INTRODUCTION

Rat RT6 proteins appear to identify a regulatory T cell population capable of modulating autoimmunity (1). Diabetes-prone BB rats are severely deficient in RT6⁺ T cells and spontaneously develop insulin-dependent diabetes mellitus (2, 3). Transfusion of spleen cells from histocompatible rats prevents diabetes provided that RT6⁺ T cells become engrafted (4, 5). Conversely, diabetes-resistant BB rats have normal numbers of RT6⁺ T cells and fail to manifest autoimmunity (1, 2). Depletion of RT6⁺ T cells induces insulitis, diabetes, and thyroiditis in these animals (6-8). RT6⁺ T cells have been suggested to be important also in the modulation of autoimmunity in the irradiated, thymectomized PVG model of insulin-dependent diabetes mellitus (9), and in the BN rat model of mercury chloride-induced autoimmune glomerulonephritis (10). The mechanism(s) by which RT6⁺ T cells regulate autoimmunity is uncertain, but may involve the RT6 protein itself.

The rat RT6 proteins are linked to the surface of T cells via a glycosylphosphatidylinositol (GPI)¹-linkage (11, 12). Deduced amino acid sequences obtained from cDNA for two rat allotypes, RT6.1 and RT6.2, with molecular masses of 25-35 kDa and 24-26 kDa, respectively, contain hydrophobic amino- and carboxyl-terminal sequences, consistent with those in GPI-linked proteins (13, 14). The internal, coding region is remarkable for consensus sequences similar to those found in bacterial toxin ADP-ribosyltransferases and believed to be involved in catalytic activity (15). Consistent with this observation, the RT6 protein exhibited NAD glycohydrolase (15) and auto-ADP-ribosyltransferase activities (16, 17). These enzymatic activities of RT6 have recently been postulated to be important in the inhibition of T cell proliferation following incubation of cells in the presence of NAD (16).

Two mouse Rt6 genes have recently been identified and are transcribed (18-20). Expression of mouse Rt6 protein by immunoblotting and RT6 gene expression appear to be restricted to T cells (18, 21). Reduced levels of Rt6 gene expression have been reported in two models of autoimmunity in mice (18, 19). In the spontaneously diabetic nonobese diabetic mouse, levels of Rt6-specific mRNA in the spleens of 8-week-old mice were lower than those of control nonobese nondiabetic mice (18). In the NZW and (NZB + NZW)F₁ mouse, which are models of systemic lupus erythematosus, defects in the structure and expression of the Rt6 gene were also observed (19). Similar to the rat homologue, mouse Rt6.1 had auto-ADP-ribosylation activity (16, 21). The immunomodulatory function of mRt6⁺ T-lymphocytes, or the role of mRt6 enzymatic activities in immune modulation, have not been determined. Recently, it was suggested that cytotoxic T cell (CTL) activity in mice may be regulated by GPI-linked ADP-ribosyltransferases that have many of the characteristics of the rat RT6 proteins (22, 23) and catalyze the modification of surface proteins (24).

Although various members of the rat RT6 and mouse Rt6 family of proteins possess NAD glycohydrolase and auto-ADP-ribosyltransferase activity, it was not evident that they could catalyze intermolecular reactions as described for CTL cells. In fact, the rat RT6.2 protein, which was shown to be an NAD glycohydrolase (15) and auto-ADP-ribosyltransferase, specifically did not catalyze the ADP-ribosylation of protein and simple guanidino compounds such as arginine and agmatine (16, 17). The present study was undertaken to determine whether the mRt6.1 protein, in contrast to the rat RT6 protein, could catalyze intermolecular reactions such as the ADP-ribosylation of free proteins and agmatine. These enzymatic activities would be consistent with a possible immunomodulatory function in mouse CTL and provide a potential mechanism by which mRt6⁺ T cells, and/or protein could influence autoimmunity.

