Maleic Anhydride-alt-methyl Vinyl Ether - American Chemical Society

Unité Mixte UMR-103, CNRS-bioMérieux, ENS-Lyon, 46, allée d'Italie, 69364 Lyon ... improved the grafting reaction which could take place under many...
34 downloads 0 Views 146KB Size
Download PDF
Bioconjugate Chem. 1998, 9, 655−661

655

Covalent Immobilization of Proteins onto (Maleic Anhydride-alt-methyl Vinyl Ether) Copolymers: Enhanced Immobilization of Recombinant Proteins Catherine Ladavie`re,† Thierry Delair,*,† Alain Domard,‡ Armelle Novelli-Rousseau,† Bernard Mandrand,† and Franc¸ ois Mallet† Unite´ Mixte UMR-103, CNRS-bioMe´rieux, ENS-Lyon, 46, alle´e d’Italie, 69364 Lyon, France, and Unite´ Mixte CNRS-Universite´ Claude Bernard, UMR-5627, LEMPB, Universite´ Claude Bernard, 43, Bd. du 11 Novembre 1918, 69612 Villeurbanne, France. Received November 28, 1997; Revised Manuscript Received August 1, 1998

Two genetically modified HIV-1 capsid p24 proteins, RH24 and RH24K, were covalently bound to maleic anhydride-alt-methyl vinyl ether (MAMVE) copolymer, under aqueous conditions. We demonstrated that the addition of a six lysine unit tag at the COOH-terminus of RH24K greatly improved the grafting reaction which could take place under many different experimental conditions. The course of the reaction was controlled by electrostatic attractive forces between the protein and the negatively charged polymer, as the chemical binding was more efficient at low ionic strength. The maximum loading capacity of the polymer depended on whether the protein bore the lysine tag (RH24K) or not (RH24). Twenty-four molecules of RH24 could be immobilized per polymer chain and 49 for RH24K. Such a difference could be explained by a difference of orientation of the protein on the polymer, side-on for RH24 and end-on for RH24K to account for the observed high packing density.

INTRODUCTION

Molecular recognition is a fundamental means of conveying information in living organisms. For applications such as in diagnostics, it may be required to immobilize proteins onto polymers, via the formation of covalent bonds, with ideally no loss of biological properties. To attain this goal, one strategy consists of directing the chemical reaction involved in the grafting, toward a specific site of the protein that is not essential for the molecular recognition process of interest. For instance, Rodwell et al. (1), used the carbohydrate moieties located at the hinge region of the immunoglobulin to target a chemical modification. Protein modifications can be used to introduce a functional group on a specific site of the macromolecule. Enzymatic digestion of antibodies with subsequent reduction of disulfide bonds leads to the formation of thio groups exactly on the opposite end of the recognition site of the immunoglobulin (2). Genetic engineering allows the introduction or the substitution, at a well-defined position on the protein, of amino acids that bear the ad hoc functional group for further chemical modification, such as coupling. Chilkoti et al. (3), for instance, incorporated a unique cysteine residue in mutant cytochrome b5 for conjugation to maleimide-terminated oligo(Nisopropylacrylamide). A closely related approach consists of introducing an oligo(amino acid) sequence, tags, at a defined position on the protein. For instance, oligo(histidine) tags were introduced in recombinant HIV-1 p24 capsid protein, to allow its purification via coordination interactions between histidines and metal ion immobilized on a solid phase (4, 5). An oligo(arginine) tag * To whom correspondence should be addressed. † CNRS-bioMe ´ rieux. ‡ Universite ´ Claude Bernard.

was reported (6) for solid-phase immobilization of the protein via electrostatic interactions with a negatively charged surface. In the lab, we used a strategy related to the second approach described above. A six lysine residue tag was added at the COOH position of a recombinant p24 HIV-1 capsid protein. The rationale was that an increase of the density of amine groups at a particular position of the protein molecule, should enhance the efficiency of the immobilization reaction, as a whole, and would direct the grafting to the lysine-enriched site of the biological molecule. In the absence of the tag, the grafting onto the polymer would take place at random, on any of the 10 lysine residues available on the macromolecule. The great advantage in using amide bond formation to tether biomolecules to carriers is that the chemistry is simple to run, reproducible, and can sustain a diversity of experimental conditions which is a preequisite for the development of a generic grafting method. Thiol chemistry, though very chemiospecific and regiospecific, suffers from the dimerization of the reactants by disulfide bond formation which easily occurs. To suppress this side reaction, strict experimental conditions have to be used (absence of traces of oxygen and divalent cations and no basic pH), which on an every day basis make this chemistry less attractive. The p24 protein, and its recombinant equivalents, are commonly used in diagnostics tests for the detection of HIV antibodies in sera. On the basis of results obtained for the hepatitis B virus (HBV) DNA detection (7), the use of polymeric p24 in diagnostics tests should allow us to increase the sensitivity of the kits, for a still more reliable investigation of HIV infected patients. We wish to report here on the comparative results obtained in the immobilization of two recombinant proteins: one with a six lysine residue tag and the other deprived of lysine tag.

