Chronic liver injury drives non-traditional intrahepatic fibrin(ogen) crosslinking via tissue transglutaminase.

Essentials Fibrin clots are often implicated in the progression of liver fibrosis. Liver fibrosis was induced in transgenic mice with defects in clot formation or stabilization. Liver fibrosis and fibrin(ogen) deposition do not require fibrin polymerization or factor XIIIa. Fibrin(ogen) is an in vivo substrate of tissue transglutaminase in experimental liver fibrosis. SUMMARY: Background Intravascular fibrin clots and extravascular fibrin deposits are often implicated in the progression of liver fibrosis. However, evidence supporting a pathological role of fibrin in hepatic fibrosis is indirect and based largely on studies using anticoagulant drugs that inhibit activation of the coagulation protease thrombin, which has other downstream targets that promote fibrosis. Therefore, the goal of this study was to determine the precise role of fibrin deposits in experimental hepatic fibrosis. Methods Liver fibrosis was induced in mice expressing mutant fibrinogen insensitive to thrombin-mediated proteolysis (i.e. locked in the monomeric form), termed FibAEK mice, and factor XIII A2 subunit-deficient (FXIII-/- ) mice. Female wild-type mice, FXIII-/- mice and homozygous FibAEK mice were challenged with carbon tetrachloride (CCl4 ) twice weekly for 4 weeks or 6 weeks (1 mL kg-1 , intraperitoneal). Results Hepatic injury and fibrosis induced by CCl4 challenge were unaffected by FXIII deficiency or inhibition of thrombin-catalyzed fibrin polymer formation (in FibAEK mice). Surprisingly, hepatic deposition of crosslinked fibrin(ogen) was not reduced in CCl4 -challenged FXIII-/- mice or FibAEK mice as compared with wild-type mice. Rather, deposition of crosslinked hepatic fibrin(ogen) following CCl4 challenge was dramatically reduced in tissue transglutaminase-2 (TGM2)-deficient (TGM2-/- ) mice. However, the reduction in crosslinked fibrin(ogen) in TGM2-/- mice did not affect CCl4 -induced liver fibrosis. Conclusions These results indicate that neither traditional fibrin clots, formed by the thrombin-activated FXIII pathway nor atypical TGM2-crosslinked fibrin(ogen) contribute to experimental CCl4 -induced liver fibrosis. Collectively, the results indicate that liver fibrosis occurs independently of intrahepatic fibrin(ogen) deposition.

Targeted inactivation of Gh/tissue transglutaminase II.

The novel G-protein, G(h)/tissue transglutaminase (TGase II), has both guanosine triphosphatase and Ca(2+)-activated transglutaminase activity and has been implicated in a number of processes including signal transduction, apoptosis, bone ossification, wound healing, and cell adhesion and spreading. To determine the role of G(h) in vivo, the Cre/loxP site-specific recombinase system was used to develop a mouse line in which its expression was ubiquitously inactivated. Despite the absence of G(h) expression and a lack of intracellular TGase activity that was not compensated by other TGases, the Tgm2(-/-) mice were viable, phenotypically normal, and were born with the expected Mendelian frequency. Absence of G(h) coupling to alpha(1)-adrenergic receptor signaling in Tgm2(-/-) mice was demonstrated by the lack of agonist-stimulated [alpha-(32)P]GTP photolabeling of a 74-kDa protein in liver membranes. Annexin-V positivity observed with dexamethasone-induced apoptosis was not different in Tgm2(-/-) thymocytes compared with Tgm2(+/+) thymocytes. However, with this treatment there was a highly significant decrease in the viability (propidium iodide negativity) of Tgm2(-/-) thymocytes. Primary fibroblasts isolated from Tgm2(-/-) mice also showed decreased adherence with culture. These results indicate that G(h) may be importantly involved in stabilizing apoptotic cells before clearance, and in responses such as wound healing that require fibroblast adhesion mediated by extracellular matrix cross-linking.

GTP binding and signaling by Gh/transglutaminase II involves distinct residues in a unique GTP-binding pocket.

