|Home | About | Journals | Submit | Contact Us | Français|
Rev is a key regulatory protein of HIV-1. Its function is to bind to viral transcripts and effect export from the nucleus of unspliced mRNA thereby allowing the synthesis of structural proteins. Despite its evident importance, the structure of Rev has remained unknown, primarily because Rev’s proclivity for polymerization and aggregation is an impediment to crystallization. Monoclonal antibody antigen-binding domains (Fabs) have proven useful for the co-crystallization of other refractory proteins. In the present study, a chimeric rabbit/human anti-Rev Fab was selected by phage display, expressed in a bacterial secretion system, and purified from the media. The Fab readily solubilized polymeric Rev. The resulting Fab/Rev complex was purified by metal ion affinity chromatography and characterized by analytical ultracentrifugation which demonstrated monodispersity and indicated a 1:1 molar stoichiometry. The Fab binds with very high affinity, as determined by surface plasmon resonance, to a conformational epitope in the N-terminal half of Rev. The complex forms crystals suitable for structure determination. The ability to serve as a crystallization aid is a new application of broad utility for chimeric rabbit/human Fab. The corresponding single chain antibody (scFv) was also prepared, offering the potential of intracellular antibody therapeutics against HIV-1.
The unique antibody repertoire development in rabbits (Oryctolagus cuniculus) has been exploited for the generation of polyclonal and monoclonal antibodies with exceptionally high avidity, affinity, and specificity 1. While rabbit polyclonal antibodies have had a long standing as research reagents, the more recent generation of rabbit monoclonal antibodies through both phage display 2; 3; 4; 5 and hybridoma technology 6 has provided access to a highly defined research reagent of unlimited supply. In addition to high affinity and specificity, rabbit monoclonal antibodies can recognize epitopes conserved between human, mouse, and rat antigens 7; 8; 9. This cross-reactivity along with the demonstration that rabbit monoclonal antibodies can be humanized has raised an interest in utilizing rabbit monoclonal antibodies for therapeutic applications 10. Their ability to target epitopes that differ from those recognized by mouse monoclonal antibodies makes rabbit monoclonal antibodies attractive research reagents for functional and biophysical studies of antigen/antibody interactions.
A rabbit monoclonal antibody format of particular interest is the chimeric rabbit/human Fab which consist of rabbit variable domains VH and Vκ and human constant domains Cγ11 and Cκ 5; 11. We have demonstrated that chimeric rabbit/human Fab libraries can be generated from spleen and bone marrow of immunized rabbits, in particular b9 allotype rabbits 7; 8; 9, and subsequently selected by phage display to yield chimeric rabbit/human Fab of high affinity, specificity, cross-reactivity, and convertibility to chimeric rabbit/human IgG1. The rabbit variable domains VH and Vκ of chimeric rabbit/human Fab can be humanized 5; 12.
While the 150-kDa IgG molecule is the most commonly used format of monoclonal antibodies in basic research as well as diagnostic, preventative, and therapeutic applications, the smaller 50-kDa Fab molecule, which can be expressed in E. coli, has facilitated the generation, affinity maturation, and humanization of monoclonal antibodies through in vitro evolution technologies, most prominently phage display 10; 13; 14. Thus, in most instances, the Fab molecule has been a facilitating format for generating and evolving IgG for particular purposes. Nonetheless, Fab have also been utilized in their own right for an increasing number of applications that exploit its smaller size and easier manufacturability compared to IgG 15. An important application in basic research is the utilization of Fab for the co-crystallization of proteins in general and transmembrane, hydrophobic, and aggregating proteins in particular. In addition to providing crystal contacts through protruding hydrophilic surfaces, Fab can support crystallization by locking in conformations and blocking aggregation. For example, Fab have been used as crystal chaperones 16 in the determination of the three-dimensional structure of transmembrane ion channels and G protein-coupled receptors 17; 18; 19. Notably, phage display has facilitated the generation and evolution of Fab with superior co-crystallization properties 16; 19.
