Immobilization of chitosan in sol–gel phases for chiral open-tubular capillary electrochromatography

Elsevier Editorial System(tm) for Analytica Chimica Acta Manuscript Draft

Manuscript Number: ACA-11-2023R2

Title: Immobilization of chitosan in sol-gel phases for chiral open-tubular capillary electrochromatography

Article Type: Full Length Article

Section/Category: SEPARATION METHODS

Keywords: Capillary electrochromatography; Chiral stationary phase; Chitosan; Open-tubular; Sol-gel Corresponding Author: Dr. Jian-Lian Chen,

Corresponding Author's Institution: China Medical University First Author: Jian-Lian Chen

Order of Authors: Jian-Lian Chen; Hong-Jie Syu Suggested Reviewers: Joseph J. Pesek

Department of Chemistry, San Jose State University [email protected]

Ute Pyell

Department of Chemistry, University of Marburg [email protected]

Norman William Smith

PhaPharmaceutical Science Division,, King's College [email protected]

Hsin-Lung Wu

School of Pharmacy, Kaoshiung Medical University [email protected]

December 26, 2011 Editorial Office

Analytica Chimica Acta

Dear Editor,

We decide not to change the form of last revision, ACA-11-2023R1 and hope the reasons given in the file “Response to Reviews” could be accepted by you and referees.

Thank you for the kind re-consideration of suitability for publication of this manuscript in your esteemed journal.

Sincerely yours,

Jian-Lian Chen Professor,

School of Pharmacy, China Medical University Cover Letter

Comment: The authors have addressed the concerns of reviewers 1 and 3 but not of reviewer 2. This reviewer questioned the apparently negative k'' values, which indicate that each enantiomer's mobility is faster than the sum of its electrophoretic mobility and the electrosomotic mobility of the electrolyte.

The authors in their response cite several papers of their own, and 3 papers by different authors where a negative k' value is observed in LC and attributed to electrostatic repulsion between the analyte and the stationary phase. It should be noted that repulsion alone is not the mechanism repsonsible for the negative k' in packed-column LC, but rather it is due to Donnan exclusion: the exclusion of the analyte from the pore volume of the stationary phase, while the void marker does permeate through the pore volume.

It is by no means clear that Donnan exclusion should lead to decreased retention times for the enantiomers in the current CEC work, since there is not a packed column but an open tubular one (for column I, the authors explicity state that the 'bonded phase was only constructed of a molecular layer of chitosan'). Why then should the (unspecified) neutral marker permeate more slowly than the analyte enantiomers (ignoring the electrophoretic influence)? It does not travel a longer route, as might be the case in packed-column LC.

I concur with reviewer 1 that the neutral marker is in fact interacting with the stationary phase leading to an incorrect value of t0.

The authors clearly have no intention of doing any additional experiments to support their work. However, if the results are to be published in their current form, I believe they must offer a fuller explanation of the negative k'' values along the lines of the above.

Response: Donnan exclusion mechanism is mostly found in ion-exclusion chromatography. When an ion-exclusion column is filled with aqueous buffers as a mobile phase, the water molecules accumulate as hydration spheres around the dissociated functional groups of the polymer support. By analogy with the Donnan membrane equilibrium, the hydrated polymer network behaves as a semi-permeable membrane between the stationary and mobile phases. Water, contained in the “pores” of the support and in the hydration spheres, is immobilized, thus *Response to Reviews

“pores”, but use of this term does not imply that a physical pore exists in the polymeric

structure. Neutral, uncharged molecules penetrate through the “pore”, while similarly charged co-ions are repulsed by the presence of dissociated functional groups immobilized in the stationary phase. [J. Sep. Sci. 2003, 26, 1547–1553; J. Chromatogr. A, 2006, 1118, 19–28] All our studied phases (I, II, and III) were synthesized through sol-gel polymerization. The sol-gel based phases are more hydrophilic than the organic polymer based ones used in ion-exclusion chromatography. There is no reason why a hydrated hypothetical Donnan membrane could not be built in our OT−CEC polymeric phases. Moreover, we did not state that “for column I, the bonded phase was only constructed of a molecular layer of chitosan”, even though we think a molecular layer of chitosan might also build a Donnan membrane. Instead, we stated a conclusion that column I and II phases were superior in the resolution and the analysis time to the phase simply bonded with a molecular layer of chitosan. In addition, the neutral marker was DMSO and had been stated in section 2.4.

1. Sequence in synthetic steps determines the chitosan loading in the sol-gel phases. 1

2. Chitosan moieties bearing carboxylic acid groups dominate the EOF. 2

3. High loading of the chitosan chiral selector caused the high k″ and α values. 3

4. Consider the hydrophobicity of the sol-gel phases in the chiral CEC separations. 4

5. Some sol-gel phases were superior in resolution and time to the monolayered phase. 5

ACA-11-2023-Rev.Highlighted 1

Immobilization of chitosan in sol-gel phases for chiral open-tubular 2

capillary electrochromatography 3

4

Jian-Lian Chen*, Hong-Jie Syu 5

6 7

School of Pharmacy, China Medical University, No. 91 Hsueh-Shih Road, Taichung 40402, 8 Taiwan 9 10 Tel: (886) 4-2205-3366. Fax: (886) 4-2203-1075. 11

E-mail address: [email protected] (J.-L. Chen). 12

13

Abstract: 14

Three different approaches for immobilizing cross-linked chitosan molecules (CS-s) in 15

sol-gel phases to form chiral OT-CEC capillaries were comparatively investigated in this 16

study. To synthesize column I, a bare capillary was first silanized with triethoxysilane (TEOS) 17

and then reacted with the reaction product of 3-glycidyloxypropyltrimethoxysilane (GTS) and 18

CS-s. Column II was prepared by the silanization of a bare capillary with a mixture of TEOS 19

and GTS silanes followed by reaction with CS-s. To obtain column III, all the reagents, 20

including TEOS, GTS, and CS-s were reacted together in a bare capillary. The SEM images 21

showed that the column I phase consisted of two distinct layers, GTS and TEOS sol-gel films, 22

while column II and III phases were homogeneous phases. By elemental analysis, the chitosan 23

contents of the columns were found to decrease in the order column I > II > III, which 24

corresponded to the order of the electroosmotic mobility values obtained from the 25

measurements of the electroosmotic flow in the columns. The retention factor and the 26

selectivity for the chiral separation of phenylglycine enantiomers in the optimized Tris 27

running buffer (100 mM, pH 7.5) also followed this decreasing order. Besides the strength of 28

the interaction with the immobilized functional chitosan, the phasehydrophobicity of the

29

column affected the resolution of enantiomeric samples. The hydrophilic alanine sample could 30

only be resolved by column III, but the hydrophobic tryptophan and catechin enantiomers 31

were better separated by columns I and II. A reverse-phase mechanism has been found in the 32

separations. Furthermore, the resolution and analysis time of column I and II phases were 33

superior to the phase simply bonded with molecular chitosan. 34

35

Keywords: Capillary electrochromatography; Chiral stationary phase; Chitosan; Open-tubular;

36 Sol-gel 37 38 Comment [c1]: Response 2 to Reviewer #3.

1. Introduction 39

The enantioseparation of chiral compounds is of continuous importance in 40

pharmacodynamics research and the pharmaceutical industry. Direct chromatographic 41

separation on a chiral stationary phase (CSP) is now the method of choice for stereoselective 42

analysis, which is a challenging task in separation science [1,2]. Besides HPLC, capillary 43

electrochromatography (CEC) is well-suited to the discovery of new CSPs using appropriate 44

column formats developed using either particulate-packed, monolithic, or open-tubular (OT) 45

columns, and many successful implementations have been reviewed [3−6]. 46

Among the column technologies, an OT column is a relatively straightforward approach 47

that does not require the arduous fabrication of frits, which are necessary in packed column 48

construction, or the precise blending of monomer reagents with suitable porogens, as shown 49

in the cases of monolithic fabrication. Although the OT column has low phase ratios, some 50

chemical bonding strategies, such as the stepwise bonding of avidin or bovine serum albumin 51

