A hierarchy of disulfide-bonded subunits: the quaternary structure of Eudistylia chlorocruorin.

The quaternary structure of the cysteine-rich, approximately 3500-kDa chlorocruorin (Chl) from the marine polychaete Eudistylia vancouverii was investigated using maximum entropy deconvolution of the electrospray ionization mass spectra (ESIMS). The native Chl provided two groups of peaks, at approximately 25 and approximately 33 kDa, and one peak at approximately 66 kDa. ESIMS of the reduced and reduced and carbamidomethylated Chl and of its subunits obtained by HPLC provided the complete subunit composition of the Chl. Two groups of nonglobin linker chains were observed: L1a-f (25 000.4, 25 017.9, 25 039.6, 25 057.0, 25 074.4 and 25 096.8 Da) and L2a-d (25 402.7, 25 446.0, 25 461.6 and 25 478.3Da) (+/-2.5 Da), with relative intensities L1:L2 = 5:2. Six globin chains were found, a1, a2, and b1-4, with reduced masses of 16 051.5, 16 172.4, 16 853.5, 17 088.9, 17 161.2 and 17 103.6 (+/-1.0 Da) and relative intensities of 8:4:1:4:2:1, respectively. Disulfide-bonded dimers and a tetramer of globin chains were identified: D1 = a1 + b3 at 33 207.1; D2 at 33 374.1, which had a cysteinylated Cys (a2 + b2 + Cys); and D3 = a1 + b4 at 33 149.4 Da (+/-3.0 Da), with relative intensities D1:D2:D3 = 5:4:1 and T = a1 + a2 + b1 + b2 at 66 154.8 +/- 4.0 Da. A 206-kDa dodecamer subunit obtained by dissociation of the Chl in 4 M urea [Qabar, A. N., et al. (1991) J. Mol. Biol. 222, 1109-1129], was found to consist only of tetramers T. A model was proposed for the Chl, based on a dimer:tetramer ratio of 2:1: four 206-kDa dodecamers (trimer of tetramers) and 48 dimers tethered to a framework of 30 L1 and 12 L2 linker chains. The 144 globin chains (2480 kDa) and 42 linker chains (1059 kDa) provide a total mass of 3539 kDa, in good agreement with the 3480 +/- 225 kDa determined previously by STEM mass mapping. The hierarchy of disulfide-bonded globin subunits observed for Eudistylia Chl provides a built-in heterogeneity of hexagonal bilayer structures.

Electrospray ionization mass spectrometric study of the multiple intracellular monomeric and polymeric hemoglobins of Glycera dibranchiata.

The intracellular hemoglobin (Hb) of the marine polychaete Glycera dibranchiata is comprised of two groups of globins differing in their primary structures and state of aggregation. About six electrophoretically and chromatographically distinct monomeric Hbs which have Leu as the distal residue, and an equal number of polymeric Hbs which have the usual distal His, have been identified to date. Deconvolution of the electrospray ionization mass spectra (ESI-MS) of the Hbs and of their carbamidomethylated, reduced, and reduced/carbamidomethylated forms, using a maximum entropy-based approach (MaxEnt), showed the presence of at least 18 peaks attributable to monomer Hbs (14,500-15,200 Da) and an approximately equal number of polymer Hb peaks (15,500-16,400 Da). Although the ratio of the monomer to polymer components in pooled Hb preparations remained constant at 60:40, Hb from individuals had generally less than 6 monomer and 6 polymer components; -2 of the 19 individuals appeared to be deficient in polymer Hbs. Taking into account possible fragmentations of the known monomeric and polymeric globin sequences, we estimate conservatively that there are 10 monomeric and an equal number of polymeric Hbs, the majority comprising a single free Cys. Surprisingly, the calculated mass of the sequence deduced from the high-resolution monomer Hb crystal structures does not correspond to any of the observed masses. ESI-MS of the monomer Hb crystal revealed 11 components, of which 5, accounting for 67% of total, were related to the three major sequences GMG2-4. These findings underline the need for routine mass spectrometric characterization of all protein preparations. The complete resolution of the Glycera Hb ESI-MS using MaxEnt processing illustrates the power of this method to resolve complex protein mixtures.

