value in water is usually observed for cation acids (BH+) such as primary amine).1 From these estimations of pKa values, the living polymers are expected to be more than 90% protonated. It must be noted that ion-pair formation may also occur in MeOH/ACN mixtures23 or in methanolic electrolytes;24 however, we did not investigate this issue further since it (22) Lundberg, R. D.; Doty, P. J. Am. Chem. Soc. 1957, 79, 3961. (23) Descroix, S.; Varenne, A.; Geiser, L.; Cherkaoui, S.; Veuthey, J.-L.; Gareil, P. Electrophoresis 2003, 24, 1577. (24) Porras, S. P.; Riekkola, M.-L.; Kenndler, E. Electrophoresis 2002, 23, 367.

5556

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

Figure 1. Separation of poly(N-TFA-L-lysine) by nonaqueous capillary electrophoresis. Electrolyte: 1 M acetic acid, 20 mM ammonium acetate in a MeOH/ACN (87.5:12.5 v/v) mixture (I ) 18 µA). Fused-silica capillary, 48 cm (39.8 cm to the detector) × 50 µm i.d. Applied voltage: +30 kV. Sample: 5 g/L in a MeOH/electrolyte (50:50 v/v) mixture. Hydrodynamic injection: 17 mbar, 3 s. UV detection at 200 nm. Temperature: 25 °C. Identification: number of monomeric units of the living poly(N-TFA-L-lysine) as indicated; (A) living polymer of high molar mass; (B, C) dead polymers (see Scheme 1 for the general chemical structure and see the text for details).

was not our purpose in this work. The living polymers were detected (peaks 3-23 and peak A) before the electroosmotic flow (EOF) marker and were separated according to their degree of polymerization n (as indicated in Figure 1 by the peak number) up to ∼20. Polymers of high molar masses (peak A) were detected close to the EOF marker since their effective mobilities tend to a zero value. The peak assignment in terms of their degree of polymerization was rendered possible by spiking the sample with n ) 1 or n ) 2 oligomers. These oligomers were respectively synthesized by a one-step reaction between n-hexylamine (respectively, n ) 1 oligomer) in excess and Lys NCA. The other peaks corresponding to the living polymers were assigned by assuming an increase in the degree of polymerization of one unit between two successive peaks. Peaks B and C in Figure 1 were assigned to dead polymers due to the following observation: when more monomers were added to the polymer sample in DMF to further continue the polymerization, peaks B and C increased in intensity while the proportion of peaks 3-23 and peak A was reduced (not shown). This was explained by a progressive death of the polymer as the polymerization continued further. Peak B and C correspond to anionic compounds since there were detected after the EOF marker. On the basis of the proposed chemical structure for the dead polymers (Scheme 1, route b), the negative

Table 1. Macromolecular Characteristics of Poly(NE-TFA-L-lysine) and Poly(L-lysine) Samples Determined from SEC, SEC-MALLS, and NACE Experiments instrumentation

peaks

Mn (g/mol)

Mw (g/mol)

SECa

1c 2c 1d

18200e 1500e 10500f

23700e 1650e 12300f

SEC-MALLSb

CEa

r1h (%)

r2i (%)

26

46

Ig 1.30 1.12 1.17

a Poly(N -TFA-L-lysine) b Poly(L-lysine) c For peak identification, see  Figure 2A. d For peak identification, see Figure 2B. e Calculated on the basis of a polystyrene calibration. f Absolute molar masses derived from SEC-MALLS measurements; dn/dC ) 0.1106 ( 0.0065 mL/g. g Polydispersity index. h r1 is the proportion in mass of dead polymers to total (dead or living) polymers. i r2 is the proportion in mass of living oligomers (n ) 1-10) to total living polymers. The polymer samples were both synthesized using a ratio of monomer to initiator of 10.

