- Email: [email protected]
Simultaneous measurement of pHWS and overloaded band profiles of small peptides when insufficiently buffered mobile phases are used in preparative liquid chromatography
Journal of Chromatography A, 1216 (2009) 8874–8882
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Simultaneous measurement of SW pH and overloaded band profiles of small peptides when insufficiently buffered mobile phases are used in preparative liquid chromatography Fabrice Gritti, Georges Guiochon ∗ Department of Chemistry, University of Tennessee, 552 Buehler Hall, Knoxville, TN 37996-1600, USA
a r t i c l e
i n f o
Article history: Received 13 July 2009 Received in revised form 5 October 2009 Accepted 9 October 2009 Available online 4 November 2009 Keywords: Peptide analysis Peptide purification Preparative HPLC of acido-basic compounds Multi pKa compounds RPLC Adsorption mechanism XBridge-C18 Methanol–water mobile phases S pH W Ionic strength Buffer adsorption System peaks Phenylalanine-alanine
a b s t r a c t We recorded simultaneously the SW pH and the elution band profiles of phenylalanine-alanine (Phe-Ala, S pKa,1 = 3.6, SW pKa,2 = 7.5) on a column packed with C18 -bonded organic (ethyl)/inorganic (silica) hybrid W porous particles (BEH), eluted with a series of buffered methanol–water mobile phases (20/80, v/v). Using different buffers, the SW pHs of the mobile phases was set to values between 2.5 and 9.8. The ionic strength of the mobile phases was kept constant at 20 mM. The injected solutions were prepared with Phe-Ala as its zwitterionic species. Analytical to overloaded sample sizes were used, under strongly or insufficiently buffered conditions. With insufficiently buffered mobile phases, the experimental results show significant variations of SW pH during band elution, which proves a significant change of the concentration ratio of the different sample species during elution. This result confirms and validates our earlier general adsorption isotherm model of acido-basic compounds in RPLC. The SW pH record shows also the system peak of the buffer component, consecutive to sample injection. The phosphate II and carbonate II ions are adsorbed and may compete with the peptide for adsorption onto the active sites of BEH-C18 . This suggests refinements of the general adsorption isotherm model, depending on the nature of the buffer components, including competitive adsorption of the sample and the buffer species. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Investigations of retention and adsorption mechanisms of acidobasic compounds in RPLC are attracting much attention [1–6], the goal of which is the optimization of optimum experimental conditions for the separation and purification of pharmaceutical and biological compounds. Under linear conditions, the pH of the mobile phase, usually an aqueous mixture of methanol or acetonitrile, imposes the concentration ratio of the compounds involved. Several general retention models have been proposed to account for the retention pattern of a broad range of acids and bases over a pH range from 2 to 11 and for large variation in the mobile phase composition [4,7,8]. It is systematically observed that, under moderately overloaded conditions, the peak shapes of the ionizable species of the sample components tail more strongly than those of their conjugated neu-
∗ Corresponding author. Fax: +1 865 974 2667. E-mail address: [email protected] (G. Guiochon). 0021-9673/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2009.10.026
tral species. Hence, acidic and basic compounds are preferentially analyzed at low and high pHs, respectively, in order to generate more symmetrical and narrower elution peaks [5]. Under severe overloaded conditions such as in preparative chromatography, the sample concentration may exceed the buffer concentration and the buffer becomes too weak to set the pH of the mobile phase. Also, if the buffer is a strong ion-pairing agent, large variations in the retention factor may eventually take place because the co-ion which accompanies the sample is different at low and at high buffer concentrations. The prediction of the band profiles eluted in insufficiently buffered mobile phases requires the knowledge of the individual adsorption isotherms of each sample species and that of their local concentrations, anywhere at any time. These different species are constantly in equilibrium but their relative concentrations depends on the total concentration of the sample component, on its pKa , and on that of the buffer. Equilibration between these species in both the mobile and the adsorbed phases is assumed to be infinitely fast, so the total adsorbed amount of acido-basic compounds is directly obtained from the molar fractions of each species in the bulk and the individual adsorption isotherms.
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
A consistent but arguable thermodynamic adsorption model was proposed, based on the analysis of the adsorption energy distributions (AEDs) of ionizable and neutral compounds in RPLC [9]. Neutral sample molecules are weakly adsorbed on a large number of adsorption sites while ionizable molecules are strongly adsorbed onto very few active sites [10]. Because the total concentration of each sample component varies during its elution, so does the relative concentration of each one of its species in a weakly buffered mobile phase. These concentrations at any position and time inside the column are obtained by solving the thermodynamics problem in the bulk solution [11,12]. The extended Debye–Hückel method was used to estimate the activity coefficients of the ionizable compounds [13]. Finally, the equilibrium-dispersive model of chromatography provides the elution profiles with the total concentration of the compound as the sum of the concentrations of all its species. The calculated band profiles agree well with the observed ones, giving unusual anti-langmuirian and S-shaped profiles in the cases of aniline and 2-phenylbutyric acid [12], although the adsorption isotherms of both the acid and the base species of these two compounds are langmuirian. The pH dependency of the shape of the overloaded band profiles of compounds with two pKa s (hence, a total of three species) such as o-phthalic acid and nicotinic acid was also well accounted for by this general adsorption model. Finally, the same approach was applied for the prediction of the elution band profile of the peptide Phe-Ala (which is a biprotic compound with SW pKa,1 = 3.6 and SW pKa,2 = 7.5) under insufficiently buffered mobile phases [14]. In this case, only qualitative agreement was observed between experimental and calculated band profiles with the basic mobile phases made with either ammonia or carbonate buffers. This implied that the buffer component was not an inert compound and eventually adsorbs onto the adsorbent, competing with at least one species of the compound studied. The arguable part of this model is our identification of the weak neutral sites with silica-bonded C18 chains and of the few high energy active sites with the accessible silanol groups amidst the C18 bonded layer. Because the method and the model used are thermodynamic, they inform on the energy of the adsorption sites involved and on their density but not on their microscopic or chemical structure. In this work, we investigated in more detail the complex adsorption mechanism of the di-peptide Phe-Ala (SW pKa,1 = 3.6 and SW pKa,2 = 7.5) onto BEH-C18 by recording simultaneously the overloaded elution band profiles (measured using UV detection) and the eluent SW pH (measured with a micro-pH electrode with a low cell volume). The main goal was to validate the model proposed for the adsorption of acido-basic compounds in RPLC and to assess its limit of applications. The measurements were performed under controlled and uncontrolled pH conditions, with six different mobile phase SW pH between 2.4 and 9.8, and the results are compared. Small and large sample volumes were injected in order to investigate the impact of the adsorption of the buffer components on the band profiles of the small peptide. 2. Experimental 2.1. Chemicals The mobile phase was a solution of methanol and water (20/80, v/v), both HPLC grade from Fisher Scientific (Fair Lawn, NJ, USA). The mobile phase was filtered before use on a surfactant-free cellulose acetate filter membrane, 0.2 m pore size (Suwannee, GA, USA). The peptides used were the dipeptide phenylalanine-alanine (Phe-Ala) (Fig. 1) and phenylalanine-valine (Phe-Val). Phosphoric acid (85% H3 PO4 ), potassium dihydrogenphosphate (KH2 PO4 ), dipotassium hydrogenphosphate (K2 HPO4 ), formic acid (98% HCOOH), sodium formate (NaHCOO), glacial acetic acid (> 99.5%
8875
Fig. 1. Chemical structures and conjugated species of the dipeptide phenylalaninealanine used in this study. The two SW pKa values indicated were determined from the SW pHs measured in this work for 44 mM solutions of Phe-Ala in different aqueous buffers containing 20% methanol in volume.
