Determination of cyclodextrins and branched cyclodextrins by reversed-phase chromatography with pulsed amperometric detection and a membrane reactor

ANALYTICALBIOCHEMISTRY

179,336-340

(1989)

Determination of Cyclodextrins and Branched Cyclodextrins by Reversed-Phase Chromatography with Pulsed Amperometric Detection and a Membrane Reactor Jun Haginaka,’ Yuki Nishimura, Junko Wakai, Hiroyuki Kyoko Koizumi, and Tetsuo Nomura*

Yasuda,

Faculty of Pharmaceutical Sciences,Mukogawa Women’sUniversity, 11-68,Koshien Kyuban-cho, Nishinomiya, Hyogo 663,Japan; and *Scientific Instrument Department, Abe Trading Company, Ltd., 6-32,3-chome, Nakanoshima, Kita-ku, Osaka530, Japan

Received

November

14,198S

A high-performance liquid chromatographic method has been developed for the determination of cyclodextrins (CDs) and branched CDs. The method involves their separation on a reversed-phase column using a mixture of water and acetonitrile as an eluant, eluant pH modification with a cation-exchange membrane reactor surrounded by 1.5 M sodium hydroxide solutions, and pulsed amperometric detection with a gold working electrode. The calibration graphs constructed by peak height versus injected amount were linear over the ranges 50-1000 pmol. The detection limits for CDs and branched CDs were about l-5 pmol at a signal-tonoise ratio of 3. The method was successfully applied to the assay of &CD in serum samples. o 1999 Academic Press.

Inc.

Triple pulsed amperometric detection (PAD)’ with a platinum working electrode for carbohydrates was first introduced by Hughes and Johnson (1). Rocklin and Pohl developed an assay method for carbohydrates using direct separation by anion-exchange chromatography combined with PAD with a gold working electrode instead of a platinum electrode (2). Recently, HPLC procedures based on PAD with a gold working electrode have been applied to the determination of reducing and nonreducing sugars (3-8). PAD is very sensitive for CHOH-bearing compounds such as carbohydrates and alcohols, but has the disadvantage of its separation mode 1 To whom correspondence should be addressed. ’ Abbreviations used: PAD, pulsed amperometric cyclodextrin. 336

detection;

CD,

being limited to a hydroxide form anion-exchange column with a highly alkaline eluant. Hughes and Johnson (9) reported an HPLC method involving separation on a cation-exchange resin loaded with calcium using water as an eluant, modification of the eluant pH to alkali by delivering sodium hydroxide solution with a pump, and detection by pulsed amperometry. However, the detector noise caused by flow noise and mixing inhomogeneity prevented sensitive detection of carbohydrates. In order to expand the use of pulsed amperometry into the detection of carbohydrates separated by the various modes, we tried modifying the eluant pH to alkali using a membrane reactor and strong alkaline solutions. As a result, we obtained a sensitive, precise, and accurate HPLC method based on PAD for the determination of carbohydrates (from monosaccharides to trisaccharides) separated by size exclusion or normal phase mode (10). Although CY-,/3-, and y-cyclodextrins (CDs) and their branched CDs can be well separated on a CIs column, giving symmetrical peaks (ll), they, except for a-CD and branched a-CDs, give tailing bands on an anion-exchange column in the hydroxide form (7). This paper introduces an HPLC procedure for the determination of CDs and branched CDs separated on a Crs column using a mixture of water and organic modifier as an eluant, based on PAD followed by eluant pH modification using a cation-exchange membrane reactor and sodium hydroxide solution. Application of this new method to the assay of P-CD in serum samples is discussed. MATERIALS

AND

METHODS

Reagents and Materials The CDs, a gift from Sanraku (Fujisawa, Japan), were purified by recrystallization from hot water. Maltosyl 0003~2697/89 $3.00 Copyright 0 1989 by Academic Press, Inc. All rights of reproduction in any form reserved,

