The α2β1 antagonist rhodocetin from Calloselasma rhodostoma is a heterotetrameric CLRP (C-type lectin-related protein) consisting of four distinct chains, α, β, γ and δ. Via their characteristic domain-swapping loops, the individual chains form two subunits, αβ and γδ. To distinguish the four chains which share similar molecular masses and high sequence homologies, we generated 11 mAbs (monoclonal antibodies) with different epitope specificities. Four groups of distinct mAbs were generated: the first targeted the rhodocetin β chain, the second group bound to the αβ subunit mostly in a conformation-dependent manner, the third group recognized the γδ subunit only when separated from the αβ subunit, whereas a fourth group interacted with the γδ subunit both in the heterotetrameric molecule and complexed with the integrin α2 A-domain. Using the specific mAbs, we have shown that the rhodocetin heterotetramer dissociates into the αβ and γδ subunit upon binding to the integrin α2 A-domain at both the molecular and cellular levels. After dissociation, the γδ subunit firmly interacts with the α2β1 integrin, thereby blocking it, whereas the rhodocetin αβ subunit is released from the complex. The small molecular interface between the αβ and γδ subunits within rhodocetin is mostly mediated by charged residues, which causes the two dissociated subunits to have hydrophilic surfaces.
- C-type lectin-like protein (CLRP)
- α2β1 integrin
- quaternary structure
Integrin α2β1 is a collagen receptor that is widely expressed on different cell types, such as fibroblasts, epidermal and endothelial cells, and platelets . As a member of the integrin family , it consists of two genetically non-related subunits, α2 and β1, the latter one of which is shared with most extracellular-matrix-binding integrins. The extracellular domains of both integrin subunits shape the head domain which harbours the ligand-interaction site. Within their α subunits, the collagen-binding integrins contain an additional A-domain, which is also located within the head region. The integrin α2 A-domain is crucially involved in collagen binding [3,4]. Upon collagen binding, it undergoes a substantial conformational change, which is conveyed through the entire molecule to the cytoplasmic tails. Thus environmental cues from the extracellular matrix are transduced into the cells in the outside–in signalling process of integrins . Consequently, several α2β1 integrin-related functions are triggered upon cell attachment to collagen, such as force transmission, and migration and expression of both collagens and collagen-degrading proteases [6–10]. Moreover, α2β1 integrin is the major collagen-binding integrin on platelets, although GP (glycoprotein) VI on platelets also binds collagen . The role of α2β1 integrin compared with GPVI in collagen-induced platelet activation and aggregation is therefore still controversial [12–14]. Owing to the redundancy of collagen receptors  and compensatory mechanisms by other collagen-binding integrins [15,16], the genetic ablation of the integrin α2 subunit in mice results in a subtle phenotype, which includes reduced collagen-induced platelet aggregation [17–19]. In addition to thrombosis, integrin α2β1 plays an important role in several physiological and pathological processes, such as wound healing and fibrosis . Hence, a search for α2β1 integrin-targeting pharmaceuticals has been initiated [20,21].
The importance of α2β1 integrin in collagen-induced platelet aggregation probably contributed to its evolution into a natural target for haemorrhagic venoms, such as from snakes. A common recognition motif that is mimicked by disintegrins in numerous snake venoms is the peptide sequence RGD (Arg-Gly-Asp) . However, α2β1 integrin binds to collagen in an RGD-independent manner and requires six amino acids, Gly-Phe-Pro-Gly-Glu-Arg, which must be presented in a collagenous triple helix [8,23,24]. So far, three naturally occuring inhibitors of α2β1 integrin have been identified from snake venoms: rhodocetin from Calloselasma rhodostoma , EMS16 from Echis multisquamatus  and VP12 from Vipera palestina . All three toxins lack a collagenous triple-helical domain. They belong to the family of CLRPs (C-type lectin-related proteins), for which the name ‘snake venom C-type lectins’ or ‘snaclecs’ has been coined [28,29].
Rhodocetin was originally described as a heterodimer consisting of two chains, α and β [25,30], that associate firmly, but non-covalently, via a loop-exchange motif typical for CLRPs from snake venoms [31,32]. Owing to similar molecular masses and high homology of the rhodocetin chains, it was subsequently realized that rhodocetin does not consist of two but of four distinct chains, α, β, γ and δ, which exist in two heterodimeric subunits, αβ and γδ, to form a unique quaternary structure . In previous studies, we have also previously demonstrated that rhodocetin strongly interacts with the integrin α2 A-domain, thereby blocking collagen binding to α2β1 integrin, and that it antagonizes α2β1 integrin-mediated cellular functions in different cell systems [34,35]. In the present study, we have developed mAbs (monoclonal antibodies) against rhodocetin and its subunits, to precisely define the molecular mechanism of action of rhodocetin. Being specifically directed against the individual subunits, these mAbs not only confirm the heterotetrameric nature of rhodocetin, but are also ideal tools to monitor the fate of the individual rhodocetin subunits during integrin binding. We demonstrate that the heterotetrameric rhodocetin dissociates upon binding to α2β1 integrin, which is likely to be accompanied by conformational changes within the rhodocetin subunits. The rhodocetin γδ subunit firmly binds to the α2β1 integrin, whereas the rhodocetin αβ subunit is released from the complex and acts as an independent entity. Moreover, rhodocetin dissociates not only in protein-interaction assays, but also under physiological conditions after it has bound to the membrane-anchored α2β1 integrin on cells.
