Structure and function of recombinant cobra venom factor.

Cobra venom factor (CVF) is the complement-activating protein from cobra venom. It is a structural and functional analog of complement component C3. CVF functionally resembles C3b, the activated form of C3. Like C3b, CVF binds factor B, which is subsequently cleaved by factor D to form the bimolecular complex CVF,Bb. CVF,Bb is a C3/C5 convertase that cleaves both complement components C3 and C5. CVF is a three-chain protein that structurally resembles the C3b degradation product C3c, which is unable to form a C3/C5 convertase. Both C3 and CVF are synthesized as single-chain prepro-proteins. This study reports the recombinant expression of pro-CVF in two insect cell expression systems (baculovirus-infected Sf9 Spodoptera frugiperda cells and stably transfected S2 Drosophila melanogaster cells). In both expression systems pro-CVF is synthesized initially as a single-chain pro-CVF molecule that is subsequently proteolytically processed into a two-chain form of pro-CVF that structurally resembles C3. The C3-like form of pro-CVF can be further proteolytically processed into another two-chain form of pro-CVF that structurally resembles C3b. Unexpectedly, all three forms of pro-CVF exhibit functional activity of mature, natural CVF. Recombinant pro-CVF supports the activation of factor B in the presence of factor D and Mg2+ and depletes serum complement activity like natural CVF. The bimolecular convertase pro-CVF,Bb exhibits both C3 cleaving and C5 cleaving activity. The activity of pro-CVF and the resulting C3/C5 convertase is indistinguishable from CVF and the CVF,Bb convertase. The ability to produce active forms of pro-CVF recombinantly ensures the continued availability of an important research reagent for complement depletion because cobra venom as the source for natural CVF will be increasingly difficult to obtain as the Indian cobra is on the list of endangered species. Experimental systems to express pro-CVF recombinantly will also be invaluable for studies to delineate the structure and function relationship of CVF and its differences from C3 as well as to generate human C3 derivatives with CVF-like function for therapeutic complement depletion ("humanized CVF").

In this manuscript we describe the recombinant expression of single-chain and two two-chain forms of pro-CVF using two different eukaryotic expression systems. Unexpectedly, all three forms of pro-CVF exhibit the functional activity of forming a C3/C5 convertase and depleting complement that is indistinguishable from natural CVF, without the need for posttranslational proteolytic processing into the three-chain form of natural CVF. These results will allow detailed studies of the structure/function relationship of CVF and C3 (23,24) as well as the generation of human C3 derivatives with CVF-like functions ("humanized CVF") as potential agents for therapeutic complement depletion (25). In addition, the ability to generate a functionally active CVF molecule by recombinant means will ensure continued availability of CVF as a research reagent to deplete complement. This is important because the Indian cobra Naja naja and related Asian cobras of the genus Naja 2 are on the list of endangered species, and cobra venom as the source of natural CVF will be increasingly difficult or impossible to obtain.

EXPERIMENTAL PROCEDURES
Materials-CVF was purified from live Naja kaouthia venom (Miami Serpentarium Laboratory; Punta Gorda, FL) as described (22). Human complement components C3, C5, factor B, and factor D were either purified as described previously (26) or obtained from Advanced Research Technologies (San Diego). A polyclonal antiserum to CVF was prepared in a goat (17). Restriction enzymes and other DNA-modifying enzymes were obtained from New England Biolabs.
