Dissecting the regions of virion-associated Kaposi's sarcoma-associated herpesvirus complement control protein required for complement regulation and cell binding

Complement, which bridges innate and adaptive immune responses as well as humoral and cell-mediated immunity, is antiviral. Kaposi's sarcoma-associated herpesvirus (KSHV) encodes a lytic cycle protein called KSHV complement control protein (KCP) that inhibits activation of the complement cascade. It does so by regulating C3 convertases, accelerating their decay, and acting as a cofactor for factor I degradation of C4b and C3b, two components of the C3 and C5 convertases. These complement regulatory activities require the short consensus repeat (SCR) motifs, of which KCP has four (SCRs 1 to 4). We found that in addition to KCP being expressed on the surfaces of experimentally infected endothelial cells, it is associated with the envelope of purified KSHV virions, potentially protecting them from complement-mediated immunity. Furthermore, recombinant KCP binds heparin, an analogue of the known KSHV cell attachment receptor heparan sulfate, facilitating infection. Treating virus with an anti-KCP monoclonal antibody (MAb), BSF8, inhibited KSHV infection of cells by 35%. Epitope mapping of MAb BSF8 revealed that it binds within SCR domains 1 and 2, also the region of the protein involved in heparin binding. This MAb strongly inhibited classical C3 convertase decay acceleration by KCP and cofactor activity for C4b cleavage but not C3b cleavage. Our data suggest similar topological requirements for cell binding by KSHV, heparin binding, and regulation of C4b-containing C3 convertases but not for factor I-mediated cleavage of C3b. Importantly, they suggest KCP confers at least two functions on the virion: cell binding with concomitant infection and immune evasion.     

The Kaposi's sarcoma-associated herpesvirus complement control protein mimics human molecular mechanisms for inhibition of the complement system

Kaposi's sarcoma-associated human herpesvirus (KSHV) is thought to cause Kaposi's sarcoma, primary effusion lymphoma, and multicentric Castleman's disease. Previously, we reported that the KSHV complement control protein (KCP) encoded within the viral genome is a potent regulator of the complement system; it acts both as a cofactor for factor I and accelerates decay of the C3 convertases (Spiller, O. B., Blackbourn, D. J., Mark, L., Proctor, D. G., and Blom, A. M. (2003) J. Biol. Chem. 278, 9283-9289). KCP is a homologue to human complement regulators, being comprised of four complement control protein (CCP) domains. In this, the first study to identify the functional sites of a viral homologue at the amino acid level, we created a three-dimensional homology-based model followed by site-directed mutagenesis to locate complement regulatory sites. Classical pathway regulation, both through decay acceleration and factor I cleavage of C4b, required a cluster of positively charged amino acids in CCP1 stretching into CCP2 (Arg-20, Arg-33, Arg-35, Lys-64, Lys-65, and Lys-88) as well as positively (Lys-131, Lys-133, and His-135) and negatively (Glu-99, Glu-152, and Asp-155) charged areas at opposing faces of the border region between CCPs 2 and 3. The regulation of the alternative pathway (via factor I-mediated C3b cleavage) was found to both overlap with classical pathway regulatory sites (Lys-64, Lys-65, Lys-88 and Lys-131, Lys-133, His-135) as well as require unique, more C-terminal residues in CCPs 3 and 4 (His-158, His-171, and His-213) and CCP 4 (Phe-195, Phe-207, and Leu-209). We show here that KCP has evolved to maintain the spatial structure of its functional sites, especially the positively charged patches, compared with host complement regulators.     

Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) open reading frame 4 protein (kaposica) is a functional homolog of complement control proteins

The genome analysis of Kaposi's sarcoma-associated herpesvirus (KSHV) has revealed the presence of an open reading frame (ORF 4) with sequence homology to complement control proteins. To assign a function to this protein, we have now expressed this ORF using the Pichia expression system and shown that the purified protein inhibited human complement-mediated lysis of erythrocytes, blocked cell surface deposition of C3b (the proteolytically activated form of C3), and served as a cofactor for factor I-mediated inactivation of complement proteins C3b and C4b (the subunits of C3 convertases). Thus, our data indicate that this KSHV inhibitor of complement activation (kaposica) provides a mechanism by which KSHV can subvert complement attack by the host.     

