Mutation of a Critical Arginine in the GTP-binding Site of Transglutaminase 2 Disinhibits Intracellular Cross-linking Activity*

Transglutaminase type 2 (TG2; also known as Gh) is a multifunctional protein involved in diverse cellular processes. It has two well characterized enzyme activities: receptor-stimulated signaling that requires GTP binding and calcium-activated transamidation or cross-linking that is inhibited by GTP. In addition to the GDP binding residues identified from the human TG2 crystal structure (Liu, S., Cerione, R. A., and Clardy, J. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, 2743-2747), we have previously implicated Ser171 in GTP binding, as binding is lost with glutamate substitution (Iismaa, S. E., Wu, M.-J., Nanda, N., Church, W. B., and Graham, R. M. (2000) J. Biol. Chem. 275, 18259-18265). Here, we have shown that alanine substitution of homologous residues in rat TG2 (Phe174 in the core domain or Arg476, Arg478, or Arg579 in barrel 1) does not affect TG activity but reduces or abolishes GTP binding and GTPγS inhibition of TG activity in vitro, indicating that these residues are important in GTP binding. Alanine substitution of Ser171 does not impair GTP binding, indicating this residue does not interact directly with GTP. Arg579 is particularly important for GTP binding, as isothermal titration calorimetry demonstrated a 100-fold reduction in GTP binding affinity by the R579A mutant. Unlike wild-type TG2 or its S171E or F174A mutants, which are sensitive to both trypsin and μ-calpain digestion, R579A is inherently more resistant to μ-calpain, but not trypsin, digestion, indicating reduced accessibility and/or flexibility of this mutant in the region of the calpain cleavage site(s). Basal TG activity of intact R579A stable SH-SY5Y neuroblastoma cell transfectants was slightly increased relative to wild-type transfectants and, in contrast to the TG activity of the latter, was further stimulated by muscarinic receptor-activated calcium mobilization. Thus, loss of GTP binding sensitizes TG2 to intracellular calcium concentrations. These findings are consistent with the notion that intracellularly, under physiological conditions, TG2 is maintained largely as a latent enzyme, its calcium-activated cross-linking activity being suppressed allosterically by guanine nucleotide binding.

Transglutaminase type 2 (TG2), 2 also known as tissue transglutaminase or G h (high molecular weight GTP-binding protein), is a multifunctional protein that is involved in diverse physiological processes, including apoptosis, bone ossification, wound repair, cell adhesion, and signal transduction and in the pathophysiology of various diseases, including gluten-induced enteropathy (celiac disease), neurodegenerative disorders, tumor growth, and diabetes (1). It is constitutively expressed by fibroblasts and endothelial and smooth muscle cells as well as a number of organ-specific cell types (2). At the subcellular level, it is found in the cytosol in association with plasma and nuclear membranes and is also found attached to the extracellular matrix (1).
TG2 has at least five distinct biological activities. It is a member of the transglutaminase family of cross-linking enzymes that catalyze posttranslational covalent linkages, the best studied of which is transamidation. Transamidation results in either a protein cross-link between a glutamine and lysine residue to form an N⑀-(␥-glutamyl) isopeptide bond, incorporation of an amine into a glutamine residue, or acylation of a lysine residue. TG2 also binds and hydrolyzes GTP (3,4),thereby mediating signaling by various G-protein-coupled receptors, including ␣ 1B and ␣ 1D -adrenergic (5-7), thromboxane A2 (8,9), and oxytocin (10) receptors. Additionally, TG2 can act as an adaptor protein that facilitates extracellular interaction between fibronectin and ␤1/3-integrins (11), as a protein disulfide isomerase (12), and as a kinase for insulin growth factor-binding protein-3 (13), p53 (14), and histones (15).
The functional switch between the transamidation (TG) and GTP binding activities of TG2 is intricately regulated by calcium and GTP. Calcium activates TG activity, and GTP inhibits TG activity at subsaturating calcium concentrations (3,16). In the presence of physiological intracellular concentrations of calcium and GTP, it has been suggested that TG2 exists largely as a latent transamidase (16 -18) but as an active G-protein involved in receptor signaling.
