Hydrolysis of gamma:epsilon isopeptides by cytosolic transglutaminases and by coagulation factor XIIIa.

Nepsilon-(gamma-glutamyl)lysine cross-links, connecting various peptide chain segments, are frequently the major products in transglutaminase-catalyzed reactions. We have now investigated the effectiveness of these enzymes for hydrolyzing the gamma:epsilon linkage. Branched compounds were synthesized, in which the backbone on the gamma-side of the cross-bridge was labeled with a fluorophor (5-(dimethylamino)-1-naphthalenesulfonyl or 2-aminobenzoyl) attached through an epsilon-aminocaproyl linker in the N-terminal position, and the other branch of the bridge was constructed with Lys methylamide or diaminopentane blocked by 2,4-dinitrophenyl at the Nalpha position. Hydrolysis of the cross-link could be followed in these internally quenched substrates by an increase in fluorescence. In addition to the thrombin and Ca2+-activated human coagulation Factor XIIIa, cytosolic transglutaminases from human red cells and from guinea pig liver were tested. All three enzymes were found to display good isopeptidase activities, with Km values of 10(-4) to 10(-5) M. Inhibitors of transamidation were effective in blocking the hydrolysis by the enzymes, indicating that expression of isopeptidase activity did not require unusual protein conformations. We suggest that transglutaminases may play a dynamic role in biology not only by promoting the formation but also the breaking of Nepsilon-(gamma-glutamyl)lysine isopeptides.

Apart from obvious differences in substrate specificities for the ␣-carbonyl groups of endo-Lys and Arg residues by papain (EC 3.4.22.2) and for the ␥-carbonyl groups of certain endo-Gln residues by transglutaminases (EC 2.3.2.13), considerable kinetic and mechanistic similarities exist between these two families of enzymes. Both operate by acylation-deacylation pathways, with a Cys thiol in the catalytic center assisted by a His residue (1)(2)(3)(4)(5)(6). However, because of the exceptional specificities of transglutaminases for amines mimicking the ⑀-amino groups of Lys side chains in proteins (7)(8)(9), this group of enzymes shows a unique ability for generating protein-to-protein N ⑀ -(␥glutamyl)lysine cross-links, a post-translational reaction of major biological significance. Transglutaminases are known to participate in various clotting phenomena (7, 10 -16), in the assembly of extracellular matrices (17) and of intracellular polymeric structures in cells under Ca 2ϩ stress (18 -22), and in apoptosis (23).
While a great deal of attention has been paid to the amine transferase activities of transglutaminases (3,4,24), i.e. to the production of N ⑀ -(␥-glutamyl)lysine bridges and the incorporation of small molecular weight amines into proteins, the isopeptide breaking potential of the enzymes has not yet been explored. Since lack of availability of appropriate substrates may have been a main reason, we embarked on synthesizing ␥-branched peptides with built-in features, which would facilitate the application of fluorescence methodologies for kinetic studies. Two cytosolic transglutaminases of different properties, isolated from human red blood cells (HTg) 1 and from guinea pig liver (GTg) respectively, and a recombinant form of the human coagulation Factor XIII a (rA 2 *) were employed as enzymes.