Rev (13 kDa) is an essential regulatory protein of the HIV-1 virus which functions by binding to, and preventing splicing of, the viral mRNA, thereby facilitating transition to the late phase of the replication cycle (for review see 20). Despite its importance, and considerable efforts directed at its elucidation, the structure of Rev remains unknown, due largely to the protein’s strong propensity to polymerize into long filaments 21; 22. Here we describe the preparation, characterization, and crystallization of a complex formed between Rev and a chimeric rabbit/human Fab selected by means of phage display. The preparation of the corresponding high-affinity single chain antibody fragment (scFv), which has anti-HIV therapeutic potential, is also described.
Following immunization with purified recombinant HIV-1 Rev, spleen and bone marrow from two rabbits were collected and processed for total RNA preparation, RT-PCR amplification of rabbit Vk, Vλ, and VH encoding sequences, Vk-Ck-VH cassette assembly, and asymmetric SfiI ligation into phage display vector pC3C essentially as described 15. The resulting library, which consisted of approximately 2 × 108 independently transformed chimeric rabbit/human Fab clones, was screened by phage display on recombinant Rev protein that had been selectively biotinylated at the C-terminus and immobilized on streptavidin-coated plates. After four to five rounds of panning, selected clones were subjected to an initial characterization by ELISA, DNA fingerprinting, and DNA sequencing. All selected clones were identical. The encoded chimeric rabbit/human Fab was termed SJS-R1. The deduced amino acid sequences of the variable domains of SJS-R1, which are shown in Fig. 1, revealed unique rabbit Vk and VH sequences with the highest similarity (85% and 77% amino acid sequence identity, respectively) to b9-allotype-derived rabbit variable domains in GenBank 8.
The Fab was expressed in E. coli using an expression cassette with two N-terminal signal sequences pelB and ompA which direct the separate secretion of the Vk-Ck and VH-CH1 chains into the periplasmic space where enzymic oxidation processes form two intramolecular disulfides per chain and one inter-chain disulfide 23. Although high expression was achieved, most of the Fab protein was retained in the periplasm with only a small amount being secreted into the media. It was found that freezing and thawing of the cells followed by a brief sonication dramatically increased the yield. The His tag on the VH-CH1 chain provided the basis for affinity purification and two cycles of Ni–Sepharose chromatography, one for batch-wise capture and the second using gradient elution, gave good results in terms of yield and purity. When further purification was required, this was carried out using Superdex S200 gel filtration in the presence of 1 M urea to increase Fab solubility (Fig. 2). Using the same expression and purification methods, the corresponding single chain Fab (scFv) was also produced.
The identities of the antibody fragments were confirmed by mass spectrometry. Reduced Fab gave mass values of 23,464 Da (23,468) and 24,472 Da (24,466) for the light and heavy chains, respectively; and oxidized scFv gave a mass of 26,429 Da (26,430). Values in parentheses are those predicted from the DNA sequences (Note: the Fab heavy chain fragment and scFv include C-terminal His tags). SDS-PAGE of non-reduced Fab (Fig. 2b) and scFv (Fig. 3a, lane scFv) also gave bands with mobilities consistent with the predicted masses. Molecular weights under native conditions were measured by sedimentation equilibrium and both Fab and scFv were found to be monomeric (Fig. 4c; data for Fab not shown). Although the antibody fragments in standard buffers and at neutral pH were well behaved below 1 mg/mL, at higher concentrations there was a tendency towards aggregation and, as mentioned above, the addition of 1 M urea helped maintain solubility without compromising conformational integrity.
The binding kinetics of the Fab to Rev were measured by surface plasmon resonance with Rev immobilized on the chip and Fab as the analyte (Fig. 5). The binding was characterized by a typical on-rate (ka = 2.2 × 105 M−1s−1) but the off-rate (kd = 0.8 × 10−5s−1) was very low: a low off-rate is characteristic of high affinity (see for example, Drake et al., 24). With such a low off-rate, and with the technical limitations of the method, the subnanomolar affinity value determined (~40 pM) needs confirmation by another approach but regardless, a high affinity interaction is certainly suggested.
Rev in solution self-associates: monomers and dimers at low protein concentration associate to form high molecular weight filaments at higher concentrations 22. The polymerization fits an isodesmic self-association model in which the association constant for the addition of a monomer to each aggregate is equal with a Kd of ~1.0 μM 25. The Rev filaments (Fig. 6a) are stable and are hollow with an outer diameter of ~15 Å 21.
