Physiological relevance of the endogenous mono(ADP-ribosyl)ation of cellular proteins


D. Corda or M. Di Girolamo, Consorzio Mario Negri Sud, Department of Cell Biology and Oncology, 66030 Santa Maria Imbaro (Chieti), Italy
Fax: +39 0872 570 412
Tel: +39 0872 570 338


The mono(ADP-ribosyl)ation reaction is a post-translational modification that is catalysed by both bacterial toxins and eukaryotic enzymes, and that results in the transfer of ADP-ribose from βNAD+ to various acceptor proteins. In mammals, both intracellular and extracellular reactions have been described; the latter are due to glycosylphosphatidylinositol-anchored or secreted enzymes that are able to modify their targets, which include the purinergic receptor P2X7, the defensins and the integrins. Intracellular mono(ADP-ribosyl)ation modifies proteins that have roles in cell signalling and metabolism, such as the chaperone GRP78/BiP, the β-subunit of heterotrimeric G-proteins and glutamate dehydrogenase. The molecular identification of the intracellular enzymes, however, is still missing. A better molecular understanding of this reaction will help in the full definition of its role in cell physiology and pathology.




ADP-ribosylating turn-turn


dinitrogenase reductase activating glycohydrolase


dinitrogenase reductase ADP-ribosyltransferase


basic fibroblast growth factor


GTP binding protein




insulin-dependent diabetes mellitus




NAD glycohydrolase


poly(ADP-ribose) polymerases


platelet-derived growth factor

Enzyme-modulated mono(ADP-ribosyl)ation was originally identified as the mechanism of action of several of the bacterial toxins [1]. The diphtheria, cholera, pertussis and clostridia toxins are mono(ADP-ribosyl)transferases (ARTs; EC, and they are known to cause various pathologies after their translocation into mammalian host cells. Once inside the cell, they act by modifying specific host cell proteins, such as elongation factor 2, the α-subunit of the heterotrimeric GTP-binding (G) proteins, the small GTPases Rho and Rac, and monomeric actin ([2–4], and references therein).

More recently, a series of enzymes that are related to these toxins have been identified in cells, and their potential physiological roles have been explored ([5,6] and references therein; Table 1). The best known of these ARTs are ectoenzymes that are either glycosylphosphatidylinositol (GPI)-anchored or secretory. Both the toxins and these toxin-related eukaryotic ARTs function through the transfer of an ADP-ribose residue from βNAD+ to a specific amino acid of the acceptor protein, with the creation of an N- or S-glycosidic linkage and the release of nicotinamide. The free amino acid arginine (the most frequently modified residue of the protein substrates) and its analogue agmatine have been widely used as substrates to characterize the enzymatic activities of these mono(ADP-ribosyl)transferases [7]. The mono(ADP-ribosyl)ation reaction is distinct from that catalysed by the poly(ADP-ribose) polymerases (PARPs; EC, which instead transfer branched polymers of ADP-ribose to their target proteins, via an O-glycosidic bond (reviewed in [7a]).

Table 1.  Mammalian ARTs. See text for details and relevant references.
EnzymesSourceSubstrateEffect of the reaction
 ART1Human, rat, mouseIntegrin, defensin, FGF-2, PDGFBB/ArgInhibits substrate activity
 ART2Rat (a,b), mouse (1,2)P2X7, LFA1/ArgRole in T-cell poliferation, apoptosis
 ART3Human, rat, mouseUnknownUnknown
 ART4Human, rat, mouseUnknownUnknown
 ART5Human, ratUnknown/ArgUnknown
 Sirtuin2HumanAlbumin/acetyl-lysineInvolved in histone deacetylation
 Arginine-specificHamster, humanGβ/Arg129Inhibits substrate activity
 Cysteine-specificHumanGDH/Cys119Inhibits substrate activity

In mammalian cells, the mono(ADP-ribosyl)ation reaction is also regulated by enzymes that are able to reverse these post-translational modifications: the cytosolic ADP-ribosyl hydrolases and the cytosolic and extracellular pyrophosphatases [8,9]. With the former, the protein-ADP-ribose linkage is hydrolysed to release the ADP-ribose moiety, while with the pyrophosphatases, it is the pyrophosphate linkage that is hydrolysed, to release AMP and thus to leave a ribosylphosphate attached to the protein.