10.1021/bc970208i CCC: $15.00 © 1998 American Chemical Society Published on Web 10/02/1998

656 Bioconjugate Chem., Vol. 9, No. 6, 1998 EXPERIMENTAL PROCEDURES

Materials. Recombinant proteins were obtained in Escherichia coli from a procedure derived from ref 5. These proteins were termed RH24 and RH24K for, respectively, recombinant p24 with a six histidine residue tag (H) for metal chelate purification and recombinant p24 with a histidine (H) tag and the six lysine (K) residue tag. For both proteins, the histidine tags were at the NH2 termini. For RH24K, the lysine tag was at the COOHterminus. The amino acid sequence for each protein is reported as follows:

The lysine groups are randomly dispersed along the protein macromolecules. NMR (8) and crystal structure data (9, 10) demonstrated that, in the 3D structure of the protein, the 10 lysine groups were accessible for grafting. Theoretical molecular weights and isoelectric points were determined with the MacVector software (version 6.01). The maleic anhydride-alt-methyl vinyl ether copolymer [P(MAMVE): Mn ) 67 000 g/mol] was from Polysciences Inc. and used as received. Other chemicals were from Aldrich and used as received. Milli-Q grade (Millipore SA, France) deionized water was used for all buffers. Coupling of Proteins to Copolymers. The appropriate amounts of polymers were dissolved in anhydrous dimethyl sulfoxide (DMSO) at 37 °C. To a protein buffer solution at the appropriate concentrations was added the required volume of polymer solution in DMSO. The reaction mixture was stirred for 3 h at 37 °C. The pH of the buffers was measured and monitored using an Ingold HA405-DXK-58/120 electrode and a Minisis 8000 pH-meter (Taccussel-Radiometer, France). As stated by the supplier, the electrode was Tris compatible. Kinetics of the Coupling Reaction. To monitor the course of the coupling of proteins onto polymers, 20 µL of the reaction mixture was pipetted off and the reaction was quenched by addition of 5 µL of 15% ammonium hydroxide solution. Protein Interaction with the Hydrolyzed Copolymer. The copolymer sample was hydrolyzed at the same concentration as for coupling (stock concentration, 1 g/L), during 2 h at 67 °C, then 48 h at 37 °C in various buffers (pH 4.0, 6.8, 7.8, 8.2, 9.2, and 10.7). Then, the coupling reaction was performed as described above. Stability of Proteins and Conjugates. This study was carried out for the proteins RH24 and RH24K and for the conjugates in the coupling medium (DMSO/ Tris buffer, pH 7, 5/95, [protein] ) 0.95 g/L, [P(MAMVE)] ) 0.048 g/L) at 37 °C, for 7 days. Prior to use, the buffer was filtered on a 0.22 µm membrane and pipet tips and Ependorff tubes were autoclaved (3 h at 120 °C). Analyses of the Coupling Reactions. The crude coupling mixture was analyzed by size-exclusion chro-

Ladavie`re et al.