Expression of thrombospondin 1 on the surface of activated platelets mediates their interaction with the heavy chains of human kininogens through Lys 244-Pro 254.

Platelet thrombospondin (TSP1) forms a complex with high (HK) and low (LK) molecular weight kininogens. We isolated a proteolytic fragment from HK and LK heavy chains (12 kDa) recognized by TSP1 with a N-terminal sequence, K244ICVGCPRDIP254. Lys244-Pro254 oxidized to cyclic form prevented binding of 125I-LK to TSP1. This effect was abolished by reduction and alkylation. Oxidized peptide KICVGCPRDIP (100 microM) reversed the known inhibitory effects of LK or HK (1 microM), on thrombin-induced platelet activation, suggesting this peptide forms part of the cell binding site on HK and LK for activated platelets. KICVGCPRDIP completely inhibited the binding of 125I-LK to activated platelets. However, the peptide only partially inhibited binding of 125I-HK to platelets, suggesting an additional binding site on the HK light chain. Fluorescein-labeled KICVGCPRDIP bound directly and specifically to activated platelets. A monoclonal antibody directed to TSP1 partially inhibited the binding of 125I-HK to activated but not inactivated platelets. We conclude residues Lys244-Pro254 on kininogen heavy chain is responsible for binding to thrombospondin on the surface of activated platelets.

Molecular symmetry of the dodecamer subunit of Lumbricus terrestris hemoglobin.

The principal functional subunit of the approximately 3500 kDa extracellular Lumbricus terrestris hemoglobin is a 213 kDa dodecamer of four chemically distinct globin chains, consisting of a non-covalent complex of three trimer submits (disulfide-bonded chains a, b and c) and three monomer subunits (chain d). X-ray diffraction of crystals of the dodecamer grown at neutral pH, were found to be monoclinic, with the unit cell dimensions: a = 112.3 A, b = 190.0 A, c = 69.6 A, beta = 102.0 degrees with h + k + l = 2n + 1 absent, characteristic of space group I121. In addition, these crystals exhibit a pseudo trigonal P321 symmetry with unit cell dimensions a = 190.5 A, b = 190.5 A, c = 69.5 A, gamma = 120.0 degrees. Assuming that the assymetric unit contains an entire dodecamer, a model of the latter was constructed that satisfies the symmetry of the trigonal pseudo cell and is consistent with the symmetry of the I121 crystallographic cell. The resulting model has strong implications concerning the hexagonal bilayer structure of the native hemoglobin.

Electron spin resonance and fluorescence studies of the conformational environment of the thiol groups of thrombospondin: interactions with thrombin.

The free thiols of platelet thrombospondin (TSP) were modified with thiol-specific spin labels and fluorescence probes. The conformational effects of thrombin complexation with TSP were monitored by thiol-specific spin labels covalently attached to TSP and active site specific spin labels on thrombin. The results provide evidence supporting speculations that the thiols of the three polypeptide chains in TSP are not conformationally identical. Studies on the effects of Ca2+ and temperature confirm that TSP exists in multiple conformations which are under dynamic equilibrium. The ESR spectra of spin-labeled TSP are sensitive to the proteolytic effects of thrombin in the presence and absence of calcium. Phenylsulfonyl fluoride spin labels specific for the active site of thrombin are excellent indicators of thrombin: TSP complex formation in the absence of calcium. The anticoagulant thrombin inhibitor hirudin competes with TSP for the same binding locus on thrombin (which includes the requirement of an intact anion exosite). The results suggest that the species observed here is the noncovalent complex formed during the first step of the TSP--thrombin interaction, showing also that thrombin activity is not essential for complex formation. ESR and fluorescence studies of thiol-labeled TSP indicate that the sulfhydryls are not affected in the noncovalent thrombin: TSP complex, although they must be playing a major role in the second step, i.e., formation of the covalent complex, through intermolecular thiol exchange.