charge is due to the partial deprotonation of the carboxylic end group. Indeed, as already reported elsewhere in a similar electrolyte,25 carboxylic acids are negatively charged in these conditions. The same conclusions can be obtained from the estimation of the pKa values. The pKa value of a carboxylic acid in methanol is ∼5 units higher than in water.1 Thus, the pKa of the carboxylic acid of the dead polymer can be reasonably estimated at about 7-7.5. Assuming that in our electrophoretic conditions the pH is ∼8, the dead polymer should be partially deprotonated. It was not possible with UV detection to clarify the differences in chemical structure or in molar mass, which may explain the separation between peaks B and C. No oligomeric distribution was observed for the dead polymer by NACE. On the contrary, the presence of dead oligomers (n < 10) was observed by a aqueous CZE separation of the polymer sample after deprotection of the TFA lateral group. Deprotected dead oligomers were detected using as electrolyte a 100 mM phosphate buffer (pH 12.5; not shown). The mass proportion of dead polymer relative to the total quantity of polymer in the sample was derived from NACE by calculating the ratio of the corrected area of peaks (B + C) to the corrected area of all polymer peaks (3-23, A, B, C). About 26% in mass (Table 1) of dead polymer was quantified. This calculation is based on the approximation that the UV absorbance is mainly due to the peptide bond and is proportional to the polymer mass concentration independently of its degree of polymerization. The information derived from NACE was compared with that coming from SEC. SEC using DMF as a mobile phase was first investigated for the characterization of poly(N-TFA-L-lysine) (Figure 2A). As suspected in Figure 1, the polymer sample was constituted of a bimodal molar mass distribution with oligomers (Mn ≈ 1500 g/mol, peaks 2a and 2b, Figure 2A) and higher molar mass polymers (Mn ≈ 18 × 104 g/mol, 〈n〉 ≈ 81, peak 1, Figure 2A). It should be noted that the average molar masses derived from Figure 2A (Table 1) were calculated by using a standard calibration curve (polystyrene standards). The use of a light (25) Tjornelund, J.; Hansen, S. H. J. Biochem. Biophys. Methods 1999, 38, 139.

Figure 2. Size exclusion chromatograms of poly(N-TFA-L-lysine) (A) and of poly(L-lysine) (B). Eluent: DMF (A) and 1 M acetate buffer pH 4.65 (B). Identifications: 1, high molar mass fraction; 2a and 2b, low molar mass fractions; 3, peak corresponding to the salts. See Table 1 for numerical results. Both polymer samples were synthesized with a ratio of monomer to initiator of 10.

scattering detector for the characterization of poly(N-TFA-L-lysine) in absolute molar mass is difficult since the dn/dC value required for the calculation is very small (e.g., 2.59 × 10-2 mL/g in DMF containing 0.1 M LiBr at 60 °C). SEC-MALLS analysis was, however, possible on poly(L-lysine) samples leading to the determination of absolute molar masses. Thus, a similar sample synthesized with the same ratio of monomer to initiator was deprotected in order to remove the TFA lateral groups. This was further analyzed by SEC-MALLS using an aqueous mobile phase (Figure 2B). All the numerical results are gathered in Table 1. Taking into account the counterion condensation, the numberaverage molar mass (Mn ≈ 10.5 × 104 g/mol, peak 1, Figure 2B) of the high molar mass fraction corresponds to an average degree of polymerization 〈n〉 ) 66. Indeed, according to the Manning theory, the fraction of monomers (49%) that are not submitted to the counterion condensation was calculated by the ratio of the average spacing between charged groups on the polyion chain (0.34 nm for polylysine26) to the Bjerrum length (0.69 nm in water at 35 °C). Thus, the average molar mass of a monomer in the acetate buffer is 0.49 × 129 + 0.51 × 188 ) 159 g/mol. The 〈n〉 value obtained by SEC-MALLS is ∼18% below that derived from SEC in DMF. This discrepancy may reflect the error due to the polystyrene calibration. Even for a ratio of monomer to initiator (26) Bordi, F.; Cametti, C.; Motta, A.; Paradossi, G. J. Phys. Chem. B 1999, 103, 5092.

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5557

Table 2. Theoretical Plate Numbers Obtained in NACE of Poly(NE-TFA-L-lysine) for Two Different Rinsing Procedures of the Capillary and Three Different Oligomers rinsing procedure using a 5-min purge with the electrolyteb

rinsing procedure using the electrophoretic desorptionb

peak no.a

run 1

run 3

run 5

〈N〉c

RSD (%)c

5 10 14

25 × 103 66 × 103 48 × 103

24 × 103 21 × 103 10 × 103

11 × 103 9 × 103 nd

122 × 103 188 × 103 238 × 103

16 15 14

a Peak identification and electrophoretic conditions as in Figure 1. b The specified rinsing procedure was applied before each injection. See the Experimental Section for more details. c The average theoretical plate number 〈N〉 and the corresponding RSD were calculated on 20 successive injections.