Role of membrane cofactor protein (CD46) in regulation of C4b and C3b deposited on cells

C4b and C3b deposited on host cells undergo limited proteolytic cleavage by regulatory proteins. Membrane cofactor protein (MCP; CD46), factor H, and C4b binding protein mediate this reaction, known as cofactor activity, that also requires the plasma serine protease factor I. To explore the roles of the fluid phase regulators vs those expressed on host cells, a model system was used examining complement fragments deposited on cells transfected with human MCP as assessed by FACS and Western blotting. Following incubation with Ab and complement on MCP(+) cells, C4b was progressively cleaved over the first hour to C4d and C4c. There was no detectable cleavage of C4b on MCP(-) cells, indicating that MCP (and not C4BP in the serum) primarily mediates this cofactor activity. C3b deposition was not blocked on MCP(+) cells because classical pathway activation occurred before substantial C4b cleavage. Cleavage, though, of deposited C3b was rapid (<5 min) and iC3b was the dominant fragment on MCP(-) and MCP(+) cells. Studies using a function-blocking mAb further established factor H as the responsible cofactor. If the level of Ab sensitization was reduced 8-fold or if Mg(2+)-EGTA was used to block the classical pathway, MCP efficiently inhibited C3b deposition mediated by the alternative pathway. Thus, for the classical pathway, MCP is the cofactor for C4b cleavage and factor H for C3b cleavage. However, if the alternative pathway mediates C3b deposition, then MCP's cofactor activity is sufficient to restrict complement activation.     

Mutations in alpha-chain of C4BP that selectively affect its factor I cofactor function

C4b-binding protein (C4BP) inhibits all pathways of complement activation, acting as a cofactor to the serine protease factor I (FI) in the degradation of activated complement factors C4b and C3b. C4BP is a disulfide-linked polymer of seven alpha-chains and a unique beta-chain, the alpha- and beta-chains being composed of eight and three complement control protein (CCP) domains, respectively. In previous studies we have localized cofactor activity and binding of C4b to alpha-chain CCP1-3 of C4BP, whereas the binding of C3b required additionally CCP4. Likewise, introduced point mutations that decreased binding of C4b/C3b caused a decrease in cofactor activity. In the present study, we describe two mutants of C4BP, K126Q/K128Q and F144S/F149S, clustered on alpha-chain CCP3, which selectively lost their ability to act as cofactors in the cleavage of both C4b and C3b. Both mutants show the same binding affinity for C4b/C3b as measured by surface plasmon resonance and have the same inhibitory effect on formation and decay of the classical pathway C3-convertase as the wild type C4BP. It appears that C4b and C3b do not undergo the same conformational changes upon binding to the C4BP mutants as during the interaction with the wild type C4BP, which then results in the observed loss of the cofactor activity.     

Analysis of Binding Sites on Complement Factor I That Are Required for Its Activity

The central complement inhibitor factor I (FI) degrades activated complement factors C4b and C3b in the presence of cofactors such as C4b-binding protein, factor H, complement receptor 1, and membrane cofactor protein. FI is a serine protease composed of two chains. The light chain comprises the serine protease domain, whereas the heavy chain contains several domains; that is, the FI and membrane attack complex domain (FIMAC), CD5, low density lipoprotein receptor 1 (LDLr1) and LDLr2 domains. To understand better how FI acts as a complement inhibitor, we used homology-based models of FI domains to predict potential binding sites. Specific amino acids were then mutated to yield 16 well expressed mutants, which were then purified from media of eukaryotic cells for functional analyses. The Michaelis constant (K_m) of all FI mutants toward a small substrate was not altered, whereas some mutants showed increased maximum initial velocity (V_max). All the mutations in the FIMAC domain affected the ability of FI to degrade C4b and C3b irrespective of the cofactor used, whereas only some mutations in the CD5 and LDLr1/2 domains had a similar effect. These same mutants also showed impaired binding to C3met. In conclusion, the FIMAC domain appears to harbor the main binding sites important for the ability of FI to degrade C4b and C3b.     