Solid tumor cells express functional "tethered ligand" thrombin receptor.

Previous work demonstrated that alpha-thrombin promoted tumor cell adhesion to endothelium and extracellular matrix as well as enhanced the metastatic capacity of tumor cells. This study was initiated to investigate whether the thrombin effect on tumor cells is mediated through the "tethered ligand" thrombin receptor. RT-PCR analysis using primers based on the human thrombin receptors detected mRNA in human colon adenocarcinoma cells (clone A), whose authenticity was confirmed by Southern hybridization. The presence of thrombin receptor mRNA in rat (W256 carcinosarcoma) and mouse (melanoma) tumor cells was demonstrated by RT-PCR/Southern blotting using species-specific PCR primers. Sequencing of the PCR fragment of clone A cells revealed complete homology with the reported human cDNA sequence. Subsequently, tumor cells derived from three species, i.e., human, rat, and mouse, were found to express the thrombin receptor protein as revealed by immunoblotting using ligand peptide-derived mAb ATAP138, whose reactivity towards the M(r) approximately 66,000, potential thrombin receptor was blocked by preincubating the antibody with the immunogen peptide SFLLRNPNDKYEPF (TRP 14). Finally, peptides TRP 14 and TRP 7 (SFLLRNP), but not TRP 5 (FLLRN), were found to mimic alpha-thrombin in stimulating tumor cell adhesion to fibronectin, suggesting that the thrombin receptors expressed on solid tumor cells are biologically functional.

Crystallization of hemoglobins II and III of the symbiont-harboring clam Lucina pectinata.

Diffraction data to 2.7 A resolution were measured on crystals of the homotetramers of components II and III of the cytoplasmic hemoglobin of the symbiont-harboring clam Lucina pectinata. Even though the crystallization conditions are different and the sequence homology of the two hemoglobins is only 63%, the crystals are isomorphous to each other and to the heterotetramer Hb II/III, implying that the residues primarily involved in the intermolecular interactions and responsible for crystal cohesion may be invariant.

Platelet thrombospondin interactions with human high and low molecular weight kininogens.

Multifunctional proteins, e.g. high molecular weight kininogen (HK, 120 kDa) and the homotrimer, thrombospondin (TSP, 540 kDa), which have more than one domain on a single polypeptide chain, are particularly well-suited to be structural elements of extracellular matrices because of their ability to bind to several macromolecules. We now demonstrate that 125I-high molecular weight kininogen (HKa) cleaved by purified kallikrein forms a complex with purified intact platelet TSP (540 kDa). HK also complexed with a proteolytic fragment (450 kDa) of TSP, lacking its three identical heparin-binding domains (HBD, 30 kDa), but failed to bind to a more extensively proteolysed molecule (210 kDa) lacking the C-terminal globular domain indicating that the binding on TSP-450 kDa is confined to the C-terminus. The binding of HK to intact TSP and to its 450 kDa fragment was of high affinity (Kd = 17-52 nM), specific, concentration dependent and saturable. Furthermore, we found both forms of the light chain (LC) of HK (56 and 46 kDa) resulting from cleavage by plasma kallikrein bound to both intact TSP and HBD independent of the presence of calcium ions. However, neither the epitope recognized by monoclonal antibody (MAb) C11C1 on domain 5 nor the prekallikrein binding site on domain 6 are involved, suggesting that the intervening proline-rich region may be the site of interaction. The heavy chain (HC) of HK required ionized calcium to bind to intact TSP or its 450 kDa homotrimer fragment.(ABSTRACT TRUNCATED AT 250 WORDS)

Evolution of prothrombin: isolation and characterization of the cDNAs encoding chicken and hagfish prothrombin.