matography (SEC) using a Shodex Protein KW-803 (Waters) or a Protein-Pak 125 (Waters), a Kontron HPLC 422 pump, a Kontron HPLC autosampler 465, and a Kontron UV diode array detector. Purifications were run in a 0.1 M phosphate buffer, pH 6.8, containing 0.5% (w/ v) of sodium dodecyl sulfate (SDS) with a flow rate of 0.5 mL/min. Detection was achieved by measuring the absorbance at 280 nm corresponding to the proteins (at the concentrations used, the polymer had no absorption). The ratio of the peak area corresponding to the polymerbound RH24(K) versus the sum of the two peaks corresponding to the unbound and to the bound RH24(K) (i.e., the total amount of protein involved in the reaction) gave the coupling yield (Y). The extinction coefficients of the bound and unbound proteins were checked by measuring the OD at 280 nm of crude conjugates obtained in a carbonate buffer compared to the reference-free protein in the same mixture. The OD data were found identical in a 10% range. Furthermore, for each HPLC analysis, the sum of the areas of the bound and unbound proteins was routinely compared to that of a reference sample recorded in the absence of polymer. To assess the average amount of protein molecules per polymer chain, N h ) nY/n′ where n and n′, respectively, correspond to the number of protein molecules and the number of polymer chains in the reaction mixture; as a first approximation, we assumed, that each polymer chain reacted according to the same pattern [which is the case for oligonucleotide immobilization (7)]. Trying to assess the molecular weight of the conjugates by acrylamide gel electrophoresis was a failure probably because of polymergel interactions. Using viscometry on-line with SEC, was not possible due to the presence of SDS in the mobile phase (unstable signals were obtained). To calculate N h, which gives access to the level of loading of the polymer, we assumed that no cross-linking had occurred. This assumption seems realistic given that the protein to polymer molar ratio is 50:1. Dynamic Light-Scattering Experiments. These experiments were performed on-line using a Protein-Pak 200 SW column (Waters) and a Waters 510 highperformance liquid chromatography pump, running in a 0.1 M phosphate buffer, pH 6.8, containing 0.5 M NaCl with a flow rate of 0.5 mL/min. The refractive index increment of proteins is ca. 0.19 mL/g (11). For the detection part, a Waters 484 absorbance detector, a Waters 410 differential refractometer, and a three angles Mini Dawn F detector (WYATT Technology) at 690 nm were used on-line. Mass Spectrometry (LC-ESI-MS). Analyses were performed with a single quadrupole API 100 mass spectrometer, 140 B pumps, and a 785 A detector (PerkinElmer). Reversed-phase liquid chromatography was performed on a C18 column (Vydac, 5 µm particle size). Solvent A was 0.1% (v/v) trifluoroacetic acid in water and solvent B was 0.1% (v/v) trifluoroacetic acid in a water/ acetonitrile mixture (20:80 v/v). A gradient from 45 to 70% of B was used. Circular Dichroism (CD). Analyses were performed in a 0.05 M Tris-HCl buffer, pH 7, with a protein concentration between 30 and 50 mM. CD measurements were performed with a Jobin-Yvon model CD6 spectrophotometer at 25 °C and with a Hellma cuvette; path length-0.2 cm. To eliminate the DMSO used for the coupling (high absorption of solvent), the conjugates were analyzed after dialysis on a polycarbonate membrane with a cutoff of 3000 g/mol (AMICON). This study was carried out with a conjugate of RH24K because the

Enhanced Immobilization of Recombinant Proteins

Bioconjugate Chem., Vol. 9, No. 6, 1998 657

Table 1. Molecular Weight and Isoelectric Point of HIV-1 Capsid Recombinant Proteins

R24 RH24 RH24K a

theoretical MW (g/mol)

LC-ESI-MS (g/mol)

SEC-MALLS (g/mol)

theoretical IPa

24 736 26 950 28 022

26 959 28 026

24 680 27 500 29 100

5.5 5.9 6.9

IP ) isoelectric point.

Figure 2. Coupling yield of the RH24 (hatched bar) and RH24K (stitched bar) to P(MAMVE) copolymer at 37 °C, 3 h [protein] ) 0.95 g/L, [P(MAMVE)] ) 0.048 g/L (protein to polymer molar ratio: 50:1). P, 0.1 M phosphate buffer; T, 0.05 M Tris buffer; C, 0.1 M carbonate buffer; B, 0.1 M borate buffer, pH 9, 0.01 M borate pH 11.

Figure 1. SEC trace of the coupling reaction mixture showing two kinds of peaks corresponding to (1) conjugate and (2) unreacted RH24K molecules.

coupling yield is high (∼88%) and as a consequence the unbound protein does not perturb the analysis. RESULTS AND DISCUSSION