CHROMATOGRAPHIC

DETERMINATION

(G,)-, maltotriosyl (G3)-, and maltotetraosyl (G,)-CDs were synthesized from CDs and maltooligosaccharides (degree of polymerization, 2-4) with the reverse action of Pseudomonas isoamylase (12) or Klebsiellu aerogenes pullulanase (13). Glucosyl (Gi)-CDs were prepared from G,- or G,-CDs by hydrolysis with purified Rhizopus delemar glucoamylase GIII (14). Stock solutions of CDS and branched CDs (2 X 10e4 M) were prepared in water and diluted with the eluant before injection onto the column. Control serum (Control Serum I) was obtained from Wako Pure Chemical Industries (Osaka, Japan). Acetonitrile, methanol, and tetrahydrofuran of HPLC grade were obtained from Nacalai Tesque (Kyoto, Japan). Water prepared by a NANOpure II unit (Barnstead, Boston, MA) was used for the preparation of the sample and the eluant. The cation-exchange membrane (AMMS-1) was obtained from Dionex (Sunnyvale, CA). Chromatography The HPLC system was composed of a Model 4000i pump and a pulsed amperometric detector (PAD-l) (both from Dionex) equipped with a gold working electrode and a silver-silver chloride reference electrode. The HPLC columns used were column A, YMC-Pack A312 ODS (5 pm, 150 X 6 mm i.d.) (Yamamura Chemical, Kyoto, Japan), for the separation of CDs and branched CDs; and column B, LiChrospher 100 RP-18e (5 pm, 125 X 4 mm i.d.) (E. Merck, Darmstadt, FRG), for @-CD in serum samples. The eluants used were as follows: eluant A, water including 1.5% acetonitrile; eluant B, water including 2.6% acetonitrile; eluant C, water including 2.0% acetonitrile; eluant D, water including 3.5% acetonitrile. Eluants A, B, and C were used for isocratic elution of (Y-, @-, and y-CDs, respectively, together with the branched CDs. Eluants A and D were used for the gradient elutions of a-CD and branched (Y-CDS; linear gradient: O10 min, eluant A, lo-40 min, O-100% eluant D. For the separation of ,&CD in serum samples, eluant A was used. The flow rate was maintained at 0.5 ml/min. The separation was carried out at 35°C using a TU-310 column oven (Japan Spectroscopic Co., Tokyo, Japan) except for the separation of P-CD in serum samples at ambient temperature. A C-R6A Chromatopac (Shimadzu, Kyoto, Japan) was used for integrating and recording chromatograms. A sodium hydroxide solution (whose concentration was 1.5 M) was used as the eluant pH modifier and delivered to a cation-exchange membrane (AMMS1) at a flow rate of 1 ml/min by pressurizing a reagent reservoir with about 2.5 psi nitrogen. Taking the signalto-noise ratio into account, we used the following applied pulse potentials and duration times for detection: E, = 100 mV (ti = 120 ms); Ez = 600 mV (t2 = 120 ms); and Es = -800 mV (t3 = 300 ms), as the optimal conditions. The PAD response time was set at 1 s. A 50-~1 portion of the sample was loaded onto the column.

OF

Preparation

CYCLODEXTRINS

337

of Serum Samples

A known amount of &CD was dissolved in human control serum. The serum samples were ultrafiltered using a Molcut II (Nihon Millipore, Tokyo, Japan) with a low binding filter (molecular weight cutoff 10,000). A 50~1 portion of the ultrafiltrate was loaded onto the column. RESULTS

AND

DISCUSSION

Separation and Detection In a previous paper (lo), we reported that when an eluant is passed through a cation-exchange membrane surrounded by a strong alkaline solution, it is more easily alkalized with both the exchange of a sodium ion with a hydrogen ion and the permeation of a hydroxide ion (which is a forbidden ion for a cation-exchanger) than when an anion-exchange membrane is used. Another advantage of the cation-exchange membrane is that a carbohydrate anion (which occurs in a strong alkaline solution because the hydroxyl groups of a carbohydrate have pK, values over the range of 12 to 14) (15) is a forbidden ion for transport. Previously (ll), CDs and branched CDs were separated on an ODS silica column using a mixture of water and methanol as an eluant. When methanol was used as an organic modifier to the eluant, the background current was large and the baseline showed drifting and much noise, or the peak obtained was negative, because of the PAD response of methanol. When a mixture of water and tetrahydrofuran was used as the eluant, the PAD response was lower than that with a mixture of water and acetonitrile. Thus, acetonitrile was selected as the organic modifier to the eluant. The optimum conditions for modifying the eluant pH to alkali were examined with respect to the concentration and flow rate of the alkali. Figure 1 shows the effect of the concentration of sodium hydroxide solution (at a flow rate of 1.0 ml/ min) on the response of P-CD and branched P-CDs. An increase in baseline noise was observed at a sodium hydroxide concentration above 2.0 M. Corresponding results were obtained with the use of the potassium hydroxide solution. The response obtained was independent of the flow rate of sodium hydroxide solutions between 0.5 and 2.0 ml/min. Thus, the optimum conditions employed were 1.5 M sodium hydroxide solution at a flow rate of 1.0 ml/min, taking into account delivery of the reagent solutions by applying pressure to the reagent reservoir. Figures 2A-2C show chromatograms of cy-, p-, and yCDs, respectively, together with branched CDs (injected amount, 100 pmol each). Table 1 gives the relative PAD response estimated from the peak areas of chromatograms in Fig. 2. The branched CDs gave higher PAD responses per glucose unit than the neat CDs. PAD responses of CDs and branched CDs in the present study

