Journal of Membrane Science 280 (2006) 116–123

Chiral separation of phenylalanine in ultrafiltration throughDNA-immobilized chitosan membranes

Yuki Matsuoka a, Naoki Kanda b, Young Moo Lee c, Akon Higuchi a,b,∗a Postgraduate Course of Applied Chemistry, Seikei University, Tokyo 180-8633, Japanb Department of Materials and Life Science, Seikei University, Tokyo 180-8633, Japan

c National Research Laboratory for Membranes, School of Chemical Engineering, Hanyang University, Seoul 133-791, South Korea

Received 29 August 2005; received in revised form 30 December 2005; accepted 4 January 2006Available online 13 February 2006

Abstract

Ultrafiltration experiments for the chiral separation of racemic phenylalanine were performed with DNA-immobilized chitosan membraneshaving various pore sizes. Atomic analysis on the membranes showed that the chitosan membranes covalently bound six times more DNA thanthe cellulose membranes used in our previous study [A. Higuchi, Y. Higuchi, K. Furuta, B.O. Yoon, M. Hara, S. Maniwa, M. Saitoh, K. Sanui,CpTt(s©

K

1

rpcioipsrpoo

8

0d

hiral separation of phenylalanine by ultrafiltration through immobilized DNA membranes, J. Membr. Sci. 221 (2003) 207–218]. d-Phenylalaninereferentially permeated through DNA-immobilized chitosan membranes with a pore size <6.4 nm [molecular weight cut-off (MWCO) <67,000].he binding affinity of a specific enantiomer due to the pore size of the DNA-immobilized membranes regulated the preferential permeation of

he enantiomer through the membranes. l-Phenylalanine was adsorbed on the DNA-immobilized chitosan membranes with a pore size <6.4 nmMWCO < 67,000), while there was little difference between the adsorption of d-phenylalanine and l-phenylalanine on the membranes with a poreize >6.4 nm (MWCO > 67,000). The DNA-immobilized chitosan membranes were categorized as channel type membranes.

2006 Elsevier B.V. All rights reserved.

eywords: Chiral separation; Membrane; DNA; Chitosan; Ultrafiltration

. Introduction

Many drug, pharmaceutical and flavoring compounds areacemic mixtures with chiral isomers having nearly identicalhysical and chemical properties. However, the enantiomers ofhiral therapeutic chemicals often exhibit significant differencesn toxicity. An example of this phenomenon was the tragedy thatccurred from the use of thalidomide in the early 1960s wheret was discovered only the (R)-enantiomer of thalidomide hadain relieving effect while the (S)-enantiomer was the cause oferious deformities in unborn children. The importance of chi-al molecules for use in production of pharmaceuticals and foodroducts has generated a considerable demand for developmentf separation techniques appropriate for large-scale purificationf chiral molecules from racemic mixtures [2–6].

∗ Corresponding author at: 3-3-1 Kichijoji Kitamachi, Musashino, Tokyo 180-633, Japan. Tel.: +81 422 37 3748; fax: +81 422 37 3748.

E-mail address: higuchi@st.seikei.ac.jp (A. Higuchi).

Membrane-based chiral separations [1,7–34] are advanta-geous because they avoid issues with low throughout and poorflow distribution in current chromatographic separation tech-niques and are energy-saving techniques and more economicalthan many other separation technologies. Examples of suchtechniques that facilitate industrial scale chiral separations areultrafiltration using chiral porous membranes with a channeltype permeation [1,15–24,35] and affinity-based chiral ultra-filtration [22–30,36,37]. Chiral separation in the affinity-basedultrafiltration method is achieved by the adsorption of one spe-cific enantiomer on the membranes with a higher binding affinitythan the opposite enantiomer, while the separation in the chan-nel type permeation occurs by the differential permeation of theenantiomers through the membranes.

In our previous studies, ultrafiltration experiments for thechiral separation of racemic amino acids were performed in asolution system (i.e., affinity ultrafiltration) [22–26] and mem-brane systems [22–25] using albumin or DNA as chiral recog-nition sites. Our previous study showed that DNA has a higherbinding affinity for l-phenylalanine (S-phenylalanine) than for

376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2006.01.013

Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123 117

d-phenylalanine (R-phenylalanine), with the equilibrium bind-ing constant for the l-form shown to be higher than the constantfor the d-form. Therefore, we investigated the chiral separa-tion of the amino acid through immobilized DNA membranesprepared from cellulose dialysis membranes with differentmolecular weight cut-offs (MWCO) [25,35]. d-Phenylalaninepreferentially permeated through DNA-immobilized cellu-lose membranes with a pore size <2.0 nm (MWCO < 5000)while l-phenylalanine preferentially permeated through DNA-immobilized cellulose membranes with a pore size >2.0 nm(MWCO > 5000) [35]. The pore size of the DNA-immobilizedcellulose membranes regulated preferential permeation of theenantiomers through the membranes. We also found that thechiral separation of the racemic phenylalanine solution wasenhanced by increasing the amount of NH2 residues in theDNA-immobilized cellulose membranes. The NH2 residuescontributed to an increase in the amount of bound Platinum (Pt)complex present on the membranes, which then led to a greateramount of immobilized DNA [1]. These results led us to the useof chitosan, which has a higher degree of NH2 residues com-pared with CNBr-activated cellulose, as base materials for theDNA-immobilized membranes.