Protein Characterizations. The molecular weight of each protein was determined by size-exclusion chromatography-multiangle laser light scattering (SECMALLS) and reversed-phase liquid chromatography coupled with electrospray ionization mass spectrometry (LC-ESI-MS). Data depicted in Table 1 show a good correlation between the two techniques and the theoretical value deduced from the amino acid composition. In comparison to RH24 and RH24K proteins, data concerning R24 that bear no tag are shown as well. CD spectra of RH24 and RH24K exhibited the same pattern as for the R24 protein reported in the literature (12), and the conformation corresponded to 43% R-helices and 17% β-sheets. This is an important result which demonstrated that no major conformational modification was induced by the addition of one or two tags. Characterization of the Protein/Polymer Conjugates. Conjugates were analyzed and purified by sizeexclusion HPLC. As shown in Figure 1, the polymer protein conjugates were recovered in the first elution peak and the unbound proteins in the second. We noticed a tendency of the protein to stick to many solid phases. That can probably be explained considering the role of a p24 protein, which is to protect the genetic code of the virus by aggregating to form a capsid. To overcome this autoassociation problem, whose dissociation constant of corresponding dimer, measured by Rose´ et al. (13), was (3-4) × 10-5 M at pH 7, we added 0.5% (w/v) of sodium dodecyl sulfate in the HPLC mobile phase.

Determination of the Coupling Conditions. On the basis of results obtained in a preliminary study on bovine serum albumin, the grafting of recombinant proteins onto MAMVE copolymer was tested in aqueous buffers containing 5% (v/v) of DMSO (14). The results obtained, depicted in Figure 2, show that, in most cases, the grafting of the lysine-tagged protein, RH24K, was more efficient than the immobilization of the RH24 HIV-1 capsid protein. In the latter case, the coupling yields were higher at pHs close to neutrality and fell with increasing pH as the hydrolysis rate increased (15). The lysine-tagged protein reacted with the anhydride moieties of the MAMVE copolymer over a wide range of experimental conditions, demonstrating the enhancing effect of the tag on the course of the reaction. Coupling yields were over 50% except for extreme conditions, i.e., low (pH 5) or high (pH 11) pHs. The Tris buffer proved the most efficient medium for the covalent grafting of both proteins, demonstrating the poor reactivity of the amine of the Tris buffer toward acylation. It is worth noting that, in the phosphate buffer, pH 5, the conjugates precipitated but the proteins did not. The effect of the lysine tag on the course of the coupling reaction was drastically marked at pHs over 7. Above pH 7, the immobilization yields of RH24 leveled off at ca. 10%, whereas 60-88% yields were obtained for RH24K. To explain the difference in behavior between both proteins, one can consider two factors: (i) different conformations, one of which would favor the coupling reaction via steric effects, and (ii) difference in reactivities due to the presence of the lysine tag in RH24K. CD spectra of RH24 and RH24K were identical, proving that the introduction of the lysine tag entailed no modification of the secondary structure of the protein. Therefore, conformational factors cannot be regarded responsible for the difference in behavior between the two proteins. Hence, reactivity factor must be evoked to account for such a large variation in behavior. The immobilization reaction onto the MAMVE copolymer occurs by nucleophilic attack of the reactive anhydride moieties by the primary amine groups of the proteins, as depicted in Figure 3. Since the reaction takes place in a mostly aqueous medium, the hydrolysis of the anhydride functions is very likely to occur. Therefore, the grafting pathway of the protein onto the polymer has to compete the hydrolysis reaction. The kinetics data reported in

658 Bioconjugate Chem., Vol. 9, No. 6, 1998

Ladavie`re et al.

Figure 3. Coupling reaction between RH24(K) molecule and copolymer P(MAMVE) (a) and hydrolysis reaction of copolymer (b). Table 3. Coupling Yields of RH24 to MAMVE Copolymer as a Function of Ionic Strength (Two Different Experiments) coupling yield (%) 0.05 M Tris buffer pH 7 [NaCl] (mol/L) 1 0.75 0.5 0.25 0.1 0

Figure 4. Kinetics of RH24 (O) and RH24K (b) coupling reaction onto P(MAMVE) sample. Buffer: 0.05 M Tris, pH 7. Table 2. RH24K Coupling Yields vs Buffer Ionic Strengtha [carbonate], pH 8.2 (mol/L) 0.1 0.05 0.02

coupling yield (%) RH24 RH24K 18 21 26

exp 2

22 33 43 52

3 12 21 33 50

a [protein] ) 0.95 g/L; [polymer] ) 0.048 g/L; protein/polymer molar ratio ) 50:1.

Table 4. Effect of the Protein Concentration on the Coupling Yields in 0.05 M Tris Buffer, pH 7 coupling yield (%)b run 1 2 3 4 5 6

53 85 94

a [protein] ) 0.95 g/L; [polymer] ) 0.048 g/L; protein/polymer molar ratio ) 50:1.