EXPERIMENTAL PROCEDURES

Materials

pAcSG2 vector was purchased from PharMingen (San Diego, CA); pMAMneo expression vector and ligation express kit from Clontech (Palo Alto, CA); gene Amp PCR reagent kit and DNA thermal cycler from Perkin-Elmer; ABI 394 DNA/RNA synthesizer from Applied Biosystems, Inc. (Foster City, CA); NuSieve and GTG-agarose from FMC Corp. (Rockland, ME); QIAEX gel extraction kit and QIAGEN plasmid purification kit from QIAGEN (Chatsworth, CA); Tris acetate-EDTA buffer, LB broth, Eagle's minimal essential medium with Earle's balanced salt solution containing L-glutamine, Dulbecco's phosphate-buffered saline (DPBS), and fetal bovine serum, heat-inactivated, from BioWhittaker, Inc. (Walkersville, MD); XhoI and ampicillin, sodium salt from Boehringer Mannheim; NheI, DH5-competent cells, G418, buffer-saturated phenol, and phenol/chloroform/isoamyl alcohol (25:24:1) from Life Technologies, Inc.; 3 M sodium acetate from Quality Biological Inc. (Gaithersburg, MD); ethanol from the Warner Graham Co. (Cockeyesville, MD); rat mammary adenocarcinoma cells (NMU cells) from American Type Culture Collection (Rockville, MD); dexamethasone sodium phosphate from MG Scientific (Buffalo Grove, IL); trypsin and 7-deaza-dGTP sequencing kit from U. S. Biochemical Corp. (Cleveland, OH); Hoechst 33258 Dye and DyNa Quant 200 fluorometer from Hoefer Pharmacia Biotech, Inc. (San Francisco, CA); -NAD, phosphatidylinositol-specific phospholipase C (PI-PLC), and agmatine sulfate from Sigma; [adenine-U-¹⁴C]NAD (243 mCi/mmol), [carbonyl-¹⁴C]nicotinamide adenine dinucleotide (53 mCi/mmol), and L-[U-¹⁴C]arginine monohydrochloride (297 mCi/mmol) from Amersham Corp.; Dowex AG1-X2 resin, 200-400 mesh, chloride form, from Bio-Rad; Long Ranger gel solution from J. T. Baker Inc.; and BCA protein assay reagent from Pierce.

Methods

mRt6.1 cDNA was subcloned into a pMAMneo expression vector as described previously (25). The mRt6.1 insert was sequenced to confirm its identity with that in GenBank^TM (accession no. X52991[GenBank]). The mRt6.1 cDNA was amplified from purified plasmid cDNA (mRt6.1 cDNA in pAcSG2 vector) by PCR, using forward (5-ACG ACG GCT AGC ATG CCA TCA AAT AAT TTC AAG-3) and reverse (5-CGT CGT CTC GAG CTA CGG CTC AGC AAG AGT AAG-3) primers (100 pmol each) under the following conditions: 60 s at 95 °C, followed by 25 cycles at 95 °C for 60 s; 55 °C, 60 s; 72 °C, 90 s; and ending at 72 °C for 7 min. The PCR product was isolated on a 3% NuSieve GTG-agarose gel followed by QIAEX purification of the gel band. The purified PCR fragment was digested with XhoI and NheI and ligated into the NheI (5) and XhoI (3) sites of the pMAMneo vector. DH5 competent cells were transformed with the ligated product and grown on LB plates containing ampicillin (50 µg/ml). The subcloned mRt6.1 cDNA in pMAMneo was amplified, and then purified using the QIAGEN plasmid purification kit. The subclone was sequenced using the 7-deaza-dGTP sequencing kit to verify orientation. NMU cells were transfected with 30 µg of the mRt6.1 subclone by the calcium phosphate/HEPES-buffered saline precipitation method. Transformed NMU cells were selected with G418 (500 µg/ml) and maintained in Eagle's minimal essential medium containing 10% fetal bovine serum with G418 (500 µg/ml).

Expression of ADP-ribosyltransferase was induced by incubating 2.5 × 10⁶ transformed NMU cells with 1 µM dexamethasone sodium phosphate for the indicated times (6, 24, 48, and 72 h). Cells were washed with DPBS and incubated with or without 2 units of PI-PLC in 1.5 ml of DPBS at 37 °C. The media were collected. Cells were trypsinized, washed twice in DPBS, and lysed in 0.7 ml of hypotonic lysis buffer (10 mM potassium phosphate, pH 7.5, 1 mM EDTA) with three freeze/thaw cycles (dry ice/room temperature). Lysed cells were centrifuged for 1 h at 100,000 × g at 4 °C. The supernatant (0.7 ml) was collected; the membrane fraction was suspended in 0.7 ml of lysis buffer and sonified. ADP-ribosyltransferase activity was determined in samples (75 µl) of the medium, cell supernatant, and cell membranes. Data are expressed as total activity per fraction (picomoles of [¹⁴C]ADP-ribosylagmatine formed in 26 h in the transferase assay).