The mammalian ecto-ADP-ribosyltransferases (ARTs)

The mammalian ARTs are coded for by a family of structurally and functionally related genes. To date, five mammalian enzymes (ART1–5) have been cloned, although only four of these are expressed in humans, due to a defective art2 gene that has a stop signal in the coding region. Conversely, there are six ARTs expressed in mouse, due to the duplication of the art2 gene ([5,6,10], and references therein).

This enzyme family shares very limited amino acid sequence identity, with 20–30% seen among the ART paralogue members within any species; the exception here is mouse ART2.1 and ART2.2, where their sequence identity (85%) indicates the recent evolutionary duplication of the mouse art2 gene [11]. The ART2 enzymes have also been cloned from rat, and in this case the two isoforms are known as ART2.a and ART2.b and they are coded for by two alleles of a single-copy gene [12]. These allelic differences between the rat art2a and art2b genes result in a sequence variation of only 10 amino acids between the ART2.a and ART2.b proteins, although this alters their enzymatic properties: while both can catalyze the hydrolysis of NAD to ADP-ribose and nicotinamide, only ART2.b is capable of auto-ADP-ribosylation [13]. As the human and mouse genome sequences have been completely determined, all of the recognizable members of these toxin-related ARTs have now been identified for these two species, with the identity among orthologues ranging from 75% to 85%. As an example, the deduced amino-acid sequence of mouse ART1, which was the first cloned and characterized mammalian arginine-specific ART, is 77% and 73% identical to human and rabbit ART1, respectively [11,14].

Despite the low similarity at the level of their amino-acid sequences, there are common structural features that characterize this family of mammalian ARTs [15,16]. The catalytic domain of these enzymes is completely coded for by a single exon in all of the ARTs, and it contains a conserved glutamate residue that has been demonstrated to be crucial for the catalytic activity of the bacterial toxins and of ART1 and ART2 (mouse and rat) by site-specific mutagenesis. In ART1 from rabbit, even the conservative glutamate 240 to aspartate (E240D) substitution abolishes the transfer of ADP-ribose to the arginine used as an acceptor; the neighbouring E238 has also been shown to be important for ADP-ribose transfer [17,18]. In several of the ARTs, the replacement of this second glutamate abolishes the ability of these transferases to use arginine as acceptors, thus further supporting the hypothesis that this region is involved in substrate recognition [19]. According to the structural model proposed by Rappuoli and colleagues [16], this catalytic domain is composed of 70–100 amino acids and consists of three regions. Region 1, which is near the N-terminal portion of the protein and is characterized either by a conserved histidine (as in diphtheria toxin, ART3 and the PARPs) or by a conserved arginine (as in pertussis toxin, cholera toxin, the heat-labile enterotoxins and the other ARTs); region 2, which is characterized either by hydrophobic amino acids that are involved in NAD+ binding or by the serine-x-serine motif (where x represents threonine, serine or alanine); and region 3, which is highly acidic and is characterized by the conserved glutamate residue. The arginine-serine-glutamate-x-glutamate motif (R-S-EQE; which spans regions 1–2–3, respectively) is present in cholera toxin and in ART1, 2 and 5, and it is typical of the arginine-specific ARTs. This motif is missing in ART3 and ART4 [20,21]. Through a comparative analysis of crystallographic structures, Han and Tainer [19] have more recently extended the significance of the region 3 sequences by identifying an ADP-ribosylating turn-turn (ARTT) motif that they have implicated in substrate recognition. Consistent with the relevance of the ARTT motif, it has been shown that the auto-ADP-ribosylation of ART2.b is abolished by mutations of its R204, which is part of the ARTT motif. Similarly, if the Y204 of ART2.a is mutated to an arginine (Y204R), it is possible to promote ART2.a auto-ADP-ribosylation [22].