(BSA) proteins [7−9], molecularly imprinted polymer [10−12], and sol-gel coating with 52

β–cyclodextrin or calixarene [7,13−15], have been adopted. In general, using macromolecules 53

with plenty of recognition sites and/or using polymerized chiral selectors can increase the 54

chirality chiral selectivity of the OT phase. 55

Chitosan (CS, a functional linear polysaccharide) and its derivatives have been 56

successfully immobilized in an HPLC/CSP system [16−18]. With regard to OT-CEC, 57

chitosan with intrinsic basic properties has mainly been physically adsorbed on the bare 58

capillaries, except for the carboxymethylchitosan covalently modified capillary, to separate 59

bioactive molecules [19−21]. For the chiral separations using chitosan as a fixed chiral 60

selector of CSP, a CEC monolith, which was composed of sol–gel/organic hybrid materials 61

with the chiral selectivities of chitosan and BSA [22], was the only example aside from our 62

recent studies [23,24]. In our previous study, an OT-CEC column was designed to have the 63

nano-sized chitosan copolymerize with methacrylamide and exhibited promising chiral 64

Comment [c2]: Response 3 to

separations of tryptophan, catechin, and α-tocopherol [23]. In another study, chitosan units 65

were cross-linked in monolayered OT-CEC phases to increase the enantiomeric resolutions of 66

tryptophan and catechin [24]. However, until now, chitosan molecules have not been 67

incorporated in a sol-gel OT-CEC phase. 68

In this study, three OT-CEC columns with in situ polymerized or post-modified 69

cross-linked chitosan molecules in sol-gel phases were fabricated and compared with regard 70

to their SEM images, elemental analysis data, electroosmotic flow measurements, and 71

enantiomeric resolutions. The difference in mechanism between the sol-gel capillaries was 72

further discussed based on the electrochromatographic parameters for the enantioseparation of 73

samples with different hydrophobicities, such as the amino acids, including phenylglycine, 74

tryptophan, and alanine, as well as polyphenolic catechin. 75 76 2. Experimental 77 2.1. Materials 78

Most chemicals were of analytical or chromatographic grades. Chitosan (CS; from shrimp 79

shells, practical grade, ≥ 75% deacetylated), 3-glycidyloxypropyltrimethoxysilane (GTS), 80

triethoxysilane (TEOS), sodium tetraborate, phosphoric acid, sodium dihydrogenphosphate, 81

hydrochloric acid, acetonitrile (ACN), and dimethylsulfoxide (DMSO) were purchased from 82

SigmaAldrich (Milwaukee, WI, USA). Boric acid, acetic acid, ammonium carbonate, 83

methanol (MeOH), 1,4-dioxane, and potassium acetate were obtained from Panreac 84

(Barcelona, Spain). Sodium hydroxide, succinic acid, disodium hydrogenphosphate, trisodium 85

phosphate, citric acid, sodium dihydrogen citrate, disodium hydrogen citrate, and trisodium 86

citrate were supplied by Merck (Garmstadt, Germany). Acetone and sodium acetate were 87

obtained from J.T. Baker (Phillipsburg, NJ, USA). Tris(hydroxymethyl)aminomethane (Tris) 88

was obtained from TEDIA (Fairfield, OH, USA). The chemical 89

1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (CDI) was obtained from Acros 90

(Thermo Fisher Scientific, Geel, Belgium). 91

The enantiomeric samples, which include phenylglycine (PG), tryptophan (Trp), alanine 92

(Ala), and catechins, (+)-(2R,3S)- and 93

(─)-(2S,3R)-2-(3,4-dihydroxyphenyl)-3,4-dihydro-1(2H)-benzopyran-3,5,7-triol (their 94

structures are shown in Fig. 1) were purchased from SigmaAldrich (Milwaukee, WI, USA). 95

Sample concentrations were 1.0 mg/mL (Trp in H2O; PG and Ala in MeOH) and 25 μg/mL

96

(catechins in MeOH). Purified water (18 MΩ cm) from a Milli-Q water purification system 97

(Millipore, Bedford, MA, USA) was used to prepare samples and buffer solutions. 98

2.2. Instrumentation

99

The laboratory-built electrophoresis apparatus was consisted of a ± 30 kV high-voltage 100

power supply (TriSep TM-2100, Unimicro Technologies, CA, USA) and a UV-Vis detector 101

(LCD 2083.2 CE, ECOM, Prague, Czech). Electrochromatograms were recorded using a 102

Peak-ABC Chromatography Data Handling System (Kingtech Scientific, Taiwan). Elemental 103

analyses were performed on an elemental carbon-hydrogen-nitrogen analyzer (elementar vario 104

EL III, Hanau, Germany). A field-emission scanning electron microscope (Joel JSM-6700F, 105

Japan) acquired the SEM images at an accelerating voltage of 3.0 kV. 106

2.3. Preparation of capillary columns

107

The preparation of the GTS-CS-s capillary without sol-gel layer but direct attachment of 108

the epoxy silane and subsequent bonding of CS has been described previously [24]. The three 109

approaches to preparing the sol-gel phases and the immobilization of chitosan are illustrated 110

in Fig. 2. 111

2.3.1. Preparation of the TEOS+GTS/CS-s capillary (column I)

112

A new, bare capillary column, 70 cm, (Polymicro Technologies, Phoenix, AZ, USA) with 113

a 375 m O.D. x 75 m I.D. was treated with 1.0 M NaOH and successively washed with 114

pure water, 0.1 M HCl, pure water, and then acetone. The clean, bare capillary was filled 115

with a TEOS solution (1.0 M in dioxane) and then kept in an oven for 1.5 h at 90°C to 116

undergo the silanization. After cooling to room temperature, the silanized capillary was 117

ready for the following sol-gel reaction. 118

A 10 mL CS-s solution was prepared by dissolving succinic acid (8 mg) in water and 119

then adjusting the solution pH to 6.5 with 0.1 M sodium hydroxide solution. After the 120

addition of the CDI (40 mg) condensation agent, the alkaline mixture was stirred at 4°C for 121

30 min and subsequently mixed with chitosan (10 mg) at room temperature. The prepared 122

CS-s solution (1 mL) was then added to the sol-gel precursor, GTS (100 μL), and reacted 123

with the epoxide ring of GTS under ultrasonic agitation for 1 h. As soon as the sonication 124

ceased, the sol-gel solution containing modified GTS silane was placed into the silanized 125

capillary, and then the capillary was heated in a GC oven with a three-stage temperature 126

program, including an initial temperature of 30oC, an intermediate temperature of 100oC 127

(holding 30 min), and a final temperature of 150oC (held for 4 h) at rate of increase of 128

2oC/min. After the heating ended, the capillary was cooled to room temperature and was 129

washed sequentially with MeOH and acetone for 30 min to complete the synthesis of the 130

designated column I. 131

2.3.2. Preparation of TEOS/GTS+CS-s capillary (column II)

132

A mixture of TEOS (80 μL), GTS silanes (100 μL), and H2O (9 μL) was used as the

133

sol-gel solution to coat the capillary wall surface with a silica layer with epoxide-ring 134

functionality that the CS-s molecules could be attached to. The sol-gel mixture was placed 135

into a bare capillary and then heated in a GC oven using the same three-stage temperature 136

program as described in section 2.3.1. After cooling the capillary to room temperature, the 137

CS-s solution was completely filled in the capillary, which was then sonicated for 1 h and 138

washed sequentially with MeOH and acetone for 30 min to complete the synthesis of the 139

designated column II. 140

2.3.2. Preparation of TEOS/GTS/CS-s capillary (column III)

A mixture of TEOS (80 μL), GTS silanes (100 μL), and CS-s solution (1 mL) was 142

sonicated for 1 h before placing it into a bare capillary. After heating in a GC oven using the 143

same three-stage temperature program as described in section 2.3.1, the capillary was washed 144

sequentially with MeOH and acetone for 30 min to complete the synthesis of the designated 145

column III. 146

2.4. CEC conditions

147

Most experiments were conducted using common CZE buffers, including Tris, acetate, 148

citrate, phosphate, ammonium carbonate, and borate buffers within a pH range of 5.5 to 11.5 149

and an ionic concentration range of 10 to 300 mM. ACN and MeOH were added to the buffers 150

as organic modifiers. All prepared buffer solutions for CEC analyses were filtered through a 151