Tetrameric rhodocetin was purified and its subunits were separated as described previously . Recombinant integrin α2 A-domain  was allowed to form a complex with the rhodocetin γδ subunit as described previously . A polyclonal antiserum against tetrameric rhodocetin was raised in rabbits according to standard protocols. Immunization was performed in compliance with German regulations on animal welfare. It recognized all four subunits of rhodocetin in ELISA and Western blot, both with and without reduction of the antigen.
The cDNA, which encodes the fusion protein of His-tagged integrin α2 A-domain with EYFP [enhanced YFP (yellow fluorescent protein)], was generated in two PCR steps. The part of the cDNA encoding the N-terminal oligoHis-α2 A-domain was amplified from pET15b-α2A  using primers containing an NcoI and an AgeI restriction site at the 5′- and 3′-end respectively. Likewise, the EYFP sequence was amplified from pIRES-EYFP (Clontech Laboratories) using primers containing AgeI and NdeI sites at the 5′- and 3′-end respectively. Subsequently, both amplicons were ligated with their AgeI sites into pET15b which was previously digested with NcoI and NdeI. Finally, the construct pET15b-α2A-YFP was verified by sequencing. Escherichia coli BL21(DE3) bacteria transformed with this plasmid by electroporation produced the recombinant fusion protein called α2A–YFP. It was purified by metal-ion-affinity chromatography using a Ni-NTA (Ni2+-nitrilotriacetate) column (GE Healthcare) and was detected by its fluorescence at 527 nm after excitation at 514 nm.
Immunization of mice and rats, and the generation of mAbs
For immunizations, two Balb/c mice and two Sprague–Dawley rats were used. Tetrameric rhodocetin was subcutaneously injected into a mouse (first fusion), the isolated rhodocetin γδ subunit was injected into another mouse and a rat (second and third fusions respectively), and the complex of rhodocetin γδ with the integrin α2 A-domain was injected into another rat (fourth fusion). The antigens were applied together with complete and incomplete Freund's adjuvants in the first and the subsequent immunization injections respectively. The splenic lymphocytes were fused with a 3-fold surplus of SP2/0 myeloma as described previously . Conditioned media from hybridomas were tested for antibody production by ELISA and isolated into single cell-derived clones in two subsequent rounds of subcloning.
ELISA for hybridoma screening and for titration assays
Not only the antigens used for immunization (tetrameric rhodocetin, rhodocetin γδ and rhodocetin γδ–integrin α2 A-domain complex), but also the other isolated rhodocetin subunits and the integrin α2 A-domain were coated on to microtitre plates at 2 μg/ml in TBS [Tris-buffered saline (50mM Tris/HCl and 150mM NaCl)] (pH 7.4) overnight at 4°C. After washing twice with TBS and blocking with 1% BSA solution in TBS (pH 7.4) containing 1 mM MgCl2, the wells were incubated with hybridoma-conditioned media (for hybridoma screening) or purified mAbs (as positive controls at 1 μg/ml or for titration experiments) in 1% BSA in TBS (pH 7.4)/MgCl2 for 90 min at room temperature (20°C). Bound mAbs were detected with secondary alkaline phosphatase-conjugated antibodies against mouse or rat antibodies diluted to 1:1000 in the same buffer for 90 min. Binding was quantified by conversion of p-nitrophenylphosphate (Sigma). After stopping the conversion with 1.5 M NaOH solution, absorbance was measured at 405 nm in an ELISA reader (Synergy HT, BioTek). Non-specific binding to BSA-coated wells was subtracted from the measured signals. The ELISA was performed four times.
Western blot analysis
Rhodocetin (2 μg) was electrophoretically separated by SDS/PAGE (12–18% gel) with and without prior reduction and was blotted on to nitrocellulose membrane. After blocking with TBST [0.05% Tween 20 in TBS (pH 7.4)] supplemented with 2% BSA, the membranes was incubated with 2 μg/ml mAb and rabbit antiserum against rhodocetin (diluted 1:2000), washed with TBST, and incubated with Alexa Fluor® 488-labelled secondary antibodies against murine, rat or rabbit antibodies (Invitrogen). After two washes with TBST and TBS, the stained bands were detected with an Ettan DIGE fluorescence reader (GE Healthcare).
