Osteopontin is linked to p65 and MMP-9 expression in pulmonary adenocarcinoma but not in malignant pleural mesothelioma.

AIMS: Osteopontin (OPN) is a matricellular protein involved in tissue remodelling, cell-mediated immunity and malignant transformation. High OPN serum levels predict poor prognosis in non-small cell carcinoma and set patients with malignant pleural mesothelioma (MM) apart from disease-free asbestos-exposed individuals. Yet neither the spectrum of tissue expression nor the signalling pathways of OPN in MM and pulmonary adenocarcinoma have been characterized, although in vitro evidence links OPN to the epidermal growth factor receptor (EGFR) pathway. The aim of this study was to address these deficiencies. METHODS AND RESULTS: OPN expression was investigated immunohistochemically in 104 adenocarcinomas and 38 MM and correlated with histological features, including tumour type, grade and proliferation and with expression of activated intermediary EGFR signalling pathway molecules p65, p-AKT, p-ERK, p-STAT-3, and of metalloproteinase (MMP)-1, MMP-2 and MMP-9. In MM, OPN expression was widespread (36/38) and independent of the molecular parameters studied. In adenocarcinoma, high OPN expression was correlated with expression of p65, p-ERK and MMP-9. CONCLUSIONS: Frequent OPN expression is typical of, but not specific for MM, whereas it appears to select adenocarcinoma cases with p65 and MMP-9 expression, suggesting a link with EGFR signalling pathways.

Tumor-infiltrating macrophages induce apoptosis in activated CD8(+) T cells by a mechanism requiring cell contact and mediated by both the cell-associated form of TNF and nitric oxide.

We have investigated the ability of different cells present in murine tumors to induce apoptosis of activated CD8(+) T cells in vitro. Tumor cells do not induce apoptosis of T cells; however, macrophages that infiltrate tumors are potent inducers of apoptosis. Tumor macrophages express cell surface-associated TNF, TNF type I (CD120a) and II (CD120b) receptors, and, upon contact with T cells which induces release of IFN-gamma from T cells, secrete nitric oxide. Killing of T cells in vitro is blocked by Abs to IFN-gamma, TNF, CD120a, or CD120b, or N-methyl-L-arginine. In concert with that finding, tumor macrophages isolated from either TNF type I or type II receptor -/- mice are not proapoptotic and do not produce nitric oxide upon contact with activated T cells. Control macrophages do not express TNF receptors or release nitric oxide. Tumor cells or tumor-derived macrophages do not express FasL, and blocking Abs to either Fas or FasL have no effect on macrophage-mediated T cell killing. These results demonstrate that macrophages which infiltrate tumors are highly proapoptotic and may be responsible for elimination of activated antitumor T cells within the tumor bed.

CD8(+) tumor-infiltrating T cells are deficient in perforin-mediated cytolytic activity due to defective microtubule-organizing center mobilization and lytic granule exocytosis.

Tumor-infiltrating lymphocytes (TIL) are well known to be functionally impaired typified by the inability to lyse cognate tumor cells in vitro. We have investigated the basis for defective TIL lytic function in transplantable murine tumor models. CD8(+) TIL are nonlytic immediately on isolation even though they express surface activation markers, contain effector phase cytokine mRNAs, and contain perforin and granzyme B proteins which are packaged into lytic granules. Ag-specific lytic capability is rapidly recovered if purified TIL are briefly cultured in vitro and tumor lysis is perforin-, but not Fas ligand mediated. In response to TCR ligation of nonlytic TIL in vitro, proximal and distal signaling events are normal; calcium flux is rapid; mitogen-activated protein/extracellular signal-related kinase kinase, extracellular regulatory kinase 2, phosphoinositide-3 kinase, and protein kinase C are activated; and IL-2 and IFN-gamma is secreted. However, on conjugate formation between nonlytic TIL and cognate tumor cells in vitro, the microtubule-organizing center (MTOC) does not localize to the immunological synapse, thereby precluding exocytosis of preformed lytic granules and accounting for defective TIL lytic function. Recovery of TCR-mediated, activation-dependent MTOC mobilization and lytic activity requires proteasome function, implying the existence of an inhibitor of MTOC mobilization. Our findings show that the regulated release of TIL cytolytic granules is defective despite functional TCR-mediated signal transduction.

