- Email: [email protected]
Adsorption of lectins on affinity membranes
Journal of Membrane Science 273 (2006) 12–19
Adsorption of lectins on affinity membranes Cristiana Boi, Francesca Cattoli, Rachele Facchini, Mirco Sorci, Giulio C. Sarti ∗ Dipartimento di Ingegneria Chimica, Mineraria e delle Tecnologie Ambientali, Universit`a degli Studi di Bologna, Viale Risorgimento 2, 40136 Bologna, Italy Received 14 June 2005; received in revised form 5 December 2005; accepted 6 December 2005 Available online 18 January 2006
Abstract The objective of this work is the development of a process for the purification of lectins with affinity membranes. To this aim affinity membranes were prepared by chemical modification of a cellulose matrix. Different ligands were tested endowed with the different affinity towards the lectins used. As a model protein a lectin obtained by chromatographic techniques from Momordica charantia seeds was mainly used; Peanut agglutinin and Ricinus communis agglutinin were also considered. Among the various ligands tested N-acetyl-d-galactosamine gave the best separation performances, whilst arabinogalactan gave the highest binding capacity. The ligand immobilized on the membrane surface is quantified indirectly by measuring the amount of protein bound to the membrane. The kinetics of adsorption and desorption of the purification process has been studied in detail for the different supports. Modified membranes have been used in separation process of lectins with good results in terms of binding capacity towards the protein of interest. © 2005 Elsevier B.V. All rights reserved. Keywords: Lectins; Purification; Affinity membranes; Binding capacity; Kinetics
1. Introduction Lectins are carbohydrate binding proteins of non-immune origin that agglutinate cells or precipitate polysaccharides or glycoproteins [1]. Although lectins were first described at the turn of XIX century, it is only in the 1960s that they became the subject of intense research all over the world [2]. There are many reasons for the current interest in lectins, first of all their usefulness in detecting and studying carbohydrates in solution and on cell surface. Lectins bind mono and oligosaccharides reversibly and with high specificity, defined in terms of the monosaccharides that inhibit lectin-induced agglutination or precipitation reactions. These glycoproteins are multivalent, that is, they possess more than one sugar binding site. Therefore, when they react with cells, for example, erythrocytes, they will not only combine with sugars on their surfaces, but will also cause cross-links of the cells and their subsequent precipitation, a phenomenon referred to as cell agglutination [3].
∗
Corresponding author. Tel.: +39 051 2093142; fax: +39 051 581200. E-mail address: [email protected] (G.C. Sarti).
0376-7388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.12.011
The seeds of Momordica charantia contain two proteins: one of these proteins is a haemagglutinating lectin inhibited by galactose, galactose-containing sugars, whilst the second protein has no haemagglutinating properties, but it is one of the most potent inhibitors of protein synthesis [4]. Separating the lectin from the toxin gives the possibility to exploit the different properties of these two proteins. Haemagglutinin has been isolated by affinity chromatography on Sepharose 4B [5], acid-treated Sepharose 4B [4], cross-linked arabinogalactan [6], cross-linked guar gum [7] and O-␣-galactosyl polyacrylamide gels [8]. The objective of this work is the development of an analogous process for the purification of M. charantia lectin with affinity membranes. The main difference between membrane and bead-based chromatography is due to the different structure of the two stationary phases: membranes are characterised by through pores, whilst beads are characterised by dead-end pores. As a consequence in membrane chromatography the rate controlling step is no longer the diffusion of the solute to the active site immobilised inside the pores, but the binding kinetics between molecule and ligand [9]. Due the typically high throughput, membrane chromatography has been successfully applied in the bioprocess industry, especially for the polishing step of the antibody production process [10,11].
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
Affinity membranes were prepared by chemical modification of two supports: a filter paper made of native cellulose and a regenerated cellulose membrane. Arabinogalactan, guar gum and N-acetyl-d-galactosamine were tested as affinity ligands due to their different affinity with respect to the protein used. Adsorption experiments were performed in batch and in continuous mode in order to determine the binding capacities of the affinity membranes. Adsorption isotherms of M. charantia lectin were obtained for all the different membranes investigated and the relevant parameters were calculated. As a comparison, Ricinus communis agglutinin (RCA) and Peanut agglutinin (PNA) that, like M. charantia lectin, show specificity for galactose, were also tested. Detailed kinetic studies of the adsorption process were performed and the kinetic parameters calculated. The different affinity membranes were compared with respect to the results obtained in terms of binding capacity, kinetic behaviour and selectivity. The feasibility of the process has been demonstrated. 2. Experimental 2.1. Materials 2.1.1. Proteins Lectin extracted from M. charantia seeds was kindly provided by the Department of Experimental Pathology of the University of Bologna [12]. The native lectin has a molecular weight of 116 kDa and consists of four sub-units linked together by cysteine bridges. Aqueous extracts of M. charantia seeds contain both a haemagglutinating lectin (MCL), able to agglutinate human erythrocytes and specific for galactose, and a non-haemagglutinating lectin, highly potent inhibitor of protein synthesis that does not show any carbohydrate specificity. In this work, other galactose-specific lectins have been tested: R. communis agglutinin and Peanut agglutinin, purchased from Vector Laboratories, USA. These glycoproteins were chosen among the commercially available lectins because they show the same specificity towards carbohydrates as MCL; their characteristics are reported in Table 1. 2.1.2. Membranes Two different matrices were used as affinity support: cellulose filter paper purchased from Whatman, with thickness of 150 m, pores capable to retain particles of 20–25 m diameter and porosity of about 62% and Sartobind epoxy-activated membrane with thickness of 250 m, average pore size of 0.45 m, a porosity of about 64% and a minimum static binding capacity for the target protein of 30 g/cm2 , kindly provided by Sartorius AG. Table 1 Lectin properties Lectin
MW (kDa)
Subunits
Specificity sugar
Momordica charantia lectin (MCL) Ricinus communis agglutinin (RCA) Peanut agglutinin (PNA)
116 120 110
4 4 4
-Gal, GalNAc Gal, GalNAc GalNAc
13
Table 2 Inhibition of lectins from MCL, RCA and PNA by saccharides Sugar
Relative inhibitory potencya MCLb
RCAc
PNAd
1 0.7 0.3
1 0.6
Monosaccharides d-Galactose d-Fucose l-Arabinose d-Galactosamine N-Acetyl-d-galactosamine
1 0.87 2.1 0.03 0.125
<0.10
Di/oligosaccharides d-Lactose d-Raffinose
5.8 1.72
7.5 1.7
Polysaccharides Arabinogalactan Guar gum
2.22 <0.10
64 53
a
Galactose is normalized to 1.0. From Mazumder et al. [6]. c From Nicolson et al. [13]: 0.15 mol of galactose are required for 50% inhibition. d From Pereira et al. [14]: 0.6 mol of galactose are required for 50% inhibition. b
2.1.3. Ligands Lectins have affinity for various polysaccharides that can be measured by haemagglutination tests [6,13,14]. Different ligands were chosen according to their affinity to the target proteins as reported in Table 2. MCL has affinity for -anomers of d-galactose and among those arabinogalactan, guar gum and N-acetyl-d-galactosamine were tested as affinity ligands. All ligands were purchased from Sigma–Aldrich, Italy. 2.1.4. Chemicals and solutions Citric acid was purchased from Carlo Erba Reagents, Italy. Sodium thiosulfate, NaBH4 , lactose, 1,4-butanediol diglycidyl ether, sodium azide, glycine were ACS reagent grade purchased from Sigma–Aldrich, Italy. BCA protein assay reagents were purchased from Pierce. The loading buffer was PBS prepared from a ten fold solution: 80 g NaCl, 2 g KCl, 14.4 g Na2 HPO4 , 2.4 g K H2 PO4 in 800 mL deionised water, brought to pH 7.4 and additioned with deionised water up to 1 L. Before use, dilution 1:10 with deionised water is applied. The elution buffer was 0.1 M lactose in PBS. All buffers were filtered prior to use through 0.45 m nitrocellulose membrane filter purchased from Millipore. Molecular weight markers for SDS-PAGE electrophoresis (Standards Broad Range) and Blue Coomassie were purchased from Bio-Rad Laboratories. 2.1.5. Equipment The dynamic experiments have been performed using an Amicon dead end membrane holder that can allocate up to 20 flat sheet membranes of 2.5 cm diameter in a stack. A peristaltic pump Gilson Minipuls 3 was used, with a head of eight channels, 10 rollers that can operate at different flow-rates: up to 26 mL/min with PVC tubes (i.d. 3.2 mm), up to 45 mL/min with silicone tubes (i.d. 2.8 mm). The maximum pressure drop is 500 kPa (75 psi).
