Functional significance of five noncanonical Ca 2+ -binding sites of human transglutaminase 2 characterized by site-directed mutagenesis Ro ´ bert Kira ´ ly 1,2 ,E ´ va Cs } osz 2 , Tibor Kurta ´ n 3 ,Sa ´ ndor Antus 3 , Krisztia ´ n Szigeti 4 , Zso ´ fia Simon-Vecsei 2 , Ilma Rita Korponay-Szabo ´ 5,6 , Zsolt Keresztessy 2 and La ´ szlo ´ Fe ´ su ¨ s 1,2 1 Apoptosis and Genomics Research Group of Hungarian Academy of Sciences, Debrecen, Hungary 2 Department of Biochemistry and Molecular Biology, Medical and Health Science Center, University of Debrecen, Hungary 3 Department of Organic Chemistry, University of Debrecen, Hungary 4 Research Group for Membrane Biology, Hungarian Academy of Sciences, Semmelweis University, Budapest, Hungary 5 Department of Pediatrics, Medical and Health Science Center, University of Debrecen, Hungary 6 Heim Pal Children Hospital, Budapest, Hungary Introduction Transglutaminase 2 (TG2), also known as tissue trans- glutaminase or Gh protein (EC 2.3.2.13), is a unique multifunctional protein with diverse biological func- tions. It is present in various cell compartments, including the cytosol, the nucleus, and the plasma membrane. TG2 has been implicated in the regulation Keywords calcium binding; celiac epitope; GTPase activity; transglutaminase activity; transglutaminase 2 (tissue transglutaminase) Correspondence R. Kira ´ ly and L. Fe ´ su ¨ s, Department of Biochemistry and Molecular Biology, Medical and Health Science Center, University of Debrecen, Nagyerdei krt. 98, POB 6, Debrecen, Hungary H-4012 Fax: +36 52 314-989 Tel: +36 52 416-432 E-mail: kiralyr@dote.hu; fesus@dote.hu (Received 28 May 2009, revised 16 September 2009, accepted 1 October 2009) doi:10.1111/j.1742-4658.2009.07420.x The multifunctional tissue transglutaminase 2 (TG2) has a four-domain structure with several Ca 2+ -regulated biochemical activities, including transglutamylation and GTP hydrolysis. The structure of the Ca 2+ -binding form of the human enzyme is not known, and its Ca 2+ -binding sites have not been fully characterized. By mutagenesis, we have targeted its active site Cys, three sites based on homology to Ca 2+ -binding residues of epider- mal transglutaminase and factor XIIIa (S1–S3), and two regions with nega- tive surface potentials (S4 and S5). CD spectroscopy, antibody-binding assay and GTPase activity measurements indicated that the amino acid substitutions did not cause major structural alterations. Calcium-45 equili- brium dialysis and isothermal calorimetric titration showed that both wild- type and active site-deleted enzymes (C277S) bind six Ca 2+ . Each of the S1–S5 mutants binds fewer than six Ca 2+ , S1 is a strong Ca 2+ -binding site, and mutation of one site resulted in the loss of more than one bound Ca 2+ , suggesting cooperativity among sites. All mutants were deficient in transglutaminase activity, and GTP inhibited remnant activities. Like those of the wild-type enzyme, the GTPase activities of the mutants were inhib- ited by Ca 2+ , except in the case of the S4 and S5 mutants, which exhibited increased activity. TG2 is the major autoantigen in celiac disease, and test- ing the reactivity of mutants with autoantibodies from celiac disease patients revealed that S4 strongly determines antigenicity. It can be con- cluded that five of the Ca 2+ -binding sites of TG2 influence its transgluta- minase activity, two sites are involved in the regulation of GTPase activity, and one determines antigenicity for autoantibodies in celiac patients. Abbreviations FXIIIa, coagulation factor XIIIa; GSE, gluten-sensitive enteropathy (celiac disease); GST, glutathione-S-transferase; ICP-OES, inductively coupled plasma–optical emission spectrometry; ITC, isothermal titration calorimetry; TG2, transglutaminase 2; TG3, epidermal transglutaminase; TG4, prostate transglutaminase; WT, wild type. FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS 7083 of cell differentiation, apoptosis, phagocytosis, cell adhesion, wound healing, and the pathophysiology of various diseases, including celiac disease [gluten-sensi- tive enteropathy (GSE)], tumor growth, and neurode- generative disorders [1,2]. In GSE, a chronic enteropathy with multiple extraintestinal manifesta- tions and 1–2% prevalence in the general population, autoantibodies of the IgA (and IgG) class are produced against TG2 in response to ingestion of gluten proteins. These antibodies contribute to disease progression and also have great importance as dia- gnostic markers [3]. TG2 has several kinds of enzymatic activity. It was recognized as a Ca 2+ -activated transglutaminase enzyme performing post-translational protein modifica- tion via the incorporation of small amines into pro- teins, forming e-(c-glutamyl)lysine isopeptide bonds between polypeptide chains, but it can deamidate glutamine side chains as well [4]. GTP and GDP inhibit the transamidating activity of the enzyme [5]. It can bind GTP and ATP [6], and its role in mediating signal transduction through G-protein-coupled recep- tors, based on its GTPase activity, has also been shown [7]. In addition, TG2 demonstrates protein kinase [8] and protein disulfide isomerase [9] activities. It also acts as a BH3-only protein, interacting with proapoptotic factors [10]. Fibronectin-bound TG2 serves as a coreceptor for integrins, contributing to the adhesive functions of cells [11]. The actual enzymatic activity of TG2 is determined by its structural state induced by the types and amounts of bound ligands [12]. The X-ray structure of TG2 in its GDP-bound and substrate analog-bound form has been described [13,14]. On comparison of these two structures, a large difference can be observed. The beginning of this large conformational change is induced by calcium ions. However, the exact sites