A novel transglutaminase-mediated post-translational modification of phospholipase A2 dramatically increases its catalytic activity.

Transglutaminases (TG), which include coagulation Factor XIIIa, are calcium-dependent ubiquitous enzymes. TGs catalyze the formation of an isopeptide bond by cross-linking a specific glutamine and a lysine residue between two proteins or within the same protein molecule. Phospholipase A2 (PLA2) is a key enzyme in the regulation of prostaglandin and leukotriene biosynthetic pathways, which catalyzes the release of free fatty acids from the sn-2 position of membrane glycerophospholipids. This enzyme has been suggested to be pathophysiologically related to the initiation and propagation of several inflammatory diseases including juvenile rheumatoid and rheumatoid arthritis. Here, we describe a novel TG-catalyzed post-translational modification of PLA2 which dramatically increases the activity of this enzyme. This increase was dependent upon the time of preincubation, the concentration of TG and the presence of Ca2+. Size exclusion chromatography of TG-treated PLA2 yielded two peaks of PLA2 activity, with apparent molecular masses of 26 and 13 kDa, respectively. The 26-kDa species, a putative PLA2 dimer, contained epsilon-(gamma-glutamyl)-lysine isopeptide in about 1:1 molar ratio to PLA2, suggesting an intramolecular rather than intermolecular cross-linking. This hypothesis was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the 26- and 13-kDa species under denaturing conditions. The specific activity of the dimeric peak was 10-fold higher with respect to that of the monomeric enzyme. These data suggest that TG-catalyzed covalent cross-linking of PLA2 is intramolecular and that this may promote a noncovalent dimerization and subsequent activation of this enzyme via a conformational change. To our knowledge, this is the first demonstration that TG-mediated post-translational modification of an enzyme (e.g. PLA2) causes a striking increase in the catalytic activity of the enzyme.


Transglutaminases
(TG), which include coagulation Factor XIIIa, are calcium-dependent ubiquitous enzymes. TGs catalyze the formation of an isopeptide bond by cross-linking a specific glutamine and a lysine residue between two proteins or within the same protein molecule.
Phospholipase Aa (PLA2) is a key enzyme in the regulation of prostaglandin and leukotriene biosynthetic pathways, which catalyzes the release of free fatty acids from the m-2 position of membrane glycerophospholipids.
This enzyme has been suggested to be pathophysiologically related to the initiation and propagation of several inflammatory diseases including juvenile rheumatoid and rheumatoid arthritis.
Here, we describe a novel TG-catalyzed post-translational modification of PLAz which dramatically increases the activity of this enzyme. This increase was dependent upon the time of preincubation, the concentration of TG and the presence of Ca2+. Size exclusion chromatography of TG-treated PLA2 yielded two peaks of PLA2 activity, with apparent molecular masses of 26 and 13 kDa, respectively.
This hypothesis was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the 26-and 13-kDa species under denaturing conditions. The specific activity of the dimeric peak was lo-fold higher with respect to that of the monomeric enzyme. These data suggest that TG-catalyzed covalent crosslinking of PLA2 is intramolecular and that this may promote a noncovalent dimerization and subsequent activation of this enzyme via a conformational change. To our knowledge, this is the first demonstration that TG-mediated post-translational modification of an enzyme (e.g. PLA2) causes a striking increase in the catalytic activity of that enzyme.
Transglutaminases (TG,' EC 2.3.2.13) are a class of enzymes which catalyze a Ca*+-dependent acyl-transfer reaction in which the y-carboxamide group of a peptide-bound glutamine residue is the acyl-donor (l-4). Primary amino groups of many amines may act as acyl-acceptors with the formation of mono-substituted y-carboxamides of peptide-bound glutamic acid. In the absence of small molecular weight amines, a cross-linking between endo-y-glutaminyl and endo-c-lysyl residues in polypeptides takes place (l-4). These enzymes have been detected both intra-and extracellularly in higher animals including man. Following blood coagulation, a plasma TG cross-links fibrin molecules via the formation of interchain c-(y-glutamyl)-lysine isopeptide bonds and stabilizes the clot by preventing its hydrolysis by proteases (5)(6)(7). This particular form of TG is variously known as fibrin-stabilizing factor, Laki-Lorand factor or coagulation Factor XIII. The active form of this enzyme (Factor XIIIa) is generated by the action of thrombin and Ca2+ on the catalytically inactive zymogen. Besides Factor XIII, several extracellular and intracellular TGs have been described (l-4).
