Arginine-specific adenosine diphosphate (ADP)-ribosyltransferase catalyzes transfer of the ADP-ribose moiety of NAD to an arginine residue of a target protein or simple guanidino compounds such as arginine, forming ADP-ribose-acceptor adducts [1,2]. On the other hand, ADP-ribosylarginine hydrolase cleaves the bond between ADP-ribose and arginine [3]. In nitrogen-fixing bacteria, Rhodospirillum rubrum, ADP-ribosylation-de-ADP-ribosylation of a specific arginine residue of dinitrogenase reductase regulates the enzyme activity [4,5]. ADP-ribosyltransferase cDNAs were cloned from rabbit [6] and human [7] skeletal muscles, chicken bone marrow cells [8] and erythroblasts [9], and mouse lymphoma cells [10,11], and ADP-ribosylarginine hydrolase cDNAs were from rat [12], mouse [13] and human [13] brains.

The sulfhydryl group of a cysteine residue plays crucial roles in catalysis of a number of enzymes, including caspases [14], aldehyde dehydrogenase [15] and phospholipase A₂[16]. Chemical modification using thiol-directed agents and site-directed mutagenesis have been used to determine the active site cysteine residues. We previously reported that arginine-specific ADP-ribosyltransferase purified from chicken polymorphonuclear granulocytes (heterophils) required DTT or 2-mercaptoethanol for activity [17]. Subsequently, the functional expression of the arginine-specific ADP-ribosyltransferase AT1 cDNA from chicken bone marrow cells revealed that 2-mercaptoethanol was required for activity of the AT1 enzyme, the heterophil transferase [8]. However, molecular determinants for the actions of sulfhydryl reducing agents on these ADP-ribosyltransferases have remained to be identified.

Arginine-specific ADP-ribosyltransferase activity has also been detected in mouse lymphoid cells, including T-cell lymphomas [18], splenocytes [18,19] and activated cytotoxic T lymphocytes (CTL) [20]. In mouse CTL, a 35-kDa glycosylphosphatidylinositol (GPI)-anchored arginine-specific ADP-ribosyltransferase was found to mediate suppression of the proliferation and cytotoxic activity of CTL [20]. As a candidate gene encoding the transferase molecule responsible for modulating CTL activity, a GPI-anchored transferase (Yac-1) cDNA was cloned from mouse T-cell lymphomas [10,21]. The Yac-1 gene is expressed predominantly in skeletal muscle, and to a lesser extent in lymphocytes [21]. Other candidate genes are mouse Rt6.1 [22] and Rt6.2 [23], the products of which have structural homology to arginine-specific ADP-ribosyltransferases. Both Rt6 genes are expressed as GPI-linked membrane proteins on mature mouse T lymphocytes, but differentially in distinct mouse strains [23,24]. Recombinant Rt6.1 and Rt6.2 proteins exhibit arginine-specific ADP-ribosyltransferase activity [24–28]. In contrast with mouse Rt6s, two rat alloantigens RT6.1 [27,29] and RT6.2 [30] are primarily NAD glycohydrolases, but not ADP-ribosyltransferases. However, by substitution of the glutamine residue at position 207 in rat RT6.1 with glutamic acid, the antigen exhibits arginine-specific ADP-ribosyltransferase activity, thus capable of modifying exogenous substrates such as l-arginine [27]. Some relations have been implicated between defects in Rt6/RT6 expression and occurrence of autoimmune diseases in rodents [31–33].

We now report that the ADP-ribosyltransferase activity of Rt6.1, but not Rt6.2, is thiol-dependent and that single amino acid replacement of the recombinant Rt6.1 alters thiol-dependency of its transferase activity.