of bound calcium ions and their functional signifi- cance have not been determined. Within the transglutaminase enzyme family, the calcium-bound X-ray structures of only the human blood coagulation factor XIIIa (FXIIIa) and epider- mal transglutaminase (TG3) are known. FXIIIa has one Ca 2+ -binding pocket, where the main chain oxy- gen of Ala457 mediates Ca 2+ binding, and the other direct coordinators are five water molecules; Asn436, Asp438, Glu485 and Glu490 are also involved in the formation of this negatively charged site [15]. TG3 has three Ca 2+ -binding sites (S1–S3): S1 is formed by Asn221–Asn229 and a water molecule – at this site, TG3 permanently binds a sole Ca 2+ with high affinity that could derive from the actual expression system [16]; S2 is similar to the heptacoordinated Ca 2+ -binding site of FXIIIa, involving Asn393, Ser415, Glu443, Glu448 and two directly coordinated water molecules; and at S3, Ca 2+ is coordinated by Asp301, Asp303, Asn305, Ser307, Asp324, and a water molecule [17]. These three sites are also represented as conserved sequences in TG2. It has been shown [18] that human red blood cell TG2 can bind six Ca 2+ , suggesting that further binding sites exist. Indeed, TG2 has several negatively charged amino acids with high surface potential that might serve as Ca 2+ -binding sites [19]. The aim of the present study was to identify the exact Ca 2+ -binding sites of human TG2, by using site- directed mutagenesis and targeting sites homologous to FXIIIa and TG3, and two other sites with highly nega- tive surface potential. We examined changes in Ca 2+ binding characteristics of the generated mutants, and investigated the role of these sites in the regulation of TG2 enzymatic activities. Our results show that each of these sites contributes to Ca 2+ binding, and that transglutaminase activities were significantly decreased or totally lost when any of these sites were mutated. Two mutants demonstrate higher GTPase activity than the wild type (WT), and one of them shows very low affinity for celiac autoantibodies. Results Ca 2+ binding of human recombinant TG2 Even after exhaustive dialysis in EDTA-containing buffer, the bacterially expressed wild-type human TG2 contains 0.45 ± 0.03 mol Ca 2+ per mol WT, as detected by inductively coupled plasma–optical emis- sion spectrometry (ICP-OES). This finding suggests that the recombinant TG2 has a tightly bound Ca 2+ , which could be derived from the expression system, similarly to the case with recombinant TG3 [16]. This tightly bound Ca 2+ has an affinity for TG2 that is comparable to its affinity for EDTA. To determine the Ca 2+ -binding properties of the recombinant WT, equi- librium dialysis measurements were performed. The results showed that the WT can bind about six Ca 2+ (Fig. 1A), similarly to the native erythrocyte TG2 [18]. The calculated affinity constant of the hyperbolic saturation curve was 560 lm. Isothermal titration calorimetry (ITC) measurements confirmed our equilibrium dialysis and ICP-OES data (Fig. 1B). The curve of integrated heats shows 0.5 mol Ca 2+ binding to TG2 per mol protein, with high affin- ity (K d = 0.1 lm). The next five Ca 2+ bind with very low and comparable affinity to the enzyme. The observed difference between the Ca 2+ -bound active form and the inactive form of TG2 suggests a large Ca 2+ -binding sites of TG2 R. Kira ´ ly et al. 7084 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS conformational change during the Ca 2+ activation process, which could be accompanied by a significant entropy change, explaining the small enthalpy change. In the presence of Ca 2+ , the WT may work as a transamidase, even during the equilibrium dialysis and ITC experiments, and could crosslink itself to a differ- ent degree, even in the absence of any other substrates (Fig. S1). This self-crosslinking occurred to a much smaller extent during the 3–4 h of ITC measurements at 25 °C than during the 2 days of equilibrium dialysis at 4 °C. We also wished to clarify whether this process altered Ca 2+ -binding properties. Therefore, we exam- ined the C277S active site mutant, which lacks any transglutaminase activity and does not have the ability to crosslink itself [20]. On the basis of our equilibrium dialysis data, the C277S mutant also binds approxi- mately six Ca 2+ , although the binding is weaker (the affinity constant is 720 lm) than in the case of the WT (Fig. 1A). The active site mutant also showed the same ITC response as the WT (data not shown). These results demonstrate that self-crosslinking and eventual polymerization does not have any significant influence on Ca 2+ binding of the recombinant WT, and that the possible heat changes related to crosslinking were probably masked by other concurrent mechanisms. Design and preparation of site-directed mutants of TG2 On the basis of the high sequence homology shared by transglutaminases and the available X-ray structures of FXIIIa, TG3, and their identified Ca 2+ -binding sites [15–17], we used homology modeling and comparative molecular modeling to design seven TG2 mutants. In these, five different surface sites were altered by intro- ducing single or multiple point mutations (Fig. 2; primers are listed in Table S1). The S1 and S3 mutants were chosen on the basis of homology with TG3 Ca 2+ -binding sites (S1 and S3, respectively). The S2 mutants were planned on the basis of homology to the 0 1 2 3 4 0 1 2 3 4 5 6 7 8 C277S WT [Ca 2+ ] free (mM) Bound Ca 2+ (mol/mol) A 024 –0.5 –0.4 –0.3 –0.2 –0.1 0.0 Molar ratio Molar ratio Molar ratio kcal·mol –1 of injectant –0.10 –0.05 0.00 kcal·mol –1 of injectant 024681012 0123456 –0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1 0.0 kcal·mol –1 of injectant B Strong binding site of wild typeTG2 Weak binding sites of