In this study, we prepared chitosan ultrafiltration membraneswith various pore sizes. We then performed ultrafiltration exper-iments for the chiral separation of racemic phenylalanine usingDNA-immobilized chitosan membranes.

2

2

pcJScrp

2.2. Preparation of chitosan membranes

Chitosan was dissolved in 2.1 wt.% aqueous acetic acid solu-tion, and the concentration was adjusted to 1.38 wt.%. PEG2000was then added to the chitosan solution. The casting solutionprepared for CS-X membranes (X = 0, 10, 25, 50, 100 and 150)contains PEG2000, which amount is 0, 10, 25, 50, 100 and 150%compared to that of chitosan in the casting solution. The chi-tosan solutions were stirred at 50 ◦C for 3 days. The solutionswere then cast into flat Petri dishes with a diameter of 9 cm at30 ◦C. After drying the chitosan membranes in the Petri dishesat 25 ◦C, the membranes were immersed into a 4 wt.% NaOHsolution for 24 h to be neutralized. Finally, the membranes weredetached from the Petri dishes and washed in ultrapure waterfour times to remove PEG2000 and stored in ultrapure water at4 ◦C.

2.3. Preparation of DNA-immobilized chitosan membranes

The chitosan membranes were immersed in 0.2 mol/L ofaqueous K2[PtCl4] solution for 24 h at 25 ◦C in the dark[25,35]. The Pt-activated chitosan membranes were immersedin 1000 ppm DNA buffer (5 mmol/L tris(hydroxymethyl)-aminomethane HCl/0.5 mmol EDTA buffer) solution for 24 h atpH 7.4, 25 ◦C, and the DNA-immobilized chitosan membranesw

2m

amMtaBpt

mob

. Experimental

.1. Materials

Chitosan (Chitosan 1000, 800-1500 cp, 039-14422) andolyethylene glycol 2000 (PEG2000, 165-09105) were pur-hased from Wako Pure Chemical Industries, Ltd (Tokyo,apan). DNA (from salmon testes, D-1626) was purchased fromigma Chemical Co. (St. Louis, MO, USA). Other chemi-als, purchased from Tokyo Chemical Co. (Tokyo, Japan), wereeagent grade and were used without further purification. Ultra-ure water was used throughout the experiments.

Fig. 1. Synthesis of DNA-im

ere finally prepared as shown in Fig. 1.

.4. Estimation of MWCO of DNA-immobilized chitosanembranes

Vitamin B12, insulin, pepsin, bovine serum albumin (BSA)nd immunoglobulin � (IgG) were used as MW markerolecules and purchased from Sigma Chemical Co. (St. Louis,O, USA). The marker molecules were dissolved in 5 mmol/L

ris(hydroxymethyl)aminomethane HCl/0.5 mmol EDTA bufferdjusted to pH 7.4 with 1 mol/L acetic acid. 200 ppm vitamin12 solution, 1000 ppm pepsin, BSA and IgG solutions were pre-ared. 10 mL of each marker molecule solution was ultrafilteredhrough the chitosan membranes (CS, CS-10, CS-25, CS-50,

ilized chitosan membranes.

118 Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123

CS-100, and CS-150) using a batch-type ultrafiltration (UF)apparatus (UHP-25K, Advantec MFS, Dublin, CA, USA) undera transmembrane pressure (�p) of 0.3 MPa at 25 ◦C [25,35].The permeate solution was sampled every 1 g (CS-50, CS-100,CS-150) or every 0.2 g (CS, CS-10, CS-25), and the flux was cal-culated. The concentration of the permeate solution (Cp) and thefeed solution (Cf) were measured using HPLC (JASCO Inter-national Co., Tokyo, Japan). The observed rejection coefficient(R) was calculated by Eq. (1):

R = 1 − Cp

Cf(1)

The molecular weight cut-off (MWCO) of the chitosan mem-branes was determined from the lowest molecular weight of themarker molecules that were rejected more than 90% in the ultra-filtration of the solutions through the chitosan membranes.

2.5. Characterization of the membranes

Atomic analysis of the surface of Pt-immobilized cellulosemembranes [35], DNA-immobilized cellulose membranes [35],Pt-immobilized chitosan membranes and DNA-immobilizedchitosan membranes was performed using XPS (ESCA-3400,Kratos Analytical Ltd, Manchester, UK).