Analysis of binding sites on complement factor I using artificial N-linked glycosylation

Factor I (FI) is a serine protease that inhibits all complement pathways by degrading activated complement components C3b and C4b. FI functions only in the presence of several cofactors, such as factor H, C4b-binding protein, complement receptor 1, and membrane cofactor protein. FI is composed of two chains linked by a disulfide bridge; the light chain comprises only the serine protease (SP) domain, whereas the heavy chain contains the FI membrane attack complex domain (FIMAC), CD5 domain, and low density lipoprotein receptor 1 (LDLr1) and LDLr2 domains. To better understand how FI inhibits complement, we used homology-based three-dimensional models of FI domains in an attempt to identify potential protein-protein interaction sites. Specific amino acids were then mutated to yield 20 recombinant mutants of FI carrying additional surface-exposed N-glycosylation sites that were expected to sterically hinder interactions. The Michaelis constant (K(m)) of all FI mutants toward a small substrate was not increased. We found that many mutations in the FIMAC and SP domains nearly abolished the ability of FI to degrade C4b and C3b in the fluid phase and on the surface, irrespective of the cofactor used. On the other hand, only a few alterations in the CD5 and LDLr1/2 domains impaired this activity. In conclusion, all analyzed cofactors form similar trimolecular complexes with FI and C3b/C4b, and the accessibility of FIMAC and SP domains is crucial for the function of FI.     

Factor I-mediated processing of complement fragments on HIV immune complexes targets HIV to CR2-expressing B cells and facilitates B cell-mediated transmission of opsonized HIV to T cells

Our study demonstrates that binding of complement-opsonized HIV to complement receptor type 1 on human erythrocytes (E) via C3b fragments is followed by a rapid normal human serum-mediated detachment of HIV from E. The release was dependent on the presence of factor I indicating a conversion of C3b fragments to iC3b and C3d on the viral surface. This in turn resulted in an efficient binding of opsonized HIV to CR2-expressing B cells, thus facilitating B cell-mediated transmission of HIV to T cells. These data provide a new dynamic view of complement opsonization of HIV, suggesting that association of virus with E might be a transient phenomenon and the factor I-mediated processing of C3b to iC3b and C3d on HIV targets the virus to complement receptor type 2-expressing cells. Thus, factor I in concert with CR1 on E and factor H in serum due to their cofactor activity are likely to be important contributors for the generation of C3d-opsonized infectious HIV reservoirs on follicular dendritic cells and/or B cells in HIV-infected individuals.     

Characterization of the complement inhibitory function of rhesus rhadinovirus complement control protein (RCP)

Rhesus rhadinovirus (RRV) is currently the closest known, fully sequenced homolog of human Kaposi sarcoma-associated herpesvirus. Both these viruses encode complement inhibitors as follows: Kaposi sarcoma-associated herpesvirus-complement control protein (KCP) and RRV-complement control protein (RCP). Previously we characterized in detail the functional properties of KCP as a complement inhibitor. Here, we performed comparative analyses for two variants of RCP protein, encoded by RRV strains H26-95 and 17577. Both RCP variants and KCP inhibited human and rhesus complement when tested in hemolytic assays measuring all steps of activation via the classical and the alternative pathway. RCP variants from both RRV strains supported C3b and C4b degradation by factor I and decay acceleration of the classical C3 convertase, similar to KCP. Additionally, the 17577 RCP variant accelerated decay of the alternative C3 convertase, which was not seen for KCP. In contrast to KCP, RCP showed no affinity to heparin and is the first described complement inhibitor in which the binding site for C3b/C4b does not interact with heparin. Molecular modeling shows a structural disruption in the region of RCP that corresponds to the KCP-heparin-binding site. This makes RRV a superior model for future in vivo investigations of complement evasion, as RCP does not play a supportive role in viral attachment as KCP does.     

Functional properties of membrane cofactor protein of complement.

