Mutation of hydrophobic residues in the factor Va C1 and C2 domains blocks membrane-dependent prothrombin activation.

The binding of factor (FVa) to phosphatidylserine (PS) membranes regulates assembly of the prothrombinase complex. Two pairs of solvent-exposed amino acids, Tyr(1956)/Leu(1957) in the C1 domain and Trp(2063)/Trp(2064) in the C2 domain, each make significant contributions to the affinity of FVa for PS membranes, but individually neither pair of amino acids is required for prothrombinase assembly on 25% PS membranes. In this study we characterize a FVa mutant with alanine substitutions in both the C1 and C2 domains: (Y1956,L1957,W2063,W2064)A. We conclude that: (i) prothrombinase assembly on PS membranes requires Trp(2063)/Trp(2064) and/or Tyr(1956)/Leu(1957); (ii) combined mutation of Trp(2063)/Trp(2064) and Tyr(1956)/Leu(1957) results in only a modest 4-fold decrease in the rate of thrombin generation in the absence of membranes; (iii) the present data provide experimental support for the joint participation of the C1 and C2 domains in the binding of FVa to phospholipid membranes as suggested by the recently solved structure for FVai (A1/A3-C1-C2).

Soluble phosphatidylserine binds to a single identified site in the C2 domain of human factor Va.

Factor V(a) (FV(a)) is a cofactor for the serine protease factor X(a) that activates prothrombin to thrombin in the presence of Ca(2+) and a membrane surface. FV(a) is a heterodimer composed of one heavy chain (A1 and A2 domains) and one light chain (A3, C1, and C2 domains). We use fluorescence, circular dichroism, and equilibrium dialysis to demonstrate that (1) the FV C2 domain expressed in Sf9 cells binds one molecule of C6PS with a k(d) of approximately 2 microM, (2) stabilizing changes occur in the FV C2 domain upon C6PS binding, (3) the C6PS binding site in the FV C2 domain is located near residue Cys(2113), which reacts with DTNB, and (4) binding to a PS-containing membrane is an order of magnitude tighter than that to soluble C6PS. Coupled with a recently published crystal structure of the C2 domain, these results support a model for the mechanism of C2-membrane interaction.

Identification of functionally important amino acid residues within the C2-domain of human factor V using alanine-scanning mutagenesis.

We have previously determined that the C2-domain of human factor V (residues 2037-2196) is required for expression of cofactor activity and binding to phosphatidylserine (PS)-containing membranes. Naturally occurring factor V inhibitors and a monoclonal antibody (HV-1) recognized epitopes in the amino terminus of the C2-domain (residues 2037-2087) and blocked PS binding. We have now investigated the function of individual amino acids within the C2-domain using charge to alanine mutagenesis. Charged residues located within the C2-domain were changed to alanine in clusters of 1-3 mutations per construct. In addition, mutants W2063A, W2064A, (W2063, W2064)A, and L2116A were constructed as well. The resultant 30 mutants were expressed in COS cells using a B-domain deleted factor V construct (rHFV des B). All mutants were expressed efficiently based on the polyclonal antibody ELISA. The charged residues, Arg(2074), Asp(2098), Arg(2171), Arg(2174), and Glu(2189) are required for maintaining the structural integrity of the C2-domain of factor V. Four of these residues (Arg(2074), Asp(2098), Arg(2171), and Arg(2174)) correspond to positions in the factor VIII C-type domains that have been identified as point mutations in patients with hemophilia A. The epitope for the inhibitory monoclonal antibody HV-1 has been localized to Lys(2060) through Glu(2069) in the factor V C2-domain. The epitope for the inhibitory monoclonal antibody 6A5 is composed of amino acids His(2128) through Lys(2137). The PS-binding site in the factor V C2-domain includes amino acid residues Trp(2063) and Trp(2064). This site overlaps with the epitope for monoclonal antibody HV-1. These factor V C2-domain mutants should provide valuable tools for further defining the molecular interactions responsible for factor V binding to phospholipid membranes.

Crystal structures of the membrane-binding C2 domain of human coagulation factor V.