2m

pwbAptecoHCJ

tof

α

wptlflmcp

(3):

α = [Cp(d)/Cp(l)]

[Cfeed(d)/Cfeed(l)](3)

The separation factor in the concentrate solution, αc was alsodefined as the concentration ratio of d-phenylalanine [Ccon(d)] tol-phenylalanine [Ccon(l)] in the concentrate solution after 80%of feed solution was ultrafiltered through CS-50, CS-100 andCS-150 membranes and 40% of feed solution was ultrafilteredthrough CS, CS-10 and CS-25 membranes.

3. Results and discussion

3.1. MWCO of DNA-immobilized chitosan membranes

Ultrafiltration membranes of chitosan with different poresizes, which were used as the base membranes for DNA-immobilized chitosan membranes, were prepared by addingdifferent amounts of PEG2000 as a nonsolvent into the chi-tosan casting solution, because our previous study showed thatthe pore size of the immobilized DNA membranes regulatedpreferential permeation of enantiomers through the membranes[1,35]. The time dependence of flux and rejection through thechitosan membranes with different pore sizes was investigatedbefore the MWCO of the chitosan membranes was estimated.Fts

adCt

mmmbtmb

trit

TM

M

CCCCCC

.6. Chiral separation through DNA-immobilized chitosanembranes

10 mL of racemic phenylalanine solution [6 �M racemichenylalanine (3 �M l-phenylalanine; 3 �M d-phenylalanine)]as ultrafiltered through the DNA-immobilized chitosan mem-ranes using a batch-type ultrafiltration apparatus (UHP-25K,dvantec MFS, Dublin, CA, USA) under a transmembraneressure (�p) of 0.3 MPa at 25 ◦C. The permeate solu-ion was sampled every 1 g (CS-50, CS-100, CS-150) orvery 0.5 g (CS, CS-10, CS-25) until eight samples wereollected, and the flux was calculated. The concentrationf l-phenylalanine and d-phenylalanine was measured usingPLC (JASCO International Co., Tokyo, Japan) with arownpak column (CR(+), Daicel Chemical Co. Ltd, Tokyo,

apan).The separation factor, α, was calculated from the concentra-

ion of samples collected every 1 g (CS-50, CS-100, CS-150)r every 0.5 g (CS, CS-10, CS-25) and was defined to be theollowing Eq. (2):

= (Jd/Jl)

(Cfeed(d)/Cfeed(l))(2)

here Cfeed(l) and Cfeed(d) were the concentrations of the l-henylalanine and d-phenylalanine in the feed solution, respec-ively. Jd and Jl were the fluxes of the d-phenylalanine and-phenylalanine through the membranes, respectively. Since theux of solute was directly related to the concentration in the per-eate [i.e., Jd/Jl = Cp(d)/Cp(l)] where Cp(l) and Cp(d) were the

oncentrations of the l-phenylalanine or d-phenylalanine in theermeate, respectively, α was reduced to be the following Eq.

ig. 2 shows the time dependence of flux [Fig. 2(a)] and rejec-ion [Fig. 2(b)] in the UF of vitamin B12, pepsin, BSA and IgGolutions through CS-25 membranes.

Flux and rejection were found to be constant at a permeatemount >0.6 g. Therefore, the rejection shown in Fig. 3 wasetermined at the permeate amount = 0.8 g for CS, CS-10 andS-25 membranes and 8.0 g for CS-50, CS-100 and CS-150 in

his study, respectively.Fig. 3 shows the dependence of the rejection of the marker

olecules (i.e., vitamin B12, pepsin, BSA and IgG) on theirolecular weight. Rejection increased with the increase of theolecular weight of the marker molecules. The chitosan mem-

ranes prepared with a lower content of PEG2000 in the chi-osan casting solution showed higher rejection of each marker

olecule, which indicated that the MWCO of the chitosan mem-ranes decreased with a lower content of PEG2000.

MWCO of the chitosan membranes, which was defined ashe lowest molecular weight where the marker molecules wereejected more than 90%, was estimated from Fig. 3, and the datas summarized in Table 1. However, we should pay attentionhat MWCO obtained from the rejection of marker molecules

able 1WCO of the chitosan membranes

embranes MWCO

S 12500S-10 43000S-25 48500S-50 67000S-100 <200000S-150 <200000

Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123 119

Fig. 2. Dependence of flux (a) and rejection (b) in the ultrafiltration of vitaminB12, BSA and pepsin solutions through CS-25 membranes. �p = 0.3 MPa, pH7.4 and 25 ◦C. Data are expressed as means ± S.D. of four independent mea-surements.

such as vitamin B12, insulin, pepsin, BSA and IgG is not anabsolute value of MWCO, but a relative scale of MWCO,because adsorption of proteins on the inner surface of poresespecially reduces the pore size of the membranes. Chitosanmembranes with a MWCO range of 10,000–200,000 were pre-pared in this study, and DNA-immobilized cellulose membraneswith a MWCO range of 1000–50,000 were prepared from com-mercially available dialysis membranes in the previous study[1,35].