One of the most thoroughly studied TGs is derived from guinea pig liver (3). Intracellular TGs with apparently identical properties to the liver enzyme are present in many tissues and organs of mammals, and these enzymes are generally known as "tissue TG." Besides fibrin clot stabilization, TG-catalyzed reactions have been proposed to be involved in the formation of seminal plug in rodents (8), cataract formation (9), aging of the erythrocyte membrane (lo), and in neuronal aging (11). Additionally, TGmediated reactions have been suggested to be involved in masking the immunogenicity of rabbit spermatozoa in the female genital tract following coitus (12) and to protect implanting embryos from maternal immunological assault (13). These results have been confirmed using rat spermatozoa (14). Moreover, it has been suggested that the maturation of human spermatozoa may involve a TG-mediated process (15). Recently, the involvement of TG in the terminal differentiation of keratinocytes has been reported (16). It has also been proposed that TG participates in the biochemical events leading to programmed cell death (apoptosis) (17). Phospholipases A2 (PLA2) (EC 3.1.1.4) are a family of lipolytic enzymes which specifically catalyze the hydrolysis of the 2-acyl ester bond in 3-sn-phosphoglycerides, including naturally occurring phospholipids, in the presence of Ca*+ (18). The most abundant sources of these enzymes are the mammalian pancreas and snake venoms (18). At the present time many of these enzymes have been extensively studied and complete amino acid sequences of more than 35 PLAzs are known (18). Characterization of a large number of these enzymes from different phyla has revealed that the members of the PLAz family are closely related. In fact, diversification has occurred with conservation of several important features which include: specificity for hydrolysis of the sn-2 position acyl group of 3-sn-phosphoglycerides, requirement for Ca*', a minimal molecular mass of 14 kDa, differential enzymatic reactivity toward substrates, a large number of intrachain disulfide bridges and a very high degree of amino acid se-Transglutaminases Activate Phospholipase A2 quence homology (18). Conservation of amino acid sequence has been discovered in more than 35 phospholipases from different sources, including those mammalian intracellular PLAzs the sequences of which have been determined so far (19)(20)(21)(22). The regions of PLA, protein structure that are generally well conserved are: the amino-terminal amphiphilic helix, the Ca*+ binding loop and the active site (18). In all the PLA, structures that have been analyzed except for the bee venom enzyme (18), the amino-terminal sequences revealed the same a-helical structure and the same pattern of amino acid residues. Recently, the crystallographic analysis of a PLA2 inhibitory protein, blastokinin (23) or uteroglobin (24) at 1.34-A resolution has revealed a remarkable similarity between the calculated molecular surfaces of this protein and that of PLA2 (25). Uteroglobin inhibits porcine pancreatic and macrophage PLA2 in vitro (26). Uteroglobin has also been shown to be an excellent substrate of TG in uitro (27), and TG treatment dramatically potentiates some of its biological activities (E-14).
On the basis of these data we decided to investigate if PLA2 is a substrate of TG and if TG treatment could affect the catalytic properties of PLA2. Here, we show that porcine pancreatic PLA2 can be dramatically activated when pretreated with guinea pig liver TG or with coagulation Factor XIIIa. The activation is dependent upon the presence of Ca*+, the incubation time and the concentrations of TG and PLA2. This dramatic activation of PLAz after TG treatment was confirmed by using two entirely different assay systems. The formation of an t-(y-glutamyl)-lysine isopeptide bond in TG-treated PLA, was established by the quantitation of t-(y-glutamyl)-lysine.
This isopeptide bond was found to be intramolecular and it involved Gln-4, the only glutamine residue present in the entire porcine pancreatic PLAz molecule. The cross-linked PLA2 formed noncovalent dimers which may be responsible for its increased enzymatic activity. in the same buffer.

Enzymes
The column was eluted with the buffer mentioned above, at 0.5 ml/min, and 0.5-ml fractions were collected. The elution of proteins was monitored by absorbance at 280 nM.
Aliquots from each fraction were tested for PLA, activity. The purified material was used for isopeptide analysis.