Figure 4 were obtained in a 50 mM Tris buffer, pH 7, at 37 °C, and the coupling yields were measured after quenching the reaction with excess ammonia. The presence of ammonia did not alter the stability of the amide bond formed during acylation with the polymer as the amount of bound protein did not decrease over the analysis period (illustrated by the presence of plateaus in Figure 4). Furthermore, the plateau values of the coupling yields obtained in the presence of ammonia perfectly matched the coupling yields obtained without ammonia as seen in Table 4, run 5, and Table 5, run 5. After 10 min of reaction, the coupling yields did not evolve any more, meaning that the reaction had reached a plateau and that all reactive groups on the polymer were consumed, either to form an amide bond with the protein or by hydrolysis by water molecules. For both proteins, the reaction did not proceed any further after 10 min. Nevertheless, for the same time period, more RH24K molecules were bound to the polymer than RH24. It was checked that no coupling between the polymer and RH24 occurred when all the anhydride groups of the

exp 1

[protein]a

(g/L)

0.095 0.19 0.48 0.76 0.95 1.90

RH24

RH24K

100 94 75 58 47 24

96 95 95 92 88 50

a Final concentration. [polymer] ) 0.048 g/L; 37 °C, 3 h. b Mean value of two experiments.

Table 5. Polymer Concentration Effect on the Coupling Yields coupling yield (%)b run

[P(MAMVE)]a (g/L)

RH24, Tris, pH 7

RH24K, carbonate, pH 8.2

1 2 3 4 5

0.0048 0.0095 0.024 0.038 0.048

11 18 32 40 46

11.5 22 50.5 70 80

a Final concentration, [protein] ) 0.95 g/L; 37 °C, 3 h. b Mean value of two experiments.

polymer were hydrolyzed (at different pHs: 4.0, 6.8, 8.2, and 10.6). Actually, the mixture of hydrolyzed polymer and RH24K always led to a precipitate (used buffers pHs: 4.0, 6.8, 7.8, 8.2, 9.2, and 10.7). This precipitate could be explained by electrostatic interactions between the negative charges of polymer chains and the cationic sites

Enhanced Immobilization of Recombinant Proteins

of the protein molecules, leading to highly cross-linked polyelectrolyte complexes insoluble in water. When the ratio polymer/protein was increased, no precipitation was observed as more polymer was available. Under these conditions, no conjugation occurred as observed by HPLC. The reproducibility of the grafting experiments is quite good, based on the following results obtained on different days. The immobilization yields of RH24K were 82, 80, and 88% for experiments described respectively in Figure 4, Table 5 (run 5), and Table 4 (run 4). For the same series of experiments, coupling yields of RH24 were 48, 46, and 47%. Role of the Ionic Strength. With the lysine-tagged protein, a series of experiments was run in a carbonate buffer in which the immobilization yield was only 53% and the coupling yield was measured as the buffer ionic strength was lowered. The results reported in Table 2 show a drastic increase in coupling yields when lowering the ionic strength and clearly emphasize the major role played by electrostatic interactions in the immobilization process of RH24K. It is worth noting that, for a 20 mM carbonate buffer, the pH of the coupling mixture only decreased 0.15 pH unit in 53 min, proving that the buffer was not exhausted by the hydrolysis of the anhydride groups of the polymer. Though at pH 8.2 the protein net charge is negative, some protonated amine groups remain. Since negative charges are formed on the polymer, either from the hydrolysis of the anhydride moieties or from the coupling reaction, attractive electrostatic forces can arise by interaction with the positive charges borne by the protein. These forces allow the macromolecular counterparts to get in contact for the chemical reaction to take place. Finally, the results obtained in Table 2 point out a fundamental difference in behavior of the two proteins. Lowering the ionic strength had a slight effect on the coupling yield for RH24, whereas a drastic increase in coupling yields was observed with RH24K. To observe any ionic strength effect with RH24, sodium chloride was added to a Tris buffer, pH 7. No data were available for the RH24K protein because, in this experiment, precipitation was observed on adding sodium chloride (the protein alone precipitated too). From the results shown in Table 3, an increase of ionic strength by adding sodium chloride resulted in a decrease of the grafting efficiency. A slight precipitation was observed in the Tris buffer at sodium chloride concentrations higher than 0.5 M. In another set of experiments, the ionic strength of the Tris buffer, pH 7, was reduced. No improvement of the immobilization yields was obtained, neither for the lysine-tagged protein, which was not surprising since the coupling yields were already as high as 80%, nor for the RH24 sample for which the maximum coupling yield was limited to ca. 40% (data not shown). This behavior was not expected on account of results obtained with BSA, for which coupling efficiency was improved by reduction of the buffer ionic strength (14). From these results, one may conclude that the covalent immobilization of RH24K is mostly governed by electrostatic interactions. These interactions are attractive forces between the negatively charged polymer and the positive charges on the protein. The role of these forces is to bring together the macromolecular reactants in order to allow the chemical reaction to take place. For RH24, the coupling process is controlled by a diffusional process as well as by the differences in reactivities between the amino groups of the protein and the water molecules of the reaction mixture that hydrolyze the reactive groups on the polymer.

