Subcellular localization of transglutaminase. Effect of collagen.

1. The subcellular distribution of transglutaminase was investigated by using the analytical approach of differential and isopycnic centrifugation as applied to three organs of the rat: liver, kidney and lung. After differential centrifugation by the method of de Duve, Pressman, Gianetto, Wattiaux & Appelmans [(1955) Biochem. J. 63, 604-617], transglutaminase is mostly recovered in the unsedimentable fraction S and the nuclear fraction N. After isopycnic centrifugation of the N fraction in a sucrose density gradient, a high proportion of the enzyme remains at the top of the gradient; a second but minor peak of activity is present in high-density regions, where a small proportion of 5'-nucleotidase, a plasma-membrane marker, is present together with a large proportion of collagen recovered in that fraction. 2. Fractions where a peak of transglutaminase was apparent in the sucrose gradient were examined by electron microscopy. The main components are large membrane sheets with extracellular matrix and free collagen fibers. 3. As these results seem to indicate that some correlation exists between particulate transglutaminase distribution and those of collagen and plasma membranes, the possible binding of transglutaminase by collagen (type I) and by purified rat liver plasma membrane was investigated. 4. The binding studies indicated that collagen is able to bind transglutaminase and to make complexes with plasma-membrane fragments whose density is higher than that of plasma-membrane fragments alone. Transglutaminase cannot be removed from such complexes by 1% Triton X-100, but can be to a relatively large extent by 0.5 M-KCl and by 50% (w/v) glycerol. 5. Such results suggest that the apparent association of transglutaminase with plasma membrane originates from binding in vitro of the cytosolic enzyme to plasma membrane bound to collagen, which takes place during homogenization of the tissue, when the soluble enzyme and extracellular components are brought together.     

Cell cycle-dependent changes in tissue transglutaminase mRNA levels in bovine endothelial cells.

The pattern of transglutaminase gene expression through the cell cycle was examined by Northern blot analysis using cultured bovine endothelial cells and a cDNA probe. When the cells reached confluency or were arrested in G0/G1 phase by nutrition deprivation, transglutaminase mRNA rose to a very high level; S- and M-phase extracts showed high and low levels, respectively. Subcellular localization studies by sucrose gradient centrifugation and immunostaining demonstrated that the majority of transglutaminase is present in cytosols throughout the cycle. The cell cycle-dependent changes in the transglutaminase mRNA levels strongly support the implicated involvement of the enzyme in cell growth, differentiation, and senescence.     

Human erythrocyte transglutaminase: purification and preliminary characterisation

Erythrocyte transglutaminase was purified by anion-exchange chromatography, size exclusion and affinity chromatography. Homogeneity was achieved by an additional step of HPLC size-exclusion chromatography. The molecular mass of the purified enzyme was calculated to be 65,000 Da by size-exclusion chromatography and sucrose-gradient centrifugation, and 92,000 Da by SDS-PAGE, thus suggesting a high degree of asymmetry. The amino-acid composition of erythrocyte transglutaminase differed substantially from that of the guinea-pig liver enzyme, notably with respect to the number of histidine, cysteine and acidic amino-acid residues. The enzyme has an absolute requirement for divalent cations for activity: calcium, manganese, and the lanthanides terbium and gadolinium activate the enzyme in decreasing order of efficacy, while no activity is displayed in the presence of magnesium. In the presence but not in the absence of calcium ions, the enzyme is rapidly inactivated by N-ethylmaleimide and by diethylpyrocarbonate suggesting that the cation influences the reactivity of amino acids essential for catalysis. When erythrocyte proteins are employed as amine acceptors in the presence of calcium, the erythrocyte transglutaminase appears to preferentially modify membrane-associated proteins, although, in the absence of calcium ions and exogenous amines, it displays a pH-dependent interaction with soluble proteins.     