Activation of Porcine Pancreatic PLA2 by Guinea Pig Liver
TG and by Rabbit and Human Factor XIIIa-Our initial approach was to investigate the possible effects of preincubation with various TGs on the catalytic activity of PLA2.   Fig. 2. The maximum activity obtained with guinea pig TG reached 350% of control at 1 h. In the absence of added Ca'+, PLA, stimulation was dramatically reduced, although not entirely abolished. In fact, some stimulation was still observed without added Ca2+ after 50 min of preincubation. This may be due to the presence of Ca2+ bound to TG and/or to PLA*. In fact, when TG was tested for enzymatic activity in the absence of added Ca2+ at the concentration used in the experiments described in Fig. 2A, it still retained about 25% of the enzymatic activity previously observed with 1 mM Ca'+ (data not shown). We avoided the use of EGTA, EDTA, or other chelating agents in the control preincubation mixture because PLAz activity is strongly dependent upon the concentration of Ca'+. As a further control, TG was denatured by boiling and then was used to perform an identical experiment in the presence of 1 mM Ca'+. In this case no increase in PLA, activity was observed ( Fig. 2A). Also in this set of experiments, human plasma Factor XIIIa seemed to produce a better stimulation of PLA, activity than guinea pig TG (Fig. 2B) at confirming the observed stimulation of PLA2 by TG and to explore some characteristics of the stimulated reaction. The time course of product formation by TG-and Factor XIIIa-treated and untreated PLA, is shown in Fig. 3. In both cases (Fig. 3, a and b) no appreciable lag period was observed, and the rate of product accumulation is clearly much higher with TG-and Factor XIIIa-treated PLA, than in control experiments.
It should be noted that the absolute values of percent hydrolysis obtained in the experiments with Factor XIIIa (Fig. 35) are lower than those obtained in the experiments with guinea pig liver TG (Fig. 3a). Since this effect was consistently observed in experiments involving thrombinactivated Factor XIII, it is probably due to the presence of thrombin, or some contaminant from the thrombin preparation. In fact, the preparation of human thrombin used in these experiments contained citrate, which may have interfered with the free Ca2+ concentration in the assay mixture. The curves are clearly biphasic, consisting of an initial phase of rapid hydrolysis followed by a prolonged phase of slower hydrolysis.
The onset of the reaction is virtually instantaneous, without any lag phase. This is commonly the case in mixed micellar systems (39). The shape of the curves is reminiscent of the time courses obtained with C. atrox PLAP and egg phosphatidylcholine vesicles as the substrate (40). In that case, the decrease in reaction rate was shown to be due to product inhibition by monomolecular and micellar lysophosphatidylcholine (40). We have not determined if product inhibition occurs in our system. If this is the case, monomolecular lysophosphatidylcholine may be responsible for the phenomenon, since the concentrations of lysophosphatidylcholine produced in our conditions are far lower than the published value (4.3 pM) of the critical micellar concentration for egg lysophosphatidylcholine (40). The rate of PLAz reaction was increased by TG and Factor XIIIa in both the initial and the slower phase of the time course (Fig. 3, a and 5). This is even more evident in the semilogarithmic plots (Fig. 3, c and d). The slopes of the two components in these plots were used as an approximate estimate of reaction rate, without any assumption about the overall order of reaction. We designated these slopes as kl (for the fast component) and kl (for the slow component).
From the data shown in Fig. 3c the calculated values for control reactions were kl = -1.96 X low4 and kS = -2.79 x 10m5 while for TG-stimulated PLA2 the values were k, = -4.99 x lo-* and kp = -1.12 X 10e4. From the data shown in Fig. 3d the values were k1 = -1.65 x 10e4 and k2 = -2.29 x 10V5 for the control curve; kl = -3.04 X 10e4 and k2 = -6.90 x 10m5 for Factor XIIIa-treated PLA2. Both rates appear to be increased by TG and by Factor XIIIa, with the increment of k2 being slightly more pronounced than that of kl in both cases.