The sensitivity of TG activity to the inhibitory effect of GTP is increased by post-translational modification, e.g. nitrosylation of cysteine residues by nitric oxide-releasing agents (19). In contrast, sphingosylphosphocholine, a relatively minor membrane phospholipid component that is increased in cells undergoing apoptosis, increases the sensitivity of TG2 to calcium (20). In pathological states, when cells become leaky to calcium and GTP generation is impaired, the TG activity of TG2 contributes to programmed cell death events, such as apoptotic body stabilization (17,21). Dysregulation of the activities of TG2, therefore, could potentially have catastrophic cellular consequences due, for example, to unrestrained protein cross-linking and activation of pro-apoptotic pathways.
Of the nine distinct, but closely related, members of the TG family (TG 1-7, Factor XIIIA and Band 4.2), only TG2, 3, (22), and 5 (23) have been shown to bind guanine nucleotides. Recent crystallographic studies have revealed the tertiary structure of the inactive form of human TG2 bound to GDP (24), which, like all other members of the TG family, is comprised of four domains: an N-terminal sandwich domain, a core domain that contains the TG active site catalytic triad (Cys 277 , His 335 , and Asp 358 ) and transition state stabilizing residue (Trp 241 ) (25), and two ␤-barrel domains (Fig. 1A). The GDP-binding site of TG2 is located in a hydrophobic pocket between the core and ␤-barrel 1, on the opposite face to the proposed glutamyl substrate-binding site (24). It has a unique architecture that, although conserved in TG3 (22), is entirely distinct from that of other GTP-binding proteins and involves mainly residues from ␤-barrel 1 (amino acids 476 -482 and 580 -583) and Phe 174 from the core. This is in agreement with our previous work, which localized GTP binding to residues 159 -173 of the core and residues 465-589 of ␤-barrel 1 and additionally showed impaired GTP binding and hydrolysis by site-directed mutants of Lys 173 (26). Although not contained within the GDP binding pocket delineated by the crystal structure, our previous studies also implicated Ser 171 in GTP binding, because such binding was lost with the substitution of this residue by a glutamate, the homologous residue in a non-GTP binding member of the mammalian TG family, fXIIIA (26).
The present study shows that several key residues of TG2 implicated in GDP binding in the crystal structure are indeed important for guanine nucleotide binding, because mutation reduces or abolishes GTP binding as well as GTP␥S inhibition of TG activity in vitro. Ser 171 does not interact directly with GTP, because nucleotide binding is unaltered with alanine substitution. Among the GTP binding residues, we have identified a particularly important residue, Arg 579 in rat TG2 (equivalent to Arg 580 in human TG2). Mutation of this residue to alanine not only renders it resistant to cleavage by the intracellular calcium-activated protease, -calpain, but also abolishes GTP regulation of TG activity in intact cells, resulting in dysregulated intracellular TG activity. This study, therefore, has provided the first physiological confirmation that the intracellular TG activity of TG2 is regulated by GTP.

EXPERIMENTAL PROCEDURES
Constructs, Cell Culture, and Transfection of Cells-Site-directed mutants were constructed in rat TG2 cDNA as described (27) using glutathione S-transferase-TG2/pGEX2T (28) as the template. Wildtype (WT) and mutant rat TG2 cDNAs were subcloned into the EcoRI-NotI sites of pcDNA3.1 and stably transfected into human neuroblastoma SH-SY5Y cells as described (29). Vector (pcDNA3.1) DNA was stably transfected as a control. After transfection, cells were maintained in RPMI medium with 10% dialysed fetal bovine serum, 2 mM L-glutamine, and 100 g/ml G-418.