wild type TG2 Weak binding sites of S1 TG2 n = 0.52 ± 0.63 K d = 0.1 ± 0.03 µM DH = –421.5 ± 83.7 Kcal·mol –1 DS = 30.5 Kcal·mol –1 K d = 4.6 ± 1.3 µM DH = –58.02 ± 23 Kcal·mol –1 DS = 24.2 Kcal·mol –1 n = 5.46 ± 0.08 K d = 6.7 ± 3.2 µM DH = –784.5 ± 38 Kcal·mol –1 DS = 21.0 Kcal·mol –1 n = 1.7 ± 0.06 Fig. 1. Ca 2+ binding of recombinant wild-type, C277S mutant and S1 mutant TG2. (A) Ca 2+ -binding curve of wild-type and C277S mutant TG2 measured by equilibrium dialysis. (B) Net heat change of ITC of Ca 2+ binding to wild-type and S1 mutant TG2. The net heat change curve of wild-type TG2 was divided into two parts to improve the quality of curve regression. For the injection scheme, see Experimental procedures. R. Kira ´ ly et al. Ca 2+ -binding sites of TG2 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS 7085 Ca 2+ -binding site of FXIIIa, which has strong similar- ities to one of the TG3 Ca 2+ -binding sequences. In the case of S2 and S3, we generated two separate mutants (S2A and S2B mutants, and S3A and S3B mutants), as here the suspected Ca 2+ -binding sites are formed by two opposing loops. We have assumed that mutations of these sites as a whole may cause significant confor- mational changes by themselves, and should be avoided. S4 and S5 were selected on the basis of sur- face patches characterized by higher local density of negatively charged amino acids on TG2 [19,21]. Mostly conservative amino acid replacements were performed to target Ca 2+ binding specifically and to prevent significant conformational changes or structural disruptions. In most cases, only negative charges were removed (e.g. Glu to Gln, or Asp to Asn) or the 432 G R N Q R Q N I T432 G R D E R E D I TS5 149 Y L N S Q Q Q R Q Q Y149 Y L D S E E E R Q E YS4 326 D K S Q M I W N326 D K S E M I W NS3B 305 H N Q S S S L305 H D Q N S N LS3A 445 Y P Q G S S Q Q R Q A445 Y P E G S S E E R E AS2B 395 A Q V S A N V395 A E V N A D VS2A 228 V S C S N N Q G V228 V N C N D D Q G VS1 Mutant sequenceOriginal sequenceMutant Fig. 2. Mutagenized sites on the surface of TG2 (upper part) and detailed location of S4 and S5 Ca 2+ -binding sites in relation to GTP binding (bottom part). On the Ca backbone of TG2, the N-terminal domain is blue, the core domain is red, the first b-barrel is cyan, and the second b-barrel is green. The red spheres show the transglutaminase active site amino acids and the purple ones indi- cate the bound GTP. The yellow balls and sticks indicate Lys173 and Phe174, and the gray spheres show the proposed location of bound calcium ions. Ca 2+ -binding sites of TG2 R. Kira ´ ly et al. 7086 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS potential for Ca 2+ complexation was decreased (e.g. Asn to Ser). According to previous results [22], this type of amino acid replacement does not alter the expression and stability of mutant proteins. For normalization of protein expression and purity, the binding of a monoclonal antibody to each mutant was examined by ELISA, as antibodies are more sensi- tive to conformational changes [23]. Although most mutants showed similar antibody binding, the S2B mutant had lower affinity, because this mutation is at the recognition site of the antibody used in the experi- ments (Fig. S2). Study of CD spectra of mutants The native states of the purified proteins were tested by CD spectroscopy (Fig. 3). The CD spectra of the mutants did not show substantial deviations from that of the WT, which suggested that their secondary struc- tures were not altered significantly by the mutations. The CD deconvolution, performed with the analysis programs continll, cdsstr, and selcon3 [24,25], showed that unordered and turn structural segments contributed about 50% to the secondary structure, and that their values were very similar for all of the studied structures (data not shown). The WT and the S2A mutant had almost identical CD curves and thus very close secondary segment contributions as well. How- ever, some minor changes could be observed with the other mutant proteins: the S4 and S5 mutants had nearly identical CD curves, but they differed from the WT in their larger helix and smaller strand contribu- tions, resulting in somewhat larger negative ellipticities in the range 200–240 nm. In contrast, the S1, S2B, S3A and S3B mutants had smaller helix and larger strand contributions than the WT, resulting in smaller elliptici- ties in the range 200–240 nm. Subtle differences could be observed among the members of this group as well; the S3A and S3B mutants had near-identical CDs and hence secondary structures, whereas the S1 and S2B mutants were slightly different from them, with slightly larger helix and smaller strand segment contributions. Ca 2+ binding of mutant TG2 proteins To compare the Ca 2+ binding of wild-type and mutant enzymes in equilibrium dialysis, a free Ca 2+ concentra- tion of 1.7 mm was used. In case of the WT, the expo- nential part of the binding curve reaches the maximum at this concentration. If the mutants showed lower Ca 2+ binding than the WT, we would see larger changes in the exponential part of the binding curve than in other parts of the curve. All mutant proteins bound less Ca 2+ than the WT at 1.7 mm free Ca 2+ concentration (Fig. 4A), and the mean values were significantly different (P < 0.0001), as calculated using ANOVA. The experimental Ca 2+ - binding values confirmed that each of the five mutage- nized sites contributes to Ca 2+ binding of TG2. It was also observed that disruption of one site by mutation leads to weaker ⁄ loss of binding to other sites, and this suggests cooperative Ca 2+ -binding properties. For instance, in the S1 mutant, the number of bound Ca 2+ dropped from six to two. Using ICP-OES, we tested whether TG2 mutated at the site homologous to the