0.45 μm cellulose ester membrane (Adventec MFS, Pleasanton, CA, USA). DMSO was used 152

as the neutral marker. The studied capillary was sequentially washed with methanol, water, 153

and running buffer between each analysis run. Prior to sample injection, a working voltage 154

was applied for 5 min to condition the charge distribution in the column. The prepared test 155

samples were introduced by siphoning using a height difference. The samples were detected 156

by UV light absorption measurements at 195 nm for alanine, 280 nm for catechin, and 214 nm 157

for DMSO, phenylglycine, and tryptophan. 158

159

3. Results and discussion 160

3.1. Characterization of the sol-gel phases

161

3.1.1. SEM image and elementary analysis

162

The GTS-CS-s powder and the cross-sections of columns I, II, and III were observed by 163

SEM, the images of which are shown in Fig. 3. The surface morphology of GTS-CS-s powder, 164

which was obtained after heating the mixture of CS-s and GTS solution in a beaker, is shown 165

in Fig. 3(A) and looks similar to the texture of the material coated on the upper layer of the 166

modified phase in column I, as shown in Fig. 3(B). There were two apparently different 167

coatings in column I, where the GTS-CS-s sol reagent was coated on the first sol-gel layer, 168

which was made of TEOS of approximately 5 μm thickness, and formed a second layer with a 169

thickness of approximately 4 μm. The formed GTS-CS-s layer shown in the central part of the 170

Fig. 3(B) image was relatively thick and the thick layer seems to have been caused by 171

capillary cutting before SEM scanning. By contrast, the cross-sectional morphology of the 172

coatings shown in Fig. 3(C) and 3(D), respectively, for columns II and III indicates the 173

“hardness” of the phase composite of the TEOS and GTS sol reagent mixture and no damage 174

to the integral phases during the cutting of the capillaries. 175

The difference in the morphology of columns II and III determined by SEM is small, but 176

the difference in nitrogen content obtained by elemental analysis (EA) is significant, 1.82% 177

(±0.03, n=5) vs. 1.14% (±0.02, n=5) for column II and III, respectively. As compared with 178

column II and III, column I had the highest nitrogen percentage, 2.64% (±0.03, n=5). Here, 179

the amount of chitosan loaded into these columns correlated with the nitrogen ratio and varied 180

with different synthetic approaches. A comparison between the EA data of columns II and III 181

revealed that the silanization and epoxide-ring opening reaction must occur stepwise, rather 182

than simultaneously, to increase chitosan loading. Furthermore, if a high chitosan loading is 183

intended, a comparison between columns I and II showed that a GTS reagent must undergo 184

the epoxide-ring opening reaction with chitosan before rather than after its silanization with 185

TEOS. In any event, very few nitrogen atoms, 0.12% (±0.04, n=5), were found in the 186

GTS-CS-s capillary, whose bonded phase was only constructed of a molecular layer of 187

chitosan. 188

3.1.2. Measurements of EOF for the sol-gel modified phases

189

To determine the EOF magnitude that contributed to solute migration in the CEC and to 190

examine some of the chemical properties of the modified capillaries, the EOF driven by the 191

capillaries under buffers with different pHs was characterized before the CS-immobilized 192

capillaries were utilized for chiral analyses. The curves shown in Fig. 4 illustrate the 193

dependence of μeo on the pH of the phosphate buffer for the bare fused-silica capillary, the

194

sol-gel capillaries, including columns I, II, and III, and the previously reported GTS-CS-s 195

capillary. 196

As shown in Fig. 4, the curve pattern of the three sol-gel capillaries reached a plateau at 197

higher pH levels and was dissimilar to that of the bare capillary, where the μeo values simply

198

increased with increasing buffer pH. Accordingly, the effect of the residual silanol groups on 199

the surface charges of the modified capillaries could be neglected as the chitosan 200

macromolecules attached to the GTS silane coupling agents might shield most of the silanols. 201

Here, the sheltered silanols could not affect the μeo values, but some of the chitosan moieties

202

bearing carboxylic acid groups derived from succinic acid dominate the EOF. Succinic acid is 203

characterized by the dicarboxylic acid moiety, which could partly act as a bridging agent to 204

cross-link the chitosan units through amidation. The cross-linking reaction enriched the 205

bonded chitosan units’ blanketing the capillary wall surface and consequently enhancing the 206

shielding effect. 207

The plateau curve at higher pH levels also occurred with the GTS-CS-s capillary. If the 208

surface charge on the Cs-s-modified capillary wall was only due to the carboxylate groups of 209

the CS-s molecules, the dissociation of the carboxylic acids would mainly determine the zeta 210

potential or the electroosmotic mobilities (μeo) of the capillaries. As a result, a further

211

examination of the curve pattern in Fig. 4 revealed that the carboxylic acids either in the 212

GTS-CS-s phase or the sol-gel phases would be dissociated within the pH range between 4.5 213

and 7.5. The range correlated to the pKa2 (5.2) of succinic acid at μ = 0.1 [25]. However, the

214

μeo values obtained at pH values higher than 8.0 in the CS-s-modified capillaries were

215

somewhat diverse. Here, column II had higher μeo values than column III. Because the EA

216

data showed that the CS-s content in column II was higher than that in column III, the surface 217

density of succinate ligands on the column II phase would be higher than that on the column 218

III phase. The column II and III phases were similarly created in the TEOS-formed silica 219

matrices, which could not contribute to an increase in zeta potential, but could reduce the μeo

220

value. By contrast, the outermost surface layer of column I was simply constructed from the 221

reaction product mixture of CS-s molecules and GTS silane without involvement of the TEOS 222

silane, and therefore had higher μeo values than columns II and III. Besides, the μeo values of

223

the GTS-CS-s capillary were close to that of column I, as they both have similar surface 224

chemistry. 225

The reproducibility of the capillary fabrication was evaluated using the μeo values

226

measured at pH 7.6 for five runs of the sol-gel capillaries. The RSD values were 4.4±0.6%, 227

3.4±0.4%, and 4.0±0.4%, respectively, for three replicate capillaries, columns I, II, and III. At 228

the 95% confidence level, no significant differences between the replicate columns were 229

observed by the Student’s t-test. 230

3.2. Enantiomeric separation of amino acids

231

3.2.1. Phenylglycine

232

Phenylglycine (PG) enantiomers were used as chiral probes to assess the CEC 233

enantioselectivity of the modified sol-gel capillaries, columns I, II, and III. After testing 234

several types of buffers (described in section 2.4), the best peak shape and resolution of the 235

PG racemate were achieved using a Tris buffer system (100 mM, pH 7.5) and are shown in 236

Fig. 5. As compared with the electrochromatograms in Fig. 5, the longest migration times of 237

the PG solutes were found in column I, although the cathodic EOF of column I was higher 238

than those of columns II and III, as shown in Fig. 4. There may be a stronger chromatographic 239

retention between the column I phase and the PG solutes. 240

Differentiating between the electrophoretic and chromatographic contributions to the CEC 241

separation is essential, particularly in this study, which focuses on the chiral selectivity 242

induced by the fixed chitosan molecules. Adopting the definition formulated by Rathore and 243

Horváth, measurements of electrophoretic migration and chromatographic retention in CEC 244

can be described by a velocity factor (ke″) and a retention factor (k″), respectively [26,27];

245

Comment [c3]: Response 1 to

these terms are expressed in equations (1) and (2): 246 ke″ = eo2 ep   (1) 247 k″ = 02 02 e M2 1 t t k t _   _      _   (2) 248

where μep and μeo2 are the electrophoretic and electroosmotic mobilities. These mobilities can

249

be obtained from open-tubular CE experiments on a bare capillary (column 1) and from the 250

CEC experiments on the CS-immobilized capillary (column 2), respectively, as follows: 251 μep =          01 M1 1 1 1 1 1 t t V L L _d (3) 252 μeo2 = 2 02 2 2 V t L L _d   (4) 253

where L is the total column length, Ld is the distance between the inlet and the detection point,

