Anal. Chem. 2001, 73, 2027-2033
Aqueous Chromatography Utilizing pH-/ Temperature-Responsive Polymer Stationary Phases To Separate Ionic Bioactive Compounds Jun Kobayashi,† Akihiko Kikuchi,‡ Kiyotaka Sakai,† and Teruo Okano*,‡
Department of Applied Chemistry, Faculty of Science and Engineering, Waseda University, 3-4-1 Ohkubo, Shinjuku, Tokyo 169-8555, Japan, and Institute of Biomedical Engineering, Tokyo Women’s Medical University, 8-1 Kawadacho, Shinjuku, Tokyo 162-8666, Japan
Cross-linked poly(N-isopropylacrylamide-co-acrylic acid) (poly(IPAAm-co-AAc))-grafted silica bead surfaces were prepared and applied as new column matrix materials that exploit temperature-responsive anionic chromatography to separate basic bioactive compounds, specifically catecholamine derivatives, in aqueous mobile phases. Since poly(IPAAm-co-AAc) has a well-known temperatureresponsive phase transition and apparent pKa shift, polymer-grafted silica bead surfaces are expected to exhibit simultaneous hydrophilic/hydrophobic and charge density alterations under thermal stimuli. Elution behavior of catecholamine derivatives from a copolymer-modified bead packed column was monitored using aqueous mobile-phase HPLC under varying temperature and pH. Catecholamine derivatives had higher retention times on poly(IPAAm-co-AAc) columns at higher pH in comparison with those on noncharged PIPAAm reference columns, suggesting an electrostatic interaction as a separation mode. Temperature also affected the retention behavior of catecholamine derivatives. Optimal separation of four catecholamine derivatives was achieved at elevated temperature, 50 °C, and at pH 7.0. This is due to the increased hydrophobicity of the stationary phase as evidenced by the elution of a nonionic hydrophobic steroid. From these results, mutual influences of both electrostatic and hydrophobic interactions between basic catecholamine derivatives and pH-/temperature-responsive surfaces are noted. Consequently, elution of weakly charged bioactive compounds is readily regulated through the modulation of stationary-phase thermoresponsive hydrophilic/hydrophobic and charge density changes. A wide variety of natural bioactive compounds, including peptides, proteins, and nucleotides, are now routinely produced in large quantity for applications in many biomedical and biotechnology fields. Reversed-phase chromatography (RPC), in which the interaction (partitioning) between stationary phase and solutes is controlled by changing the polarity of the mobile phase, is * Corresponding author: (phone) +81-3-3353-8111 Ext. 30233; (fax) +81-33359-6046; (e-mail)
[email protected]. † Waseda University. ‡ Tokyo Women’s Medical University. 10.1021/ac0013507 CCC: $20.00 Published on Web 03/27/2001
© 2001 American Chemical Society
commonly used as an effective separation tool, particularly in pharmaceutics and biochemistry. RPC, however, sometimes has limited applications because solute bioactivity (particularly for proteins and peptides) is frequently compromised by the use of organic solvent mobile phases. Development of effective separation methods for bioactive compounds using only aqueous mobile phases is an important new technology need. We have previously applied the thermoresponsive polymer; poly(N-isopropylacrylamide) (PIPAAm), as a new stationary column phase for separation of bioactive compounds in aqueous media.1-4 PIPAAm exhibits thermally reversible soluble/insoluble changes across a lower critical solution temperature (LCST) at 32 °C in aqueous solution.5,6 Three-dimensional cross-linked PIPAAm gels show reversible swelling and collapse changes in aqueous solution in response to temperature. These gels, therefore, are now commonly applied to control drug release rates by using external temperature changes.7,8 In addition, we have developed thermoresponsive PIPAAm-modified surfaces by chemical immobilization or electron beam irradiation, demonstrating both control of blood platelet activation/inactivation9-11 and noninvasive recovery of cultured cells and cell sheets for tissue engineering applications.12-15 Furthermore, we previously demonstrated a new form of hydrophobic chromatography using (1) Kanazawa, H.; Yamamoto, K.; Matsushima, Y.; Takai, N.; Kikuchi, A.; Sakurai, Y.; Okano, T. Anal. Chem. 1996, 68, 100-105. (2) Kanazawa, H.; Yamamoto, K.; Kashiwase, Y.; Matsushima, Y.; Takai, N.; Kikuchi, A.; Sakurai, Y.; Okano, T. J. Pharm. Biomed. Anal. 1997, 15, 15451550. (3) Kanazawa, H.; Kashiwase, Y.; Yamamoto, K.; Matsushima, Y.; Kikuchi, A.; Sakurai, Y.; Okano, T. Anal. Chem. 1997, 