Immunological and structural conservation of mammalian skeletal muscle glycosylphosphatidylinositol-linked ADP-ribosyltransferases

Increase in ADP-ribosyltransferase activity of rat T lymphocyte alloantigen RT6.1 by a single amino acid mutation

A family of glycosylphosphatidylinositol-linked ADP-ribosyltransferases, of which cDNAs were cloned from various mammalian cells, possess a common Glu-rich motif (EEEVLIP) near their carboxyl termini. Although the first Glu in the common motif is replaced by Gln (Q207EEVLIP) in rat T lymphocyte alloantigens RT6.1 and RT6.2, the two RT6s appear to have both activities of NAD+ glycohydrolase and ADP-ribosyltransferase to a lesser extent. To investigate the significance of the Glu-rich motif in the two enzyme activities, we produced a mutant RT6.1 (Q207E), in which Gln207 was replaced by Glu, together with wild-type RT6s, in Escherichia coli. Kinetic analysis revealed that there were no marked differences in the Vmax and Km values of NAD+ glycohydrolases among the three recombinant proteins. The recombinant RT6.1 and RT6.2 displayed extremely low auto-ADP-ribosylation, although the latter modification was somewhat higher than the former. In contrast, much greater auto-modification was observed for the Q207E mutant. Moreover, the mutant could effectively ADP-ribosylate agmatine as a substrate. Thus, the single amino acid mutation of RT6.1 caused a marked increase in its ADP-ribosyltransferase activity, indicating that the Glu-rich motif near the carboxy terminus plays an important role in the enzyme activity.

Molecular characterization of mouse T-cell ecto-ADP-ribosyltransferase Rt6: cloning of a second functional gene and identification of the Rt6 gene products

RT6 is an enzymatically active GPI-anchored membrane protein that was originally discovered in the rat as a peripheral T cell alloantigen. It has attracted interest as an activation antigen and because defective RT6-expression coincides with increased susceptibility for autoimmune type I diabetes in the BB rat. Southern blot analyses indicate that the rat carries a single copy RT6 gene whereas the mouse carries a duplication of the homologous locus. We had previously cloned and sequenced a RT6-homologous cDNA from BALB/c mouse spleen. We now report the cloning and characterization of a second RT6-homologue from BALB/c and 129/Sv mice. The two mouse Rt6 genes (designated Rt6-1 and Rt6-2) encode similar open reading frames that are disrupted by conserved introns. The nucleotide sequences of the Rt6-1 and Rt6-2 coding regions show 87% sequence identity, the deduced amino acid sequences 79% identity. The amino acid sequences reveal significant similarity to recently cloned ADP-ribosylating ectoenzymes from rabbit and human skeletal muscle as well as chicken bone marrow cells. RT-PCR analyses reveal that the two Rt6 genes are differentially expressed in distinct inbred mouse strains and that their transcripts are properly processed. Western blot analyses demonstrate that the respective gene products are released from cells by treatment with PI-PLC. The results further show that both mouse Rt6 genes are translated into GPI-anchored cell surface molecules and that Rt6 gene expression is restricted to peripheral lymphoid tissues.

Arginine-specific mono(ADP-ribosyl)transferase activity on the surface of human polymorphonuclear neutrophil leucocytes

An Arg-specific mono(ADP-ribosyl)transferase activity on the surface of human polymorphonuclear neutrophil leucocytes (PMNs) was confirmed by the use of diethylamino-(benzylidineamino)guanidine (DEA-BAG) as an ADP-ribose acceptor. Two separate HPLC systems were used to separate ADP-ribosyl-DEA-BAG from reaction mixtures, and its presence was confirmed by electrospray mass spectrometry. ADP-ribosyl-DEA-BAG was produced in the presence of PMNs, but not in their absence. Incubation of DEA-BAG with ADP-ribose (0.1-10 mM) did not yield ADP-ribosyl-DEA-BAG, which indicates that ADP-ribosyl-DEA-BAG formed in the presence of PMNs was not simply a product of a reaction between DEA-BAG and free ADP-ribose, due possibly to the hydrolysis of NAD+ by an NAD+ glycohydrolase. The assay of mono(ADP-ribosyl)transferase with agmatine as a substrate was modified for intact PMNs, and the activity was found to be approx. 50-fold lower than that in rabbit cardiac membranes. The Km of the enzyme for NAD+ was 100.1 30.4 microM and the Vmax 1.4 0.2 pmol of ADP-ribosylagmatine/h per 10(6) cells. The enzyme is likely to be linked to the cell surface via a glycosylphosphatidylinositol anchor, since incubation of intact PMNs with phosphoinositol-specific phospholipase C (PI-PLC) led to a 98% decrease in mono(ADP-ribosyl)transferase activity in the cells. Cell surface proteins were labelled after exposure of intact PMNs to [32P]NAD+. Their molecular masses were 79, 67, 46, 36 and 26 kDa. The time course for labelling was non-linear under these conditions over a period of 4 h. The labelled products were identified as mono(ADP-ribosyl)ated proteins by hydrolysis with snake venom phosphodiesterase to yield 5'-AMP.