Rat gastrointestinal transglutaminase: demonstration of enzyme activity and cell and tissue distributions

The properties, tissue and cellular distribution of intestinal transglutaminase have been investigated. Transglutaminase was assayed with dimethylcasein and [14C]putrescine as substrates. The enzyme has maximum activity at pH 10, although more reliable assays are made at pH 9. Transglutaminase showed an absolute requirement for Ca2+ and exhibited linear assay kinetics. The Km for putrescine was approx. 0.15 mmol/l. Tissue distribution studies suggest transglutaminase is more active in the more muscular segments of the gut. The cellular localization in jejunum was investigated by sequential cell release techniques. Approximately 2 per cent of the total activity was found in the enterocytes and crypt cells. Most of the activity was in the submucosa and serosa suggesting an interstitial cell localization. Acute hypoplastic enteropathy induced by methotrexate was accompanied by a striking decrease in mucosal transglutaminase but the activity returned to control values by 72 h. There was no significant increase in activity during the period of intense crypt cell hyperplasia and it is concluded that intestinal transglutaminase is not implicated in crypt cell proliferation.     

In vivo cross-linking of nucleosomal histones catalyzed by nuclear transglutaminase in starfish sperm and its induction by egg jelly triggering the acrosome reaction.

A histone heterodimer, designated as p28, which contains an Nepsilon(gamma-glutamyl)lysine cross-link between Gln9 of histone H2B and Lys5 or Lys12 of histone H4, is present in starfish (Asterina pectinifera) sperm. Treatment of sperm nuclei with micrococcal nuclease produced soluble chromatin, which was size-fractionated by sucrose-gradient centrifugation to give p28-containing oligonucleosome and p28-free mononucleosome fractions, indicating that the cross-link is internucleosomal. When sperm nuclei were incubated with monodansylcadaverine, a fluorescent amine, in the presence or absence of Ca(2+), histone H2B was modified only in the presence of Ca(2+). Gln9, in the N-terminal region, was modified, but the other Gln residues located in the internal region were not, suggesting that the modification takes place on the surface of the nucleosome core by the in situ action of a Ca(2+)-dependent nuclear transglutaminase. Treatment of sperm with the egg jelly, which activates Ca(2+) influx to induce the acrosome reaction, resulted in a significant elevation of the p28 content in the nucleus. This is the first demonstration of an in vivo activation of transglutaminase leading to the formation of a cross-link in intracellular proteins.     

Characterization of purified rat testicular transglutaminase and age-dependent changes of the enzyme activities

The Ca2+-dependent tissue transglutaminase is widely distributed in various tissues and has been reported to participate in many cellular growth and differentiation processes. In the past decade, tissue transglutaminase is also identified as a G protein, G(alphah), for intercellular signaling. To further characterize testicular transglutaminase, the rat testicular transglutaminase was purified by ammonium sulfate precipitation, DEAE ion-exchange, heparin-agarose, and GTP-agarose affinity chromatographies. This purification protocol resulted in a 8400-fold enrichment of the enzyme with a reproducible 15% yield. The purified enzyme showed as a single band of 78kDa on SDS-polyacrylamide gel. Western blot analysis using anti-liver tissue transglutaminase monoclonal antibody also recognized the enzyme, indicating it is a t-TGase in nature. The Km values of purified testicular transglutaminase for putrescine and N,N-dimethylcasein were determined to be 35 and 17 microM, respectively. Its transglutaminase cross-linking activity was strongly inhibited by EGTA, GTP, polyamines, and cystamine, as well as moderately by ATP and NaCl. The enzyme exhibited a magnesium-dependent GTP-hydrolyzing capacity, but its GTP-binding activity did not require magnesium. Furthermore, the enzyme activity was found to be closely related with the first wave of spermatogenesis. Thus, testicular transglutaminase is speculated to participate in the event of spermatogenesis. In conclusion, the purified testicular transglutaminase displays property of either the tissue-type transglutaminase, or the GTP-binding and hydrolyzing characteristics. The activity of testicular transglutaminase is age-dependent, greatly stimulated during the first wave of spermatogenesis.     