14
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
Absorbance readings were performed using a UV–vis spectrophotometer Shimadzu UV-1601. SDS-PAGE analysis of the protein solutions was performed with Criterion electrophoresis system from Bio-Rad Laboratories, using precast gels. 2.2. Membrane modification 2.2.1. Membrane activation Whatman filters 541 have been modified following the two stage protocol described in ref. [15]. First a spacer arm, 1,4butanediol diglycidyl ether, was bound to the surface. The concentration of active epoxy groups onto the membranes was measured by titration with sodium thiosulfate [16] and then the ligand was bound to the epoxy active groups present. In order to improve the support properties the effects of a mercerization pre-treatment were also investigated. Mercerization modifies the crystalline structure of cellulose and increases the amount of amorphous regions that have more accessible OH groups for reaction with the bisoxirane. This treatment is often used in the textile industry since it allows easier dying of cotton fabrics. The membranes were immersed in a 8 M NaOH solution for 20 min and then extensively washed with deionised water [17]. The concentration of epoxy groups on the membrane surface was determined by titration with sodium thiosulfate [16]. As expected, the number of epoxy groups increases with the pretreatment, as it can be seen in Fig. 1 in which the comparison of the number of available epoxy groups before and after mercerization is reported for two different types of filter paper. Sartobind epoxy is a regenerated cellulose membrane preactivated for ligand coupling. The different ligands were then coupled with the same protocol for the two different matrices, but with different procedures depending on the ligand chosen. 2.2.2. Ligand coupling Guar gum activation: 3.75 g of guar gum were dissolved in 150 mL 0.3N NaOH, the suspension was centrifuged at 12,000 rpm for 20 min and the supernatant was collected and used for coupling [18]. From the initial solution 100 mL of supernatant were obtained. Epoxy activated membranes were
Fig. 1. Effect of mercerization on the active epoxy groups present on the membrane surface. Comparison between mercerized and non-mercerized membranes for two different supports, W541 and W542.
immersed in the solution and the coupling reaction started by adding 1 mg/mL of NaBH4 . The system was shaken overnight at 40 ◦ C. After about 20 h the temperature was increased up to 70 ◦ C and the reaction continued for other 6 h [19]. The membranes were thoroughly washed with deionised water before use. Arabinogalactan activation: 4.5 g of arabinogalactan were dissolved in 150 mL 0.3N NaOH. Epoxy activated membranes were immersed in the solution and the coupling reaction started by adding 1 mg/mL of NaBH4 . The system was shaken 24 h at 40 ◦ C. The membranes were thoroughly washed with deionised water before use [20]. N-Acetyl-d-galactosamine activation: 7.5 g of N-acetyl-dgalactosamine were dissolved in 150 mL 0.1N NaOH. Epoxy activated membranes were immersed in the solution and shaken 24 h at 40 ◦ C. The membranes were thoroughly washed alternatively with deionised water and 0.1 M sodium borate buffer before use [21]. The affinity membranes can be stored in 0.02% sodium azide at 4 ◦ C. 2.3. Protein concentration measurement Lectin concentration was measured either by reading the absorbance at 280 nm or by using the bicinchoninic acid assay (BCA assay, from Pierce). The BCA assay is used for the colorimetric detection and quantitation of total protein; this method combines the well-known reduction of Cu2+ to Cu1+ by protein in an alkaline medium (Biuret reaction) with the highly sensitive and selective colorimetric detection of cuprous cation using a reagent containing BCA. The purple-colored reaction product of this assay is formed by the chelation of two molecules of BCA with one cuprous ion. This water-soluble complex exhibits a strong absorbance at 562 nm that is linear with protein concentrations over a broad working range from 20 to 2000 g/mL. Calibration curves were prepared for both methods by using pure lectin solutions at known concentration values. 2.4. Batch adsorption experiments Affinity membranes were tested for adsorption in batch experiments in order to measure binding kinetics and static binding capacity. The membranes, 20 cm2 total area, were immersed in a beaker containing 5 mL of pure protein solution and gently agitated in an orbital shaker. By monitoring the decrease of protein concentration in solution, kinetic curves are obtained. The duration of the adsorption step depends on the concentration of the protein solution; in any case, for these systems equilibrium is reached within 2 h. After adsorption, the membranes were washed with PBS buffer at pH 7.4 and eluted with 3 mL of 0.2 M lactose in PBS. The experiments were repeated for the different affinity membranes obtained; the native cellulose matrices were tested for non-specific adsorption. The effect of mercerization of native filter paper was also investigated. To investigate the feasibility of the process and test the selectivity of the affinity membranes an artificial protein mixture was prepared with 0.5 mg/mL MCL, 0.5 mg/mL BSA and 0.5 mg/mL
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
15
Fig. 3. SDS-PAGE electrophoresis: adsorption experiments of MCL in mixture with BSA and lysozyme, on Sartobind epoxy membranes modified with arabinogalactan (SarEpo-AraGal) and N-acetyl-d-galactosamine (SarEpo-GalNAc). M, molecular weight markers; I, solution before the adsorption step; F, solution after the adsorption step; W, washing step; E, concentrated eluted fraction. Fig. 2. Experimental set-up used for dynamic experiments.
lysozyme in PBS. A batch purification of MCL was performed with Sartobind epoxy modified with N-acetyl-d-galactosamine and Sartobind epoxy modified with arabinogalactan. 2.5. Flow adsorption experiments A second series of experiments has been performed in a continuous flow system configuration as illustrated in Fig. 2. The experiments were conducted with an initial protein concentration of 0.5 mg/mL, with a volume of 1 mL of solution and with 30 cm2 of membrane area. Since lectins are rather expensive and MCL is difficult to obtain, the protein solution was recirculated to the membrane cell. The experiments were performed with the same conditions but at two different superficial velocities, i.e. 0.9 and 0.18 cm/min, respectively. 3. Results and discussion
N-acetyl-d-galactosamine is reported in Fig. 4 as a function of the protein concentration at equilibrium. The concentration of lectin in solution has been monitored with two different techniques, absorbance reading at 280 nm and BCA protein assay. The results are in good agreement with one another, indicating that both techniques can be equivalently used for this purpose. The experimental data follow a Langmuir type isotherm: qm ceq qeq = (1) Kd + ceq The relevant parameters, qm and Kd , were calculated by least square regression and reported in Fig. 4. Since N-acetyl-dgalactosamine is generally used for lectin purification by affinity chromatography [13] it was the natural choice as a ligand for affinity membranes. However, this ligand is rather expensive and other carbohydrates were also tested as possible substitute ligands. Arabinogalactan has been coupled onto Sartobind epoxy membranes and the resulting affinity membranes gave
The SDS-PAGE electrophoresis reports the results of MCL purification in batch experiments with an artificial protein mixture for two different affinity membranes: Sartobind epoxy modified with N-acetyl-d-galactosamine and Sartobind epoxy modified with arabinogalactan, Fig. 3. The bands of the MCL sub-units, about 30 kDa MW, are present in both eluted fractions; however, MCL eluted from Sartobind epoxy modified with Nacetyl-d-galactosamine is rather pure, whilst bands of BSA and lysozyme are also visible in the fractions eluted from Sartobind epoxy modified with arabinogalactan. In any case, the protein mixture used represents an artificial example and is not representative of the real extract from M. charantia seeds in which the non-haemagglutinating lectin needs to be isolated from MCL. 3.1. Static binding capacity The static binding capacity of MCL onto affinity membranes obtained with Sartobind epoxy activated supports modified with
Fig. 4. Adsorption isotherm of MCL on N-acetyl-d-galactosamine-Sartobind epoxy (SarEpo-GalNAc) obtained from batch experiments. The protein concentration was measured with absorbance reading at 280 nm and by BCA assay. The calculated data were obtained by non-linear, least squares regression of experimental data (R2 = 0.967).