Construction of a Full-length Pro-CVF cDNA Clone-The CVF venom gland library cDNA clones (CVF65, CVF72, and pCVF106) used for the construction of a full-length CVF cDNA were described previously (19). gt11 clone CVF65 was digested with EcoRI, and the resulting 2.5-kb fragment spanning bp 849 -3350 of the CVF cDNA sequence was subcloned into pUCl8, resulting in the plasmid clone pCVF65/9. gt11 clone CVF72 was digested with SacI and KpnI. The resulting 4-kb fragment (containing ϳ2 kb of CVF sequence at the 5Ј-end) was ligated into pUC18, resulting in the plasmid construct pCVF72. The plasmid clones pCVF65/9 and pCVF106 were both digested with SalI and MunI. The proper fragments were separated and ligated, resulting in the plasmid construct pCVF-MKI. pCVF-MKI was digested with BamHI and NotI and dephosphorylated using calf intestine alkaline phosphatase. The 4.3-kb fragment spanning from bp 1711 to the 3Ј-end of the CVF sequence was isolated. pCVF72 was digested with DraI and BamHI. The proper fragment was isolated and ligated to the pCVF-MK1 fragment. An ATP concentration of 5 mM was used in the ligation reaction to prevent dimerization and formation of concatamers. The ligation product was digested with SalI and subcloned into a NotI/SalI-precut pSportl vector (Invitrogen), resulting in the vector pCVF-FL3. A 1-kb remainder of gt11 sequence at the CVF 5Ј-end was removed by partial digestion with EcoRI. The CVF cDNA has three internal EcoRI sites, so this was achieved by digestion with a high amount of enzyme at room temperature for a short time (30 s) followed by ligase-mediated recircularization. Because a unique SmaI site was removed along with the desired deletion of the gt11 sequence, the background of untruncated plasmids was reduced by linearization with SmaI prior to transformation into Escherichia coli. Plasmid DNA isolated from a polyclonal culture was linearized by digestion with KpnI. Plasmids with a correct length of 10 kb were isolated, religated, and again transformed into E. coli. Monoclonal cultures were screened by plasmid minipreparation, restriction mapping, and T7 sequencing. A correct clone was obtained and named pCVF-FL3⌬. The full-length pro-CVF cDNA can be isolated from the plasmid by digestion with NotI and Kpn2I. The clone was completely resequenced. All cloning operations were done by standard methods (27).
Construction of Baculovirus Expression Transfer Vector for Pro-CVF-pCVF-FL3⌬ was digested with NotI and Kpn2I. The CVF fragment was isolated and ligated into the pVL1393 vector (Pharmingen), which was digested with XmaI and EagI and dephosphorylated with calf intestine alkaline phosphatase. Transfer vector DNA was purified by a cesium chloride gradient. Recombinant baculoviruses encoding full-length pro-CVF were generated using the BaculoGold TM transfection kit (Pharmingen). Monoclonal virus strains were isolated by at least two rounds of plaque purification.
Construction of Plasmids for Expression of Pro-CVF in Drosophila S2 Cells-To express pro-CVF in Drosophila melanogaster S2 cells (Invitrogen) using the native CVF signal sequence, the plasmid pCVF-FL3⌬ was cut with KpnI and NotI. The two fragments were separated by agarose gel electrophoresis, and the 6-kb fragment was isolated from the gel using a QiaQuick TM gel extraction column (Qiagen). This was then ligated into the vector pMT/V5-HisA (Invitrogen) that had been cut with the same enzymes. After confirmation of the junction sequences, the plasmid was stably transfected into Drosophila S2 cells using the calcium phosphate method (28) with 25 g/ml blasticidin (Invitrogen) as a selection agent.
To express pro-CVF in S2 cells using the Drosophila Bip signal sequence from the Drosophila expression vector pMT-Bip/V5-HisA (29), we engineered a new KpnI site in pMT-Bip/V5-HisA, using the Gen-eTailor TM site-directed mutagenesis kit from Invitrogen and following the manufacturer's directions. Primers used for mutagenesis were pMT-Bip-Kpn-F1 (TTGTTGGCCTCTCGCTCGGGGTACCTCCATGGC-CCG) and pMT-Bip-Kpn-R1 (CCCGAGCGAGAGGCCAACAAAGGCC-ACGAC). The signal sequence of natural CVF was removed and an in-frame KpnI site engineered (30), using the primers CVF-KpnSig-R1 (GGGGTACCCTACACCCTCATC) and CVFK-F1 (CCTTCCACTTC-CTCTCCA). After the PCR, the product was purified using the Qiagen PCR purification kit. It was then cut with KpnI and PmlI and ligated into pCVF-FL3⌬ that had been cut with the same enzymes. The 6-kb Kpn-NotI CVF fragment was then transferred into the S2 expression vector as described above.