CH3 COOH), sodium acetate (NaCH3 COO), ammonium chloride (NH4 Cl), ammonium hydroxide (30% NH3 ), potassium hydrogencarbonate (99% KHCO3 ), and disodium carbonate (Na2 CO3 ), all from Fisher Scientific, were used to prepare the buffer solutions (see Table 1 for the SW pH of the buffers). 2.2. Materials An Agilent 1090 liquid chromatograph (Agilent, Palo Alto, CA, USA) was used to perform the measurements. This instrument includes a ternary solvent delivery system, an auto-sampler with a
8876
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Table 1 Buffer solutions in a water/methanol (80/20, v/v) mixture. Ionic strength I=20 mM. The measured 44 mM solution of dipeptide Phe-Ala (the concentrations refer to the water/methanol volume). Buffer number I II III IV V VI
Name Phosphate I Formate Acetate Phosphate II Ammonia Carbonate II
Acid [M]
Base [M]
H3 PO4 0.020 HCOOH 0.020 CH3 COOH 0.020 KH2 PO4 0.005 NH4 Cl 0.020 KHCO3 0.005
are given for the neat buffer solution, the 2 mM solution, and the
Buffer Exp.
K2 HPO4 0.020 HCOONa 0.020 CH3 COONa 0.020 K2 HPO4 0.005 NH4 OH 0.020 Na2 CO3 0.005
250 L sample loop, a diode-array UV detector (cell volume 1.7 L, sampling rate 25 Hz), a column oven, and a data station running the HP data software. From the needle seat to the column inlet and from the column outlet to the detector cell, the total extra-column volume of the instrument is 45 L, measured as the apparent hold-up volume of a zero-volume union connector in place of the column. The chromatographic column (150 mm × 4.6 mm packed with XBridge-C18 particles, average size 3.5 m, average pore size 136 Å, specific surface area 172 m2 /g, bonding density 16.95% C, surface coverage 3.13 mol/m2 ) was a generous gift from Waters (Milford, MA, USA). The total porosity t = 0.637 of this column was estimated from pycnometric measurements (made using methanol and dichloromethane), giving a column hold-up volume of 1.588 mL [13]. The packing material was endcapped according to a proprietary process. Table 2 summarizes the manufacturer’s data. The SW pHs of the mobile phases and sample solutions were measured with a flow through micro-pH-electrode (Lazar Research Laboratories Inc.), calibrated in pure water with two standard aqueous solutions at pH = 4.00 and 9.00. The cell, through which the eluent circulates and is in contact with the electrode, has a volume of 20 L. The plots of the eluent pH versus the elution volume were independent of the flow rate when this flow rate was set below 0.25 mL/min, demonstrating a sufficiently fast response of the pH electrode. The pH data were recorded every second.
S pHs W
2 mM, Phe-Ala, Sample Exp.
44 mM, Phe-Ala, Sample Exp.
S pH W
S pH W
S pH W
2.53 3.89 4.83 6.91 9.40 9.80
2.58 3.91 4.84 6.75 9.17 9.25
3.71 4.24 4.89 5.35 6.71 6.87
volumetric glasses. So, the largest injected sample concentration of Phe-Ala was 10 g/L. 2.4. Injections The injected sample volumes were 30 and/or 180 L, unless indicated otherwise. The flow rate was set at 0.5 mL/min (inlet column pressure close to 140 bar) and the temperature was maintained at 295 ± 1 K by the air-conditioning system of the laboratory. 3. Results and discussion In the next section, we discuss the relationship between the profile recorded with the flow-through micro-pH-electrode and the experimental UV-band profiles of the dipeptide Phe-Ala. In all cases, the solution injected was prepared by dissolving the zwitterionic species of Phe-Ala directly into the six buffer solutions. The ionic strength of all the buffer solutions was equal to 20 mM. Two different conditions are examined: (1) the concentration of Phe-Ala in the sample is smaller than the buffer concentration (2 mM Phe-Ala) and (2) its concentration is larger than that of the buffer (44 mM Phe-Ala). We analyze the results one by one for each buffered mobile phase at SW pH = 2.53 (phosphate I), 3.89 (formate), 4.83 (acetate), 6.90 (phosphate II), 9.03 (amoniac) and 9.80 (carbonate II). The aqueous solutions contain all 20% methanol in volume. The flow rate was constant at 0.50 mL/min in all the experiments.
S pH W
2.3. Buffer and sample preparation Six buffer solutions (total volume 150 mL each) were prepared by dissolving the appropriate amounts of the acid and its conjugated base in order to obtain SW pH values increasing from about 2 to 10 at a constant ionic strength of 20 mM. All the measured SW pH of the buffers are listed in Table 1. For each buffer solution, 2 mM and 44 mM solutions of Phe-Ala were prepared by dissolving 4.5 and 100 mg, respectively, in 10 mL Table 2 Physico-chemical properties of the column used in this work according to the manufacturer. Neat silica
Xbridge 3.5 m
Particle size [m] Pore diameter [Å] Surface area [m2 /g]
3.5 136 172
Bonded phase
Xbridge -C1 8
Total carbon [%] Surface coverage [mol/m2 ] Endcapping
16.95 3.13 Yes
Packed column Serial number Dimension (mm × mm) Total porosity a a
128B3812113683 4.6× 150 0.637
Measured by pycnometry (CH3 OH/CH2 Cl2 ) at 295 K.
3.1. Buffer I: Phosphate I, SW pH = 2.53 At infinite dilution, Phe-Ala is mostly present in solution as its cationic species, + Phe-Ala. The SW pH of the 2 mM solution barely differs from that of the neat buffer, at 2.58 rather than 2.53. When 180 L of this solution is injected into the BEH-C18 column, the S pH remains almost constant during elution, except at the holdW up column time (starting at t = 2 × 1.588 = 3.18 min) and during the elution of Phe-Ala, from 6.7 min to 8.6 min (Fig. 2). These pH perturbations are very small (< 0.03 pH unit). The first perturbation is related to the difference in the concentration ratio of the phosphate buffer species H3 PO4 and H2 PO4 − caused by the slight difference between the SW pH of the injected solution and that of the neat buffer. A perturbation of the equilibrium concentration of phosphate takes place and the usual system peak of the additive compound phosphate is detected [15]. The second perturbation is directly related to the change in the concentration ratio of + Phe-Ala and + Phe-Ala− during the elution of its band. The very slight increase of the SW pH (< 0.1 pH unit) suggests that the molar fraction of + Phe-Ala− slightly increases. This was expected due to the slightly larger SW pH of the 2 mM sample solution. When 180 L of a 44 mM solution of Phe-Ala is injected, the amplitude of these two pH perturbations are largely increased because the SW pH of the injected sample (3.71) exceeds that of the neat buffer by more than 1 pH unit. Accordingly, the initial concentration ratio of the H3 PO4 and H2 PO4 − species is smaller than that of the neat buffer. At the hold-up time, the SW pH of the elu-
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 2. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid red line, right scale) of Phe-Ala eluted from the 150 mm × 4.6 mm I.D. BEH-C18 column with a 20 mM phosphate I buffer (SW pH = 2.53) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. The SW pH of the injected peptide solutions are given in Table 1. Note the two positive pH disturbances, one before, the second during elution of the peptide. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) + Phe-Ala
and ent rises to 3.0. Similarly, the concentration ratio of + Phe-Ala− increases. As the total Phe-Ala concentration increases in the bulk mobile phase, the molar fraction of the zwitterion increases, explaining the second increase of the pH observed during the elution of the peptide. This demonstrates that the front spike (high total concentration) of the chromatogram is richer in zwitterions than its tail (small total concentrations). This last observation is very important because it comforts the hypothesis made in designing the general adsorption isotherm model of acido-basic compounds and predicting their band profiles in insufficiently buffered mobile phases [11,12,9,13,16,14]. The deformation of the band profiles originates from (1) the difference in the adsorption isotherms of the different species onto the adsorbent and (2) the variation of their respective molar fractions in the mobile phase, which depends on the total sample concentration. 3.2. Buffer II: Formate I, SW pH = 3.89 At a buffer SW pH of 3.89, the zwitterion of Phe-Ala predominates. Accordingly, the SW pH of the 2 mM and 44 mM sample solutions of Phe-Ala should not differ much from 3.89. Actually, they are 3.91 and 4.24, respectively.