Protection against tick-transmitted Lyme disease in dogs vaccinated with a multiantigenic vaccine.

In an effort to develop a safe and effective vaccine for the prevention of Lyme borreliosis that addresses concerns raised over currently available vaccines, dogs were vaccinated twice with a multiantigenic preparation of Borrelia burgdorferi, strain N40, on days 0 and 20 of the experiment. About 70 and 154 days after the first immunization, dogs were challenged by exposing them to field-collected Ixodes scapularis ticks harboring B. burgdorferi. Vaccinated dogs were completely protected from infection by all criteria utilized to assess infection, developed high-titer anti-B. burgdorferi serum antibodies and growth inhibitory activity which persisted for over 200 days, and did not demonstrate any untoward consequence of vaccination. Serum absorption experiments revealed that borreliacidal and most likely protective antibodies in dogs receiving the multiantigenic preparation were not only elicited against the OspA antigen, but were also produced against additional yet to be determined targets on B. burgdorferi organisms. These data demonstrate that a multiantigenic vaccine is effective in preventing Lyme disease transmitted via the natural vector.

CD8+ tumor-infiltrating lymphocytes are primed for Fas-mediated activation-induced cell death but are not apoptotic in situ.

Induction of Fas-mediated activation-induced cell death in antitumor T cells has been hypothesized to permit tumor escape from immune destruction. Several laboratories have proposed that expression of Fas ligand (L) by tumor is the basis for this form of T cell tolerance. In this study, we characterized murine tumor-infiltrating lymphocytes (TIL) for activation status, cell cycle status, level of apoptosis, cytokine secretion, and proliferative capacity. TILs express multiple activation markers (circa CD69, CD95L, CD122, and LFA-1) and contain IL-2 and IFN-gamma mRNAs, but are neither cycling nor apoptotic in situ. In addition, TIL are dramatically suppressed in proliferative response and do not secrete IL-2 and IFN-gamma. However, upon purification and activation in vitro, TIL secrete high levels of IL-2 and IFN-gamma, enter S phase, and then die by Fas-mediated apoptosis. Activation by injection of anti-TCR Ab or IL-2 into tumor-bearing mice induced TIL entrance into S phase preceding apoptosis, showing that TIL have functional TCR-mediated signal transduction in situ. Our data demonstrate that TIL, not tumor, express both Fas and FasL, are arrested in G(1), do not secrete cytokine in situ, and, upon activation in vitro and in vivo, rapidly die by activation-induced cell death.

Distinct requirements for IFNs and STAT1 in NK cell function.

NK cell functions were examined in mice with a targeted mutation of the STAT1 gene, an essential mediator of IFN signaling. Mice deficient in STAT1 displayed impaired basal NK cytolytic activity in vitro and were unable to reject transplanted tumors in vivo, despite the presence of normal numbers of NK cells. IL-12 enhanced NK-mediated cytolysis, but poly(I:C) did not, and a similar phenotype occurred in mice lacking IFNalpha receptors. Molecules involved in activation and lytic function of NK cells (granzyme A, granzyme B, perforin, DAP10, and DAP12) were expressed at comparable levels in both wild-type and STAT1(-/-) mice, and serine esterase activity necessary for CTL function was normal, showing that the lytic machinery was intact. NK cells with normal cytolytic activity could be derived from STAT1(-/-) bone marrow progenitors in response to IL-15 in vitro, and enhanced NK lytic activity and normal levels of IFN-gamma were produced in response to IL-12 treatment in vivo. Despite these normal responses to cytokines, STAT1(-/-) mice could not reject the NK-sensitive tumor RMA-S, even following IL-12 treatment in vivo. Whereas in vitro NK cytolysis was also reduced in mice lacking both type I and type II IFN receptors, these mice resisted tumor challenge. These results demonstrate that both IFN-alpha and IFN-gamma are required to maintain NK cell function and define a STAT1-dependent but partially IFN-independent pathway required for NK-mediated antitumor activity.

Cancer-induced defective cytotoxic T lymphocyte effector function: another mechanism how antigenic tumors escape immune-mediated killing.