We also tested whether TG-induced stimulation of PLAP activity could be observed by using an entirely different assay with a detergent-free substrate. For these experiments we used a modification of the procedure described by Haigler et al. (34,35), which involves the use of autoclaved E. coli membranes metabolically labeled with ["Hloleic acid as PLA, substrate. We studied the time course of PLA, reaction at two different temperatures and the results of this group of experiments are shown in Fig. 4. At 37 "C, no lag period was observed in the hydrolysis of E. coli phospholipids (Fig. 4A). However, the reaction catalyzed by TG-pretreated PLA2 was much faster, reaching about 28% hydrolysis in 5 min, while the same value was reached in 15 min by TG-untreated PLA2. In addition, the time course of hydrolysis by TG-treated enzyme was linear from time 0, while with TG-untreated enzyme the time course did not reach linearity until 10 min. Indeed, when the activity of TG-treated enzyme, expressed as percent of control uersus time was plotted (Fig. 4B), it was evident that the main difference between control and TGtreated enzyme was in the first 10 min of the reaction. When the assays were performed at 12 "C ( Fig. 4C), the control curve showed a triphasic behavior, very similar to the one described by  for the hydrolysis of dimiristoylphosphatidylcholine vesicles by porcine pancreatic PLAz at 17 "C. In fact, after a small initial burst, a long latency phase was observed where essentially no product accumulated until 10 min when the reaction started to progress again. Interestingly, under the same conditions TGpretreated PLA, showed no latency phase, although the rate of hydrolysis did increase after 10 min. Accordingly, the activity of TG-treated enzyme, expressed as percent of control, increased from 0 to 10 min, and decreased slightly thereafter (Fig. 40). These data indicated that TG-treated PLA, also had an enhanced enzymatic activity on detergentfree substrates. Moreover, these observations gave us some indirect information on the mechanism of stimulation of PLA, activity. The observed higher activity of TG-treated PLA, during the first 2 min of reaction at 37 "C and particu- larly the absence of a true latency phase at 12 "C, suggest that TG-treated enzyme had a much faster "penetration" and/or "interfacial activation" than untreated enzyme (42). These data may be interpreted to suggest that TG-catalyzed activation of PLA, could effectively bypass, and/or dramatically increase the speed of the activation step that takes place upon binding of PLA, to organized substrates (18,42).
Mechanism of TG-induced PLAz Stimulation-Although it was clear that a TG-catalyzed reaction was activating PLA*, there still remained several possibilities which needed to be explored in order to delineate the more probable mechanism(s) of this activation.
These included: (i) the formation of an intermolecular t-(y-glutamyl)-lysine isopeptide bond between molecules of PLAz or between one molecule of PLA, and TG; (ii) a hydrolytic reaction with deamidation of Gln-4 in PLA,; and (iii) the formation of an intramolecular c-(yglutamyl)-lysine cross-link within the same PLA2 molecule. Our initial approach was to incubate PLA2 with TG under appropriate conditions and then analyze the reaction mixture by size exclusion chromatography, so that individual fractions could be assayed for PLA* activity. A preliminary experiment was performed in which PLA, (5 nM) was incubated with 5 milliunits of TG for 1 h in a total volume of 720 ~1. All other conditions were identical to those used for the experiments described in Fig. 2a. The incubation mixture was analyzed by size exclusion chromatography on Sephacryl S-200, and each fraction was tested for PLA2 activity. In this experiment, PLA, activity eluted in two poorly resolved peaks of approximately equal height, with apparent molecular masses of about 26 and 10 kDa, respectively (data not shown). Since the molecular mass of porcine pancreatic PLA, is about 14.9 kDa, this was an indication that PLAr may have been partially modified by TG into a higher molecular weight form. This experiment was then repeated with a higher concentration of PLAz (140 nM) and the reaction volume was doubled (Fig.  5A). Again, after size exclusion chromatography PLA? activities were eluted as two poorly resolved peaks with apparent molecular masses of 26 and 13 kDa, respectively. When PLA:! preincubated with buffer only was run through the same column, it eluted as a single peak at 14 kDa. To improve the resolution, we reduced the flow rate of the column from 6 to 4 ml/h, and repeated the experiment. Fig. 5b shows that in this case both peaks of PLA, activity were retarded with respect to the profile shown in Fig. 5A, probably because of adsorption of this enzyme to the gel matrix due to the low ionic strength used. While peak 1 was only slightly retarded, peak 2 eluted with the total bed volume of the