high-affinity Ca 2+ -binding site of TG3 (S1) still binds Ca 2+ after purification. The result clearly showed that the S1 mutant cannot bind Ca 2+ after dialysis with EDTA (< 0.03 mol Ca 2+ per mol TG2), whereas the WT binds 0.5 mol Ca 2+ per mol TG2 under the same conditions. This result means that TG2 also has a Ca 2+ -binding site with high affinity and that this is S1. The ITC measurements of the S1 mutant show a stoichiometry of 1.7, which is the same value as obtained by equilib- rium dialysis (Figs 1B and 4A). Ca 2+ -dependent transglutaminase activity of mutant TG2 proteins As the transglutaminase activity of TG2 is Ca 2+ - dependent, decreased activity of the mutants could be expected (Fig. 4B). In accordance with this, the trans- glutaminase activity of each mutant decreased to vari- ous extents, and the S3, S4 and S5 mutants lost their activity completely in the microtiter plate as well as in the filter paper assay (Fig. 4C). Interestingly, in the case of the S2A mutant and, mainly, the S2B mutant, a substantial difference was observed between the results of the two methods, which differ in the use of amine substrate and the availability of Gln substrate 190 200 210 220 230 240 250 –10 –5 0 5 10 15 [θ] x 10 –3 deg cm 2 decimol –1 Wavelength (nm) S5 S4 W S2A S1, S2B S3A,S3B Fig. 3. CD spectra of recombinant TG2 proteins. R. Kira ´ ly et al. Ca 2+ -binding sites of TG2 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS 7087 in solution versus bound to the surface. At higher Ca 2+ concentrations, there was no significant increase in the activities, which means that increased Ca 2+ con- centrations cannot compensate for the loss of specific Ca 2+ -binding side chains. Transglutaminase activity is inhibited by GTP and GDP. As a proof of GTP sensitivity, we wished to see a decrease in this remaining, lower transglutaminase activity of mutant enzymes as compared with the WT. To obtain a sufficient starting transglutaminase activ- ity, a relatively high Ca 2+ concentration (5 mm) was used (Fig. 4D): under these conditions, the presence of 100 lm GTP decreased the transglutaminase activity of both the WT and the S1, S2A and S2B mutants by  40%. These results suggested that GTP can effec- tively bind to the mutants. GTPase activity of mutant TG2 proteins Ca 2+ binding also influences both GTP binding and the GTPase activity of TG2 [5]. Initially, we per- formed photoaffinity GTP-labeling experiments. As shown by autoradiography in Fig. 5A, the S2B, S3A and S3B mutants had similar GTP incorporation to the WT, whereas the S1 and S2A mutants had lower GTP incorporation than the WT after UV light exposure. The GTPase activity of these mutants correlated well with photoaffinity GTP labeling. As expected, we also could see a slight decrease in GTPase activity at increasing Ca 2+ concentrations for most of the mutants, similarly to what was seen with the WT (Fig. 5B). A P = 0.0023 P = 0.0016 P < 0.0001 P < 0.0001 P < 0.0001 P < 0.0001 P < 0.0001 P values 3.0 ± 1.4 3.2 ± 1.1 2.5 ± 0.5 2.3 ± 0.9 1.6 ± 0.3 4.1 ± 0.7 1.7 ± 0.4 5.6 ± 0.7 Bound Ca 2+ at 1.67 mM [Ca 2+ ] free mol/mol S5S4S3BS3AS2BS2AS1WT Mutants 0 20 40 60 80 100 120 WT S1 S2A S2B S3A S3B S4 S5 Relative activity (%) Filter paper method Microtiter plate method C D 0 1 2 3 4 5 6 7 8 9 0 10203040 [Ca 2+ ] (mM) Specific activity (DAbs·min·mg –1 ) WT S1 S2A S2B S3A S3B S4 S5 B 0 20 40 60 80 100 120 WT S1 S2A S2B Relative activity (%) GTP 0 µM/Ca 2+ 5 mM GTP 100 µM/Ca 2+ 5 mM P = 0.0135 P = 0.0140 P = 0.0005 P < 0.0001 Fig. 4. Ca 2+ -binding and Ca 2+ -dependent transglutaminase activities of wild-type and mutant TG2s. (A) Ca 2+ binding by wild-type and mutant TG2s at 1.67 m M free Ca 2+ concentration as measured by equilibrium dialysis. Data are presented as means from four separate experiments performed in duplicate. The mean values are significantly different (P < 0.0001), as calculated using ANOVA. Unpaired t-tests were per- formed to compare the Ca 2+ binding of the mutants with Ca 2+ binding of the wild-type enzyme, and these show that each difference is highly significant. (B) Ca 2+ -dependent transglutaminase activity of recombinant TG2s as determined by using a microtiter plate method. Varia- tion between experimental values was less than 10%. (C) Transglutaminase activity is shown as a percentage of the activity of wild-type TG2. One hundred per cent specific activity of wild-type TG2 was 8.4 DA 405 (minÆmg) )1 protein in the case of the microtiter plate method, and 77.4 pmol putrescine (minÆmg) )1 protein in the case of the filter paper method in the presence of 5 mM Ca 2+ . (D) Inhibition of residual transglutaminase activity of recombinant TG2s by GTP, using the microtiter plate method. Activity is shown as a percentage of the activity of wild-type TG2; the Ca 2+ concentration was 5 mM. Ca 2+ -binding sites of TG2 R. Kira ´ ly et al. 7088 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS Interestingly, the S4 and S5 mutants showed 1.5-fold to two-fold increased specific GTPase activity despite an apparent lack of stable UV-induced GTP incorpora- tion as determined by photoaffinity labeling, and their GTPase activity was not inhibited by increasing Ca 2+ concentrations. Neither longer UV irradiation nor a higher amount of the protein led to photolabeling of these proteins. To confirm increased GTPase activity of the S4 and S5 mutants, we also expressed them in gluta- thione-S-transferase (GST)-fused forms with high purity. The transglutaminase activities of the GST–S4 and GST–S5 mutants were measured using the two methods described above, and both failed to detect any crosslinking activity (data not shown). We found slightly increased GTPase activity in the case of the GST–S4 mutant (130.7% ± 3.4% as compared with GST–WT as 100%), and greatly