The cDNA sequences of chicken and hagfish prothrombin have been determined. The sequences predict that prothrombin from both species is synthesized as a prepro-protein consisting of a putative Gla domain, two kringle domains, and a two-chain protease domain. Chicken and hagfish prothrombin share 51.6% amino acid sequence identity (313/627 residues). Both chicken and hagfish prothrombin are structurally very similar to human, bovine, rat, and mouse prothrombin and all six species share 41% amino acid sequence identity. Amino acid sequence alignments of human, bovine, rat, mouse, chicken, and hagfish prothrombin suggest that the thrombin B-chain and the propeptide-Gla domain are the regions most constrained for the common function(s) of vertebrate prothrombins.

The role of thrombin in tumor cell metastasis.

The relationship between malignancy and thrombosis and hemostasis has long been recognized but is poorly understood. The interaction of tumor cells and the proteins and cells of coagulation is symbiotic and complex. Selected cancers have been known to respond to antithrombotic and anticoagulant therapies such as warfarin, heparin and most recently, hirudin. The identification of the functional 'tethered ligand' thrombin receptor on platelets, other cells and recently tumor cells has provided additional opportunities to examine and control the course of tumor proliferation. Thrombin, at concentrations which precede fibrin formation, is a potent inducer of tumor cell expression of various integrin receptors and tumor cell adhesion to the matrix and other activated cells. New strategies based on specifically controlling the action of thrombin may be beneficial in arresting the proliferation of these cells.

Purification and characterization of recombinant polymeric hemoglobin P1 of Glycera dibranchiata.

The apoprotein of component P1 of the polymeric fraction of the intracellular hemoglobin of the marine polychaete Glycera dibranchiata has been expressed at a high level in Escherichia coli. The expressed globin was reconstituted with heme and purified. The N-terminal sequence of the recombinant P1 is identical to the cDNA-derived sequence of cloned P1 (Zafar et al., Biochem. Biophys. Acta, 1041, 117-123, 1990). Gel filtration, SDS-PAGE, optical spectra over the range 200-650 nm, and circular dichroism over the range 200-250 nm of the purified recombinant P1 were very similar to the polymeric fraction of native Glycera hemoglobin. The molar ellipticity at 222 nm provided an estimate of 77% for the alpha-helical content of the recombinant P1, in excellent agreement with that calculated from the crystal structure of Glycera monomeric component M-II. Although the oxygen binding affinity of the recombinant P1 is higher than that of the polymeric fraction of Glycera hemoglobin (3-4 torr vs 7-13 torr), which consists of at least six different single-chain hemoglobins, the Hill coefficient is lower (1.0-1.2 vs 1.2-1.4).

Purification and properties of the hemoglobins of the platyhelminth Isoparorchis hypselobagri (Trematoda: Isoparorchidae) and its host Wallagu attu (catfish).

1. The hemoglobins of the trematode Isoparorchis hypselobagri and of its host Wallagu attu (catfish) were isolated and purified. 2. SDS-polyacrylamide gel electrophoresis showed both to consist of single, 15-17 kDa chains, having different electrophoretic mobilities. 3. Isoelectric focusing showed the trematode hemoglobin to be homogeneous with a pI of 4.2 and the host hemoglobin to consist of several components. 4. Gel filtration of freshly prepared trematode hemoglobin revealed one peak corresponding to M(r) approximately 17 kDa; gel filtration of a preparation which had been stored for 2-3 months demonstrated the presence of two peaks, whose elution volumes corresponded to M(r) of ca 35 and 17 kDa, respectively. 5. Reversed-phase chromatography of carboxymethylated 35 and 17 kDa peaks on a C8 column, gave a single peak a and two peaks b and c, respectively. 6. Edman degradation of peaks a, b and c obtained provided identical sequences of 27 amino acid residues for peaks a and c and another sequence differing at 10 of the 27 positions, for peak b. Edman degradation of the freshly prepared Isoparorchis hemoglobin provided the first 15 amino acid residues found for peaks a and c. The host hemoglobin gave an N-terminal sequence completely different from the trematode sequences. 7. Since gel filtration of the 35 and 17 kDa peaks showed no sign of an interconversion equilibrium, it appears that the 35 kDa peak and peak a represent a disulfide-bonded dimer of a monomer globin chain which shares the 27 N-terminal residues with chain c, the major monomer globin component of the 17 kDa peak.(ABSTRACT TRUNCATED AT 250 WORDS)