Rapid and controlled clot formation is achieved through sequential activation of circulating serine proteinase precursors on phosphatidylserine-rich procoagulant membranes of activated platelets and endothelial cells. The homologous complexes Xase and prothrombinase, each consisting of an active proteinase and a non-enzymatic cofactor, perform critical steps within this coagulation cascade. The activated cofactors VIIIa and Va, highly specific for their cognate proteinases, are each derived from precursors with the same A1-A2-B-A3-C1-C2 architecture. Membrane binding is mediated by the C2 domains of both cofactors. Here we report two crystal structures of the C2 domain of human factor Va. The conserved beta-barrel framework provides a scaffold for three protruding loops, one of which adopts markedly different conformations in the two crystal forms. We propose a mechanism of calcium-independent, stereospecific binding of factors Va and VIIIa to phospholipid membranes, on the basis of (1) immersion of hydrophobic residues at the apices of these loops in the apolar membrane core; (2) specific interactions with phosphatidylserine head groups in the groove enclosed by these loops; and (3) favourable electrostatic contacts of basic side chains with negatively charged membrane phosphate groups.

Partial glycosylation at asparagine-2181 of the second C-type domain of human factor V modulates assembly of the prothrombinase complex.

Thrombin-activated factor Va exists as two isoforms, factor Va(1) and factor Va(2), which differ in the size of their light chains and their affinity for biological membranes. The heterogeneity of the light chain remained following incubation of factor Va with N-glycanase. However, we found that the factor V C2 domain, which contains a single potential glycosylation site at Asn-2181, was partially glycosylated when expressed in COS cells. To confirm the structural basis for factor Va(1) and factor Va(2), we mutated Asn-2181 to glutamine (N2181Q) and expressed this mutant using a B domain deletion construct (rHFV des B) in COS cells. Thrombin activation of N2181Q released a light chain with mobility identical to that of factor Va(2) on SDS-PAGE. The functional properties of purified N2181Q were similar to those of factor Va(2) in prothrombinase assays carried out in the presence of limiting concentrations of phosphatidylserine. The binding of human factor Va(1) and factor Va(2) to 75:25 POPC/POPS vesicles was also investigated in equilibrium binding assays using proteins containing a fluorescein-labeled heavy chain. The affinity of human factor Va(2) binding to POPC/POPS vesicles was approximately 3-fold higher than that of factor Va(1). These results indicate that partial glycosylation of factor V at asparagine-2181 is the structural basis of the light chain doublet and that the presence of this oligosaccharide reduces the affinity of factor Va for biological membranes.

Inhibitory anti-factor V antibodies bind to the factor V C2 domain and are associated with hemorrhagic manifestations.

Factor V inhibitors may develop as spontaneous autoantibodies, as alloantibodies after exposure to bovine thrombin preparations, or in factor V-deficient patients after plasma therapy. Clinical manifestations range from asymptomatic laboratory abnormalities to life-threatening hemorrhage. We have characterized the anti-factor V antibodies from 12 patients diagnosed with factor V inhibitors. In 8 patients, hemorrhagic complications (5 autoantibodies and 3 bovine thrombin-induced alloantibodies) developed, and 4 were asymptomatic (2 autoantibodies and 2 alloantibodies). The IgG fractions from all 12 patients immunoprecipitated the factor Va light chain, but only the 8 IgG fractions associated with hemorrhage inhibited factor V activity in a prothrombinase assay. Nine IgG fractions, including the 8 patients with hemorrhage, immunoprecipitated the isolated second C-type domain (C2). The 8 IgG fractions from the symptomatic patients also immunoprecipitated recombinant chimeras containing only the N-terminal third of the factor V C2 domain, and isolated recombinant C2 domain abrogated the inhibitory effect of the antibodies. Five of the inhibitory IgG fractions blocked binding of factor V to phosphatidylserine. These results suggest that inhibitory anti-factor V antibodies are associated with hemorrhagic manifestations and frequently bind to a common region within the C2 domain, whether originating spontaneously or after exposure to bovine thrombin.