Isothermal Titration Calorimetry-Samples of purified WT and mutant TG2 were prepared by dialysis against high magnesium buffer (25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 5 mM MgCl 2 , 2 mM DTT) for 16 h at room temperature, followed by exhaustive dialysis at 4°C against isothermal titration calorimetry (ITC) buffer (25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 1 mM MgCl 2 , 0.5 mM DTT). GTP␥S (Roche Applied Science) was dissolved in ITC dialysis buffer, and the concentration was checked by absorbance at 256 nm (⑀ ϭ 1. Blank runs were also performed in which TG2 was omitted. Base-linesubtracted data were fitted by a single-site model using Origin ITC Analysis software (Microcal Software, Northampton, MA). Data for R579A TG2 were weaker and fitted with fixed stoichiometries (n ϭ 0.3-1), and an average estimated K d from these fits is presented.
Data Analysis-Results shown are the means Ϯ S.E. Statistical differences were determined by analysis of variance followed by Bonferroni's post-test, with p Յ 0.05 being considered significant.

Mutation of GDP-binding Site Residues
Affects GTP Binding-We mutated some of the key GDP-binding site residues (Phe 174 , Arg 476 , Arg 478 , and Arg 579 ) as well as Ser 171 in rat TG2 (Fig. 1A) and tested their ability to bind GTP. Phe 174 in the core domain forms part of a hydrophobic pocket around the guanine base, where it is thought to stabilize one side of the guanine ring through an aromatic stacking interaction (24). In agreement with this interpretation, removal of the phenylalanine side chain by substitution with alanine abolished [␣-32 P]GTP photolabeling, a qualitative measure of GTP binding, whereas tryptophan substitution retained wild-type binding (Fig. 1B). In ␤-barrel 1, Arg 580 , one of four positively charged residues surrounding the phosphate groups of GDP in the crystal structure, forms two ion pairs with the ␣and ␤-phosphates of GDP, as well as interacting with the guanine base (24). Mutation of Arg 579 of rat TG2 (equivalent to human Arg 580 ) to alanine abolished [␣-32 P]GTP photolabeling, as did a more conservative mutation to lysine (the homologous residue in the GTP binding TG member, TG5 (23)), which retains the charge and much of the bulk of showing the four domains, N-terminal ␤-sandwich (blue), core (magenta), ␤-barrel 1 (dark green), and ␤-barrel 2 (light green). GDP is shown in red. Right, the GDP-binding site of human TG2. Core domain backbone, magenta; ␤-barrel 1 domain, green; carbon atoms of mutated residues, gray, and those of GDP, light blue. Functional groups include oxygen, red; amino groups, deep blue; phosphorous, magenta. Hydrogen bonds and ion pair interactions are shown as dashed lines. Arg 580 in human TG2 is equivalent to Arg 579 in rat TG2. B, photoaffinity labeling of WT or mutant TG2 with GTP was carried out as described under "Experimental Procedures." Gels were subjected to autoradiography to discern radiolabeling (top panel) and to Coomassie Blue staining to visualize the proteins directly (bottom panel). C, isothermal titration calorimetry of GTP␥S binding to TG2. GTP␥S was titrated into solutions of WT, F174A, and R579A as described under "Experimental Procedures." Raw data from injections is shown in the upper panels. Peaks were integrated to give the heat change associated with each injection, and buffer control was subtracted to give the data plotted in the lower panels. the arginine side chain. This confirms the importance of the arginine residue for nucleotide binding. In contrast, substitution of alanine for Arg 478 , which interacts with the ␤-phosphate, reduced but did not abolish [␣-32 P]GTP photolabeling. Similarly, alanine substitution of Arg 476 (Fig. 1B) or leucine substitution of Lys 173 (26), residues postulated to stabilize the ␥-phosphate of GTP during binding and hydrolysis (24), caused only a minor reduction in GTP photolabeling. The double mutant R476A/R478A photolabeled similarly to R478A alone. Although glutamate substitution of Ser 171 prevents GTP binding (26), alanine substitution at the same position had no effect on GTP photolabeling, indicating Ser 171 does not interact directly with GTP. Taken together, these results indicate that residues interacting only with phosphate groups contribute less to guanine-nucleotide binding than residues interacting with the guanine base.