Adventitious variability? The amino acid sequences of nonvertebrate globins.

1. The more than 140 amino acid sequences of non-vertebrate hemoglobins (Hbs) and myoglobins (Mbs) that are known at present, can be divided into several distinct groups: (1) single-chain globins, containing one heme-binding domain; (2) truncated, single-chain, one-domain globins; (3) chimeric, one-domain globins; (4) chimeric, two-domain globins; and (5) chimeric multi-domain globins. 2. The crystal structures of eight nonvertebrate Hbs and Mbs are known, all of them monomeric, one-domain globin chains. Although these molecules represent plants, prokaryotes and several metazoan groups, and although the inter-subunit interactions in the dimeric and tetrameric molecules differ from the ones observed in vertebrate Hbs, the secondary structures of all seven one-domain globins retain the characteristic vertebrate "myoglobin fold". No crystal structures of globins representing the other four groups have been determined. 3. Furthermore, a number of the one-, two- and multi-domain globin chains participate in a broad variety of quaternary structures, ranging from homo- and heterodimers to highly complex, multisubunit aggregates with M(r) > 3000 kDa (S. N. Vinogradov, Comp. Biochem. Physiol. 82B, 1-15, 1985). 4. (1) The single-chain, single-domain globins are comparable in size to the vertebrate globins and exhibit the widest distribution. (A) Intracellular Hbs include: (i) the monomeric and polymeric Hbs of the polychaete Glycera; (ii) the tetrameric Hb of the echiuran Urechis; (iii) the dimeric Hbs of echinoderms such as Paracaudina and Caudina; and (iv) the dimeric and tetrameric Hbs of molluscs, the bivalves Scapharca, Anadara, Barbatia and Calyptogena. (B) Extracellular Hbs include: (i) the multiple monomeric and dimeric Hbs of the larva of the insect Chironomus; (ii) the Hbs of nematodes such as Trichostrongylus and Caenorhabditis; (iii) the globin chains forming tetramers and dodecamers and comprising approximately 2/3 of the giant (approximately 3600 kDa), hexagonal bilayer (HBL) Hbs of annelids, e.g. the oligochaete Lumbricus and the polychaete Tylorrhynchus and of the vestimentiferan Lamellibrachia; and (iv) the globin chains comprising the ca 400 kDa Hbs of Lamellibrachia and the pogonophoran Oligobrachia. (C) Cytoplasmic Hbs include: (i) the Mbs of molluscs, the gastropods Aplysia, Bursatella, Cerithedea, Nassa and Dolabella and the chiton Liolophura; (ii) the three Hb of the symbiont-harboring bivalve Lucina; (iii) the dimeric Hb of the bacterium Vitreoscilla; and (iv) plant Hbs, including the Hbs of symbiont-containing legumes (Lgbs), the Hbs of symbiont-containing non-leguminous plants and the Hbs in the roots of symbiont-free plants.(ABSTRACT TRUNCATED AT 400 WORDS)

Thrombin increases the metastatic potential of tumor cells.