The other common structural features of the ARTs include an α-helix-rich N-terminal region, which represents a signal sequence for extracellular proteins, and a C-terminal region folded into β-sheets, which is characteristic of GPI-anchored membrane proteins [23]. As mammalian (human and mouse) ART5 does not contain this hydrophobic C-terminal signal sequence, it is a secreted protein [24]. Finally, there are four cysteines involved in disulfide bridge formation that are conserved among all of the ART isoforms. Thus, according to the rat ART2.2 crystal structure, C21 (C53 in mART1) forms a disulfide bond with another conserved cysteine at 223 (C272 in mART1) at the C-terminus of the molecule, which stabilizes the folding of the α-helix-rich domain [11]. C121 and C173 form the second disulfide bond, which is also located at the protein surface and is also important for protein stabilization [25].

The possibility that the N- and C-terminal domains of the ARTs are involved in the regulation of ART activity has been recently investigated by measuring both the transferase and NAD glycohydrolase (NADase) activities of truncated mutants of ART1 [26]. In mouse ART1, the amino acids at 24–38 (an ART1-specific extension) modulate both the transferase and NADase activities, and amino acids 39–45 (a common ART coil) are essential for both activities. The removal of the C-terminal basic domain decreases the transferase, but enhances the NADase activity. The N- and C-terminal regions of ART1 are therefore required for its transferase activity, while the enhanced NADase activity of the shorter mutants indicates that there are sequences outside of the catalytic site that exert structural constraints, and that modulate the substrate specificity and catalytic activity [26].

The ecto-ARTs: expression and function

ART1 is predominantly expressed in skeletal muscle, heart and lung, and in neutrophils and T-cell lymphoma cells [27,28]. Its substrates include integrin α7 in mouse skeletal muscle cells, where ADP-ribosylation has been proposed to have a role in myogenesis, as an increase in arginine-specific ADP-ribosylation has been observed during their differentiation into myotubes [29].

Of note, the human defensin HNP-1 is among the most recently identified substrates of ART1 [30]. The defensins are 2–6 kDa cationic peptides that are considered to be the major components of innate antimicrobial immunity, and that are thought to act by disrupting the microbial membrane [31]. They can also be considered to be components of adaptive immunity, because cytokine stimulation of human natural killer cells and T- and B-lymphocytes leads to the production of the defensins. Interestingly, α-defensins 1–3 are secreted by CD8 T-cells from immunologically stable HIV-1-infected individuals (long-term nonprogressors) and they are able to suppress HIV-1 replication [32].

Thus ART1 has been shown to modify HNP-1 on R14 in an in vitro assay [30]. This ADP-ribosylated HNP-1 loses its antimicrobial and cytotoxic activity, although it significantly increases the release of IL-8 from A549 cells, as compared to unmodified HNP-1. Conversely, the two peptides (unmodified and ADP-ribosylated) have similar chemotactic activities when evaluated for their ability to recruit T-lymphocytes [30]. These data are consistent with the concept that, once modified, HNP-1 acquires specific biological activities that can result in the recruitment of neutrophils (by the release of IL-8 from epithelial cells) and in the modulation of its own antimicrobial and cytotoxic activities. These latter aspects are particularly relevant, as this study also identified ADP-ribosylated HNP-1 in the bronchoalveolar lavage fluid from smokers (but not from nonsmokers); this would indicate that ADP-ribosylated HNP-1 is produced during the inflammatory response (and loses its antimicrobial activity). The relevance of these data also resides in the fact that this was the first demonstration of endogenous ADP-ribosylation in humans [30,33].

Additional substrates of ART1 have been identified in various different cell lines overexpressing this ectoenzyme, and these include growth factors and membrane receptors. ART1-transfected rat adenocarcinoma (NMU) cells were used to demonstrate ADP-ribosylation of basic fibroblast growth factor (FGF-2), which had been detected initially on the surface of adult bovine aortic endothelial and human hepatoma cells [34,35]. As FGF-2 has a high affinity for heparin, it is localized and possibly sequestered by the heparin sulfates on the cell surface and in the extracellular matrix. Heparin also inhibits the ADP-ribosylation reaction, which would imply that the heparin binding of FGF-2 and its ADP-ribosylation are mutually exclusive. Furthermore, the ADP-ribosylated site of FGF-2 is in its receptor-binding domain, and so it is possible that ADP-ribosylation modulates the binding of FGF-2 to its receptor and to heparin, thus regulating its availability to the cell [34,35].