254

V is the applied voltage, tM is the migration time of the solute, and t0 is the migration time of

255

the neutral marker. The electrochromatographic parameters for the PGs separated under the 256

conditions of Fig. 5 are summarized in Table 1. Here, the pI (6.56 at μ= 0.1) of PG is lower 257

than the pH (7.5) of Tris running buffer [28], leading to the electrophoretic movement of PGs 258

toward the anode and to negative ke″ values. Moreover, the ke″ values of the DL solutes in all

259

of the columns were identical and indicated that the electrophoretic action did not contribute 260

to the enantioseparation. By contrast, chromatographic selectivity due to the different k″ 261

values of the DL solutes contributed to the enantioseparation. The negative k″ values are most 262

likely to arise from the repulsive interaction between the negatively charged PGs and ionized 263

succinate groups on the column phases. In addition, the high loading of the chitosan chiral 264

selector would be responsible for the higher α and N values observed in the column I phase 265

than those in the column II and III phases. 266

3.2.2. Tryptophan

267

Tryptophan (Trp) is more hydrophobic than PG [29]. As shown in Fig. 6, the optimal 268

conditions for the separation of Trp enantiomers varied with the types of columns and, 269

evidently, different amounts of MeOH was required in the Tris running buffers. The addition 270

of MeOH into the running buffers would be expected to affect the chiral selectivity between 271

the enantiomeric solutes in the CEC capillaries; the presence of the organic modifier not only 272

altered the electrophoretic and electroosmotic flows, but it was also of interest to the 273

chromatographic partitioning between the solute molecules and the stationary phases. 274

As shown in Fig. 6(A), the Tris buffer (50 mM) used in column I reached a high level of 275

pH 10.0 but no MeOH was required. At pH 10.0, Trp molecules, with pKa1 (2.35) and pKa2

276

(9.33) at μ = 0.1 [25],will be dissociated into their anionic form. Accordingly, the repulsive 277

interaction between the Trp anions and the negatively charged succinate groups on the column 278

I phase would greatly affect the enantiomer selectivity as in the chiral separation of PGs in 279

section 3.2.1. As compared with the optimal buffer used in column I, the buffer pH level used 280

for column II was lowered to 9.5, and 20% (v/v) MeOH was added to the buffer. These 281

changes would decrease the EOF magnitude and increase the proportion of neutral to ionized 282

Trp molecules. As a consequence, the repulsion between ionic solutes and succinate groups 283

would be reduced, and the retention between neutral solutes and the immobilized chitosan 284

selector would be enhanced. This conversion also exchanged the migration order from L/D to 285

D/L Trp, as shown in Fig. 6(C) and (D), where the Tris buffers were optimized at pH 9.0 with 286

50% MeOH and at pH 8.5 without MeOH, respectively. Furthermore, the increase in retention 287

with increased MeOH percentage from 40 to 50% could be observed for column III, as shown 288

in Fig. 7. If the MeOH percentage was over 50%, the retention would start to decrease as the 289

reverse phase mechanism contributed significantly here. 290

3.2.3. Alanine

291

Alanine (Ala) is more hydrophilic than PG, and its acid dissociation constants are pKa1

(2.35) and pKa2 (9.33) at μ = 0.1 [25,29]. Although Tris buffers of various pH values and

293

MeOH ratios were tried in all of the modified sol-gel columns, only column III could achieve 294

a distinct separation of Ala enantiomers in the optimized conditions, as shown in Fig. 8. In 295

comparison with column III, the lack of resolution in columns I and II could be due to their 296

higher loading of chitosan or their higher hydrophobicity, which is not favorable for 297

interaction with the hydrophilic Ala. 298

In comparison with the phenyl substituent in PG and the indole substituent in Trp, the 299

methyl group in the Ala structure could only provide a little retention with CS-immobilized 300

phases. The plot of k″ versus MeOH percentage is also shown in Fig. 7 and is an inverted 301

U-curve, which was caused by a balance between an increasing ratio of neutral to ionized 302

forms of Ala solutes and an increasing amount of neutral Ala solutes partitioning in MeOH as 303

the MeOH percentage was increased. 304

3.3. Chiral separation of (±)-catechin

305

(+)-(2R,3S)- and (─)-(2S,3R)-catechinsare a category of flavonoids and have different 306

bioavailability and bioactivity [30,31]. Their hydrophobicity is higher than that of the amino 307

acids and they have acid dissociation constants of 8.16 (pKa1) and 9.2 (pKa2) [32]. As shown

308

in Fig. 9, they could be separated in the CS-immobilized columns with Tris running buffers. 309

Of the sol-gel capillaries, columns I and II had a better resolution than the GTS-CS-s capillary. 310

Here, the high loading of the chitosan chiral selector and the hydrophobic characteristic in 311

columns I and II were favorable factors for the separation of (±)-catechins. 312

Although the optimal conditions in different columns differed from each other, the effect 313

of the addition of MeOH into Tris buffer on the retention factor was similar. As shown in Fig. 314

10, the retention factors decreased with the increasing percentage of MeOH modifier. This 315

suggests that the reverse phase mechanism determined the chromatographic retention during 316

the CEC separation. Moreover, the strength of the retention factor decreased as column I > II 317

> III at a given MeOH proportion, as did the amount of chitosan in the columns. 318

319

4. Conclusions 320

Three OT-CEC capillaries, columns I, II, and III, with different approaches of 321

immobilizing chitosan chiral selectors in sol-gel phases were successfully characterized and 322

applied to chiral separations in this study. In addition to observing the morphology of the 323

studied OT-CEC sol-gel capillaries by SEM, the chitosan contents were measured by 324

elemental analyses of the nitrogen percentage and the μeo values were obtained by EOF

325

measurements. The nitrogen percentage and the μeo values decreased in the order of column I

326

> II > III. In the same Tris buffer system, column I, which had the highest loading of chitosan, 327

showed a higher retention factor and selectivity (α) for the enantiomeric separation of 328

phenylglycine than columns II and III. By contrast, column III could resolve the hydrophilic 329

alanine enantiomers but columns I and II could not. For the hydrophobic samples of 330

tryptophan and catechin enantiomers, the MeOH percentage in the running buffer greatly 331

affected the resolutions, and the reversed phase mechanism was found in the column phases. 332

Here, column I and II phases were superior in the resolution and the analysis time to the phase 333

simply bonded with a molecular layer of chitosan. Although the peak performances in some 334

separations did not meet the practical utility requirement, the optimization of the coating 335

thickness and sol-gel composition ratio in the capillaries will be the possible ways to 336

overcome the disadvantages. 337

338

Acknowledgements 339

Support of this work by the National Science Council of Taiwan 340

(NSC−98−2113−M−039−003−MY3) is gratefully acknowledged. 341

References 343

[1] T.J. Ward, B.A. Baker, Anal. Chem. 80 (2008) 4363–4372. 344

[2] M. Lämmerhofer, J. Chromatogr. A, 1217 (2010) 814–856. 345

[3] G. Gübitz, M.G. Schmid, J. Chromatogr. A 1204 (2008) 140–156. 346

[4] B. Preinerstorfer, M. Lämmerhofer, W. Lindner, Electrophoresis 30 (2009) 100–136. 347

[5] Z. Zhang, R. Wu, M. Wu, H. Zou, Electrophoresis 31 (2010) 1457–1466 348

[6] H. Lu, G. Chen, Anal. Methods, 3 (2011) 488–508. 349

[7] R. Dai, L. Tang, H. Li, Y. Deng, R. Fu, Z. Parveen, J. Appl. Polym. Sci. 106 (2007) 350

2041–2046. 351

[8] S.K. Wiedmer, T. Bo, M.-L. Riekkola, Anal. Biochem. 373 (2008) 26–33. 352

[9] J. Olsson, J.G. Blomberg, J. Chromatogr. B 875 (2008) 329–332. 353

[10] S.A. Zaidi, W.J. Cheong, Electrophoresis 30 (2009) 1603–1607. 354

[11] S.A. Zaidi, K.M. Han, D.G. Hwang, W.J. Cheong, Electrophoresis 31 (2010) 1019–1028. 355