BACKGROUND: The notion that a deficit in immune cell functions permits tumor growth has received experimental support with the discovery of several different biochemical defects in T lymphocytes that infiltrate cancers. Decreased levels of enzymes involved with T-cell signal transduction have been reported by several laboratories, suggesting that tumors or host cells recruited to the tumor site actively down-regulate antitumor T-cell immune response. This permits tumor escape from immune-mediated killing. The possibility that defects in T-cell signal transduction can be reversed, which would potentially permit successful vaccination or adoptive immunotherapy, motivates renewed interest in the field. Summarizing the literature concerning tumor-induced T-cell dysfunction, we focus on the end stage of immune response to human cancer, that of defective cytotoxic T lymphocyte killing function. Based on the data from several laboratories, we hypothesize a biochemical mechanism that accounts for the unusual phenotype of antitumor T-cell accumulation in tumors, but with defective killing function.

Mice bearing late-stage tumors have normal functional systemic T cell responses in vitro and in vivo.

Immune suppression in tumor-bearing hosts is considered to be one factor causally associated with the growth of antigenic tumors. Support for this hypothesis has come from reports that spleen T cells in tumor-bearing mice are deficient in either priming or effector phase functions. We have reexamined this hypothesis in detail using multiple murine tumor models, including transplantable adenocarcinoma, melanoma, sarcoma, and thymoma, and also a transgenic model of spontaneous breast carcinoma. In both in vitro and in vivo assays of T cell function (proliferation, cytokine production, induction of CD8+ alloreactive CTL, and development of anti-keyhole limpet hemocyanin CD4+ T cells, rejection of allogeneic or syngeneic regressor tumors, respectively) we show that mice bearing sizable tumor burdens are not systemically suppressed and do not have diminished T cell functions. Therefore, if immune suppression is a causal function in the growth of antigenic tumor, the basis for escape from immune destruction is likely to be dependent upon tumor-induced T cell dysfunction at the site of tumor growth.

Type I IFN modulates innate and specific antiviral immunity.

IFNs protect from virus infection by inducing an antiviral state and by modulating the immune response. Using mice deficient in multiple aspects of IFN signaling, we found that type I and type II IFN play distinct although complementing roles in the resolution of influenza viral disease. Both types of IFN influenced the profile of cytokines produced by T lymphocytes, with a significant bias toward Th2 differentiation occurring in the absence of responsiveness to either IFN. However, although a Th1 bias produced through inhibition of Th2 differentiation by IFN-gamma was not required to resolve infection, loss of type I IFN responsiveness led to exacerbated disease pathology characterized by granulocytic pulmonary inflammatory infiltrates. Responsiveness to type I IFN did not influence the generation of virus-specific cytotoxic lymphocytes or the rate of viral clearance, but induction of IL-10 and IL-15 in infected lungs through a type I IFN-dependent pathway correlated with a protective response to virus. Combined loss of both IFN pathways led to a severely polarized proinflammatory immune response and exacerbated disease. These results reveal an unexpected role for type I IFN in coordinating the host response to viral infection and controlling inflammation in the absence of a direct effect on virus replication.

NKT cell cytokine imbalance in murine diabetes mellitus.

A minor subset of murine MHC class I-restricted T cells which express both the alphabeta form of the T cell receptor and a NK lineage marker, termed NKT cells, is capable of secreting significant amounts of Interleukin-4 and Interferon-y upon activation. As such NKT cells may play a role in development of Th1 and Th2 cells during T cell ontogeny or expansion of T cells expressing a dominant cytokine pattern in the effector phase. We have studied the role of NKT cells in a murine model of disease multidose streptozotocin induced diabetes mellitus (MDSDM). In MDSDM thymic and splenic NKT cells are present at normal levels but have greatly reduced capacity to secrete Interleukin-4 upon stimulation with anti-TCR antibody compared to control mice; conversely, Interferon-y secretion is maintained. By analysis of cytokine RNA production we found that treatment of several strains of mice with streptozotocin changes the peripheral helper T cell phenotype elicited after immunization with Keyhole Limpet Hemocyanin from a mixed Th1- and Th2-type cytokine pattern (characterized by IFN-gamma and IL-4 and IL-5 expressions, respectively) to predominately Th1-type. Furthermore, susceptibility to MDSDM is significantly enhanced when NKT cells are selectively eliminated in vivo by administration of depleting anti-CD122 antibody TMbeta-1. In addition, antibody depletion of NKT cells from non-obese diabetic mice significantly accelerates onset of disease. Collectively these data support a model for development of murine diabetes mellitus in which NKT cell cytokine expression influences the development of Th1-type diabetogenic T cells.