Antibody interference assay
The detecting mAbs were biotinylated with a 25–500 molar excess of EZ-link sulfo-NHS-LC-biotin (Pierce, Thermo Scientific) in 1 ml of PBS for 1 h at room temperature. Biotinylation was stopped by adding 40 mM Tris/HCl (pH 7.4). Excess biotinylation reagent was removed by dialysis. For the mAb-interference assay, 2 μg/ml rhodocetin tetramer or γδ subunit were coated on to a microtitre plate in TBS (pH 7.4) overnight at 4°C. After washing and blocking with 1% BSA in TBS (pH 7.4) and 1 mM MgCl2, wells were incubated with each biotinylated detecting mAb at 1 μg/ml in the same buffer in the absence or presence of non-biotinylated interfering mAbs. Bound detecting mAbs were quantified in an ELISA with alkaline-phosphatase-conjugated ExtrAvidin (Sigma–Aldrich), diluted 1:1000 in 1% BSA in TBS (pH 7.4) with 150 mM NaCl and 1 mM MgCl2. Conversion of phosphatase substrate was colorimetrically detected in a microtitre plate reader. The signals were corrected for background values, measured in BSA-coated wells, and normalized to the non-inhibited controls, measured in the absence of any interfering mAb. The experiment was carried out in duplicate.
mAb inhibition test
To test the potential of mAbs to inhibit the binding of rhodocetin to α2β1 integrin, the integrin α2 A-domain was immobilized to the microtitre plate at 10 μg/ml. After washing and blocking with 1% BSA, rhodocetin was added at increasing concentrations in both the absence and presence of the murine or rat mAb, in at least a 2-fold molar excess. Bound rhodocetin was fixed with 2.5% glutaraldehyde in 25 mM Hepes (pH 7.4), 150 mM NaCl and 2 mM MgCl2 for 10 min and detected in an ELISA using the rabbit antiserum against rhodocetin.
Immunoprecipitation of rhodocetin
Crude C. rhodostoma venom (20 mg) (Sigma–Aldrich) was dissolved in PBSE [20 mM sodium phosphate (pH 7.4), 150 mM NaCl and 5 mM EDTA), supplemented with leupeptin, pepstatin, aprotinin (each at 1 μg/ml), phenanthroline and PMSF (each at 1 mM). Then, 100 μg of each mAb (including species-matched control antibodies) were incubated with 30 μl of this crude venom solution in 1% BSA in PBS (pH 7.4) at a final volume of 300 μl for 2 h at room temperature. Then, 50 μl of equilibrated Protein G beads (Pierce, Thermo Scientific) were added and incubation was continued for another 90 min. Beads were washed three times with PBS, suspended and boiled in SDS/PAGE sample buffer without 2-mercaptoethanol, electrophoretically separated by SDS, blotted on to a nitrocellulose membrane, stained with the rabbit antiserum against rhodocetin and fluorescently detected with an Alexa Fluor® 488-labelled secondary antibody as described above.
Analytical size-exclusion chromatography of rhodocetin
Rhodocetin (50 μg), either alone or in an equimolar ratio with YFP-tagged integrin α2 A-domain (40 μg), was cross-linked with 1 mM BS3 (Pierce, Thermo Scientific) for 1 h at 26°C in PBS (pH 7.4), supplemented with 1 mM MgCl2. The cross-linking reaction was stopped with 20 mM Tris/HCl (pH 7.4) and proteins were separated on a TSK 2000 GWXL column (TosoHass) in PBS at 0.5 ml/min. Elution of proteins was monitored by measuring absorbance at 219 nm and fluorescence at 527nm (excitation at 514 nm). The eluate fractions were further analysed in the rhodocetin subunit sandwich ELISA.
Rhodocetin subunit sandwich ELISA
The mAbs VIIG2 and IC3 were coated at 5 μg/ml in TBS at 4°C overnight. The microtitre plate was subsequently washed with TBS (pH 7.4), and blocked with 1% BSA in TBS (pH 7.4) and 1 mM MgCl2, before the gel-filtration fractions were applied in a 1:100 dilution for 90 min. Captured rhodocetin was quantified with the rabbit antiserum against rhodocetin and an alkaline-phosphate-conjugated secondary antibody, both diluted 1:2000, as described above.