69, 823-830. (4) Yakushiji, T.; Sakai, K.; Kikuchi, A.; Aoyagi, T.; Sakurai, Y.; Okano, T. Anal. Chem. 1999, 71, 1125-1130. (5) Heskins, M.; Guillet, J. E.; James, E. J. Macromol. Sci. Chem. 1968, A2, 1441-1445. (6) Bae, Y. H.; Okano, T.; Kim, S. W. J. Polym. Sci. B: Polym. Phys. 1990, 28, 923-936. (7) Yoshida, R.; Sakai, K.; Okano, T.; Sakurai, Y.; Bae, Y. H.; Kim, S. W. J. Biomater. Sci. Polym. Ed. 1991, 3, 155-162. (8) Yoshida, R.; Sakai, K.; Okano, T.; Sakurai, Y. J. Biomater. Sci. Polym. Ed. 1992, 3, 243-252. (9) Takei, Y. G.; Aoki, T.; Sanui, K.; Ogata, N.; Sakurai, Y.; Okano, T. Biomaterials 1995, 16, 667-673. (10) Okano, T.; Kikuchi, A.; Sakurai, Y.; Takei, Y. G.; Ogata, N. J. Controlled Release 1995, 36, 125-133. (11) Uchida, K.; Sakai, K.; Ito, E.; Kwon, O. H.; Kikuchi, A.; Yamato, M.; Okano, T. Biomaterials 2000, 21, 923-929. (12) Yamada, N.; Okano, T.; Sakai, H.; Karikusa, F.; Sawasaki, Y.; Sakurai, Y. Makromol. Chem., Rapid Commun. 1990, 11, 571-576.
Analytical Chemistry, Vol. 73, No. 9, May 1, 2001 2027
PIPAAm-grafted HPLC columns to separate bioactive steroids using water as a mobile phase.1-4 While Gewehr et al.16 and Hosoya et al.17 also independently reported chromatography systems that utilize PIPAAm-modified surfaces, these systems utilized a coil-globule transition of PIPAAm molecules on porous matrix beads to thermally control pore size in permeation chromatography. By contrast, solute separation is driven mainly by hydrophobic partitioning changes with PIPAAm-modified surfaces in our system, with analyte retention times dramatically altered using temperature changes. Our system distinguishes itself from conventional RPC from the standpoints of retaining solute bioactivity and for environmental reasons (e.g., reduced organic solvent use/disposal) because of the exclusive use of aqueous mobile phases. Moreover, transition temperatures for hydrophilic/hydrophobic stationaryphase surface alterations are modulated by introduction of the comonomer:9 interactions between hydrophobic molecules and grafted copolymer surfaces was enhanced by incorporating the hydrophobic comonomer, butyl methacrylate (BMA).3 Polymer graft conformation and PIPAAm surface density greatly affect thermoresponsive wettability changes of PIPAAm-modified surfaces and subsequent interaction with hydrophobic steroids.4,18 Through these studies, we have found that component chemistry and graft architecture of thermoresponsive polymers on surfaces are great influences to modulate surface properties. The primary purpose of the present contribution is to describe the design of new stationary phases for effective separation of cationic bioactive compounds. As most bioactive compounds contain ionizable as well as hydrophobic residues, introduction of charged groups on the column surface produces electrostatic interactions with solutes. PIPAAm derivatives containing acrylic acid (AAc) anionic comonomer moieties produces an alterable stationary phase with both thermally regulated hydrophobicity and charge density for separation of basic bioactive compounds, specifically the neurotransmitter, catecholamine. This temperatureresponsive polymer stationary phase should exhibit a hydrophilic, negatively charged property at temperatures below the LCST. As temperature is increased, the polymer collapses and dehydrates, preventing charged groups access to the mobile-phase hydrophilic environment, resulting in decreased polymer surface charge density. Urry et al.19 reported hydrophobicity-induced pKa shifts in synthetic polypeptides, and such shifts are routinely observed for charged groups at interfaces or within small compartments. Thus, it appears reasonable to create novel stimuli-responsive chromatography stationary phases modified with PIPAAm derivatives containing AAc that exhibit substantial changes in partitioning capabilities in aqueous milieu. (13) Okano, T.; Yamada, N.; Sakai, H.; Sakurai, Y. J. Biomed. Mater. Res. 1993, 27, 1243-1251. (14) Okano, T.; Yamada, N.; Okuhara, M.; Sakai, H.; Sakurai, Y. Biomaterials 1995, 16, 297-303. (15) Kikuchi, A.; Okuhara, M.; Karikusa, F.; Sakurai, Y.; Okano, T. J. Biomater. Sci. Polym. Ed. 1998, 9, 1331-1348. (16) Gewehr, M.; Nakamura, K.; Ise, N.; Kitano, H. Makromol. Chem. 1992, 193, 249-256. (17) Hosoya, K.; Sawada, E.; Kimata, K.; Araki, T.; Tanaka, N.; Frechet, J. M. J. Macromolecules 1994, 27, 3973-3976. (18) Yakushiji, T.; Sakai, K.; Kikuchi, A.; Aoyagi, T.; Sakurai, Y.; Okano, T. Langmuir 1998, 14, 4657-4662. (19) Urry, D. W.; Peng, S.: Parker, T. J. Am. Chem. Soc. 1993, 115, 75097510.