16
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 Table 3 Membrane capacity for the different supports tested with MCL
Fig. 5. Adsorption isotherm of MCL on arabinogalactan-Sartobind epoxy (SarEpo-AraGal) obtained from batch experiments. The protein concentration was measured with absorbance reading at 280 nm and by BCA assay. The calculated data were obtained by non-linear, least squares regression of experimental data (R2 = 0.898).
good results in terms of static binding capacity, as illustrated in Fig. 5 in which the adsorption data at equilibrium of MCL are reported. Even in this case both absorbance at 280 nm and BCA assay were used to measure the protein concentration in solution. The experimental data points are reported together with the adsorption isotherm interpolated with Langmuir equation. Adsorption isotherms of MCL onto Whatman 541 membranes modified with arabinogalactan are reported in Fig. 6 in which the static binding capacity is reported as a function of the concentration of protein in solution at equilibrium. The experimental data points follow a Langmuir type isotherm and the relevant parameters, qm and Kd , were calculated and reported in Fig. 6. Mercerization of the native support increases the capacity of the affinity membranes but not as much as expected from the titration results in which an increase of 80% of the available epoxy groups was obtained. From adsorption experiments a 38% increase of the static binding capacity was calculated. This might
Membrane
Ligand
qm (mg/cm2 )
qm (mg/mL)
W541 W541M SarEpo SarEpo
AraGal AraGal AraGal GalNAc
0.016 0.022 0.031 0.022
1.067 1.467 1.240 0.880
W541 W541M Basic Sar
– – –
0.007 0.006 0.007
0.453 0.367 0.258
be due to the yield of the reaction for ligand coupling, which takes place after mercerization. Another effect that should be taken into consideration is the effect of mercerization on the cellulose structure: the fibres increase their diameter whilst the diameter of the membrane pores is reduced [17]. This could well have an effect on the accessibility of the active sites within the pores. In any case an increase of static binding capacity of 38% is a relevant result and mercerization should be further investigated as a possible pre-treatment on a process scale. The static binding capacity of MCL on the native cellulose matrices was measured to account for non-specific adsorption on the different supports. The results are reported in Table 3, in which the static binding capacity, generally calculated in terms of mg of protein per cm2 of membrane, has been calculated also in terms of mg of protein per mL of membrane to account for the different thickness of the two films. The static binding capacity is of the same order of magnitude for all the membranes used, and the highest value of capacity is obtained for the mercerized membranes. Arabinogalactan gives higher binding capacity than N-acetyl-d-galactosamine; this result has been confirmed by data obtained with different proteins, and in particular with PNA. The binding capacity of Whatman 541, modified with the different ligands and for all the proteins tested, is reported in Table 4. Indeed, N-acetyl-d-galactosamine is the ligand that gave lowest binding capacity, whilst guar gum and arabinogalactan gave similar performances. However, guar gum is very difficult to handle because it tends to gel and, due to this reason, arabinogalactan has been preferred. 3.2. Dynamic binding capacity Continuous flow experiments were performed in order to calculate the dynamic binding capacity of the membranes and to study the effects of superficial velocity on the dynamic binding Table 4 W541 membrane capacity for the different ligands tested with different lectins Static binding capacity, qm (mg/cm2 )
Fig. 6. Adsorption isotherms of MCL on arabinogalactan-Whatman 541 (W541AraGal) obtained from batch experiments. The calculated data were obtained by non-linear, least squares regression of experimental data. Comparison between mercerized and non-mercerized supports (R2W541 = 0.780; R2W541M = 0.852).
MCL RCA120 PNA
AraGal
GalNAc
Guar gum
0.016 – 0.030
0.006 0.009 0.006
– 0.022 0.019
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 Table 5 Dynamic binding capacity for the different supports tested Membrane
W541 SarEpo SarEpo
Ligand
AraGal AraGal GalNAc
with a kinetic equation. On a rather general basis, the binding–unbinding kinetic model considers the adsorption and desorption reactions as:
Dynamic binding capacity (mg/mL) v1 = 0.09 cm/min, tads = 180 min
v2 = 0.18 cm/min, tads = 90 min
dq = k1 [(qm − q)(c0 − αq) − Kd q] dt
0.400 0.448
0.333 0.456 0.328
By assuming that adsorption is the result of the direct interaction between the protein in solution and the active sites, the forward reaction is linear with the protein concentration in solution and linear with free active site concentration onto the membrane surface. If we consider only the initial instants of adsorption, the desorption reaction becomes negligible and Eq. (3) reduces to:
capacity. The results obtained for the affinity membranes tested in this configuration are reported in Table 5 in which the dynamic binding capacity for the different supports is reported for the two different superficial velocities investigated, v1 = 0.9 cm/min and v2 = 0.18 cm/min. It is apparent that the dynamic binding capacity does not vary much with the superficial velocity, in the small range of velocities investigated. In any case the dynamic binding capacity is always smaller than the static binding capacity and this holds true for all the affinity membranes tested. In Fig. 7 a typical concentration profile for a dynamic experiment of MCL is reported. During the adsorption step, the feed solution is continuously recirculated to the membrane cell; this is the reason why the adsorption curve is different from a typical breakthrough curve. The small elution peak represents one of the main drawbacks of this process and the quantity of protein that was eluted is not comparable to the amount of protein adsorbed. Different buffers were tested in order to improve the elution step, among those 0.1 M glycine pH 2.8 and 0.1 M citric acid pH 2.5, but the amount of protein recovered was always lower than the amount obtained with 0.2 M lactose. 3.3. Kinetic study In order to investigate the kinetic behaviour of the different supports, for some batch experiments, the concentration of protein in solution has been monitored as a function of time. The interpolation curve has been obtained by coupling the mass balance equation for the batch system: V
dc dq = −A dt dt
17
dq = k1 qm c0 dt
(3)
(4)
Short-times experimental data have been interpolated by using Eq. (4), together with the mass balance, Eq. (2), in order to obtain directly k1 , the kinetic constant of the adsorption reaction, which is the most representative parameter of the kinetic behaviour of the system. In Fig. 8 the decrease of dimensionless concentration of MCL is reported as a function of time for Sartobind epoxy membranes modified with two different ligands. These experiments have been conducted at the same operating conditions: membrane area of 20 cm2 , initial protein concentration 0.3 mg/mL, adsorption time 2 h. Similar experiments have been performed for W541, with the aim to investigate the effects of mercerization on the adsorption kinetics. We used arabinogalactan as ligand, because it is cheaper than N-acetyl-d-galactosamine and shows higher binding capacity. In Fig. 9 the comparison between non-mercerized and mercerized supports is reported. The operating conditions were the same for both experiments: membrane area of 9.8 cm2 , initial protein concentration 0.3 mg/mL, adsorption time 2 h. From Fig. 9, we can conclude that mercerized membranes, which have higher binding capacity with respect to non-mercerized membranes and have also a faster binding kinetics.
(2)
Fig. 7. Concentration profiles of the adsorption, washing and elution curves during a dynamic experiment of MCL.