increased GTPase activity in the case of the GST–S5 mutant (353 ± 38% as compared with GST–WT). Nonspecific GTP degra- dation was excluded by the use of appropriate controls. Antigenicity of mutant TG2 forms TG2 is recognized as an autoantigen in GSE. The epitopes are conformational [26], and the presence of Ca 2+ can increase the binding of celiac autoantibodies to TG2 [27,28], although there are some contradictory results [29,30]. In an attempt to resolve this discrep- ancy, our Ca 2+ -binding site mutants were tested in ELISA with a large panel (n = 62) of serum samples obtained from GSE patients (age 1.1–69 years, mean 10.4) prior to treatment (Fig. 6; for statistical analysis, see Table S2). The S4, S5, S3B and S3A mutants showed decreased affinity for celiac autoantibodies (P < 0.001), and the S4 mutant showed the lowest binding (11.5 ± 8.2% as compared with the WT as 100%). The binding of celiac autoantibodies to TG2 is influ- enced by the presence or absence of Ca 2+ in the case of guinea pig TG2 [27]. Therefore, we also examined the effect of Ca 2+ and GDP on the binding of celiac autoantibodies to mutant TG2s. The presence of 2 mm EDTA, 20 lm GDP or 5 mm Ca 2+ failed to alter the antigenicity of the enzymes (Fig. 6, insert). It has been recently reported [31] that mutation of the transgluta- minase catalytic triad of the active site decreased the binding of celiac autoantibodies to the enzyme. In our experiments, celiac autoantibodies could bind to the C277S mutant and to the WT with similar affinity (data not shown). Discussion In this article, we describe the Ca 2+ -binding properties of TG2 and its mutants and the role of five Ca 2+ - binding sites in the regulation of transglutaminase and GTPase activities, as well as the binding of celiac 0 20 40 60 80 100 120 140 160 180 A B WT S1 S2A S2B S3A S3B S4 S5 Relative activity (%) Without Ca 2+ With 3.5 mM Ca 2+ With 7 mM Ca 2+ TG2 TG2 WT S1 S2A S2B S3A S3B S4 S5 Fig. 5. Photoaffinity GTP labeling and GTPase activity of recombi- nant TG2s. (A) Photoaffinity labeling of TG2 proteins; 2.2 lg protein per lane (upper panel). Proteins were visualized with Coomassie BB staining (lower panel). (B) GTPase activity and effect of Ca 2+ on the GTPase activity of recombinant TG2s. GTPase activity was calculated as percentage of activity of the WT [99.1 ± 7.4 pmol GTP (minÆmg protein) )1 ] in the absence of Ca 2+ . Data are presented as means with ± standard deviations from three separate experiments performed in triplicate. S1 S2A S2B S3A S3B S4 S5 WT 0 25 50 75 100 125 150 175 Relative binding (%) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 WT S4 Relative binding (%) +Ca 2+ +EDTA +GDP Fig. 6. Binding of IgA class celiac antibodies to wild-type and mutant TG2s. Binding to wild-type TG2 is 100%, serum dilution is 1 : 200, and n = 62 from biopsy-proven untreated celiac disease patients. The mean values are significantly different (P < 0.0001), as calculated using ANOVA. Effects of 2 m M EDTA or 20 lM GDP are compared with those of 5 m M Ca 2+ on antibody binding (insert). Results were similar for IgG class antibodies (data not shown). R. Kira ´ ly et al. Ca 2+ -binding sites of TG2 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS 7089 autoantibodies to the enzyme. We found five nonca- nonical Ca 2+ -binding sites of TG2, and determined that one of these, S1, has a tightly bound Ca 2+ . There are canonical and noncanonical protein structures that bind Ca 2+ , and common to all of them is a high nega- tive surface potential, derived mainly from Asp or Glu residues. Ca 2+ , as a ‘hard’ metal ion, can coordinate six or seven ligands with negative character or charge in a pentagonal bipyramidal arrangement. There are some well-known canonical Ca 2+ -binding domain structures: EF-hand domain, C-type lectin-like domain, Ca 2+ -dependent phosphatidylserine-binding domains (C2, annexin and Gla domains), and EGF-like domain, which have been characterized in detail by X-ray diffraction and NMR spectroscopy. These known Ca 2+ -binding domains, which are present in a large number of Ca 2+ -binding proteins, do not share significant similarities with the Ca 2+ -binding motifs of the transglutaminase family. Interestingly, TG2 also has a GTP-binding site and can hydrolyze GTP, but does not have a typical GTP-binding site [13]. Members of the transglutaminase family have some highly conserved negatively charged amino acids with high surface potential. Ikura et al. [22] mutated two highly conserved anionic sites of the guinea pig TG2 that were earlier proposed, on the basis of sequence comparison, as putative Ca 2+ -binding sites [21]; how- ever, their data showed that these sites are not essen- tial for or directly involved in Ca 2+ binding. The Ca 2+ -binding sites of human TG2 studied here partially overlap with these negatively charged surface patches in the guinea pig enzyme sequence. Every mutated site investigated by us is located on a loop or border of a loop, which could allow the appropriate coordination of Ca 2+ and, in addition, may induce a change in the structure of the protein. Without a Ca 2+ -bound X-ray structure, however, it is not possi- ble to establish the exact participation of the different side chains in Ca 2+ binding and selected functions. For TG2, the most important regulatory function of bound Ca 2+ is the initiation of transglutaminase activity. The tightly bound Ca 2+ at S1 is not enough for transglutaminase activity in the case of TG3; addi- tional Ca 2+ binding to S3 is needed to open the active site and to form a substrate channel [16]. According to our data, the measurable transglutamin- ase activity of the S1 mutant suggests that, although Ca 2+ binding to this site is important for this activity, binding of Ca 2+ to other sites