[12] S.A. Zaidi, S.M. Lee, W.J. Cheong, J. Chromatogr. A 1218 (2011) 1291–1299. 356

[13] Y. Wang, Z. Zeng, N. Guan, J. Cheng, Electrophoresis, 22 (2001) 2167–2172. 357

[14] C.-Q. Shou, J.-F. Kang, N.-J. Song, Chin. J. Anal. Chem. 36 (2008) 297–300. 358

[15] K. Hu, Y. Tian, H. Yang, J. Zhang, J. Xie, B. Ye, Y. Wu, S. Zhang, J. Liq. Chromatogr. 359

Relat. Technol. 32 (2009) 2627–2641. 360

[16] Y. Liu, H. Zou, J. Haginaka, J. Sep. Sci. 29 (2006) 1440–1446. 361

[17] S.-H. Son, J. Jegal, J. Appl. Polym. Sci. 106 (2007) 2989–2996. 362

[18] C. Yamoto, M. Fujisawa, M. Kamigaito, Y. Okamoto, Chirality 20 (2008) 288–294. 363

[19] X. Huang, Q. Wang, B. Huang, Talanta 69 (2006) 463–468. 364

[20] X. Fu, Y. Liu, W. Li, N. Pang, H. Nie, H. Liu, Z. Cai, Electrophoresis 30 (2009) 365

1783–1789. 366

[21] S. Zhou, J. Tan, Q. Chen, X. Lin, H. Lü, Z. Xie, J. Chromatogr. A 1217 (2010) 367

8346–8351. 368

[22] M. Kato, H. Saruwatari, K. Sakai-Kato, T. Toyo’oka, J. Chromatogr. A 1044 (2004) 369

267–270. 370

[23] J.-L. Chen, K.-H. Hsieh, Electrophoresis 32 (2011) 398–407. 371

[24] J.-L. Chen, Talanta, 85 (2011) 2330–2338. 372

[25] A.E. Martell, R.M. Smith (Eds.), Critical Stability Constants, Plenum Press, New 373

York, 1989. 374

[26] A.S. Rathore, Cs. Horváth, J. Chromatogr. A 743 (1996) 231–246. 375

[27] A.S. Rathore, Cs. Horváth, Electrophoresis 23 (2002) 1211–1216. 376

[28]A.A. Mohamed, F.I. El-Dossoki, H.A. Gumaa, J. Chem. Eng. Data 55 (2010) 673–678. 377

[29] E.S.J. Rudolph, M. Zomerdijk, M. Ottens, L.A.M. van der Wielen, Ind. Eng. Chem. Res. 378

40 (2001) 398–406. 379

[30] J.L. Donovan, V. Crespy, M. Oliveira, K.A. Copper, B.B. Gibson, G. Williamson, Free 380

Radic. Res. 40 (2006) 1029–1034. 381

[31]F. Nyfeler, U.K. Moser, P. Walter, Biochim. Biophys. Acta 763 (1983) 50–57. 382

[32] D.A. El-Hady, N.A. El-Maali, Talanta 76 (2008) 138–145. 383

Figure captions and legends 385

Figure 1. Chemical structures of the chiral samples. 386

Figure 2. Schemes to synthesize the CS-immobilized sol-gel capillaries. 387

Figure 3. SEM images of GTS-CS-s powder (A) and coatings on the inner wall of column I 388

(B), column II (C), and column III (D). 389

Figure 4. Dependence of electroosmotic mobility on buffer pH in different columns. 390

Columns: (▲) a bare fused-silica capillary; (∆) the GTS-CS-s capillary; (●) column I; (♦) 391

column II; and (■) column III. Conditions: BGE, phosphate buffer, 50 mM; neutral marker, 392

DMSO; hydrostatic injection, 5 cm, 2 sec; applied voltage, 15 kV; detection, 214 nm. The (▲) 393

and (∆) data were obtained from [24]. 394

Figure 5. Enantioseparations of (D/L)-PG in the CS-immobilized sol-gel capillaries. Columns: 395

(A) column I (55 cm (50 cm) x 75 μm I.D.); (B) column II (60 cm (55 cm) x 75 μm I.D.); and 396

(C) column III (41 cm (38 cm) x 75 μm I.D.). BGE: Tris buffer, 100 mM, pH 7.5. The applied 397

voltage was 8 kV. Samples: hydrostatic injection of 15 cm for 10 sec and detection at 214 nm. 398

Peaks correspond to (1) L-PG, and (2) D-PG. 399

Figure 6. Enantioseparations of (D/L)-Trp in the CS-immobilized sol-gel capillaries. (A) 400

column I (60 cm (55 cm) x 75 μm I.D.); (B) column II (60 cm (55 cm) x 75 μm I.D.); (C) 401

column III (44 cm (38 cm) x 75 μm I.D.); and (D) GTS-CS-s (60 cm (55 cm) x 75 μm I.D.). 402

BGE: Tris buffer, 50mM, (A) pH 10.0; (B) pH 9.5 with 20% MeOH; (C) pH 9.0 with 50% 403

MeOH; and (D) pH 8.5. The applied voltage was 10 kV except for (D), which was 15 kV. 404

Samples: hydrostatic injection of 15 cm for 10 sec and detection at 214 nm. Peaks correspond 405

to (1) L-Trp, and (2) D-Trp. 406

Figure 7. Effect of the addition of MeOH into the Tris buffer on the retention factor (k″) of 407

(D/L)-Trp and (D/L)-Ala in the column III. (◊) and (■) represent the k″ values of (d)-Trp and 408

(l)-Trp, respectively, observed under the conditions in Fig. 6(C). (○) and (▲) represent the k″ 409

values of D-Ala and L-Ala, respectively, observed under the conditions in Fig. 8. 410

Figure 8. Enantioseparations of (D/L)-Ala in column III (44 cm (39 cm) x 75 μm I.D.). BGE: 411

Tris buffer, 50 mM, pH 10.5. The applied voltage was 15 kV. Samples: hydrostatic injection 412

of 15 cm for 10 sec and detection at 195 nm. Peaks correspond to (S) MeOH solvent, (1) 413

D-Ala, and (2) L-Ala. 414

Figure 9. Enantioseparations of (±)-catechins in the CS-immobilized sol-gel and GTS-CS-s 415

capillaries. (A) column I (60 cm (55 cm) x 75 μm I.D.); BGE: Tris (pH 8.5, 50 mM) with 416

MeOH (50%, v/v); 20 kV applied voltage; (B) column II (60 cm (55 cm) x 75 μm I.D.); BGE: 417

Tris (pH 9.5, 50 mM) with MeOH (70%, v/v); 15 kV applied voltage; (C) column III (45 cm 418

(40 cm) x 75 μm I.D.); BGE: Tris (pH 9.5, 50 mM) with MeOH (40%, v/v); 15 kV applied 419

voltage; (D) GTS-CS-s (65 cm (60 cm) x 75 μm I.D.); BGE: Tris (pH 8.5, 100 mM) with 420

MeOH (60%, v/v); 15 kV applied voltage. Samples: hydrostatic injection of 15 cm for 10 sec 421

and detection at 280 nm. Peaks correspond to (S) MeOH solvent, (1) (–)-catechin, and (2) 422

(+)-catechin. 423

Figure 10. Effect of the addition of MeOH into the Tris buffer on the retention factor (k″) of 424

(±)-catechins in the CS-immobilized sol-gel capillaries. (♦)/(◊), (▲)/(∆), and (■)/(□) were 425

observed under the conditions in Fig. 9(A), 9(B), and 9(C), respectively. Black-filled symbols, 426

(♦), (▲), and (■), and white-filled symbols, (◊), (∆), and (□), represent the k″ values of (–)- 427

and (+)-catechin, respectively. 428

Immobilization of chitosan in sol-gel phases for chiral open-tubular

capillary electrochromatography

3

Jian-Lian Chen*, Hong-Jie Syu 4

5 6

School of Pharmacy, China Medical University, No. 91 Hsueh-Shih Road, Taichung 40402, 7 Taiwan 8 9 Tel: (886) 4-2205-3366. Fax: (886) 4-2203-1075. 10