column, allowing a virtually complete separation of the two peaks. To test whether the appearance of peak 1 was due to a TGcatalyzed reaction, we repeated the experiment with TG which was inactivated by boiling for 30 min (Fig. 5B). In this case, peak 1 was virtually absent, except for a slight elevation of Size exclusion chromatography was performed using a Sephacryl S-200 superfine column (see "Materials and Methods"). The column was calibrated as shown in the inset. n, the flow rate of the column was of 6 ml/h. PLA, assay was performed on each fraction (20 ~1) at 37 "C for 10 min.   the base line. Essentially all PLA, activity eluted as a peak corresponding to peak 2. We tentatively interpreted these results as indirect evidence that the formation of a PLA? dimer was being catalyzed by TG. Therefore, we decided to confirm this hypothesis by testing peaks 1 and 2 for the presence of t-(y-glutamyl)-lysine isopeptide. Peak 1 was found to contain 0.86 mole of e-(y-glutamyl)-lysine/mol of PLAL and that peak 2 contained a much lower amount (0.1 mol of t-(y-glutamyl)-lysine/mol of PLAJ. These data strongly supported the possibility that TG catalyzes the formation of an t-(y-glutamyl)-lysine isopeptide bond in PLAr. However, assuming that an intermolecular isopeptide bond is formed by TG, the highest theoretical molar ratio of t-(y-glutamyl)lysine to PLAe monomer would have been 0.5, and peak 2 should not have contained any t-(y-glutamyl)-lysine. One of the possible explanations may be that the t-(y-glutamyl)lysine cross-link was within the same molecule of PLA,. In such a case, the dimerization could be a noncovalent process promoted by the conformational changes induced by the intramolecular cross-link. This hypothesis was also consistent with other findings, such as the "skewed" aspect of the elution profile in Fig. 5A, and the noticeable decrease in the height of peak 1 after complete resolution from peak 2. The latter phenomena could be due to an alteration in the equilibrium between the monomer and dimer in solution as a result of the chromatographic process. To test this hypothesis, we attempted a semipreparative separation of peak 1, so that electrophoretic analysis under denaturing conditions and isopeptide analysis could be performed using the material from the same peak. After size exclusion chromatography on Sephacryl S-200, peak 1 was further purified by Superose 12 chromatography. Also in this column, PLAe activity eluted with an apparent molecular mass of 25-30 kDa. When FPLCpurified PLALz from peak 1 was analyzed by SDS-PAGE under denaturing conditions, it was found to be apparently homogeneous and its molecular weight was indistinguishable from that of TG-untreated PLAZ (Fig. 6). Fig. 6 shows the results of electrophoretic analysis and Western blot of peak 1 after Sephacryl S-200 and Superose I2 chromatography. The purified material from Superose 12 chromatography was found to contain t-(y-glutamyl)-lysine in a molar ratio of about 0.89 to 1 to PLAe. These data strongly supported the possibility that TG catalyzed the formation of an intramolecular c-(7glutamyl)-lysine isopeptide bond in PLAe. The formation of this intramolecular cross-link appeared to induce an increased tendency for PLAr to form dimers in solution. Dimerization Transglutaminases Activate Phospholipase A2 of porcine pancreatic PLA? has been suggested to be the basis of interfacial activation of this enzyme (18,39,(43)(44)(45)(46). Thus, it is not unreasonable to speculate that this noncovalent dimerization of intramolecularly cross-linked PLA2 may play a role in the observed TG-induced increase of PLA, activity. To test this hypothesis, we measured the specific activity of the putative PLA, dimer eluting in peak 1 after TG treatment and Sephacryl S-200 chromatography.
We treated PLAz (690 nM) with 45 milliunits of guinea pig liver TG in a total volume of 1.4 ml for 1 h at 37 "C and then separated peaks 1 and 2 by size exclusion chromatography on Sephacryl S-200. These conditions were found to provide a good compromise between clearcut resolution of the peaks and detectability of protein in the eluate. It should be noted that the relative proportion of the total PLA2 activity eluting in peak 1 was found to increase considerably with the concentration of TG-treated PLA2 loaded. The relative proportion of PLA, activity eluting in peak 1 reached 85% of the total recovered PLA2 activity in the preparative experiments (data not shown). This is consistent with a concentration-dependent self-association of TG-treated PLA,. When the specific activity of the two peaks was measured with the deoxycholate/phosphatidylcholine mixed micellar substrate, peak 1 (the putative dimer) was found to be about lo-fold more active than peak 2 (monomer). Although this observation cannot be considered definitive evidence that the dimerization of cross-linked PLA2 was the cause of its increased activity, it is consistent with such a hypothesis.