8877
Fig. 3. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid red line, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM formate buffer (SW pH = 3.89) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the decrease in amplitude of the pH disturbances compared to those observed in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Fig. 3A and B shows the variation of SW pH. The system peak of the formate buffer is clearly detectable at the hold-up time. The SW pH increases by no more than 0.02 and 0.15 pH unit when 2 mM and 44 mM sample concentrations are injected, respectively. Interestingly, the system peak of the buffer formate tails (Fig. 3B), which suggests that the anion HCOO− is retained on the BEH-C18 adsorbent. This is consistent with the adsorption properties of this column [16,14]. Note that the buffer perturbation signal in Fig. 2B does not tail, which suggests that the phosphate buffer is only slightly retained. The system peak will interfere with the peptide band, leading to more complex SW pH profiles during the elution of the peptide. At high concentrations, the band profile corresponds to an S-shaped isotherm because the adsorption isotherm of the zwitterion alone is anti-langmuirian. At low concentrations, the band tails because the cation has a Langmuir adsorption isotherm [14]. 3.3. Buffer III: Acetate I, SW pH = 4.83 The zwitterion + Phe-Ala− predominates also at the SW pH of the acetate buffer. The SW pH change upon dissolving 2 mM and 44 mM Phe-Ala in this buffer are the smallest (see Table 1). Fig. 4A and B shows the corresponding UV and pH profiles. The retention of the zwitterion is so small (minimum) and the injected volume so large (180 L) that the system peak of acetate and the band profile of the
8878
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 4. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM acetate buffer (SW pH = 4.83) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the negligible amplitude of the pH disturbances.
peptide eventually interfere. The system peak of acetate is barely positive. The passage of this compound is accompanied by a very small but detectable pH decrease similar to that observed with the buffer formate. Fig. 4B and the fronting of the band of + Phe-Ala− confirms the anti-langmuirian shape of the adsorption isotherm of the zwitterion. This behavior is most probably related to the ionic interactions between the positive and the negative charges of two zwitterionic molecules. Overall, the SW pH of the eluent during elution is the least affected by the injection of concentrated solution of the peptide since the Phe-Ala dissolved does not react with the neat buffer solution. We simply observe the band profile of a single species, the zwitterion, with no significant change in the molar fractions of the other ionic species in solution.
Fig. 5. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM phosphate II buffer (SW pH = 6.90) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 30 L. (C) Peptide concentration: 44 mM, volume injected: 180 L. Note the interference between the two pH disturbances.
3.4. Buffer IV: Phosphate II, SW pH = 6.91 Dilute solutions (2 mM) of the peptide in this buffer contain both the zwitterion and the anion Phe-Ala− . The SW pH of the 2 mM and 44 mM sample solutions are smaller than that of the neat buffer because the zwitterion can react with the phosphate buffer and displaces the initial H2 PO4 − /HPO4 2− equilibrium toward the formation of more H2 PO4 − anions. Accordingly, the SW pH profiles in Fig. 5A–C shows a negative peak, the intensity of which grows with increasing sample volume and/or concentration. Inter-
estingly, the eluent SW pH increases when the front shock layer of the peptide is eluted. We could have expected the opposite because the ratio of the concentrations of the zwitterions and of the anions increases with increasing total sample concentrations (see Table 1). This shows that the general adsorption model proposed for Phe-Ala [14] has limits. There is a serious disagreement between the experimental band profiles shown in Fig. 5C and those
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
8879
single-component adsorption behavior compete for adsorption [15]. 3.5. Buffer V: Ammonia, SW pH = 9.40 The mobile phase SW pH was increased further to 9.40 by using the ammonia buffer. In this case, the pH is well above the second S pK of the peptide (7.5) and the dissolved neutral compound Phea W Ala reacts strongly with the buffer to be quantitatively transformed into the anion Phe-Ala− , unless the initial sample concentration largely exceeds the buffer capacity. Fig. 6A–C shows the corresponding SW pH and band profiles recorded for the three different sample loadings used. The size of the system peak of NH4 + /NH3 increases with increasing sample size because the dissolution of 2 mM and 44 mM of Phe-Ala in the ammonia buffer causes a decrease of the buffer SW pH (see Table 1) and an increase of the concentration ratio of NH4 + to NH3 . Due to the ammonia and ammonium ion lack of chromophores, the relative change in the concentration ratio of the two buffer species is not detected. Would the injection volume exceed 180 L, the system peak of the ammonia buffer would interfere with the band profile of the peptide. Note that the SW pH slightly decreases during elution of the peptide, demonstrating the progressive increase of the ratio of the concentration of + Phe-Ala− to that of Phe-Ala− when the total peptide concentration increases. The SW pH
Fig. 6. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM ammonia buffer (SW pH = 9.30) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 30 L. (C) Peptide concentration: 44 mM, volume injected: 180 L. Note the small and negative amplitude of the pH disturbance during the sample elution.
calculated in the intermediate concentrations region. More specifically, the band profiles calculated with the phosphate II buffer should have corresponded to a front spike with an S-shaped profile. Clearly, this is not the experimental profile recorded. This suggests that the competitive adsorption model involving the zwitterions (anti-Langmuir behavior) and the anions (Langmuir behavior) that we used in the calculations is too simple. This situation is expected when compounds having so different
Fig. 7. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM carbonate II buffer (SW pH = 9.80) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the absence of pH disturbance during the sample elution.
8880
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 8. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of a mixture of Phe-Ala and Phe-Val eluted from the same column as in Fig. 2, with four different 20 mM buffers in a methanol:water solution (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. The sample concentrations are 10 g/L for each compound. The sample volume injected is 250 L (injection time, 30 s). (A) Phosphate I buffer (SW pH = 2.4). (B) Acetate buffer (SW pH = 5.1). (C) Phosphate II buffer (SW pH = 7.3). (D) Carbonate II buffer (SW pH = 10.3).
profile mirrors closely the UV profile and shows the direct relationship between the total peptide concentration and the solution SW pH. This confirms the theoretical assumption made in the derivation of the overloaded band profiles of acido-basic compounds [11,13] when recalculating at each step of the simulation process (at time t and at the axial coordinate z) the molar fraction of each sample species as a function of the total sample concentration.
3.6. Buffer VI: Carbonate II, SW pH = 9.80 The SW pH of the neat carbonate buffer is slightly larger than that of the ammonia buffer. Fig. 7A and B shows that the system peaks of the carbonate buffer are more severely distorted and more strongly retained than those of the ammonia buffer (Fig. 6A–C). As the bulk concentration of hydrogencarbonate increases and that of bicarbonate decreases, the amount of buffer adsorbed increases. We observed that the pH is perturbed from t = 3.2 min to t = 6 min. If the buffer species were both unretained on the stationary phase, the pH perturbation would start at 3.2 min and end at 3.6 min since the injection lasts only 0.36 min. A comparison of Figs. 6C and 7B suggests that the carbonate buffer is more retained than the ammo-
nia buffer, which is consistent with the fact that BEH-C18 retains singly and negatively charged species as previously reported for nicotinate [16] and Phe-Ala− [14]. Since the SW pH largely decreases before elution of the peptide and the concentration ratio of the monovalent bicarbonate anion and the divalent carbonate anion increases in the mobile phase, the retention of the buffer increases, showing that HCO3 − is more retained than CO3 2− . The interference between the system peak of the carbonate buffer and the elution band profile of Phe-Ala clearly occurs all along the chromatographic column. It can seriously affect the band shape of the main component [17–20]. In Fig. 7B, the elution front of the peptide begins before the mobile phase SW pH has completely returned to its initial value of 9.8. Note that there is no obvious discontinuity in the SW pH profile at that moment suggesting that the peptide was almost entirely converted into its anion at elution. The variation of the SW pH of the eluent is solely due to the adsorption behavior of the monohydrogencarbonate/bicarbonate buffer onto BEH-C18 and to the disturbance of the initial equilibrium due to injection of the peptide, the SW pH of which is significantly smaller than that of the neat buffer (6.9 versus 9.8). This contrasts with the result shown in Fig. 6B and C, where the elution of the dipeptide is accompanied by a change of the eluent SW pH.