Residues in Torpedo californica acetylcholinesterase necessary for processing to a glycosyl phosphatidylinositol-anchored form

Acetylcholinesterase from Torpedo californica (TcAChE) can be found as a glycosyl phosphatidylinositol (GPI)-anchored, membrane associated form. The C-terminal amino-acid sequence of the precursor protein resembles the signal peptide sequence found in proteins and enzymes destined for GPI-modification. Characteristics of such a signal peptide are a relatively polar stretch of amino acids, separating a cleavage- and modification-site (omega-site) residue from a hydrophobic C-terminus. We have introduced mutations, both at putative omega-sites and in the hydrophobic region, and analysed their effects on GPI-anchoring of TcAChE. Our results show that substitution of all three Ser residues in the region Ser542-Ser544 prevents GPI-modification and membrane anchoring. Individual substitution of each of these residues resulted in no or only a minor effect on the modification. We therefore conclude that more than one residue within this sequence can be utilised as the omega-site. Our analyses of double substitutions indicated that Ser543 and Ser544 are the preferred residues for GPI-modification. Moreover, the hydrophobic region is shown to be essential for GPI-anchoring of TcAChE.

Immunochemical detection of ADP-ribosylating enzymes in the archaeon Sulfolobus solfataricus

Polyclonal antibodies raised against eukaryotic mono-(ADPribose)transferase and poly(ADPribose)polymerase were used to test the presence of antigenic determinants in a crude extract of Sulfolobus solfataricus, a thermophilic archaeon. Samples from eukaryotic (bull testis) and bacterial (E. coli) sources were analysed for comparison. All tested antibodies reacted with the sulfolobal sample with a specificity comparable to that of the eukaryotic preparation, as revealed by ELISA test, activity assays in the presence of antibodies and immunoblot experiments. After electrophoresis and western blot of sulfolobal proteins, a band at a mass around 50 kDa was detected by immunostaining.

Mono-ADP-ribosylation: a reversible posttranslational modification of proteins

Mono-ADP-ribosyltransferase activity has been detected in numerous vertebrate tissues and transferase cDNAs from a few species have recently been cloned. In vitro ADP-ribosylation has been demonstrated with diverse substrates such as phosphorylase kinase, actin, and Gs alpha resulting in the alteration of substrate function. ADP-ribosylation of endogenous target proteins has been observed in chicken heterophils, rat brain, and human platelets, and integrin alpha 7 was found to be the endogenous substrate of the GPI-anchored rabbit skeletal muscle transferase. The reversibility of ADP-ribosylation is made possible by ADP-ribosylarginine hydrolases which have been isolated and cloned from rodent and human tissues. The transferases and hydrolases could in principle form an intracellular ADP-ribosylation regulatory cycle. In the case of the skeletal muscle transferases, however, processing of ADP-ribosylated integrin alpha 7 is carried out by phosphodiesterases and possibly phosphatases (Fig. 1). Most bacterial toxin and eukaryotic mono-ADP-ribosyltransferases, and perhaps other NAD-utilizing enzymes such as the RT6 family of proteins, share a common catalytic-site structure despite a lack of overall sequence identity. The transferases that have been studied thus far possess a critical glutamic acid and other amino acids at the catalytic cleft which function to position NAD for nucleophilic attack at the N-glycosidic linkage for either ADP-ribose transfer or NAD hydrolysis. The amino acid differences among transferases at the active site may reflect different catalytic mechanisms of ADP-ribosylation or may be required for accommodating the different ADP-ribose acceptor molecules.