For pro-CVF expression in Drosophila S2 cells, the stably transformed S2 cell cultures were first expanded in serum-free medium (Invitrogen) in the presence of 25 g/ml blasticidin into 150-cm 2 flasks. Once these cultures reached a cell density of 5 ϫ 10 6 cells/ml, they were expanded into Spinner flasks in serum-free medium at a final density of 1 ϫ 10 6 cells/ml in the absence of blasticidin. Cell growth was monitored, and serum-free medium added to maintain a cell density of 1 ϫ 10 6 cells/ml until the final culture volume reached 1 liter. Cells were then allowed to grow until they reached a final density of 5 ϫ 10 6 cells/ml. At this point, the production of pro-CVF was induced by the addition of copper sulfate to a final concentration of 25 M. After induction, the culture was maintained for 4 -5 days, and the pro-CVF purified as described above.
In Vitro Proteolytic Processing of Single-chain Pro-CVF in Sf9 Culture Supernatants-At the 4th day of infection, the culture supernatant was collected by centrifugation, diluted 1:1 with deionized water, adjusted to pH 7.0, and filtered through a 0.2-m cellulose acetate membrane (Sartorius) into a sterile bottle. This solution was incubated at 4°C for 4 days in the dark. Proteolytically processed recombinant pro-CVF was purified as described for the unprocessed molecule.
Glycosylation Analysis-Glycosylation of recombinant CVF was analyzed by lectin blotting using the DIG Glycan Differentiation Kit (Roche Applied Science). To investigate the effect of inhibiting glycosylation on the secretion of recombinant CVF, Sf9 cells were grown in the presence of 5 g/ml tunicamycin (Calbiochem). At the times indicated, cells and supernatants were harvested and subjected to 7.5% SDS-PAGE under nonreducing conditions with subsequent blotting and development using anti-CVF antiserum.
Complement Consumption Assay-The assay is based on the anticomplementary activity of CVF (31). Briefly, 10 l of normal guinea pig serum (Advanced Research Technologies) was added to a 10-l sample containing CVF or pro-CVF in VBS 2ϩ buffer. The mixture was incubated for 30 min at 37°C. Subsequently, the mixture was diluted with 100 l of VBS 2ϩ containing 0.1% gelatin, and 30 l of sensitized sheep erythrocytes (Advanced Research Technologies) in gelatin containing VBS 2ϩ buffer at a cell concentration of 5 ϫ 10 8 cells/ml was added and incubated at 37°C in a shaking water bath until lysis was apparent in control samples. Subsequently, the reaction was stopped by the addition of 1 ml of ice-cold gelatin containing VBS 2ϩ buffer. Unlysed erythrocytes were sedimented by centrifugation for 2 min at 2,000 ϫ g at 4°C, and the released hemoglobin was determined spectrophotometrically in the supernatant at 412 nm.
Bystander Lysis Assay-This assay for CVF is based on the fluid phase C5 cleaving activity of the CVF,Bb convertase with subsequent bystander lysis of unsensitized guinea pig erythrocytes. It was performed as described previously (25). Briefly, a 20-l sample containing CVF or pro-CVF was incubated for 30 min at 37°C with 20 l of normal guinea pig serum and 20 l of a suspension of guinea pig erythrocytes (HemoStat Laboratories; Dixon, CA) in gelatin containing VBS 2ϩ at a cell concentration of 5 ϫ 10 8 cells/ml. Controls included heat-inactivated guinea pig serum, guinea pig serum supplemented with 2.5 mM Mg-EGTA, and guinea pig serum supplemented with 10 mM EDTA. Subsequently, the reaction was stopped by the addition of 1 ml of ice-cold gelatin containing VBS 2ϩ buffer. Unlysed erythrocytes were sedimented by centrifugation for 2 min at 2,000 ϫ g at 4°C, and the released hemoglobin was determined spectrophotometrically in the supernatant at 412 nm.