Addition of equimolar Fab to the filaments causes rapid depolymerization with the formation of small, uniformly sized complexes (Fig. 6b). In some views these complexes appear to have a central stain-penetrable hole (Fig. 6b, inset). In a similar manner the scFv also depolymerized Rev filaments (data not shown). In both cases, the facile disruption of the protein-protein interactions is consistent with a high affinity interaction.
To approximately locate the Rev epitope, the C-terminal deletion mutant Rev Δ60–116 was used. Electron microscopy showed that this mutant by itself cannot assemble into a filament, probably due to the absence of the carboxy-terminal half. When the scFv was added uniform complexes were formed that appeared similar, though smaller, to those observed with full-length Rev (data not shown). This suggested that the epitope is present, and located in the N-terminal 1–59 residues of Rev.
Immune complexes were prepared by mixing the antibody fragments with a several-fold molar excess of Rev and then purifying them by means of metal chelate chromatography on Ni-Sepharose taking advantage of the C-terminal His tags. SDS-PAGE of the non-reduced Fab/Rev complex gave two main bands of ~45 kDa and ~16 kDa corresponding to oxidized Fab and Rev (Fig. 3a, lane Fab/Rev). It should be noted that the predicted mass of Rev is 12,905 Da but that it appears to have an anomalously low mobility in gel electrophoresis. Reduction of the complex also produced two bands, in this case Rev and an ~25 kDa band corresponding to unresolved Fab heavy chain fragment (~24.5 kDa) and light chain (~23.5 kDa; Fig. 3b). Densitometry indicated a Fab/Rev 1:0.9 molar ratio, assuming equal Coomassie dye binding capacity. A similar analysis of the scFv/Rev complex under non-reducing (Fig. 3a, lane scFv/Rev) and reducing conditions (Fig. 3c) indicated a 1:0.87 molar ratio. Thus, both the Fab and scFv antibodies appear to form equimolar complexes with Rev.
Following affinity purification, it is common to use gel filtration to finalize purification (polish), and at the same time confirm physical homogeneity as evidenced by symmetrical elution peaks. Unfortunately, both Rev and Rev complexed with antibody fragments adsorb strongly to the commonly used gel filtration matrices ruling out this approach. As the primary aim of this work was to crystallize the immune complex it was important to ascertain monodispersity. For this purpose we used sedimentation velocity analysis after removing any protein aggregates by centrifugation at 100,000 ×g for 1–2 hours. Typical data (Fig. 7, insert) show a single moving boundary equivalent to a single species and characteristic of a monodisperse system. In a more detailed analysis, the sedimentation coefficient distribution plot (Fig. 7) also indicates a single species, analogous to a gel filtration peak 26. From fitting the data, the diffusion coefficient can be obtained and, hence, a molecular weight estimate made ~63 kDa (Fig. 7). This result is consistent with a complex of Rev and Fab in an equimolar ratio with a predicted mass of 60.81 kDa.
A more robust approach to mass and constituent stoichiometry determination is sedimentation equilibrium analysis. Protein gradients of Fab/Rev were analyzed assuming the system was ideal, for example, no reversible association, with a mass determination of 68 kDa (Fig. 4a). Small amounts of aggregate, evidenced as systematic error in residuals at the bottom of cell, accumulate during the course of the centrifugation. We have not examined in detail variations in buffer composition and pH that might stabilize the complex during analysis, although using lower temperatures had no effect. Even when the aggregated protein absorbance is truncated prior to data fitting, mass estimates are ~65–68 kDa. If the gradient is analyzed as monomer-dimer system a better fit is obtainedwith a Kd ~0.1 mM, indicating a weak dimerization potential probably mediated by Rev (Fig. 4a). Native mass spectrometry also detects a Fab/Rev complex as the main component (with lesser amounts of oligomers of this 1:1 complex) but where none of the complexes contained dimeric Rev (Uetrecht, C. and Heck, A., personal communication). The mass and stoichiometry determinations with the single chain antibody complex are clearer, the scFv/Rev having a determined mass of 40.4 kDa (Fig. 4b), which is close to that predicted for an equimolar complex (39.34 kDa).