Organic Synthesis
Reagents, solvents, and blocked amino acid derivatives were purchased from Aldrich, Sigma, and Bachem Bioscience Inc. TLC (thinlayer chromatography) was performed on Whatman K 6 F-Silica gel glass plates (0.25 mm) using the following solvent systems (v/v): (A) chloroform/methanol/glacial acetic acid (10:3:1); (B) chloroform/methanol/2propanol (10:4:4); (C) n-butanol/glacial acetic acid/water (15:6:5); (D) 1-propanol/water (7:3); (E) propanol/water/concentrated ammonium hydroxide/ethanol (7:4:2:3). Plates were viewed under UV light (at 254 nm and 366 nm for detection of UV absorbing and fluorescent moieties, respectively) or were developed by ninhydrin (0.25% in 1-butanol for N-deblocked peptides) or by hypochlorite (10%) followed by starch/KI spray for N-blocked peptides (26). Melting points were determined with a Bü chi apparatus and are uncorrected. After acid hydrolysis, amino acid analyses were kindly carried out by Dr. Thomas J. Lukas of the Department of Molecular Pharmacology and Biochemistry, Northwestern University Medical School, Chicago, IL. Elemental analyses were performed by G. D. Searle Laboratories, Skokie, IL. 1 The abbreviations used are: HTg, human red blood cell transglutaminase; GTg, guinea pig liver transglutaminase; Boc, tert-butyloxycarbonyl; Z, benzyloxycarbonyl; Dns or dansyl, 5-(dimethylamino)-1naphthalenesulfonyl; Cad, cadaverine (or 1,5-diaminopentane); Dns-Cad or Cad-Dns, dansylcadaverine (or N- (5- (obtained by acidolytic deblocking of the Boc-peptide benzyl ester described below) in dry DMF, which was pre-neutralized with N-methylmorpholine. The reaction mixture was stirred at 0°C for 1 h and then at room temperature for 18 -36 h. The mixture was evaporated under reduced pressure to remove DMF, and the residue was stirred with 3% sodium bicarbonate for 15 min. The precipitated product was filtered off, or separated by centrifugation, washed with 5% NaHCO 3 , water, cold 0.5 N HCl, water and dried in vacuo in a P 2 O 5 desiccator for ϳ18 h. When necessary, the product was reprecipitated from DMF-water. Alternatively, the product was extracted with ethyl acetate, and the organic extract was washed as above, dried over anhydrous Na 2 SO 4 , and the solvent evaporated to give the product.

Deblocking of the Boc Group
To 1.0 mmol of the Boc-peptide benzyl ester was added 2 ml of 50% trifluoroacetic acid in anhydrous dichloromethane. The solution was allowed to stand at room temperature for 1 h, and excess trifluoroacetic acid was removed by adding fresh dichloromethane to the mixture followed by evaporation under reduced pressure, then by addition of anhydrous ether to the residue. The precipitated trifluoroacetic acid salt of the peptide benzyl ester was filtered off, washed with anhydrous ether, and dried under vacuum in a desiccator for 2 h before proceeding with the coupling reaction.

Deblocking of the Benzyl Ester Group
The blocked peptide benzyl ester was dissolved in a mixture of DMF/ethanol/water in the approximate volume ratio of 10:20:2 at 50°C and hydrogenated in presence of 10% Pd/C catalyst for 3.5 h at 50°C with stirring. After cooling to room temperature, water and 1 N NH 4 OH to pH 9 were added and the mixture filtered to remove the catalyst through a Celite filter pad. The filtrate was evaporated under reduced pressure, the residue was taken up in a minimum volume of absolute ethanol, and about 10 volumes of anhydrous ether added and cooled to 0°C. The precipitated product was filtered off, washed with anhydrous ether, and dried in vacuo.

Compounds Used as First Substrates
Dns-Eaca-Gln-Gln-Ile-Val (I)-This was synthesized as described in Ref. 27.
Dns-Eaca-Glu-[␥-(Cad-Dnp)]-Gln-Ile-Val (II)-This was synthesized by a multistep procedure as follows. First, Boc-Glu-␣-benzyl ester was coupled to mono-Z-cadaverine (24), essentially according to the general procedure above and the product isolated from an ethyl acetate extract of the reaction mixture to give an 80% yield of Boc-Glu . This was then reacted in 50% aqueous ethanol with a 40% molar excess of 2,4-dinitrofluorobenzene and NaHCO 3 . The reaction mixture was stirred at 0°C for 0.5 h and at room temperature for 14 h while protected from light. About 15 drops of concentrated Na 2 CO 3 solution were added to the mixture to raise the pH to 9 -9.5 (for decomposing excess 2,4-dinitrofluorobenzene), and ethanol was evaporated under reduced pressure. After adding 5 ml of water, the solution was centrifuged to remove a small amount of precipitate. The clear supernatant was separated, cooled (0°C), and acidified to pH 2.5 with cold 2 N HCl with stirring. The yellow precipitate was collected by centrifugation, washed with cold water (3 ϫ 3 ml), and finally washed with ether (5 ϫ 5 ml). The precipitate was triturated with ethanol-ether