Kinetic SPR (surface plasmon resonance) measurements
The oligoHis-free integrin α2 A-domain was covalently immobilized to a CM5 chip (GE Healthcare) according to the manufacturer's instructions. Using BiaCore X (GE Healthcare), sensograms were recorded while solutions of different concentrations of rhodocetin tetramer or rhodocetin γδ subunit in 50 mM Hepes/NaOH (pH 7.5), 150 mM NaCl and 1 mM MgCl2, were flown over the integrin α2 A-domain-coated CM5 chip. The firmly attached rhodocetin molecules were washed off with 50 mM DTT (dithiothreitol) in 50 mM Tris/HCl (pH 9.5) and 300 mM NaCl after every binding cycle. The rhodocetin γδ subunit was dissolved as stock solution in the acetonitrile-containing buffer of the reversed-phase purification. Hence, correspondingly diluted buffer controls were performed and the respective sensograms were substracted. The sensograms were assesses with the BIAevaluation v3.1 software (GE Healthcare).
HT1080 cells (1×105) were incubated with 2 μg/ml rhodocetin in PBS (pH 7.4), containing 2% BSA and 1% horse serum. After two washes with ice-cold PBS, the cells were incubated with mAbs directed against rhodocetin or species-matched control antibodies, each at 2 μg/ml. Bound murine and rat mAbs were detected with pycoerythrin-conjugated and Alexa Fluor® 568-labelled secondary antibodies against anti-mouse IgG and anti-rat IgG respectively in a CyFlowSL (Partec) with 1×104 events counted per sample.
Generation of mAbs against rhodocetin and the integrin α2 A-domain–rhodocetin γδ complex
Stable hybridomas (11) were generated that produced mAbs against rhodocetin. These included nine murine mAbs (VIIG2, VD10, IIIB6, VIIF4, VIIF9, IXH7, VIIIB9, IIIG5 and IIC9), all of which were IgG1, and two rat mAbs, IC3 and ID10, that were IgG2a and IgG1 respectively (Supplementary Figure S1 at http://www.BiochemJ.org/bj/440/bj4400001add.htm). Three rat hybridomas were obtained which produced mAbs (VIIIA8, XD8 and VIH7) against the integrin α2 A-domain. The IgG2 VIIIA8, and the two IgMs XD8 and VIH7, were difficult to purify without loss of activity. They are therefore not used in the present study.
Specificity of the newly generated mAbs against rhodocetin
The specificities of the murine and rat mAbs were determined in an ELISA, in which the tetrameric rhodocetin, its individual α and β chains, its subunits αβ and γδ, its complex with the integrin α2 A-domain or the integrin α2 A-domain alone were used as substrates (Figure 1). The nine murine and two rat mAbs recognized the individual rhodocetin subunits differently and were sorted into four subgroups. The first subgroup comprised the murine mAbs VIIG2 and VD10 which recognized the rhodocetin β subunit, irrespective of whether it was coated alone or as the rhodocetin αβ dimer. By contrast, the murine mAbs IIIB6, VIIF4 and VIIF9, constituting the second subgroup of mAbs, recognized the rhodocetin αβ dimeric subunit only, but not the α or β chains alone. A similar antigen-recognition behaviour was observed for the mAbs directed against the rhodocetin γδ subunit. Although the mAbs IXH7, VIIIB9 and IIIG5 bound preferentially to the rhodocetin γδ subunit, but not the entire rhodocetin tetramer (third subgroup of mAbs), the mAbs IIC9 from mouse and ID10 from rat showed greater binding to the rhodocetin αβγδ tetramer than to the rhodocetin γδ subunit. The rat mAb IC3 recognized both the rhodocetin tetramer and rhodocetin γδ subunit equally well. The latter three mAbs, IIC9, ID10 and IC3, form the fourth subgroup of antibodies, which also detect the complex of the rhodocetin γδ subunit with the integrin α2 A-domain, without interacting with the integrin-binding partner of rhodocetin.
The mAbs VIIG2, VD10 and IIIG5 recognize a sequence epitope
To distinguish whether the mAbs recognize a conformational or sequence epitope, immunoblots were carried out in which the electrophoretically separated rhodocetin subunits, both without and with prior reduction (left- and right-hand lanes respectively in Supplementary Figure S2 at http://www.BiochemJ.org/bj/440/bj4400001add.htm) were blotted on to nitrocellulose membranes and probed with the test mAb and the polyclonal rabbit anitiserum against rhodocetin. Although the two mAbs VIIG2 and VD10 recognized both the non-reduced and reduced rhodocetin β chain, the other anti-β subunit mAb IIIB6 recognized only the non-reduced rhodocetin β chain. The mAb IIIB6 seemed to recognize a certain folding pattern within the rhodocetin β chain. This was further corroborated by the fact that IIIB6 shows a binding signal in ELISA, only if the conformation of the β chain is preserved within the αβ dimeric subunit or the entire tetrameric rhodocetin molecule (Figure 1). A similar conformational epitope restriction in Western blot analysis was seen for the mAbs VIIIB9, IC3 and ID10, all of which exclusively recognized the non-reduced rhodocetin γδ subunit. Moreover, they also recognized the disulfide-linked (γδ)2 subunit dimers, which run at an apparent molecular mass of 55 kDa on SDS/PAGE. The mAb IIIG5 was the only antibody that detected a sequence epitope within the rhodocetin γ chain, which is exposed only after reductive separation of the rhodocetin γ chains.