Based on the physiochemical analysis, Fab/Rev appears homogeneous when freshly prepared, with the caveat that this assessment by analytical centrifugation was made using protein concentrations less than 1 mg/mL. In order to screen for homogeneity at higher protein concentrations, dynamic light scattering was used (Fig. 4d). This technique, which measures the translational diffusion coefficient, is widely used to assess the suitability of samples for crystallization 27. The Fab/Rev complex at concentrations between 1–9 mg/mL gave a main peak (~97% total protein) with mass estimations close to those obtained from centrifugation. (The mass determination by the DLS plate reader is not as rigorous as that by sedimentation equilibrium but in this case was consistent). Aggregates did accumulate slowly when the complex was incubated at room temperature for up to 24 hours but these were easily removed by filtration or low speed centrifugation.
Fab/Rev was crystallized (see Material and Methods) forming long rods (Fig. 8). The crystals were suitable for X-ray structure determination and the diffraction pattern (Fig. 9) indicated resolution to beyond 3.3 Å. The data were indexed in a primitive triclinic crystal system, with unit-cell parameters a = 87.7 Å, b = 87.6 Å, c = 177.4 Å, α = 95.3°, β = 94.9°, γ =104.3°. The structural determination of the Fab/Rev complex will be described elsewhere (Dimattia et al., manuscript in submission).
The composition of the epitope was derived from the structure determination of the Fab/Rev complex at 3.2 Å resolution (Dimattia et al., manuscript in submission). Briefly, the epitope is conformational and located in the N-terminal region of Rev. Specific interactions with the Fab involve 16 residues encompassing amino acids within the N-terminal 63 residues that contact either one or both domains (Vκ and VH) of the Fab (Fig. 10; Table 1). The paratope (Fig. 10) reveals a Fab/Rev interface of ~720Å2. This large binding footprint is certainly consistent with high affinity binding and stabilization of the Fab/Rev 1:1 molar complex.
The selection of Fab SJS-R1 is remarkable in that, although we have so far only obtained this single clone, it has such advantageous properties. The antibody binds to monomeric Rev with very high affinity producing a stable immune complex. As we have previously mentioned, at concentrations >80 μg/mL Rev rapidly polymerizes to form filaments. The Rev used for the rabbit immunizations was below this critical concentration and was therefore probably a monomer-dimer mixture. For selection of the chimeric rabbit/human Fab library that was generated from Rev-immunized rabbits, solid phase Rev immobilized via a C-terminal biotinylated Avitag was used. This modified Rev does not form filaments and although confirmation of its association state was not done, the immobilized protein at high dilution was likely to be either monomeric or dimeric and binding to these non-polymeric species was selected for. In future selections from this library, we will use Rev and mutant variants in other physical states in order to pan for clones binding to other epitopes. The high affinity of Fab SJS-R1, estimated to be ~40 pM, is primarily due to an exceptionally low dissociation rate (Fig. 5). Because our panning protocol 11 enriches clones with low dissociation rates it is conceivable that the Fab outcompeted other anti-Rev Fab early in the selection process. Mutant variants of Rev without the epitope of Fab SJS-R1 (Table 1) or epitope masking 28 with purified Fab will be used to identify additional chimeric rabbit/human Fab clones.
Antibody fragment-mediated crystallization with high affinity reagents directed at conformation-sensitive epitopes is commonly used to improve crystallization by reducing protein flexibility and providing different surface contacts. This approach has been particularly important for membrane proteins 29 but is also potentially useful for protein systems which exhibit physical heterogeneity due to self-association. Such systems may include proteins, as in the present study, that polymerize into filaments of indefinite length. Crystallization conditions may increase the propensity for self-association and antibody fragments which bind to an epitope located at or near protein-protein interaction interfaces may form stable, crystallizable complexes. The rapid depolymerization of Rev by the Fab or scFv clearly indicates binding to N-terminal protein oligomerization sites (Fig. 6) and indicates high affinity interactions. The Fab/Rev complex is monodisperse (Fig. 7) but during protracted sedimentation (>16 hours), some dimerization can occur. For this system, velocity sedimentation analysis (2 hours) appears to give a more reliable mass estimate. Nevertheless, the complex readily formed crystals (Fig. 8) which diffract to ~3 Å (Fig. 9) and have been suitable for structural determination which will be reported elsewhere (Dimattia et al., manuscript in submission).