338

HAGINAKA

ET

AL. TABLE

1

Relative PAD Response of CDs and Branched CDs”

CD

1

0

Concentration

2 of NaOH

3 (W)

FIG. 1. Effect of sodium hydroxide concentration on the PAD response of @-CD and branched @-CDs. P-CD and branched &CDs (100 pmol each) were injected onto a column, and the peak heights obtained were plotted. Chromatographic conditions: column, column A; eluant, eluant B. Other conditions, see Materials and Methods. A sodium hydroxide solution used as the eluant pH modifier was delivered to a cation-exchange membrane (AMMS-1) at a flow rate of 1 ml/min. Key: p, &CD; G1, G,-B-CD; Ga, G,-P-CD; GI, G&CD; Gq, G&-CD.

Y&-YG-YGl-YG-Y-

0.37 0.93 1.4 1.6 1.7

0.43 0.93 1.2 1.2 1.2

1.0 1.2 1.6 1.6 2.1

1.0 1.1 1.2 1.1 1.3

1.2 1.6 2.2 2.4 2.5

1.1 1.2 1.5 1.5 1.5

’ Chromatographic conditions were 100 pmol each. b All the values were normalized

paralleled those found previously using separation by an anion-exchange column and a strong alkaline eluant (7), except for a lower response of wCD in the present study. Figure 3 shows the gradient elution profile of a-CD and branched a-CDs. Although a slight drift in the baseline was observed, the gradient elution could be employed in the present study.

Precision,

Relative PAD response per glucose unit

Relative PAD b response

Linearity,

as in Fig.

2. The

injected

amounts

to P-CD.

and Detection

Limits

Table 2 shows the precision of T-CD and branched yCDs. Figure 4 shows the calibration graphs of y-CD and branched y-CDs constructed with peak height versus injected amount. The graphs were linear with correlation

s

I

15nA

I

0

I

10

I

I

I

I

20

30

40

50

Time FIG. 2. (Y, a-CD; detection

(mid

I

0

I

10

do Time

(mid

3b

I

0

I

10 Time

I

1

20

30

(mid

Isocratic elution profiles of (U-CD and branched (Y-CDS (A), @CD and branched @-CDs (B), and y-CD and branched 8, &CD; y, r-CD; G,, glucosyl; Gx, maltosyl; GB, maltotriosyl; Gq, maltotetraosyl. Injected amounts were 100 pmol was by PAD at 300 nA full scale. Chromatographic conditions, see Materials and Methods.

y-CDs (C). Key: each. In all cases,

CHROMATOGRAPHIC

DETERMINATION

OF

339

CYCLODEXTRINS

I

100

nA

T

hpcted

%

FIG. 4. Fig. 2.

Calibration

graphs

of r-CD

Amount (rmols)

and branched

y-CDs.

Key as in

a

ishing is not necessary. We are now searching for an eluant system which can offer good sensitivity without polishing.

I

i

a-- :

Application

to the Assay of P-CD in Serum Samples

":I

I 0

I 20 Time

i 40 (min)

FIG. 3. Gradient elution profile of or-CD and branched (Y-CDS. Key as in Fig. 2. Injected amounts were 500 pmol each. Detection was by PAD at 1000 nA full scale. Chromatographic conditions, see Materials and Methods.