Affinities of mAbs towards rhodocetin and the effect of rhodocetin γδ-directed mAbs on the rhodocetin–integrin interaction
The affinities of mAbs towards their antigen were determined by titration of rhodocetin or its respective subunit in an ELISA (Figure 2A). The titration curves were approximated by a linearization algorithm developed by Heyn and Weischet . The Kd values calculated by this approximation are summarized in Table 1 and range from 0.5 nM (VD10) to almost 700 nM (IIIG5).
As the rhodocetin γδ subunit contains the binding site for the integin α2 A-domain, the γδ subunit-targeting mAbs IIC9, IC3 and ID10 were tested for their ability to inhibit the interaction of rhodocetin with the integrin-binding domain. Although the titration curves of rhodocetin to immobilized α2 A-domain were shifted to higher rhodocetin concentrations in the presence of mAbs (Figure 4B), thus indicating increased Kd values (IIC2, 4.2±0.4 nM; IC3, 6.9±0.3 nM; ID10, 5.7±0.2 nM compared with 2.7±0.1 nM in the absence of mAb), none of the three mAbs were able to abrogate the binding of rhodocetin to the α2 A-domain.
Epitope specificity of the anti-rhodocetin mAbs
To determine whether the newly generated mAbs have overlapping epitope specificities, we performed antibody-interference assays. To distinguish the detecting mAb from the interfering mAbs, we biotinylated the detecting mAbs with an amine-reactive biotinylation reagent. Biotinylated detecting mAbs against rhodocetin were allowed to bind to immobilized rhodocetin in the absence and presence of various concentrations of an interfering non-biotinylated mAb. When binding of the biotinylated mAb was compromized by the other mAb, the two mAbs were assigned to the same competition group and possess an overlapping or even identical epitope within the rhodocetin molecule. Representative interference curves for VIIG2 and VD10 (Figure 3A) and for IIC9 (Figure 3B) are shown as examples with mAbs directed against the αβ and γδ subunit respectively. Comprehensive interference tables of anti-rhodocetin αβ and anti-rhodocetin γδ mAbs (Figures 3C and 3D respectively) summarize the results. Probably owing to sterical hindrance, VIIG2 and VD10, VIIF4 and VIIIF9, IXH7 and VIIIB9, and IIC9 and ID10 excluded each other mutually from binding to rhodocetin (Figures 3C and 3D). Rather than being inhibitory, the mAbs IIIB6, VIIF4 and VIIF9 enhanced binding of the mAbs VIIG2 and VD10. Interestingly, all of the former mAbs recognized the αβ heterodimer, but not the individual α or β chain, whereas the latter recognized a sequence epitope within the rhodocetin β chain. Moreover, albeit belonging to the same antibody group, the binding intensity of IIIB6 was increased in the presence of VIIF4 and VIIF9. The enhanced binding signals suggest conformational changes within rhodocetin. Thus binding of one mAb leads to a conformational change within the rhodocetin molecule, thereby generating the conformational epitope or unmasking the epitope of the other antibody. Within the group of antibodies recognizing the rhodocetin γδ subunit, only one positive interference was observed (Figures 3B and 3D). IC3 supported the binding of the detecting IIC9 strongly, whereas a reciprocal activation was not observed. Interestingly, such a reciprocal activation did not occur between the rhodocetin αβ subunit-recognizing antibodies (Figure 3C).
Tetrameric rhodocetin dissociates into its αβ and γδ subunits upon mAb binding
To study the interaction of mAbs with rhodocetin in solution, we performed immunoprecipiation experiments (Figure 4A). All mAbs directed against the rhodocetin β chain, VIIG2, VD10, IIIB6, VIIF4 and VIIF9, precipitated the rhodocetin α and β chains. However, the precipitates contained hardly any rhodocetin γδ subunit. Complementarily, the mAbs IIC9, IC3 and ID10, directed against the rhodocetin γδ subunit, preferentially pulled down the γδ subunit, with a much smaller amount of accompanying rhodocetin αβ subunit. This observation suggested that our mAbs against rhodocetin induced a dissociation of the rhodocetin tetramer into the αβ and γδ subunits after binding their respective epitopes. Rhodocetin is a heterotetrameric protein in solution and dissociates only after being bound by the antibodies. In support of this hypothesis, the mAbs against the rhodocetin γδ subunit, IXH7, VIIIB9 and IIIG5, which only recognize their epitopes within the isolated rhodocetin γδ subunit but not in the entire tetrameric rhodocetin molecule (Figure 2), failed to precipiate any rhodocetin chain. However, when the precipitations with these mAbs were carried out in the presence of the anti-rhodocetin β-subunit mAb VIIG2, they pulled down the rhodocetin γδ subunit completely (Figure 4B). This suggests that their epitopes within the rhodocetin γδ subunit are masked by the rhodocetin αβ subunit and become accessible only after dissociation of the rhodocetin tetramer into its two heterodimeric subunits.