Structure and function of eukaryotic mono-ADP-ribosyltransferases

ADP-ribosylation of proteins has been observed in numerous animal tissues including chicken heterophils, rat brain, human platelets, and mouse skeletal muscle. ADP-ribosylation in these tissues is thought to modulate critical cellular functions such as muscle cell development, actin polymerization, and cytotoxic T lymphocyte proliferation. Specific substrates of the ADP-ribosyltransferases have been identified; the skeletal muscle transferase ADP-ribosylates integrin alpha 7 whereas the chicken heterophil enzyme modifies the heterophil granule protein p33 and the CTL enzyme ADP-ribosylates the membrane-associated protein p40. Transferase sequence has been determined which should assist in elucidating the role of ADP-ribosylation in cells. There is sequence similarity among the vertebrate transferases and the rodent RT6 alloantigens. The RT6 family of proteins are NAD glycohydrolases that have been shown to possess auto-ADP-ribosyltransferase activity whereas the mouse Rt6-1 is also capable of ADP-ribosylating histone. Absence of RT6+ T cells has been associated with the development of an autoimmune-mediated diabetes in rodents. Humans have an RT6 pseudogene and do not express RT6 proteins. The reversal of ADP-ribosylation is catalyzed by ADP-ribosylarginine hydrolases, which have been purified and cloned from rodent and human tissues. In principle, the transferases and hydrolases could form an intracellular ADP-ribosylation regulatory cycle. In skeletal muscle and lymphocytes, however, the transferases and their substrates are extracellular membrane proteins whereas the hydrolases described thus far are cytoplasmic. In cultured mouse skeletal muscle cells, processing of the ADP-ribosylated integrin alpha 7 was carried out by phosphodiesterases and possibly phosphatases, leaving a residual ribose attached to the (arginine)protein. Several bacterial toxin and eukaryotic mono-ADP-ribosyltransferases, and perhaps other NAD-utilizing enzymes such as the RT6 alloantigens share regions of amino acid sequence similarity, which form, in part, the catalytic site. The catalytic cleft, found in the bacterial toxins that have been studied thus far, contains a critical glutamate and other amino acids that function to position NAD for nucleophilic attack at the N-glycosidic linkage, for either ADP-ribose transfer or NAD hydrolysis. Amino acid differences among the transferases at the active site may be required for accommodating the different ADP-ribose acceptor molecules.

Sequence of a chicken erythroblast mono(ADP-ribosyl)transferase-encoding gene and its upstream region

We have cloned the MADPRT gene encoding the 300-amino-acid mono(ADP-ribosyl)transferase (MADPRT) from chicken erythroblasts. The protein has homology to the rabbit and human skeletal muscle (50% identity) and two chicken heterophil (52% identity) NAD+:arginine MADPRT. The active site region is particularly conserved. The upstream region of the MADPRT gene from erythroblasts has several features characteristic of promoter sequences.

NAD(+)-dependent ADP-ribosylation of T lymphocyte alloantigen RT6.1 reversibly proceeding in intact rat lymphocytes

Rat T lymphocyte alloantigen 6.1 (RT6.1), which was synthesized as the fusion protein with a maltose-binding protein in Escherichia coli, displayed NAD(+)-dependent auto-ADP-ribosylation in addition to an enzyme activity of NAD+ glycohydrolase. Such ADP-ribosylation of RT6.1 was also observed in lymphocytes isolated from rat tissues as follows. When intact rat lymphocytes expressing RT6.1 mRNA were incubated with [alpha-32P]NAD+, its radioactivity was incorporated into a cell surface protein with the M(r) of 31,000. The radiolabeled 31-kDa protein was released from the cell surface by treatment of the cells with phosphatidylinositol-specific phospholipase C and immunoprecipitated with anti-RT6.1 antiserum. The radioactivity incorporated into the 31-kDa protein was recovered as 5'-[32P]AMP upon incubation with snake venom phosphodiesterase and also removed by NH2OH treatment. These results suggested that the NAD(+)-dependent modification of the 31-kDa protein was due to ADP-ribosylation of glycosylphosphatidylinositol-anchored RT6.1 at an arginine residue. When intact lymphocytes, in which RT6.1 had been once modified by [32P]ADP-ribosylation, were further incubated in the absence of NAD+, there was reduction of the radioactivity in the [32P]ADP-ribosylated RT6.1. The reduced radioactivity was recovered from the incubation medium as [32P]ADP-ribose. This reduction was effectively inhibited by the addition of ADP-ribose to the reaction mixture. Moreover, readdition of NAD+ caused the ADP-ribosylation of RT6.1 again. Thus, the ADP-ribosylation of RT6.1 appeared to proceed reversibly in intact rat lymphocytes.