Materials and methods

Preparation of recombinant proteins

The site-directed mutants of mouse Rt6.1 were made essentially as described [27], using pMAL-p2-MRT6H [27] as a template and oligonucleotide primers: C80S-Rt6.1, 5′-G GAG ATC AAA AAT AGT ATG AGT TAT CCG GC-3′; C201F-Rt6.1, 5′-G GCT CAC ATC AAA CAT TTT TCC TAC TAT ACT C-3′, where underlined codons were altered. Mouse Rt6.2 cDNA was amplified from BALB/c mouse spleen poly(A)⁺ RNA by reverse transcription/polymerase chain reaction, using 5′-GGT ACC GTC GAC GGC CTT GCA GTG CCT TTC-3′ (sense) and 5′-TCT AGA AGC TTC AAC TGA AAT CAA TGT TGA C-3′ (antisense). Construction and sequence of the expression plasmids were confirmed by entire sequencing, in both directions. A pMAL-p2 plasmid vector carrying mutant Rt6.1 or wild-type Rt6.2 was introduced and expressed in Escherichia coli K12TB1, producing a fusion protein with a maltose-binding protein (MBP) [27]. As noted for wild-type Rt6.1 (MBP-Rt6.1) [27], mutant Rt6.1 and wild-type Rt6.2 (MBP-Rt6.2) were not transported into the periplasm. Thus, E. coli expressing a recombinant protein was sonicated in the presence of 20 mm Tris/Cl^– (pH 7.5), 1 mm EDTA, 1% Triton X-100, 1 mm PhCH₂SO₂F, and 10 μg·mL^–1 leupeptin. After centrifugation at 100 000 g for 60 min, a clear supernatant (E. coli crude extract) was used as the source of MBP-Rt6.1, mutant Rt6.1 or MBP-Rt6.2. Expression of the recombinant proteins with their expected molecular masses (70 kDa) was confirmed by Western blot analysis [27], using antimouse Rt6 peptide antibodies (see below) and ECL detection system (Amersham, Buckinghamshire, UK). For quantitative analysis of ADP-ribosyltransferase activity (see Fig. 4), the recombinant proteins were purified from the crude extracts by sequential chromatographies on DE52 anion exchanger, hydroxyapatite and amylose resin (New England Biolabs, Bevely, MA, USA). MBP-Rt6.2 was further purified on Resourse Q (Pharmacia Biotech, Uppsala, Sweden). Concentrations of the recombinant proteins were estimated by comparison of Coomassie-stained bands in SDS/polyacrylamide gels using a dilution series of BSA as a standard. Recombinant wild-type rat RT6.1 (MBP-RT6.1) and mutant rat RT6.1 (Q207E-RT6.1) were prepared, as described [27].

Results

Splenocytes from a BALB/c mouse were treated with PI-PLC and ADP-ribosyltransferase activity in the medium was determined. PI-PLC liberated 86.0 ± 3.9% (n = 3) of arginine-specific ADP-ribosyltransferase activity from BALB/c mouse splenocytes. As shown in Table 1, the transferase activity in the medium was stimulated fivefold with a sulfhydryl reducing agent DTT (2 mm), indicating the presence of a GPI-anchored thiol-dependent ADP-ribosyltransferase on the surface of BALB/c mouse splenocytes. The medium was then subjected to immunoprecipitation with the increasing amounts of anti-RT6 antibodies, and the ADP-ribosyltransferase activities in both precipitates and supernatants were determined in the presence of DTT. As shown in Fig. 1A, most of the activity was detected in the fraction precipitated from the medium by the anti-RT6 antibodies (87.0 ± 1.7%, n = 3). The immunoprecipitated activity was stimulated by DTT, in a dose-dependent manner (Fig. 1B). With 0.2 mm DTT, ADP-ribosylarginine formation was stimulated 6.1-fold (Fig. 1B). Half-maximal stimulation was observed with 15–20 μm DTT (Fig. 1B). As shown in Fig. 1C, basal and stimulated activities of the immunoprecipitated transferase were inhibited by a sulfhydryl alkylating agent NEM. In contrast, ADP-ribosyltransferase activity liberated from C57BL/6 mouse splenocytes with PI-PLC was not increased by DTT (Table 1). ADP-ribosyltransferase activity in BALB/c mouse skeletal muscle was neither stimulated by DTT (Table 1) nor immunoprecipitated by the anti-RT6 antibodies (see Materials and methods). In the C57BL/6 mouse, the coding sequence of Rt6.1, but not Rt6.2, contains an in-frame stop codon [24], resulting in the lack of enzymatically active Rt6.1 on C57BL/6 mouse splenocytes. Thus, the thiol-dependent ADP-ribosyltransferase activity liberated from BALB/c mouse splenocytes with PI-PLC may be due to Rt6.1, rather than Rt6.2.

To determine whether ADP-ribosyltransferase activity of mouse Rt6.1, but not Rt6.2, indeed exhibits thiol dependency, we expressed Rt6.1 and Rt6.2 as an MBP-linked fusion protein (MBP-Rt6.1 and MBP-Rt6.2, respectively) in E. coli and examined the effects of DTT and/or NEM on transferase activities of the recombinant proteins. As noted earlier [24,27,28], both MBP-Rt6.1 and MBP-Rt6.2 catalyzed ADP-ribosylarginine formation (Fig. 2). The ADP-ribosyltransferase activity of MBP-Rt6.1 was stimulated with DTT, in a dose-dependent manner (Fig. 2), in much the same manner as observed for ADP-ribosyltransferase immunoprecipitated with anti-RT6 antibodies from PI-PLC supernatant of BALB/c mouse splenocytes (Fig. 1A). With 0.2 mm DTT, ADP-ribosylarginine formation was stimulated 6.4 ± 0.9-fold (n = 3). Half-maximal stimulation was observed with 15–20 μm DTT (Fig. 2). In contrast to MBP-Rt6.1, the transferase activity of MBP-Rt6.2 was not increased by DTT (Fig. 2). The Glu-207 mutant of rat RT6.1 (Q207E-RT6.1), which is an arginine-specific ADP-ribosyltransferase and capable of modifying exogenous substrates such as l-arginine and histone [27], did not require DTT for maximal activity (Fig. 2). These results indicate that recombinant mouse Rt6.1 but not Rt6.2 is a thiol-dependent arginine-specific ADP-ribosyltransferase.