also contributes to the effective induction of an active transglutaminase con- formation. Binding of Ca 2+ to S2 plays only a minor role in the formation of the active state of TG2, because mutation of S2 resulted in the highest resid- ual transglutaminase activity. The loss of S3 Ca 2+ binding leads to an enzyme without transglutaminase activity, suggesting that the binding of Ca 2+ to S3 in TG2 plays a significant role in the induction of this activity, similarly to the case of TG3. It is very likely that Glu329 (replaced in the S3B mutant) plays a crucial role in Ca 2+ coordination and regulation of transglutaminase activity. It is interesting to note that, by activation of TG2 [14], S3 undergoes significant dislocation, just like the GTP-binding site, which is also composed of two or three loops. Datta et al. [32] studied, without determining actual Ca 2+ binding, how three Ca 2+ -binding site mutants of TG2 influence cell survival; these sites correspond to our S1, S2, and S3A. They changed two amino acids to Ala at targeted sites, and this resulted in decreased transglutaminase activity; similarly to our results, there was no change in GTPase activity and GTP binding, except for the N229A ⁄ D233A mutant (labeled S1 by us). Each Ca 2+ -binding site is in the core domain of TG2, and they could influence each other, leading to an energetically favorable arrangement of the enzyme structure (Fig. 2). Our finding that mutation of one site leads to the loss of more than one bound Ca 2+ supports an assumption of positive cooperativity [33] among the Ca 2+ -binding sites of TG2. Ahvazi et al. [16] also found indications that S2 and S3 may coop- erate in TG3. S4 and S5 may have similar roles in the process of fine tuning cooperativity, as mutation of these sites also leads to the loss of Ca 2+ -inducible transglutaminase activity. Our data suggest that the cooperativity may be strong between S3, S4, and S5, because loss of any of these results in binding of about three Ca 2+ and total inactivation, which means that Ca 2+ binding at these sites is needed for the active conformation of transglutaminase activity. This also raises the possibility that a sequential mechanism of site occupancy may operate in Ca 2+ binding of TG2. How can weak Ca 2+ -binding sites play such an important role in determining transglutaminase activity when various biophysical measurements did not show significant changes of TG2 after Ca 2+ binding [34]? In the case of the canonical C2 domain, it is known that a third Ca 2+ binds with lower affinity to the domain, but in the presence of an interaction partner – phos- pholipid in the case of C2, but it could be any appro- priate substrate in the case of TG2 – the affinity for Ca 2+ is higher, owing to completed coordination spheres [35] of Ca 2+ . Further study is required to clar- ify whether substrates, other interacting partners or lipid molecules can regulate Ca 2+ affinity of TG2. Ca 2+ -binding sites of TG2 R. Kira ´ ly et al. 7090 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS Interestingly, the S1–S3 mutants and the WT showed Ca 2+ -sensitive GTPase activity, but this was not observed in the case of the S4 and S5 mutants. In the TG3 structure, Asp324, which coordinates the S3 analog Ca 2+ -binding site directly and is located on a loop forming a part of the S3B analog site, is responsi- ble for a switch between GTP and Ca 2+ binding by opening a channel for the acyl Gln donor substrate [16,17]. As Ca 2+ binding can decrease GTP binding and GTPase activity in the case of TG2, too, S4 and S5 could be responsible for regulatinon of GTPase activity and the proper regulation of the distinct trans- glutaminase and GTPase activities. When these two sites could not bind Ca 2+ , the GTPase activity of TG2 was not inhibited by increasing the Ca 2+ concentra- tion. Moreover, these two mutations resulted in increased basal GTPase activity and altered GTP ⁄ GDP binding. The two mutated sites are steri- cally close to the hydrophobic pocket for GTP ⁄ GDP binding, which is formed by the side chains from Phe174, Val479, Met483, Leu582, and Tyr583 [13]. They may conformationally influence GTP binding and the GTPase activity of TG2 by changing the posi- tion of Phe174, the docking amino acid, and of Lys173, which is the nucleophilic attacking group in GTP hydrolysis (Fig. 2, lower panel). The mutations can result in a conformational state that speeds up GDP ⁄ GTP exchange via decreasing the docking time of GTP and facilitating the release of GDP, which ulti- mately results in higher GTPase activity and lower GTP binding. Such a change could also be responsible for the lack of GTP incorporation signal in our phot- olabeling experiments. However, it cannot be excluded that this happened because the altered surface did not support the UV light-induced artificial trapping of GDP after hydrolysis via its guanosine group. A simi- lar finding was described when the core domain of TG2 was expressed alone and tested for GTP bind- ing ⁄ hydrolysis with the same methods [36]. Interest- ingly, two shorter alternatively spliced forms of TG2 that have lower GTP-binding affinity also have higher GTPase activity [36a]. Mutagenesis of some of the Ca 2+ -binding sites leads to decreased binding of celiac autoantibodies against TG2 to the enzyme. GSE is a chronic disorder of the small intestine in genetically susceptible individuals. Wheat gliadin and related prolamins in other cereals can trigger an autoimmune reaction to TG2 [37,38], and the resulting autoantibodies might play a role in the pathogenesis of GSE by modifying the enzyme’s activities or other functions [39]. Previous results sug- gested that binding of Ca 2+ to TG2 is needed to pro- mote the binding of celiac antibodies to the enzyme [27,28]. S1, S2 