Subcellular localization of a membrane-associated transglutaminase activity in rat liver

Fractionation of rat liver by homogenization and differential centrifugation revealed that only about 83% of the transglutaminase activity in the tissue is in a soluble form, and that the remainder is associated with the particulate fraction. This latter activity remained with the membranes even after they were extensively washed to remove 99% of such soluble enzymes as lactate dehydrogenase and aldolase. Subsequent fractionation of the membranes by isopycnic density gradient centrifugation in sucrose resulted in a single band of transglutaminase activity at a density of 1.194 g/cm3. This activity was coincident with the major band of plasma membranes, which was identified by its content of 5'-nucleotidase, alkaline phosphodiesterase I, alkaline phosphatase and leucine aminopeptidase activities. After treatment with digitonin and fractionation on sucrose gradients, the transglutaminase activity and the plasma membrane marker enzyme activities were found at a new density of 1.210 g/cm3, while the enzyme markers for the other membrane fractions remained unchanged. From these data, we conclude that approximately 17% of the transglutaminase activity in rat liver is specifically associated with the plasma membranes.     

Plasma membrane transglutaminase and cornified envelope competence in cultured human keratinocytes

When confluent cultures of the transformed human keratinocyte line SV-K14 are shifted to serum-free medium the cells achieve, within 4 days, the ability to synthesize a cornified envelope after challenge with the Ca2+ ionophore A23187. During these 4 days the enzyme transglutaminase (EC 2.3.2.13), which catalyses the cross-linking of different envelope precursor proteins, is partially transferred from the cytosolic pool into the plasma membrane. The association of the enzyme with the plasma membrane proves to be an essential step in the envelope formation since a direct correlation between plasma membrane-bound transglutaminase and envelope competence is observed. Retinoids block the insertion of the enzyme and therefore prevent envelope formation.     

A role for transglutaminase in glucose-stimulated insulin release from the pancreatic beta-cell.

Preincubation of rat islets of Langerhans with the potent inhibitors of islet transglutaminase activity, monodansylcadaverine (30-100 microM) and N-(5-aminopentyl)-2-naphthalenesulphonamide (100-200 microM), led to significant inhibition of glucose-stimulated insulin release from islets. In contrast, the respective N'-dimethylated derivatives of these two compounds, which did not inhibit islet transglutaminase activity, were much less effective as inhibitors of glucose-stimulated insulin release. None of the compounds inhibited rat spleen protein kinase C activity at concentrations which gave rise to inhibition of glucose-stimulated insulin release. When tested for their effects on calmodulin-stimulated bovine heart phosphodiesterase activity, of the compounds that inhibited insulin release, only monodansylcadaverine did not act as an effective antagonist of calmodulin at concentrations (up to 50 microM) that gave rise to significant inhibition of glucose-stimulated insulin release. Furthermore, at 50 microM, monodansylcadaverine did not inhibit methylation of islet lipids. The inhibition of glucose-stimulated insulin release by monodansylcadaverine is therefore likely to be attributable to its interference with islet transglutaminase activity. The sensitivity of islet transglutaminase to activation by Ca2+ was investigated by using a modified assay incorporating dephosphorylated NN'-dimethylcasein as a substrate protein. The Km for Ca2+ obtained (approx. 3 microM) was an order of magnitude lower than previously reported for the islet enzyme [Bungay, Potter & Griffin (1984) Biochem. J. 219, 819-827]. Mg2+ (2 mM) was found to have little effect on the sensitivity of the enzyme to Ca2+. Investigation of the endogenous substrate proteins of islet transglutaminase by using the Ca2+-dependent incorporation of [14C]methylamine into proteins of islet homogenates demonstrated that most of the incorporated radiolabel was present in cross-linked polymeric aggregates which did not traverse 3% (w/v) acrylamide gels. The radiolabelled polymeric aggregates were present in 71 000 g-sedimented material of homogenates, and their formation was transglutaminase-mediated. These findings provide new evidence for the involvement of islet transglutaminase in the membrane-mediated events necessary for glucose-stimulated insulin release.     

Stimulus-response coupling in human platelets. Changes evoked by platelet-activating factor in cytoplasmic free calcium monitored with the fluorescent calcium indicator quin2.