Amine Compounds Used as Second Substrates for Studying the Enzyme-catalyzed Formation of Isopeptides
Dnp-Cadaverine (XI)-This was prepared according to the published procedure (24). Dbc-Cadaverine (XIV)-First, 4-(4-dimethylaminophenylazo)benzoic acid sodium salt was coupled to Boc-cadaverine hydrochloride (32) in presence of two equivalents of N-methylmorpholine by a slightly modified peptide coupling method. A 2-fold molar excess of 1-hydroxybenzotriazole was used in the coupling, and the reaction was carried out at room temperature for 24 h and the product isolated from an ethyl acetate extract of the reaction mixture to give dabcylcadaverine-Boc, a bright red solid, yield 71%, m.p. 159 -161°C, R F 0.97 (A), 0.94 (C), 0.87 (D), and 0.93 (E). 1 H NMR (Me 2 SO-d 6 ) was in agreement with the structure. This was then deblocked with 50% trifluoroacetic acid in anhydrous dichloromethane as described above to give a 77% yield of dabcylcadaverine trifluoroacetate as a red solid, m.p. 177-179°C, R

Enzymatic Studies
Guinea pig liver transglutaminase (GTg; Refs. 33 and 34) and human red blood cell transglutaminase (HTg; Refs. 35 and 36) were purified as described previously and stored at Ϫ80°C. Protein concentrations for GTg were measured by absorbance at 280 nm (⑀ 1 cm 1% ϭ 15.8). Protein concentrations for HTg were determined with the BCA protein assay (Pierce), utilizing bicinchoninic acid with bovine serum albumin for standard. Calculations were based on molecular weights of 76,600 for GTg (37) and 80,000 for HTg (35).
Transglutaminase activity was measured in a CytoFluor model 2300 (Millipore, Bedford, MA), upgraded to a model 2350, by monitoring the rate of increase in fluorescence during the transglutaminase-catalyzed incorporation of dansylcadaverine (38)

Monitoring Transglutaminase-catalyzed Isopeptide Formation or the Breaking of Isopeptide Bonds by Changes in Fluorescence
Conditions for individual experiments are specified in the figure legends. Fluorescence measurements in the CytoFluor instrument were carried out with an excitation filter ϭ 360 Ϯ 40 nm and an emission filter ϭ 590 Ϯ 35 nm at sensitivities of 7 or 8. Incubations were set up in a CytoPlate at 37°C in 150-l reaction mixtures, which, in addition to the specified substrates and inhibitors, comprised 50 mM buffer (either Tris-HCl, N-methylmorpholine-HCl or sodium acetate:acetic acid), 1 mM DTT, 1 mM CaCl 2 , and one of the transglutaminases or rA 2 Ј. Ionic strength was adjusted with NaCl as specified. For monitoring isopeptide formation, fluorescence readings were expressed as a percentage of the initial fluorescence measured in the absence of enzyme.