G(h) is a dual function protein. It has receptor signaling activity that requires GTP binding and Ca(2+)-activated transglutaminase (TGase) activity that is inhibited by GTP binding. G(h) shows no homology with other GTP-binding proteins, and its GTP-binding site has not been defined. Based on sequence analysis of [alpha-(32)P]GTP-photolabeled and proteolytically released internal peptide fragments, we report localization of GTP binding to a 15-residue segment ((159)YVLTQQGFIYQGSVK(173)) of the G(h) core domain. This was confirmed by site-directed mutagenesis; a G(h)/fXIIIA chimera (in which residues 162-179 of G(h) were substituted with the equivalent but nonhomologous region of the non-GTP-binding TGase factor XIIIA) and a G(h) point mutant, S171E, retained TGase activity but failed to bind and hydrolyze GTP and did not support alpha(1B)-adrenergic receptor signaling. Slight impairment of GTP binding (1.5-fold) and hydrolysis (10-fold) in the absence of altered TGase activity did not affect signaling by the mutant K173N. However, greater impairment of GTP binding (6-fold) and hydrolysis (50-fold) abolished signaling by the mutant K173L. Mutant S171C exhibited enhanced GTP binding and signaling. Thus, residues Ser(171) and Lys(173) are critical for both GTP binding and signaling but not TGase activity. Mutagenesis of residues N-terminal to Gly(170) impaired both GTP binding and TGase activity. From computer modeling of G(h), it is evident that the GTP-binding region identified here is distinct from, but interacts with, the TGase active site. Together with structural considerations of G(h) versus other GTP-binding proteins, these findings indicate that G(h) has a unique GTP-binding pocket and provide for the first time a mechanism for GTP-mediated regulation of the TGase activity of G(h).

Organization and chromosomal mapping of mouse Gh/tissue transglutaminase gene (Tgm2).

The mouse Gh/tissue transglutaminase gene (Tgm2), coding a dual-function protein that both binds guanosine triphosphate (GTP) and catalyzes the posttranslational modification of proteins by transamidation of glutamine residues, has been cloned. Sequence analysis of Tgm2 and comparison with the TGase sequences of other species allowed correction of several apparent sequencing artifacts in the Tgm2 cDNA. Tgm2 spans approximately 34 kb and has 13 exons and 12 introns. Although the structure of Tgm2 shows similarity to that of other transglutaminase genes, with introns ranging from 921 bp to >5 kb, several introns differ considerably in size from those of the human Gh gene, TGM2. Tgm2 maps to the distal region of mouse chromosome 2, a region syntenic to human chromosome 20q containing TGM2. Tgm2 is in the vicinity of two uncloned mouse mutations, diminutive (dm) and blind-sterile (bs). Genomic DNA from dm mice was unavailable; however, Southern blot analysis of bs DNA showed no gross rearrangements of Tgm2.

Oxidation regulates the inflammatory properties of the murine S100 protein S100A8.

The myeloid cell-derived calcium-binding murine protein, S100A8, is secreted to act as a chemotactic factor at picomolar concentrations, stimulating recruitment of myeloid cells to inflammatory sites. S100A8 may be exposed to oxygen metabolites, particularly hypochlorite, the major oxidant generated by activated neutrophils at inflammatory sites. Here we show that hypochlorite oxidizes the single Cys residue (Cys41) of S100A8. Electrospray mass spectrometry and SDS-polyacrylamide gel electrophoresis analysis indicated that low concentrations of hypochlorite (40 microM) converted 70-80% of S100A8 to the disulfide-linked homodimer. The mass was 20,707 Da, 92 Da more than expected, indicating additional oxidation of susceptible amino acids (possibly methionine). Phorbol 12-myristate 13-acetate activation of differentiated HL-60 granulocytic cells generated an oxidative burst that was sufficient to efficiently oxidize exogenous S100A8 within 10 min, and results implicate involvement of the myeloperoxidase system. Moreover, disulfide-linked dimer was identified in lung lavage fluid of mice with endotoxin-induced pulmonary injury. S100A8 dimer was inactive in chemotaxis and failed to recruit leukocytes in vivo. Positive chemotactic activity of recombinant Ala41S100A8 indicated that Cys41 was not essential for function and suggested that covalent dimerization may structurally modify accessibility of the chemotactic hinge domain. Disulfide-dependent dimerization may be a physiologically significant regulatory mechanism controlling S100A8-provoked leukocyte recruitment.