Membrane cofactor protein (MCP or gp45-70) of the complement system is a cofactor for factor I-mediated cleavage of fluid-phase C3b and C3b-like C3, which opens the thioester bond. In the present study the activity of MCP was further characterized. Unexpectedly, in the absence of factor I, MCP stabilized the alternative- and, to a lesser extent, the classical-pathway cell-bound C3 convertases and thereby enhanced C3b deposition. Soluble MCP, if added exogenously, hardly functioned as cofactor for the cleavage of erythrocyte-bound C3b to iC3b; i.e. its activity, compared with the cofactor activity of factor H, was inefficient, since less than 10% of the bound C3b was MCP-sensitive. Further, exogenously added soluble MCP was also a weak cofactor for the cleavage of C3b bound to zymosan. Likewise, factor I, in the presence of cells bearing MCP, cleaved fluid-phase C3b inefficiently. These results imply that MCP has very little extrinsic cofactor activity for factor I. In contrast, exogenously added MCP and factor I mediated efficient cleavage of erythrocyte-bound C3b if the concentration of Nonidet P40 was sufficient to solubilize the cells. Interestingly, soluble MCP and factor I degraded C3b attached to certain solubilized acceptor membrane molecules more readily than others. The cleavage reaction of fluid-phase and cell-bound C3b by soluble MCP and factor I produced iC3b, but no C3c and C3dg. These and prior data indicate that soluble MCP has potent cofactor activity for fluid-phase C3b or C3b bound to solubilized molecules, but acts inefficiently towards C3b on other cells. This functional profile is unique for a C3b/C4b binding protein and, taken together with its wide tissue distribution, suggests an important role for MCP in the regulation of the complement system.     

Physiologic inactivation of fluid phase C3b: isolation and structural analysis of C3c, C3d,g (alpha 2D), and C3g

The fragments that result from the inactivation of C3b have not been completely characterized. Initial inactivation is catalyzed by the protease factor I, which, in the presence of its cofactor (factor H), cleaves two peptide bonds in the alpha'-chain of C3b. This results in the release of a small peptide (C3f, Mr 3000) from iC3b, which consists of the C3 beta chain covalently bonded to two alpha'-chain-derived peptides (Mr 68,000 and Mr 43,000). Surface-bound iC3b is cleaved at a third site by factor I to produce C3c and C3d,g (or alpha 2D). The factor I cofactor for this cleavage is the C3b receptor that is present on erythrocyte and leukocyte membranes. This report describes the isolation and initial structural characterization of C3c and C3d,g generated in whole blood after complement activation with cobra venom factor. These fragments were compared with the C3 fragments isolated from the serum and plasma of a patient with complement activation in vivo. The fragments were isolated with two solid phase monoclonal antibodies, one of which recognizes a determinant on C3g (clone 9) and one of which recognizes a determinant on C3c (clone 4). C3c isolated from normal blood showed three polypeptides that had apparent m.w. of 75,000, 43,000, and 27,000. The C3d,g consisted of a single polypeptide chain with a m.w. of 40,000. Amino terminal sequence analysis showed that the Mr 27,000 peptide from C3c is derived from the amino terminal portion of the alpha'-chain of C3b, whereas the Mr 43,000 peptide is derived from the carboxy terminus of the same chain. Amino terminal sequence analysis showed also that C3g is derived from the amino terminus of C3d,g. The C3 fragments isolated from a patient with partial lipodystrophy, nephritic factor activity, low serum C3 levels, and circulating C3 cleavage products showed a more complicated pattern on SDS-PAGE. The fragment isolated with clone 9 had an apparent m.w. of 40,000, identical to C3d,g generated in vitro, and it had the same amino terminal sequence as C3d,g generated in vitro. The eluate from insolubilized clone 4, however, showed prominent bands with Mr of 75,000, 56,000, 43,000, and 27,000, together with a triple-banded pattern at 68,000 and a minor band at 80,000. This eluate thus appears to contain C3c, and iC3b or an iC3b-like product. The origin of the Mr 56,000 and Mr 80,000 peptides have not yet been determined. These studies, with previous data, definitively order the C3c and C3d,g peptides in the alpha-chain of C3.(ABSTRACT TRUNCATED AT 400 WORDS)     

Complement is an essential component of the immune response to adeno-associated virus vectors