Isothermal titration calorimetry was used to quantitate the GTP binding affinities of the F174A and R579A mutants relative to that of WT TG2, as both of the mutants showed significant loss of GTP binding by photolabeling. Although it has been shown that TG2 binds GTP with a 1:1 stoichiometry (32), the best fit to our initial calorimetry data gave a stoichiometry value of around 0.1 for GTP binding to both WT and F174A, indicating that only ϳ10% of the protein was available to bind nucleotide. Native gel electrophoresis showed that almost all of the samples migrated in the faster GTP-bound form, as previously demonstrated (27), indicating that the proteins were competent to bind GTP and that most of the protein was already nucleotide bound. Attempts to strip nucleotide from the proteins by dialysis against 2.5 M KCl (33) resulted in the appearance of high molecular weight bands on native gels indicating aggregation, and less than 5% of the protein was available to bind GTP. As we have shown previously that TG2 binds less GTP in the presence of 5 mM MgCl 2 (26), samples were dialysed against a buffer containing 5 mM MgCl 2 prior to exchange into ITC buffer (as described under "Experimental Procedures"), resulting in ϳ30% of the protein being available to bind GTP, and better quality data were obtained. WT TG2 bound GTP␥S with a K d of 1.6 M, whereas the affinities of F174A (K d , 2.9 M) and R579A (K d , 150 M) for GTP␥S were around 2-and 100-fold weaker, respectively (Fig. 1C). These dissociation constants were very similar to those obtained initially without magnesium pretreatment.
In Vitro Inhibition of TG Activity by GTP Correlates with GTP Binding Affinity-All mutants exhibited dose-dependent Ca 2ϩ -stimulated TG activity comparable with WT ( Fig. 2A and Table 1), indicating that the proteins were correctly folded and that their TG activity was not compromised by the mutation(s). GTP inhibition of TG2 activity is inversely proportional to Ca 2ϩ activation and is greatest under conditions where TG activity is minimal (26). GTP␥S inhibition of TG activity was evaluated in the presence of 75-125 M free Ca 2ϩ (which stimulates ϳ40% of maximal activity; Fig. 2, B and C). Under these conditions, GTP␥S inhibited the TG activity of WT TG2 with an IC 50 of 27 M. The TG activity of R476A and R478A was less sensitive to GTP␥S inhibition than WT (Fig. 2B), yielding IC 50 values Ͼ350 M (R476A) and 450 M (R478A). The TG activity of R476A/R478A was not measurably inhibited by GTP␥S (Fig. 2B; IC 50 Ͼ500 M), indicating that mutation of both arginines further reduced GTP binding. The TG activities of the S171E (26), F174A, and R579A mutants were insensitive to GTP␥S inhibition (Fig. 2C). Thus, loss of GTP inhibition of TG activity correlated with the degree of [␣-32 P]GTP photolabeling of the mutants (Fig. 1B).
Resistance of GTP Binding Mutants to Trypsin and Calpain Cleavage-As demonstrated previously (3) and as shown in Fig. 3A, resistance of WT TG2 to proteolytic digestion by trypsin is increased with increasing concentrations of GTP␥S. Consistent with little or no GTP binding, the sensitivity of the S171E and R579A mutants to trypsin digestion was unaltered even at 1 mM GTP␥S (Fig. 3A), whereas F174A, which has  2). B and C, GTP␥S inhibition of TG activity was assayed at ϳ40% maximal TG activity (symbols as in panel A). Activities as a percentage of maximal activity were 36% for WT, 43% for R478A, and 49% for R579A at 75 M free Ca 2ϩ ; 48% for R476A, 47% for R476A/R478A, and 40% for F174A at 98 M free Ca 2ϩ ; and 49% for S171E at 124 M free Ca 2ϩ . Data are presented as means Ϯ S.E. of one experiment performed in triplicate.