3.2. Characterization of the DNA-immobilized chitosanmembranes

Atomic analysis of the surface of Pt-immobilized chitosanmembranes and DNA-immobilized chitosan membranes, as wellas Pt-immobilized cellulose membranes and DNA-immobilizedcellulose membranes prepared in the previous study, [1,35] wasperformed using XPS, and the results are shown in Table 2.The atomic mole fraction of Pt to carbon (Pt1s/C1s) on the Pt-immobilized chitosan membranes was detected to be 0.0824,which was a much higher ratio than on the Pt-immobilized

Fig. 3. Dependence of rejection of the marker molecules (i.e., vitamin B12,insulin, pepsin, BSA and IgG) on the molecular weight of the marker molecules.The broken lines in the figure were polynomial equations (n = 3), fitted with Robtained experimental by using a least-squares method. Data are expressed asmeans ± S.D. of four independent measurements.

cellulose membrane. The atomic mole fraction of phosphateatom to carbon (P1s/C1s), which originated from the phos-phate group of DNA, was detected to be 0.0470 on the DNA-immobilized chitosan membrane. This value of P1s/C1s wasalso a higher ratio than that on the DNA-immobilized cellu-lose membranes. Because chitosan membranes have one aminogroup per repeated sugar unit, which is a higher amount of aminogroups than the CNBr-activated cellulose membranes (i.e., pre-cursor membranes of DNA-immobilized cellulose membranes),more Pt and DNA were immobilized compared to that onthe DNA-immobilized cellulose membranes studied previously[1,35].

The atomic mole fraction of chloride to carbon (Cl1s/C1s)on the DNA-immobilized chitosan membranes was found to belower than that on the Pt-immobilized chitosan membranes. Thisindicates that chloride atoms on the Pt-immobilized chitosanmembranes were removed by the reaction between DNA and the

Table 2Atomic analysis on the surface of the membranes by XPS

Membranes Cl1s/C1s N1s/C1s O1s/C1s P1s/C1s Pt1s/C1s

Pt-immobilizedcellulosea

0.0064 0.0146 0.8451 0.0116 0.0056

DNA-immobilizedcellulosea

0.0047 0.0294 0.8332 0.0151 0.0079

P

D

t-immobilizedchitosan (Pt-CS)

0.1194 0.1650 0.6224 0.0260 0.0824

NA-immobilizedchitosan (DNA-CS)

0.0636 0.2271 0.6515 0.0470 0.0795

a MWCO of cellulose membranes is 1000 Da [35].

120 Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123

Fig. 4. Dependence of the separation factor in the permeate solutionthrough DNA-immobilized chitosan membranes (DNA-CS-25) at Cfeed = 6 �M,�p = 0.3 MPa, pH 7.0 and 25 ◦C. The permeate flux was constant during thechiral separation experiment and was 0.005 ± 0.001 m3 m−2 day−1. Data areexpressed as means ± S.D. of four independent measurements.

membranes in the synthesis of the DNA-immobilized chitosanmembranes, which is shown in Fig. 1.

3.3. Chiral separation in DNA-immobilized chitosanmembranes

A racemic mixture of phenylalanine was separated by ultrafil-tration using the DNA-immobilized chitosan membranes (DNA-CS, DNA-CS-25, DNA-CS-50, DNA-CS-100) at �p = 0.3 MPa,pH 7.0 and 25 ◦C. Fig. 4 shows the time dependence of theseparation factor in the permeate solution through the DNA-immobilized chitosan membranes (DNA-CS-25) plotted as afunction of collected permeate at Cfeed = 6 �M. The separationfactor in the permeate solution was greater than unity. This indi-cates that d-phenylalanine preferentially existed in the permeatesolution, which was the same result as the ultrafiltration throughDNA-immobilized cellulose membranes with a small pore size(MWCO < 5000) as reported in the previous study [1].

The concentration ratio of d-phenylalanine to l-phenyl-alanine in the concentrate solution (the ratio of concentrationsin the concentrated feed solution after the ultrafiltration ofracemic phenylalanine) was also plotted against αc in Fig. 4.l-Phenylalanine was present in the concentrate solution withan αc < 1, whereas d-phenylalanine preferentially existed in thepermeate solution with an α > 1. Based on these results, theDiam[bsf

Fig. 5. Dependence of separation factor in the permeate (�, � ) and the con-centrate (©) solutions on the MWCO of the base membranes and on thepore size of the DNA-immobilized chitosan membranes in the ultrafiltrationof racemic phenylalanine through unmodified chitosan membranes (� ) andDNA-immobilized chitosan membranes (�, ©) and at Cfeed = 6 �M. Data areexpressed as means ± S.D. of four independent measurements.