Rat pancreatic islets contain a Ca2+-activated and thiol-dependent transglutaminase (EC 2.3.2.13) comparable in activity with that found in rat liver, lung and spleen. The Ca2+-dependence of this enzyme is such that half-maximal velocity was obtained in the region of 40 microM. Preincubation of rat islets with primary-amine substrates of transglutaminase (monodansylcadaverine, methylamine, ethylamine, propylamine and cystamine) led to an inhibition of glucose-stimulated insulin release by these amines. Kinetic analysis of the competitive substrates methylamine, monodansylcadaverine, propylamine and ethylamine for their ability to inhibit islet transglutaminase activity indicated a potency that matched their ability to inhibit glucose-stimulated insulin release. When these amines were tested for their effects on glucose-stimulated protein synthesis and glucose utilization, the most potent inhibitor of insulin release, monodansylcadaverine, had no effect on either process at 100 microM. The amines cystamine, ethylamine, methylamine and propylamine had variable effects on these metabolic processes. For ethylamine, methylamine and propylamine, concentrations were found which inhibited glucose-stimulated insulin release in a manner which was found to be independent of their effects on either glucose oxidation or protein synthesis. Primary amines may therefore inhibit insulin release through their incorporation by islet transglutaminase into normal cross-linking sites. A role for protein cross-linking in the secretory mechanism is suggested.     

Ligand-induced conformational changes in tissue transglutaminase: Monte Carlo analysis of small-angle scattering data

Small-angle neutron and x-ray scattering experiments have been performed on type 2 tissular transglutaminase to characterize the conformational changes that bring about Ca(2+) activation and guanosine triphosphate (GTP) inhibition. The native and a proteolyzed form of the enzyme, in the presence and in the absence of the two effectors, were considered. To describe the shape of transglutaminase in the different conformations, a Monte Carlo method for calculating small-angle neutron scattering profiles was developed by taking into account the computer-designed structure of the native transglutaminase, the results of the Guinier analysis, and the essential role played by the solvent-exposed peptide loop for the conformational changes of the protein after activation. Although the range of the neutron scattering data is rather limited, by using the Monte Carlo analysis, and because the structure of the native protein is available, the distribution of the protein conformations after ligand interaction was obtained. Calcium activation promotes a rotation of the C-terminal with respect to the N-terminal domain around the solvent-exposed peptide loop that connects the two regions. The psi angle between the longest axes of the two pairs of domains is found to be above 50 degrees, larger than the psi value of 35 degrees calculated for the native transglutaminase. On the other hand, the addition of GTP makes possible conformations characterized by psi angles lower than 34 degrees. These results are in good agreement with the proposed enzyme activity regulation: in the presence of GTP, the catalytic site is shielded by the more compact protein structure, while the conformational changes induced by Ca(2+) make the active site accessible to the substrate.     

Two antigenic sites of tissue transglutaminase.

The immunization of rabbits with purified guinea pig liver transglutaminase resulted in the appearance of two antibody populations against the enzyme: one which reacted only with the Ca2+-enzyme complex and another which reacted with the intact as well as the Ca2+-enzyme. The Ca2+-induced confomrational change of the enzyme molecule exposes a new antigenic determinant which initiates the production of a specific antibody population. When the glutamine substrate of the enzyme was a dipeptide, the result of the interaction of the Ca2+-enzyme and its isolated specific antibody was an apparent activation of the catalytic activity. However, when protein substrates were used, an inhibition was observed. The characterization of the mechanism of the activation and the inhibition has led to the conclusion that the consequence of the interaction of Ca 2+-enzyme and its specific antibody is not only a limited steric hindrance of the active center but, besides that, a stabilization of the otherwise labile Ca2+-enzyme. The other antibody population reacts with both forms of the enzyme and its inhibitory effect, which has been observed in each assay, could be due to a prevention of the Ca2+-induced formation of the active enzyme.     