Michaelis-Menten kinetics of TG-treated and untreated PLAz were performed with the mixed micellar assay, since in such systems hyperbolic Michaelis-Menten curves are usually obtained (18,42,47). We used phosphatidylcholine concentrations between 2.8 and 90 pM, with a fixed concentration of 1 mM deoxycholate.
Keeping the deoxycholate concentration fixed prevented possible artifacts due to interference of the detergent carboxylate groups with the pH and/or the free Ca2+ concentration.
In addition, this allowed us to avoid possible changes in the size or shape of the micelles at high detergent concentrations.
Since at the highest concentration of phosphatidylcholine used the molar ratio of deoxycholate: phosphatidylcholine was still >lO:l, we assumed that the "quality of the interface" and the size of micelles did not dramatically change within this range of phospholipid concentrations (47). Moreover, the total molarity of phospholipid + detergent, which is approximately proportional to the total area of the lipid-water interface (47), changed from 1.0029 to 1.090 mM over this range of concentrations of phospholipid, while the molar fraction [phospholipid]/[detergent + phospholipid] changed from 2.72 X 10e3 to 8.26 X lo-*. Therefore, for practical purposes we assumed the total area of the interface to be constant, so that the molar concentrations of phospholipid could be considered proportional to the interfacial concentrations in moles/unit of surface area. Under these conditions, we obtained straight double-reciprocal plots for PLA, preincubated with buffer and with TG, without chromatographic separation of the reaction products (not shown). Apparent K,,, values of 42 and 53 pM for control and TGtreated PLA?, respectively, were obtained. Apparent V,,, values were 0.32 pmol x liter-' x min-' for PLA, alone and 0.8 pmol x liter-' x min-' for TG-treated PLA,. From these experiments it appeared that the kinetic parameter which was significantly altered by TG was the apparent V,,,. The same conclusion was reached by analyzing these data by iterated weighted fit with the computer program ENZYME (48). With this program we obtained apparent K, values of 27.7 f 7.3 pM for PLA, and 23.6 f 10.9 FM for TG-treated PLA2, while V,,,,, values were 0.26 f 0.03 for untreated PLAz and 0.52 + 0.07 pmol x liter-' x mine1 for the TG-treated enzyme. The effect of TG on the V,,, was found by ENZYME to be statistically significant (p < 0.05). However, these results represent the kinetic properties of a mixture of (i) non-TGmodified monomers and (ii) TG-modified monomers and dimers of PLA2. Thus, we performed kinetic experiments using the putative PLAz dimer peak obtained after size exclusion chromatography on Sephacryl S-200. In this case the results were more complex. Double-reciprocal plots clearly deviated from linearity below 10 pM (not shown) suggesting a negative cooperativity or the presence of more than one class of active sites. Therefore, in calculating estimates of apparent K,,,(41.6 + 23.9 FM) and V,., (0.82 f 0.26 rmol X liter-' X min-') ENZYME attributed very low relative weights to the points below 10 pM. These values must therefore be considered only as approximate estimates. The Hill plot was also nonlinear (not shown), with limiting slopes of 0.44 and 1.4. These data might indicate that at low substrate concentrations (2-10 pM) only one active site per dimer binds the substrate. The apparent K, value of this site may be lower than that of monomeric PLA2. At higher substrate concentrations all active sites of PLA, tend to be saturated. However, the second site may have a lower affinity for phosphatidylcholine than the first site, thus causing an increase in the overall estimate of apparent K,,,. The apparent V,,,,, of the putative PLA, dimer is certainly higher than that of monomeric PLA, by a factor of at least 4-5-fold.
