Anal Bioanal Chem (2004) 378 : 84–88 DOI 10.1007/s00216-003-2320-4

PA P E R I N F O R E F R O N T

Takuya Kubo · Nobuo Tanaka · Ken Hosoya

Target-selective ion-exchange media for highly hydrophilic compounds: a possible solution by use of the “interval immobilization technique”

Received: 27 August 2003 / Revised: 17 September 2003 / Accepted: 30 September 2003 / Published online: 13 November 2003 © Springer-Verlag 2003

Abstract We prepared a crosslinked polymer as a separation and/or adsorption medium for CYN, as shown in Fig. 1. The polymers were evaluated by high-performance liquid chromatography (HPLC) and adsorption was examined under batch conditions. Results from detailed HPLC evaluation and measurement of the difference between the binding affinity for CYN and for other compounds showed the prepared polymer had specific recognition ability for CYN. Keywords Ion-exchange media · Highly hydrophilic compounds · Interval immobilization technique · Cylindrospermopsin (CYN) · Cyanobacterium hepatotoxin

Introduction Ion-exchange has been often used for separation and/or adsorption of ionic compounds, for example hydrophilic polypeptides, DNA, or other hydrophilic ionic compounds. Because of the growing need for separation of such compounds, ion-exchange separation and/or adsorption media with selectivity for target compounds will be required. Molecular imprinting (MI) is a successful means of preparing selective ion-exchange media for a particular target compound. In this method we can easily obtain specific molecular recognition of the molecule used as the template molecule. Many examples and useful applications of MI have been reported [1, 2, 3, 4]. However, MI can be rarely be applied for highly hydrophilic compounds, rare naturally occurring compounds, and highly toxic compounds, because it is necessary to use the real target molecule as the template molecule in MI [5].

T. Kubo · N. Tanaka · K. Hosoya (✉) Department of Polymer Science and Engineering, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan e-mail: [email protected]

We have reported several applications of MI in which the real target molecule is not used [6, 7, 8, 9, 10]. Another problem of MI using non-covalent-type interaction between the template molecule and the functional monomers is the formation of rather heterogeneous recognition sites. The formation of heterogeneous recognition sites in MI has been studied [11, 12, 13]. In MI an appropriate organic solvent is usually used as polymerization solvent (porogenic solvent) to dissolve the monomers and the template molecule. In a relatively hydrophobic organic solvent, hydrogen bonding works effectively between the template molecule and the functional monomers, resulting in formation of more accurate recognition sites. MI can, therefore, hardly be applied for highly hydrophilic template molecules such as amino acids, hydrophilic polypeptides, and organic ionic material, because those compounds dissolve only in water, in which typical monomers do not dissolve. For example, many natural toxic compounds occur in environmental water. Cyanobacterium toxins, shellfish toxins, and fish toxins are examples. These toxic compounds are highly hydrophilic because of their ionic properties, and so conventional adsorption methods based on hydrophobic interaction are not effective at trapping them. The occurrence of cyanobacterium toxins in freshwater could be a serious danger to humans and domestic animals if the water of lakes or ponds is used as drinking water [14, 15, 16, 17]. To avoid this danger quantitative analysis and effective removal of each toxin will be required. Environmental water is contaminated by many species, however, so direct analysis and selective removal of the target toxins is very difficult and novel separation media with selectivity for the highly hydrophilic toxins will be necessary. In this report, we propose a novel separation medium for cylindrospermopsin (CYN), a powerful cyanobacterium hepatotoxin (Fig. 1), as an example [18, 19, 20], by use of a novel technique, called the “interval immobilization technique”. By use of this technique we believe that ionic functional monomers can be immobilized at regular distances and that these immobilized groups specifically

85

Fig. 1 Concept of interval immobilization technique for CYN

recognize target molecules with ionic groups separated by the same distance as the immobilized ionic groups.