Bioconjugate Chem., Vol. 9, No. 6, 1998 659

Figure 5. Number of protein molecules bound per polymer chain (N h ) for RH24 (O) and RH24K (b) as a function of protein concentration during coupling. Buffer: 0.05 M Tris, pH 7. Reaction time and temperature 3 h, 37 °C.

Figure 6. Representation of end-on (a) and side-on (b) orientations.

Role of the Reactant Concentration. The relative concentrations of both reactants are factors that can have an effect on the course of the binding reaction. The increase of the protein concentration in the reaction mixture resulted in a more marked decrease of the immobilization yields of RH24 than of RH24K as seen in Table 4. More striking is the effect on the average amount of protein loaded per polymer chain, as represented by N h in Figure 5. The amount of bound RH24 per polymer chain leveled off at 24 protein molecules, whereas no plateau value was reached yet for RH24K, though coupling yields started to significantly decrease at a concentration of 0.95 g/L (Table 4). The plateau value, obtained on increasing RH24 concentrations, suggests that the grafting reaction is controlled by a steric effect: the polymer chain is covered with proteins which adopted some sort of side-on orientation. For the same polymer chain, more than 49 molecules of RH24K can be tethered, which must have adopted an end-on type orientation. Due to the presence of positive charges on the six lysine residue terminus of RH24K, the tag favors the approach of a protein molecule toward a negatively charged P(MAMVE) chain, in some sort of an oriented manner. In other words, the tag acts as a leash to bind the protein to the polymer. The amino groups of the lysine tag are the first to get in contact, and react, with some anhydride moieties of the polymer which ensures an oriented immobilization of RH24K. When the concentration of RH24K is high, many proteic species can be grafted onto one polymer chain, preventing, thus, the protein to adopt a side-on orientation (Figure 6). The next factor to be investigated was the polymer concentration in the coupling medium. This factor was investigated in a carbonate buffer for RH24K, in which RH24 very poorly binds (Table 2), which makes this buffer more selective toward the orientation of the grafting of the lysine-tagged protein. For RH24, a Tris buffer was selected which afforded a high grafting efficiency, whose variations should be quite significant. The coupling yields increased with the polymer concentrations, as reported in Table 5, and concomitant decrease

660 Bioconjugate Chem., Vol. 9, No. 6, 1998

Figure 7. Amount of RH24(K) molecules bound per polymer chain (N h ) in a 0.05 M Tris buffer, pH 7, for RH24 (O) and in a 0.02 M carbonate buffer pH 8.2 for RH24K (b) as a function of the polymer concentration. [protein] ) 0.95 g/L.