Recognition and lysis of human melanoma by a CD3+, CD4+, CD8- T-cell clone restricted by HLA-A2.

The in vitro cytotoxic response to human melanoma is characterized by CD3+ CD8+ T-cells which recognize shared peptide antigens presented in the context of HLA class-I-encoded gene products. We report here studies of a CD3+, CD4+, CD8-, HLA-A2-restricted, melanoma-specific cytotoxic T-cell clone derived by limiting dilution from a T-cell line induced in PBLs from a melanoma patient following in vitro stimulation with an HLA-A2-matched melanoma cell line. The CD4+ cytotoxic T-cell clone is lytic only for melanomas which share the HLA-A2 allele, and the cytotoxicity is blocked by antibody to the T-cell receptor and by antibody to HLA class I. The clone proliferates only following stimulation with HLA-A2-matched melanoma tumor cells. The data suggest that cytotoxic CD4+ T-cells may play a significant role in immunity to melanoma, and HLA class-I-restricted recognition of melanoma may not necessarily require the CD8 molecule on the lytic T-cell.

Thrombin-catalyzed activation of recombinant human factor V.

Proteolytic activation of human factor V by thrombin results from the cleavage of three peptide bonds at Arg709, Arg1018, and Arg1545. In order to define the functional importance of these sites, mutants with isoleucine substitutions blocking thrombin cleavage at one, two, or all three activation sites were expressed in COS-7 cells. The wild type protein is activated approximately 10-fold by thrombin or Russell's viper venom (RVV-V). Thrombin cleavage at Arg709 alone did not result in an increase in procoagulant activity. Cleavage at both Arg709 and Arg1018 resulted in an approximately 3.4-fold increase in activity. Cleavage at these sites was required for rapid cleavage by thrombin at Arg1545, however, which resulted in maximal activation of the factor V molecule. In contrast, isolated cleavage at Arg1545 by RVV-V was sufficient for efficient and complete activation of factor V. The effect of isoleucine substitutions at one or both thrombin cleavage sites in a B-domain deletion mutant lacking amino acids 811-1491 was also investigated. The specific activity of all four mutants was approximately 30% compared to thrombin activated factor V, indicating that these isoleucine substitutions do not drastically alter the structure of the protein and that cleavage at these sites is not required for the expression of partial procoagulant activity.

Localization of functionally important epitopes within the second C-type domain of coagulation factor V using recombinant chimeras.

Coagulation factor V, an integral component of the prothrombinase complex, possesses two C-type domains at the carboxyl-terminal end of the molecule. Homologous C-type domains are present in factor VIII as well as several non-coagulation proteins. Deletion of the second C-type domain of factor V results in the loss of procoagulant activity and the ability to bind phosphatidylserine. We now report the effect of substitution of all or a portion of the C2 domain of factor V with the corresponding regions of factor VIII or the human breast carcinoma protein BA46. Substitution of the entire domain with a heterologous C2 domain does not restore significant procoagulant activity, although smaller, exon-size substitutions do result in chimeras with partial activity (approximately 10% of factor Va). Using chimeras with partial substitutions, we determined that the amino-terminal region of the domain is involved in binding to phosphatidylserine. In contrast, the central region of the domain is not involved in phosphatidylserine binding, but an antibody binding at or near this site inhibits procoagulant activity, suggesting that this region is involved in a separate function. Lastly, the molecular basis for the light chain doublet, which is important in the expression of full procoagulant activity, is located within the carboxyl-terminal region of the C2 domain.

Characterization of an acquired inhibitor to coagulation factor V. Antibody binding to the second C-type domain of factor V inhibits the binding of factor V to phosphatidylserine and neutralizes procoagulant activity.