Protection from Lyme neuroborreliosis in nonhuman primates with a multiantigenic vaccine.

In an effort to develop an effective and safe vaccine for lyme disease, rhesus macaques were injected with a multiantigenic preparation of Borrelia burgdorferi, strain N40. One month later animals were boosted before intradermal challenge with infectious B. burgdorferi. Challenges were performed at 1 and again at 5 months after the booster vaccination. Vaccinated and control nonvaccinated animals were monitored for development of systemic infection by measurement of serum anti-spirochetal antibodies by ELISA and Western blotting, and neurological involvement was monitored by testing of cerebrospinal fluid (CSF) and PCR analysis of central nervous system (CNS) tissue obtained at necropsy. Two control nonhuman primates (NHPs), given saline injections instead of vaccine and then challenged with B. burgdorferi, developed CSF pleocytosis, PCR positivity of the brain, and high levels of specific anti-B. burgdorferi antibody in the serum and CSF. Necropsy studies revealed widespread invasion of the CNS of one of these animals by the spirochete. In contrast, none of the four vaccinated animals developed evidence of invasion of the CNS after either of two challenge inoculations with infectious B. burgdorferi. In addition to resisting infection, no vaccinated animal demonstrated any untoward consequence of vaccination. These data demonstrate that a multiantigenic vaccine is effective in preventing systemic infection and lyme neuroboreliosis in NHPs and suggest that a successful vaccine could be developed in humans which would prevent lyme disease.

Multidose streptozotocin induction of diabetes in BALB/cBy mice induces a T cell proliferation defect in thymocytes which is reversible by interleukin-4.

Thymic T cell function in streptozotocin-treated (STZ) diabetic mice has been examined. STZ administration suppresses thymic T cell proliferation in response to mitogen stimulation in vitro. Secretion of IL-4 was dramatically reduced; however, secretion of IL-2 or IFN-gamma was not significantly inhibited. RT-PCR analysis of thymocyte RNA revealed that levels of IL-4 mRNA were dramatically decreased in STZ-treated mice. Levels of mRNA encoding IFN-gamma were similar, but the appearance was delayed in thymocytes derived from STZ-treated mice, implying differential regulation of IL-4 and IFN-gamma. Defective thymocyte proliferation was partially restored by exposure to IL-2 in vitro; however, IL-4 completely reversed the STZ-induced defect. Administration in vivo of IL-4 before STZ treatment reversed the STZ-induced thymocyte proliferation defect and prevented both pancreatic islet destruction and hyperglycemia. Thymocyte cell surface differentiation markers were not appreciably different from control mice. Collectively these experiments suggest that STZ treatment of mice reduces expression of IL-4 which is associated with development of autoimmune diabetes.

Protection of human breast cancer cells from complement-mediated lysis by expression of heterologous CD59.

CD59, decay accelerating factor (DAF) and membrane cofactor protein (MCP) are widely expressed cell surface glycoproteins that protect host cells from the effects of homologous complement attack. Complement inhibitory activity of these proteins is species-selective. We show that the human breast cancer cell line MCF7 is relatively resistant to lysis by human complement, but is effectively lysed by rat or mouse complement. CD59, DAF and MCP were all shown to be expressed by MCF7. The species-selective nature of CD59 activity was used to demonstrate directly the effectiveness of CD59 at protecting cancer cells from complement-mediated lysis. cDNAs encoding rat and mouse CD59 were separately transfected into MCF7 cells, and cell populations expressing high levels of the rodent CD59 were isolated by cell sorting. Data show that rat and mouse CD59 were highly effective at protecting transfected MCF7 cells from lysis by rat and mouse complement, respectively. Data further reveal that rat CD59 is not effective against mouse complement, whereas mouse CD59 is effective against both mouse and rat complement. These studies establish a model system for relevant in vivo studies aimed at determining the effect of complement regulation on tumourigenesis, and show that for effective immunotherapy using complement-activating anti-tumour antibodies, the neutralization of CD59 and/or other complement inhibitory molecules will probably be required.

Rat NKR-P1+ CD3+ T cells: selective proliferation in interleukin-2, diverse T-cell-receptor-Vbeta repertoire and polarized interferon-gamma expression.