Assay for Activation of Factor B-To detect cleavage of factor B into Ba and Bb, CVF or pro-CVF (at 1 M) was incubated for up to 24 h in the presence of a 3-fold molar excess of human factor B in the presence of 0.5 M human factor D and 5 mM MgCl 2 at 37°C (32). The reaction mixture was analyzed by 7.5% SDS-PAGE gels under nonreducing conditions. Gels were stained with Coomassie Blue (Sigma). The disappearance of factor B and the appearance of the cleavage product Bb were quantified by densitometric analysis of stained cells.
Assays for Fluid Phase Cleavage of C3 and C5-To investigate the fluid phase cleavage of C3 and C5, convertases were formed with recombinant CVF and incubated with purified human C3 or C5 as substrate. Cleavage of C3 or C5 was demonstrated by the conversion of the respective ␣-chain into the respective ␣Ј-chain by 7% SDS-PAGE. The assays were performed as described previously (6,33). Briefly, CVF or pro-CVF (15 pmol) was incubated for 1 h at 37°C with human factor B (45 pmol) and human factor D (1.5 pmol) in a total volume of 20 l of VBS 2ϩ supplemented with 5 mM MgCl 2 . Convertase formation was stopped by the addition of EDTA to a concentration of 7.5 mM. Subsequently, 5 g of human C3 or human C5 was added to the convertase solution at a total volume of 25 l. The mixture was incubated for 3 h at 37°C, and aliquots were taken and subjected to 7% SDS-PAGE under reducing conditions. Gels were stained with Coomassie Blue.
Assay for the Presence of an Intramolecular Thioester Bond-The presence of an intramolecular thioester bond in CVF or pro-CVF was analyzed by incorporation of [ 14 C]methylamine (17). [ 14 C]Methylamine (250 Ci; specific activity 54 mCi/mmol; Amersham Biosciences) was dissolved in 0.5 M sodium phosphate buffer (pH 8.0) to a concentration of 54 mM. Single-chain pro-CVF, CVF, or human C3 (0.5-1 nmol) in 40 l of VBS 2ϩ buffer was incubated with 15 l of the [ 14 C]methylamine solution for 10 h at 37°C. Subsequently, excess methylamine was removed by diafiltration against VBS 2ϩ buffer using Ultrafree-MC filter units (Millipore) with a 30-kDa molecular mass cutoff. Radioactivity was measured using a scintillation counter, and the molar incorporation of methylamine was calculated.
Assay for Free Sulfhydryl Groups-The presence of a free sulfhydryl group was analyzed in CVF or pro-CVF by incorporation of iodo-[1- Protein Sequencing by Mass Spectroscopy-To determine the amino acid at position 974 (CVF numbering) of recombinant CVF, the ␥/␤precursor chain of two-chain pro-CVF form Sf9 cells was separated by 7% SDS-PAGE, excised, and subjected to in-gel tryptic digestion (porcine trypsin; Sigma). The gel digests were extracted twice with 10 l of a solution of 50% acetonitrile and 5% trifluoroacetic acid. The extracts were pooled and desalted using a C-18 resin (Ziptip TM , Millipore) and subjected to mass spectroscopy in tandem mode using electrospray ionization (Q-TOF2 mass spectrometer; Micromass; Manchester, UK). On the basis of the predicted tryptic fragmentation at the thioester site, tandem mass spectrometric sequencing was only acquired on selected precursor ions. The spectra were interpreted according to the known fragmentation pathway of peptides under self-ionization, low energy decomposition (34).
Other Methods-Normal guinea pig serum was heat-inactivated by incubation for 30 min at 56°C. SDS-PAGE and Western blotting were performed as described. Isoelectric focusing was performed as described (17,35).

RESULTS
Expression of Recombinant CVF-A full-length prepro-CVF cDNA clone was expressed in two insect cell expression systems: in baculovirus-infected Sf9 S. frugiperda cells and in stably transfected D. melanogaster S2 cells. The recombinant protein was purified from both supernatants to a purity of Ͼ80%. The yield was ϳ500 g/liter. Replacement of the natural CVF signal sequence with the Drosophila Bip signal sequence significantly increased the export efficiency of recombinant CVF from S2 cells resulting in a yield of ϳ1.5 mg/l. A corresponding attempt in Sf9 cells using the signal sequence of the gp67 insect protein instead of the natural CVF signal sequence did not result in a higher yield of the recombinant protein.