All family members of eucaryotic arginine-specific ADP-ribosyltransferases have four conserved cysteine residues [26]. Mouse Rt6 and rat RT6 antigens also have these conserved cysteine residues (amino acid residues 41, 141, 193, and 246 in Rt6.1). In addition to these cysteines, mouse Rt6.1 has two extra cysteine residues (Cys-80 and Cys-201), while amino acid residues corresponding to Cys-80 and Cys-201 of Rt6.1 are serine and phenylalanine, respectively, in both mouse Rt6.2 and rat RT6.1 (Fig. 3). Thus, we hypothesized that Cys-80 and/or Cys-201 in mouse Rt6.1 may account for thiol dependency of the transferase activity. To gain support for this hypothesis, we introduced site-directed mutations into mouse Rt6.1 cDNA to replace two cysteine residues (Cys-80 and Cys-201) with serine (C80S-Rt6.1) and phenylalanine (C201F-Rt6.1), respectively, and then examined the effects of DTT and/or NEM on transferase activities of these mutant proteins expressed in E. coli. As shown in Fig. 4C, ADP-ribosyltransferase activity of C201F-Rt6.1 was neither stimulated by DTT, nor inhibited by NEM, as was the case for MBP-Rt6.2 (Fig. 4B) and Q207E-RT6.1 (data not shown). In contrast, DTT stimulated the transferase activity of C80S-Rt6.1 twofold (Fig. 4D), and NEM inhibited the stimulated as well as basal activities of C80S-Rt6.1 (Fig. 4D), as was seen in the case for MBP-Rt6.1 (Fig. 4A). Thus, Cys-201 in mouse Rt6.1 confers thiol dependency on Rt6.1 transferase. Using purified recombinant proteins, we determined the specific activities of these transferases. The results revealed that basal activity of Rt6.1 was similar to that of Rt6.2 (Fig. 4A,B). The activity of Rt6.1 in the presence of DTT was eightfold higher than that of Rt6.2.

Discussion

In the present study, we found that recombinant mouse Rt6.1 is a thiol-dependent arginine-specific ADP-ribosyltransferase. The transferase activity of the mutant Rt6.1 in which Cys-201 was replaced with phenylalanine lost thiol dependency (Fig. 4C) while that of the other mutant in which Cys-80 was replaced with serine remained thiol-dependent (Fig. 4D). Thus, we concluded that the Cys-201 in Rt6.1 confers thiol dependency on the ADP-ribosyltransferase activity of Rt6.1. As the Ser-80 mutant was not stimulated in the presence of DTT so much as wild-type Rt6.1, Cys-80 may somewhat contribute to thiol dependency of Rt6.1 transferase. The requirement of a sulfhydryl reducing agent for maximum activity of Rt6.1, together with inactivation by sulfhydryl alkylation, suggests that the Cys-201 plays important parts in the catalytic reaction. However, replacement of the Cys-201 with phenylalanine did not abolish the ADP-ribosyltransferase activity of Rt6.1 (Fig. 4C). Thus, although the sulfhydryl group of Cys-201 in Rt6.1 is not essential for the catalysis, reduction of the sulfhydryl group of the cysteine residue facilitates the rate of arginine-specific ADP-ribosylation. In this regard, the mouse Rt6.1 transferase differs from the chicken heterophil ADP-ribosyltransferase which exhibits an absolute requirement of a sulfhydryl reducing agent for the activity [17].