In ART1-transfected V79 Chinese hamster lung fibroblasts, platelet-derived growth factor-BB (PDGF-BB) is the best substrate for ART1, whereas its structural homologue PDGF-AA is not a substrate [36]. ADP-ribosylated PDGF-BB loses its ability to stimulate mitogenic and chemotactic responses in human pulmonary smooth muscle cells, and it shows a reduced capacity for binding to PDGF receptors in competition-binding experiments, as compared to unmodified PDGF-BB [36]. This indicates that PDGF-BB-dependent signalling can be regulated by ART1 activity at the cell surface.

When the EL-4 mouse T-cell lymphoma cell line was stably transfected with ART1, the T-cell receptor signalling was inhibited in the presence of NAD via the ADP-ribosylation of integrin LFA-1 and other coreceptor proteins [37,38]. These effects have been proposed to result from a failure of the T-cell receptors and coreceptors to associate into a functional receptor cluster. Thus these T-cell responses would be modulated by mono(ADP-ribosyl)ation of cell surface proteins [37,38].

In general, the role of the ARTs in T-cell signalling is not clear, although the expression of ART2 on T-cells has been well characterized. Thus ART2 is known to be expressed in resting T-cells and in natural killer cells, and it appears to be specific to the immune system. This presence of ART2 on the surface of immune cells would thus suggest an immunomodulatory activity, and indeed, a significant disposition to develop autoimmune diabetes has been shown to depend on the absence of ART2 expression on rat T-cells [39–42].

What can perhaps be defined as the most intriguing function of ART2 was recently uncovered by Koch-Nolte and coworkers, namely that ADP-ribosylation activates the P2X7 purinoceptor ([43], Fig. 1A). P2X7 is a member of the P2X family of ATP-gated ion channels, and it is widely expressed on several types of blood cells [44]. This specific purinoceptor has attracted interest because of its particular ability to induce the formation of large membrane pores. Thus the activation of P2X7 with millimolar concentrations of ATP triggers calcium fluxes, phosphatidylserine exposure and apoptosis [44]. These same effects are triggered by NAD at micromolar concentrations via the ADP-ribosylation of P2X7. However, these effects are not seen in ART2-deficient T-cells, demonstrating that the activation of P2X7 by NAD is ART2-dependent [43]. These data provide an explanation for the previous demonstrations that extracellular NAD induces rapid apoptosis in naive T-cells by a mechanism involving ADP-ribosylation of cell surface molecules [45]. Altogether, these data show that not only are ART1 and ART2 expressed in cells of the immune system, but also that these two arginine-specific ARTs have a clear role in the regulation of the immune response. However, it is somewhat disappointing that one of the most interesting physiological roles has been defined for an ART that is not expressed in human cells. Thus, it is important to understand what the human counterpart of mouse ART2 might be, and whether these mouse T-cell effects can be extended to human cells.

Figure 1.

(A) Schematic representation of the mammalian mono(ADP-ribosyl)ation reactions. The figure shows both the extracellular mono(ADP-ribosyl)ation that is catalysed by the ARTs and the intracellular reaction that is catalysed by the yet undefined mono(ADP-ribosyl)transferases. The upper section (extracellular space, OUT) shows the ART2-dependent ADP-ribosylation of the P2X7 purinergic receptor. The ADP-ribosylated receptor is activated and leads to T-cell apoptosis. The lower section (intracellular space, IN) shows the ADP-ribosylation/deribosylation cycle of the heterotrimeric G-protein β subunit that is catalysed by a membrane-associated, intracellular ADP-ribosyltransferase (iART) and by a cytosolic ADP-ribosylhydrolase (ARH). The dashed arrow indicates possible hormonal regulation of this iART. The effectors that are uncoupled from the βγ dimer by ADP-ribosylation are indicated by the red line, while the red arrow indicates coupling (see text for details). (B) Schematic representation of the product of a mono(ADP-ribosyl)ation reaction. The N-glycosidic linkage between the ADP-ribose residue and Arg129 on the heterotrimeric G-protein β-subunit is illustrated.