2028
Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
EXPERIMENTAL SECTION Materials. N-Isopropylacrylamide (IPAAm) was kindly provided from Kojin Co. Ltd. (Tokyo, Japan) and recrystallized from n-hexane. Acrylic acid (AAc) (Wako Pure Chemical Industries, Ltd., Osaka, Japan) was purified by distillation at 42 °C (7 mmHg). N,N′-Methylenebisacrylamide (MBAAm) was purchased from Eastman Kodak Co. (Rochester, NY). 2,2′-Azobisisobutyronitrile (AIBN; Wako) was purified by recrystallization from methanol. N,N-Dimethylformamide (DMF) and tetrahydrofuran (THF), both from Kanto Chemical Co. (Tokyo, Japan), were purified by distillation at 53 (9 mmHg) and at 67 °C (760 mmHg), respectively. 4,4′-Azobis(4-cyanovaleric acid) (ACV; Wako), 1-(ethoxycarbonyl)2-ethoxy-1,2-dihydroquinoline (EEDQ; Tokyo Kasei Kogyo Co., Tokyo, Japan), diethyl ether (Wako), ethanol (Wako), and all other chemicals were used as received. Aminopropyl silica beads (average diameter, 5 µm; pore size, 120 Å) were purchased from Nishio Industry (Tokyo, Japan). Sulfosuccinimidyl-4-o-(4,4′-dimethoxytrityl) butyrate (s-SDTB) was obtained from Pierce (Rockford, IL). Synthesis of Linear PIPAAm Copolymers. Poly(IPAAm-coAAc)s with varying feed ratios (IPAAm/AAc mole ratio, 99/1, 98/ 2, and 97/3) were synthesized in THF (monomer, 1.43 mol/L) with AIBN as an initiator (1.4 mmol of AIBN/mol of monomer). The reaction mixture was degassed by triplicate freeze-thaw cycles and sealed under reduced pressure to remove dissolved oxygen. Polymerization was then carried out at 70 °C for 2 h. Polymer was precipitated in diethyl ether and dried under vacuum. Obtained crude polymers were further purified by dialysis from 5 wt % aqueous solution against distilled water at 4 °C for 3 days using cellulose ester dialysis membrane tube (molecular weight cutoff 500, Spectrum, California, CA), with water changed every day. Polymer products were recovered by freeze-drying. Titrations. AAc content in each copolymer was determined by acid-base titration. Poly(IPAAm-co-AAc) (100 mg) was dissolved in 20 mL of distilled water. Two milliliters of 0.1 M HCl was added to completely protonate carboxyl groups. Titration with 0.05 M NaOH was performed using pH meter (HM-30V; TOA Electronics, Tokyo, Japan) at 4 °C under N2 gas bubbling. The amount of AAc in the copolymer was calculated from the amount of NaOH required for neutralization of carboxyl groups. Apparent dissociation constants (Ka′) of the linear copolymers were determined from titration curves using the following Henderson-Hasselbalch equation:
pKa′ ) pH - log(R/(1 - R))
(1)
where R is the dissociation degree for carboxyl groups. Polymer LCST Measurement. Solutions of poly(IPAAm-coAAc) containing different amounts of AAc were prepared in phosphate/citrate buffer (1.0 w/v%). Optical transmittance changes of the solutions were monitored at 500 nm with a UV/visible spectrometer (V-530, Japan Spectroscopic, Co. Ltd., Tokyo, Japan). The sample cell was thermostated with a Peltier-effect cell holder (EHC-477, Japan Spectroscopic) with heating rate of 0.1 °C/min. The LCST was defined as the temperature at 90% transmittance of solution. Immobilization of Initiator on Silica Surfaces. Aminopropyl silica beads were used as base matrixes for liquid chromatography.