TABLE 1 Kinetic parameters for calcium activation of TG activity
TG calcium-activation kinetics for WT TG2 or mutant TG2 proteins (10 mM each) were determined as described under "Experimental Procedures." Kinetic values (EC 50 , the concentration resulting in 50% of maximal activation) were calculated from best-fit plots using GraphPad Prism. Data are presented as means Ϯ S.E. of one representative experiment performed in triplicate. 2-fold lower affinity for GTP than WT (Fig. 1C), was resistant to digestion at GTP␥S concentrations greater than 50 M. Similarly, GTP␥S prevented cleavage of WT TG2 by -calpain (34) (Fig. 3B). Again consistent with reduced GTP binding, the increased resistance to -calpain digestion in the presence of GTP␥S was moderate for F174A and negligible for S171E and R579A (Fig. 3B). Interestingly, and in contrast to trypsin digestion, R579A demonstrated some inherent resistance to -calpain digestion. In the absence of GTP␥S, R579A was more resistant to calpain digestion than WT, and in the presence of GTP␥S R579A was as resistant as F174A. This is likely due to the alanine substitution of Arg 579 resulting in a local conformational change that renders the resulting protein more resistant to calpain, but not trypsin, cleavage.

Protein
R579A TG2 Has Dysregulated TG Activity in Intact Cells-To determine the effect of the R579A mutation on intracellular TG activity, SH-SY5Y human neuroblastoma cells were stably transfected with either pcDNA3.1 (vector control) or pcDNA3.1 containing WT or R579A TG2 cDNA. Two independent clonal cell lines of each type were analyzed. Lysates from clones transfected with WT or R579A TG2 showed robust TG2 expression (Fig. 4A) and TG activity (Fig. 4B) relative to vector-transfected clones. Expression of endogenous human TG2, which runs on SDS-PAGE at a higher molecular weight than rat TG2 (35), was unaltered (Fig. 4A).
Stably transfected clones were stimulated with the muscarinic cholinergic receptor agonist carbachol (18), which acts on endogenous M3 receptors to stimulate calcium release from the endoplasmic reticulum via phospholipase C-mediated production of inositol 1,4,5-trisphosphate. Although both WT-and R579A-transfected clones had higher basal TG activity than vector-transfected clones because of the increased amount of TG2 protein in these cells, neither vector-nor WT-transfected clones showed significantly increased TG activity in response to low dose carbachol stimulation (Fig. 4C). R579A-transfected clones, on the other hand, displayed slightly increased basal TG activity compared with WT-transfected clones, as well as significantly increased TG activity upon low dose carbachol stimulation (Fig. 4C). Western blot analysis indicated an absence of TG2 protein degradation in WT-or R579A-transfected cells during the course of carbachol stimulation (Fig. 4D). Intracellular calcium concentrations were assessed using Fura-2 and found not to be statistically significantly different at base line or at the plateau after peak stimulation, although WT-transfected clones generally showed a higher calcium spike than R579Atransfected clones (Fig. 4E). Chronic treatment with low doses of calcium ionophore A23187 (36), which transports calcium across the cell membrane, activated the TG activity of both WT-and R579A-transfected clones (Fig. 4F). Taken together, these results demonstrate that the loss of GTP binding by R579A TG2 renders the TG activity of R579A-transfected clones more sensitive than that of WT-transfected clones to small increases in intracellular calcium such as those induced by low dose carbachol treatment. This loss of GTP inhibition of R579A TG activity thus results in dysregulated TG activity in vivo.

DISCUSSION
Alanine substitution of individual ␤-barrel 1 domain residues Arg 476 , Arg 478 , and Arg 579 and the core domain loop residue Phe 174 impaired GTP binding by TG2 (Figs. 1-3). Residues binding the guanine ring (Arg 579 and Phe 174 ) are individually more important for GTP binding than residues postulated to stabilize the ␥-phosphate (Lys 173 (26), Arg 476 ) or those that stabilize the ␤-phosphate (Arg 478 ). Arg 579 is particularly important for binding GTP, as evidenced by its 100-fold lower GTP binding affinity (Fig. 1C) and the insensitivity of its TG activity to GTP␥S inhibition, both in vitro (Fig. 2C) and in intact cells (Fig. 4C).
Consistent with ITC results for F174A, which showed ϳ2-fold weaker affinity for GTP␥S than WT (Fig. 1C), F174A exhibited some resistance to protease cleavage in the presence of GTP␥S (Fig. 3). However, F174A did not photolabel (Fig. 1B), and its TG activity was insensitive to GTP␥S inhibition (Fig. 2C). This apparent discrepancy is most likely because of the different assay conditions and may be explained by the documented inhibitory effect of Mg 2ϩ on GTP binding in the standard photolabeling assay and the inverse relationship between Ca 2ϩ activation and GTP inhibition of TG activity (26).