Chain Structure of Recombinant CVF-In both expression systems, recombinant CVF is expressed initially as a singlechain polypeptide with a molecular mass of ϳ210 kDa upon SDS-PAGE under nonreducing conditions (Fig. 2). Under reducing conditions, the molecular mass is ϳ185 kDa. The Nterminal sequence of the single-chain pro-CVF is identical to the N-terminal sequence of the CVF ␣-chain, indicating proper removal of the signal peptide. In both expression systems, single-chain pro-CVF is subsequently proteolytically processed by removal of the four arginine residues at positions 628 through 631 (CVF numbering) resulting in a two-chain pro-CVF molecule with the two chains exhibiting molecular masses of ϳ115 and 70 kDa, respectively (Fig. 2). N-terminal sequencing identified the N terminus of the 115-kDa chain as the C3a-like domain of pro-CVF and the N terminus of the 70-kDa chain as the CVF ␣-chain. The 115-kDa chain (termed ␥/␤precursor chain) of this two-chain form of pro-CVF is therefore homologous to the C3 ␣-chain. This two-chain form of pro-CVF structurally resembles a mature C3 protein. In both expression systems, the C3-like two-chain form of pro-CVF is further proteolytically processed by removal of the C3a-like domain of the CVF ␥/␤-precursor chain into a second form of a two-chain pro-CVF, consisting of a 105-kDa chain and the 70-kDa CVF ␣-chain. N-terminal sequencing identified the N terminus of the 105-kDa chain as the N terminus of the CVF ␥-chain. 4 These results identify the second form of two-chain pro-CVF as a C3b-like form of pro-CVF, with the 105-kDa chain (termed CVF ␥/␤Ј-precursor chain) being homologous to the ␣Ј-chain of C3b (Fig. 2).
The two eukaryotic expression systems differ in the extent of the proteolytic processing of the single-chain pro-CVF. In Sf9 cells, single-chain pro-CVF is the predominant form identified in the culture supernatant. Incubation of the cell-free supernatant from Sf9 cells for 4 days at 4°C and pH 7.0 resulted in proteolytic processing of the single-chain pro-CVF molecule into a mixture of the two two-chain pro-CVF molecules resembling C3 and C3b, respectively (Fig. 2). Prolonged continued incubation of the supernatant for 2 weeks led to virtually complete conversion of the 115-kDa CVF ␥/␤-precursor chain into the 105-kDa CVF ␥/␤Ј-precursor chain (not shown). In S2 cells, the single-chain pro-CVF is readily processed into a mixture of the two two-chain pro-CVF molecules, with the C3-like two-chain pro-CVF being the predominant product (Fig. 2). No further processing into a three-chain molecule resembling natural CVF was observed in either of the two expression systems.
Biochemical Characteristics of Recombinant CVF-Recombinant CVF produced in both Sf9 and S2 cells was glycosylated (Fig. 3). Lectin blot analyses revealed the presence of N-linked oligosaccharides of the high mannose type and O-linked oligosaccharides of the simple galactose-␤(1-3)-N-acetylgalactosamine type. No complex N-or O-glycosylation was detected. The glycosylation pattern was qualitatively similar between recombinant CVF produced in Sf9 and S2 cells, with the exception that sialic acid residues, terminally linked to galactose residues, could only be detected in recombinant CVF from S2 cells. Treatment of Sf9 cells with tunicamycin inhibited the secretion of recombinant CVF into the supernatant (Fig. 4) and caused its intracellular accumulation and degradation, indicating a functional role of the oligosaccharide chains for folding and/or intracellular sorting and secretion of recombinant CVF.