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
In conclusion, one cannot accurately predict the large overloaded band profiles of Phe-Ala in presence of the carbonate buffer because the adsorption of Phe-Ala− eventually depends on the concentration of the buffer species HCO3 − , which continuously varies during elution of the peptide, from the inlet to the outlet of the column. Differences between calculated and experimental band profiles were observed under such conditions [14]. 3.7. Application to the purification of mixtures of two small peptides The profiles of the SW pH and the peptide concentrations during the elution of a 250 L sample of a mixture of Phe-Ala and PheVal (10 g/L each) were recorded. Fig. 8A–D shows these profiles at four different SW pHs, 2.43, 5.10, 7.30, and 10.3 of the mobile phase. The ionic strength of the buffer was kept constant (20 mM). Except at SW pH = 5.10, at which the presence of either peptide does not affect much the mobile phase pH, the elution of the first eluted compound (Phe-Ala) strongly interferes with the system peak of the buffer additive. Although the two peptides differ only by one methylene group in the side-chain, they are well resolved and the concentration ratio of the two species of each peptide at elution is different. Nevertheless, the complex profile of the two-component band can be interpreted based on the adsorption behavior of the two single peptides. At SW pH = 2.43, the system peak of the formate buffer is spread over a large fraction of the chromatogram. The eluent pH-spike following elution of the first compound (Phe-Ala) is much higher than that accompanying the front shock of the second one (PheVal) (Fig. 8A). The band profile of Phe-Ala is typical of those of compounds having an S-shaped isotherm, showing that a significant fraction of Phe-Ala is eluted as the zwitterion. In contrast, the elution profile of Phe-Val corresponds to that of a compound having a langmuirian isotherm, hence for which the relative abundance of the cation in the bulk during its elution is larger than that of the zwitterion. This illustrates the importance of the interference between the bands of the additive and of the main components when the mobile phase is insufficiently buffered and large volumes of sample solution are injected, and the injection last for a time that is not negligible compared to the hold-up time. This interference can strongly affect the band shapes of the main components, to an extent that depends on the injected volume. At SW pH = 5.10 (Fig. 8B), the presence of large concentrations of Phe-Ala and Phe-Val do not affect the SW pH of the neat buffer. The two compounds are well resolved and both band shapes are anti-langmuirian, as expected from their individual simple Moreau adsorption isotherms [14]. The adsorbate–adsorbate interactions between the adsorbed zwitterions are stronger with Phe-Val than with Phe-Ala (see the enhanced peak fronting in Fig. 8B). At SW pH = 7.30 (Fig. 8C), the system peak of the phosphate buffer interferes with the band of Phe-Ala during nearly its whole elution time. The SW pH drops by almost 1 pH unit, to values at which this peptide is entirely under the zwitterion state. This explains the anti-langmuirian shape observed in Fig. 8C, between 4.5 min and 6.0 min. Beyond this time, the SW pH returns to more neutral values (higher than 7) and the anion concentration in the bulk is no longer negligible. Then the band of Phe-Ala ends in a langmuirian shape that is characteristic of the adsorption behavior of ionic species. The elution of Phe-Val is accompanied with a slight increase in the buffer pH, demonstrating that, in contrast to PheAla, the anion concentration is always significant during its elution. Note that the spike observed in the chromatogram is an artefact of the UV detector and is not due to a sharp increase of the peptide concentration.
8881
Finally, at SW pH = 10.3 (Fig. 8D), the system peak of the buffer carbonate interferes with the elution bands of both compounds, affecting seriously the shape of the least retained peptide (Fig. 8D). The SW pH drops to about 7, at which pH the zwitterions + Phe-Ala− concentration in the mobile phase is significant. This explains the large increase of the Phe-Ala concentration at about 5 min. Beyond this time, the SW pH returns to a basic value and both bands tail strongly, according to the adsorption behavior of the anions.
4. Conclusion The simultaneous records of the SW pH and the elution profiles of overloaded bands of Phe-Ala permits the assessment of the causes of the variations of the SW pH during elution of this peptide with an insufficiently buffered mobile phase. From a general point of view, we conclude that the initial equilibrium SW pH imposed by the buffered mobile phase is disturbed twice after injection of the peptide sample 1. The first SW pH disturbance is directly related to the system peak of the buffer species, which is not retained or only weakly so. Although often undetected by the UV detector, this system peak is systematically detected by the pH micro-electrode when the mobile phase is insufficiently buffered. This is due to the reaction of the peptide with the neat buffer. The concentration ratio of the acidic and the basic buffer species is significantly different in the presence of the peptide and in the neat buffer. This causes a perturbation of the bulk/adsorbent equilibrium, which propagates along the column. The shape of this perturbation depends on the adsorption behavior of the buffer species on the solid adsorbent. 2. The second SW pH disturbance is caused by the elution of the sample component. The ratio of the concentrations of the acidic and basic peptide species depends on the total peptide concentration. The analysis of the SW pH-concentration profiles during elution of the peptide confirms an assumption made in the calculation of the overloaded band profiles of acido-basic compounds when using insufficiently buffered mobile phases. In the calculation of the total amount adsorbed the molar fraction of each species in the bulk phase is computed as a function of the total bulk concentration at each time t and for each position z along the column. These molar fractions change during elution, causing a parallel drift of the mobile phase SW pH. The validity of this assumption is unambiguously confirmed by our results. The limit of validity of our general adsorption isotherm of acidobasic compounds is due to its neglecting the adsorption of the buffer components. When the SW pH of the injected sample significantly differs from that of the buffered mobile phase, the buffer concentration changes during sample elution. When the retention of the buffer species is significant compared to that of the sample and when the injected volume is large, the two SW pH disturbances or composition waves of the buffer and the sample species interfere. This leads to more complex band profiles than could initially be expected. The sample is not eluted at a well-defined buffer composition and the general adsorption isotherm model previously reported is only an approximation of the true adsorption behavior, for the neglect of the competition with the buffer components. Agreement between calculated and experimental band profiles is merely qualitative but major improvement would be made by accounting for the competitive adsorption between the buffer and the sample components.
8882
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Acknowledgements This work was supported in part by grant CHE-06-08659 of the National Science Foundation and by the cooperative agreement between the University of Tennessee and the Oak Ridge National Laboratory. We thank Uwe Neue (Waters, Milford, MA) for the generous gift of the new hybrid BEH-C18 packing material and for informative discussions. References [1] N.H. Davies, M.R. Euerby, D.V. McCalley, J. Chromatogr. A 1119 (2006) 11. [2] D.V. McCalley, J. Chromatogr. A 1075 (2005) 57. [3] N.H. Davies, M.R. Euerby, D.V. McCalley, J. Chromatogr. A 1178 (2008) 71. [4] U.D. Neue, C.H. Phoebe, K. Tran, Y.-F. Cheng, Z. Lu, J. Chromatogr. A 925 (2001) 49.