The ADP-ribosyltransferase reaction was carried out for 26 h in 0.3 ml of reaction mix containing 50 mM potassium phosphate, pH 7.5, 0.1 mM nicotinamide [adenine-U-¹⁴C]NAD (0.05 µCi), with or without 20 mM agmatine. Duplicate samples were incubated at 30 °C, and at the end of the incubation, two 100-µl samples from each assay were applied to columns (1 ml bed volume) of Dowex AG1-X2 resin (26). [¹⁴C]ADP-ribosylagmatine was eluted with 5 ml of H₂O for radioassay.

Protein was determined using BCA protein assay reagent (Pierce) with bovine serum albumin as a standard.

ADP-ribosylarginine hydrolase, purified from rat brain as described earlier, contained one major 39-kDa protein band on SDS-polyacrylamide gel electrophoresis (27). Turkey erythrocyte ADP-ribosyltransferase was purified as described previously (28).

Recombinant mRt6.1 and RT6.2 were synthesized in Sf9 cells under the control of a baculovirus promoter, as described previously (16). Two rat RT6.2 genes were expressed in Sf9 cells, RT6.2-GB (accession nos. M85193[GenBank] and M30311[GenBank]) and RT6.2-LR. The latter gene, from Lewis.B (RT6.2) rats, differs from RT6.2-GB at position 29 (lysine in place of threonine), position 71 (arginine in place of lysine), position 96 (alanine in place of valine) and position 251 (a stop signal in place of arginine).

RESULTS AND DISCUSSION

mRt6.1 was synthesized in Sf9 insect cell/baculovirus system and in transformed rat NMU cells. Transferase activity was found in the Sf9 cell pellet, cell supernatant, and the culture supernatant (data not shown). Using mRt6.1 isolated from Sf9 medium in assays containing either [adenine-U-¹⁴C]NAD or [carbonyl-¹⁴C]NAD to monitor the incorporation of ADP-ribose into ADP-ribosylagmatine and the release of nicotinamide from NAD, respectively, it was observed that the rate of formation of ADP-ribosylagmatine was closely correlated with the rate of release of nicotinamide, with a ratio of [adenine-U-¹⁴C]ADP-ribosylagmatine formation to [carbonyl-¹⁴C]nicotinamide release of 0.99 (Table I). These data are compatible with the conclusion that, in the presence of an ADP-ribose acceptor, the enzyme is primarily a transferase, with only minimal, uncoupled NAD glycohydrolase activity. In the absence of agmatine or other ADP-ribose acceptors, the rate of release of [carbonyl-¹⁴C]nicotinamide from [carbonyl-¹⁴C]NAD was significantly less than that observed in the presence of acceptor (Table I). Again, these data support the hypothesis that the enzyme is not as efficient as an NAD glycohydrolase as it is an ADP-ribosyltransferase. In contrast, the rat RT6.2, synthesized in the Sf9 system, did not appear to catalyze the formation of [adenine-U-¹⁴C]ADP-ribosylagmatine from [adenine-U-¹⁴C]NAD and agmatine, but did, as reported elsewhere, release [carbonyl-¹⁴C]nicotinamide from [carbonyl-¹⁴C]NAD (Table I). For NAD, the K_m of RT6.2-LR was 23.4 µM ± 3.2; the K_m of RT6.2-GB was 41.0 µM ± 9.4; both were significantly lower than that of mRt6.1 (1.9 mM ± 0.3) (Table II). The K_m for NAD of mRt6.1 was similar to that of cholera toxin (29), a bacterial NAD:arginine ADP-ribosyltransferase. The mRt6.1 K_m for agmatine of 4.8 mM ± 0.3 was significantly lower than that reported for the toxin (30).