Table 1. Characterization of Poly(IPAAm-Co-AAc)s AAc content (mol %)
c
Figure 1. Preparation of cross-linked thermosensitive poly(IPAAmco-AAc) gel grafted onto silica bead surfaces.
ACV, polymerization initiator, was immobilized on aminopropyl silica beads through amide bond formation (first step in Figure 1). The detailed procedure for ACV immobilization was previously reported.4 Introduction of Copolymer Hydrogels on Silica Bead Surfaces. Cross-linked copolymer grafted gels of IPAAm and AAc were prepared by radical polymerization on the surface of initiatorimmobilized silica beads (second step in Figure 1). Typical preparation procedure was as follows: IPAAm (19.4 g, 171 mmol), AAc (0.382 g, 5.30 mmol), and MBAAm (0.272 g, 1.76 mmol) (IPAAm/AAc ) 97/3 mol/mol, 1.0 mol % MBAAm to monomer) were dissolved in 200 mL of ethanol. ACV-immobilized silica beads (5.00 g) were added to the monomer solution. The reaction mixture was then degassed by being subjected to three freezethaw cycles and was sealed under reduced pressure. Polymerization was carried out at 70 °C for 15 h with vigorous stirring. Copolymer hydrogel-modified silica beads were collected by filtration, washed three times with ethanol to remove unreacted monomer and unimmobilized polymer, and then dried for 12 h under vacuum at 25 °C. PIPAAm hydrogel-modified silica beads as a control were also prepared in a similar manner. Elemental analyses of polymer hydrogel-modified surfaces were carried out by electron spectroscopy for chemical analysis (ESCA; ESCA750, Shimadzu, Kyoto, Japan) with a takeoff angle of 90°. The proportion of nitrogen to carbon atoms was calculated from core-level high-resolution spectra areal integration of C1s and N1s peaks collected with sampling time of 200 ms and analyzed with ESPAC210 software (Shimadzu).
code
in feed
in copolymera
pKa′a,b
LCSTc (°C)
IA1.4 IA2.1 IA3.0
1.00 2.00 3.00
1.36 2.09 3.00
4.90 4.87 4.84
32.4 34.1 36.4
a Determined by acid-base titration (n ) 3). b Measured at 4 °C. Determined at 90% transmittance.
pH-/Temperature-Responsive Elution of Basic Analytes. Polymer hydrogel-modified silica beads were packed into a stainless steel column (150 mm × 4.6 mm i.d.) from a slurry of beads in water/methanol (1:1) using a column packer at 350 kg/ cm2 for 1 h followed by equilibration with distilled water for 24 h. This column was connected to an HPLC system (PU-980, AS-950, UV-970, Japan Spectroscopic Co.) controlled by personal computer with Borwin analysis software version 1.21 (Japan Spectroscopic Co.). Milli-Q water was used for preparation of buffer and sample solution. Mobile phase comprised phosphate/citrate buffer adjusted to constant ionic strength (I ) 0.1) by addition of KCl. Four model analytes containing amino groups were used to produce the chromatograms at a concentrations of 1.0 mg/mL for D,LDOPA, adrenaline, dopamine hydrochloride, and tyramine with 2.7 mg/mL Na2SO3 used to prevent sample oxidization. Two types of steroidsscortisone and prednisolone (both 0.25 mg/mL)swere dissolved in Milli-Q water and used as nonionic hydrophobic reference samples. Thermoresponsive elution behavior for these substances was monitored at 254 nm at a flow rate of 0.50 mL/ min at various temperatures. Column temperature was thermostated with a water jacket connected to a Coolnics circulator (CTE42A, Yamato-Komatsu, Tokyo, Japan) within a deviation of (0.1 °C. RESULTS AND DISCUSSION Syntheses and Characterization of pH-/TemperatureResponsive Copolymers. Ionic group content, pH, and ionic strength of solutions all influenced ionic thermoresponsive polymer LCST behavior. Effective design of the desired HPLC stationary phase requires confirmation of the specific LCST of grafted poly(IPAAm-co-AAc)s at various pH aqueous