Coagulation Factor V is an essential component of the prothrombinase complex, which activates the zymogen prothrombin to thrombin. A patient was described who developed a Factor V inhibitor that neutralized the procoagulant activity of Factor V and resulted in a fatal hemorrhagic diathesis (Coots, M. C., A. F. Muhleman, and H. I. Glueck. 1978. Am. J. Hematol. 4:193-206). This inhibitor was shown to be an IgG antibody that bound to the light chain of Factor V. Using a series of light chain deletion mutants, we have found that this antibody binds to the second C-type domain of the light chain. Both inhibitor IgG and Fab fragments rapidly neutralized the procoagulant activity of Factor Va, implying that the neutralization resulted from specific binding to the C2 domain. We have previously demonstrated that deletion of the C2 domain results in loss of procoagulant activity, as well as loss of phosphatidylserine-specific binding. Confirming these results, both inhibitor IgG and Fab fragments interfered with phosphatidylserine-specific binding of Factor V. Conversely, preincubation of Factor Va with procoagulant phospholipids protected the cofactor from inactivation by the inhibitor. Our results suggest that this inhibitor neutralizes the procoagulant activity of Factor Va by interfering with the C2-mediated interaction with phospholipid surfaces, thereby disrupting formation of the prothrombinase complex.

Human melanoma-mediated inhibition of autologous CD4+ helper tumor-infiltrating lymphocyte growth in vitro.

Tumor-infiltrating lymphocytes (TIL) were isolated from a human melanoma metastatic to the abdomen. The TIL were 99% CD3+ and 99% CD4+ and CD8-. They were dependent on interleukin-2 (IL-2) for growth, as measured in a thymidine uptake assay, and were not cytotoxic to autologous or allogeneic melanoma or K562. When co-cultured with irradiated autologous tumor cells, or tumor cell supernatants, the TIL not only did not respond, but the IL-2-dependent growth was inhibited significantly. Inhibition occurred during the first 24 hours of co-culture and persisted as long as the tumor was present. After being washed free of inhibitory tumor cells, the TIL again were able to grow in the presence of IL-2, indicating that the inhibition was not caused by irreversible toxicity mediated by the tumor. Addition of excess IL-2 did not reverse the inhibitory effect, but addition of indomethacin, an inhibitor of cyclooxygenase and prostaglandin synthesis, partially blocked the inhibition. These data show melanoma-mediated inhibition of induction and expansion of human T-cells in vitro, which may reflect one of the mechanisms of inhibition of cellular responses in vivo. These results stress the need to examine the techniques for optimal in vitro expansion of tumor-specific TIL or cytotoxic T-cells for adoptive immunotherapy.

MHC-restricted recognition of autologous melanoma by tumor-specific cytotoxic T cells. Evidence for restriction by a dominant HLA-A allele.

Autologous melanoma-specific CTL recognize a common tumor-associated Ag (TAA) in the context of HLA class I antigens. We have demonstrated that HLA-A2 can be a restricting Ag and, in T cell lines homozygous for HLA-A2, that CTL can be generated by stimulation with HLA-A2 allogeneic melanomas. In the current study, we have investigated T cell lines from patients who are heterozygous at HLA-A region locus, to determine the relative importance of each A-region allele in this MHC-restricted recognition of tumor. We have shown that HLA-A1 can be a restricting Ag, and that allogeneic melanomas expressing HLA-A1 can substitute for the autologous tumor in the generation of HLA-A1-restricted CTL. However, when T cell lines express both HLA-A1 and HLA-A2, the HLA-A2 allele governed restriction of the melanoma TAA. Three autologous-stimulated HLA-A1, A2 CTL lines all demonstrated restriction by the HLA-A2 allele, when examined in cytotoxicity assays, cold-competition assays, and proliferation assays. There was no evidence of restriction by the second HLA-allele, HLA-A1. Although the autologous-stimulated CTL use a single A-region allele for tumor recognition, the autologous HLA-A1, A2 tumors are lysed by both HLA-A1-restricted and HLA-A2-restricted CTL. The dominance of restricting alleles was further demonstrated when HLA-matched allogeneic melanomas were used as the stimulating tumor to generate tumor-specific CTL. Stimulation of the heterozygous (HLA-A1, A2) lymphocytes with HLA-A2-matched allogeneic melanomas resulted in CTL specific for the autologous tumor, and restricted by the HLA-A2 Ag. However, stimulation with an HLA-A1-matched allogeneic melanoma failed to induce tumor-specific CTL restricted by the HLA-A1 Ag. The data suggest there is a dominance of HLA-A region Ag at the level of the T cell, such that only one is restricting in the recognition of the autologous melanoma. At the level of the tumor, however, the TAA is expressed in the context of both HLA-A region alleles. We can generate specific CTL from lymph node cells or PBL and HLA-A region matched allogeneic melanomas; however, because most patients are heterozygous at the HLA-A region locus, an understanding of the dominant restricting alleles must be obtained so that an appropriately matched allogeneic melanoma can be selected.