Fig. 8. Adsorption kinetics of MCL on Sartobind epoxy membranes (SarEpo) obtained from batch experiments: comparison between supports, modified with two different ligands: arabinogalactan (AraGal) and N-acetyl-d-galactosamine (GalNAc).
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
18
The resulting affinity membranes were tested for adsorption in batch and dynamic experiments and were characterised in terms of binding capacities and kinetic parameters. Among the carbohydrates tested, N-acetyl-d-galactosamine gave the worst performances in terms of binding capacity but it was the ligand that showed highest selectivity; guar gum gave promising results but it was used only in a few experiments because it gels and it was very difficult to manipulate. Arabinogalactan was the ligand of choice due to its good binding capacity, easiness of use and price. Elution remains the critical step of the process and still needs to be optimised. The affinity membranes presented in this work should be tested with the aqueous extract from M. charantia seeds and this is the first objective for future work. Fig. 9. Adsorption kinetics of MCL on arabinogalactan-Whatman 541 obtained from batch experiments: comparison between mercerized (W541M) and nonmercerized (W541) supports. Table 6 Kinetic constant k1 for the different supports tested Membrane
Ligand
k1 mL/(mg min)
STD mL/(mg min)
W541 W541M SarEpo SarEpo
AraGal AraGal AraGal GalNAc
0.167 0.426 0.082 0.107
0.057 0.160 0.015 0.008
Acknowledgements This work was financially supported by MIUR, Italian Ministry of Education, University and Research (PRIN 2004-05) and by the University of Bologna, Italy. Special thanks go out to Prof. Stirpe of the Department of Experimental Pathology of the University of Bologna, for providing the M. charantia lectin used in the experiments. Sartorius A.G. is thankfully acknowledged for supplying the Sartobind epoxy-activated membranes.
Nomenclature In order to make quantitative comparison among the different behaviours, all the kinetic experiments have been interpolated with the short times analysis model, obtaining the kinetic constant of the adsorption reaction, k1 . In Table 6 the mean values obtained for the different affinity membranes studied are reported. Indeed, the mercerization pre-treatment has the effect to change the structure of cellulose, making it more amorphous, that leads not only to an increase in the number of the active sites, but it also increases the accessibility of the active sites. Not important differences have been observed using different ligands: Sartobind epoxy membranes modified with arabinogalactan and N-acetyl-d-galactosamine are characterized by kinetic constants of the same order of magnitude, although the latter ligand has k1 30% higher. A greater difference results by making a comparison between the two supports: Whatman filters show a binding kinetics much faster than the Sartobind epoxy membranes; the kinetic constants is about five time higher. So far there is not a satisfactory explanation of this behaviour: a possible reason could be the different accessibility of the active sites, that may be affected by different factors as support, spacer arm or external mass transfer phenomena.
A c k1 k2 Kd q q* t v V
frontal membrane area (cm2 ) protein concentration in the liquid solution (mg mL−1 ) rate constant of the complex formation reaction (mL mg−1 min−1 ) rate constant of the complex breakdown reaction (min−1 ) dissociation constant (mg mL−1 ) protein surface concentration (mg cm−2 ) binding capacity (mg mL−1 ) time (min) superficial velocity (cm min−1 ) solution volume (mL)
Greek symbol α working parameter defined as A/V (cm) Subscripts 0 initial condition ads adsorption eq equilibrium condition m maximum
4. Conclusions Affinity membranes for the adsorption of M. charantia lectins were obtained starting from two different cellulosic matrices. Mercerization of the native cellulose support was also investigated as a possible beneficial pre-treatment. Arabinogalactan, N-acetyl-d-galactosamine and guar gum were the ligands coupled to the supports.
References [1] H.B.F. Dixon, Defin. Lectin, Nat. 292 (1981) 192. [2] I.E. Liener, N. Sharon, I.J. Goldstein, The Lectins: Properties, Functions and Applications in Biology and Medicine, Academic Press, Inc., Orlando, 1986.
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 [3] H. Lis, N. Sharon, Lectins: carbohydrate-specific proteins that mediate cellular recognition, Chem. Rev. 98 (1998) 637. [4] L. Barbieri, M. Zamboni, E. Lorenzoni, L. Montanaro, S. Sperti, F. Stirpe, Inhibition of protein synthesis in vitro by proteins from the seeds of Momordica charantia (Bitter pear melon), J. Biochem. 186 (1980) 443. [5] M. Tomita, T. Kurokawa, K. Onozaki, N. Ichiki, T. Osawa, T. Ukita, Purification of galactose-binding phytoagglutinins and phytotoxin by affinity column chromatography using Sepharose, Experientia 28 (1972) 84. [6] T. Mazumder, N. Gaur, A. Surolia, The physicochemical properties of the galactose-specific lectin from Momordica charantia, Eur. J. Biochem. 113 (1981) 463. [7] P. Padma, S.S. Komath, S.K. Nadimpalli, M.J. Swamy, Purification in high yield ann characterisation of a new galactose-specific lectin from the seeds of Trichosanthes cucumerina, Phytochemistry 50 (1999) 363. [8] V. Hoˇrejˇs´ı, M. Tich´a, J. Novotny, J. Kocourek, Studies on lectins XLVII. Some properties of d-galactose binding lectins isolated from the seeds of Butea frondosa, Erythrina indica and Momordica charantia, Biochim. Biophys. Acta 623 (1980) 439. [9] E. Klein, Affinity Membranes, Wiley, 1991. [10] M. Phillips, J. Cormier, J. Ferrence, C. Dowd, R. Kiss, H. Lutz, J. Carter, Performance of a membrane adsorber for trace impurity removal in biotechnology manufacturing, J. Chromatogr. A 1078 (2005) 74. [11] H.L. Knudsen, R.L. Fahrner, Y. Xu, L.A. Norling, G.S. Blank, Membrane ion-exchange chromatography for process-scale antibody purification, J. Chromatogr. A 907 (2001) 145.
19
[12] L. Barbieri, E. Lorenzoni, F. Stirpe, Inhibition of protein synthesis in vitro by a lectin from Momordica charantia and by other haemagglutinins, J. Biochem. 182 (1979) 633. [13] G.L. Nicolson, J. Blaustein, M.E. Etzler, Characterization of two plant lectins from Ricinus communis and their quantitative interaction with a murine lymphoma, Biochemistry 13 (1974) 196. [14] M.E.A. Pereira, E.A. Kabat, Immunochemical studies on blood groups LXII. Fractionation of hog and human A, H, and AH blood group active substance on insoluble immunoadsorbents of Dolichos and Lotus lectins, J. Exp. Med. 143 (1976) 422. [15] F. Cattoli, G.C. Sarti, Purification of MBP--galactosidase and MBPrubredoxin through affinity membrane separation, Sep. Sci. Technol. 37 (2002) 1699. [16] L. Sundberg, J. Porath, Preparation of adsorbent for biospecific affinity chromatography. I. Attachment of group-containing ligands to insoluble polymers by means of bifunctional oxiranes, J. Chromatgr. 90 (1974) 87. [17] E. Ruckenstein, W. Guo, Crosslinked mercerized cellulose membranes and their application to membrane affinity chromatography, J. Membr. Sci. 193 (2001) 131. [18] R. Tyagi, R. Agarwal, M.N. Gupta, Purification of peanut lectin using guar gum as an affinity ligand, J. Biotechnol. 46 (1996) 79. [19] K.C. Gupta, M.K. Shani, B.S. Rathaur, C.K. Narang, N.K. Mathur, Gel filtration medium derived from guar gum, J. Chromatgr. 169 (1979) 183. [20] T. Majumdar, A. Surolia, A large scale preparation of peanut agglutinin on a new affinity matrix, Prep. Biochem. 8 (1978) 119. [21] G.T. Hermanson, A.K. Mallia, P.K. Smith, Immobilized Affinity Ligand Techniques, Academic Press, 1992.