Experimental For preparation of adsorbents, ethylene glycol dimethacrylate (EDMA) and 2-(diethylamino)ethyl methacrylate (DAEMA) were purified by vacuum distillation to remove polymerization inhibitor. 2,2′-Azobis-(2,4-dimethylvaleronitrile) (ADVN), methyl (bromomethyl)benzoate, 4-(bromomethyl)phenyl acetic acid, trimethylsilyldiazomethane, and tetrabutylammonium chloride were purchased from Tokyo Kasei (Tokyo Japan). Methyl iodide, p-styrene sulfonic acid sodium salt (Ssa), methanol (MeOH), acetonitrile (AN), ethanol (EtOH), chloroform, dimethyl sulfoxide (DMSO), and NaCl were purchased from Wako Chemicals (Kyoto Japan) and used as received. Synthesis of tributyl-(4-carboxybenzyl)ammonium (Tcba) as possible pseudo-template for CYN Methyl (4-bromomethyl)benzoate (6.9 mmol, 1.5 g) and K2CO3 (1.0 g) were dissolved in 50 mL AN. Tributylamine (4.2 mmol, 1.0 mL) was added slowly to the reaction mixture with stirring. The mixture was then stirred at 100 °C, under reflux, for 24 h under a nitrogen atmosphere. Generation of the target compound was confirmed by TLC with a detection reagent. When the reaction was complete the mixture was separated by silica gel column chromatography and tributyl-(4-methoxycarbonylbenzyl)-ammonium chloride was obtained. The tributyl-(4-methoxycarbonylbenzyl)ammonium chloride was dissolved in hydrochloric acid (1.0 mol L–1, 50 mL) and heated at 120 °C under reflux for 24 h under a nitrogen atmosphere. After confirmation, by TLC, of loss of the methyl ester the resulting tributyl-(4-carboxybenzyl)ammonium (Tcba), was extracted with chloroform and purified by silica gel column chromatography.

Yield 90.2%. 1H NMR (in CD3OD) δ (ppm): 0.97 (m, 12H), 1.36 (m, 8H), 1.76 (m, 8H), 3.13 (m, 8H), 4.78 (s, 2H), 7.42 (d, 2H), 7.97 (d, 2H). Synthesis of tributyl-(4-carboxymethylbenzyl)ammonium (Tcmba) as possible looser pseudo-template for CYN 4-(Bromomethyl)phenylacetic acid (2.0 g) and trimethylsilyldiazomethane (2.0 mol L–1 solution in hexane, 10 mL) were dissolved in 50 mL MeOH–Benzene, 15:35. The mixture was stirred at room temperature for 2 h. The solvent was then evaporated and the residue was treated with tributylamine at 100 °C for 24 h. Tributyl(4-methoxycarbonylmethylbenzyl)ammonium was isolated by silica gel column chromatography and the resulting compound was hydrolyzed with 1.0 mol L–1 aqueous NH3 solution. Finally, the target compound, tributyl-(4-carboxymethylbenzyl)ammonium, was isolated by silica gel column chromatography. Yield 64.3%. 1H NMR (in CD3OD) δ (ppm): 1.03 (m, 12H), 1.41 (m, 8H), 1.82 (m, 8H), 3.18 (m, 8H), 3.60 (s, 2H), 4.75 (s, 2H), 7.45 (d, 4H). Isolation of CYN A toxic strain of Cylindrospermopsis raciborskii (CRJ-1=AWT 205) was obtained from the Microbial Culture Collection (MCC-NIES) and grown in CT medium. Cells were separated from the medium by centrifugation and lyophilized. CYN was extracted according to a method reported elsewhere [18]. The extracted CYN was purified by HPLC on an Amide-80 column (10 mm×250 mm, Tosoh Corporation, Japan) with a 100 to 60% aqueous AN linear gradient in 20 min at 4.0 mL min–1. The isolated CYN was identified by NMR and MS. The spectrometric data were in good agreement with those of CYN [18]. Preparation of polymer adsorbents The complexes of p-styrene sulfonic acid sodium salt (Ssa) and the ammonium compound were prepared as follows. Ssa was dis-

86 Table 1 Composition of polymers Symbol

Composition

P-Base

EDMA (5 mL, 26.5 mmol), 1:1 AN–EtOH (5 mL) EDMA (5 mL, 26.5 mmol), 1:1 AN–EtOH (5 mL), DAEMA (2 mmol), Complex of Ssa and Tba (2 mmol) EDMA (5 mL, 26.5 mmol), 1:1 AN–EtOH (5 mL), DAEMA (2 mmol), Complex of Ssa and Tcba (2 mmol) EDMA (5 mL, 26.5 mmol), 1:1 AN–EtOH (5 mL), DAEMA (2 mmol), Complex of Ssa and Tcmba (2 mmol)

P-BL (Non-MIP, blank) P-Tcba (M1P 1)

P-Tcmba (M1P 2)