[5] J. Samuelsson, A. Franz, B.J. Stanley, T. Fornstedt, J. Chromatogr. A 1163 (2007) 177. [6] L. Pan, R. LoBrutto, Y.V. Kazakevich, R. Thompson, J. Chromatogr. A 1049 (2004) 63. [7] U.D. Neue, A. Méndez, K. Tran, P.W. Carr, J. Chromatogr. A 1063 (2005) 35. [8] Y.V. Kazakevich, J. Chromatogr. A 1126 (2006) 232. [9] F. Gritti, G. Guiochon, Russ. Bull. 71 (2009) 1. [10] F. Gritti, G. Guiochon, J. Chromatogr. A 1099 (2005) 1. [11] F. Gritti, G. Guiochon, J. Sep. Sci. 31 (2008) 3657. [12] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 63. [13] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 3175. [14] F. Gritti, G. Guiochon, Anal. Chem., accepted for publication. [15] G. Guiochon, S.G. Shirazi, A. Felinger, A.M. Katti, Fundamentals of Preparative and Non-linear Chromatography, Academic Press, Boston, MA, 2006. [16] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 6917. [17] T. Fornstedt, G. Guiochon, Anal. Chem. 66 (1994) 2116. [18] T. Fornstedt, G. Guiochon, Anal. Chem. 66 (1994) 2686. [19] P. Sajonz, Y. Tong, G. Zhong, T. Fornstedt, G. Guiochon, J. Chromatogr. A 734 (1996) 75. [20] G. Zhong, T. Fornstedt, G. Guiochon, J. Chromatogr. A 734 (1996) 63.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Simultaneous measurement of SW pH and overloaded band profiles of small peptides when insufficiently buffered mobile phases are used in preparative liquid chromatography Fabrice Gritti, Georges Guiochon ∗ Department of Chemistry, University of Tennessee, 552 Buehler Hall, Knoxville, TN 37996-1600, USA
a r t i c l e
i n f o
Article history: Received 13 July 2009 Received in revised form 5 October 2009 Accepted 9 October 2009 Available online 4 November 2009 Keywords: Peptide analysis Peptide purification Preparative HPLC of acido-basic compounds Multi pKa compounds RPLC Adsorption mechanism XBridge-C18 Methanol–water mobile phases S pH W Ionic strength Buffer adsorption System peaks Phenylalanine-alanine
a b s t r a c t We recorded simultaneously the SW pH and the elution band profiles of phenylalanine-alanine (Phe-Ala, S pKa,1 = 3.6, SW pKa,2 = 7.5) on a column packed with C18 -bonded organic (ethyl)/inorganic (silica) hybrid W porous particles (BEH), eluted with a series of buffered methanol–water mobile phases (20/80, v/v). Using different buffers, the SW pHs of the mobile phases was set to values between 2.5 and 9.8. The ionic strength of the mobile phases was kept constant at 20 mM. The injected solutions were prepared with Phe-Ala as its zwitterionic species. Analytical to overloaded sample sizes were used, under strongly or insufficiently buffered conditions. With insufficiently buffered mobile phases, the experimental results show significant variations of SW pH during band elution, which proves a significant change of the concentration ratio of the different sample species during elution. This result confirms and validates our earlier general adsorption isotherm model of acido-basic compounds in RPLC. The SW pH record shows also the system peak of the buffer component, consecutive to sample injection. The phosphate II and carbonate II ions are adsorbed and may compete with the peptide for adsorption onto the active sites of BEH-C18 . This suggests refinements of the general adsorption isotherm model, depending on the nature of the buffer components, including competitive adsorption of the sample and the buffer species. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Investigations of retention and adsorption mechanisms of acidobasic compounds in RPLC are attracting much attention [1–6], the goal of which is the optimization of optimum experimental conditions for the separation and purification of pharmaceutical and biological compounds. Under linear conditions, the pH of the mobile phase, usually an aqueous mixture of methanol or acetonitrile, imposes the concentration ratio of the compounds involved. Several general retention models have been proposed to account for the retention pattern of a broad range of acids and bases over a pH range from 2 to 11 and for large variation in the mobile phase composition [4,7,8]. It is systematically observed that, under moderately overloaded conditions, the peak shapes of the ionizable species of the sample components tail more strongly than those of their conjugated neu-
∗ Corresponding author. Fax: +1 865 974 2667. E-mail address: [email protected] (G. Guiochon). 0021-9673/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2009.10.026
tral species. Hence, acidic and basic compounds are preferentially analyzed at low and high pHs, respectively, in order to generate more symmetrical and narrower elution peaks [5]. Under severe overloaded conditions such as in preparative chromatography, the sample concentration may exceed the buffer concentration and the buffer becomes too weak to set the pH of the mobile phase. Also, if the buffer is a strong ion-pairing agent, large variations in the retention factor may eventually take place because the co-ion which accompanies the sample is different at low and at high buffer concentrations. The prediction of the band profiles eluted in insufficiently buffered mobile phases requires the knowledge of the individual adsorption isotherms of each sample species and that of their local concentrations, anywhere at any time. These different species are constantly in equilibrium but their relative concentrations depends on the total concentration of the sample component, on its pKa , and on that of the buffer. Equilibration between these species in both the mobile and the adsorbed phases is assumed to be infinitely fast, so the total adsorbed amount of acido-basic compounds is directly obtained from the molar fractions of each species in the bulk and the individual adsorption isotherms.
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
A consistent but arguable thermodynamic adsorption model was proposed, based on the analysis of the adsorption energy distributions (AEDs) of ionizable and neutral compounds in RPLC [9]. Neutral sample molecules are weakly adsorbed on a large number of adsorption sites while ionizable molecules are strongly adsorbed onto very few active sites [10]. Because the total concentration of each sample component varies during its elution, so does the relative concentration of each one of its species in a weakly buffered mobile phase. These concentrations at any position and time inside the column are obtained by solving the thermodynamics problem in the bulk solution [11,12]. The extended Debye–Hückel method was used to estimate the activity coefficients of the ionizable compounds [13]. Finally, the equilibrium-dispersive model of chromatography provides the elution profiles with the total concentration of the compound as the sum of the concentrations of all its species. The calculated band profiles agree well with the observed ones, giving unusual anti-langmuirian and S-shaped profiles in the cases of aniline and 2-phenylbutyric acid [12], although the adsorption isotherms of both the acid and the base species of these two compounds are langmuirian. The pH dependency of the shape of the overloaded band profiles of compounds with two pKa s (hence, a total of three species) such as o-phthalic acid and nicotinic acid was also well accounted for by this general adsorption model. Finally, the same approach was applied for the prediction of the elution band profile of the peptide Phe-Ala (which is a biprotic compound with SW pKa,1 = 3.6 and SW pKa,2 = 7.5) under insufficiently buffered mobile phases [14]. In this case, only qualitative agreement was observed between experimental and calculated band profiles with the basic mobile phases made with either ammonia or carbonate buffers. This implied that the buffer component was not an inert compound and eventually adsorbs onto the adsorbent, competing with at least one species of the compound studied. The arguable part of this model is our identification of the weak neutral sites with silica-bonded C18 chains and of the few high energy active sites with the accessible silanol groups amidst the C18 bonded layer. Because the method and the model used are thermodynamic, they inform on the energy of the adsorption sites involved and on their density but not on their microscopic or chemical structure. In this work, we investigated in more detail the complex adsorption mechanism of the di-peptide Phe-Ala (SW pKa,1 = 3.6 and SW pKa,2 = 7.5) onto BEH-C18 by recording simultaneously the overloaded elution band profiles (measured using UV detection) and the eluent SW pH (measured with a micro-pH electrode with a low cell volume). The main goal was to validate the model proposed for the adsorption of acido-basic compounds in RPLC and to assess its limit of applications. The measurements were performed under controlled and uncontrolled pH conditions, with six different mobile phase SW pH between 2.4 and 9.8, and the results are compared. Small and large sample volumes were injected in order to investigate the impact of the adsorption of the buffer components on the band profiles of the small peptide. 2. Experimental 2.1. Chemicals The mobile phase was a solution of methanol and water (20/80, v/v), both HPLC grade from Fisher Scientific (Fair Lawn, NJ, USA). The mobile phase was filtered before use on a surfactant-free cellulose acetate filter membrane, 0.2 m pore size (Suwannee, GA, USA). The peptides used were the dipeptide phenylalanine-alanine (Phe-Ala) (Fig. 1) and phenylalanine-valine (Phe-Val). Phosphoric acid (85% H3 PO4 ), potassium dihydrogenphosphate (KH2 PO4 ), dipotassium hydrogenphosphate (K2 HPO4 ), formic acid (98% HCOOH), sodium formate (NaHCOO), glacial acetic acid (> 99.5%
8875
Fig. 1. Chemical structures and conjugated species of the dipeptide phenylalaninealanine used in this study. The two SW pKa values indicated were determined from the SW pHs measured in this work for 44 mM solutions of Phe-Ala in different aqueous buffers containing 20% methanol in volume.
CH3 COOH), sodium acetate (NaCH3 COO), ammonium chloride (NH4 Cl), ammonium hydroxide (30% NH3 ), potassium hydrogencarbonate (99% KHCO3 ), and disodium carbonate (Na2 CO3 ), all from Fisher Scientific, were used to prepare the buffer solutions (see Table 1 for the SW pH of the buffers). 2.2. Materials An Agilent 1090 liquid chromatograph (Agilent, Palo Alto, CA, USA) was used to perform the measurements. This instrument includes a ternary solvent delivery system, an auto-sampler with a
8876
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Table 1 Buffer solutions in a water/methanol (80/20, v/v) mixture. Ionic strength I=20 mM. The measured 44 mM solution of dipeptide Phe-Ala (the concentrations refer to the water/methanol volume). Buffer number I II III IV V VI
Name Phosphate I Formate Acetate Phosphate II Ammonia Carbonate II
Acid [M]
Base [M]
H3 PO4 0.020 HCOOH 0.020 CH3 COOH 0.020 KH2 PO4 0.005 NH4 Cl 0.020 KHCO3 0.005
are given for the neat buffer solution, the 2 mM solution, and the
Buffer Exp.