Adsorption of lectins on affinity membranes Cristiana Boi, Francesca Cattoli, Rachele Facchini, Mirco Sorci, Giulio C. Sarti ∗ Dipartimento di Ingegneria Chimica, Mineraria e delle Tecnologie Ambientali, Universit`a degli Studi di Bologna, Viale Risorgimento 2, 40136 Bologna, Italy Received 14 June 2005; received in revised form 5 December 2005; accepted 6 December 2005 Available online 18 January 2006
Abstract The objective of this work is the development of a process for the purification of lectins with affinity membranes. To this aim affinity membranes were prepared by chemical modification of a cellulose matrix. Different ligands were tested endowed with the different affinity towards the lectins used. As a model protein a lectin obtained by chromatographic techniques from Momordica charantia seeds was mainly used; Peanut agglutinin and Ricinus communis agglutinin were also considered. Among the various ligands tested N-acetyl-d-galactosamine gave the best separation performances, whilst arabinogalactan gave the highest binding capacity. The ligand immobilized on the membrane surface is quantified indirectly by measuring the amount of protein bound to the membrane. The kinetics of adsorption and desorption of the purification process has been studied in detail for the different supports. Modified membranes have been used in separation process of lectins with good results in terms of binding capacity towards the protein of interest. © 2005 Elsevier B.V. All rights reserved. Keywords: Lectins; Purification; Affinity membranes; Binding capacity; Kinetics
1. Introduction Lectins are carbohydrate binding proteins of non-immune origin that agglutinate cells or precipitate polysaccharides or glycoproteins [1]. Although lectins were first described at the turn of XIX century, it is only in the 1960s that they became the subject of intense research all over the world [2]. There are many reasons for the current interest in lectins, first of all their usefulness in detecting and studying carbohydrates in solution and on cell surface. Lectins bind mono and oligosaccharides reversibly and with high specificity, defined in terms of the monosaccharides that inhibit lectin-induced agglutination or precipitation reactions. These glycoproteins are multivalent, that is, they possess more than one sugar binding site. Therefore, when they react with cells, for example, erythrocytes, they will not only combine with sugars on their surfaces, but will also cause cross-links of the cells and their subsequent precipitation, a phenomenon referred to as cell agglutination [3].
∗
Corresponding author. Tel.: +39 051 2093142; fax: +39 051 581200. E-mail address: [email protected] (G.C. Sarti).
0376-7388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.12.011
The seeds of Momordica charantia contain two proteins: one of these proteins is a haemagglutinating lectin inhibited by galactose, galactose-containing sugars, whilst the second protein has no haemagglutinating properties, but it is one of the most potent inhibitors of protein synthesis [4]. Separating the lectin from the toxin gives the possibility to exploit the different properties of these two proteins. Haemagglutinin has been isolated by affinity chromatography on Sepharose 4B [5], acid-treated Sepharose 4B [4], cross-linked arabinogalactan [6], cross-linked guar gum [7] and O-␣-galactosyl polyacrylamide gels [8]. The objective of this work is the development of an analogous process for the purification of M. charantia lectin with affinity membranes. The main difference between membrane and bead-based chromatography is due to the different structure of the two stationary phases: membranes are characterised by through pores, whilst beads are characterised by dead-end pores. As a consequence in membrane chromatography the rate controlling step is no longer the diffusion of the solute to the active site immobilised inside the pores, but the binding kinetics between molecule and ligand [9]. Due the typically high throughput, membrane chromatography has been successfully applied in the bioprocess industry, especially for the polishing step of the antibody production process [10,11].
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
Affinity membranes were prepared by chemical modification of two supports: a filter paper made of native cellulose and a regenerated cellulose membrane. Arabinogalactan, guar gum and N-acetyl-d-galactosamine were tested as affinity ligands due to their different affinity with respect to the protein used. Adsorption experiments were performed in batch and in continuous mode in order to determine the binding capacities of the affinity membranes. Adsorption isotherms of M. charantia lectin were obtained for all the different membranes investigated and the relevant parameters were calculated. As a comparison, Ricinus communis agglutinin (RCA) and Peanut agglutinin (PNA) that, like M. charantia lectin, show specificity for galactose, were also tested. Detailed kinetic studies of the adsorption process were performed and the kinetic parameters calculated. The different affinity membranes were compared with respect to the results obtained in terms of binding capacity, kinetic behaviour and selectivity. The feasibility of the process has been demonstrated. 2. Experimental 2.1. Materials 2.1.1. Proteins Lectin extracted from M. charantia seeds was kindly provided by the Department of Experimental Pathology of the University of Bologna [12]. The native lectin has a molecular weight of 116 kDa and consists of four sub-units linked together by cysteine bridges. Aqueous extracts of M. charantia seeds contain both a haemagglutinating lectin (MCL), able to agglutinate human erythrocytes and specific for galactose, and a non-haemagglutinating lectin, highly potent inhibitor of protein synthesis that does not show any carbohydrate specificity. In this work, other galactose-specific lectins have been tested: R. communis agglutinin and Peanut agglutinin, purchased from Vector Laboratories, USA. These glycoproteins were chosen among the commercially available lectins because they show the same specificity towards carbohydrates as MCL; their characteristics are reported in Table 1. 2.1.2. Membranes Two different matrices were used as affinity support: cellulose filter paper purchased from Whatman, with thickness of 150 m, pores capable to retain particles of 20–25 m diameter and porosity of about 62% and Sartobind epoxy-activated membrane with thickness of 250 m, average pore size of 0.45 m, a porosity of about 64% and a minimum static binding capacity for the target protein of 30 g/cm2 , kindly provided by Sartorius AG. Table 1 Lectin properties Lectin
MW (kDa)
Subunits
Specificity sugar
Momordica charantia lectin (MCL) Ricinus communis agglutinin (RCA) Peanut agglutinin (PNA)
116 120 110
4 4 4
-Gal, GalNAc Gal, GalNAc GalNAc
13
Table 2 Inhibition of lectins from MCL, RCA and PNA by saccharides Sugar
Relative inhibitory potencya MCLb
RCAc
PNAd
1 0.7 0.3
1 0.6
Monosaccharides d-Galactose d-Fucose l-Arabinose d-Galactosamine N-Acetyl-d-galactosamine
1 0.87 2.1 0.03 0.125
<0.10
Di/oligosaccharides d-Lactose d-Raffinose
5.8 1.72
7.5 1.7
Polysaccharides Arabinogalactan Guar gum
2.22 <0.10
64 53
a
Galactose is normalized to 1.0. From Mazumder et al. [6]. c From Nicolson et al. [13]: 0.15 mol of galactose are required for 50% inhibition. d From Pereira et al. [14]: 0.6 mol of galactose are required for 50% inhibition. b
2.1.3. Ligands Lectins have affinity for various polysaccharides that can be measured by haemagglutination tests [6,13,14]. Different ligands were chosen according to their affinity to the target proteins as reported in Table 2. MCL has affinity for -anomers of d-galactose and among those arabinogalactan, guar gum and N-acetyl-d-galactosamine were tested as affinity ligands. All ligands were purchased from Sigma–Aldrich, Italy. 2.1.4. Chemicals and solutions Citric acid was purchased from Carlo Erba Reagents, Italy. Sodium thiosulfate, NaBH4 , lactose, 1,4-butanediol diglycidyl ether, sodium azide, glycine were ACS reagent grade purchased from Sigma–Aldrich, Italy. BCA protein assay reagents were purchased from Pierce. The loading buffer was PBS prepared from a ten fold solution: 80 g NaCl, 2 g KCl, 14.4 g Na2 HPO4 , 2.4 g K H2 PO4 in 800 mL deionised water, brought to pH 7.4 and additioned with deionised water up to 1 L. Before use, dilution 1:10 with deionised water is applied. The elution buffer was 0.1 M lactose in PBS. All buffers were filtered prior to use through 0.45 m nitrocellulose membrane filter purchased from Millipore. Molecular weight markers for SDS-PAGE electrophoresis (Standards Broad Range) and Blue Coomassie were purchased from Bio-Rad Laboratories. 2.1.5. Equipment The dynamic experiments have been performed using an Amicon dead end membrane holder that can allocate up to 20 flat sheet membranes of 2.5 cm diameter in a stack. A peristaltic pump Gilson Minipuls 3 was used, with a head of eight channels, 10 rollers that can operate at different flow-rates: up to 26 mL/min with PVC tubes (i.d. 3.2 mm), up to 45 mL/min with silicone tubes (i.d. 2.8 mm). The maximum pressure drop is 500 kPa (75 psi).