E-mail address: [email protected] (J.-L. Chen). 11

12 *Manuscript

Abstract:

13

Three different approaches for immobilizing cross-linked chitosan molecules (CS-s) in 14

sol-gel phases to form chiral OT-CEC capillaries were comparatively investigated in this 15

study. To synthesize column I, a bare capillary was first silanized with triethoxysilane (TEOS) 16

and then reacted with the reaction product of 3-glycidyloxypropyltrimethoxysilane (GTS) and 17

CS-s. Column II was prepared by the silanization of a bare capillary with a mixture of TEOS 18

and GTS silanes followed by reaction with CS-s. To obtain column III, all the reagents, 19

including TEOS, GTS, and CS-s were reacted together in a bare capillary. The SEM images 20

showed that the column I phase consisted of two distinct layers, GTS and TEOS sol-gel films, 21

while column II and III phases were homogeneous phases. By elemental analysis, the chitosan 22

contents of the columns were found to decrease in the order column I > II > III, which 23

corresponded to the order of the electroosmotic mobility values obtained from the 24

measurements of the electroosmotic flow in the columns. The retention factor and the 25

selectivity for the chiral separation of phenylglycine enantiomers in the optimized Tris 26

running buffer (100 mM, pH 7.5) also followed this decreasing order. Besides the strength of 27

the interaction with the immobilized functional chitosan, the hydrophobicity of the column 28

affected the resolution of enantiomeric samples. The hydrophilic alanine sample could only be 29

resolved by column III, but the hydrophobic tryptophan and catechin enantiomers were better 30

separated by columns I and II. A reverse-phase mechanism has been found in the separations. 31

Furthermore, the resolution and analysis time of column I and II phases were superior to the 32

phase simply bonded with molecular chitosan. 33

34

Keywords: Capillary electrochromatography; Chiral stationary phase; Chitosan; Open-tubular;

35

Sol-gel 36

1. Introduction

38

The enantioseparation of chiral compounds is of continuous importance in 39

pharmacodynamics research and the pharmaceutical industry. Direct chromatographic 40

separation on a chiral stationary phase (CSP) is now the method of choice for stereoselective 41

analysis, which is a challenging task in separation science [1,2]. Besides HPLC, capillary 42

electrochromatography (CEC) is well-suited to the discovery of new CSPs using appropriate 43

column formats developed using either particulate-packed, monolithic, or open-tubular (OT) 44

columns, and many successful implementations have been reviewed [3−6]. 45

Among the column technologies, an OT column is a relatively straightforward approach 46

that does not require the arduous fabrication of frits, which are necessary in packed column 47

construction, or the precise blending of monomer reagents with suitable porogens, as shown 48

in the cases of monolithic fabrication. Although the OT column has low phase ratios, some 49

chemical bonding strategies, such as the stepwise bonding of avidin or bovine serum albumin 50

(BSA) proteins [7−9], molecularly imprinted polymer [10−12], and sol-gel coating with 51

β–cyclodextrin or calixarene [7,13−15], have been adopted. In general, using macromolecules 52

with plenty of recognition sites and/or using polymerized chiral selectors can increase the 53

chiral selectivity of the OT phase. 54

Chitosan (CS, a functional linear polysaccharide) and its derivatives have been 55

successfully immobilized in an HPLC/CSP system [16−18]. With regard to OT-CEC, 56

chitosan with intrinsic basic properties has mainly been physically adsorbed on the bare 57

capillaries, except for the carboxymethylchitosan covalently modified capillary, to separate 58

bioactive molecules [19−21]. For the chiral separations using chitosan as a fixed chiral 59

selector of CSP, a CEC monolith, which was composed of sol–gel/organic hybrid materials 60

with the chiral selectivities of chitosan and BSA [22], was the only example aside from our 61

recent studies [23,24]. In our previous study, an OT-CEC column was designed to have the 62

nano-sized chitosan copolymerize with methacrylamide and exhibited promising chiral 63

separations of tryptophan, catechin, and α-tocopherol [23]. In another study, chitosan units 64

were cross-linked in monolayered OT-CEC phases to increase the enantiomeric resolutions of 65

tryptophan and catechin [24]. However, until now, chitosan molecules have not been 66

incorporated in a sol-gel OT-CEC phase. 67

In this study, three OT-CEC columns with in situ polymerized or post-modified 68

cross-linked chitosan molecules in sol-gel phases were fabricated and compared with regard 69

to their SEM images, elemental analysis data, electroosmotic flow measurements, and 70

enantiomeric resolutions. The difference in mechanism between the sol-gel capillaries was 71

further discussed based on the electrochromatographic parameters for the enantioseparation of 72

samples with different hydrophobicities, such as the amino acids, including phenylglycine, 73

tryptophan, and alanine, as well as polyphenolic catechin. 74 75 2. Experimental 76 2.1. Materials 77

Most chemicals were of analytical or chromatographic grades. Chitosan (CS; from shrimp 78

shells, practical grade, ≥ 75% deacetylated), 3-glycidyloxypropyltrimethoxysilane (GTS), 79

triethoxysilane (TEOS), sodium tetraborate, phosphoric acid, sodium dihydrogenphosphate, 80

hydrochloric acid, acetonitrile (ACN), and dimethylsulfoxide (DMSO) were purchased from 81

SigmaAldrich (Milwaukee, WI, USA). Boric acid, acetic acid, ammonium carbonate, 82

methanol (MeOH), 1,4-dioxane, and potassium acetate were obtained from Panreac 83

(Barcelona, Spain). Sodium hydroxide, succinic acid, disodium hydrogenphosphate, trisodium 84

phosphate, citric acid, sodium dihydrogen citrate, disodium hydrogen citrate, and trisodium 85

citrate were supplied by Merck (Garmstadt, Germany). Acetone and sodium acetate were 86

obtained from J.T. Baker (Phillipsburg, NJ, USA). Tris(hydroxymethyl)aminomethane (Tris) 87

was obtained from TEDIA (Fairfield, OH, USA). The chemical 88

1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (CDI) was obtained from Acros 89

(Thermo Fisher Scientific, Geel, Belgium). 90

The enantiomeric samples, which include phenylglycine (PG), tryptophan (Trp), alanine 91

(Ala), and catechins, (+)-(2R,3S)- and 92

(─)-(2S,3R)-2-(3,4-dihydroxyphenyl)-3,4-dihydro-1(2H)-benzopyran-3,5,7-triol (their 93

structures are shown in Fig. 1) were purchased from SigmaAldrich (Milwaukee, WI, USA). 94

Sample concentrations were 1.0 mg/mL (Trp in H2O; PG and Ala in MeOH) and 25 μg/mL

95

(catechins in MeOH). Purified water (18 MΩ cm) from a Milli-Q water purification system 96

(Millipore, Bedford, MA, USA) was used to prepare samples and buffer solutions. 97

2.2. Instrumentation

98

The laboratory-built electrophoresis apparatus was consisted of a ± 30 kV high-voltage 99

power supply (TriSep TM-2100, Unimicro Technologies, CA, USA) and a UV-Vis detector 100

(LCD 2083.2 CE, ECOM, Prague, Czech). Electrochromatograms were recorded using a 101

Peak-ABC Chromatography Data Handling System (Kingtech Scientific, Taiwan). Elemental 102

analyses were performed on an elemental carbon-hydrogen-nitrogen analyzer (elementar vario 103

EL III, Hanau, Germany). A field-emission scanning electron microscope (Joel JSM-6700F, 104