The core domain of the tissue transglutaminase Gh hydrolyzes GTP and ATP.

Tissue transglutaminase (TGase II) catalyzes the posttranslational modification of proteins by transamidation of available glutamine residues and is also a guanosinetriphosphatase (GTPase) and adenosinetriphosphatase (ATPase). Based on its homology with factor XIIIA, an extracellular transglutaminase, the structure of TGase II is likely composed of an N-terminal beta-sandwich domain, an alpha/beta catalytic core, and two C-terminally located beta-barrels. Here we used a domain-deletion approach to identify the GTP and ATP hydrolytic domains of TGase II. Full-length TGase II and two domain-deletion mutants, one retaining the N-terminal beta-sandwich and core domains (betaSCore) and the other retaining only the core domain, were expressed as glutathione S-transferase (GST) fusion proteins and purified. GST-Full and GST-betaSCore exhibited calcium-dependent TGase activity, whereas GST-Core had no detectable TGase activity, indicating the beta-sandwich domain is required for TGase activity but the C-terminal beta-barrels are not. All three GST-TGase II fusion proteins were photoaffinity-labeled with [alpha-32P]-8-azidoGTP and were able to bind GTP-agarose. The GTPase activity of GST-betaSCore was equivalent to that of GST-Full, whereas the ATPase activity was approximately 40% higher than GST-Full. GST-Core had approximately 50% higher GTPase activity and approximately 75% higher ATPase activity than GST-Full. The GTPase and ATPase activities of each of the GST-TGase II fusion proteins were inhibited in a dose-dependent manner by both GTPgammaS and ATPgammaS. These results demonstrate that the GTP and ATP hydrolysis sites are localized within the core domain of TGase II and that neither the N-terminal beta-sandwich domain nor the C-terminal beta-barrels are required for either GTP or ATP hydrolysis. Taken together with previous work [Singh, U. S., Erickson, J. W., & Cerione, R. A. (1995) Biochemistry 34, 15863-15871; Lai, T.-S., Slaughter, T. F., Koropchak, C. M., Haroon, Z. A., & Greenberg, C. S. (1996) J. Biol. Chem. 271, 31191-31195] the results of this study indicate that the GTP and ATP hydrolysis sites are localized to a 5. 5 kDa (47 amino acid) region at the start of the core domain.

Induction of the S100 chemotactic protein, CP-10, in murine microvascular endothelial cells by proinflammatory stimuli.

Microvascular endothelial cells (EC) have multiple functions in inflammatory responses, including the production of chemoattractants that enhance leukocyte transmigration into tissues. Chemotactic protein, 10 kD (CP-10), is an S100 protein with potent chemotactic activity for myeloid cells in vitro and in vivo and is expressed in neutrophils and lipopolysaccharide (LPS)-activated macrophages. We show here that CP-10 is induced in murine endothelioma cell lines (bEnd-3, sEnd-1, and tEnd-1) after activation with LPS and interleukin-1 (IL-1) but not tumor necrosis factor alpha (TNFalpha) or interferon gamma (IFNgamma). Induction was not mediated by endogenous release of IL-1 or TNFalpha and was not directly upregulated by phorbol myristate acetate, calcium ionophore, or vitamin D3. EC were exquisitely sensitive to IL-1 activation (3.4 U/mL) and CP-10 mRNA induction with IL-1 occurred earlier (8 hours) than with LPS (12 hours). Furthermore, some microvessels and capillaries in delayed-type hypersensitivity lesions expressed cytoplasmic CP-10. Responses to LPS and not IL-1 in vitro were regulated by the degree of cell confluence and by TNFalpha costimulation. The related MRP-14 mRNA had a different induction pattern. Monomeric and homodimeric CP-10 upregulated by activation was predominantly cell-associated. EC-derived CP-10 may contribute to amplification of inflammatory processes by enhancing leukocyte shape changes and transmigration in the microcirculation.

Recombinant and cellular expression of the murine chemotactic protein, CP-10.