14
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
Absorbance readings were performed using a UV–vis spectrophotometer Shimadzu UV-1601. SDS-PAGE analysis of the protein solutions was performed with Criterion electrophoresis system from Bio-Rad Laboratories, using precast gels. 2.2. Membrane modification 2.2.1. Membrane activation Whatman filters 541 have been modified following the two stage protocol described in ref. [15]. First a spacer arm, 1,4butanediol diglycidyl ether, was bound to the surface. The concentration of active epoxy groups onto the membranes was measured by titration with sodium thiosulfate [16] and then the ligand was bound to the epoxy active groups present. In order to improve the support properties the effects of a mercerization pre-treatment were also investigated. Mercerization modifies the crystalline structure of cellulose and increases the amount of amorphous regions that have more accessible OH groups for reaction with the bisoxirane. This treatment is often used in the textile industry since it allows easier dying of cotton fabrics. The membranes were immersed in a 8 M NaOH solution for 20 min and then extensively washed with deionised water [17]. The concentration of epoxy groups on the membrane surface was determined by titration with sodium thiosulfate [16]. As expected, the number of epoxy groups increases with the pretreatment, as it can be seen in Fig. 1 in which the comparison of the number of available epoxy groups before and after mercerization is reported for two different types of filter paper. Sartobind epoxy is a regenerated cellulose membrane preactivated for ligand coupling. The different ligands were then coupled with the same protocol for the two different matrices, but with different procedures depending on the ligand chosen. 2.2.2. Ligand coupling Guar gum activation: 3.75 g of guar gum were dissolved in 150 mL 0.3N NaOH, the suspension was centrifuged at 12,000 rpm for 20 min and the supernatant was collected and used for coupling [18]. From the initial solution 100 mL of supernatant were obtained. Epoxy activated membranes were
Fig. 1. Effect of mercerization on the active epoxy groups present on the membrane surface. Comparison between mercerized and non-mercerized membranes for two different supports, W541 and W542.
immersed in the solution and the coupling reaction started by adding 1 mg/mL of NaBH4 . The system was shaken overnight at 40 ◦ C. After about 20 h the temperature was increased up to 70 ◦ C and the reaction continued for other 6 h [19]. The membranes were thoroughly washed with deionised water before use. Arabinogalactan activation: 4.5 g of arabinogalactan were dissolved in 150 mL 0.3N NaOH. Epoxy activated membranes were immersed in the solution and the coupling reaction started by adding 1 mg/mL of NaBH4 . The system was shaken 24 h at 40 ◦ C. The membranes were thoroughly washed with deionised water before use [20]. N-Acetyl-d-galactosamine activation: 7.5 g of N-acetyl-dgalactosamine were dissolved in 150 mL 0.1N NaOH. Epoxy activated membranes were immersed in the solution and shaken 24 h at 40 ◦ C. The membranes were thoroughly washed alternatively with deionised water and 0.1 M sodium borate buffer before use [21]. The affinity membranes can be stored in 0.02% sodium azide at 4 ◦ C. 2.3. Protein concentration measurement Lectin concentration was measured either by reading the absorbance at 280 nm or by using the bicinchoninic acid assay (BCA assay, from Pierce). The BCA assay is used for the colorimetric detection and quantitation of total protein; this method combines the well-known reduction of Cu2+ to Cu1+ by protein in an alkaline medium (Biuret reaction) with the highly sensitive and selective colorimetric detection of cuprous cation using a reagent containing BCA. The purple-colored reaction product of this assay is formed by the chelation of two molecules of BCA with one cuprous ion. This water-soluble complex exhibits a strong absorbance at 562 nm that is linear with protein concentrations over a broad working range from 20 to 2000 g/mL. Calibration curves were prepared for both methods by using pure lectin solutions at known concentration values. 2.4. Batch adsorption experiments Affinity membranes were tested for adsorption in batch experiments in order to measure binding kinetics and static binding capacity. The membranes, 20 cm2 total area, were immersed in a beaker containing 5 mL of pure protein solution and gently agitated in an orbital shaker. By monitoring the decrease of protein concentration in solution, kinetic curves are obtained. The duration of the adsorption step depends on the concentration of the protein solution; in any case, for these systems equilibrium is reached within 2 h. After adsorption, the membranes were washed with PBS buffer at pH 7.4 and eluted with 3 mL of 0.2 M lactose in PBS. The experiments were repeated for the different affinity membranes obtained; the native cellulose matrices were tested for non-specific adsorption. The effect of mercerization of native filter paper was also investigated. To investigate the feasibility of the process and test the selectivity of the affinity membranes an artificial protein mixture was prepared with 0.5 mg/mL MCL, 0.5 mg/mL BSA and 0.5 mg/mL
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
15
Fig. 3. SDS-PAGE electrophoresis: adsorption experiments of MCL in mixture with BSA and lysozyme, on Sartobind epoxy membranes modified with arabinogalactan (SarEpo-AraGal) and N-acetyl-d-galactosamine (SarEpo-GalNAc). M, molecular weight markers; I, solution before the adsorption step; F, solution after the adsorption step; W, washing step; E, concentrated eluted fraction. Fig. 2. Experimental set-up used for dynamic experiments.
lysozyme in PBS. A batch purification of MCL was performed with Sartobind epoxy modified with N-acetyl-d-galactosamine and Sartobind epoxy modified with arabinogalactan. 2.5. Flow adsorption experiments A second series of experiments has been performed in a continuous flow system configuration as illustrated in Fig. 2. The experiments were conducted with an initial protein concentration of 0.5 mg/mL, with a volume of 1 mL of solution and with 30 cm2 of membrane area. Since lectins are rather expensive and MCL is difficult to obtain, the protein solution was recirculated to the membrane cell. The experiments were performed with the same conditions but at two different superficial velocities, i.e. 0.9 and 0.18 cm/min, respectively. 3. Results and discussion
N-acetyl-d-galactosamine is reported in Fig. 4 as a function of the protein concentration at equilibrium. The concentration of lectin in solution has been monitored with two different techniques, absorbance reading at 280 nm and BCA protein assay. The results are in good agreement with one another, indicating that both techniques can be equivalently used for this purpose. The experimental data follow a Langmuir type isotherm: qm ceq qeq = (1) Kd + ceq The relevant parameters, qm and Kd , were calculated by least square regression and reported in Fig. 4. Since N-acetyl-dgalactosamine is generally used for lectin purification by affinity chromatography [13] it was the natural choice as a ligand for affinity membranes. However, this ligand is rather expensive and other carbohydrates were also tested as possible substitute ligands. Arabinogalactan has been coupled onto Sartobind epoxy membranes and the resulting affinity membranes gave
The SDS-PAGE electrophoresis reports the results of MCL purification in batch experiments with an artificial protein mixture for two different affinity membranes: Sartobind epoxy modified with N-acetyl-d-galactosamine and Sartobind epoxy modified with arabinogalactan, Fig. 3. The bands of the MCL sub-units, about 30 kDa MW, are present in both eluted fractions; however, MCL eluted from Sartobind epoxy modified with Nacetyl-d-galactosamine is rather pure, whilst bands of BSA and lysozyme are also visible in the fractions eluted from Sartobind epoxy modified with arabinogalactan. In any case, the protein mixture used represents an artificial example and is not representative of the real extract from M. charantia seeds in which the non-haemagglutinating lectin needs to be isolated from MCL. 3.1. Static binding capacity The static binding capacity of MCL onto affinity membranes obtained with Sartobind epoxy activated supports modified with
Fig. 4. Adsorption isotherm of MCL on N-acetyl-d-galactosamine-Sartobind epoxy (SarEpo-GalNAc) obtained from batch experiments. The protein concentration was measured with absorbance reading at 280 nm and by BCA assay. The calculated data were obtained by non-linear, least squares regression of experimental data (R2 = 0.967).