Japan) acquired the SEM images at an accelerating voltage of 3.0 kV. 105

2.3. Preparation of capillary columns

106

The preparation of the GTS-CS-s capillary without sol-gel layer but direct attachment of 107

the epoxy silane and subsequent bonding of CS has been described previously [24]. The three 108

approaches to preparing the sol-gel phases and the immobilization of chitosan are illustrated 109

in Fig. 2. 110

2.3.1. Preparation of the TEOS+GTS/CS-s capillary (column I)

111

A new, bare capillary column, 70 cm, (Polymicro Technologies, Phoenix, AZ, USA) with 112

a 375 m O.D. x 75 m I.D. was treated with 1.0 M NaOH and successively washed with 113

pure water, 0.1 M HCl, pure water, and then acetone. The clean, bare capillary was filled 114

with a TEOS solution (1.0 M in dioxane) and then kept in an oven for 1.5 h at 90°C to 115

undergo the silanization. After cooling to room temperature, the silanized capillary was 116

ready for the following sol-gel reaction. 117

A 10 mL CS-s solution was prepared by dissolving succinic acid (8 mg) in water and 118

then adjusting the solution pH to 6.5 with 0.1 M sodium hydroxide solution. After the 119

addition of the CDI (40 mg) condensation agent, the alkaline mixture was stirred at 4°C for 120

30 min and subsequently mixed with chitosan (10 mg) at room temperature. The prepared 121

CS-s solution (1 mL) was then added to the sol-gel precursor, GTS (100 μL), and reacted 122

with the epoxide ring of GTS under ultrasonic agitation for 1 h. As soon as the sonication 123

ceased, the sol-gel solution containing modified GTS silane was placed into the silanized 124

capillary, and then the capillary was heated in a GC oven with a three-stage temperature 125

program, including an initial temperature of 30oC, an intermediate temperature of 100oC 126

(holding 30 min), and a final temperature of 150oC (held for 4 h) at rate of increase of 127

2oC/min. After the heating ended, the capillary was cooled to room temperature and was 128

washed sequentially with MeOH and acetone for 30 min to complete the synthesis of the 129

designated column I. 130

2.3.2. Preparation of TEOS/GTS+CS-s capillary (column II)

131

A mixture of TEOS (80 μL), GTS silanes (100 μL), and H2O (9 μL) was used as the

132

sol-gel solution to coat the capillary wall surface with a silica layer with epoxide-ring 133

functionality that the CS-s molecules could be attached to. The sol-gel mixture was placed 134

into a bare capillary and then heated in a GC oven using the same three-stage temperature 135

program as described in section 2.3.1. After cooling the capillary to room temperature, the 136

CS-s solution was completely filled in the capillary, which was then sonicated for 1 h and 137

washed sequentially with MeOH and acetone for 30 min to complete the synthesis of the 138

designated column II. 139

2.3.2. Preparation of TEOS/GTS/CS-s capillary (column III)

A mixture of TEOS (80 μL), GTS silanes (100 μL), and CS-s solution (1 mL) was 141

sonicated for 1 h before placing it into a bare capillary. After heating in a GC oven using the 142

same three-stage temperature program as described in section 2.3.1, the capillary was washed 143

sequentially with MeOH and acetone for 30 min to complete the synthesis of the designated 144

column III. 145

2.4. CEC conditions

146

Most experiments were conducted using common CZE buffers, including Tris, acetate, 147

citrate, phosphate, ammonium carbonate, and borate buffers within a pH range of 5.5 to 11.5 148

and an ionic concentration range of 10 to 300 mM. ACN and MeOH were added to the buffers 149

as organic modifiers. All prepared buffer solutions for CEC analyses were filtered through a 150

0.45 μm cellulose ester membrane (Adventec MFS, Pleasanton, CA, USA). DMSO was used 151

as the neutral marker. The studied capillary was sequentially washed with methanol, water, 152

and running buffer between each analysis run. Prior to sample injection, a working voltage 153

was applied for 5 min to condition the charge distribution in the column. The prepared test 154

samples were introduced by siphoning using a height difference. The samples were detected 155

by UV light absorption measurements at 195 nm for alanine, 280 nm for catechin, and 214 nm 156

for DMSO, phenylglycine, and tryptophan. 157

158

3. Results and discussion

159

3.1. Characterization of the sol-gel phases

160

3.1.1. SEM image and elementary analysis

161

The GTS-CS-s powder and the cross-sections of columns I, II, and III were observed by 162

SEM, the images of which are shown in Fig. 3. The surface morphology of GTS-CS-s powder, 163

which was obtained after heating the mixture of CS-s and GTS solution in a beaker, is shown 164

in Fig. 3(A) and looks similar to the texture of the material coated on the upper layer of the 165

modified phase in column I, as shown in Fig. 3(B). There were two apparently different 166

coatings in column I, where the GTS-CS-s sol reagent was coated on the first sol-gel layer, 167

which was made of TEOS of approximately 5 μm thickness, and formed a second layer with a 168

thickness of approximately 4 μm. The formed GTS-CS-s layer shown in the central part of the 169

Fig. 3(B) image was relatively thick and the thick layer seems to have been caused by 170

capillary cutting before SEM scanning. By contrast, the cross-sectional morphology of the 171

coatings shown in Fig. 3(C) and 3(D), respectively, for columns II and III indicates the 172

“hardness” of the phase composite of the TEOS and GTS sol reagent mixture and no damage 173

to the integral phases during the cutting of the capillaries. 174

The difference in the morphology of columns II and III determined by SEM is small, but 175

the difference in nitrogen content obtained by elemental analysis (EA) is significant, 1.82% 176

(±0.03, n=5) vs. 1.14% (±0.02, n=5) for column II and III, respectively. As compared with 177

column II and III, column I had the highest nitrogen percentage, 2.64% (±0.03, n=5). Here, 178

the amount of chitosan loaded into these columns correlated with the nitrogen ratio and varied 179

with different synthetic approaches. A comparison between the EA data of columns II and III 180

revealed that the silanization and epoxide-ring opening reaction must occur stepwise, rather 181

than simultaneously, to increase chitosan loading. Furthermore, if a high chitosan loading is 182

intended, a comparison between columns I and II showed that a GTS reagent must undergo 183

the epoxide-ring opening reaction with chitosan before rather than after its silanization with 184

TEOS. In any event, very few nitrogen atoms, 0.12% (±0.04, n=5), were found in the 185

GTS-CS-s capillary, whose bonded phase was only constructed of a molecular layer of 186

chitosan. 187

3.1.2. Measurements of EOF for the sol-gel modified phases

188

To determine the EOF magnitude that contributed to solute migration in the CEC and to 189

examine some of the chemical properties of the modified capillaries, the EOF driven by the 190

capillaries under buffers with different pHs was characterized before the CS-immobilized 191

capillaries were utilized for chiral analyses. The curves shown in Fig. 4 illustrate the 192

dependence of μeo on the pH of the phosphate buffer for the bare fused-silica capillary, the

193

sol-gel capillaries, including columns I, II, and III, and the previously reported GTS-CS-s 194

capillary. 195

As shown in Fig. 4, the curve pattern of the three sol-gel capillaries reached a plateau at 196

higher pH levels and was dissimilar to that of the bare capillary, where the μeo values simply

197

increased with increasing buffer pH. Accordingly, the effect of the residual silanol groups on 198

the surface charges of the modified capillaries could be neglected as the chitosan 199

macromolecules attached to the GTS silane coupling agents might shield most of the silanols. 200

Here, the sheltered silanols could not affect the μeo values, but some of the chitosan moieties

201

bearing carboxylic acid groups derived from succinic acid dominate the EOF. Succinic acid is 202

characterized by the dicarboxylic acid moiety, which could partly act as a bridging agent to 203

cross-link the chitosan units through amidation. The cross-linking reaction enriched the 204

bonded chitosan units’ blanketing the capillary wall surface and consequently enhancing the 205

shielding effect. 206

The plateau curve at higher pH levels also occurred with the GTS-CS-s capillary. If the 207

surface charge on the Cs-s-modified capillary wall was only due to the carboxylate groups of 208

the CS-s molecules, the dissociation of the carboxylic acids would mainly determine the zeta 209

potential or the electroosmotic mobilities (μeo) of the capillaries. As a result, a further

210

examination of the curve pattern in Fig. 4 revealed that the carboxylic acids either in the 211

GTS-CS-s phase or the sol-gel phases would be dissociated within the pH range between 4.5 212

and 7.5. The range correlated to the pKa2 (5.2) of succinic acid at μ = 0.1 [25]. However, the