We further investigated whether the pore size of the DNA-immobilized chitosan membranes regulated preferential perme-ation of the enantiomers through the membranes. Ultrafiltrationof a racemic phenylalanine solution was performed throughDNA-immobilized chitosan membranes with various MWCOs.Fig. 5 shows the dependence of the separation factor in the per-meate and the concentrate solutions on the MWCO of the basemembranes and on the pore size of the DNA-immobilized chi-tosan membranes. The pore size of the DNA-immobilized chi-tosan membranes was estimated from the size of the molecularmarker molecules corresponding to the MWCO of the mem-branes. The separation factors in the permeate solution shownin Fig. 5 are measured from the total permeate solution dur-ing the ultrafiltration of racemic phenylalanine solution througheach membrane.

A pore size in the DNA-immobilized chitosan membranesless than 6.4 nm (MWCO < 67,000) led to a separation factor inthe permeate solution greater than unity, and a pore size over6.4 nm (MWCO > 67,000) led to a separation factor less thanunity. The separation factor in the concentrate solution showedexactly the opposite tendency as the separation factor in thepermeate solution.

The separation factor is expected to approach unitythrough the DNA-immobilized membranes having extremelylarger pore sizes such as the membranes of pore sizesover MWCO > 1,000,000 (see Fig. 5), because the interac-titot

NA-immobilized chitosan membranes as well as the DNA-mmobilized cellulose membranes reported previously [1,35]re categorized as channel type membranes and not as affinityembranes as reported for immobilized albumin membranes

22–24] because (α − 1) × (αc − 1) shows a negative value foroth membranes (i.e., the separation factor in the concentrateolution showed exactly the opposite tendency as the separationactor in the permeate solution).

ion between phenylalanine and DNA decreases in the DNA-mmobilized chitosan membranes of larger pore sizes due tohe decrease of internal surface in the pores and the decreasef residual time of phenylalanine in the DNA-immobilized chi-osan membranes.

Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123 121

Table 3Adsorption of l- and d-phenylalanine on the DNA-immobilized and theunmodified chitosan membranes after ultrafiltration, and the flux throughthe DNA-immobilized and unmodified chitosan membranes at Cfeed = 6 �M,�p = 0.3 MPa, pH 7.0 and 25 ◦C

Membranes Adsorption ratios (%) Flux (m3 m−2 day−1)

l-Phenylalanine d-Phenylalanine

CS 4.1 ± 2.0 9.2 ± 2.5 0.008 ± 0.001DNA-CS 51.9 ± 3.0 43.1 ± 3.0 0.004 ± 0.001DNA-CS-25 25.5 ± 3.0 18.8 ± 2.5 0.005 ± 0.001DNA-CS-50 17.3 ± 3.0 19.7 ± 4.0 0.017 ± 0.001DNA-CS-100 1.1 ± 1.0 2.3 ± 1.0 0.018 ± 0.002

Data are expressed as means ± S.D. of four independent measurements.

The flux through the DNA-immobilized and unmodified chi-tosan membranes in the ultrafiltration of a racemic phenylala-nine solution was summarized in Table 3. The flux increasedthrough the DNA-immobilized chitosan membranes havinglarger pore sizes. The decrease of chiral separation through theDNA-immobilized chitosan membranes having larger pore sizesmay be explained by the increase of flux through the mem-branes, because the interaction between phenylalanine and DNAdecreases in the DNA-immobilized chitosan membranes hav-ing larger pore sizes. Because concentration of feed solutionof racemic phenylalanine was extremely dilute in this study(6 �M), flux through the membranes (i.e., DNA-immobilizedand unmodified chitosan membranes) in the ultrafiltration ofracemic phenylalanine solution was found to be the same to thatin the ultrafiltration of phosphate buffer solution. This indicatesthat neither fouling nor concentration polarization generatedon the DNA-immobilized and unmodified chitosan membranes.Therefore, the fouling and concentration polarization did notcontribute to the chiral separation of racemic phenylalanine solu-tion through the DNA-immobilized chitosan membranes.