of 10, relatively high molar masses were detected by SEC. Even so, the average number molar mass of the polymer sample as calculated by proton NMR (2300 g/mol, 〈n〉 ) 10 monomers) was in accordance with the ratio of monomer to initiator. In fact, the sample contained both a high proportion of small oligomers (46% in mass of the living polymers are living oligomers (n e 10) as calculated by NACE) and an important quantity of polymers of higher molar masses. So far, it can be inferred from this study that NACE permitted (i) the separation of the living oligo(N-TFA-L-lysine) according to their molar mass and (ii) the separation of the poly(N-TFA-Llysine) as a function of the nature of the end groups. This latter possibility is of high interest for the polymer chemist in order to quantify the proportion of dead polymer contained in polymer samples. On the other hand, SEC experiments allowed the characterization of the polymer samples according to the molar mass on a larger range than NACE did. However, SEC did not permit the separation according to the polymer functionalities and did not allow the baseline separation of the oligomers (peaks 2a and 2b in Figure 2A vs Figure 1). Thus, NACE and SEC appeared to be complementary techniques allowing a better characterization of the polymer. One of the challenging issues in the separation of polypeptides and proteins is prevention of the strong tendency of the solute to interact with the fused-silica capillary wall.27 Despite the presence of trifluoroacyl lateral groups, a study of the repeatability demonstrated that, from run to run, the migration times increased dramatically (Figure 3A-C), and the peak efficiency was also seriously affected (see Table 2). Since the initial interaction of the polymer on the capillary surface was not strong/critical enough to necessitate the use of a coated capillaries, a specific rinsing procedure was developed in order to keep the capillary surface as clean as possible. This rinsing procedure is based on the so-called electrophoretic desorption, which was originally used for quantifying the adsorption of proteins onto a capillary surface.28,29 This method consists of applying a voltage in the presence of a strong denaturating aqueous electrolyte containing SDS micelles (see Experimental Section). The anionic micelles migrate from the inlet to the outlet end of the capillary, draining with them the adsorbed solutes. In our procedure, 5 M urea was added in the denaturating electrolyte since it significantly helped to desorb (27) Dolnik, V.; Hutterer, K. M. Electrophoresis 2001, 22, 4163. (28) Barberi, R.; Bonvent, J. J.; Bartolino, R.; Roeraade, J.; Capelli, L.; Righetti, P. G. J. Chromatogr., B 1996, 683, 3. (29) Verzola, B.; Gelfi, C.; Righetti, P. G. J. Chromatogr., A 2000, 868, 85.

5558 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

Figure 3. Electropherograms of poly(N-TFA-L-Lysine) obtained by rinsing the capillary with the electrolyte (A, run 1; B, run 5; C, run 10) or by using a rinsing procedure based on electrophoretic desorption between each run (D, run 10; E, run 20). Rinsing conditions before injection: (A-C) 5 min electrolyte; (D, E) see Experimental Section for the electrophoretic desorption. Fused-silica capillary, 48.5 cm (40 cm to the detector) × 50 µm i.d. Other conditions as in Figure 1.

the solute. This method was found to be the most effective procedure for cleaning the capillary wall of any other, such as the following: rinse with NaOH, rinse with HCl, electrophoretic desorption using SDS without urea, and electrophoretic desorption using a neutral surfactant. When this specific procedure was applied between each run, then the repeatability was highly improved in terms of migration times (Figure 3D,E vs B,C) and in terms of efficiency (∼2 × 105 theoretical plates; see Table 2). However, the rinsing procedure including the electrophoretic desorption was quite long (30 min, see Experimental Section). To optimize the poly(N-TFA-L-lysine) separation by NACE, we first tried to study the influence of the solvent composition in the electrolyte. As shown in Figure 4, the increase of the MeOH proportion in the MeOH/ACN mixture induced better resolutions and an increase in the migration times. This was essentially due to the lowering effect of MeOH on the EOF since the effective mobility of the living polymer was found to be almost constant between 75 and 100% MeOH. A decrease in the effective electrophoretic mobility of the living polymers was also observed in 50% MeOH and may be explained by the lower protonation of living polymers when the proportion of ACN is increased. Indeed, ACN is expected to increase the pKa value of acetic acid to a higher extent than that of the amine end group.1 On the other hand, the peak symmetry deteriorated as the proportion of MeOH increased.