EDMA, ethylene glycol dimethacrylate; DAEMA, 2-(diethylamino)ethyl methacrylate; Ssa, p-styrene sulfonic acid sodium salt; Tba, tetrabutylammonium; Tcba, tributyl-(4-carboxybenzyl)ammonium; Tcmba, tributyl-(4-carboxymethylbenzyl)ammonium

solved in water and extracted with chloroform containing one half the mole ratio of tetrabutylammonium chloride (or Tcba or Tcmba) relative to Ssa, by the phase transfer effect. The solvent was then removed to give the complexes. The polymer adsorbents were prepared with 1.0% w/w ADVN as radical initiator, at 50 °C, for 24 h. The polymers were ground, washed with MeOH, then treated with CH3I to generate the alkylammonium groups separate from base

Fig. 2 Structures of the pseudo-template molecules and distance between ionic groups

polymer (P-base). The reaction was carried out in DMSO at 70 °C for 24 h as shown in Fig. 1. The symbols used for the polymers, and the polymer compositions, are also summarized in Table 1. HPLC evaluation of polymer adsorbents Each polymer was isolated in the size range 25–45 µm. The resulting polymer particles were packed into stainless steel columns by a slurry method and evaluated by HPLC using a 90% methanol aqueous solution containing NaCl as mobile phase, because retention of the ionic solutes was too large to be detected under salt-free conditions. Chromatographic data were acquired with a Shimadzu (Japan) HPLC system consisting of an LC-6A pump, an SPDM10A photodiode-array detector, and a CTO-10AC column oven. Scatchard plot To construct a Scatchard plot CYN solutions were prepared at concentrations of 1.0–10–4 mmol L–1 in 90% MeOH aqueous solution. Each CYN solution (1.0 mL) was added to vials containing 10 mg P-BL or P-Tcba. After 12 h, during which the vials were shaken at regular time intervals, the amount of free CYN in the supernatant was quantified by HPLC–MS with external standard calibration. Quantitative determination of CYN was performed by HPLC with an Amide-80 column (100 mm×2.0 mm, Tosoh, Japan) with 100 to 60% AN linear gradient, in 20 min, at 0.2 mL min–1. CYN was detected by MS in SIM mode at m/z 414.

87

Results and discussion To immobilize ionic functional groups on to the polymer adsorbents, two possible pseudo-template molecules were designed by computer modeling taking into consideration the distance (interval) between the two ionic groups of CYN. The structures of the functional groups and the distance between them are shown in Fig. 2. A possible pseudo-template molecule (Tcba) and two functional monomers in polymerization solvent were examined by FTIR. Comparison of the spectra of Tcba and the complexes with Ssa and Tcba revealed the specific adsorption band attributed to the sulfonyl group at approximately 1200 cm–1. Moreover, although Ssa was not soluFig. 3 Comparison of separation factor for CYN relative to other ionic solutes. HPLC conditions: mobile phase 9:1 MeOH–1.0 mol L–1 NaCl aq; column size 100 mm×4.6 mm (i.d.); flow rate 0.5 mL min–1; detection photo-diode array; temperature 30 °C

Fig. 4 Scatchard plots for CYN on each polymer. The association constant (Ka) and number of binding sites were determined from the slope and y intercept, respectively, of the fitted line (n/Cf=–Kan+KaN) obtained by least-squares regression

ble in the solvent the complex of Ssa and Tcba was readily soluble. Therefore it seemed that the complex was stable in this solvent. Comparison of the spectra of DAEMA and complex revealed the specific adsorption band attributed to hydrogen bonding between the carboxylic acid and amine at approximately 1600 cm–1. On the basis of these results it was therefore believed that the pseudo-print molecule and each functional monomer could interact in each other in the polymerization mixture. Columns packed with the polymer adsorbents were prepared and retention factors, k′, and separation factors, α, were determined for CYN and some ionic solutes. The results obtained from these examinations revealed that both ionic functional groups had been immobilized on

88

each polymer (P-BL, P-Tcba, and P-Tcmba) – this was shown by comparison of retention factors for ionic solutes on P-BL, P-Tcba, and P-Tcmba with those on P-Base with ionic functional groups attached. Moreover, as is shown in Fig. 3, high selectivity was observed for CYN on P-Tcba, prepared with Tcba as the possible pseudo-template molecule, for which the distance between the functional groups was closer than those of CYN. This selectivity suggests that the distance between the immobilized ionic groups obtained by use of the possible pseudo-template enabled selective recognition of the CYN molecule. The dependence of the retention factor for CYN on the concentration of NaCl in the mobile phase was also examined. It was shown that the slope of k′ for CYN on P-Tcba was much less steep than that for CYN on P-BL. This suggests that recognition sites on P-Tcba recognize CYN more strongly. Thus retention of CYN was lower even for higher NaCl concentrations.