Cells expressing markers of both natural killer and T lymphocytes (NK T cells) in humans and mice express a restricted T-cell receptor (TCR) repertoire, are of CD4- CD8- or CD4+ CD8- phenotype, and upon anti-CD3 stimulation secrete large amounts of interleukin-4 (IL-4) and interferon-gamma (IFN-gamma). NK T cells may be the primary source of IL-4-promoting T helper type 2 (Th2) responses and/or they might be involved in regulating the balance between Th1- and Th2-type immune responses, and may consequently affect susceptibility to autoimmune diseases associated with a skewed Th phenotype. We show that rat NK T cells selectively proliferate to IL-2, and use this fact to analyse cytokine production by NK T cells in two rat strains differentially susceptible to Th1- or Th2-type autoimmune diseases. Analysis by reverse transcription-polymerase chain reaction revealed that, in contrast to mouse, rat NK T cells secrete exclusively IFN-gamma and not IL-4 after anti-CD3 stimulation, and use a wider TCR-Vbeta repertoire, suggesting that rat NK T cells are not essential for the development of Th2-type CD4+ T-cell responses.

Soluble proteins isolated from Borrelia burgdorferi by extraction with Triton X-114 confer resistance to experimental infection.

Fractionation of Borrelia burgdorferi was made by extraction of infectious spirochetes using the detergent Triton X-114. Gel electrophoresis analysis of hydrophilic and hydrophobic proteins demonstrated that detergent extraction resulted in two populations of proteins with nonoverlapping electrophoretic profiles. Immunoblot analysis with monoclonal antibodies reactive with two abundant membrane proteins demonstrated that hydrophilic proteins were uncontaminated with hydrophobic proteins. In addition, assay of thymidine incorporation into and secretion of tumor necrosis factor-alpha from splenocytes cocultured in vitro with either detergent or aqueous phase proteins showed that lymphocyte mitogenic and macrophage activation activities of B. burgdorferi were completely absent from the hydrophilic phase proteins. The Triton X-114 aqueous and detergent phase proteins were used to immunize BALB/c and separately microMT/microMT (B cell knockout) mice that were subsequently challenged with infectious B. burgdorferi. The hydrophilic phase proteins were able to induce protective resistance to infection in either strain of mice demonstrating that potential candidate vaccine antigens are contained in the biochemical class of antigens which is devoid of both lymphocyte mitogen activity and major outer surface proteins. Furthermore, the ability to vaccinate B cell knockout mice suggests that the humoral antispirochete immune response is not the exclusive basis for protective immunity.

Administration of silica sensitizes lipopolysaccharide responsiveness of murine macrophages but inhibits T and B cell priming by inhibition of antigen presenting function.

Macrophages play a key role in natural host defense against infection by a variety of pathogens. In addition, macrophages initiate the development of acquired immunity via antigen processing and presentation. The role of macrophages in resistance to pathogens, the development of autoimmune diseases and the induction of acquired immunity has been studied by treatment of rodents with reagents which are cytotoxic. We have studied the effects of one such reagent, silica, on the function of spleen macrophages and peritoneal exudate cells (PEC). Intraperitoneal administration of silica caused the accumulation of spleen macrophages and neutrophils, reduction in the number of B cells and had a modest effect on T cell abundance. The percentage of CD11b+ PEC was not affected by silica treatment but total PEC recovery was diminished 5-8 fold. Silica treatment did not cause release of TNF-alpha or IL-1-beta but, when stimulated with lipopolysaccharide (LPS) in vitro after silica treatment, PEC or spleen macrophages produced elevated levels of both cytokines compared to controls. In contrast, release of IL-12 from non-LPS treated PEC was stimulated 4-5 fold by silica treatment. In addition, sensitivity to LPS toxicity in vivo was significantly enhanced by silica. The ability of macrophages to present antigen to a T cell clone in vitro was found to be dramatically inhibited by silica treatment, as was the ability to prime antigen-specific T cells and B cells by antigen injection. Collectively these data demonstrate that silica treatment enhances macrophage sensitivity to LPS exposure but inhibits antigen processing and presentation.

Repression of interleukin-2 mRNA translation in primary human breast carcinoma tumor-infiltrating lymphocytes.