Modulation of in vitro autologous melanoma-specific cytotoxic T-cell responses by phorbol dibutyrate and ionomycin.

Human melanoma-specific, HLA restricted, cytotoxic T-cell lines can be generated by in vitro stimulation and culturing of peripheral lymphocytes, or lymph node cells, with autologous or HLA-A region matched melanomas in the presence of a low concentration (5 U/ml) of IL-2. Stimulation is followed by a period of clonal expansion and differentiation into cytotoxic T-cells specific for melanoma. We investigated the effect of the PKC modulating drug phorbol dibutyrate combined with the calcium ionophore Ionomycin on growth and differentiation of the cell lines. The growth of the T-cell lines was substantially augmented in the presence of the drugs with increases of 10-fold or more in clonal expansion by 3 weeks of culture. The cell lines were IL-2 dependent for growth in the presence or absence of the drugs and the phenotypic distribution remained predominantly CD3+ T-cells of mixed CD4 and CD8 phenotypes. In spite of the increased rate of growth in the presence of the drugs, autologous melanoma-specific cytotoxicity was almost completely abrogated in those cultures. The cells were, however, nonspecifically lytic in the presence of concanavalin A. The melanoma-specific cytotoxic response was completely restored following culture with IL-2 alone. The results suggest that the human tumor-specific cytotoxic T-cell response can be induced and amplified in the presence of immune modulating drugs.

Human cytotoxic T cells specific for autologous melanoma cells: successful generation from lymph node cells in seven consecutive cases.

Human T-cell populations specifically cytotoxic for autologous melanoma cells have been successfully generated from lymph node cells obtained from seven consecutive patients. The lymph node cells were stimulated in vitro with autologous irradiated melanoma cells; stimulation was repeated every 10-15 days at a tumor cell-to-lymphocyte ratio of approximately 1:20. Cytotoxic activity was assessed by a 4-hour 51Cr release assay. Mean lysis of autologous tumor cells was 47% at an effector-to-target cell ratio of 20:1, while mean lyses of the human myeloid leukemia cell line K562, allogeneic melanoma cells, and an osteosarcoma cell were 20%, 13%, and 11%, respectively. There was no lysis of autologous fibroblasts, fresh lymphocytes, or phytohemagglutinin-stimulated blasts. Three grades of specificity developed sequentially. In grade I, lysis of autologous tumor cells exceeded lysis of allogeneic tumor cells but did not exceed lysis of K562 cells. In grade II, lysis of autologous tumor cells exceeded lysis of K562 cells and all allogeneic tumor cells tested. In grade III, potent lysis of autologous tumor cells (greater than 40%) exceeded lysis of K562 cells and of all allogeneic tumor cells tested. All seven lymphocyte populations reached or exceeded grade I. Six reached or exceeded grade II. Two progressed to grade III. The generated cells were T cells, as determined by phenotypic analysis with flow cytometry. CD4+ cells and CD8+ cells accounted for 83%-100% of the cells. CD8+ T cells were separated from CD4+ T cells by panning with OKT8 and OKT4 antibodies. The resulting CD8-enriched and CD4-enriched populations were compared as effectors in cytotoxicity assays. The results suggest that the cell responsible for lysis of autologous tumor cells is CD8+. The methods used in this study have repeatedly resulted in the successful generation of cytotoxic T lymphocytes specifically cytotoxic for autologous melanoma cells; it is suggested that these cells have potential application for adoptive immunotherapy of melanoma.