Effect of TG on PLAzs from Different Sources-To determine if the observed TG-induced activation was specific for porcine pancreatic PLA2 only, we tested if PLAzs from other species could also be stimulated by TG. The results of these experiments are summarized in Table I. Interestingly, a considerable stimulation was observed with the N. naja enzyme tested in three different concentrations, while no significant stimulation was observed with the C. atron enzyme. N. naja PLAz exists as a monomer in solution, and forms dimers only at concentrations of 4.5 pM or above (18). On the other hand, C. atrox PLA, is a dimeric protein at concentrations as low as 2 nM (18). Thus, these observations support the hypothesis that the mechanism by which TG-mediated post-translational modification of PLA2 stimulates PLAz activity is by promoting a noncovalent dimerization of this enzyme. Further support for this hypothesis was provided by the fact that the observed stimulation of N. nuja PLA, is dependent on the final concentration of the enzyme in the assay (Table I). This is consistent with a concentration-dependent self association of PLA*. Since some stimulation of N. naja PLAz was observed at final concentrations as low as 0.1 nM, this might indicate that the TG-induced shift in the equilibrium of dimerization of this enzyme was quite dramatic. Finally, it is noteworthy that some degree of stimulation was also observed with bee venom PLA2 at 0.2 nM. This enzyme could not be assayed at higher concentrations because its activity was too high, even in the absence of TG treatment. Bee venom PLA, lacks several amino acid residues present in the amino terminus of most pancreatic and snake venom enzymes (l&49), including Gln-4. This residue is most likely the acyl donor in the TG reaction (see "Discussion"). It has been recently proposed, on the basis of computer-generated alignments and other considerations (49) that bee venom PLAz could be evolutionarily related to the pancreatic and snake venom enzymes. In this model, a region in the COOH-terminal sequence of bee venom PLA2 was suggested to be structurally and functionally analogous to the amino-terminal a-helix of the pancreatic and snake venom enzymes. This sequence includes Gln-121 which may be a putative homologue of . Should bee venom PLAz prove to be a substrate for TG, it would be interesting to test whether Gln-121 is indeed the acyl donor and functionally a counterpart of Gln-4 of N. naja enzyme. All in all, our observations suggest that TGcatalyzed activation of PLA, is not restricted to porcine pancreatic PLA2 only, but may be a more generalized phenomenon.

DISCUSSION
The data presented in this paper demonstrate that both Factor XIIIa and tissue TG, when used to pretreat porcine pancreatic PLA2, dramatically increase the activity of this enzyme in a time, concentration and Ca'+-dependent manner. We have shown that guinea pig TG can catalyze the formation of an t-(y-glutamyl)-lysine isopeptide bond when PLA2 was used as a substrate. The data obtained by size exclusion chromatography, SDS-PAGE, and Western blot indicate that the isopeptide bond is intramolecular, and that the modified PLA, containing the c-(y-glutamyl)-lysine isopeptide has an increased tendency to form noncovalent dimers. This interpretation has been confirmed by isolating the putative dimeric PLA, and showing that the purified material: (i) contained e-(y-glutamyl)-lysine in a molar ratio of 0.89 to 1; (ii) when analyzed by size exclusion chromatography it showed an apparent molecular mass of 25-30 kDa; (iii) migrated as a monomer with an apparent molecular mass of 15 kDa when analyzed by SDS-PAGE under denaturing conditions; and (iv) reacted with an anti-PLAz antibody after electrophoresis and immunoblot.
These data strongly support the hypothesis that the TG-catalyzed post-translational modification of PLA, is the chemical basis for the observed increase in PLA, enzymatic activity. TG-modified PLAz seems to have a faster rate of interfacial activation and a faster overall rate of catalysis than unmodified PLA2. Steady state kinetics of TGmodified PLA2 in our system appear to be complex. The TGmodified enzyme shows a higher apparent V,,, than TGuntreated PLA,. However, it is possible that its affinity for the phospholipid substrate may also be higher, particularly at low substrate concentration (<lo PM). Transglutaminases are ubiquitous but very selective enzymes in terms of the glutamine residues that can serve as acyl donors (2,3,50). In those proteins which are TG substrates usually only one specific glutamine residue is utilized as acyl donor, even if more than one residue of glutamine is present. This extreme specificity for a particular glutamine residue suggests that TG-mediated post-translational modifications in vitro are likely to be functionally important in uiuo (50). Since Gln-4 is the only glutamine residue in porcine pancreatic PLA1, this residue is the only possible acyl donor for the TG-catalyzed reaction. Gln-4 is an almost invariant residue in the PLA, family across the phyla, including a recently cloned human lung PLAz (19). This could be another indication that TG-mediated post-translational modifications of PLA, might be of physiological importance. We did not identify a particular Lys residue involved in the formation of the intramolecular isopeptide bond. However, structural considerations suggest that this residue may be Lys-10.