Table I. Comparison of [adenine-U-14C]ADP-ribosylagmatine formation and release of [carbonyl-14C]nicotinamide catalyzed by mRt6.1 and rat RT6.2 Medium (10 µl) from Sf9 cells expressing rat RT6.2-LR was assayed for transferase activity by incubation in 50 mM potassium phosphate, pH 7.5, 0.1 mM [adenine-U-14C]NAD (~43,000 cpm) with or without 20 mM agmatine (Ag) for 1 h at 30°C in a total volume of 150 µl. Similar results were obtained with RT6.2-GB. RT6.2 medium (20 µl) was assayed for NADase activity by replacing [adenine-U-14C]NAD with 0.1 mM [carbonyl-14C]NAD (~37,000 cpm), and incubating as above for 1.5 h. Medium from Sf9 cells expressing mouse Rt6.1 (25 µl) was assayed for transferase activity as described above except for the presence of 1 mM [adenine-U-14C]NAD (~45,000 cpm) with or without 100 mM agmatine. NADase activity was determined by substituting 1 mM [carbonyl-14C]NAD (~40,000 cpm) for the radiolabel. Two 50-µl samples of the assay mix were quantified as described in the legend to Table II. Values are mean ± S.E. (n = 3). Experiments were repeated three times with similar results. Protein [adenine-U-14C]ADP-ribosylagmatine [carbonyl-14C]Nicotinamide No Ag Plus Ag No Ag Plus Ag pmol.h1 mRt6.1 0 4155 ± 409 612  ± 154 4205  ± 789 RT6.2 0 0 5452  ± 970 4492  ± 299

Table II. Kinetic constants for mRt6.1 and rat RT6.2 ADP-ribosylagmatine transferase activity of mRt6.1 from supernatant obtained from Sf9 cells infected with virus containing the mRT6.1 construct was assayed in 50 mM potassium phosphate, pH 7.5, with [adenine-14C]NAD (0.025 µCi), 100 mM agmatine, and an average of 10 concentrations of NAD or with 1 mM NAD and nine different concentrations of agmatine for 1 h at 30°C in a total volume of 150 µl. 50-µl samples of reaction mix were applied to columns (0.5 × 4 cm) of AG1-X2 to separate [14C]ADP-ribosylagmatine for radioassay. NAD glycohydrolase activity of rat RT6.2-LR from supernatant of transfected Sf9 cells was incubated in 50 mM potassium phosphate, pH 7.5, [carbonyl-14C]NAD (0.025 µCi) with seven NAD concentrations as required in a total volume of 150 µl at 30°C for 1.5 h. The reaction mix was quantified as above. NAD glycohydrolase activity of rat RT6.2-GB was assayed as given above except with nine concentrations of NAD. The Km values were determined from a double-reciprocal plot with a least-squares fit of the initial velocity (glycohydrolase or transferase activity), versus NAD or agmatine concentration. All assays were done in duplicate. Experiments were repeated two or three times as noted. Data are means ± one-half the range or ± S.D. Protein Km NAD Agmatine mRt6.1 1.9 mM  ± 0.3 (n = 3) 4.8 mM  ± 0.3 (n = 2) RT6.2-LR 23.4 µM  ± 3.2 (n = 3) RT6.2-GB 41.0 µM  ± 9.4 (n = 4)

To verify that mRt6.1 was indeed catalyzing the formation of an ADP-ribose-linkage to simple guanidino compounds and that the ADP-ribose-guanidino linkage was identical to that formed by erythrocyte and cardiac muscle NAD:arginine ADP-ribosyltransferases (25, 31), [¹⁴C]arginine and saturating concentrations of unlabeled NAD were incubated with or without mRt6.1, and the products resolved by SAX-HPLC (Fig. 1). [¹⁴C]Arginine eluted at 2-4 min; with mRt6.1, a new radiolabeled peak, corresponding to 6.4% of the total radiolabel, was observed at 7-9 min (Fig. 1). To determine whether this peak represented ADP-ribosyl-[¹⁴C]arginine, it was concentrated and incubated with or without Mg²⁺ and DTT in the presence of an ADP-ribosylarginine hydrolase, which cleaves the ADP-ribose-arginine linkage (27, 32). ADP-ribosylarginine hydrolase is dependent on Mg²⁺ and DTT for activity, and degrades stereospecifically the -anomeric product of the erythrocyte transferase-catalyzed reaction (27, 33). The reaction products were again subjected to SAX-HPLC. In the presence of Mg²⁺, DTT, and hydrolase a new peak, eluting at 2-4 min, corresponding to [¹⁴C]arginine and representing 72% of the total radiolabel, was formed (Fig. 2).