16
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 Table 3 Membrane capacity for the different supports tested with MCL
Fig. 5. Adsorption isotherm of MCL on arabinogalactan-Sartobind epoxy (SarEpo-AraGal) obtained from batch experiments. The protein concentration was measured with absorbance reading at 280 nm and by BCA assay. The calculated data were obtained by non-linear, least squares regression of experimental data (R2 = 0.898).
good results in terms of static binding capacity, as illustrated in Fig. 5 in which the adsorption data at equilibrium of MCL are reported. Even in this case both absorbance at 280 nm and BCA assay were used to measure the protein concentration in solution. The experimental data points are reported together with the adsorption isotherm interpolated with Langmuir equation. Adsorption isotherms of MCL onto Whatman 541 membranes modified with arabinogalactan are reported in Fig. 6 in which the static binding capacity is reported as a function of the concentration of protein in solution at equilibrium. The experimental data points follow a Langmuir type isotherm and the relevant parameters, qm and Kd , were calculated and reported in Fig. 6. Mercerization of the native support increases the capacity of the affinity membranes but not as much as expected from the titration results in which an increase of 80% of the available epoxy groups was obtained. From adsorption experiments a 38% increase of the static binding capacity was calculated. This might
Membrane
Ligand
qm (mg/cm2 )
qm (mg/mL)
W541 W541M SarEpo SarEpo
AraGal AraGal AraGal GalNAc
0.016 0.022 0.031 0.022
1.067 1.467 1.240 0.880
W541 W541M Basic Sar
– – –
0.007 0.006 0.007
0.453 0.367 0.258
be due to the yield of the reaction for ligand coupling, which takes place after mercerization. Another effect that should be taken into consideration is the effect of mercerization on the cellulose structure: the fibres increase their diameter whilst the diameter of the membrane pores is reduced [17]. This could well have an effect on the accessibility of the active sites within the pores. In any case an increase of static binding capacity of 38% is a relevant result and mercerization should be further investigated as a possible pre-treatment on a process scale. The static binding capacity of MCL on the native cellulose matrices was measured to account for non-specific adsorption on the different supports. The results are reported in Table 3, in which the static binding capacity, generally calculated in terms of mg of protein per cm2 of membrane, has been calculated also in terms of mg of protein per mL of membrane to account for the different thickness of the two films. The static binding capacity is of the same order of magnitude for all the membranes used, and the highest value of capacity is obtained for the mercerized membranes. Arabinogalactan gives higher binding capacity than N-acetyl-d-galactosamine; this result has been confirmed by data obtained with different proteins, and in particular with PNA. The binding capacity of Whatman 541, modified with the different ligands and for all the proteins tested, is reported in Table 4. Indeed, N-acetyl-d-galactosamine is the ligand that gave lowest binding capacity, whilst guar gum and arabinogalactan gave similar performances. However, guar gum is very difficult to handle because it tends to gel and, due to this reason, arabinogalactan has been preferred. 3.2. Dynamic binding capacity Continuous flow experiments were performed in order to calculate the dynamic binding capacity of the membranes and to study the effects of superficial velocity on the dynamic binding Table 4 W541 membrane capacity for the different ligands tested with different lectins Static binding capacity, qm (mg/cm2 )
Fig. 6. Adsorption isotherms of MCL on arabinogalactan-Whatman 541 (W541AraGal) obtained from batch experiments. The calculated data were obtained by non-linear, least squares regression of experimental data. Comparison between mercerized and non-mercerized supports (R2W541 = 0.780; R2W541M = 0.852).
MCL RCA120 PNA
AraGal
GalNAc
Guar gum
0.016 – 0.030
0.006 0.009 0.006
– 0.022 0.019
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 Table 5 Dynamic binding capacity for the different supports tested Membrane
W541 SarEpo SarEpo
Ligand
AraGal AraGal GalNAc
with a kinetic equation. On a rather general basis, the binding–unbinding kinetic model considers the adsorption and desorption reactions as:
Dynamic binding capacity (mg/mL) v1 = 0.09 cm/min, tads = 180 min
v2 = 0.18 cm/min, tads = 90 min
dq = k1 [(qm − q)(c0 − αq) − Kd q] dt
0.400 0.448
0.333 0.456 0.328
By assuming that adsorption is the result of the direct interaction between the protein in solution and the active sites, the forward reaction is linear with the protein concentration in solution and linear with free active site concentration onto the membrane surface. If we consider only the initial instants of adsorption, the desorption reaction becomes negligible and Eq. (3) reduces to:
capacity. The results obtained for the affinity membranes tested in this configuration are reported in Table 5 in which the dynamic binding capacity for the different supports is reported for the two different superficial velocities investigated, v1 = 0.9 cm/min and v2 = 0.18 cm/min. It is apparent that the dynamic binding capacity does not vary much with the superficial velocity, in the small range of velocities investigated. In any case the dynamic binding capacity is always smaller than the static binding capacity and this holds true for all the affinity membranes tested. In Fig. 7 a typical concentration profile for a dynamic experiment of MCL is reported. During the adsorption step, the feed solution is continuously recirculated to the membrane cell; this is the reason why the adsorption curve is different from a typical breakthrough curve. The small elution peak represents one of the main drawbacks of this process and the quantity of protein that was eluted is not comparable to the amount of protein adsorbed. Different buffers were tested in order to improve the elution step, among those 0.1 M glycine pH 2.8 and 0.1 M citric acid pH 2.5, but the amount of protein recovered was always lower than the amount obtained with 0.2 M lactose. 3.3. Kinetic study In order to investigate the kinetic behaviour of the different supports, for some batch experiments, the concentration of protein in solution has been monitored as a function of time. The interpolation curve has been obtained by coupling the mass balance equation for the batch system: V
dc dq = −A dt dt
17
dq = k1 qm c0 dt
(3)
(4)
Short-times experimental data have been interpolated by using Eq. (4), together with the mass balance, Eq. (2), in order to obtain directly k1 , the kinetic constant of the adsorption reaction, which is the most representative parameter of the kinetic behaviour of the system. In Fig. 8 the decrease of dimensionless concentration of MCL is reported as a function of time for Sartobind epoxy membranes modified with two different ligands. These experiments have been conducted at the same operating conditions: membrane area of 20 cm2 , initial protein concentration 0.3 mg/mL, adsorption time 2 h. Similar experiments have been performed for W541, with the aim to investigate the effects of mercerization on the adsorption kinetics. We used arabinogalactan as ligand, because it is cheaper than N-acetyl-d-galactosamine and shows higher binding capacity. In Fig. 9 the comparison between non-mercerized and mercerized supports is reported. The operating conditions were the same for both experiments: membrane area of 9.8 cm2 , initial protein concentration 0.3 mg/mL, adsorption time 2 h. From Fig. 9, we can conclude that mercerized membranes, which have higher binding capacity with respect to non-mercerized membranes and have also a faster binding kinetics.
(2)
Fig. 7. Concentration profiles of the adsorption, washing and elution curves during a dynamic experiment of MCL.