3.4. Chiral adsorption in DNA-immobilized chitosanmembranes

The amount of phenylalanine (Qm(l) for l-phenylalanine andQ (d) for d-phenylalanine) adsorbed on the DNA-immobilizedcDDpltaf

Q

Q

Ttlbi

phenylalanine than d-phenylalanine. On the other hand, theDNA-immobilized chitosan membranes with a pore size over6.4 nm (MWCO > 67,000), i.e. DNA-CS-50 and DNA-CS-100membranes, adsorbed slightly more d-phenylalanine than l-phenylalanine. The difference of the adsorption amount of d-phenylalanine to l-phenylalanine was less than 5%. The lowadsorption difference between the enantiomers on the DNA-CS-50 and DNA-CS-100 membranes contributed to the lowseparation factors shown in Fig. 5, which probably originatedfrom the larger pore size of the DNA-immobilized chitosanmembranes than that of the DNA-CS, DNA-CS-10 and DNA-CS-25 membranes. High separation factors through the DNA-CS, DNA-CS-10 and DNA-CS-25 membranes originated fromthe high adsorption difference between the enantiomers, as DNAand l-phenylalanine preferentially interact on the small pore sizeof the DNA-immobilized chitosan membranes. This is becausethe DNA-immobilized membranes having smaller pore sizes areexpected to have higher internal surface areas in the membranes,which lead to generate higher adsorption sites of phenylalanine.

When the pore size of the DNA-immobilized membrane isless than 6.4 nm (MWCO < 67,000, CS, CS25 membranes), d-phenylalanine preferentially permeates through the membranes.As chiral separation was not achieved in the chitosan-castmembranes without DNA (see Fig. 5), it is thought that l-phenylalanine binds to DNA on the top surface of the DNA-immobilized chitosan membranes and the free d-phenylalanineppp1qb

iaaib

4

cricb

ifdi(mwtt

mhitosan membranes (DNA-CS, DNA-CS-25, DNA-CS-50,NA-CS-100) and on the chitosan membranes (CS) withoutNA was calculated from the mass balance of the amount of thehenylalanine enantiomer in the initial feed solution (Qf(l) for-phenylalanine and Qf(d) for d-phenylalanine), permeate solu-ion (Qp(l) for l-phenylalanine and Qp(d) for d-phenylalanine)nd concentrate solution (Qc(l) for l-phenylalanine and Qc(d)or d-phenylalanine):

m(l) = Qf(l) − Qc(l) − Qp(l) (4)

m(d) = Qf(d) − Qc(d) − Qp(d) (5)

he results are summarized in Table 3. The unmodified chi-osan (CS) membranes adsorbed more d-phenylalanine than-phenylalanine, while the DNA-immobilized chitosan mem-ranes with a pore size less than 6.4 nm (MWCO < 67,000),.e. DNA-CS and DNA-CS-25 membranes, adsorbed more l-

ermeates through the pores. l-Phenylalanine preferentiallyermeates through the DNA-immobilized membranes with aore size >6.4 nm (MWCO > 67,000, DNA-CS-50, DNA-CS-00 membranes). However, chiral separation was found to beuite small compared to the DNA-immobilized chitosan mem-ranes with a pore size <6.4 nm (MWCO < 67,000).

These results indicate that the pore size of the DNA-mmobilized chitosan membranes regulates preferential perme-tion of a specific enantiomer through the membranes, and that,long with the DNA-immobilized cellulose membranes reportedn the previous study [1,35], they function as channel-type mem-ranes.

. Conclusions

We showed that the DNA-immobilized chitosan membranesan be used for chiral separations. DNA was selected as aecognition site for optical enantiomers as in our previous stud-es [1,20,35]. The DNA-immobilized chitosan membranes areategorized as channel-type membranes and not as affinity mem-ranes.

The pore size of the membranes and the binding affin-ty of a specific enantiomer are the most important factorsor the preparation of channel-type membranes. In this study,-phenylalanine preferentially permeated through the DNA-

mmobilized chitosan membranes with a pore size <6.4 nmMWCO < 67,000), while l-phenylalanine preferentially per-eated through the DNA-immobilized chitosan membranesith a pore size >6.4 nm (MWCO > 67,000). The pore size of

he DNA-immobilized chitosan membranes regulated preferen-ial permeation of the enantiomer through the membranes.

122 Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123

The DNA-immobilized chitosan membranes adsorbed l-phenylalanine preferentially when the membranes had pore size<6.4 nm (MWCO < 67,000), and showed good chiral separa-tion. However, with a pore size >6.4 nm (MWCO > 67,000) onlya small difference between adsorbed l-phenylalanine and d-phenylalanine was measured, and chiral separation was poor.The binding affinity of a specific enantiomer also regulated itspreferential permeation through the membranes.

Atomic analysis on the membranes indicated that the chi-tosan membranes have the advantage of covalent binding ofPt and DNA compared to the cellulose membranes reported inthe previous study [1,35]. This indicates that DNA-immobilizedchitosan membranes are expected to show a higher ability forchiral separation than the DNA-immobilized cellulose mem-branes, because membranes with a higher amount of DNAcan achieve greater chiral separation. However, we experiencedmore difficulty regulating the pore size of the chitosan mem-branes than with the DNA-immobilized cellulose membranes.The DNA immobilized cellulose membrane with a MWCOof 1000 showed excellent chiral separation ability in the pre-vious studies [1,35]. On the contrary, the minimum MWCOof the DNA-immobilized chitosan membranes was as high as12,500 in this study (CS membrane). DNA-immobilized chi-tosan membranes prepared from smaller MWCO such as 1000is expected to demonstrate superior chiral separation of aminoacids.