As CDs have unique properties such as enhancing the bioavailability or improving stability of drugs, they have been used as additives to pharmaceutical preparations. For the biopharmaceutical andpharmacokinetic studies, a sensitive assay method would be required to measure their levels in biological fluids. We tested our present method by applying it to assay P-CD in serum samples. Figures 5A and 5B show chromatograms of P-CD (2.5 nmol) in standard and serum samples, respectively. The

coefficients above 0.998 and passed through the origin. The other CDs and branched CDs gave similar results. The detection limits for CDs and branched CDs were about 1-5 pmol at a signal-to-noise ratio of 3. One disadvantage of the present method is that the working electrode surface must be polished once or twice a month to prevent a decrease in sensitivity and increases in baseline noise and drift due to the use of acetonitrile in the eluant. In a system using a mixture of water and tetrahydrofuran for the eluant, such frequent polA

TABLE

2

Precision of the Assay of -y-CD and Branched r-CDs” Injected CD

amount

(pmol)

100

500

2.2 3.5 3.3 4.0 2.2

1.1

v Y-

G--fG--YG,--V G-Y-

a Relative

standard

deviation

of five replicates.

1.8 2.3 2.1 1.5

I

0

I

10 Time

I

20 (mid

r

0

10 Time

20 (mid

FIG. 5. Chromatograms of P-CD in standard (A) and serum (B) samples. Peak 1, P-CD. Injected amount, 2.5 nmol. Detection was by PAD at 1000 nA full scale. Chromatographic conditions, see Materials and Methods.

340

HAGINAKA

recovery of /?-CD from serum samples was almost 100%. For a P-CD amount of 2.5 nmol, the precision of the assay was 2.5% (relative standard deviation, n = 5). Since amino acids and sulfur compounds as well as carbohydrates can be detected by PAD (16), elution of the background components of serum, especially amino acids in serum, due to the preceding injection could interfere with the assay of ,&CD at a lower concentration. Thus, sample pretreatment in the assay should be performed at the picomole level. ACKNOWLEDGMENT The authors are indebted to Professor S. Hizukuri versity) for preparing the branched CDs.

(Kagoshima

Uni-

REFERENCES 1. Hughes, 22.

S., and Johnson,

2. Rocklin, 1590.

R. D., and Pohl,

D. C. (1981)

Anal.

Chim.

Acta

132, ll-

ET

AL.

5. Hardy, M. R., Townsend, R. R., and Lee, Y. C. (1988) chem.170,54-62. 6. Larew, L. A., and Johnson, D. C. (1988) Anal. Chem.

Anal. 60,

Bio1867-

1872.

7. Koizumi, Chromatogr.

K., Kubota, Y., Tanimoto, 464,303-310.

8. Chen,

L.-M., 2819-2821.

9. Hughes,

Yet,

M.-G.,

S., and Johnson,

and

T., and Okada,

Shao,

D. C. (1982)

M.-C.

(1988)

J. Agric.

Y. (1988) FASEB

Food

Chem.

J.

J. 2, 30,

712-714. 10. Haginaka, 271.

J., and Nomura,

T. (1988)

11. Koizumi, K., Kubota, Y., Okada, and Abe J. (1988) J. Chromatogr. 12. Abe, J., Mizowaki, T. (1986) Carbohydr.

J. Chromatogr.

Y., Utamura, 437,47-57.

N., Hizukuri, S., Koizumi, Res. 154,81-92.

447,

268-

T., Hizukuri,

S.,

K., and Utamura,

S., Kawano, S., Abe, J., Koizumi, K., and Tanimoto, 13. Hizukuri, (1989) Biotecnol. Appl. Biochem. 11,60-73. 14. Abe, J., Nagano, H., and Hizukuri, S. (1985) J. Appl. B&hem.

T. 7,

235-247. C. A. (1983)

3. Ohsawa, K., Yoshimura, Y., Watanabe, A., Tamura, K., and Imaeda, K. (1986) 4. Hardy, M. R., and Townsend, USA 85,3289-3293.

J. Liq. Chromatogr.

6,1577-

S., Tanaka, H., Yokota, Anal. Sci. 2,165168.

R. R. (1988)

Proc.

N&l.

Acad.

Sci.

15. Rendleman, in Chemistry Chem. Sot.,

J. A. (1973) in Carbohydrates in Solution, Advances Series (Gould, R. F. Ed.), Vol. 117, pp. 51-69, Amer. Washington, DC.

16. Johnson, D. C., Polta, J. A., Polta, son, J., Tang, A. P-C., Yeo, I.-H., Sot. Farady Trans. 182,1081-1098.

T. Z., Neuburger, G. G., Johnand Baur, J. (1986) J. Chem.