Figure 4. Influence of the proportion of methanol in the electrolyte on the separation of poly(N-TFA-L-lysine). Electrolyte: 1 M acetic acid, 20 mM ammonium acetate in a MeOH/ACN mixture, (A) 50:50 v/v, (B) 75:25 v/v, (C) 87.5:12.5 v/v, (D) 100:0 v/v. Fused-silica capillary, 30 cm (20 cm to the detector) × 50 µm i.d. Applied voltage: +20 kV. Hydrodynamic injection: 0.1 psi, 3 s. UV detection at 214 nm. Other conditions as in Figure 1.

Figure 6. Variations of (A) the frictional coefficient and of (B) the apparent selectivity as a function of the number of monomers n for poly(N-TFA-L-lysine). Electrolyte: 1 M acetic acid, 20 mM ammonium acetate in a MeOH/ACN (87.5:12.5 v/v) mixture (b) without and (2) with 30 mM trifluoroacetic acid. Other experimental conditions as in Figure 1.

Figure 5. Effect of the acetic acid concentration and of the addition of trifluoroacetic acid on the separation of poly(N-TFA-L-lysine). Fused-silica capillary 50 µm i.d., (A, B) 48 cm (39.8 cm to the detector); (C) 48.5 cm (40 cm to the detector). Electrolyte: 20 mM ammonium acetate in a MeOH/ACN (87.5:12.5 v/v) mixture containing (A) 0.13 M (1%) acetic acid, (B) 1 M acetic acid, and (C) 1 M acetic acid + 30 mM trifluoroacetic acid. Other experimental conditions as in Figure 1.

A good compromise between resolution, analysis time, solute ionization, and peak dissymmetry could be obtained at 87.5% MeOH (Figure 4C). Next, the influence of acetic acid concentration in the electrolyte was investigated (Figure 5A,B). The increase in acetic acid concentration from 0.13 (1% w/w) to 1 M led to better resolutions for the separation of the living oligomers, mainly because of the decrease in the EOF intensity due to the higher acidic conditions. As already mentioned, the effective mobilities were not affected by the increase in acetic acid concentration from 0.13 M up to 2 M, keeping constant the concentration in ammonium acetate at 20 mM. These results prove that the living polymers were fully charged in these electrolytes. Contrary to the addition of MeOH, acetic acid magnified selectivity without sacrificing peak efficiency. Concentrations in acetic acid higher than 1 M did not further reduce the EOF. Thus, the concentration was set at its optimal value of 1 M.

To go even deeper into the optimization of the oligomeric separation, we tried to find other additives that could reduce the EOF. A stronger acid than acetic acid, namely, trifluoroacetic acid (TFA), was added in the electrolyte in order to facilitate the protonation of the silanolate groups. Unexpected improvement of the resolution was observed with the addition of 30 mM TFA (Figure 5C) allowing the baseline separation of the living polymers up to a degree of polymerization of ∼50 (∼104 g/mol). However, it should be noted that a slow drift in the EOF was observed when separations in these conditions were repeated. Nevertheless, the surprising enhancement in the resolution was not only due to the moderating effect of TFA on the EOF intensity but also to the increase in the effective mobility of the living polymers for n > 9. To explain this increase in effective mobility, a change in the ionization of the polymer end group was discarded since the amine end group of the living polymer was found to be already fully protonated before the addition of TFA. Figure 6A displays the variation of the frictional coefficient, i.e., the ratio of the charge to the effective mobility, as a function of the degree of polymerization n of the living polymer in two different electrolytes, with and without TFA. For end-charged polymers, the frictional coefficient is expected to be mainly representative of the hydrodynamic frictional coefficient. Without TFA, the frictional coefficient of poly(N-TFA-L-lysine) was a linear function of n with a break in the slope for n ) 9. A linearity of the frictional coefficient with n was previously reported for different end-charged oligomers Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5559