K2 HPO4 0.020 HCOONa 0.020 CH3 COONa 0.020 K2 HPO4 0.005 NH4 OH 0.020 Na2 CO3 0.005
250 L sample loop, a diode-array UV detector (cell volume 1.7 L, sampling rate 25 Hz), a column oven, and a data station running the HP data software. From the needle seat to the column inlet and from the column outlet to the detector cell, the total extra-column volume of the instrument is 45 L, measured as the apparent hold-up volume of a zero-volume union connector in place of the column. The chromatographic column (150 mm × 4.6 mm packed with XBridge-C18 particles, average size 3.5 m, average pore size 136 Å, specific surface area 172 m2 /g, bonding density 16.95% C, surface coverage 3.13 mol/m2 ) was a generous gift from Waters (Milford, MA, USA). The total porosity t = 0.637 of this column was estimated from pycnometric measurements (made using methanol and dichloromethane), giving a column hold-up volume of 1.588 mL [13]. The packing material was endcapped according to a proprietary process. Table 2 summarizes the manufacturer’s data. The SW pHs of the mobile phases and sample solutions were measured with a flow through micro-pH-electrode (Lazar Research Laboratories Inc.), calibrated in pure water with two standard aqueous solutions at pH = 4.00 and 9.00. The cell, through which the eluent circulates and is in contact with the electrode, has a volume of 20 L. The plots of the eluent pH versus the elution volume were independent of the flow rate when this flow rate was set below 0.25 mL/min, demonstrating a sufficiently fast response of the pH electrode. The pH data were recorded every second.
S pHs W
2 mM, Phe-Ala, Sample Exp.
44 mM, Phe-Ala, Sample Exp.
S pH W
S pH W
S pH W
2.53 3.89 4.83 6.91 9.40 9.80
2.58 3.91 4.84 6.75 9.17 9.25
3.71 4.24 4.89 5.35 6.71 6.87
volumetric glasses. So, the largest injected sample concentration of Phe-Ala was 10 g/L. 2.4. Injections The injected sample volumes were 30 and/or 180 L, unless indicated otherwise. The flow rate was set at 0.5 mL/min (inlet column pressure close to 140 bar) and the temperature was maintained at 295 ± 1 K by the air-conditioning system of the laboratory. 3. Results and discussion In the next section, we discuss the relationship between the profile recorded with the flow-through micro-pH-electrode and the experimental UV-band profiles of the dipeptide Phe-Ala. In all cases, the solution injected was prepared by dissolving the zwitterionic species of Phe-Ala directly into the six buffer solutions. The ionic strength of all the buffer solutions was equal to 20 mM. Two different conditions are examined: (1) the concentration of Phe-Ala in the sample is smaller than the buffer concentration (2 mM Phe-Ala) and (2) its concentration is larger than that of the buffer (44 mM Phe-Ala). We analyze the results one by one for each buffered mobile phase at SW pH = 2.53 (phosphate I), 3.89 (formate), 4.83 (acetate), 6.90 (phosphate II), 9.03 (amoniac) and 9.80 (carbonate II). The aqueous solutions contain all 20% methanol in volume. The flow rate was constant at 0.50 mL/min in all the experiments.
S pH W
2.3. Buffer and sample preparation Six buffer solutions (total volume 150 mL each) were prepared by dissolving the appropriate amounts of the acid and its conjugated base in order to obtain SW pH values increasing from about 2 to 10 at a constant ionic strength of 20 mM. All the measured SW pH of the buffers are listed in Table 1. For each buffer solution, 2 mM and 44 mM solutions of Phe-Ala were prepared by dissolving 4.5 and 100 mg, respectively, in 10 mL Table 2 Physico-chemical properties of the column used in this work according to the manufacturer. Neat silica
Xbridge 3.5 m
Particle size [m] Pore diameter [Å] Surface area [m2 /g]
3.5 136 172
Bonded phase
Xbridge -C1 8
Total carbon [%] Surface coverage [mol/m2 ] Endcapping
16.95 3.13 Yes
Packed column Serial number Dimension (mm × mm) Total porosity a a
128B3812113683 4.6× 150 0.637
Measured by pycnometry (CH3 OH/CH2 Cl2 ) at 295 K.
3.1. Buffer I: Phosphate I, SW pH = 2.53 At infinite dilution, Phe-Ala is mostly present in solution as its cationic species, + Phe-Ala. The SW pH of the 2 mM solution barely differs from that of the neat buffer, at 2.58 rather than 2.53. When 180 L of this solution is injected into the BEH-C18 column, the S pH remains almost constant during elution, except at the holdW up column time (starting at t = 2 × 1.588 = 3.18 min) and during the elution of Phe-Ala, from 6.7 min to 8.6 min (Fig. 2). These pH perturbations are very small (< 0.03 pH unit). The first perturbation is related to the difference in the concentration ratio of the phosphate buffer species H3 PO4 and H2 PO4 − caused by the slight difference between the SW pH of the injected solution and that of the neat buffer. A perturbation of the equilibrium concentration of phosphate takes place and the usual system peak of the additive compound phosphate is detected [15]. The second perturbation is directly related to the change in the concentration ratio of + Phe-Ala and + Phe-Ala− during the elution of its band. The very slight increase of the SW pH (< 0.1 pH unit) suggests that the molar fraction of + Phe-Ala− slightly increases. This was expected due to the slightly larger SW pH of the 2 mM sample solution. When 180 L of a 44 mM solution of Phe-Ala is injected, the amplitude of these two pH perturbations are largely increased because the SW pH of the injected sample (3.71) exceeds that of the neat buffer by more than 1 pH unit. Accordingly, the initial concentration ratio of the H3 PO4 and H2 PO4 − species is smaller than that of the neat buffer. At the hold-up time, the SW pH of the elu-
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 2. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid red line, right scale) of Phe-Ala eluted from the 150 mm × 4.6 mm I.D. BEH-C18 column with a 20 mM phosphate I buffer (SW pH = 2.53) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. The SW pH of the injected peptide solutions are given in Table 1. Note the two positive pH disturbances, one before, the second during elution of the peptide. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) + Phe-Ala
and ent rises to 3.0. Similarly, the concentration ratio of + Phe-Ala− increases. As the total Phe-Ala concentration increases in the bulk mobile phase, the molar fraction of the zwitterion increases, explaining the second increase of the pH observed during the elution of the peptide. This demonstrates that the front spike (high total concentration) of the chromatogram is richer in zwitterions than its tail (small total concentrations). This last observation is very important because it comforts the hypothesis made in designing the general adsorption isotherm model of acido-basic compounds and predicting their band profiles in insufficiently buffered mobile phases [11,12,9,13,16,14]. The deformation of the band profiles originates from (1) the difference in the adsorption isotherms of the different species onto the adsorbent and (2) the variation of their respective molar fractions in the mobile phase, which depends on the total sample concentration. 3.2. Buffer II: Formate I, SW pH = 3.89 At a buffer SW pH of 3.89, the zwitterion of Phe-Ala predominates. Accordingly, the SW pH of the 2 mM and 44 mM sample solutions of Phe-Ala should not differ much from 3.89. Actually, they are 3.91 and 4.24, respectively.