Ser 171 in the core domain does not appear to interact directly with GTP, as S171A was unaffected in GTP binding (Fig. 1B). The dramatic loss of GTP binding by S171E (26) (Fig. 2C) is thus likely due to an indirect conformational effect on the positioning of Lys 173 and Phe 174 , which are C-terminal to Ser 171 on the same ␤-strand.
Unlike the rat isoform, human S171E is poorly expressed and/or lacks TG activity (37,38), which has led to the hypothesis that TG activity is stabilized by, and critically dependent on, GTP binding (38). This is clearly not the case for rat TG2, as indicated by both our in vitro ( Fig. 2A) and intact cell (Fig. 4F) findings, which demonstrate TG activity comparable with that of the WT protein not only for S171E but for all GTP binding-deficient mutants examined. This is also not the general case for human TG2, as the R580A mutant (equivalent of rat R579A) has robust TG activity in intact cells. 3 Thus, the reduced TG activity/expression of human S171E may be peculiar to this particular mutant but in any case is unrelated to loss of GTP binding. . In vivo TG activity of SH-SY5Y cells stably transfected with R579A TG2 is dysregulated. A and B, lysates from two independent SH-SY5Y clonal cell lines, each stably transfected with either pcDNA3.1 (vector control) or pcDNA3.1 plus WT or R579A rat TG2 cDNA, were centrifuged (10 min, 10,000 ϫ g, 4°C). A, supernatants (2.5 g) were transferred to a Western blot with purified recombinant human (h, 40 ng) or rat (r, 20 ng) as standards, and the blot was developed using anti-TG2 monoclonal antibody CUB 7402. B, supernatants (15 g) were assayed for TG activity, calculated as nanomoles of putrescine incorporated/mg of protein/h; data for vector clones 3 and 8, WT clones 2 and 41, and R579A clones 3 and 6 are means Ϯ S.E. of five, four, four, one, three, or four experiments, respectively, performed in triplicate. C, stably transfected SH-SY5Y cells were treated for 90 min with 0, 0.1, or 0.25 mM carbachol, and TG activity of intact cells was quantitated by enzyme-linked immunosorbent assay as described under "Experimental Procedures" using 5 g of crude cell lysates. Data are means Ϯ S.E. of one experiment performed in triplicate and are representative of three, nine, nineteen, five, six, and nineteen experiments performed in triplicate using vector clones 3 and 8, WT clones 2 and 41, and R579A clones 3 and 6, respectively. Statistically significant differences, *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, are indicated by asterisks above the bars (vector versus WT or R579A) or brackets (ns, not significant). D, crude cell lysates (5 g) from panel C were transferred to the same Western blot, and the blot was developed with an anti-TG2 polyclonal antibody. Immunoblots shown are from the same autoradiograph and have been separated to allow alignment with activity measurements in panel C. E, effect of 0.25 mM carbachol stimulation on intracellular calcium levels measured using Fura-2 as described under "Experimental Procedures." Data are means Ϯ S.E. from six-eight replicates for each stable clone. Statistically significant differences, # or *, p Ͻ 0.05; **, p Ͻ 0.01; ### or ***, p Ͻ 0.001, are indicated above the bars (asterisk, relative to vector 3; hash, relative to vector 8) or brackets. Inset, representative 6-min trace of Fura-2-loaded cells showing base-line (1 min), peak, and plateau fluorescence after 0.25 mM carbachol stimulation (arrow). F, stably transfected SH-SY5Y cells were treated for 60 min with 0, 0.1, or 0.25 M A23187 and in vivo TG activity assayed as described under "Experimental Procedures." Data are means Ϯ S.E. of one experiment performed in triplicate. Statistically significant differences, **, p Ͻ 0.01; ***, p Ͻ 0.001 are indicated by asterisks above the bars (vector versus WT or R579A) or brackets.