The subcellular localization of transglutaminase in normal liver and in glucagon-treated and partial hepatectomized rats

Rat liver was homogenized in isotonic sucrose and subjected to analytical subcellular fractionation by sucrose density-gradient centrifugation. Transglutaminase, when assayed with putrescine and dimethylcasein as substrates, showed three distinct localizations, cytosol (73%), plasma membrane (20%), and nuclei (7%). The distribution was unaffected by homogenization in the presence of potassium chloride, indicating that the particulate activity was not due to adsorbed cytosolic enzyme. The specific activity and subcellular distribution of transglutaminase in rats which had received intra-peritoneal glucagon, stimulating endocytosis, or which had been subjected to sub-total hepatectomy 2, 16, or 32 h previously, showed no significant difference from control animals.     

Ca2+: a stabilizing component of the transglutaminase activity of Galphah (transglutaminase II)

Galphah (transglutaminase type II; tissue transglutaminase) is a bifunctional enzyme with transglutaminase (TGase) and guanosine triphosphatase (GTPase) activities. The GTPase function of Galphah is involved in hormonal signaling and cell growth while the TGase function plays an important role in apoptosis and in cross-linking extracellular and intracellular proteins. To analyze the regulation of these dual enzymatic activities we examined their calcium-dependence and thermal stability in enzymes from several cardiac sources (mouse heart, and normal, ischemic and dilated cardiomyopathic human hearts). The GTP binding activity of Galphah was markedly inhibited by Ca2+ whereas the TGase activity was strongly stimulated, suggesting that Ca2+ acts as a regulator, switching Galphah from a GTPase to a TGase. The TGase function of Galphah of both mouse and human hearts was more thermostable in the presence of Ca2+.     

Expression of the cytosolic and particulate forms of transglutaminase during chemically induced rat liver carcinogenesis

Transglutaminase (EC 2.3.2.13) activity in chemically induced rat hepatocellular carcinomas was reduced by some 65% when compared to normal rat livers. The majority of the remaining activity (approx. 85%) was found in the particulate fraction. The use of non-ionic detergent to extract the transglutaminase activity present in both normal and tumour tissue followed by its separation on a Mono-Q column revealed two distinct peaks of activity. These peaks of activity were equivalent to those previously identified as a membrane-bound transglutaminase and the more characteristic cytosolic or tissue transglutaminase. The ratio of the activity of the cytosolic enzyme to that of the membrane-bound enzyme in normal liver was calculated as 5:1. In hepatocellular carcinomas, this ratio was reduced to 0.4:1. No significant change in the activity of the membrane-bound enzyme was detectable in tumour tissue. Comparison of the cytosolic enzyme found in hepatocellular carcinomas with that found in normal liver indicated no change in its molecular weight, Km,app for putrescine incorporation into N,N'-dimethylcasein and sensitivity to activation by Ca2+. These observations suggest that the reduction in transglutaminase activity observed in the hepatocellular carcinoma is due to a selective reduction in the expression of the cytosolic transglutaminase.     

The kinetics of the thermal deactivation of transglutaminase from guinea-pig liver

By incubating native (N) transglutaminase from guinea-pig liver at various temperatures and assaying it at 25 degrees C, two steps in the irreversible deactivation process to the denatured form (D) have been found. The fitting of the data to the equations of two possible models (the two-steps model and the two-isoenzymes model) is only compatible with the first one (N----X----D). It is shown that the structure of the active intermediate, X, depends on the deactivation temperature and on the thermal history of the enzyme. This may mean that transglutaminase exists in a large number of microstates. Surprisingly, the activation energy of deactivation is lower than that of activity (36.6 +/- 3.4 against 47.2 +/- 2.2 kJ.mol-1). By deactivating transglutaminase at a constant temperature (55 degrees C) and assaying it at variable temperatures, the activation energy of the intermediate, (X55), has been determined to be 40.2 +/- 5 kJ.mol-1, of the same order of magnitude as the native form. Among several agents assayed, only Ca2+ had a positive effect on the thermal stability of this enzyme. At 40 degrees C, transglutaminase was quite stable in the presence of Ca2+ (in its absence, the half-life was 65 min) and at 45 degrees C, its thermostability had been considerably increased, the half-life being raised from 47 min to 275 min.     