form high-affinity binding sites for CYN, as illustrated in Fig. 1. According to the results from the Scatchard plot it seems that a specific distance between binding sites for CYN was achieved on P-Tcba by the interval immobilization technique for ionic groups, resulting in a high association constant then the polymer P-Tcba was used. Although the N value was larger on P-BL than that on P-Tcba, the Scatchard plot based on batch adsorption strongly suggests that specific recognition sites for CYN were formed within the polymer prepared by the interval immobilization technique, for example P-Tcba, whereas on P-BL non-specific adsorption sites are dominant. We suggest that the technique proposed in this paper can be applied to other highly hydrophilic compounds, for example as natural toxins. Moreover, the technique can also be used for analysis of other biological compounds. For example, recognition of part of a protein by ionic groups as a result of the interval immobilization technique might lead to analysis of specific groups of proteins.

Scatchard plot

References To obtain more detailed information about the mechanism of recognition of CYN, the Scatchard plot was constructed under batch conditions, because this evaluation is often performed to examine the association constant in molecular imprinting [11, 12, 13]. On the basis of the results from this analysis, the association constant Ka and the number of binding sites N for the polymers were calculated according to the Langmuir model based on the Scatchard plot. The Scatchard plots are shown in Fig. 4 and the results are summarized briefly in Table 2. Table 2 shows that the association constants and number of binding sites were similar on both polymers, and the distribution of the plots was much broader at higher concentrations. On the other hand, for P-Tcba the association constant was higher and the distribution of the plots was narrower than for P-BL at lower concentrations. These notable differences suggest that the polymer prepared by the interval immobilization technique, P-Tcba, has uniTable 2 Binding data determined by Scatchard analysis Ka1 Ka2 N1 N2 (mol–1 L) (mol–1 L) (µmol g–1) (µmol g–1) P-BL (non-MIP, blank) 2.7×104 P-Tcba (MIP 1) 8.9×104

2.3×103 3.2×103

1.4 0.82

8.1 7.6

The value of Ka and N were determined from the Scatchard plots of each polymer

1. Nilsson K, Lindell J, Norlow O, Sellergren B (1994) J Chromatogr A 680:57 2. Kempe M, Mosbach K (1995) J Chromatogr A 691:317 3. Kempe M, Fischer L, Mosbach K (1993) J Mol Recognit 6:25 4. Spivak D, Shea KJ (1999) J Org Chem 64:4627 5. Haginaka J, Sanbe H (2000) Anal Chem 72:5206 6. Yoshizako K, Hosoya K, Iwakoshi Y, Kimata K, Tanaka N (1998) Anal Chem 70:386 7. Hosoya K, Yoshizako K, Sasaki H, Kimata K, Tanaka N (1998) J Chromatogr A 828:91 8. Hosoya K, Iwakoshi Y, Yoshizako K, Kimata K, Tanaka N, Takehira H, Haginaka J (1999) J High Resolut Chromatogr 22: 256 9. Hosoya K, Yoshizako K, Kubo T, Ikegami T, Tanaka N, Haginaka J (2002) Anal Sci 18:55 10. Kubo T, Hosoya K, Watabe Y, Ikegami T, Tanaka N, Sano T, Kaya K (2003) J Chromatogr A 987:389 11. Sajionz P, Kele M, Zhong G, Sellergren B, Guiochon G (1998) J Chromatogr A 810:1 12. Quaglis M, Chenon K, Hall AJ, Lorenzi ED, Sellergren B (2001) J Am Chem Soc 123:2146 13. Takeuchi T, Mukawa T, Matsui J, Higashi M, Shimizu KD (2001) Anal Chem 73:3869 14. Kaya K, Watanabe MM (1994) Microbiol Cult Coll June, Review 15. Codd GA, Metcalf JS, Kaya K (2001) J AOAC Int 84:1626 16. Hunter PR (1992) J Med Microbiol 36:301 17. Elder GH, Hunter PR, Codd GA (1993) Lancet 341:1519 18. Harada K, Ohtani I, Terao K (1994) Toxicon 32:73 19. Hawkins PR, Chandrasena NR, Fralconer IR (1997) Toxicon 35:341 20. Li R, Carmichael WW, Brittain S, Kaya K, Watanabe MM (2001) Toxicon 39:973