8877
Fig. 3. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid red line, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM formate buffer (SW pH = 3.89) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the decrease in amplitude of the pH disturbances compared to those observed in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Fig. 3A and B shows the variation of SW pH. The system peak of the formate buffer is clearly detectable at the hold-up time. The SW pH increases by no more than 0.02 and 0.15 pH unit when 2 mM and 44 mM sample concentrations are injected, respectively. Interestingly, the system peak of the buffer formate tails (Fig. 3B), which suggests that the anion HCOO− is retained on the BEH-C18 adsorbent. This is consistent with the adsorption properties of this column [16,14]. Note that the buffer perturbation signal in Fig. 2B does not tail, which suggests that the phosphate buffer is only slightly retained. The system peak will interfere with the peptide band, leading to more complex SW pH profiles during the elution of the peptide. At high concentrations, the band profile corresponds to an S-shaped isotherm because the adsorption isotherm of the zwitterion alone is anti-langmuirian. At low concentrations, the band tails because the cation has a Langmuir adsorption isotherm [14]. 3.3. Buffer III: Acetate I, SW pH = 4.83 The zwitterion + Phe-Ala− predominates also at the SW pH of the acetate buffer. The SW pH change upon dissolving 2 mM and 44 mM Phe-Ala in this buffer are the smallest (see Table 1). Fig. 4A and B shows the corresponding UV and pH profiles. The retention of the zwitterion is so small (minimum) and the injected volume so large (180 L) that the system peak of acetate and the band profile of the
8878
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 4. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM acetate buffer (SW pH = 4.83) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the negligible amplitude of the pH disturbances.
peptide eventually interfere. The system peak of acetate is barely positive. The passage of this compound is accompanied by a very small but detectable pH decrease similar to that observed with the buffer formate. Fig. 4B and the fronting of the band of + Phe-Ala− confirms the anti-langmuirian shape of the adsorption isotherm of the zwitterion. This behavior is most probably related to the ionic interactions between the positive and the negative charges of two zwitterionic molecules. Overall, the SW pH of the eluent during elution is the least affected by the injection of concentrated solution of the peptide since the Phe-Ala dissolved does not react with the neat buffer solution. We simply observe the band profile of a single species, the zwitterion, with no significant change in the molar fractions of the other ionic species in solution.
Fig. 5. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM phosphate II buffer (SW pH = 6.90) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 30 L. (C) Peptide concentration: 44 mM, volume injected: 180 L. Note the interference between the two pH disturbances.
3.4. Buffer IV: Phosphate II, SW pH = 6.91 Dilute solutions (2 mM) of the peptide in this buffer contain both the zwitterion and the anion Phe-Ala− . The SW pH of the 2 mM and 44 mM sample solutions are smaller than that of the neat buffer because the zwitterion can react with the phosphate buffer and displaces the initial H2 PO4 − /HPO4 2− equilibrium toward the formation of more H2 PO4 − anions. Accordingly, the SW pH profiles in Fig. 5A–C shows a negative peak, the intensity of which grows with increasing sample volume and/or concentration. Inter-
estingly, the eluent SW pH increases when the front shock layer of the peptide is eluted. We could have expected the opposite because the ratio of the concentrations of the zwitterions and of the anions increases with increasing total sample concentrations (see Table 1). This shows that the general adsorption model proposed for Phe-Ala [14] has limits. There is a serious disagreement between the experimental band profiles shown in Fig. 5C and those
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
8879
single-component adsorption behavior compete for adsorption [15]. 3.5. Buffer V: Ammonia, SW pH = 9.40 The mobile phase SW pH was increased further to 9.40 by using the ammonia buffer. In this case, the pH is well above the second S pK of the peptide (7.5) and the dissolved neutral compound Phea W Ala reacts strongly with the buffer to be quantitatively transformed into the anion Phe-Ala− , unless the initial sample concentration largely exceeds the buffer capacity. Fig. 6A–C shows the corresponding SW pH and band profiles recorded for the three different sample loadings used. The size of the system peak of NH4 + /NH3 increases with increasing sample size because the dissolution of 2 mM and 44 mM of Phe-Ala in the ammonia buffer causes a decrease of the buffer SW pH (see Table 1) and an increase of the concentration ratio of NH4 + to NH3 . Due to the ammonia and ammonium ion lack of chromophores, the relative change in the concentration ratio of the two buffer species is not detected. Would the injection volume exceed 180 L, the system peak of the ammonia buffer would interfere with the band profile of the peptide. Note that the SW pH slightly decreases during elution of the peptide, demonstrating the progressive increase of the ratio of the concentration of + Phe-Ala− to that of Phe-Ala− when the total peptide concentration increases. The SW pH
Fig. 6. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM ammonia buffer (SW pH = 9.30) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 30 L. (C) Peptide concentration: 44 mM, volume injected: 180 L. Note the small and negative amplitude of the pH disturbance during the sample elution.
calculated in the intermediate concentrations region. More specifically, the band profiles calculated with the phosphate II buffer should have corresponded to a front spike with an S-shaped profile. Clearly, this is not the experimental profile recorded. This suggests that the competitive adsorption model involving the zwitterions (anti-Langmuir behavior) and the anions (Langmuir behavior) that we used in the calculations is too simple. This situation is expected when compounds having so different
Fig. 7. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of Phe-Ala eluted from the same column as in Fig. 2, with a 20 mM carbonate II buffer (SW pH = 9.80) in a methanol:water mixture (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. (A) Peptide concentration in the injected sample: 2 mM, volume injected: 180 L. (B) Peptide concentration: 44 mM, volume injected: 180 L. Note the absence of pH disturbance during the sample elution.
8880
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Fig. 8. Simultaneous records of the SW pH (thick solid line, left scale) and the overloaded band profiles (thin solid right, right scale) of a mixture of Phe-Ala and Phe-Val eluted from the same column as in Fig. 2, with four different 20 mM buffers in a methanol:water solution (20/80, v/v). T = 295 K. Flow rate: 0.5 mL/min. The sample concentrations are 10 g/L for each compound. The sample volume injected is 250 L (injection time, 30 s). (A) Phosphate I buffer (SW pH = 2.4). (B) Acetate buffer (SW pH = 5.1). (C) Phosphate II buffer (SW pH = 7.3). (D) Carbonate II buffer (SW pH = 10.3).
profile mirrors closely the UV profile and shows the direct relationship between the total peptide concentration and the solution SW pH. This confirms the theoretical assumption made in the derivation of the overloaded band profiles of acido-basic compounds [11,13] when recalculating at each step of the simulation process (at time t and at the axial coordinate z) the molar fraction of each sample species as a function of the total sample concentration.
3.6. Buffer VI: Carbonate II, SW pH = 9.80 The SW pH of the neat carbonate buffer is slightly larger than that of the ammonia buffer. Fig. 7A and B shows that the system peaks of the carbonate buffer are more severely distorted and more strongly retained than those of the ammonia buffer (Fig. 6A–C). As the bulk concentration of hydrogencarbonate increases and that of bicarbonate decreases, the amount of buffer adsorbed increases. We observed that the pH is perturbed from t = 3.2 min to t = 6 min. If the buffer species were both unretained on the stationary phase, the pH perturbation would start at 3.2 min and end at 3.6 min since the injection lasts only 0.36 min. A comparison of Figs. 6C and 7B suggests that the carbonate buffer is more retained than the ammo-
nia buffer, which is consistent with the fact that BEH-C18 retains singly and negatively charged species as previously reported for nicotinate [16] and Phe-Ala− [14]. Since the SW pH largely decreases before elution of the peptide and the concentration ratio of the monovalent bicarbonate anion and the divalent carbonate anion increases in the mobile phase, the retention of the buffer increases, showing that HCO3 − is more retained than CO3 2− . The interference between the system peak of the carbonate buffer and the elution band profile of Phe-Ala clearly occurs all along the chromatographic column. It can seriously affect the band shape of the main component [17–20]. In Fig. 7B, the elution front of the peptide begins before the mobile phase SW pH has completely returned to its initial value of 9.8. Note that there is no obvious discontinuity in the SW pH profile at that moment suggesting that the peptide was almost entirely converted into its anion at elution. The variation of the SW pH of the eluent is solely due to the adsorption behavior of the monohydrogencarbonate/bicarbonate buffer onto BEH-C18 and to the disturbance of the initial equilibrium due to injection of the peptide, the SW pH of which is significantly smaller than that of the neat buffer (6.9 versus 9.8). This contrasts with the result shown in Fig. 6B and C, where the elution of the dipeptide is accompanied by a change of the eluent SW pH.