213

μeo values obtained at pH values higher than 8.0 in the CS-s-modified capillaries were

214

somewhat diverse. Here, column II had higher μeo values than column III. Because the EA

215

data showed that the CS-s content in column II was higher than that in column III, the surface 216

density of succinate ligands on the column II phase would be higher than that on the column 217

III phase. The column II and III phases were similarly created in the TEOS-formed silica 218

matrices, which could not contribute to an increase in zeta potential, but could reduce the μeo

219

value. By contrast, the outermost surface layer of column I was simply constructed from the 220

reaction mixture of CS-s molecules and GTS silane without involvement of the TEOS silane, 221

and therefore had higher μeo values than columns II and III. Besides, the μeo values of the

222

GTS-CS-s capillary were close to that of column I, as they both have similar surface 223

chemistry. 224

The reproducibility of the capillary fabrication was evaluated using the μeo values

225

measured at pH 7.6 for five runs of the sol-gel capillaries. The RSD values were 4.4±0.6%, 226

3.4±0.4%, and 4.0±0.4%, respectively, for three replicate capillaries, columns I, II, and III. At 227

the 95% confidence level, no significant differences between the replicate columns were 228

observed by the Student’s t-test. 229

3.2. Enantiomeric separation of amino acids

230

3.2.1. Phenylglycine

231

Phenylglycine (PG) enantiomers were used as chiral probes to assess the CEC 232

enantioselectivity of the modified sol-gel capillaries, columns I, II, and III. After testing 233

several types of buffers (described in section 2.4), the best peak shape and resolution of the 234

PG racemate were achieved using a Tris buffer system (100 mM, pH 7.5) and are shown in 235

Fig. 5. As compared with the electrochromatograms in Fig. 5, the longest migration times of 236

the PG solutes were found in column I, although the cathodic EOF of column I was higher 237

than those of columns II and III, as shown in Fig. 4. There may be a stronger chromatographic 238

retention between the column I phase and the PG solutes. 239

Differentiating between the electrophoretic and chromatographic contributions to the CEC 240

separation is essential, particularly in this study, which focuses on the chiral selectivity 241

induced by the fixed chitosan molecules. Adopting the definition formulated by Rathore and 242

Horváth, measurements of electrophoretic migration and chromatographic retention in CEC 243

can be described by a velocity factor (ke″) and a retention factor (k″), respectively [26,27];

these terms are expressed in equations (1) and (2): 245 ke″ = eo2 ep





1

t

t

k

t

_







_

_











_





where μep and μeo2 are the electrophoretic and electroosmotic mobilities. These mobilities can

248

be obtained from open-tubular CE experiments on a bare capillary (column 1) and from the 249

CEC experiments on the CS-immobilized capillary (column 2), respectively, as follows: 250 μep =



















1

1

t

t

V

L

L

V

t

L

L





where L is the total column length, Ld is the distance between the inlet and the detection point, 253

V is the applied voltage, tM is the migration time of the solute, and t0 is the migration time of

254

the neutral marker. The electrochromatographic parameters for the PGs separated under the 255

conditions of Fig. 5 are summarized in Table 1. Here, the pI (6.56 at μ= 0.1) of PG is lower 256

than the pH (7.5) of Tris running buffer [28], leading to the electrophoretic movement of PGs 257

toward the anode and to negative ke″ values. Moreover, the ke″ values of the DL solutes in all

258

of the columns were identical and indicated that the electrophoretic action did not contribute 259

to the enantioseparation. By contrast, chromatographic selectivity due to the different k″ 260

values of the DL solutes contributed to the enantioseparation. The negative k″ values are most 261

likely to arise from the repulsive interaction between the negatively charged PGs and ionized 262

succinate groups on the column phases. In addition, the high loading of the chitosan chiral 263

selector would be responsible for the higher α and N values observed in the column I phase 264

than those in the column II and III phases. 265

3.2.2. Tryptophan

266

Tryptophan (Trp) is more hydrophobic than PG [29]. As shown in Fig. 6, the optimal 267

conditions for the separation of Trp enantiomers varied with the types of columns and, 268

evidently, different amounts of MeOH was required in the Tris running buffers. The addition 269

of MeOH into the running buffers would be expected to affect the chiral selectivity between 270

the enantiomeric solutes in the CEC capillaries; the presence of the organic modifier not only 271

altered the electrophoretic and electroosmotic flows, but it was also of interest to the 272

chromatographic partitioning between the solute molecules and the stationary phases. 273

As shown in Fig. 6(A), the Tris buffer (50 mM) used in column I reached a high level of 274

pH 10.0 but no MeOH was required. At pH 10.0, Trp molecules, with pKa1 (2.35) and pKa2

275

(9.33) at μ = 0.1 [25],will be dissociated into their anionic form. Accordingly, the repulsive 276

interaction between the Trp anions and the negatively charged succinate groups on the column 277

I phase would greatly affect the enantiomer selectivity as in the chiral separation of PGs in 278

section 3.2.1. As compared with the optimal buffer used in column I, the buffer pH level used 279

for column II was lowered to 9.5, and 20% (v/v) MeOH was added to the buffer. These 280

changes would decrease the EOF magnitude and increase the proportion of neutral to ionized 281

Trp molecules. As a consequence, the repulsion between ionic solutes and succinate groups 282

would be reduced, and the retention between neutral solutes and the immobilized chitosan 283

selector would be enhanced. This conversion also exchanged the migration order from L/D to 284

D/L Trp, as shown in Fig. 6(C) and (D), where the Tris buffers were optimized at pH 9.0 with 285

50% MeOH and at pH 8.5 without MeOH, respectively. Furthermore, the increase in retention 286

with increased MeOH percentage from 40 to 50% could be observed for column III, as shown 287

in Fig. 7. If the MeOH percentage was over 50%, the retention would start to decrease as the 288

reverse phase mechanism contributed significantly here. 289

3.2.3. Alanine

290

Alanine (Ala) is more hydrophilic than PG, and its acid dissociation constants are pKa1

(2.35) and pKa2 (9.33) at μ = 0.1 [25,29]. Although Tris buffers of various pH values and

292

MeOH ratios were tried in all of the modified sol-gel columns, only column III could achieve 293

a distinct separation of Ala enantiomers in the optimized conditions, as shown in Fig. 8. In 294

comparison with column III, the lack of resolution in columns I and II could be due to their 295

higher loading of chitosan or their higher hydrophobicity, which is not favorable for 296

interaction with the hydrophilic Ala. 297

In comparison with the phenyl substituent in PG and the indole substituent in Trp, the 298

methyl group in the Ala structure could only provide a little retention with CS-immobilized 299

phases. The plot of k″ versus MeOH percentage is also shown in Fig. 7 and is an inverted 300

U-curve, which was caused by a balance between an increasing ratio of neutral to ionized 301

forms of Ala solutes and an increasing amount of neutral Ala solutes partitioning in MeOH as 302

the MeOH percentage was increased. 303

3.3. Chiral separation of (±)-catechin

304

(+)-(2R,3S)- and (─)-(2S,3R)-catechins are a category of flavonoids and have different 305

bioavailability and bioactivity [30,31]. Their hydrophobicity is higher than that of the amino 306

acids and they have acid dissociation constants of 8.16 (pKa1) and 9.2 (pKa2) [32]. As shown

307

in Fig. 9, they could be separated in the CS-immobilized columns with Tris running buffers. 308

Of the sol-gel capillaries, columns I and II had a better resolution than the GTS-CS-s capillary. 309

Here, the high loading of the chitosan chiral selector and the hydrophobic characteristic in 310

columns I and II were favorable factors for the separation of (±)-catechins. 311

Although the optimal conditions in different columns differed from each other, the effect 312

of the addition of MeOH into Tris buffer on the retention factor was similar. As shown in Fig. 313

10, the retention factors decreased with the increasing percentage of MeOH modifier. This 314

suggests that the reverse phase mechanism determined the chromatographic retention during 315

the CEC separation. Moreover, the strength of the retention factor decreased as column I > II 316

> III at a given MeOH proportion, as did the amount of chitosan in the columns. 317