Human breast carcinoma tumor-infiltrating lymphocytes (TIL) express activation antigens in situ indicative of ongoing immune response-CD28, CD45RO, CD69, CD71, and DR. However, interleukin 2 (IL-2) receptor was poorly expressed: CD25 was detected in only 1/24 samples and CD122 in only 2/24 samples. Furthermore, isolated breast cancer TIL were defective in proliferative response but recover when treated with recombinant IL-2. Nineteen of 24 tumor samples expressed B7-1, B7-2, and CD28 protein, showing that absence of costimulator proteins or counter ligand was not the basis for TIL proliferative deficit. Expression of IL-2 activity was not detected; however, mRNA encoding IL-2 was produced and translatable in vitro. These findings show that human breast cancer tumor-induced repression of IL-2 RNA translation is the basis of failure of TIL to express the IL-2 receptor and subsequent T cell hyporesponsiveness.

Study of immune response to tumors in the rat.

Use of the rat as host for the study of cancer has become popular for several reasons. The larger body size compared to mice is especially convenient for lines of experiments involving surgical manipulation, transplantation, or biochemical purification of molecules of interest. Immune response to cancer is also studied in rat models, and this article focuses on the methodological aspects of in vivo and in vitro protocols related to rat tumor immunology.

Killing of rat adenocarcinoma 13762 in situ by adoptive transfer of CD4+ anti-tumor T cells requires tumor expression of cell surface MHC class II molecules.

CD4+ anti-tumor T cells reactive with rat adenocarcinoma 13762 kill tumor in vitro and cause regression of tumor in vivo. The role of various host immune cells in CD4+ T-cell-mediated tumor elimination in vivo was investigated by adoptive transfer of anti-tumor T cell clones to recipients that were selectively depleted of individual immune cell types. By these means, macrophages and NK cells were found to be required for tumor killing. Depletion of host CD4+ T cells, CD8+ T cells, or neutrophils was without effect on tumor elimination by anti-tumor T cells. An essential role for antigen receptor-negative NK cells is likely dependent upon secretion of IFN-gamma from NK cells since treatment of tumor recipients with anti-IFN-gamma antibody prior to adoptive transfer and tumor challenge abrogated T cell killing, resulting in progressive tumor growth. Viability of adenocarcinoma 13762 or anti-tumor T cells was unaffected by treatment with either IFN-gamma or anti-IFN-gamma antibody in vitro, but cell surface MHC class II expression was induced in tumor cells by exposure to IFN-gamma. In addition, tumor cells were isolated from tumor-bearing animals by absorption using anti-MHC class II antibody, demonstrating that 13762 tumor expresses cell surface MHC class II antigens in situ. However, if hosts were depleted of NK cells before tumor challenge, MHC class II+ tumor was not recovered. Collectively these results suggest that adenocarcinoma 13762 is eliminated by MHC class II-restricted CD4+ T cells by direct tumor killing.

Analysis of the structure and function of the von Willebrand factor A1 domain using targeted deletions and alanine-scanning mutagenesis.

von Willebrand factor (vWF) mediates the primary adhesion of platelets to sites of vascular damage through interaction with glycoprotein Ib (GPIb) of the platelet GPIb/IX complex. To investigate the vWF/GPIb interaction we introduced both in-frame deletions and substitutions into the vWF A1 domain. The introduction of nine sequential 20-amino acid deletions within the Cys509-Cys695 loop of the A1 domain caused the defective secretion of vWF from mammalian cells, and resulted in multimeric vWF without platelet-binding activity. In other experiments we substituted alanine for charged amino acids (residues 524, 534, 549, 552, 569-573, and 642-645) in proposed functional domains within the Cys509-Cys695 loop. All six substitution mutants showed normal secretion from transfected mammalian cells and bound to fixed platelets in the presence of botrocetin. In contrast, only mutants vWF-R524A and vWF-K549A showed significant binding to platelets in the presence of ristocetin. Mutant vWF-K549A showed increased platelet-binding at suboptimal concentrations of both botrocetin and ristocetin. These results suggest that the substituted amino acids do not play a critical role in the activation of vWF by botrocetin or in the direct interaction of vWF with the GPIb/IX complex. However, the charged amino acids at positions 534, 552, 569-573, and 642-645 do play an important role in the ristocetin-induced binding of vWF to platelets. The interaction of vWF with heparin was significantly reduced by substitution of Lys residues 642-645, indicating that these residues may form part of a heparin-binding domain in the carboxy-terminal half of the Cys509-Cys695 loop.