Although the primary motivation of the present study was the production of anti-Rev antibody fragments for structural studies, the high affinity and site of binding of Fab and scFv SJS-R1 predict that it may have anti-HIV-1 therapeutic potential. Although high affinity alone cannot be predictive of efficacy we suggest that binding to the Rev epitope, which so effectively blocks Rev oligomerization, is significant. The function of Rev appears to depend on its ability to assemble as a multiprotein complex on the viral RNA targeting sequence RRE (see for example, Jain and Belasco 30) Recent work suggests that following Rev binding to the RRE, it assembles into a tetramer and probably higher order complexes by oligomerization one Rev molecule at a time 31. Direct visualization of the Rev-RRE by atomic force microscopy indicates a complex containing up to 13 Rev molecules 32. The binding affinities of the initial Rev–RRE interaction are 0.26 nM with subsequent bindings to tetramer in the range of 0.79-0.48 nM 31. The Rev-Rev interactions on the RRE complex appear to be of much higher affinity than the simple self-association of Rev of 10 μM per monomer 25. From this work, the affinity of both the initial Rev-RNA binding and the subsequent Rev oligomerization are weaker than the sub-nanomolar affinity of the Fab-Rev interaction (Fig. 5). The anti-HIV-1 potential of Fab and scFv SJS-R1 is, thus, based on its blocking the protein-protein interactions that are essential for Rev action. This could be achieved through anti-Rev intracellular immunization33 with scFv SJS-R1 targeted to the cytoplasm or nucleus. In a previous study, two murine-based anti-Rev scFv, which were targeted to the cytoplasm, inhibited HIV-1 replication in human T-cells and peripheral blood mononuclear cells 34. These antibodies had relatively low affinities (0.1-0.01 μM) and bound to epitopes in the C-terminal region of Rev (residues 69–94). Interestingly, Fab SJS-R1 binds to the N-terminal domain of Rev and the C-terminal domain (residues 69–116) is not observed in the Fab/Rev crystal structure being either non-structured or highly mobile (Dimattia et al., manuscript in submission). Effective anti-HIV-1 agents based on anti-Rev intracellular antibodies may not be limited to any particular epitope but very high affinity binding would seem to be a prerequisite attribute. The anti-HIV-1 therapeutic potential of Fab SJS-R1 is not restricted to intracellular antibodies. The crystal structure of epitope and paratope of the Fab/Rev complex facilitates the computational modeling of lead peptides, peptidomimetics, or other small synthetic molecules that subsequently can be optimized by high-throughput screening of corresponding chemical libraries35. A small synthetic molecule that interferes with Rev action by mimicking Fab SJS-R1 would provide an anti-HIV-1 agent with potential oral bioavailability.
For the first time a Fab with rabbit variable domains and human constant domains derived by phage display has been produced against an HIV-1 protein. The antibody binds with very high affinity to a unique conformational epitope located in the N-terminal half of HIV-1 Rev. Both the Fab and its scFv derivative potently depolymerize Rev filaments indicating that their binding blocks a protein-protein interaction interface. The Fab/Rev complex is stable and readily forms crystals suitable for structural determination. This is also the first example of a chimeric/rabbit human Fab employed as a crystallization chaperone, promising broad utility for such agents in structural studies using X-ray diffraction. Although the anti-HIV-1 activities of the Fab and scFv have not yet been tested, based on their binding properties and the central role that Rev plays in viral infection, they also have therapeutic potential.
Rev (clone BH10) was expressed in E. coli and purified as previously described 22. Protein in 4 M urea was diluted with 6 M urea to a concentration of 69 μg/mL (i.e. below the polymerization concentration of 80 μg/mL) and then folded by dialysis against 50 mM sodium phosphate (pH 7), 150 mM sodium chloride, 600 mM ammonium sulfate, 1 mM EDTA, 1 mM DTT at 4 °C. The Rev was then dialyzed extensively against 50 mM sodium phosphate, pH 7.0, 150 mM sodium chloride. The material was sterile-filtered (0.2 μm) and then snap frozen in liquid nitrogen in aliquots at 66 μg/mL. A C-terminally truncated form of the protein, Rev ΔC60–116, was also expressed and purified in a similar manner.