RESULTS AND DISCUSSION
Three representative preparations from the family of transglutaminases, exhibiting different kinetic and physical proper-ties, were employed in this study. The cytosolic transglutaminases were purified from human red cells and from guinea pig liver, whereas the activated form of human fibrin stabilizing factor (Factor XIII a ϭ A 2 *) was generated by treatment with thrombin and Ca 2ϩ from the recombinant placental rA 2 zymogen. The findings presented in Fig. 1 serve as evidence that transglutaminases can be effective in catalyzing the cleavage of the ␥:⑀ isopeptide bond. In panel A (lane 2), thin layer chromatographic separation was used to demonstrate the production of fluorescent ␣-Dns-Lys-NHCH 3 , marked as P 1 , during the liver transglutaminase-catalyzed hydrolysis of the branched peptide substrate: Boc-Glu-[␥-⑀(␣-Dns-Lys-NHCH 3 )]-Gln-Ile-Val-Gly-Pro-Leu (VIII). No release of P 1 , identified by the mobility (R F ϳ 0.6) of the reference compound XIII in lane 3, occurred with the enzyme when Ca 2ϩ was replaced by EDTA, as in lane 1. Analysis by HPLC, presented in panel B, confirmed the formation of ␣-Dns-Lys-NHCH 3 as the fluorescent product P 1 (graph 2) eluting at the position (ϳ14.5 min) of the reference compound XIII (graph 3), while the amount of the starting fluorescent substrate S (i.e. compound VIII, eluting at ϳ36.5 min) diminished by about 60% (compare graphs 1 and 2). Monitoring by absorbance at 220 nm revealed the production of a non-fluorescent peak, representing the second product of hydrolysis in the Ca 2ϩ -containing enzymatic mixture, with an elution time (ϳ32.5 min) corresponding to that of the Boc-Glu-Gln-Ile-Val-Gly-Pro-Leu reference (XV; data not shown). These findings established the catalytic potential of transglutaminases for hydrolyzing the isopeptide bond. However, it became obvious that a more penetrating evaluation of branched substrates would require a different analytical approach, and, with this in mind, fluorescence quenching procedures were explored first for the enzyme-catalyzed conventional reaction of a Gln residue-containing acceptor with a primary amine as the donor substrate.
Isopeptide Formation Monitored by the Quenching of Fluo-rescence-Acceptor substrates were synthesized with a fluorescent N-terminal blocking group (Dns), whereas the donor substrates (cadaverine or Lys-NHCH 3 ) contained some quenching moiety (Dnp or Dbc) in the N-terminal position. Affinities of the new substrates for the enzymes were sufficiently favorable so that they could be employed at low enough concentrations for minimizing bimolecular quenching in the starting mixtures. However, as the coupling reaction for forming the ␥:⑀ product progressed with the addition of enzymes, a significant drop in fluorescence ensued. This was attributed to the intramolecular quenching effect exerted by the Dnp or Dbc group on the Dns fluorophore concomitant with forming the isopeptide linkage. This approach, tested for a variety of substrate pairs, proved to be highly sensitive for measuring the rate of isopeptide formation. Catalysis by the human red cell and the guinea pig liver enzyme was explored either with Dnp-cadaverine (XI), Dbccadaverine (XIV), or ␣-Dnp-Lys-NHCH 3 (XII) as donor, in conjunction with Dns-Eaca-Gln-Gln-Ile-Val (I) as the acceptor substrate. Figs. 2 and 3 illustrate the findings for the tissue type of transglutaminases. Fig. 2 pertains to the reaction of the Dnslabeled acceptor (I) with Dnp-Lys-NHCH 3 (XII) as promoted by guinea pig liver transglutaminase, whereas Fig. 3 presents the data for the human red cell enzyme-catalyzed reaction between I and Dbc-cadaverine (XIV). Dns-Eaca-Pro-Ala-Gln-Gln-Ile-Val (X) could be used as first substrate either with Dbc-cadaverine (XIV) or with Dnp-cadaverine (XI) as second substrate for following the coupling reactions catalyzed by rA 2 * (data not shown). These results guided our synthetic work for designing ␥-branched peptides for examining the isopeptide breaking potentials of transglutaminases.