Treatment of BALB/c mouse splenocytes with PI-PLC liberated 86% of arginine-specific ADP-ribosyltransferase activity into the medium, which was almost completely precipitated by anti-RT6 antibodies (Fig. 1A). The transferase activity precipitated with the anti-RT6 antibodies was stimulated by DTT (Fig. 1B), in much the same manner as observed for MBP-Rt6.1 (2, 4). cDNAs for arginine-specific ADP-ribosyltransferases heretofore cloned from lymphoid tissues are those for Rt6.1 [22], Rt6.2 [23], Yac-1 [10] and Yac-2 [11]. Yac-1 is a GPI-anchored transferase [10], as in the case of Rt6.1 and Rt6.2. Yu et al. cloned the same gene as Yac-1 from murine T-cell lymphoma SL12 [21]; the gene is expressed predominantly in skeletal muscle, but to a lesser extent in lymphocytes [21]. The activity of mouse skeletal muscle ADP-ribosyltransferase (Yac-1) was not stimulated by DTT (Table 1) and the anti-RT6 antibodies used in our study did not cross-react with the mouse skeletal muscle transferase, as noted above. In addition, Yac-2 is not GPI-anchored, although it is membrane-bound [11]. From these observations, neither Yac-1 nor Yac-2 would account for the transferase activity immunoprecipitated by anti-RT6 antibodies from PI-PLC supernatant of BALB/c mouse splenocytes. Thus, most of arginine-specific ADP-ribosyltransferase activity detected on BALB/c mouse splenocytes is attributed to Rt6.1 antigen, such being consistent with much higher expression of Rt6.1 than Rt6.2 in BALB/c mouse splenocytes [23]. Soman et al. previously reported the presence of a guanidino group-specific ADP-ribosyltransferase on the murine T-cell lymphomas and stimulation of the activity by DTT [18]. The transferase described by Soman et al. may possibly be Rt6.1.

Our evidence shows that recombinant mouse Rt6.1 is a thiol-dependent arginine-specific ADP-ribosyltransferase and such is the case for native Rt6.1 expressed on BALB/c mouse splenocytes. In the C57BL/6 mouse, coding sequence of Rt6.1, but not Rt6.2, has been reported to contain an in-frame stop codon [24], resulting in the lack of enzymatically active Rt6.1 on C57BL/6 mouse splenocytes. Another GPI-anchored lymphoid transferase Yac-1 does not seem to be a thiol-dependent ADP-ribosyltransferase (Table 1). Assuming that the native Rt6.2 is not a thiol-dependent ADP-ribosyltransferase, as observed for the recombinant Rt6.2 (2, 4), ADP-ribosyltransferase activity detected in the PI-PLC supernatant from C57BL/6 mouse splenocytes would not be stimulated by DTT. Indeed, addition of 2 mm DTT to the reaction mixture did not increase ADP-ribosyltransferase activity liberated with PI-PLC from C57BL/6 mouse splenocytes (Table 1). This is consistent with the idea that the native Rt6.2 is not a thiol-dependent ADP-ribosyltransferase. Thus, even though Rt6.1 and Rt6.2 both catalyze arginine-specific ADP-ribosylation [24,27,28], Rt6.1 differs from Rt6.2, in enzymatic properties of the transferase and perhaps physiological functions or regulatory mechanisms of the enzyme activity, as we have described.

It has been demonstrated that activated macrophages tightly bind T cells during a course of immune response and release thiol compounds into culture medium to modulate functions of T cells in their vicinity [35]. When a T cell expressing Rt6.1 on the surface interacts with the macrophage in the presence of NAD, which may be released from cells as a consequence of cell lysis during inflammation, the activity of Rt6.1 transferase would be stimulated by thiols released from the macrophage. As ADP-ribosylation of T-cell surface proteins inhibits its functions such as cytotoxicity [20], homing [36] and antigen-stimulated responses [36], the modification of the target proteins by thiol-stimulated Rt6.1 ADP-ribosyltransferase may result in down-regulation of T-cell functions. Recently, a hydrolase has been described in the mouse that cleaves the ADP-ribosylarginine bond, and the hydrolase activity has been shown to be thiol-dependent [12]. The hydrolase activity was detected in rat cerebrospinal fluid, suggesting the presence of extracellular hydrolase [37]. Removal of ADP-ribose from target proteins on the surface of T cells by the hydrolase may serve to reverse the effects of ADP-ribosylation, but the significance of this reaction in T-cell functions remains to be elucidated. In contrast with Rt6.1, the activity of Rt6.2 transferase is not stimulated by thiols, but there may exist unknown factors or signaling pathways to stimulate its activity, thus enabling Rt6.2 to function as an immunoregulatory molecule. Whether these regulatory mechanisms mediated by Rt6 ADP-ribosyltransferases are operating in vivo to modulate T-cell functions, to what extent Rt6.1 and Rt6.2 could contribute to regulation of T-cell functions, and whether defective regulation of the enzymatic activities of Rt6s could develop abnormal immune responses, such as immunity to self-antigens are the subjects of ongoing study.

Acknowledgements

We thank H. Osago for technical assistance. This work was supported by Grants-in-Aid for Scientific Research, 08680687 from the Ministry of Education, Science, Sports and Culture, Japan.