Adeno-associated virus (AAV) vectors are associated with relatively mild host immune responses in vivo. Although AAV induces very weak innate immune responses, neutralizing antibodies against the vector capsid and transgene still occur. To understand further the basis of the antiviral immune response to AAV vectors, studies were performed to characterize AAV interactions with macrophages. Primary mouse macrophages and human THP-1 cells transduced in vitro using an AAV serotype 2 (AAV2) vector encoding green fluorescent protein did not result in measurable transgene expression. An assessment of internalized vector genomes showed that AAV2 vector uptake was enhanced in the presence of normal but not heat-inactivated or C3-depleted mouse/human serum. Enhanced uptake in the presence of serum coincided with increased macrophage activation as determined by the expression of NF-κB-dependent genes such as macrophage inflammatory protein 2 (MIP-2), interleukin-1β (IL-1β), IL-8, and MIP-1β. AAV vector serotypes 1 and 8 also activated human and mouse macrophages in a serum-dependent manner. Immunoprecipitation studies demonstrated the binding of iC3b complement protein to the AAV2 capsid in human serum. AAV2 did not activate the alternative pathway of the complement cascade and lacked cofactor activity for factor I-mediated degradation of C3b to iC3b. Instead, our results suggest that the AAV capsid also binds complement regulatory protein factor H. In vivo, complement receptor 1/2- and C3-deficient mice displayed impaired humoral immunity against AAV2 vectors, with a delay in antibody development and significantly lower neutralizing antibody titers. These results show that the complement system is an essential component of the host immune response to AAV.     

Identification of functional domains in kaposica, the complement control protein homolog of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8)

Recently it has been shown that kaposica, an immune evasion protein of Kaposi's sarcoma-associated herpesvirus, inactivates complement by acting on C3-convertases by accelerating their decay as well as by acting as a cofactor in factor I-mediated inactivation of their subunits C3b and C4b. Here, we have mapped the functional domains of kaposica. We show that SCRs 1 and 2 (SCRs 1-2) and 1-4 are essential for the classical and alternative pathway C3-convertase decay-accelerating activity (DAA), respectively, while the SCRs 2-3 are required for factor I cofactor activity (CFA) for C3b and C4b. SCR 3 and SCRs 1 and 4, however, contribute to optimal classical pathway DAA and C3b CFA, respectively. Binding data show that SCRs 1-4 and SCRs 1-2 are the smallest structural units required for measuring detectable binding to C3b and C4b, respectively. The heparin-binding site maps to SCR 1.     

Effects of zinc on factor I cofactor activity of C4b-binding protein and factor H

Complement inhibition is to a large extent achieved by proteolytic degradation of activated complement factors C3b and C4b by factor I (FI). This reaction requires a cofactor protein that binds C3b/C4b. We found that the cofactor activity of C4b-binding protein towards C4b/C3b and factor H towards C3b increase at micromolar concentrations of Zn(2+) and are abolished at 2 mM Zn(2+) and above. 65Zn(2+) bound to C3b and C4b molecules but not the cofactors or FI when they were immobilized in a native form on a nitrocellulose membrane. Zn(2+) binding constants for C3met (0.2 microM) and C4met (0.1 microM) were determined using fluorescent chelator. It appears that higher cofactor activity at low zinc concentrations is due to an increase of affinity between C4b/C3b and cofactor proteins as assessed by surface plasmon resonance. Inhibition of the reaction seen at higher concentrations is due to aggregation of C4b/C3b.     

Enhancement of complement activation and opsonophagocytosis by complexes of mannose-binding lectin with mannose-binding lectin-associated serine protease after binding to Staphylococcus aureus

Human mannose-binding lectin (MBL) is a serum protein of the innate immune system that circulates as a complex with a group of so-called MBL-associated serine proteases (MASP-1, MASP-2, and MASP-3). Complexes of MBL-MASP2 are able to activate the complement system in an Ab and C1-independent fashion after binding of the lectin to appropriate microbial sugar arrays. We have evaluated the additive effect of the lectin pathway relative to other complement activation pathways and the subsequent effect on neutrophil phagocytosis. Complement activation in the sera of MBL-deficient individuals was studied with and without the addition of exogenous MBL-MASP. Flow cytometry was used to measure the deposition of C4, factor B, C3b, and iC3b on Staphylococcus aureus. Deposition of the first cleavage product of the lectin pathway, C4b, was increased using the sera of three different MBL-deficient individuals when exogenous MBL-MASP was added. Factor B was deposited in association with C4, but there was no evidence of independent alternative pathway activation. Similar enhancement of C3b deposition was also observed, with evidence of elevated amounts of C3b processed to iC3b. The increase in opsonic C3 fragments mediated by MBL was associated with a significant increase in the uptake of organisms by neutrophils. We also observed significant increases in phagocytosis with MBL-MASPs that were independent of complement activation. We conclude that MBL-MASP makes a major contribution to complement-mediated host defense mechanisms.     