conditions. In this regard, we used linear copolymers as models for LCST determinations of the surface grafted hydrogels. Poly(IPAAm-coAAc) samples with different monomer compositions were obtained by radical polymerization. AAc content and pKa′ values of these copolymers at 4 °C determined from acid-base titration are summarized in Table 1. Copolymer composition is close to monomer feed values. Values for copolymer pKa′ are seen to increase over that of the AAc monomer (pKa 4.26) and exhibited similar values regardless of copolymer composition at 4 °C. LCST values for the copolymers at pH 7.0 and I ) 0.1 were estimated from temperature-dependent solution transmittance changes and are also summarized in Table 1. Copolymer LCST increases with increasing AAc content. Furthermore, the temperature range for transmittance changes became wider with increasing AAc contents in the copolymer (data not shown). Incorporation of a small amount of “hydrophilic” comonomer, AAc, produces a Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
2029
Figure 2. Temperature-dependent optical transmittance changes of IA3.0 in phosphate/citrate buffer at pH 4.0 (diamond), 5.0 (square), 6.0 (circle), and 7.0 (triangle) (I ) 0.1).
large increase in LCST due to the increasing hydrophilicity of the random polymer.20-22 We previously proposed that the increasing LCSTs of poly(IPAAm-co-AAc) are attributed to the disruption of the hydrophobic aggregation of PIPAAm sequences by incorporation of hydrophilic AAc units.22 Figure 2 shows the effect of pH on the optical transmittance changes of the copolymer containing 3 mol % AAc (IA3.0). Ionic strength was set at 0.1 regardless of buffer pH. Higher LCST is observed with an increase in buffer pH due to the increasing dissociation of carboxylate anions in the copolymer. At lower pH, protonated carboxyl groups in the copolymer are electrically neutral and start to dissociate, producing increasing negative charge with increasing pH. Increasing dissociation produces increasing ionic side-chain hydration, resulting in an increase of overall polymer hydrophilicity. From the data indicated in Figure 2, the copolymer containing 3 mol % AAc (sample IA3.0) exhibits sufficient response to both pH and temperature for HPLC application. Further incorporation of AAc units into PIPAAm diminishes thermoresponse.21, 22 To confirm temperature-induced pKa′ shifts, the pKa′ of sample copolymer IA3.0 was measured as a function of temperature (Figure 3). A substantial decrease in copolymer acidity is observed above 35 °C. Above this temperature, IA3.0 is insoluble in water and solutions of linear IA3.0 become turbid. Feil et al.23 reported pKb shifts for a thermoresponsive terpolymer composed of IPAAm, hydrophobic BMA, and cationic N,N-diethylaminoethyl methacrylate (DEAEMA). The basicity of DEAEMA in the terpolymer decreases strongly as temperature increases, although the pKb of DEAEMA was independent of temperature in the copolymer with more hydrophilic comonomer, acrylamide. This phenomenon can be explained by the increasing hydrophobicity of IPAAm copolymer sequences and subsequent increasing dehydration (reduced local dielectric constant) with increasing temperature. Several studies have also shown that incorporation of hydrophobic comonomers into polyelectrolytes leads to decreases in the acidity (20) Gutowska, A.; Bae, Y. H.; Feijen, J.; Kim, S. W. J. Controlled Release 1992, 22, 95-104. (21) Kaneko, Y.; Nakamura, S.; Sakai, K.; Aoyagi, T.; Kikuchi, A.; Sakurai, Y.; Okano, T. Macromolecules 1998, 31, 6099-6105. (22) Aoyagi, T.; Ebara, M.; Sakai, K.; Sakurai, Y.; Okano, T. J. Biomater. Sci., Polym. Ed. 2000, 11, 101-110. (23) Feil, H.; Bae, Y. H.; Feijen, J.; Kim, S. W. Macromolecules 1992, 25, 55285530.