Dissociation of rhodocetin upon α2 A-domain binding
To determine whether the dissociation of the heterotetrameric rhodocetin into its αβ and γδ subunits is only artifically induced by the mAbs, or whether it represents a physiological process, we performed analytical size-exclusion chromatography of rhodocetin without and with its target molecule, the integrin α2 A-domain. To avoid dissociation of the rhodocetin subunits during gel filtration, we cross-linked rhodocetin with the homo-bifunctional cross-linker BS3 prior to size-exclusion chromatography. The subunits of rhodocetin in the eluate fractions were identified by sandwich ELISAs, in which the immobilized mAbs VIIG2 and IC3 specifically captured the rhodocetin αβ or γδ subunit respectively. On the gel-filtration column, rhodocetin separated into two major peaks, I and II (Figure 5A). Peak I with an apparent molecular mass of 72.3±1.4 kDa represented the heterotetrameric rhodocetin molecule with a higher molecuar mass shoulder at 108 kDa, which is likely to represent disulfide-linked rhodocetin (γδ)2 including associated αβ subunit(s). Peak II contained dissociated rhodocetin αβ and γδ subunits with an apparent molecular mass of 36.3±2.0 kDa. As analysed by sandwich ELISA and immunoblotting, the composition of rhodocetin (subunits) in peaks I and II did not change irrespective of divalent cations during the pre-column cross-linkage. A consistent shift of peak I and II by approximately 10 kDa to higher apparent molecular masses was observed when rhodocetin was cross-linked in the presence of divalent cations as compared with the divalent-cation-free cross-linking conditions. This indicated that only the conformation, but not the quaternary structure, of the rhodocetin is altered by divalent cations.
In contrast, the quaternary structure of the rhodocetin tetramer changed drastically when rhodocetin was incubated with the integrin α2 A-domain prior to cross-linkage and gel filtration (Figure 5B). To ease detection of the α2 A-domain we used a YFP-tagged α2 A-domain, which has an apparent molecular mass of 55 kDa and is prone to homodimerization (110 kDa) similar to the untagged α2 A-domain. After incubation with rhodocetin and cross-linkage, α2 A-domain was detected by fluorescence at 527 nm at higher apparent molecular masses of 165±1, 130±1 and 86.5±1.7 kDa (peaks F-I, F-II and F-III respectively) (Figure 5B). When compared with the rhodocetin-free control, the peaks of integrin-α2 A-domain (peaks F-I and F-III) were shifted by approximately 30–35 kDa to higher apparent molecular masses, indicating that the YFP-tagged integrin α2 A-domain was associated with a rhodocetin subunit. Interestingly, all three fluorescent peaks, F-I to F-III, contained rhodocetin γδ, but virtually no rhodocetin αβ subunit, as measured by sandwich ELISA (Figure 5B). The rhodocetin αβ subunit was detected at a molecular mass of 31.2±1.9 kDa, with a slightly lower apparent molecular mass than peak II in Figure 5(A) of the dissociated rhodocetin subunits. This peak did not contain any rhodocetin γδ subunit and therefore contained only the rhodocetin αβ subunit. Upon integrin binding, the heterotetrameric rhodocetin dissociated, whereby its γδ subunit formed an avid complex with the α2 A-domain and its αβ subunit was released. Hence, ligand binding to rhodocetin alters its quaternary structure.
Rhodocetin αβ subunit accelerates rhodocetin γδ binding to α2 A-domain
SPR measurement (Figure 6) demonstrated that the rhodocetin tetramer bound to immobilized integrin α2 A-domain approximately 280000 times faster than the rhodocetin γδ subunit alone, whereas the rhodocetin γδ subunit alone dissociated from its integrin target several orders of magnitudes slower than the rhodocetin tetramer (Table 2). Although the rhodocetin αβ subunit aids in forming the complex of rhodocetin teramer with α2β1 integrin, the subsequent dissociation of the rhodocetin αβ subunit renders the remaining rhodocetin γδ subunit alone with a very slow dissociation rate, thus resulting in an almost irreversible blocking complex of rhodocetin γδ with the α2β1 integrin.