and S3A do not have a role in antibody binding, because the S1, S2 and S3A mutants were recognized equally as well as the WT. Also, the S3B and S5 mutants retained considerable antigenicity towards patient serum samples, making a direct role in antibody binding improbable for the majority of patients. In contrast, binding of celiac serum samples to TG2 was greatly affected by changing S4, suggest- ing that it may be needed to form a main celiac epitope. Further clarification of potential anchor points in this region may help us to understand the role of antibodies in the pathogenesis of GSE and in designing new therapy for it. Members of the mammalian transglutaminase family have evolved through duplication of a single gene and subsequent redistribution to distinct chromosomes [40]. On the basis of the available and presented data, a description of the subsequent evolution of the Ca 2+ - binding sites of the human enzymes can be attempted. Sequence comparison (Fig. 7) clearly shows that S2 is conserved in each transglutaminase, and this by itself can determine the Ca 2+ dependency of transglutamin- ase activity, as FXIIIa has only the S2-equivalent site. Similarly, prostate transglutaminase (TG4) seems to have only this site. It is likely that these two secreted enzymes are sufficiently activated by Ca 2+ through this site in the extracellular space, where the Ca 2+ concentration is high. Transglutaminase 1 works in the terminally differentiating keratinocytes, where Ca 2+ concentration rises; sequence data show that, in addi- tion to S2, it may have S1 as well. It seems that intra- cellular transglutaminases need more sophisticated Ca 2+ regulation. We propose that, for intracellular transglutaminase activation, S1, which binds Ca 2+ tightly, is essential, as all intracellular forms have potential S1s. Actually, sequence comparison suggests that even the red sea bream and invertebrate Drosoph- ila transglutaminases have S1 and S2. Sequence com- parisons also explain why FXIIIa does not have S1: FXIIIa has a positively charged amino acid (Lys) in this region. A similar sequence difference may preclude Ca 2+ binding at S1 of TG4. There are some amino acids with apolar or positive side chains in S3, S4 and S5 of FXIIIa, transglutaminase 1 and TG4, and S3 and S5 in TG3, suggesting that they do not bind Ca 2+ there. S3 is needed to open the substrate channel in intracellular transglutaminases. Transglutaminase 5 and transglutaminase 7 probably lost this site; these two enzymes are located on a different arm of the phy- logenetic tree of transglutaminases than TG2 or TG3 and transglutaminase 6 [40], and may use another site for this purpose. Transglutaminase 5, transglutaminase 6 and transglutaminase 7 also have S4 and S5, and R. Kira ´ ly et al. Ca 2+ -binding sites of TG2 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS 7091 therefore may have similar Ca 2+ regulatory mecha- nisms as TG2. This perhaps explains how these trans- glutaminases may compensate for the loss of TG2 in knockout mice [41]. Experimental procedures Materials All materials were purchased from Sigma (St Louis, MO, USA) unless otherwise indicated. Transglutaminase enzyme preparations Wild-type recombinant human TG2s were expressed in N-terminally (His)6-tagged [42] and GST-fused forms [19]. The fusion tags, which were not found to alter the enzymo- logical properties of TG2 [36], were left on the protein. Site-directed mutants were constructed using the Quik- Change Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). Mutant constructs were checked by restriction analysis and DNA sequencing (ABI PRISM, Applied Biosystems, Foster City, CA, USA). Rosetta 2 (Novagen, Darmstadt, Germany) strains were transformed with wild-type or mutant TG2 containing pET-30 Ek ⁄ LIC– TG2 vectors. The (His)6-tagged proteins were expressed in a similar way to that described previously [42], using Pro- Bond Ni 2+ –nitrilotriacetic acid resin (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. The protein concentration was determined using the Brad- ford method (Bio-Rad, Mu ¨ nchen, Germany). The purity and self-crosslinking activity of proteins were checked by Coomassie BB staining of SDS ⁄ polyacrylamide gels and by western blotting (Fig. S1). Equilibrium dialysis Ca 2+ binding to TG2 was measured by equilibrium dialy- sis, with modification of a published procedure [18]. For every Ca 2+ -binding experiment, only EDTA-rinsed plastic- ware and high-purity water (Millipore, Billerica, MA, USA) were used, to prevent Ca 2+ contamination. Recom- binant TG2 ( 1.7 mgÆmL )1 ) was dialyzed for 48 h at 4 °C in a 96-well equilibrium dialyzer plate (molecular mass cutoff 10 kDa; Harvard Bioscience, Holliston, MA, USA) against 150 lL of dialysis buffer (50 mm Tris ⁄ HCl, 5 mm mercaptoethanol, pH 7.5) supplemented with 8.3 lCi of 45 CaCl 2 per mL (PerkinElmer, Boston, MA, USA) and containing different concentrations of cold CaCl 2 . After the equilibration, the radioactivity was measured by liquid scintillation counting using Tritosol [43]. The results were normalized for the protein content of the sample deter- mined by Bradford reagent and protein purity, which was measured with alpha imager software ( 90%). The free Ca 2+ concentration was calculated by maxchelator and Fabiato and Fabiato’s computer program [44,45]. The Ca 2+ -binding curves and binding parameters were fitted and calculated using graphpad prism software (GraphPad Software, Inc., La Jolla, CA, USA). To reduce the possible incidental errors of the radioactive method, we used Ca 2+ solutions of high specific activity: this was 10 000– 23 000 c.p.m. per nmol Ca 2+ in the case of measurements at 1.67 mm Ca 2+ , and  3.2 000 000 c.p.m. per nmol Ca 2+ at smaller Ca 2+ concentrations, to increase the