mRt6.1 utilized histones as ADP-ribose acceptors and, in the presence of [³²P]NAD, catalyzed the formation of [³²P]ADP-ribosylhistones (Fig. 3). Free ADP-ribose did not inhibit [³²P]ADP-ribosylation of histone, suggesting that a nonenzymatic pathway resulting from the initial generation of ADP-ribose by an NAD glycohydrolase, followed by the nonenzymatic addition of ADP-ribose to protein, was not responsible for the result. In addition, rat RT6.2 is more active as an NAD glycohydrolase than is the mRt6.1 and it did not, under these conditions, cause ADP-ribosylation of histone (Fig. 3). It is noted also that auto-ADP-ribosylation of rat RT6.2 was not inhibited by ADP-ribose and is, in all likelihood, enzymatic. For the avian erythrocyte transferase A, histones serve as both ADP-ribose acceptors and activators of the enzyme (26, 34). To determine whether histones enhanced mRt6.1 activity, the enzyme was incubated with [adenine-U-¹⁴C]NAD and agmatine, and radiolabeled ADP-ribosylagmatine was isolated (Table III). Histones inhibited ADP-ribosylagmatine formation, presumably, in part, by serving as an alternative ADP-ribose acceptor and competing for the catalytic site, but no enhancement of mRT6.1 activity was observed.

Table III. Effect of histones on mRt6.1 and turkey erythrocyte ADP-ribosyltransferase activities in the presence or absence of agmatine Duplicate samples of medium from Sf9 cells infected with virus containing the mouse Rt6.1 construct (25 µl) or virus alone (25 µl) or purified turkey transferase (30 ng) were incubated for 1 h at 30°C in 50 mM potassium phosphate (pH 7.5), 0.1 mM [adenine-14C]NAD (~107,000 cpm) with or without 20 mM agmatine or 50 µg of histone in a total volume of 300 µl. 100-µl samples were applied to columns (0.5 × 4 cm) of AG1-X2 resin, which were eluted with water. Apparent ADP-ribosylagmatine formation in columns 3 and 4 reflects modified histone and/or background. Sample ADP-ribosylagmatine formed 20 mM agmatine No agmatine Plus histone No histone Plus histone No histone pmol/h Sf9, mRt6.1 4390  ± 230 4859  ± 132 152  ± 10 65  ± 5 Sf9, vector 0 0 55  ± 6 55  ± 9 Turkey transferase 6121  ± 389 2756  ± 388 228  ± 17 46  ± 8

Expression of mRt6.1 protein in lymphocytes has been demonstrated (21). It appeared, therefore, that the signal sequence at the carboxyl terminus was recognized by the enzymes involved in the formation of a GPI-linked protein (35). mRt6.1 was expressed under the control of a dexamethasone-regulated promoter in rat NMU cells. Under the assay conditions, NMU cells transfected with vector alone did not exhibit ADP-ribosyltransferase activity (data not shown). The cells, however, possessed significant activities that metabolized NAD, and caused release of nicotinamide (data not shown). For that reason, ADP-ribosyltransferase activity was measured directly with [adenine-U-¹⁴C]NAD and agmatine as substrates, and the formation of [adenine-U-¹⁴C]ADP-ribosylagmatine was monitored. As shown in Table IV, following induction with dexamethasone, NMU cells transfected with the mRt6.1 gene exhibited a time-dependent increase in ADP-ribosyltransferase activity. At 6 h following induction, activity was primarily associated with the cell pellet and was not released with PI-PLC. At 24 h, transferase activity was increased further and was released by PI-PLC. Although the total transferase activity declined after 48- and 72-h incubations with dexamethasone, the fraction susceptible to release by incubation with PI-PLC remained constant. These studies suggest that, following induction by dexamethasone, there is a delay in processing the mRt6.1 transferase to a PI-PLC-releasable form by NMU cells, although the carboxyl-terminal signal sequence in mRt6.1 is sufficient for the addition of the GPI anchor.