Rhodocetin dissociates upon binding to cell-anchored α2β1 integrin
HT1080 fibrosarcoma cells abundantly express α2β1 integrin. Upon incubation with rhodocetin, the HT1080 cells were decorated only with the rhodocetin γδ subunit, whereas the rhodocetin αβ subunit could not be detected on the cells by flow cytometry (Figure 7). Therefore, not only the isolated α2 A-domain, but also cell-anchored α2β1 integrin induced the dissociation of rhodocetin with release of rhodocetin αβ and with firm and kinetically inert binding of rhodocetin γδ to the integrin.
As a natural product rhodocetin lacks any tag sequence, which usually facilitates the isolation and detection of recombinant proteins. So far, rhodocetin has to have been detected by its capability to bind to α2β1 integrin and to inhibit collagen-induced cell adhesion. These tests were complex and impossible to perform with the crude venom because of the potential presence of contaminating cell-toxic venom components. In addition to the α2β1 integrin inhibition tests, we had raised a polyclonal rabbit antiserum against the rhodocetin tetramer. However, owing to the high homology of the four subunits of rhodocetin , the fates of individual rhodocetin chains were impossible to analyse. The newly generated mAbs to the individual rhodocetin chains and their characterization have been instrumental in revealing the molecular mechanism of rhodocetin binding to α2β1 integrin. In the present paper, we describe 11 mAbs, nine from mouse and two from rat, specifically directed against different rhodocetin chains. According to their recognition pattern, they can be sorted into four groups. The first and second group recognize the rhodocetin β chain and the rhodocetin heterodimeric αβ subunit respectively, whereas the third and fourth group detect an epitope within the rhodocetin γδ subunit only after or irrespective of dissociation from the rhodocetin αβ subunit respectively. Owing to their specificity, they are well suited to the detection of individual rhodocetin chains. In a newly established sandwich ELISA in which the mAbs are used as capturing antibodies, very low concentrations of rhodocetin chains can be quantified in the range of 0.1 μg/ml in protein-rich samples, such as blood and urine. Hence, these mAbs might be useful in developing a fast detection assay for envenomed patients, either on the basis of the established sandwich ELISA or even in a test-stick format. Although none of the mAbs inhibited rhodocetin binding to α2β1 integrin and thus will not aid in developing an effective antivenom therapy, they may help diagnostically in identifying the envenoming species. Nevertheless, generating neutralizing mAbs against venom components with low immunogenicity might be an essential and live-saving improvement in antivenom therapy .
In the present study, the mAbs were instrumental in elucidating the changes of the quaternary structure of rhodocetin and even provided evidence for conformational changes of the rhodocetin subunits, which occur upon α2β1 integrin binding. Interestingly, every mAb of the second group markedly increased the binding of mAbs of the first group indicating a potential conformational change within the rhodocetin αβ subunit. By contrast, only one positive interference occurred within the groups of mAbs recognizing the rhodocetin γδ subunit, as the rat mAb IC3 increased the binding signal of the murine mAb IIC9. Such conformational changes of CLRPs are not unusual and have been described for the structurally related rhodocytin/aggretin from the same species  and for other heterodimeric venom CLRPs [29,41]. However, they have never been described for heterotetrameric CLRPs in the context of subunit dissociation. Conspiciously, the mAbs against rhodocetin not only detect conformational changes of rhodocetin, but also induce its dissociation when used in immune precipitation. The mAb-induced dissociation recapitulates the changes of the quaternary structure of the rhodocetin tetramer upon integrin binding, which can only be studied with these subunit-specific mAbs. Upon binding to the A-domain of α2β1 integrin, the rhodocetin subunits αβ and γδ dissociate, whereas the rhodocetin γδ subunits associate tightly with the integrin α2 A-domain. From kinetic measurements, we concluded that the rhodocetin αβ subunit accelerates the binding of the rhodocetin γδ subunit to the integrin α2 A-domain. After dissociation into two subunits, the rhodocetin γδ subunit remains at the integrin target with a very slow off-kinetics, thus forming an almost irreversible inhibitory complex. Moreover, with its higher solubility (results not shown), the rhodocetin αβ seems to keep the rhodocetin γδ subunit in solution until it forms the complex with α2β1 integrin on the cell surface. We do not yet know the newly formed interface of the rhodocetin γδ–integrin α2 A-domain complex, but the interface between the two heterodimeric rhodocetin subunits was visualized by the crystal structure of the heterotetrameric rhodocetin . The overall buried surface area for the heterotetramer adds up to 2120Å2 (1 Å=0.1 nm) and is comparatively small, whereby the α and γ subunits contribute to ~80% of this interface (Figure 8). The contact between the rhodocetin subunits αβ and γδ is dominated by electrostatic interactions between the momoners α and γ respectively. The amino acid stretches NKG76QR (α chain residues 74–78) and KEQQC (γ chain residues 77–81) form the centre of the interdomain stabilization core (Figure 8C). Only one additional van der Waals contact (Tyr98 in the α chain and Val94 in the γ chain) can be detected. The small interaction face with a rather low content of hydrophobic patches facilitates the dissociation of rhodocetin tetramer into two soluble subunits, αβ and γδ, with hydrophilic surfaces. This dissociation even occured upon binding of the mAbs, and physiologically upon α2β1 integrin binding.