Rev with a 14-residue biotin ligase substrate domain (Avitag) appended to its C-terminus 36 was expressed in E. coli and purified by ion-exchange and gel filtration chromatographies in buffers supplemented with 2 M urea. Biotinylation with biotin ligase (Avidity, LLC) was done according to the manufacturer’s protocol. Following the reaction, the protein was gel filtrated on Superdex S200 using 20 mM Tris (pH 7.4) containing 2 M urea. The integrities of the Rev-AviTag and biotinylated proteins were confirmed by mass spectrometry.
All immunization protocols were reviewed and approved by the Animal Care and Use Committees of the NIAID (ASP LI-6) and Spring Valley Laboratories (Sykesville, MD) where the animals were housed and injected. Two rabbits (1QQ174-1 and 1QQ82-1) homozygous for immunoglobulin allotypes VHa1 and Cκb9 were bled for serum prior to immunization and seven days after each immunization and each rabbit was immunized with 0.5 mL of 66 μg/mL Rev that was stored frozen as 1.2 mL aliquots for injection of two rabbits. Precautions were taken to not warm the protein when it was mixed with an equal volume of Ribi adjuvant and the mixture was placed on ice. One mL was injected per rabbit. Boosts were then given at 3-week intervals using the same amounts of antigen and Ribi adjuvant. After four boosts, when serum titers had stabilized in both rabbits, a final boost was given and 5 or 6 days later the rabbits (1QQ174-1; 1QQ82-1) were euthanized and bone marrow and spleens collected and immediately stored in TRIzol (Invitrogen).
Antibody titers were monitored by dot blot. Rev was immobilized directly on a 0.45 μm PVDF membrane at a density of 500 ng per dot. Sera from both animals were diluted serially (10-folds) and antibody levels were determined with an anti-rabbit antibody kit (WesternBreeze, Invitrogen).
Spleen and bone marrow from both rabbits were processed for total RNA preparation and RT-PCR amplification of rabbit Vκ, Vλ, and VH encoding sequences using established primer combinations and protocols 11. For each rabbit, rabbit VL/human Cκ/rabbit VH segments were assembled in one PCR amplification step, digested with SfiI, and cloned into phage display vector pC3C 9. Electrotransformation of E. coli strain XL-1 Blue (Stratagene) yielded approximately 2 × 108 independent transformants. Based on established protocols 11, the library was selected against biotinylated HIV-1 Rev immobilized on streptavidin-coated microtiter plates (Sigma) at 100 ng/well. After four rounds of panning, 11 out of 12 clones that were analyzed by ELISA 11 revealed strong binding to biotinylated Rev. One clone was negative. AluI restriction analysis showed that all 11 anti-Rev ELISA-positive clones had the same restriction pattern. The negative clone had a different pattern. Alu1 restriction analysis was done on 19 additional clones. Thirteen had the same fingerprint as the 11 original positive clones. The six clones with a different fingerprint were all negative. A fifth round of panning did not yield additional positive clones. Six positive clones (three from the fourth and three from the fifth round of panning) were sequenced as described 11. The sequences were all identical. Surprisingly, all clones had a stop codon (TAG) at the position of the first codon of the VH-CH1 chain, which immediately follows the pelB signal sequence in pC3C (as the Fab was obviously being expressed, the host, E. coli strain XL1 Blue, must be suppressing this amber codon through its supE44 genotype).