Analysis of C3-receptor activity on human B-lymphocytes and isolation of the complement receptor type 2 (CR2).

The rosetting of defined C3-fragment-coated sheep erythrocytes to B-cell-enriched tonsil lymphocytes was measured. The rosetting lymphocytes were homogeneous with respect to expression of C3b, iC3b and C3d receptors. Isolation of receptors for C3 fragments from surface-radioiodinated lymphocytes by affinity chromatography on immobilized C3u, iC3b and C3d,g produced two proteins with partially overlapping specificities. A protein of 240 000 Mr, recognized by the monoclonal antibody To5 and identified as CR1 (complement receptor type 1), had affinity for C3u and iC3b. A protein of 145 000 Mr, recognized by the monoclonal antibody B2, had affinity for all three C3 fragments. Inhibition of rosetting by antibodies to these proteins indicates that CR1 is responsible for C3b-mediated rosetting and that the 145000-Mr receptor (CR2) is responsible for C3d-mediated rosetting. Partial inhibition by both anti-CR1 and anti-CR2 antibodies of iC3b-mediated rosetting indicates that both receptors are involved in iC3b-mediated rosetting. No other protein appears to be involved in tonsil B-cell rosetting to C3-fragment-coated cells. A method for preparing CR2 from tonsil lymphocytes based on affinity chromatography on C3d,g-Sepharose has been developed. Forty tonsil pairs (2 X 10(10) B-cells) yield about 40 micrograms of pure protein equivalent to a purification of 6500-fold from a detergent extract.     

Inhibition of factor I by diisopropylfluorophosphate. Evidence of conformational changes in factor I induced by C3b and additional studies on the specificity of factor I

The factor I-mediated cleavage of C3b, using factor H as a cofactor was completely inhibited by diisopropylfluorophosphate (DFP) when factor I and C3b were incubated with DFP before the addition of factor H. Inhibition, although to a lesser degree, was observed when factor H was present during DFP-exposure. No inhibition in factor I activity was seen when factor I and H were incubated with DFP either alone or together. It was also demonstrated that the 38-kDa subunit of factor I bound radiolabeled DFP when factor I and C3b together were exposed to DFP. These observations suggest that factor I interacts with C3b in a manner that exposes its catalytic site to DFP, an interaction that is independent of factor H. The inhibitory effect by DFP on factor I led us to further investigate the factor I cleavage products of iC3b, inasmuch as previous reports were ambiguous as to whether digestion occurs in the presence of DFP. Digestion of C3b bound to activated thiol Sepharose (ATS-C3b) in the presence of factor H at low pH and ionic strength and in serum by complement activation produced C3d,g-like fragments with apparent molecular mass of 41 and 43 kDa. These fragments were shown to have three different N-terminal and two different C-terminal ends. The major fragments had N-terminal sequences starting with Glu933, as shown by sequence determination. Traces of fragments extending beyond this point were also found, shown by Western blot analysis using a panel of mAb previously shown to bind to epitopes exposed within a region of C3 spanning residues 929 to 943, as well as a shorter fragment starting with Glu938. When digestion of C3b is carried out in the presence of DFP, the factor I level necessary for digestion is elevated and may explain how the first two cleavages producing iC3b but not the following giving C3d,g, can occur. The finding of several factor I cleavage sites in the C3d,g region of C3 demonstrates that factor I has a broad specificity, mainly for arginyl bonds. It has also been shown to digest a lysyl bond exposed in ATS-bound C3b.     

Isolation of an iC3b forming enzyme from peritoneal polymorphonuclear leukocytes of guinea pigs

We have investigated fluid phase cleavage of C3b by peritoneal polymorphonuclear leukocytes of guinea pigs and found that polymorphonuclear leukocytes expressed an iC3b forming enzyme as well as C3b receptor with maturation in peritoneal cavity. The iC3b forming enzyme was found to be distinct from C3bINA, a physiological iC3b forming enzyme in plasma, since the activity was inhibited by monoiodoacetic acid and did not require a cofactor plasma protein, beta 1H, for the cleavage of C3b into iC3b. The iC3b forming enzyme is gradually released upon incubation of PMN at 37 degrees C. The molecular weight of the iC3b forming enzyme was estimated to be 48,000 from gel filtration on Sephadex G-200.