Processing of ADP-ribosylated integrin alpha 7 in skeletal muscle myotubes

Integrin alpha 7 is a major substrate in skeletal muscle cells for the cell surface, glycosylphosphatidylinositol-anchored, arginine-specific ADP-ribosyltransferase. Since ADP-ribosylarginine hydrolase, the enzyme responsible for cleavage of the ADP-ribosylarginine bond and a component with the transferase of a putative ADP-ribosylation cycle, is cytosolic, the processing of ADP-ribosylated integrin alpha 7 was investigated. Following incubation of differentiated mouse C2C12 myoblasts with [adenylate-32P]NAD and analysis by SDS-polyacrylamide gel electrophoresis under reducing conditions, two [32P]ADP-ribosylated forms of integrin alpha 7 were resolved. By pulse-chase and purification of the radiolabeled proteins on a laminin affinity column, it was demonstrated that a 105-kDa ADP-ribosylated form originated from a mono-ADP-ribosylated 102-kDa form and represented integrin alpha 7 modified at more than one site. The additional site(s) of modification, utilized at higher NAD concentrations, were located in the 63-kDa N-terminal segment of integrin alpha 7. Both [32P]ADP-ribosylated integrins were loosely associated with the cytoskeleton, bound to laminin affinity columns, and immunoprecipitated with antibodies to integrin beta 1. 32P label was rapidly removed from [32P]ADP-ribosylated integrin alpha 7 at either site of modification, a process inhibited by free ADP-ribose or p-nitrophenylthymidine-5'-monophosphate, an alternative substrate of 5'-nucleotide phosphodiesterase. The processed integrin alpha 7 was unavailable for subsequent ADP-ribosylation, although the amount of surface integrin alpha 7 remained constant. During the processing, no loss of label was observed from integrin alpha 7 radiolabeled with [14C]NAD, containing 14C in the nicotinamide proximal ribose, consistent with degradation of the ADP-ribose moiety by a cell surface 5'-nucleotide phosphodiesterase. Thus, cell surface ADP-ribosylation, in contrast to intracellular ADP-ribosylation, is not readily reversed by ADP-ribosylarginine hydrolase and seems to operate outside the postulated ADP-ribosylation cycle.

Conservation of a common motif in enzymes catalyzing ADP-ribose transfer. Identification of domains in mammalian transferases

Bacterial toxin ADP-ribosyltransferases, e.g. diphtheria toxin (DT) and pertussis toxin, have in common consensus sequences involved in catalytic activity, which are localized to three regions. Region I is notable for a histidine or arginine; region II, approximately 50-75 amino acids downstream, is rich in aromatic/hydrophobic amino acids; and region III, further downstream, has a glutamate and other acidic amino acids. A similar motif was observed in the sequence of the glycosylphosphatidylinositol-linked muscle ADP-ribosyltransferase. Site-directed mutagenesis was performed to verify the role of this motif. Proteins were expressed in rat adenocarcinoma cells, released from the cell with phosphatidylinositol-specific phospholipase C, and quantified with polyclonal antibodies. Transferase His114 in region I aligned with His21 of DT; as with DT, the H114N mutant was active. Aromatic/hydrophobic amino acids (region II) were found approximately 30-50 amino acids downstream of this histidine. Although transferase has a Glu278-Tyr-Ile sequence characteristic of region III in DT, Glu278 was not critical for activity. In an alternative region III containing Glu238-Glu239-Glu240, Glu238 and Glu240 but not Glu239 were critical. Glu240 aligned with critical glutamates in DT, Pseudomonas exotoxin, and C3 transferase. Thus, the mammalian ADP-ribosyltransferases have motifs similar to toxin ADP-ribosyltransferases, suggesting that these sequences are important in ADP-ribose transfer reactions.

Vertebrate mono-ADP-ribosyltransferases

Mono-ADP-ribosylation appears to be a reversible modification of proteins, which occurs in many eukaryotic and prokaryotic organisms. Multiple forms of arginine-specific ADP-ribosyltransferases have been purified and characterized from avian erythrocytes, chicken polymorphonuclear leukocytes and mammalian skeletal muscle. The avian transferases have similar molecular weights of approximately 28 kDa, but differ in physical, regulatory and kinetic properties and subcellular localization. Recently, a 38-kDa rabbit skeletal muscle ADP-ribosyltransferase was purified and cloned. The deduced amino acid sequence contained hydrophobic amino and carboxy termini, consistent with known signal sequences of glycosylphosphatidylinositol (GPI)-anchored proteins. This arginine-specific transferase was present on the surface of mouse myotubes and of NMU cells transfected with the cDNA and was released with phosphatidylinositol-specific phospholipase C. Arginine-specific ADP-ribosyltransferases thus appear to exhibit considerable diversity in their structure, cellular localization, regulation and physiological role.