The S100 protein CP-10 (chemotactic protein, 10 kD), a potent chemotactic factor for murine and human polymorphonuclear cells (PMN) and murine monocytes, has been purified in small amounts from supernatants of activated murine spleen cells (Lackmann et al., 1992). To obtain a more abundant source of the protein, CP-10 was expressed in Escherichia coli as a fusion protein with glutathione S-transferase (GST). The property of S100 proteins to undergo calcium-dependent conformational changes was used in a novel approach to optimize the release of recombinant (r) CP-10 by thrombin cleavage. Purified rCP-10 was characterized by amino-terminal sequence analysis and bioassays. Optimal chemotactic activity of rCP-10 for murine PMN and WEHI-265 monocytoid cells was 10(-11) M (native protein has optimal chemotactic activity between 10(-11) and 10(-13) M). Immunization of rabbits with the GST/CP-10 fusion protein bound to glutathione-agarose beads resulted in high titer, specific antibodies that neutralized CP-10-initiated chemotaxis and were suitable for immunoblotting. A combination of Western and Northern analyses identified CP-10 in murine peritoneal exudate PMN and macrophages, splenocytes, bone marrow cells, and WEHI-265 cells (all of myeloid origin), but not in thymus, liver, lung, 3T3 fibroblasts, EL4 lymphoma cells, or bEND 3 brain endothelial cells, indicating cell-specific regulation of CP-10 expression.

Azotobacter vinelandii mutS: nucleotide sequence and mutant analysis.

An Azotobacter vinelandii homolog to the Salmonella typhimurium mutS gene was discovered upstream of the fdxA gene. The product of this gene is much more similar to S. typhimurium MutS than either is to the HexA protein of Streptococcus pneumoniae. An A. vinelandii delta mutS mutant strain was shown to have a spontaneous mutation frequency 65-fold greater than that of the wild type.

Identification of a chemotactic domain of the pro-inflammatory S100 protein CP-10.

We previously reported the purification and partial amino acid sequence of a novel murine cytokine designated CP-10, which has chemotactic activity for murine polymorphonuclear cells (PMN) and macrophages. The complete cDNA encoding an 88-amino acid polypeptide has been isolated and the sequence is presented here. Transient transfection of CP-10 cDNA into CV-1 cells confirmed the chemotactic activity of rCP-10 for murine PMN. CP-10 has sequence homology with members of the S100 family of Ca(2+)-binding proteins with pronounced amino acid sequence similarities within the putative N- and C-terminal Ca(2+)-binding sites, but differences within their connecting hinge and C-terminal regions. We have confirmed the hypothesis of Kligman and Hilt that functional specificity of individual members of the S100 protein family may reside in the hinge region. A synthetic peptide corresponding to the hinge region of CP-10 (CP-10(42-55) was compared with native CP-10 in chemotaxis and skin test assays. Native CP-10 had potent activity for phagocytic cells, but not lymphocytes, in vitro (optimal activity, 10(-11) to 10(-13) M) and elicited a sustained recruitment of neutrophils and mononuclear cells over 24 h in vivo. The hinge-region peptide had strong chemotactic activity for murine phagocytic cells (optimal activity, 10(-10) - 10(-11) M) but elicited only a transient infiltration of neutrophils over 4 to 8 h after intradermal injection. Results indicate that although the hinge region contributes significantly to the functional specificity of the S100 protein CP-10, sustained cellular recruitment typical of a delayed type hypersensitivity response is apparently dependent on the structural integrity of the protein.

Site-directed mutagenesis of Azotobacter vinelandii ferredoxin I. Changes in [4Fe-4S] cluster reduction potential and reactivity.