such as oligosaccharides,30 fatty acids,31 and poly(ethylene glycol)s.32 The rupture in the slope that we observed in Figure 6A corresponds to the transition from the disordered form to the helical conformation of the polymer. This transition was accompanied by an increase in the slope of the frictional coefficient. We report in this work the opportunity to monitor or to detect changes in the polymer conformation using this graphic representation. It is worth noting that, for n < 9, the addition of TFA did not induce any increase in the electrophoretic mobility since the ionization and the conformation of the polymer were not affected. Thus, the slope of the first linear section (n < 9) remained unchanged while the slope corresponding to the helical conformation (n > 9) was ∼1.7 times smaller after the addition of TFA. This change in the slope was explained by the destabilization of the helical conformation when TFA was added. These results are consistent with the well known consideration that, for a given molar mass, the hydrodynamic frictional coefficient of a random coil is smaller than that measured for a more elongated structure such as a rod-like or a helical conformations. Moreover, the influence of TFA on polypeptide structures is well known and has been currently used for studying the helix-coil transition region as exemplified by the work of Nagayama et al. on poly-γbenzyl-L-glutamate.33 As for the separation performances, TFA had two advantages: the reduction of the e.o.f. intensity and the increase in the polymer effective mobility for n > 9. These two effects enlarge the time window of the oligomeric separation and led to much better selectivity as confirmed for n > 12 by plotting the apparent selectivity as a function of n with or without TFA in the electrolyte (Figure 6B). The apparent selectivity is defined for a pair of solutes as34

∆µep/(< µep > + µeo)

(1)

where ∆µep is the difference in apparent electrophoretic mobility between the two solutes, is the average of the effective mobilities, and µeo is the EOF mobility. For the large oligomers (n > 15), the selectivity is 2-3 times higher in the electrolyte containing TFA. In addition to its interest for analyzing the performances of the separation, Figure 6B also has the merit of emphasizing the changes in polymer conformations as exemplified by the positive peak for n ) 9 corresponding to the onset of the (30) Sudor, J.; Novotny, M. V. Anal. Chem. 1995, 67, 4205. (31) Cottet, H.; Gareil, P. Electrophoresis 2000, 21, 1493. (32) Oudhoff, K. A.; Schoenmakers; P. J.; Kok, W. T. J. Chromatogr., A 2003, 985, 479. (33) Nagayama, K.; Wada, A. Chem. Phys. Lett. 1972, 16, 50. (34) Bocek, P.; Vespalec; R.; Giese, R. W. Anal. Chem. 2000, 587 A. (35) Mitchell, J. C.; Woodward, A. E.; Doty, P. J. Am. Chem. Soc. 1957, 79, 3955.

5560

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

helical conformation. This finding is in accordance with the observation of Mitchell et al.35 on poly(γ-benzyl-L-glutamate). They reported that the minimum degree of polymerization required for stability of the R-helical configuration in DMF solution was found to be 10 ((3).35 In Figure 6B, a negative peak of small intensity for n ) 26 may also reflect a small change in the polymer conformation not yet elucidated. CONCLUSION We have demonstrated that NACE can be advantageously used for analyzing non-water-soluble synthetic polymers. The information concerning the characterization of the polymer derived from the NACE experiments was complementary to that obtained from SEC experiments. NACE allowed a baseline separation of the living oligomers and a quantification of the proportion of dead polymer in the sample. The potential of NACE for the characterization of synthetic polypeptides is of high interest for the polymer chemist since it may contribute to shedding more light on the mechanisms of NCA polymerization. The separation of the polymer according to its functionalities was possible by the use of a nonaqueous electrolyte in which amine groups were fully protonated (cationic) while carboxylic groups were partially deprotonated (anionic). Considering the importance of the polymer conformation on its reactivity, NACE may also be an attractive tool for analyzing the influence of the polymer conformation on the polymer synthesis at different stages of the process. Further works are in progress in order to establish a relationship between electrophoretic mobility and the geometrical characteristics of the polymer. Considering the composition of the nonaqueous electrolyte used in this work and the performances obtained for the polymer separation, a coupling of the CE apparatus with a mass spectrometer (MS) using an electrospray ionization interface should lead to a very powerful analytical instrumentation for the characterization of synthetic non-water-soluble polypeptides. The CE-MS coupling may also help to identify more clearly the chemical structure of the dead polymers. ACKNOWLEDGMENT The authors thank Dr. Andre´ Deratani from the Institut Europe´en des Membranes (IEM, Universite´ de Montpellier 2, France) for the SEC-MALLS experiment.

Received for review May 19, 2003. Accepted July 18, 2003. AC034526O