In order to maximize the expression of the selected anti-Rev Fab, the stop codon was changed, using PCR and the appropriate primers, from TAG (stop) to CAG (glutamine), the customary first codon of the VH variable domain. A modified ompA-Vκ-Cκ-pelB-VH-CH1-polyHis cassette (without HA tag) was transferred from pC3C into E. coli expression vector pET11a (Novagen) between the NdeI–BamHI restriction sites. The anti-Rev Fab Vκ and VH sequences were also cloned into pET11a such that scFv versions of them would be expressed joined by the 18-residue linker GGSSRSSSSGGGGSGGGG, i.e. ompA-Vκ-linker-VH-polyHis. Similar to anti-Rev Fab, these scFv also had a C-terminal polyHis to facilitate purification. The expression plasmids for Fab or scFv production were transfected into E. coli strain BL21CodonPlusRIL (Stratagene) and the resulting transfectants fermented in a Biostat B 2-L bench-top fermentor (Braun Biotech) using a glycerol carbon source. Cells were grown at 37 °C and induced with IPTG with a typical cell yield of 50–60 g wet weight/L. The fermentation broth was frozen at −80 °C, thawed, briefly sonicated and then clarified by centrifugation in a JA-10 rotor at 14,000×g for 30 minutes. The supernatant was added to ~100 mL of Chelating Sepharose Big Beads resin (GE Healthcare Life Sciences) charged with NiSO4 and equilibrated with 20 mM sodium phosphate buffer (pH 7.2). The resin was gently mixed for ~30 minutes, filtered through a Buchner funnel, washed with column buffer containing 1 M urea and then packed into a Ni-Sepharose Fast Flow column (3.5 cm diameter × 10 cm length). After washing the column, the protein was eluted using a 10–500 mM imidazole gradient. Pooled fractions were further purified by gel filtration on a Superdex S200 column (2.6 cm diameter × 60 cm length) equilibrated in 20 mM sodium phosphate (pH 7.2), 30 mM sodium chloride and 1 M urea. Column fractions were assayed by reducing (plus DTT) and non-reducing (minus DTT) SDS-PAGE. Note: the anti-Rev Fab is formally identified as Fab SJS-R1 (for simplicity hereafter ‘Fab’), and the derived anti-Rev scFv as scFv SJS-R1 (hereafter ‘scFv’).
The Fab/Rev complex was produced by combining Fab and a 5-fold molar excess of Rev in 1 M Urea. The urea was necessary to prevent the precipitation of Rev. The mixture was dialyzed against PBS, any precipitated protein (usually excess Rev) was removed by centrifugation, and then applied to a Ni-Sepharose Fast Flow column equilibrated in dialysis buffer. The Fab/Rev complex was eluted with imidazole, dialyzed against 20 mM HEPES (pH 8.0) and concentrated to 8–9 mg/mL using an Amicon Ultra-15 10K NMWL centrifugal filter (Millipore). The scFv/Rev complex was produced and purified in a similar manner except that the final dialysis was performed against PBS.
Gel electrophoresis was performed using NuPAGE Novex 4–12% polyacrylamide gels (Invitrogen). Gels were scanned and TIF image files analyzed using Kodak Molecular Imaging software V4.0 (Carestream Health, Rochester, NY).
A Beckman Optima XL-I analytical ultracentrifuge, absorption optics, an An-60 Ti rotor and standard double-sector centerpiece cells were used. Equilibrium measurements were taken at 20 °C and concentration profiles recorded after 16–20 hours at either 12,000–14,500 (Fab/Rev) or 18,000–21,500 (scFv/Rev or scFv) rpm. Baselines were established by over-speeding at 45,000 rpm for 3 hours. Data (the average of five scans collected using a radial step size of 0.001 cm) were analyzed using the standard Optima XL-I data analysis software. Protein partial specific volumes were calculated from the amino acid compositions 37 and solvent densities were estimated using the public domain software program SEDNTERP (http://www.rasmb.bbri.org/). Sedimentation velocity measurements at 20 °C were taken at 45,000 rpm for 3 hours with data collection at 5–10 minute intervals. Data (radial step size 0.003 cm) was analyzed using the program DCDT+ version 2.2.1 26.
Fab and Rev filaments prepared as described previously 21 in cold 50 mM HEPES, 150 mM sodium chloride, 25 mM sodium citrate (pH 7.0) were mixed in equimolar ratio. The mixtures were incubated overnight at 4 °C (conditions under which Rev filament controls remained polymerized). Specimens were applied to 400-mesh carbon-coated copper grids made hydrophilic by air glow discharge and negatively stained with 1% uranyl acetate. Images were recorded with a Philips CM120 electron microscope by CCD at 45,000× magnification.