Transglutaminase-catalyzed Breaking of the Isopeptide Bond-We synthesized a number of intramolecularly quenched compounds containing an isopeptide linkage (described under "Materials and Methods") and followed the actions of the enzymes by the increase in fluorescence resulting from the release of the quenching moiety (Dnp) attached to the leaving amine group. On the ␥ side of the backbone the substrates carried either a Dns or an Abz fluorophor in an N-terminal position. ing the hydrolytic cleavage of compound III (0.1 mM) in Fig. 4 than the concentration of the enzyme (0.21 M) for generating the highest rate of cross-bridge formation in the coupling reaction between compounds (I; 0.24 mM) and XII (0.5 mM), depicted by solid triangles (ç) in Fig. 2. Because of inherent ambiguities in comparing rate constants for the reaction of a single substrate (as in the experiment in Fig. 4) with those for a two-substrate reaction (as in Fig. 2), we did not attempt to draw kinetic comparisons between these different enzymatic processes.
Some of the ␥-branched peptides also satisfied the more restrictive specificity requirement of human Factor XIII a . Hydrolysis of Dns-Eaca-Glu[␥-(Cad-Dnp)]-Gln-Ile-Val (II) by the human red cell transglutaminase proceeded fastest in the pH 6.5-7.5 range (ionic strength ϳ 0.1), with reaction rates falling off on either side of this pH range (Fig. 8). For the guinea pig liver enzyme, the apparent optimum for the hydrolysis of the same substrate was in the pH 5.5-7 range (data not shown).
Tissue transglutaminases are negatively regulated by GTP for incorporating amines into protein substrates (42)(43)(44). We tested the effects of equimolar GTP⅐Mg 2ϩ complexes on the isopeptidase activities of the enzymes. As illustrated in Fig. 9, GTP was found to exert a very strong inhibitory effect on the hydrolysis of Dns-Eaca-Glu[␥-(Cad-Dnp)]-Gln-Ile-Val (II) by the human red cell enzyme (I 50 ϳ 5-10 ϫ 10 Ϫ6 M GTP). However, even the ϳ90% inhibition caused by 2 ϫ 10 Ϫ5 M GTP⅐Mg 2ϩ at low Ca 2ϩ (10 Ϫ3 M), could be overcome substantially at higher concentrations of Ca 2ϩ (Fig. 10). This finding supports the concept (43) that the binding of GTP causes a reduction in the Ca 2ϩ sensitivity of transglutaminase. The inhibitory effect of GTP␥S was indistinguishable from that of GTP (data not shown).
Transglutaminases are known to catalyze the hydrolysis of protein or peptide-bound Gln to Glu residues (45), and also to hydrolyze nitrophenyl and thiocholine esters (3,4,24). The hydrolytic nature of transglutaminases is brought even more to the forefront by the examples provided in this paper for the breaking of isopeptide linkages, with high affinities for the ␥-branched substrates (e.g. K m ϳ 10 Ϫ5 M for substrate II by the human red cell enzyme; Fig. 5). Inhibition by the active-site directed blocking agent: 1,3,4,5-tetramethyl-2-[(2-oxopropyl)thio]-imidazolium chloride (Fig. 7) and also by GTP or GTP␥S (Figs. 9 and 10), which modulate the Ca 2ϩ sensitivities of cytosolic transglutaminases, show that the same functional domains are involved in the expression of isopeptidase activities as in the well studied transamidating reactions promoted by these enzymes.
Claims have been made (46,47) and refuted (48) for having isolated an isopeptidase, called destabilase, from leech saliva with specificity for hydrolyzing the N ⑀ -(␥-glutamyl)lysine bonds between the ␥-chains of solubilized fibrin. The primary structure of destabilase, derived from the cDNA clone, does not share significant homology with transglutaminase (49). Based on the findings described in the present paper, it may be suggested that if a select group of enzymes exists with properties that would uniquely define them as isopeptidases, they would probably also display transamidating activities, the characteristic attributes of transglutaminase. One of the enzymes, Factor XIII a , which was shown in our experiments to exhibit isopeptidase activity (Figs. 6 and 7), has been reported to actually hydrolyze the cross-link formed between ␣ 2 -plasmin inhibitor and fibrinogen, and to a lesser extent also the crosslink between the inhibitor and fibrin (50,51). Altogether, our findings with the cytosolic enzymes suggest that transglutaminases may play a more dynamic role in cell biology than hitherto envisaged, not only by catalyzing the formation but also the breaking of N ⑀ -(␥-glutamyl)lysine bonds.