Initial arrest of tumor cells in the microvasculature and their attachment to the endothelium and subendothelial matrix (SEM) are essential prerequisites for metastasis to occur. Factors mediating these interactions are viewed as important determinants of the tumor-cell metastatic phenotype. In this work we have studied the effects of thrombin, its analogs and its precursors on the adhesive properties and metastatic potential of tumor cells. We show that alpha-thrombin, the native form of the key coagulation enzyme, is capable of enhancing tumor-cell adhesion to both the endothelium and SEM components represented by fibronectin. Subclotting, physiological concentrations of alpha-thrombin produced a 2- to 5-fold increase in tumor-cell adhesion. A bell-shaped dose-response curve was observed, with maximal effect at 0.1 U/ml. Maximum effect occurred when cells were exposed to the agonist for 15 min and exposure for up to 4 hr resulted in enhanced tumor-cell adhesion. Prolonged incubation with thrombin resulted in a decline in the thrombin-enhanced adhesion which reached unstimulated control levels by 24 hr. Thrombin precursors and active-site-inhibited thrombin analogs only had minimal adhesion-enhancing activity; nitro- and exosite-alpha-thrombin, which retain a functional active site, mimicked, although to a lesser degree, the action of alpha-thrombin. Tumor-cell incubation with thrombin resulted in an upregulated cell-surface expression of the alpha11b beta 3 integrin, a receptor mediating interactions between tumor cells and endothelial cells, and between tumor cells and SEM. Antibodies against alpha 11b beta 3 integrin effectively inhibited thrombin-enhanced tumor-cell adhesion. Thrombin effects on tumor cells involved the PKC signal transduction pathway as thrombin-enhanced adhesion was inhibited by pre-incubation with PKC inhibitors and a transient PKC translocation from cytosol to membrane was observed following thrombin challenge. In vivo, thrombin-treated tumor cells demonstrated a 2-fold increase in their lung-colonizing ability. In contrast to the adhesion results, the metastasis-enhancing effects of alpha-thrombin were mimicked by a thrombin precursor (prothrombin) and thrombin analogs.

The amino acid sequence of hemoglobin III from the symbiont-harboring clam Lucina pectinata.

The cytoplasmic hemoglobin III from the gill of the symbiont-harboring clam Lucina pectinata consists of 152 amino acid residues, has a calculated Mm of 18,068, including heme, and has N-acetyl-serine as the N-terminal residue. Based on the alignment of its sequence with other vertebrate and nonvertebrate globins, it retains the invariant residues Phe45 at position CD1 and His98 at the proximal position F8, as well as the highly conserved Trp16 and Pro39 at positions A12 and C2, respectively. The most likely candidate for the distal residue at position E7 is Gln66. Lucina hemoglobin III shares 95 identical residues with hemoglobin II (J. D. Hockenhull-Johnson et al., J. Prot. Chem. 10, 609-622, 1991), including Tyr at position B10, which has been shown to be capable of entering the distal heme cavity and placing its hydroxyl group within a 2.8 A of the water molecule occupying the distal ligand position, by modeling the hemoglobin II sequence using the crystal structure of sperm whale metmyoglobin. The amino acid sequences of the two Lucina globins are compared in detail with the known sequences of mollusc globins, including seven cytoplasmic and 11 intracellular globins. Relative to 75% homology between the two Lucina globins (counting identical and conserved residues), both sequences have percent homology scores ranging from 36-49% when compared to the two groups of mollusc globins. The highest homology appears to exist between the Lucina globins and the cytoplasmic hemoglobin of Busycon canaliculatum.

Organization of non-vertebrate globin genes.