Regulation of endometrial transglutaminase activity during the menstrual cycle

When human endometrial transglutaminase was measured a 10-fold higher activity was detected during the secretive phase. This change was not related to either differences in solubility of the enzyme or to selective contamination by plasma factor XIII and rather appears to depend on the expression of the tissue form of transglutaminase, suggesting that this enzyme is regulated in vivo by progesterone.     

Hierarchies of protein cross-linking in the extracellular matrix: involvement of an egg surface transglutaminase in early stages of fertilization envelope assembly

The involvement of transglutaminase activity in fertilization envelope (FE) formation was investigated using eggs from the sea urchin, Strongylocentrotus purpuratus. Eggs fertilized in the presence of the transglutaminase inhibitors, putrescine and cadaverine, had disorganized and expanded FEs with inhibition of the characteristic I-T transition. The permeability of the FE was increased by these agents, as revealed by the loss of proteins from the perivitelline space and the appearance of ovoperoxidase activity in supernates from putrescine- treated eggs. [3H]putrescine was incorporated into the FE during fertilization in a reaction catalyzed by an egg surface transglutaminase that could also use dimethylcasein as a substrate in vitelline layer-denuded eggs. Egg secretory products alone had no transglutaminase activity. The cell surface transglutaminase activity was transient and maximal within 4 min of activation. The enzyme was Ca2+ dependent and was inhibited by Zn2+. We conclude that sea urchin egg surface transglutaminase catalyzes an early step in a hierarchy of cross-linking events during FE assembly, one that occurs before ovoperoxidase-mediated dityrosine formation (Foerder, C. A., and B. M. Shapiro. 1977. Proc. Natl. Acad. Sci. USA. 74:4214-4218). Thus it provides a graphic example of the physiological function of a cell surface transglutaminase.     

Proteolytic release of keratinocyte transglutaminase.

Human keratinocytes express a particulate transglutaminase that can be released from the membrane by limited proteolysis with trypsin or plasmin to yield a form that is congruent to 80 kDa. The enzyme from cultured cells was also releasable by endogenous proteolysis to yield a catalytically active fragment of congruent to 80 kDa. Endogenous release was strongly dependent upon temperature and Ca2+ concentration and was inhibited by iodoacetate, but not by leupeptin, antipain or phenylmethanesulphonyl fluoride. These phenomena raise the possibility of partial translocation of transglutaminase activity to the cytoplasm by proteolysis to which the enzyme is subject during terminal differentiation. In addition, hydrodynamic measurements showed that the endogenously released enzyme was monomeric in solution (79 kDa), whereas that solubilized by hydroxylamine without proteolysis appeared dimeric (190 kDa). The latter dimeric state may reflect either an altered conformation of the enzyme or post-translational modification beyond fatty acid esterification.     

Secretion of active-form Streptoverticillium mobaraense transglutaminase by Corynebacterium glutamicum: processing of the pro-transglutaminase by a cosecreted subtilisin-Like protease from Streptomyces albogriseolus

The transglutaminase secreted by Streptoverticillium mobaraense is a useful enzyme in the food industry. A fragment of transglutaminase was secreted by Corynebacterium glutamicum when it was coupled on a plasmid to the promoter and signal peptide of a cell surface protein from C. glutamicum. We analyzed the signal peptide and the pro-domain of the transglutaminase gene and found that the signal peptide consists of 31 amino acid residues and the pro-domain consists of 45 residues. When the pro-domain of the transglutaminase was used, the pro-transglutaminase was secreted efficiently by C. glutamicum but had no enzymatic activity. However, when the plasmid carrying the S. mobaraense transglutaminase also encoded SAM-P45, a subtilisin-like serine protease derived from Streptomyces albogriseolus, the peptide bond to the C side of 41-Ser of the pro-transglutaminase was hydrolyzed, and the pro-transglutaminase was converted to an active form. Our findings suggest that C. glutamicum has potential as a host for industrial-scale protein production.