Pro-CVF does not react with [ 14 C]methylamine, indicating the absence of a thioester bond. Consistent with this observation, pro-CVF was found to react with iodo-[ 14 C]acetamide (0.68 mol/mol), indicating the presence of a single free sulfhydryl group (Table I). Using mass spectroscopy analysis of a tryptic digest of the ␥/␤-precursor chain of pro-CVF, the   lanes 1-7)  amino acid residue at position 974 (CVF numbering) was identified as glutamine in ϳ80% of the molecules, indicating that an intramolecular thioester had never been formed in recombinant CVF.
Single-chain pro-CVF focused between pH 6.5 and 6.8 upon isoelectric focusing. The isoelectric point is slightly more basic than that of natural CVF, consistent with the difference in the amino acid composition of natural CVF and single-chain pro-CVF, the latter of which additionally contains the four arginine residues and the C3a-like and C3d-like domains.
Functional Activity of Recombinant CVF-Pro-CVF causes the consumption of serum complement activity indistinguishable from natural CVF (Fig. 5). Pro-CVF supports factor B activation in the presence of factor D and Mg 2ϩ like natural CVF (Fig. 6). The resulting bimolecular complex pro-CVF,Bb is a C3/C5 convertase that cleaves purified human C3 (Fig. 7) and C5 (Fig. 8) like the C3/C5 convertase CVF,Bb formed with natural CVF. Consistent with the ability of pro-CVF,Bb to cleave C5, pro-CVF induces bystander lysis of erythrocytes (Fig. 9). No activity differences were observed between singlechain pro-CVF and the two forms of two-chain pro-CVF or any mixtures thereof in any of the activity assays.

Pro-CVF can be expressed recombinantly in insect cells using baculovirus-infected Sf9 cells or stably transformed S2
cells. In either case, the signal peptide is removed from the single-chain prepro-CVF, and the recombinant protein is secreted into the culture supernatant. Dependent on the cellular system and the experimental conditions used, pro-CVF is found in the culture supernatant as a single-chain pro-CVF, a twochain pro-CVF homologous to C3, a two-chain pro-CVF homologous to C3b, or mixtures thereof. The first two-chain pro-CVF molecule generated is C3-like, consisting of a 115-kDa ␥/␤precursor chain that is homologous to the C3 ␣-chain, and a 70-kDa chain that is the ␣-chain of natural CVF and which is homologous to the C3 ␤-chain. Further proteolytic processing of the ␥/␤-precursor chain of the C3-like two-chain form of pro-CVF by removal of the C3a-like domain results in a second form of a two-chain pro-CVF molecule which is homologous to C3b. It consists of the 105-kDa ␥/␤Ј-precursor chain which is homologous to the C3 ␣Ј-chain, and the 70-kDa ␣-chain of natural CVF. No further processing of the C3b-like form of pro-CVF by removal of a C3d-like domain into the mature three-chain form of natural CVF was observed in either of the two insect cell expression systems. Fig. 10 schematically depicts the chain structures of the three forms of pro-CVF as well as native CVF.
The conversion of the single-chain pro-CVF molecule into the C3-like two-chain pro-CVF molecule involves the removal of the four arginine residues at positions 628 -631 (CVF numbering) (19). The removal of the four arginine residues usually occurs by furin-type proteases in the trans-Golgi network (36). The efficiency of this cleavage differs between Sf9 cells and S2 cells, and it is relatively poor in Sf9 cells. Relatively poor processing was also observed for human pro-C3 expressed in Sf9 cells using the baculovirus system (37). The amount of unprocessed pro-C3 increased even further in Sf9 cells if chimeric pro-C3 molecules were expressed by replacing regions of human C3 with corresponding regions of C3 from other species (37). It therefore appears that the efficiency of cleavage of single chain pro-CVF in Sf9 cells is affected by the secondary or tertiary structure at or near the cleavage site. However, other effects such as differential intracellular trafficking resulting from differences in oligosaccharide trimming may also play a role in the relatively inefficient processing of single-chain pro-CVF in Sf9 cells. Nevertheless, virtually complete cleavage at the four-Arg site in single-chain pro-CVF produced in Sf9 cells can be accomplished by incubation of the cell supernatant for 4 days. In contrast to Sf9 cells, S2 cells efficiently process singlechain pro-CVF into the C3-like form of two-chain pro-CVF.