The organization of non-vertebrate globin genes exhibits substantially more variability than the three-exon, two-intron structure of the vertebrate globin genes. (1) The structures of genes of the single-domain globin chains of the annelid Lumbricus and the mollusc Anadara, and the globin gene coding for the two-domain chains of the clam Barbatia, are similar to the vertebrate plan. (2) Genes for single-domain chains exist in bacteria and protozoa. Although the globin gene is highly expressed in the bacterium Vitreoscilla, the putative globin gene hmp in E. coli, which codes for a chimeric protein whose N-terminal moiety of 139 residues contains 67 residues identical to the Vitreoscilla globin, may be either unexpressed or expressed at very low levels, despite the presence of normal regulatory sequences. The DNA sequence of the globin gene of the protozoan Paramecium, determined recently by Yamauchi and collaborators, appears to consist of two exons separated by a short intron. (3) Among the lower eukaryotes, the yeasts Saccharomyces and Candida have chimeric proteins consisting of N-terminal globin and C-terminal flavoprotein moieties of about the same size. The structure of the gene for the chimeric protein of Saccharomyces exhibits no introns. According to Riggs, the presence of chimeric proteins in E. coli and other prokaryotes, such as Alcaligenes and Rhizobium, as well as in yeasts, suggests a previously unrecognized evolutionary pathway for hemoglobin, namely that of a multipurpose heme-binding domain attached to a variety of unrelated proteins with diverse functions. (4) The published globin gene sequences of the insect larva Chironomus have an intron-less structure and are present as clusters of multiple copies; the expression of the globin genes is tissue and developmental stage-specific. Furthermore, the expression of many of these genes has not yet been demonstrated despite the presence of apparently normal regulatory sequences in the two flanking regions. Unexpectedly, Bergtrom and collaborators have recently shown that at least three Ctt globin II beta genes contain putative introns. (5) Pohajdak and collaborators have found a seven-exon and six-intron structure for the globin gene of the nematode Pseudoterranova which codes for a two-domain globin chain. Although the second and fourth introns of the N-terminal domain correspond to the two introns found in vertebrate globin genes, the position of the third intron is close to that of the central intron in plant hemoglobins.(ABSTRACT TRUNCATED AT 400 WORDS)

Thrombospondin as a mediator of cancer cell adhesion in metastasis.

Thrombospondin (TSP) is a 450 kDa adhesive glycoprotein. It is present in high concentrations in the platelet alpha-granule and can readily be secreted following platelet activation where local concentrations can be increased by 3-4 orders of magnitude. TSP is also synthesized by a variety of other cells and is incorporated into their extracellular matrix. TSP is a homotrimer with a number of functional domains, at least four of which might serve as receptor recognizing regions. The amino-terminal heparin binding domain interacts with heparin, other glycosaminoglycans and glycolipids and likely recognizes specific cell surface proteoglycans. The central disulfide cross-linked region, 210 kDa non-reduced and 70 kDa reduced, contains a peptide motif CSVTCG which is apparently responsible for binding to glycoprotein IV (CD36) with high affinity. Immediately adjacent to the calcium binding region of TSP, which undergoes considerable molecular relaxation in the absence of calcium, is an RGDA sequence. TSP has been demonstrated to bind to integrins of the alpha v beta 3 and alpha IIb beta 3 class. The carboxy-terminal region of TSP also contains at least one binding epitope for a cell receptor. There are 2 well characterized genes for TSP and truncated forms of TSP have been detected which have inhibitory effects on angiogenesis. Finally, TSP can interact with fibrinogen and fibronectin, perhaps on cellular surfaces, which might serve as secondary receptor-like mechanisms for TSP binding and subsequent mediation of cell adhesion.

Thrombin enhances tumor cell adhesive and metastatic properties via increased alpha IIb beta 3 expression on the cell surface.

The association between blood coagulation and cancer growth and metastatic dissemination is not yet completely understood. In this study we demonstrate that thrombin is capable of enhancing tumor cell adhesive properties and thereby increases tumor cell metastatic potential. Following exposure to alpha-thrombin, Walker 256 carcinosarcoma cells and B16 amelanotic melanoma cells became more adherent to both endothelial cell monolayers and the subendothelial matrix component, fibronectin. Preincubation of W256 and B16a cells with doses of alpha-thrombin from 0.01 to 10.0 U/ml produced a bell shape dose-response curve with the maximal effect (a 2-5-fold increase in adhesion) observed at 0.1 U/ml (corresponding to 0.8 nM). Complexes of alpha-thrombin with its inhibitors, hirudin and antithrombin III-heparin, diminished its effect on tumor cell adhesion. The effect of thrombin on tumor cell adhesion may be mediated by the alpha IIb beta 3 integrin as thrombin increased cell surface expression of the alpha IIb beta 3 complex. The significance of the in vitro observations was further substantiated by results of in vivo studies. Pretreatment of B16a cells with alpha-thrombin resulted in a 2-fold increase in the number of metastatic lung colonies in an experimental metastasis model. The data indicate a new role for thrombin in the metastatic spread of cancer.