The biological functions of ART3, ART4 and ART5 remain poorly defined [20,24]. ART3 and ART5 are strongly expressed in human testis, whereas ART4 is preferentially expressed in human lymphatic tissue. In human monocytes, the cell-surface ADP-ribosylated proteins are modified on their cysteine residues, suggesting that ART3 and ART4 are cysteine-specific ARTs [21]. This is consistent with the observation that in in vitro assays neither of these two ARTs displays arginine-specific enzymatic activity when expressed in and purified from Sf9 insect cells. In the same study, human ART5 was seen to be an arginine-specific ART, unlike mouse ART5, which shows a potent NADase activity [5].

A point that still needs to be clarified is the occurrence of the extracellular NAD+ that is required to sustain the ADP-ribosylation reaction. The steady-state concentration of NAD+ in the serum of healthy individuals is around 0.1 µm, and it can be kept low by the extracellular NAD-glycohydrolase CD38 (both soluble and membrane-associated) [46]; thus, to be utilized by ecto-ARTs, extracellular NAD+ should reach the concentration of 1–10 µm that is required for ADP-ribosylation of P2X7[43], or higher if the Km of the ARTs (from in vitro assays) is considered [47]. The probable mechanism is that NAD+ is released from cells, where its concentration is in the range of 0.5–1.0 mm, as a consequence either of cell lysis during inflammatory immune reactions and apoptosis, or of nonlytic release, for example through the connexin 43 channels [48].

Intracellular mono(ADP-ribosyl)tranferases and endogenous substrates

Although the ARTs that are able to modify extracellular proteins are the only well characterized family, mono(ADP-ribosyl)ation has also been demonstrated for intracellular proteins involved in cell signalling and metabolism (Table 1; [6] and references therein). The enzymatic activities involved here have been shown to be both cytosolic and membrane associated, although there is very little further information available concerning their identities. The first example of a well defined intracellular ADP-ribosylation cycle was reported in prokaryotes. An intracellular ART activity (dinitrogenase reductase ADP-ribosyltransferase; DRAT) was characterized in the photosynthetic bacterium Rodospirillum rubrum[49], where it regulates nitrogen fixation through mono(ADP-ribosyl)ation on R101 of the dinitrogenase reductase [50]. This reaction is reversible and the dinitrogenase reductase is fully reactivated by an ADP-ribosylarginine-hydrolase known as dinitrogenase reductase activating glycohydrolase (DRAG). Surprisingly, there is no significant amino acid sequence similarity between DRAT and the bacterial toxins that have ADP-ribosyltransferase activity; only a few key residues are conserved across the two families [23].

The same scenario could occur for the two families of mammalian ADP-ribosyltransferases: the ecto-ARTs and the endo-ARTs. These endo-ARTs appear to be part of a completely different family of proteins that shows no structural relationship to the ecto-ARTs described above. An example consistent with this is seen in the sirtuin family. It has recently been shown that yeast silent information regulator 2 protein (Sir2p, a NAD+-dependent histone/protein deacetylase) has ADP-ribosyltransferase activity, and while it deacetylates histones, it also catalyses the mono(ADP-ribosyl)ation of the removed acetyl group [51,52]. Clearly, this is not a ‘classical’ reaction that involves the modification of a target protein, but it involves small molecule substrates. It is therefore similar to that seen for a bacterial ADP-ribosyltransferase that is able to ADP-ribosylate and inactivate the antibiotic rifampicin [53], and a yeast enzyme that is able to transfer ADP-ribose from NAD+ to a phosphate group in tRNA [54]. To date, seven human homologues of Sir2p have been described (sirtuins 1–7), and they are characterized by ART activity [51,52], although they share no obvious sequence homologies with the ARTs themselves. Obviously, the sirtuins could represent the prototypes of a novel intracellular ART family. The alternative possibility that the ectocellular ARTs can modify intracellular substrates can also be considered. In this situation, either there needs to be a search for new isoforms that do not contain the signal peptide, or it needs to be shown that one or more of the ecto-ARTs can be shed from the membrane and can translocate into the cytoplasm. This could be achieved in a way similar to that of the bacterial toxins, which have their own specific receptors on the plasma membrane [55]. We are now actively working to identify and define potential new intracellular ART isoforms.

The intracellular mono(ADP-ribosyl)ation reactions have been associated with cell signalling and metabolism in intact cells. They modify three substrate proteins: the endoplasmic reticulum-resident chaperone GRP78/BiP, the β-subunit of heterotrimeric G-proteins, and the mitochondrial glutamate dehydrogenase GDH.