The kinetics of the Fab binding to immobilized Rev was studied by surface plasmon resonance using a Biacore X (GE Healthcare). Rev was immobilized on the surface of a CM5 sensor chip at 1500 RU (1000 RU ~1 ng bound protein/mm2,38) by EDC-NHS coupling chemistry according to the manufacturer’s protocol. HBS-EP buffer (GE Healthcare) was used as the running buffer and the analyte Fab (90 μl) in running buffer was passed over immobilized Rev at a flow rate of 30 μL/min. Analyte was injected at concentrations between 0.5–500 nM. For all assays, at least two replicate injections at the same concentrations were employed to calculate the kinetic data. Analysis was done using BIAevaluation software, version 3.1. The interaction was globally fitted to a 1:1 interaction Langmuir model with the association and dissociation phases of the interaction fitted simultaneously. The goodness of fit between the fitted curves and the experimental curves was assessed by visual comparison. The rate constants (ka and kd) and the equilibrium constant (Kd) were calculated from the best-fit kinetic parameters (BIAevaluation handbook BR 1002-29, Biacore AB).
An HP1100 LC-MS electrospray mass spectrometer (Agilent) coupled to a Zorbax C-3, 2.1 mm diameter × 15 cm length column was used. Protein (0.1–0.5 mg/mL) was diluted between 1:20 to 1:50 with either H2O or 1% formic acid and samples (5 μL) were applied to the column equilibrated in 0.1% formic acid, 5% acetonitrile. The column was washed for 15 minutes with column solvent and then a 35-minute gradient of 5% to 100% acetonitrile in 0.1% formic acid was applied. The flow rate was 0.2 mL/min and gradient eluate was analyzed by mass spectrometry.
Proteins (1–9 mg/mL) were either centrifuged or filtered prior to analysis and then 10 μL samples were manually pipetted into a 384 well plate (Greiner Bio-One). Measurements were made at 20 °C using a DynaPro Plate Reader Plus and the data analyzed using Dynamics V6 software (Wyatt Technology Corporation).
Purified Fab/Rev in 20 mM HEPES (pH 8.0) at a concentration of 12.9 mg/mL was used. Crystallization trials were at 21°C in hanging drops containing 200 nL protein and 200 nL precipitant solution equilibrated against 50 μL reservoirs in 96-well plates. Over 1600 reservoir conditions were assayed and only one resulted in crystalline nucleations. Microcrystals of Fab/Rev initially grew in 12% PEG 6000, 100 mM di-ammonium phosphate (DAP), and 100 mM Tris-HCl (pH 8.5) and were optimized with a screen varying concentration and pH of the initial reservoir components as well as screening of the initial crystallization condition against the 96-well Hampton Research (Laguna Niguel, CA, USA) additive screen kit. Optimized crystals were grown with Fab/Rev sample at 7.1 mg/mL in 20 mM HEPES, pH 8.0 in 1–2 weeks with 50 mM spermidine added to the drop and in 9–14% PEG 600, 100–200 mM DAP, 100 mM Tris (pH 8.5) and were cryoprotected by a quick pass through reservoir solution supplemented with 25% (v/v) ethylene glycol before flash cryo-cooling in a cold (100 K) stream of nitrogen gas. Diffraction data were recorded from a single Fab/Rev crystal at λ = 0.97960 Å at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The crystal used for diffraction was grown in 9.5% PEG 6000, 150 mM DAP, 100 mM Tris-HCl (pH 8.5), and 50 mM spermidine. Diffraction data were integrated and scaled using HKL2000 software (HKL Research, Inc.).
The sequences for the Fab SJS-R1 L- and H-chains have been deposited in GenBank with accession numbers GU223201 and GU223202, respectively.
We thank Dr. J. Vethanayagam (NIAMS), C. Alexander, and B. Newman (NIAID), and the staff at the Diamond Light Source, for their expert technical assistance. We also thank Drs N. Noinaj and S. Buchanan (NIDDK) for advice and provisionof resources during the initial crystallization and crystal evaluation experiments, andDr. E. Kandiah (NIAMS) for insightful discussions. This work was supported by the Intramural Research Programs of NIAMS, NCI, and NIAID, and by the Intramural Targeted Antiviral Program of the National Institutes of Health, as well as by the NIH-Oxford Scholars Program. Support was also provided by the UK Medical Research council (DIS) and by SPINE2COMPLEXES (LSHGCT-2006-031220) (JMG).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.