In both insect cell expression systems, further processing of the C3-like two-chain pro-CVF molecule into the C3b-like twochain form of pro-CVF by removal of the C3a-like domain from the ␥/␤-precursor chain was observed. In both systems, this conversion seems rather inefficient. However, virtually complete conversion could be achieved by incubating the Sf9 cell supernatant for an extended period of time (2 weeks). N-terminal sequencing of the ␥/␤Ј-precursor chain of pro-CVF revealed Arg-710 as the N-terminal residue rather than Asp-711, which is the N terminus of the CVF ␥-chain, indicating that the proteolytic processing occurs between Ala-709 and Arg-710 rather than between Arg-710 and Asp-711 as is the case in the venom gland for natural CVF. The enzyme responsible for the conversion of the C3-like form of pro-CVF into the C3b-like form of pro-CVF in the insect cell supernatants is not known.
The most surprising aspect of this work is the observation that single-chain pro-CVF as well as the two forms of two-chain pro-CVF exhibit functional activity of forming a C3/C5 convertase that is indistinguishable from natural CVF, a three-chain molecule structurally resembling C3c (17). Although there are a number of pro-proteins exhibiting similar functional activity to the corresponding mature and fully processed proteins (e.g. bovine pancreatic RNase (38), porcine pancreatic ␣-amylase (39), human insulin pro-receptor (40), fish anti-freeze protein (41), and human myeloperoxidase (42)), the functional activity of pro-CVF raises the question about the biological function of the complex proteolytic processing of pro-CVF with removal of the C3a-like and C3d-like domains into natural CVF as it occurs in the venom gland. Accidental proteolytic processing of pro-CVF in the venom gland appears to be an unlikely explanation because cobra venom does not contain proteases as components of its toxic activity, and proteolytic activity is virtually absent in cobra venom (43,44). All the more surprising in this context is the presence of a highly specific metalloprotease, termed cobrin, in cobra venom, which cleaves human C3 into a three-chain structure resembling but not identical to C3c, termed C3o (32,45,46). In contrast to C3c, C3o supports the activation of factor B in the presence of factor D and Mg 2ϩ (32). Presumably, cobrin is responsible for the processing of pro-CVF into the mature three-chain CVF molecule in the cobra venom gland. Future research will be necessary to elucidate the biological function of the processing of pro-CVF into the mature three-chain CVF molecule and the role of cobrin in this process.
The structural basis required for any C3 derivative to form a convertase with factor B is unknown. Although C3(H 2 O), C3b, and C3o support activation of factor B, C3, the two forms of iC3b, and C3c do not. In contrast, all known forms of CVF (single-chain pro-CVF, which resembles pro-C3; the two twochain pro-CVF molecules, which resemble C3 and C3b, respectively; and natural CVF, which resembles C3c and C3o) support the activation of factor B. It therefore appears that in those forms of C3 that do not support factor B activation, the responsible structure is altered or masked for biological reasons that relate to the regulation of complement activation. Native C3 as it circulates in plasma is obviously not wanted to activate factor B. The intramolecular thioester appears to be the molecular structure responsible for preventing factor B activation. Hydrolysis of the thioester in both native C3 and nascent C3b allows the resulting C3 derivatives C3(H 2 O) and C3b to form a convertase. When further convertase formation is no longer warranted it is achieved by cleavage of C3(H 2 O) and C3b into iC3b and eventually C3c. The molecular structure responsible for preventing iC3b (and possibly even C3c) from activating factor B and forming more convertase molecules has not been identified. Interestingly, the ability to support factor B activation must be harbored but masked in iC3b because C3o supports factor B activation (32), indicating that cobrin removes or alters the molecular structure responsible for preventing iC3b from activating factor B.