accu- racy of measurement. The standard errors of parallel coun- ter measurements and protein determinations were lower than 6%, mostly less than 3%, in every case. The differ- ence in radioactivity between the two chambers was always highly significant (P < 0.01). Fig. 7. Multiple sequence alignment of Ca 2+ -binding sites of transglutaminases. Sequence alignments of Ca 2+ -binding sites of TG2 com- pared with the other members of transglutaminase family using CLUSTALW. The bold characters mark the proven Ca 2+ -binding sites, and the underlined characters indicate the amino acids that coordinate Ca 2+ in known crystal structures. Characters in bold italics indicate potential Ca 2+ -binding sites as compared with those in TG2. The amino acids in the frames may preclude Ca 2+ binding as compared with the homo- logous sites in other members of the transglutaminase family in which Ca 2+ -binding sites have been verified. Invertebrate transglutaminase used in the alignments: Q9VLU2_DROME is the A isoform of Drosophila melanogaster transglutaminase. TGM2_PAGMA is the red sea bream (Pagrus major) TG2. ‘x’ indicates the amino acids that are conserved in the enzyme family. Ca 2+ -binding sites of TG2 R. Kira ´ ly et al. 7092 FEBS Journal 276 (2009) 7083–7096 ª 2009 The Authors Journal compilation ª 2009 FEBS [...]... disease recognize distinct functional FEBS Journal 27 6 (20 09) 7083–7096 ª 20 09 The Authors Journal compilation ª 20 09 FEBS Ca2+-binding sites of TG2 ´ly R Kira et al 27 28 29 30 31 32 33 34 35 36 36a 37 38 domains of the autoantigen tissue transglutaminase Clin Exp Immunol 125 , 21 6 22 1 Roth EB, Sjoberg K & Stenberg P (20 03) Biochemical and immuno-pathological aspects of tissue transglutaminase in coeliac... modulation by ligands of protein turn-over in vivo Amino Acids 33, 415– 421 13 Liu S, Cerione RA & Clardy J (20 02) Structural basis for the guanine nucleotide-binding activity of tissue 7094 14 16 17 18 19 20 21 22 23 24 25 26 transglutaminase and its regulation of transamidation activity Proc Natl Acad Sci USA 99, 27 43 27 47 Pinkas DM, Strop P, Brunger AT & Khosla C (20 07) Transglutaminase 2 undergoes... properties of the Ca2+-binding helix–loop–helix EF-hand motifs Biochem J 405, 199– 22 1 ` Di Venere A, Rossi A, De Matteis F, Rosato N, Agro AF & Mei G (20 00) Opposite effects of Ca (2+ ) and GTP binding on tissue transglutaminase tertiary structure J Biol Chem 27 5, 3915–3 921 Rizo J & Sudhof TC (1998) C2-domains, structure and ¨ function of a universal Ca2+-binding domain J Biol Chem 27 3, 15879–158 82 Iismaa... Biosciences, University of Newcastle upon Tyne, UK This work was supported by the following grants: Hungarian Scientific Research Funds (OTKA NI 67877, K 61868) and EU grants MRTN-CT -20 06-0360 32, MRTN-CT -20 06´ 035 624 , and LSHB-CT -20 07-037730 R Kiraly was the FEBS Journal 27 6 (20 09) 7083–7096 ª 20 09 The Authors Journal compilation ª 20 09 FEBS 7093 Ca2+-binding sites of TG2 ´ly R Kira et al recipient of a travel... Experimental procedures Fig S1 Self-crosslinking of TG2 variants Fig S2 Purity of and monoclonal antibody binding to recombinant wild-type and mutant TG2s Table S1 Primers used for mutagenesis FEBS Journal 27 6 (20 09) 7083–7096 ª 20 09 The Authors Journal compilation ª 20 09 FEBS 7095 Ca2+-binding sites of TG2 ´ly R Kira et al Table S2 Statistical analysis of celiac antibody binding in Fig 6 This supplementary... enzyme J Biol Chem 27 8, 23 834 23 841 Ahvazi B, Kim HC, Kee SH, Nemes Z & Steinert PM (20 02) Three-dimensional structure of the human transglutaminase 3 enzyme: binding of calcium ions changes structure for activation EMBO J 21 , 20 55 20 67 Bergamini CM (1988) GTP modulates calcium binding and cation-induced conformational changes in erythrocyte transglutaminase FEBS Lett 23 9, 25 5 25 8 ´ Ambrus A, Banyai... Farrace MG, Garofano E, Piredda L, Fimia GM, Malorni W & Piacentini M (20 04) Tissue transglutaminase is a multifunctional BH3-only protein J Biol Chem 27 9, 54783–547 92 11 Zemskov EA, Janiak A, Hang J, Waghray A & Belkin AM (20 06) The role of tissue transglutaminase in cell–matrix interactions Front Biosci 11, 1057–1076 12 Bergamini CM (20 07) Effects of ligands on the stability of tissue transglutaminase: ... experiments were performed at 25 °C; a 40 lm wild-type TG2 sample was placed into the sample chamber, and then, from a 2 mm CaCl2 solution (in ITC buffer), the following volumes were injected into it: 5 · 2 lL, 6 · 5 lL, 5 · 15 lL, 7 · 20 lL, and 2 · 25 lL In the case of S1 mutants, a 23 lm TG2 sample was used, and from a 0.6 mm CaCl2 solution, 4 · 2 lL, 5 · 4 lL, 6 · 8 lL, 5 · 20 lL and 5 · 30 lL volumes... dependence on reaction environment and enzyme fitness J Autoimmun 26 , 27 8 28 7 Grenard P, Bates MK & Aeschlimann D (20 01) Evolution of transglutaminase genes: identification of a transglutaminase gene cluster on human chromosome 15q15 Structure of the gene encoding transglutaminase X and a novel gene family member, transglutaminase Z J Biol Chem 27 6, 33066–33078 Szondy Z, Sarang Z, Molnar P, Nemeth T, Piacentini... grant from the European Science Foundation’s Transglutaminases Program and Erasmus ´ Program and the Deak Ferenc Fellowship of the Hungarian Ministry of Education and Culture 15 References 1 Fesus L & Piacentini M (20 02) Transglutaminase 2: an enigmatic enzyme with diverse functions Trends Biochem Sci 27 , 534–539 2 Griffin M, Casadio R & Bergamini CM (20 02) Transglutaminases: nature’s biological glues . Functional significance of five noncanonical Ca 2+ -binding sites of human transglutaminase 2 characterized by site-directed mutagenesis Ro ´ bert. alignment of Ca 2+ -binding sites of transglutaminases. Sequence alignments of Ca 2+ -binding sites of TG2 com- pared with the other members of transglutaminase