Localization of two binding domains for thrombospondin within fibronectin.

Thrombospondin is a major glycoprotein of the platelet alpha-granule and is secreted during platelet activation. Several protease-resistant domains of thrombospondin mediate its interactions with components of the extracellular matrix including fibronectin, collagen, heparin, laminin, and fibrinogen. Thrombospondin, as well as fibronectin, is composed of several discretely located biologically active domains. We have characterized the thrombospondin binding domains of plasma fibronectin and determined the binding affinities of the purified domains; fibronectin has at least two binding sites for thrombospondin. Thrombospondin bound specifically to the 29-kDa amino-terminal heparin binding domain of fibronectin as well as to the 31-kDa non-heparin binding domain located within the larger 40-kDa carboxy-terminal fibronectin domain generated by chymotrypsin proteolysis. Platelet thrombospondin interacted with plasma fibronectin in a specific and saturable manner in blot binding as well as solid-phase binding assays. These interactions were independent of divalent cations. Thrombospondin bound to the 29-kDa fibronectin heparin binding domain with a Kd of 1.35 x 10(-9) M. The Kd for the 31-kDa domain of fibronectin was 2.28 x 10(-8) M. The 40-kDa carboxy-terminal fragment bound with a Kd of 1.65 x 10(-8) M. Heparin, which binds to both proteins, inhibited thrombospondin binding to the amino-terminal domain of fibronectin by more than 70%. The heparin effect was less pronounced with the non-heparin binding carboxy-terminal domain of fibronectin. By contrast, the binding affinity of the thrombospondin 150-kDa domain, which itself lacked heparin binding, was not affected by the presence of heparin. Based on these data, we conclude that thrombospondin binds with different affinities to two distinct domains in the fibronectin molecule.

Evidence for the existence of O-linked sugar chains consisting of glucose and xylose in bovine thrombospondin.

We have recently discovered unusual sugar chains [xylose-glucose and (xylose)2-glucose] linked to a serine residue in the first epidermal growth factor (EGF)-like domains of human and bovine coagulation factors VII, IX, and protein Z. The sequence surrounding this serine residue has a common -Cys-X-Ser-X-Pro-Cys- structure. Since one (residues 533-538) of the three EGF-like domains found in human thrombospondin contains the conserved sequence, we examined the presence of such O-linked sugar chains in bovine thrombospondin (bTSP) and its 210-kDa fragment. Component sugar analysis after pyridylamination (PA) of the acid hydrolysates of the S-aminoethylated proteins revealed that the proteins contain glucose (Glc) and xylose (Xyl). The oligosaccharide moieties released from intact bTSP by hydrazinolysis followed by pyridylamination were separated into two PA-oligosaccharides by high performance liquid chromatography (HPLC). Component sugar analysis of these PA-oligosaccharides indicated that they consist of Glc and Xyl in molar ratios of 1:1 and 1:2 (or 1:3). The reducing ends of both PA-sugar chains were found to be PA-Glc, as judged from the retention time of the HPLC peak of their hydrolysates. The presence of these PA-sugar chains in bTSP was confirmed by HPLC mapping with two different columns, using standard PA-di- or PA-trisaccharide derived from coagulation factors. From these results, we concluded that bTSP contains O-linked sugar chains consisting of Glc and Xyl in one of its three EGF-like domains.