Pro-CVF does not have the intramolecular thioester present in C3. Furthermore, the encoded amino acid glutamine was found in 80% of the pro-CVF molecules at position 974 (CVF numbering) rather than glutamic acid, which was found in the remaining 20% of the molecules. This observation indicates that the thioester in most cases had never formed and, most likely, did not form at all as 20% glutamic acid residues found at position 974 are likely to be the consequence of spontaneous deamination during storage rather than hydrolysis of a preexisting intramolecular thioester. The lack of thioester formation is consistent with the biological role of CVF. There is no need for an inactive precursor form with regard to convertase formation as is the case for C3. Quite to the contrary, CVF is meant to form a convertase and consume complement as soon as it comes in contact with the plasma of a prey animal.
The formation of the intramolecular thioester is a spontaneous process rather than enzymatic (47). Although pro-CVF exhibits the tetrapeptide sequence CGEQ (residues 971-974, CVF numbering) which forms the 14-member thiolactone structure in C3, the thioester does not form. The CGEQ sequence is not sufficient to form the thioester because many more proteins without intramolecular thioesters have this sequence. There is significant sequence conservation in regions immediately flanking the tetrapeptide sequence in proteins that have an intramolecular thioester such as C3, C4, and ␣ 2 -macroglobulin. Of particular relevance appears to be a proline residue at position 1020 (C3 numbering) (48). Site-directed mutagenesis at position 1020 from proline to glycine resulted in a C3 phenotype unable to form the thioester bond (48). In contrast to human C3 (21), cobra C3 (20), and a large number of other C3, C4, and ␣ 2 -macroglobulin molecules from various species with a thioester, CVF encodes an alanine at this position (981 CVF numbering) (19). This observation lends further support that the proline residue at this position is essential for forming an intramolecular thioester.
Single-chain pro-CVF exhibits upon SDS-PAGE under reducing conditions an apparent molecular mass of 185 kDa, which is ϳ25 kDa smaller than under nonreducing conditions. Such anomalous faster migration has also been found for other proteins (e.g. sporamin A (49)). Especially in single-chain molecules, the reducing of intramolecular disulfide bonds can lead to major conformational changes and exposition of core regions with different polarity.
The glycosylation of recombinant CVF was found to consist of N-linked chains of the high mannose type and simple O-linked chains of the galactose-␤(1-3)-N-acetylgalactosamine type. No complex N-or O-glycosylation was found. These findings are consistent with oligosaccharide structures normally produced in insect cells (50). The oligosaccharide structures of recombinant CVF are significantly different from natural CVF, which contains predominantly N-linked oligosaccharide chains of the complex type and no O-glycosylation (51,52). The finding that recombinant CVF with its insect-type glycosylation exhibits functional activity indistinguishable from natural CVF represents independent confirmation that glycosylation of CVF is not required for its functional activity because partial and complete deglycosylation of natural CVF as well as sialylation of the oligosaccharide chains of natural CVF did not affect its activity (53). However, glycosylation of recombinant CVF in insect cells is important for its correct folding and subsequent secretion into the supernatant as incubation of the cells with tunicamycin inhibits the secretion of recombinant CVF. Oligosaccharides may also have a function for correct sorting of the protein in the endoplasmic reticulum (54).
The ability to produce a functionally active form of CVF by recombinant means now provides an important experimental system to conduct detailed studies of the structure and function relationship of CVF and C3 molecules (23,24). In addition, because the Indian cobra N. naja has been placed on the list of endangered species, the ability to produce recombinantly a functionally active form of CVF ensures continued availability of this protein, which has become an important tool to study the biological functions of complement in the immune response, in host defense, and in the pathogenesis of disease (1), as well as a gold standard for evaluating the anticomplementary activity of experimental drugs. Significant effort is devoted to develop drugs for complement inhibition (16). Whereas these experimental drugs aim to inhibit complement activation, CVF inhibits complement by depleting it. The ability to produce recombinantly functionally active forms of CVF also allows the production of human C3 derivatives with CVF-like functions ("humanized CVF") (25). A humanized CVF represents a different concept for an anticomplementary agent and may be a promising drug for therapeutic complement depletion.