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
In conclusion, one cannot accurately predict the large overloaded band profiles of Phe-Ala in presence of the carbonate buffer because the adsorption of Phe-Ala− eventually depends on the concentration of the buffer species HCO3 − , which continuously varies during elution of the peptide, from the inlet to the outlet of the column. Differences between calculated and experimental band profiles were observed under such conditions [14]. 3.7. Application to the purification of mixtures of two small peptides The profiles of the SW pH and the peptide concentrations during the elution of a 250 L sample of a mixture of Phe-Ala and PheVal (10 g/L each) were recorded. Fig. 8A–D shows these profiles at four different SW pHs, 2.43, 5.10, 7.30, and 10.3 of the mobile phase. The ionic strength of the buffer was kept constant (20 mM). Except at SW pH = 5.10, at which the presence of either peptide does not affect much the mobile phase pH, the elution of the first eluted compound (Phe-Ala) strongly interferes with the system peak of the buffer additive. Although the two peptides differ only by one methylene group in the side-chain, they are well resolved and the concentration ratio of the two species of each peptide at elution is different. Nevertheless, the complex profile of the two-component band can be interpreted based on the adsorption behavior of the two single peptides. At SW pH = 2.43, the system peak of the formate buffer is spread over a large fraction of the chromatogram. The eluent pH-spike following elution of the first compound (Phe-Ala) is much higher than that accompanying the front shock of the second one (PheVal) (Fig. 8A). The band profile of Phe-Ala is typical of those of compounds having an S-shaped isotherm, showing that a significant fraction of Phe-Ala is eluted as the zwitterion. In contrast, the elution profile of Phe-Val corresponds to that of a compound having a langmuirian isotherm, hence for which the relative abundance of the cation in the bulk during its elution is larger than that of the zwitterion. This illustrates the importance of the interference between the bands of the additive and of the main components when the mobile phase is insufficiently buffered and large volumes of sample solution are injected, and the injection last for a time that is not negligible compared to the hold-up time. This interference can strongly affect the band shapes of the main components, to an extent that depends on the injected volume. At SW pH = 5.10 (Fig. 8B), the presence of large concentrations of Phe-Ala and Phe-Val do not affect the SW pH of the neat buffer. The two compounds are well resolved and both band shapes are anti-langmuirian, as expected from their individual simple Moreau adsorption isotherms [14]. The adsorbate–adsorbate interactions between the adsorbed zwitterions are stronger with Phe-Val than with Phe-Ala (see the enhanced peak fronting in Fig. 8B). At SW pH = 7.30 (Fig. 8C), the system peak of the phosphate buffer interferes with the band of Phe-Ala during nearly its whole elution time. The SW pH drops by almost 1 pH unit, to values at which this peptide is entirely under the zwitterion state. This explains the anti-langmuirian shape observed in Fig. 8C, between 4.5 min and 6.0 min. Beyond this time, the SW pH returns to more neutral values (higher than 7) and the anion concentration in the bulk is no longer negligible. Then the band of Phe-Ala ends in a langmuirian shape that is characteristic of the adsorption behavior of ionic species. The elution of Phe-Val is accompanied with a slight increase in the buffer pH, demonstrating that, in contrast to PheAla, the anion concentration is always significant during its elution. Note that the spike observed in the chromatogram is an artefact of the UV detector and is not due to a sharp increase of the peptide concentration.
8881
Finally, at SW pH = 10.3 (Fig. 8D), the system peak of the buffer carbonate interferes with the elution bands of both compounds, affecting seriously the shape of the least retained peptide (Fig. 8D). The SW pH drops to about 7, at which pH the zwitterions + Phe-Ala− concentration in the mobile phase is significant. This explains the large increase of the Phe-Ala concentration at about 5 min. Beyond this time, the SW pH returns to a basic value and both bands tail strongly, according to the adsorption behavior of the anions.
4. Conclusion The simultaneous records of the SW pH and the elution profiles of overloaded bands of Phe-Ala permits the assessment of the causes of the variations of the SW pH during elution of this peptide with an insufficiently buffered mobile phase. From a general point of view, we conclude that the initial equilibrium SW pH imposed by the buffered mobile phase is disturbed twice after injection of the peptide sample 1. The first SW pH disturbance is directly related to the system peak of the buffer species, which is not retained or only weakly so. Although often undetected by the UV detector, this system peak is systematically detected by the pH micro-electrode when the mobile phase is insufficiently buffered. This is due to the reaction of the peptide with the neat buffer. The concentration ratio of the acidic and the basic buffer species is significantly different in the presence of the peptide and in the neat buffer. This causes a perturbation of the bulk/adsorbent equilibrium, which propagates along the column. The shape of this perturbation depends on the adsorption behavior of the buffer species on the solid adsorbent. 2. The second SW pH disturbance is caused by the elution of the sample component. The ratio of the concentrations of the acidic and basic peptide species depends on the total peptide concentration. The analysis of the SW pH-concentration profiles during elution of the peptide confirms an assumption made in the calculation of the overloaded band profiles of acido-basic compounds when using insufficiently buffered mobile phases. In the calculation of the total amount adsorbed the molar fraction of each species in the bulk phase is computed as a function of the total bulk concentration at each time t and for each position z along the column. These molar fractions change during elution, causing a parallel drift of the mobile phase SW pH. The validity of this assumption is unambiguously confirmed by our results. The limit of validity of our general adsorption isotherm of acidobasic compounds is due to its neglecting the adsorption of the buffer components. When the SW pH of the injected sample significantly differs from that of the buffered mobile phase, the buffer concentration changes during sample elution. When the retention of the buffer species is significant compared to that of the sample and when the injected volume is large, the two SW pH disturbances or composition waves of the buffer and the sample species interfere. This leads to more complex band profiles than could initially be expected. The sample is not eluted at a well-defined buffer composition and the general adsorption isotherm model previously reported is only an approximation of the true adsorption behavior, for the neglect of the competition with the buffer components. Agreement between calculated and experimental band profiles is merely qualitative but major improvement would be made by accounting for the competitive adsorption between the buffer and the sample components.
8882
F. Gritti, G. Guiochon / J. Chromatogr. A 1216 (2009) 8874–8882
Acknowledgements This work was supported in part by grant CHE-06-08659 of the National Science Foundation and by the cooperative agreement between the University of Tennessee and the Oak Ridge National Laboratory. We thank Uwe Neue (Waters, Milford, MA) for the generous gift of the new hybrid BEH-C18 packing material and for informative discussions. References [1] N.H. Davies, M.R. Euerby, D.V. McCalley, J. Chromatogr. A 1119 (2006) 11. [2] D.V. McCalley, J. Chromatogr. A 1075 (2005) 57. [3] N.H. Davies, M.R. Euerby, D.V. McCalley, J. Chromatogr. A 1178 (2008) 71. [4] U.D. Neue, C.H. Phoebe, K. Tran, Y.-F. Cheng, Z. Lu, J. Chromatogr. A 925 (2001) 49.
[5] J. Samuelsson, A. Franz, B.J. Stanley, T. Fornstedt, J. Chromatogr. A 1163 (2007) 177. [6] L. Pan, R. LoBrutto, Y.V. Kazakevich, R. Thompson, J. Chromatogr. A 1049 (2004) 63. [7] U.D. Neue, A. Méndez, K. Tran, P.W. Carr, J. Chromatogr. A 1063 (2005) 35. [8] Y.V. Kazakevich, J. Chromatogr. A 1126 (2006) 232. [9] F. Gritti, G. Guiochon, Russ. Bull. 71 (2009) 1. [10] F. Gritti, G. Guiochon, J. Chromatogr. A 1099 (2005) 1. [11] F. Gritti, G. Guiochon, J. Sep. Sci. 31 (2008) 3657. [12] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 63. [13] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 3175. [14] F. Gritti, G. Guiochon, Anal. Chem., accepted for publication. [15] G. Guiochon, S.G. Shirazi, A. Felinger, A.M. Katti, Fundamentals of Preparative and Non-linear Chromatography, Academic Press, Boston, MA, 2006. [16] F. Gritti, G. Guiochon, J. Chromatogr. A 1216 (2009) 6917. [17] T. Fornstedt, G. Guiochon, Anal. Chem. 66 (1994) 2116. [18] T. Fornstedt, G. Guiochon, Anal. Chem. 66 (1994) 2686. [19] P. Sajonz, Y. Tong, G. Zhong, T. Fornstedt, G. Guiochon, J. Chromatogr. A 734 (1996) 75. [20] G. Zhong, T. Fornstedt, G. Guiochon, J. Chromatogr. A 734 (1996) 63.