A

G“CCSpS

R

[8] C. Thoelen, M. De bruyn, E. Theunissen, Y. Kondo, I.G.J. Vankele-com, P. Grobet, M. Yoshikawa, P.A. Jacobs, Membranes based onpoly(�-methyl-l-glutamate): synthesis, characterization and use in chiralseparations, J. Membr. Sci. 186 (2001) 153–163.

[9] M. Yoshikawa, T. Ooi, J. Izumi, Alternative molecularly imprinted mem-branes from a derivative of natural polymer, cellulose acetate, J. Appl.Polym. Sci. 72 (1999) 493–499.

[10] K. Inoue, A. Miyahara, T. Itaya, Enantioselective permeation of aminoacids across membranes prepared from 3a-helix bundle polygluta-mates with oxyethylene chains, J. Am. Chem. Soc. 119 (1997) 6191–6192.

[11] Y. Aoki, M. Ohshima, K. Shinohara, T. Kaneko, E. Oikawa, Enantiose-lective permeation of racemates through a solid (+)-poly{{2-[dimethyl-(10-pinanyl)silyl]norbornadiene}} membrane, Polymer 38 (1997)235–238.

[12] B.B. Lakshmi, C.R. Martin, Enantioseparation using apoenzymes immo-bilized in a porous polymeric membrane, Nature 388 (1997) 758–760.

[13] M. Newcomb, J.L. Toner, R.C. Helgeson, D.J. Cram, Host-guest com-plexation. 20. Chiral recognition in transport as a molecular basis for acatalytic resolving machine, J. Am. Chem. Soc. 101 (1979) 4941–4947.

[14] L.J. Keurentjes, W.M. Nabuurs, E.A. Vegter, Liquid membrane technol-ogy for the separation of racemic mixtures, J. Membr. Sci. 113 (1996)351–360.

[15] J.H. Kim, J.H. Kim, J. Jegal, K.-H. Lee, Optical resolution of �-aminoacids through enantioselective polymeric membranes based on polysac-charides, J. Membr. Sci. 213 (2003) 273–283.

[16] S. Tone, T. Masawaki, T. Hamada, The optical resolution of amino acidsby ultrafiltration membranes fixed with plasma polymerized �-menthol,J. Membr. Sci. 103 (1995) 57–63.

[17] T. Aoki, S. Tomizawa, E. Oikawa, Enantioselective permeation throughpoly[�-[3-(pentamethyldisiloxanyl)propyl]-l-glutamate] membranes, J.

[

[

[

[

[

[

[

[

[

[

[

cknowledgements

We gratefully acknowledge financial support through arant-in-Aid for Scientific Research on Priority Areas (B,

Novel Smart Membranes Containing Controlled Molecularavity”, No. 13133202, 2001–2004) and High-Tech Researchenter Project from the Ministry of Education, Culture, Sports,cience, and Technology of Japan. This research was also sup-orted from Asahi Glass Foundation, Iwatani Foundation andalt Science Foundation.

eferences

[1] A. Higuchi, Y. Higuchi, K. Furuta, B.O. Yoon, M. Hara, S. Maniwa,M. Saitoh, K. Sanui, Chiral separation of phenylalanine by ultrafiltra-tion through immobilized DNA membranes, J. Membr. Sci. 221 (2003)207–218.

[2] G.L.A. Rikken, E. Raupach, Enantioselective magnetochiral photochem-istry, Nature 405 (2000) 932–935.

[3] M.C. Millot, Separation of drug enantiomers by liquid chromatogra-phy and capillary electrophoresis, using immobilized proteins as chiralselectors, J. Chromatogr. B 797 (2003) 131–159.

[4] K.B. Jirage, C.R. Martin, New developments in membrane-based sepa-rations, Trend Biotechnol. 17 (5) (1999) 197–200.

[5] O. Hofstetter, J.K. Hertweck, H. Hofstetter, Detection of enantiomericimpurities in a simple membrane-based optical immunosensor, J.Biochem. Biophys. Meth. 63 (2005) 91–99.

[6] N.M. Maier, P. France, W. Lindner, Separation of enantiomers: needs,challenges, perspectives, J. Chromatogr. A 906 (2001) 3–33.

[7] A. Maruyama, N. Adachi, T. Takehisa, M. Torii, K. Sanui, N.Ogata, Enantioselective permeation of �-amino acid isomers throughpoly(amino acid)-derived membranes, Macromolecules 23 (1990)2748–2752.