We have used site-directed mutagenesis to obtain two variants of Azotobacter vinelandii ferredoxin I (AvFdI), whose x-ray structures are now available. In the C20A protein, a ligand to the [4Fe-4S] cluster was removed whereas in the C24A mutant a free cysteine next to that cluster was removed. Like native FdI, both mutants contain one [4Fe-4S] cluster and one [3Fe-4S] cluster. The structure of C24A is very similar to that of native FdI, while the structure of C20A is rearranged in the region of the [4Fe-4S] cluster to allow it to use the free Cys-24 as a replacement ligand. Here we compare the properties of the native, C20A, and C24A proteins. Although all three proteins are O2 stable in vitro, the C20A protein is much less stable toward proteolysis than the other two in vivo. Spectroscopic results show that all three proteins exhibit the same general redox behavior during O2-oxidation and dithionite reduction. Electrochemical data show that the [3Fe-4S] clusters in all three proteins have the same pH-dependent reduction potentials (-425 mV versus SHE, pH 7.8), whereas the [4Fe-4S] cluster potentials vary over a approximately 150 mV range from -600 mV (C24A) to -647 mV (native) to -746 mV (C20A). Despite this variation in potential both the C20A and C24A proteins appear to be functional in vivo. Native FdI reacts with three equivalents of Fe(CN)3-(6) to form a paramagnetic species previously proposed to be a cysteinyl-disulfide radical. Neither the C20A nor the C24A variant undergoes this reaction, strongly suggesting that it involves the free Cys-24.

Molecular linkage of the nif/fix and nod gene regions in Rhizobium leguminosarum biovar trifolii.

Nucleotide sequence analysis of a 2.5kb region downstream of the nifA gene from Rhizobium leguminosarum biovar trifolii has resulted in linkage, at the DNA sequence level, of the nifEN, nifHDK, fixABCX, nifA gene cluster with the nodEF, nodD, nodABCIJ genes. Four genes have been identified within this intervening region. Immediately 3' to the nifA gene is the nifB gene and the nifB-linked ferredoxin-encoding fdxN gene. Downstream of fdxN in R. leguminosarum bv. trifolii and in Rhizobium meliloti, we have identified an open reading frame which has not been described previously and which we propose to designate fixU. Downstream of fixU in R. leguminosarum bv. trifolii is a nod gene, nodT, which is contiguous with nodJ (B. Surin et al., manuscript in preparation). As a result of this study, the linkage relationships of 22 symbiotic genes spanning a 24 kb region of the symbiotic plasmid from R. leguminosarum bv. trifolii are now known.

The nifA gene product from Rhizobium leguminosarum biovar trifolii lacks the N-terminal domain found in other NifA proteins.

The nifA gene has been identified between the fixX and nifB genes in the clover microsymbiont Rhizobium leguminosarum biovar trifolii (R.I. bv. trifolii) strain ANU843. Expression of the nifA gene is induced in the symbiotic state and site-directed mutagenesis experiments indicate that nifA expression is essential for symbiotic nitrogen fixation. Interestingly, the predicted R.I. bv. trifolii NifA protein lacks an N-terminal domain that is present in the homologous proteins from R.I. bv. viciae, Rhizobium meliloti, Bradyrhizobium japonicum, Klebsiella pneumoniae and all other documented NifA proteins. This indicates that this N-terminal domain is not essential for NifA function in R.I. bv. trifolii.

Construction and characterization of an Azotobacter vinelandii strain with mutations in the genes encoding flavodoxin and ferredoxin I.

Flavodoxin and ferredoxin I have both been implicated as components of the electron transport chain to nitrogenase in the aerobic bacterium Azotobacter vinelandii. Recently, the genes encoding flavodoxin (nifF) and ferredoxin I (fdxA) were cloned and sequenced and mutants were constructed which are unable to synthesize either flavodoxin (DJ130) or ferredoxin I (LM100). Both single mutants grow at wild-type rates under N2-fixing conditions. Here we report the construction of a double mutant (DJ138) which does not synthesize either flavodoxin or ferredoxin I. When plated on ammonium-containing medium, this mutant had a very small colony size when compared with the wild type, and in liquid culture with ammonium, this double mutant grew three times slower than the wild type or single mutant strains. This demonstrated that there is an important metabolic function unrelated to nitrogen fixation that is normally carried out by either flavodoxin or ferredoxin. If either one of these proteins is missing, the other can substitute for it. The double mutant phenotype can now be used to screen site-directed mutant versions of ferredoxin I for functionality in vivo even though the specific function of ferredoxin I is still unknown. The double mutant grew at the same slow rate under N2-fixing conditions. Thus, A. vinelandii continues to fix N2 even when both flavodoxin and ferredoxin I are missing, which suggests that a third as yet unidentified protein also serves as an electron donor to nitrogenase.