of N h was observed for both proteins as shown in Figure 7. It is worth noting that the maximum amount of RH24(K) bound to the polymer (i.e., at the lowest polymer concentration, run 1 in Table 5) was similar in both cases. The maximum loading capacity of ca. 58 protein molecules chain (Figure 7) could be reached when the reaction was run with a large excess of proteins (protein to polymer molar ratio of 500:1, run 1 in Table 5). At first, this result could challenge the side-on/end-on hypothesis postulated in the preceding paragraph. But one has to take into account that in run 1, Table 5, the protein to polymer molar ratio was 5-fold higher than in run 6, Table 4. Thus, under an extremely high concentration in protein (run 1, Table 5), RH24 was forced to adopt an end-on orientation during coupling. When the immobilization conditions were less drastic (run 6, Table 4, protein to polymer molar ratio of 100:1), the high level of loading of RH24K onto the polymer (49 proteins/chain) can only be explained if the protein binds in an end-on orientation. This orientation arises from attractive electrostatic forces generated by the interaction of the lysine tag and the polymer. This result clearly demonstrates that orientation of RH24K on binding to the polymer is achieved via electrostatic factors, and that the coupling reaction is indeed governed by these factors. In other words, binding of RH24 in an end-on orientation is under kinetic control, when a very large excess of proteins is used, the side-on orientation being obtained when the reaction is under thermodynamic control, i.e., at the lowest protein concentration. When the polymer concentration in the reaction mixture was increased, more polymer chains were available for the same initial amount of protein and we observed a decrease in N h as already observed with BSA (14). But the decrease in the case of RH24K is slower than for RH24. This difference in behavior could be explained by the already suspected higher reactivity of RH24K compared to the non-lysine-tagged protein. An increase in the polymer concentration of the reaction mixture results in an increase of the hydrolysis rate of the reactive groups (15) of the polymer. RH24 does not compete with hydrolysis as well as RH24K. When the polymer concentration was high, in large excess to the protein, the immobilization yields were maximum, and once again, higher for RH24K. It is worth mentioning that with RH24K three buffers were investigated in this series of experiments: 0.02 M carbonate buffer, pH 8.2, 0.05 M Tris buffer, pH 7, and 0.1 M borate buffer, pH 9.2. At low polymer concentrations in the Tris and borate buffers, an insoluble precipitate was formed. When enough polymer was present (0.024 g/L in Tris buffer, 0.038 g/L

Ladavie`re et al.

in the borate buffer), no insoluble precipitate was formed, and the amounts of bound proteins in the three buffers were closely related. Stability of Proteins and Conjugates. When RH24 and RH24K were incubated at 37 °C in the coupling medium (DMSO/Tris buffer, pH 7, 5/95), new peaks at higher elution times appeared, corresponding to degradation fragments of the proteins. The degradation yield increased with time at 37 °C, to reach 16% for RH24 and 20% for RH24K after 7 days. The protein/polymer conjugates remained stable at 37 °C over the same time period. In the meantime, the unbound protein contained in the crude conjugate underwent degradation. So the conjugation of the proteins to the polymer was a means of stabilizing them and protecting them from degradation. CONCLUSION

This work demonstrated that the introduction of a six lysine residue tag into a recombinant protein drastically modified the immobilization conditions onto a (maleic anhydride-alt-methyl vinyl ether) copolymer, though it only corresponded to ca. 3% of the global amino acid composition. The nontagged protein (RH24) was immobilized at buffer pHs close to neutrality in order to disfavor the competing hydrolysis of the polymer reactive groups. For the lysine-tagged protein (RH24K), the hydrolysis reaction had less impact, and the protein could be tethered under a wide variety of experimental conditions, including high pH values. For a sufficient polymer concentration, the presence of the polylysine end allowed the immobilization of up to 50 protein molecules per polymer chain. When densely packed, the protein has probably adopted an end-on orientation. The maximal density of polymer-bound nontagged protein was shown to be lower than for its tagged counterpart. This upper limit was due to steric effects as an increase of the protein concentration in the coupling medium did not allow it to bind more RH24. Taking this last result into account, one may think that the orientation of the nontagged protein on the polymer is more of a side-on type. For a weak polymer concentration (protein-to-polymer ratio of 500:1), the maximal amount of polymer-bound protein was independent of the protein nature and attained 58 protein molecules/polymer chain. So when proteins are in large excess, they are forced to bind in an end-on orientation, affording the high loading value of 58 proteins per polymer chain. When coupling conditions were less drastic (protein to polymer molar ratio of 100: 1), discrimination between the nature of the two proteins can occur. Thus, RH24K only, can bind in an end-on orientation, as shown by the high loading value of 49 proteins/chain, because of the orientation effect due to interaction of the tag with the copolymer. In the absence of these interactions, as in the case of RH24, the proteins bind according to a random orientation, which corresponds to a side-on orientation as shown by the reduced loading value of RH24 molecules. The study on the stability of protein/polymer conjugates at 37 °C showed that the polymer-bound proteins were more stable than the unbound ones. Finally, on a conformational standpoint, CD spectra of the conjugates with RH24K were identical to those of the unbound proteins, so the conjugation process occurred with no modification of the secondary structure of the bioactive species. Therefore, the evaluations of the

Enhanced Immobilization of Recombinant Proteins

biological properties of the conjugates could take place and will be reported in due course. ACKNOWLEDGMENT