2030 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
Figure 3. Polymer carboxylate pKa′ shifts observed for IA3.0 in 100 mM KCl solution as a function of temperature.
or basicity of weak acids or bases.24,25 Thus, observed changes in pKa′ should be predictably attributed to the reduced dielectric constant around the polymer side-chain environment in water at higher temperature. Copolymer IA3.0 with those characteristics is then introduced onto silica beads as discussed in the next section. Preparation of Hydrogel-Modified Silica. The amount of immobilized initiator, ACV, was estimated by measuring the consumption of amino groups on silica bead surfaces with s-SDTB.26 According to this method, the amount of amino groups on native aminopropyl silica beads was 250 µmol/g of beads. Amino groups were consumed by immobilization of ACV. Residual amino groups after immobilization of ACV amounted to 6.8 µmol/g of beads. Thus, ACV was introduced onto silica bead surfaces with calculated 97% efficiency. Poly(IPAAm-co-AAc) hydrogels having a monomer composition analogous to linear sample IA3.0 (IPAAm/AAc ) 97/3 mol/mol, 1.0 mol % MBAAm/mol of monomer; designated IA3G) and pure PIPAAm hydrogel (IG) as a control surface were introduced onto silica bead surfaces, respectively. Successful introduction of these hydrogels onto silica bead surfaces was confirmed by ESCA analyses. Detected surface composition from elements, C, N, O, and Si is summarized in Table 2. Slightly larger amounts of nitrogen and smaller amounts of oxygen and silicon were observed on polymer-grafted silica beads than those on native amino silica beads. N/C ratios calculated from the ESCA data were almost identical to theoretical values for both types of polymer-modified silica beads. This supports polymer immobilization as full coatings on these bead surfaces. Separation of Basic Solutes Using pH-/TemperatureResponsive Stationary Phases. We previously reported that hydrophobic interactions between steroids and PIPAAm hydrogelmodified surfaces is readily modulated by temperature-dependent hydrophilic/hydrophobic surface property alteration on these stationary phases, modulating retention times and facilitating effective separation in aqueous media.1-4 Most bioactive compounds, however, have differences not only in polarities (hydrophobicities) but also in ionic properties. Therefore, separation of a wide variety of bioactive compounds is restricted to some extent (24) Siegel, R. A.; Firestone, B. A. Macromolecules 1988, 21, 3254-3259. (25) Pradny, M.; Kopecek, J. Makromol. Chem. 1990, 191, 1887-1897. (26) Gaur, R. K.; Gupta, K. C. Anal. Biochem. 1989, 180, 253-258.
Table 2. Elemental Analyses on Unmodified and Two Types of Polymer-Modified Silica Beads by ESCAa element (%) code
C
O
N
Si
N/C
N/Cb (theor)
amino silica IG IA3G
41.95 (2.09) 43.25 (3.49) 48.29 (4.22)
33.97 (1.73) 34.39 (2.78) 27.47 (3.80)
5.49 (1.93) 6.78 (1.09) 9.67 (1.47)
18.60 (2.45) 15.59 (2.19) 14.57 (0.80)
0.131 (0.043) 0.157 (0.036) 0.200 (0.038)
0.168 0.166
a Data are calculated as mean of four samples. The values in parentheses indicate the standard deviation (n ) 4). b Calculated by theoretical molar ratios.
Figure 4. Chromatograms of model basic analyte compounds on (a) IG and (b) IA3G columns at 10, 30, and 50 °C. Mobile phase is phosphate/citrate buffer (pH 7.0, I ) 0.1). Peaks: 1, DOPA; 2, adrenaline; 3, dopamine; 4, tyramine.