Fig. 8. Adsorption kinetics of MCL on Sartobind epoxy membranes (SarEpo) obtained from batch experiments: comparison between supports, modified with two different ligands: arabinogalactan (AraGal) and N-acetyl-d-galactosamine (GalNAc).
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19
18
The resulting affinity membranes were tested for adsorption in batch and dynamic experiments and were characterised in terms of binding capacities and kinetic parameters. Among the carbohydrates tested, N-acetyl-d-galactosamine gave the worst performances in terms of binding capacity but it was the ligand that showed highest selectivity; guar gum gave promising results but it was used only in a few experiments because it gels and it was very difficult to manipulate. Arabinogalactan was the ligand of choice due to its good binding capacity, easiness of use and price. Elution remains the critical step of the process and still needs to be optimised. The affinity membranes presented in this work should be tested with the aqueous extract from M. charantia seeds and this is the first objective for future work. Fig. 9. Adsorption kinetics of MCL on arabinogalactan-Whatman 541 obtained from batch experiments: comparison between mercerized (W541M) and nonmercerized (W541) supports. Table 6 Kinetic constant k1 for the different supports tested Membrane
Ligand
k1 mL/(mg min)
STD mL/(mg min)
W541 W541M SarEpo SarEpo
AraGal AraGal AraGal GalNAc
0.167 0.426 0.082 0.107
0.057 0.160 0.015 0.008
Acknowledgements This work was financially supported by MIUR, Italian Ministry of Education, University and Research (PRIN 2004-05) and by the University of Bologna, Italy. Special thanks go out to Prof. Stirpe of the Department of Experimental Pathology of the University of Bologna, for providing the M. charantia lectin used in the experiments. Sartorius A.G. is thankfully acknowledged for supplying the Sartobind epoxy-activated membranes.
Nomenclature In order to make quantitative comparison among the different behaviours, all the kinetic experiments have been interpolated with the short times analysis model, obtaining the kinetic constant of the adsorption reaction, k1 . In Table 6 the mean values obtained for the different affinity membranes studied are reported. Indeed, the mercerization pre-treatment has the effect to change the structure of cellulose, making it more amorphous, that leads not only to an increase in the number of the active sites, but it also increases the accessibility of the active sites. Not important differences have been observed using different ligands: Sartobind epoxy membranes modified with arabinogalactan and N-acetyl-d-galactosamine are characterized by kinetic constants of the same order of magnitude, although the latter ligand has k1 30% higher. A greater difference results by making a comparison between the two supports: Whatman filters show a binding kinetics much faster than the Sartobind epoxy membranes; the kinetic constants is about five time higher. So far there is not a satisfactory explanation of this behaviour: a possible reason could be the different accessibility of the active sites, that may be affected by different factors as support, spacer arm or external mass transfer phenomena.
A c k1 k2 Kd q q* t v V
frontal membrane area (cm2 ) protein concentration in the liquid solution (mg mL−1 ) rate constant of the complex formation reaction (mL mg−1 min−1 ) rate constant of the complex breakdown reaction (min−1 ) dissociation constant (mg mL−1 ) protein surface concentration (mg cm−2 ) binding capacity (mg mL−1 ) time (min) superficial velocity (cm min−1 ) solution volume (mL)
Greek symbol α working parameter defined as A/V (cm) Subscripts 0 initial condition ads adsorption eq equilibrium condition m maximum
4. Conclusions Affinity membranes for the adsorption of M. charantia lectins were obtained starting from two different cellulosic matrices. Mercerization of the native cellulose support was also investigated as a possible beneficial pre-treatment. Arabinogalactan, N-acetyl-d-galactosamine and guar gum were the ligands coupled to the supports.
References [1] H.B.F. Dixon, Defin. Lectin, Nat. 292 (1981) 192. [2] I.E. Liener, N. Sharon, I.J. Goldstein, The Lectins: Properties, Functions and Applications in Biology and Medicine, Academic Press, Inc., Orlando, 1986.
C. Boi et al. / Journal of Membrane Science 273 (2006) 12–19 [3] H. Lis, N. Sharon, Lectins: carbohydrate-specific proteins that mediate cellular recognition, Chem. Rev. 98 (1998) 637. [4] L. Barbieri, M. Zamboni, E. Lorenzoni, L. Montanaro, S. Sperti, F. Stirpe, Inhibition of protein synthesis in vitro by proteins from the seeds of Momordica charantia (Bitter pear melon), J. Biochem. 186 (1980) 443. [5] M. Tomita, T. Kurokawa, K. Onozaki, N. Ichiki, T. Osawa, T. Ukita, Purification of galactose-binding phytoagglutinins and phytotoxin by affinity column chromatography using Sepharose, Experientia 28 (1972) 84. [6] T. Mazumder, N. Gaur, A. Surolia, The physicochemical properties of the galactose-specific lectin from Momordica charantia, Eur. J. Biochem. 113 (1981) 463. [7] P. Padma, S.S. Komath, S.K. Nadimpalli, M.J. Swamy, Purification in high yield ann characterisation of a new galactose-specific lectin from the seeds of Trichosanthes cucumerina, Phytochemistry 50 (1999) 363. [8] V. Hoˇrejˇs´ı, M. Tich´a, J. Novotny, J. Kocourek, Studies on lectins XLVII. Some properties of d-galactose binding lectins isolated from the seeds of Butea frondosa, Erythrina indica and Momordica charantia, Biochim. Biophys. Acta 623 (1980) 439. [9] E. Klein, Affinity Membranes, Wiley, 1991. [10] M. Phillips, J. Cormier, J. Ferrence, C. Dowd, R. Kiss, H. Lutz, J. Carter, Performance of a membrane adsorber for trace impurity removal in biotechnology manufacturing, J. Chromatogr. A 1078 (2005) 74. [11] H.L. Knudsen, R.L. Fahrner, Y. Xu, L.A. Norling, G.S. Blank, Membrane ion-exchange chromatography for process-scale antibody purification, J. Chromatogr. A 907 (2001) 145.
19
[12] L. Barbieri, E. Lorenzoni, F. Stirpe, Inhibition of protein synthesis in vitro by a lectin from Momordica charantia and by other haemagglutinins, J. Biochem. 182 (1979) 633. [13] G.L. Nicolson, J. Blaustein, M.E. Etzler, Characterization of two plant lectins from Ricinus communis and their quantitative interaction with a murine lymphoma, Biochemistry 13 (1974) 196. [14] M.E.A. Pereira, E.A. Kabat, Immunochemical studies on blood groups LXII. Fractionation of hog and human A, H, and AH blood group active substance on insoluble immunoadsorbents of Dolichos and Lotus lectins, J. Exp. Med. 143 (1976) 422. [15] F. Cattoli, G.C. Sarti, Purification of MBP--galactosidase and MBPrubredoxin through affinity membrane separation, Sep. Sci. Technol. 37 (2002) 1699. [16] L. Sundberg, J. Porath, Preparation of adsorbent for biospecific affinity chromatography. I. Attachment of group-containing ligands to insoluble polymers by means of bifunctional oxiranes, J. Chromatgr. 90 (1974) 87. [17] E. Ruckenstein, W. Guo, Crosslinked mercerized cellulose membranes and their application to membrane affinity chromatography, J. Membr. Sci. 193 (2001) 131. [18] R. Tyagi, R. Agarwal, M.N. Gupta, Purification of peanut lectin using guar gum as an affinity ligand, J. Biotechnol. 46 (1996) 79. [19] K.C. Gupta, M.K. Shani, B.S. Rathaur, C.K. Narang, N.K. Mathur, Gel filtration medium derived from guar gum, J. Chromatgr. 169 (1979) 183. [20] T. Majumdar, A. Surolia, A large scale preparation of peanut agglutinin on a new affinity matrix, Prep. Biochem. 8 (1978) 119. [21] G.T. Hermanson, A.K. Mallia, P.K. Smith, Immobilized Affinity Ligand Techniques, Academic Press, 1992.