Rhodocetin is likely to share its structure prototypically with the two other known heterotetrameric CLRPs agglucetin [42,43] and alboaggretin-A . In their primary structure, all three CLRPs conspiciously possess a characteristic glycine residue at position 76 in their α subunits with a central role in subunit interaction (Figure 8C). Lacking the crystal structure of agglucetin and alboaggretin-A, we can only surmise that this residue also comes to lay in the typical loop-in-the-loop motif within the subunit interface of rhodocetin . No information is so far available as to whether agglucetin and alboaggretin-A would dissociate into their heterodimeric subunits.
When bound to cell-anchored α2β1 integrin, we could not detect any rhodocetin αβ subunit bound to the α2β1 integrin-bearing HT1080 cells. This indicates that after binding to cellular α2β1 integrin, tetrameric rhodocetin undergoes integrin binding and dissociation in subsequent and very fast, or even concommitant, steps. The integrin α2 A-domain is able to change its conformation upon collagen binding . Although not proven for rhodocetin binding yet, rhodocetin-induced conformational changes within the A-domain might in turn lead to dissociation of the two rhodocetin subunits. In fact, conformational changes within the integrin α2 A-domain are proven for EMS16, the only other crystallized α2β1 integrin-inhibiting CLRP to date . Binding properties of rhodocetin to the integrin α2 A-domain, such as the effects of divalent cations, support rhodocetin-induced conformational changes within the integrin target . The crystal structure of the complex of rhodocetin γδ with the integrin α2 A-domain will address this question.
The release of the rhodocetin αβ subunit after complex formation which occurs on the cell surface after envenomation prompts the question of potential functions of rhodocetin αβ subunit. In a recent paper , we have demonstrated that the rhodocetin αβ subunit can bind to the platelet glycoprotein GPIb-containing complex on platelets, thereby blocking its interaction with vWF (von Willebrand factor). Thus tetrameric rhodocetin blocks both α2β1 integrin and GPIb with its γδ and αβ subunit respectively, thereby blocking collagen- and vWF-induced platelet aggregation much more efficiently than a venom component targeting a single receptor only. The two other snaclecs, agglucetin [42,43] and alboaggretin-A , which possess a heterotetrameric quaternary structure, block GPIb. However, nothing is known about whether they are able to inhibit α2β1 integrin in a similar manner to rhodocetin. Therefore a general molecular mechanism for heterotetrameric snaclecs similar to the one shown for rhodocetin cannot to date be deduced. The elucidation of the mode of action of rhodocetin in the present study explains the very effective inhibition of platelet aggregation by both collagen and vWF. To develop specific inhibitors to α2β1 integrin, the interaction of rhodocetin γδ subunit with the integrin requires further structural information.
Thilo Bracht, Lydia Sorokin and Johannes Eble generated the hybridomas; Flávia Figueiredo de Rezende produced the YFP-tagged integrin α2 A-domain; and Jörg Stetefeld performed the structural analysis of the rhodocetin subunit interface. Experiments were planned, performed and evaluated mostly by Thilo Bracht and Johannes Eble. Johannes Eble wrote the paper.
This work was supported by the Deutsche Forschungsgemeinschaft through the Excellence Cluster Cardio-Pulmonary System [grant numbers EXC147/1, SFB/TR23, project A8 (both to J.A.E.)]; and the medical faculty of the University of Muenster, Muenster, Germany (to L.M.S.). J.S. is a Canada Research Chair in Structural Biology and receives support from the Heart and Stroke Foundation Manitoba.
Abbreviations: CLRP, C-type lectin-related protein; EYFP, enhanced yellow fluorescent protein; GP, glycoprotein; mAb, monoclonal antibody; RGD, Arg-Gly-Asp; SPR, surface plasmon resonance; TBS, Tris-buffered saline; TBST, 0.05% Tween 20 in TBS; vWF, von Willebrand factor; YFP, yellow fluorescent protein
- © The Authors Journal compilation © 2011 Biochemical Society