Membr. Sci. 99 (1995) 117–125.18] S. Kiyohara, M. Nakamura, K. Saito, K. Sugita, T. Sugo, Binding of dl-

tryptophan to BSA adsorbed in multilayers by polymer chains graftedonto a porous hollow-fiber membrane in a permeation mode, J. Membr.Sci. 152 (1999) 143–149.

19] F. Garnier, J. Randon, J.L. Rocca, Enantiomeric separation by ultrafil-tration: complexation mechanism of tryptophan analogs to bovine serumalbumin, Sep. Purif. Technol. 16 (1999) 243–250.

20] Y. Osada, F. Ohta, A. Mizumoto, M. Takase, Y. Kurimura, Specific per-meation and adsorption of aqueous amino acids by plasma-polymerizedmembranes of d-camphor and l-menthol, Nippon Kagaku Kaishi (1986)866–872.

21] T. Kobayashi, H.Y. Wang, N. Fujii, Molecular imprinting of theophyllinein acrylonitrile-acrylic acid copolymer membrane, Chem. Lett. (1995)927.

22] A. Higuchi, Y. Ishida, T. Nakagawa, Surface modified polysulfone mem-branes: separation of mixed proteins and optical resolution of tryptophan,Desalination 90 (1993) 127–136.

23] A. Higuchi, M. Hara, T. Horiuchi, T. Nakagawa, Optical resolution ofamino acids by ultrafiltration membranes containing serum albumin, J.Membr. Sci. 93 (1994) 157–164.

24] A. Higuchi, P. Wang, T.-M. Tak, T. Nohmi, T. Hashimoto, Preparationand properties of surface-modified polyacrylonitrile hollow fibers, in: I.Pinnau, B.D. Freeman (Eds.), Membrane Formation and Modification,ACS Symposium Series 744, ACS, Washington, DC, 1999, p. 228.

25] A. Higuchi, H. Yomogita, B.O. Yoon, T. Kojima, M. Hara, S. Maniwa,M. Saitoh, Optical resolution of amino acid by ultrafiltration using recog-nition sites of DNA, J. Membr. Sci. 205 (2002) 203–212.

26] A. Higuchi, T. Hashimoto, M. Yonehara, N. Kubota, K. Watanabe, S.Uemiya, T. Kojima, M. Hara, Effect of surfactant agents and lipids onoptical resolution of amino acid by ultrafiltration membranes containingbovine serum albumin, J. Membr. Sci. 130 (1997) 31–39.

27] B. Mattiason, M. Ramstorp, Ultrafiltration affinity purification, Ann.N.Y. Acad. Sci. 413 (1983) 307–309.

28] S. Poncet, J. Randon, J. Rocca, Enantiomeric separation of tryptophanby ultrafiltration using the BSA solution system, Sep. Sci. Technol. 32(1997) 2029–2038.

Y. Matsuoka et al. / Journal of Membrane Science 280 (2006) 116–123 123

[29] J. Romero, A.L. Zydney, Affinity ultrafiltration: effects of ligand bind-ing on process optimization, Biotechnol. Bioeng. 88 (2002) 256–265.

[30] J. Romero, A.L. Zydney, Staging of affinity ultrafiltration processes forchiral separations, J. Membr. Sci. 209 (2002) 107–119.

[31] P. Hadik, L.-P. Szabo, E. Nagy, Zs. Farkas, Enantioseparation of d, l-lactic acid by membrane techniques, J. Membr. Sci. 251 (2005) 223–232.

[32] S. Tone, T. Masawaki, K. Eguchi, The optical resolution of amino acidsby plasma polymerised terpene membranes, J. Membr. Sci. 118 (1996)31–40.

[33] M. Bryjak, J. Kozlowski, P. Wieczorek, P. Kafarski, Enantioselectivetransport of amino acid 14 through supported chiral liquid membranes,J. Membr. Sci. 85 (1993) 221–228.

[34] M. Yoshikawa, Chiral separation membranes, in: A. Higuchi (Ed.), Med-ical Materials and Functional Membranes, CMC Publisher, Tokyo, 2005,pp. 215–230.

[35] A. Higuchi, A. Hayashi, N. Kanda, K. Sanui, H. Kitamura, Chiral separa-tion of amino acids in ultrafiltration through DNA-immobilized cellulosemembranes, J. Mol. Struct. 739 (2005) 145–152.

[36] M. Nakamura, S. Kiyohara, K. Saito, K. Sugita, T. Sugo, Chiral separa-tion of dl-tryptophan using porous membranes containing multi-layeredbovine serum albumin crosslinked with glutaraldehyde, J. Chromatogr.A 822 (1998) 53–58.

[37] S. Kiyohara, M. Nakamura, K. Saito, K. Sugita, T. Sugo, Binding of dl-tryptophan to BSA adsorbed in multilayers by polymer chains graftedonto a porous hollow-fiber membrane in a permeation mode, J. Membr.Sci. 152 (1999) 143–149.