It has been suggested that all PLA+, including those which exist in solution as monomers such as the pancreatic enzymes, could act on organized substrates in dimeric or aggregated form (1642). Strong kinetic evidence has been presented that the "interfacial activation" of porcine pancreatic PLAP involves dimerization of the enzyme in the presence of large unilamellar phospholipid vesicles (43) and of mixed vesicles containing deoxycholate (39,44). Recently, the chemical basis for this activation of porcine pancreatic PLA, has been shown to consist of a slow autocatalytic acylation of Lys-56, which in turn promotes dimerization (46). Our data demonstrate the presence of a novel enzymatic mechanism by which the slow autoactivation step may be bypassed or accelerated through a TG-mediated post-translational modification of PLA*. Like autoacylation, transglutamination promotes dimerization and activation of PLA*. It is tempting to speculate that in uivo tissue TG or Factor XIIIa may convert a slowly auto-activating PLAz into a pre-activated or rapidly autoactivating form with enhanced enzymatic activity. There are several physiological and pathological situations in which such a phenomenon could be important.
PLAz has been suggested to be involved in the onset and propagation of acute inflammation, and inflammatory cells such as the neutrophils release soluble PLAz upon activation (18,51,52). Macrophages also release arachidonate and PLAz upon phagocytosis (52). It is well known that inflammation and blood coagulation are intimately related. Since activation of the coagulation cascade may be triggered by inflammatory processes, it is not unreasonable to speculate that soluble PLAP released by phagocytes or by necrotic cells could be activated by Factor XIIIa. Moreover, activation of macrophages is accompanied by a striking increase in tissue TG activity (50,53). TG-activated PLAz in the extracellular spaces during inflammation could participate in: (i) releasing arachidonate from cell membranes, thereby potentiating the inflammatory response; (ii) killing microorganisms in the inflamed area; and (iii) lysing dead or dying cells in the exudate. It is also possible that intramolecular crosslinking of PLAP may be a mechanism for a rapid turnover of this active enzyme in uiuo. This may prevent the propagation of local inflammation to adjoining healthy tissues. As for the possible biological relevance of TG-catalyzed modification of intracellular PLAzs is concerned, it has been suggested that intracellular TG activity is related to terminal differentiation and cell death (17,50). It is entirely possible that this increased TG activity may also trigger the activation of intracellular PLA,. The irreversible activation of an intracellular PLA, could be part of a series of biochemical events leading to programmed cell death and lysis. In this regard, it is worth noting that PLA*s are active components of most animal poisons with cytolytic activity. Some poisons also contain peptides, such as melittin or mastoparan, which are well known activators of PLA:! (18). Moreover, it has been recently shown that cell death induced by the toxic peptide of vegetal origin Pyrularia thionin is mediated by Ca2+ influx and activation of an intracellular PLA, (54). It is possible that the final biochemical sequence of events leading to cell death may include an irreversible, TG-catalyzed activation of an intracellular PLA*, converting the latter enzyme into a more cytolytic species. It should be emphasized that TG-mediated activation of PLA2 is readily observed with PLA, concentrations as low as 5 nM (see "Results"). The fact that this reaction can take place at such low molar concentrations of PLA, may Trunsglutuminases Activate PhosphoEipase AS also suggest its physiological relevance. Moreover, in the case of a membrane-associated PLA, the actual concentration of this enzyme per unit of surface area could be much higher than the observed molar concentration. In this model an influx of Ca*+ would have the effect of activating TG and PLAQ, since both are Ca'+-dependent enzymes. Although this model is at present entirely speculative, we feel that the possible role of TG-mediated activation of PLA2 in inflammation and in programmed cell death deserves a thorough investigation.
There is compelling evidence to suggest that in inflammatory diseases, such as rheumatoid and juvenile rheumatoid arthritis, activation of PLAz may be pathophysiologically involved (52). The mechanism of PLA2 activation in these diseases is as yet unclear. Since inflammatory cells (e.g. neutrophils and macrophages) are known to secrete PLAz and this enzyme may cause synovial tissue damage releasing TG and further PLA2 activation may occur as this process continues. On the other hand, activation of extracellular PLA2 may also be catalyzed by Factor XIIIa present in the exudate. From the clinical standpoint it would be important to determine whether the synovial PLA, in rheumatoid and juvenile rheumatoid arthritis is a transglutaminated form.
Acknowledgments-We are grateful to Dr. Simone Beninati for performing the determination of the c-(y-glutamyl)-lysine isopeptide. We also thank Dr. Norman Dubin for a critical reading of the manuscript and Dr. Soo I1 Chung for the gift of purified rabbit Factor XIII.