by use of nonionic hydrophobic chromatography. We introduced ionic groups into thermoresponsive stationary phases to investigate separation of ionic bioactive compounds, specifically catecholamine derivatives. Anionically modified poly(IPAAm-co-AAc) hydrogel-grafted HPLC columns were used to probe the contributions of electrostatic interactions to temperature-dependent elution, compared to the nonionic PIPAAm hydrogel-modified control column. Since PIPAAm-based hydrogels possess large thermoresponsive gel volume transitions, the thermoresponsive void volume change of the hydrogel-modified columns was evaluated using deuterium oxide (D2O) as a mobile-phase dopant. We confirmed that the D2O retention time was constant regardless of temperature and pH, indicating that column void volume was not affected by either temperature or pH change. This suggests that the gel volume phase transition is negligible and that a very thin grafted polymer gel layer is formed on the silica surface. Figure 4a shows chromatograms of several basic catecholamine compounds separated on the PIPAAm hydrogel (IG) column. Little change in retention times is observed for all solutes examined on this IG column regardless of temperature. Shorter retention times with overlapping unresolved peaks resulted in poor resolution for the catecholamine derivatives. Elution times increased in the order DOPA < adrenaline < dopamine < tyramine. Analyte hydrophobicity (polarity), typically represented as the log P value, where P is the partition coefficient in a 1-octanol/water system,27 is as follows for the catecholamine derivatives: -2.74 for DOPA, -0.685 for adrenaline, 0.019 for dopamine, and 0.616 for tyramine, respectively. This order is in agreement with the
observed order of elution times on IG columns. Hydrophobic partitioning of these solutes into the IG stationary phase, however, was significantly weakened due to the higher solubility of basic compounds readily protonated at this pH 7.0. Figure 4b shows chromatograms of the same four compounds on the charged IA3G column at pH 7.0 under varying column temperatures. Basic compounds (not the zwitterionic DOPA) were all retained on the IA3G column at pH 7.0 with longer retention times observed compared to the IG column. This improved peak resolution is proposed to result from the additional electrostatic interaction between dissociated IA3G carboxylate groups on stationary-phase surfaces and the basic protonated analytes. In Figure 4, observed analyte peaks are broader with peak tailing. We previously discussed broadening of elution peaks on thin PIPAAm gel grafted columns.4 We considered that there are two factors affected to peak shapes: (1) analyte partitioning toward poly(IPAAm-co-AAc) hydrogel layers and (2) stronger hydrophobic interaction between analytes and hydrophobized matrix surfaces at elevated temperature. Both of them have substantial influence on peak shapes. It should be improved by optimizing column dimensions as well as elution conditions. Effect of pH on Retention Time of Catecholamine Derivatives on Hydrophilic Stationary Phases. Changes in retention times of basic compounds are shown in Figure 5 at various buffer pH values. Temperature was set at 10 °C where grafted stationaryphase polymer is in a swollen state and thus hydrophilic regardless of solution pH. IA3G carboxyl side-chain dissociation degree (R, indicated on the top of Figure 5) increases with an increase in buffer pH. DOPA retention time remained constant regardless of solution pH, a reasonable result since DOPA exists as an ampholyte under the examined pH range (i.e., pKa (NH3+) 8.72, (COOH) 2.32). The high solubility of DOPA in buffer limits interaction with the swollen, hydrophilic stationary phase, resulting in void volume elution. Retention times for catecholamines containing only amino groups (adrenaline, dopamine, tyramine) were prolonged with increasing buffer pH, especially above pH 5.0. Copolymer AAc carboxylate groups in stationary-phase gel dissociate above this pH as apparent from the change in R, supporting electrostatic interaction mechanisms between protonated analytes with anionic IA3G stationary phase under these elution conditions. It is also seen that the analyte with higher hydrophobicity (tyramine) exhibited longer retention time. After neutralization of analyte through electrostatic interaction (salt formation) with the anionic IA3G surface, analyte hydration may (27) Hansch, C.; Leo, A.; Hoekman, D. Exploring QSAR-Hydrophobic, Electronic, and Steric Constants; ACS Professional Reference Book; American Chemical Society: Washington, DC, 1995.
Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
2031
Figure 7. Temperature-dependent retention time changes for steroids on the anionic IA3G column at (a) pH 5.0 and (b) 7.0 (I ) 0.1); cortisone (circle) and prednisolone (diamond).
Figure 5. pH-dependent retention time changes for catecholamine derivatives on the anionic IA3G column at 10 °C (I ) 0.1). Change in dissociation degree (R) of acrylic acid carboxyls is also indicated on this figure. DOPA (circle), adrenaline (square), dopamine (diamond), and tyramine (triangle).