Table IV. ADP-ribosyltransferase activity following dexamethasone induction of transformed NMU cells Following incubation with 1 µM dexamethasone for 6, 24, 48, or 72 h, NMU cells (2.5 × 106) transformed with mRt6.1 as described under "Methods" were incubated in DPBS with or without 2 units of PI-PLC for 1 h at 37°C. DPBS fractions were collected, cells lysed, and the lysates centrifuged at 100,000 × g × 1 h. ADP-ribosyltransferase activity was determined in samples (75 µl) of the DPBS (1.5 ml, total volume), cell supernatant (0.7 ml, total volume) and membranes (cell pellet; 0.7 ml, total volume). Induction time ADP-ribosyltransferase activity PI-PLC Medium Supernatant Pellet Total h units pmol/26 h 6 2 (0.20)a (1.06) 656 (0.63) 656 (1.88) 0 (0.17) (1.14) 561 (0.48) 561 (1.79) 24 2 3696 (0.20) (1.77) 1087 (0.87) 4783 (2.84) 0 (0.17) 202 (1.74) 3885 (0.75) 4087 (2.66) 48 2 1194 (0.20) 171 (1.95) 328 (0.78) 1693 (2.92) 0 (0.22) (2.06) 2205 (1.18) 2205 (3.45) 72 2 693 (0.20) 95 (2.09) 176 (1.20) 964 (3.49) 0 (0.22) (1.82) 705 (1.05) 705 (3.08) a  In parentheses, total protein (mg) in each fraction.

The mouse Rt6 and rat RT6 families of proteins share both structural and functional properties (13-17, 21, 36). The deduced amino acid sequences of all four proteins contain signal sequences at their amino and carboxyl termini believed to be involved, respectively, in export to the endoplasmic reticulum and attachment of a GPI anchor (13, 14, 36). The internal regions of the coding domain contain the consensus sequences found in ADP-ribosyltransferases and NAD glycohydrolases, which are believed to be involved in the formation of the catalytic site (37, 38). Although the RT6 proteins, other mammalian and avian ADP-ribosyltransferases, and bacterial toxin ADP-ribosyltransferases contain the consensus sequences, their enzymatic activities appear clearly to differ (25, 31, 37-41). The rat RT6.1 and RT6.2 appear to be NAD glycohydrolases (15, 17); RT6.2 catalyzes auto-ADP-ribosylation, but there is some disagreement as to whether RT6.1 possesses this activity (17, 42). Unlike the avian erythrocyte and rabbit muscle ADP-ribosyltransferases (28, 31, 43), rat RT6.2 did not appear to use agmatine or similar, simple guanidino compounds as acceptors (15). A recent report on mouse Rt6.2 demonstrated that it had NAD glycohydrolase activity that was inhibited by agmatine; it was not established whether agmatine was used as an ADP-ribose acceptor (21). mRt6.1 protein was observed to have auto-ADP-ribosyltransferase activity (21); based on chemical stability of the ADP-ribose-protein bond, it was concluded that arginine was the ADP-ribose acceptor (21). It is not clear whether the auto-ADP-ribosylation reaction represents a single or multiple turnover reaction. Rat RT6.2, on the other hand, is an auto-ADP-ribosyltransferase but does not catalyze the ADP-ribosylation of arginine (15). The current report clearly demonstrates that mRt6.1 differs significantly in enzymatic properties from rat RT6.1 (17) and RT6.2 (15). Unlike the rat proteins, mRt6.1 catalyzes the stereospecific formation of ADP-ribosylagmatine, similar to cholera toxin and eukaryotic NAD:arginine ADP-ribosyltransferases (38, 44, 45). In the presence of agmatine, nicotinamide release parallels ADP-ribosylagmatine formation. As shown by the NMU cell transfection experiments, the putative carboxyl-terminal signal sequence for mRt6 is processed appropriately in mammalian cells leading to the addition of a GPI anchor. Thus, the mRt6.1 could serve a regulatory role in the ADP-ribosylation of cell surface proteins, suggesting a possible mechanism by which mRt6 protein on the surface of Rt6⁺ cells might regulate immune function.