The authors are grateful to Dr. L. Ladavie`re for CD measurements, Dr. G. Gervasi for analyses of proteins by mass spectrometry, and C. Hebrard for the preparation of proteins. C.L. is thankful to Fondation Me´rieux for financial support. LITERATURE CITED (1) Rodwell, J. D., Alvarez, V. L., Lee, C., Lopes, A. D., Goers, J. W. F., King, H. D., Powsner, H. J., and McKearn, T. J. (1986) Site-specific covalent modification of monoclonal antibodies: in vitro and in vivo evaluations. Proc. Natl. Acad. Sci. 83, 2632-2636. (2) Delair, Th., Pichot, C., and Mandrand, B., (1994) Synthesis and characterization of cationic latexes particles bearing sulfhydryl groups and their use in the immobilization of Fab antibody fragments. Colloid Polym. Sci. 272, 72-81. (3) Chilkoti, A., Chen, G., Stayton, P. S., and Hoffman, A. (1994) Site-specific conjugation of a temperature-sensitive polymer to a genetically engineered protein. Bioconjugate Chem. 5, 504-507. (4) Ng, K., Pack, D. W., Sasaki, D. Y., and Arnold, F. H. (1995) Engineering protein-lipid interactions: targeting of histidinetagged proteins to metal-chelating lipid monolayers. Langmuir 11, 4048-4055. (5) Cheynet, V., Verrier, B., and Mallet, F. (1993) Overexpression of HIV-1 proteins in Escherichia coli by a modified expression vector and their one-step purification. Protein Expression Purif. 4, 367-372. (6) Stempfer, G., Holl-Neugebauer, B., Kopetzki, E., and Rudolph, R. (1996) A fusion protein designed for noncovalent immobilization: stability, enzymatic activity, and use in an enzyme reactor. Nat. Biotechnol. 14, 481-484.

Bioconjugate Chem., Vol. 9, No. 6, 1998 661 (7) Erout, M. N., Troesch, A., Kurfurst, R., Pichot, C., and Cros, P. (1996) Preparation of conjugates between oligonucleotides and N-vinylpyrrolidone/N-acryloxysuccinimide copolymers and applications in nucleic acid assays to improve sensitivity. Bioconjugate Chem. 5, 568-575. (8) Gitti, R. K., Lee, B. M., Walker, J, Summers, M. F., Yoo, S., and Sundquist, W. I. (1996) Structure of the amino-terminal core domain of the HIV-1 capsid protein. Science 273, 231235. (9) Gamble, T. R., Vajdos, F. F., Yoo S., Worthylake, D. K., Houseweart, M., Sundquist, W. I., and Hill, C. P. (1996) Crystal structure of human Cyclophilin A bound to the aminoterminal domain of HIV-1 capsid. Cell 87, 1285-1294. (10) Gamble, T. R., Yoo, S., Vajdos, F. F., von Schwedker, U.K., Worthylake, D. K., Wang, H., McCutcheon, J. P., Sundquist, W. I., and Hill, C. P. (1997) Structure of the carboxyl-terminal dimerization domain of the HIV-1 capsid protein. Science 278, 849-853. (11) Brandrup, J., Immergut, E. H. (1989) Polymer Handbook, 3rd ed.; p 461. Wiley, New York. (12) Misselwitz, R., Hausdorf, G., Welfle, K., Hohne, W. E., and Welfle, H. (1995) Conformation and stability of recombinant HIV-1 capsid protein p24 (rp24). Biochim. Biophys. Acta 1250, 9-18. (13) Rose´, S., Hensley, P., O′Shannessy, D. J., Culp, J., Debouck, C., and Chaiken, I. (1992) Characterization of HIV-1 p24 selfassociation using analytical affinity chromatography. Proteins: Struct., Funct., and Genet. 13, 112-119. (14) Ladavie`re, C., Delair, Th., Domard, A., Pichot C., Mandrand B. Covalent immobilization of Bovine Serum Albumin onto (maleic anhydride-alt-methyl vinyl ether) copolymers. J. Appl. Polym. Sci. (in press). (15) Ladavie`re, C., Delair, Th., Domard, A., Pichot, C., and Mandrand, B. Covalent immobilization of biological molecules to maleic anhydride and methyl vinyl ether copolymers. A physicochemical approach. J. Appl. Polym. Sci. (in press).

BC970208I