Figure 6. Temperature-dependent retention time changes of model basic analytes on the anionic IA3G column at (a) pH 5.0, (b) 6.0, and (c) 7.0 (I ) 0.1). Symbols remain the same as Figure 6 except for hydrophilic model compound, adenosine (closed circle).
be disrupted, producing a hydrophobic mode of retention. Therefore, longer retention times are observed for more hydrophobic analytes. Effect of Temperature on Retention Time at Various pH. Figure 6 shows temperature-dependent retention times for the model analytes at pH 5.0, 6.0, and 7.0. Every basic compound has an entirely protonated amino group, becoming highly hydrophilic below pH 7.0. At pH 5.0, dissociation of carboxyl groups on the stationary phase is incomplete and is further suppressed with temperature, resulting in relatively faster elution of hydrophilic ionized solutes. Retention times of solutes at pH 5.0 and those except tyramine at pH 6.0 are slightly reduced with increasing temperature. At pH 7.0, by contrast, longer sample retention times (except for DOPA) were observed over those at pH 5.0 and 6.0. The retention time for tyramine at pH 7.0 was delayed with increasing temperature. As discussed in the previous section, the primary force of interaction between model catecholamine derivatives and the anionic IA3G surface is electrostatic. With increasing temperature, surface charge density is considered to decrease, since carboxyl group surface pKa′ increases with temperature as shown in Figure 3. This is supported by the observed retention behavior of basic, hydrophilic adenosine shown in Figure 6c (closed circle). Reten2032 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
tion of adenosine decreased with temperature, indicating weakened electrostatic attraction to IA3G surfaces at higher temperature. As described above, it is also reasonable to suppose that IG and IA3G stationary-phase surface hydrophobicity increases with increasing temperature, inducing other hydrophobic interaction with these solutes. To confirm temperature-dependent surface hydrophilic/ hydrophobic changes, hydrophobic neutral steroid retention was also examined. Figure 7 shows temperature-dependent retention times for nonionic hydrophobic steroids at pH 5.0 and 7.0. Retention times for each steroid increased at higher temperature and depended on the order of analyte hydrophobicity (log P values for steroids are 1.47 for cortisone and 1.62 for prednisolone, respectively). This indicates that the main interaction force for steroid partitioning is hydrophobic in origin. Slightly longer retention times for steroids are observed at pH 5.0 over those at pH 7.0. Especially above 40 °C (near the IA3G LCST), steroid retention times increase significantly due to strong partitioning with the stationary phase with observable peak tailing. Increased steroid interaction is observed on the IA3G surface with temperature even though stationary-phase carboxyl groups are largely dissociated at pH 7.0. This is strong evidence that IA3G surfaces become hydrophobic at higher temperature even at pH 7.0. Hence, both electrostatic and hydrophobic interactions simultaneously influence partitioning of basic catecholamine derivatives. These two interaction forces compete with each other, resulting in constant retention times with temperature observed for adrenaline and dopamine at pH 7.0 in Figure 6c. Tyramine is the most hydrophobic solute among the four samples tested at elevated temperature, showing longer retention times. Successful separation on the IA3G column is achieved at pH 7.0 and 50 °C through mutual influences of both electrostatic and hydrophobic interactions. CONCLUSIONS We prepared silica stationary-phase surfaces modified with grafted poly(IPAAm-co-AAc) thin hydrogel layers to investigate temperature- and pH-dependent elution behavior of basic bioactive compounds (catecholamine derivatives) in aqueous mobile phases. Catecholamine derivatives were retained on poly(IPAAm-co-AAc) columns above pH 5.0 due mainly to electrostatic interactions while weak interactions with insufficient resolution occurred on the nonionic PIPAAm column. Retention behavior of these basic compounds is readily modulated by temperature-responsive
hydrophobic and electrostatic property alterations of the stationary phase. Since chromatographic separation is performed using a mild aqueous mobile phase, this system would have the advantages of preserving analyte bioactivity and low environmental impact (no organic mobile-phase disposal issues). As thermoresponsive surface property alterations are weakened by increasing amounts of acrylic acid copolymerized into PIPAAm, modification of stationary-phase hydrophobicity using copolymerization of hydrophobic comonomer and/or modulation of copolymer carboxyl group dissociation would produce more effective separation. Details will be reported in a separate paper. In conclusion, we demonstrate a new concept of aqueous chromatography to separate ionic compounds through the modulation of stationaryphase properties with only temperature changes.
ACKNOWLEDGMENT Part of this work was financially supported by the Japan Chemical Innovation Institute (JCII) under the New Energy and Industrial Technology Development Organization (NEDO), Industrial Science and Technology Frontier Program, “Technology for Novel High-Functional Materials”. Technical critique of the manuscript by Prof. David W. Grainger (Colorado State University) is appreciated.
Received for review November 20, 2000. Accepted February 17, 2001. AC0013507
Analytical Chemistry, Vol. 73, No. 9, May 1, 2001
2033