The mono(ADP-ribosyl)ation of GRP78/BiP leads to its inactivation. The modified GRP78/BiP has been detected in response to conditions that deplete the endoplasmic reticulum of processible proteins or that result in nutritional stress (such as lowered temperature, amino acid and glucose starvation), and has been related to the rate of protein synthesis and processing in intact Swiss 3T3 and GH3 pituitary cells [56–58]. According to the model proposed by Laitusis and colleagues [56], in cells with high rates of protein synthesis, unmodified GRP78/BiP is complexed with protein folding intermediates; a slowing of protein synthesis results in accumulation of the free, active form of GRP78/BiP, which is subjected to subsequent inactivation by ADP-ribosylation. The ADP-ribosylated form of the chaperone thus provides a buffering system that allows the rates of protein processing to be balanced with those of protein synthesis. It should be noted that while this mono(ADP-ribosyl)ation occurs intracellularly, from a topological point of view the catalytic domain of the enzyme involved (that has not yet been characterized) needs to be located in the lumen of the endoplasmic reticulum to modify its substrate, GRP78/BiP. Thus, this intracellular reaction occurs out of the cytosolic compartment.

Direct evidence of functional, intracellular mono(ADP-ribosyl)ation has been reported for the G-protein β-subunit ([59], Fig. 1). This reaction modifies R129 of the β-subunit (Fig. 1B) and is catalysed by a plasma-membrane-associated, but not GPI-anchored, intracellular ART that has not yet been molecularly characterized. The mono(ADP-ribosyl)ated β-subunit becomes the substrate of a cytosolic, ADP-ribosylhydrolase ([59], Fig. 1), which completes a cellular ADP-ribosylation/de-ADP-ribosylation cycle that controls the activation/inactivation of the βγ-dimer. Importantly, β-subunit mono(ADP-ribosyl)ation has also been detected in intact cells, under both resting [59] and stimulated conditions, thus indicating the physiological potential of this reaction. In intact cells under resting conditions, approximately 0.2% of the total βγ-heterodimer is modified; this could correspond to a cellular pool of free βγ-heterodimer that remains inactive. This hypothesis is supported by the demonstration that the β-subunit is modified only as a free heterodimer, and that mono(ADP-ribosyl)ation inactivates the β-subunit by impairing its interactions with its effector enzymes. This has been shown directly in the case of type 1 adenylyl cyclase, phosphoinositide 3-kinase and phospholipase C [59,60]. Thus, the ADP-ribosylation/deribosylation cycle modulates the function of the β-subunit. It is of particular interest here that the ADP-ribosylation of the β-subunit has also been shown to be under hormonal control: it can be increased upon activation of specific G-protein-coupled receptors (e.g. thrombin, serotonin and cholecystokinin receptors), indicating that the active βγ-heterodimer released from different classes of G-proteins can be a substrate for the endogenous mono(ADP-ribosyl)transferase [60]. Thus, while activation of these receptors will lead to the activation and dissociation of the G-protein α- and the βγ-subunits, this can initiate a parallel inactivation of βγ-subunit function that would potentially regulate the duration of βγ and α signalling, through the selective termination of the βγ function.

An ADP-ribosylation/deribosylation cycle has also been proposed to occur in mitochondria, and involves the cysteine-specific ADP-ribosylation of mitochondrial GDH in intact Hep-G2 cells [61]. The modified cysteine has been recently identified as Cys119 [61a]. As for GRP78/BiP and for the G-protein β-subunit, however, the nature of this ADP-ribosyltransferase activity remains uncharacterized. This cycle appears to be completed by an ADP-ribosylcysteine hydrolase that is also present in mitochondria.

Further substrates of mono(ADP-ribosyl)ation have also been identified, including the membrane-fissioning protein CtBP3/BARS [62,63], and the cytoskeletal proteins actin, tubulin and desmin [64–66]. However, to date there has been no direct evidence for their in vivo modification. Overall, a better understanding of the various and diverse biological roles of ADP-ribosylation of cellular proteins and peptides will be essential to fully define the role of this modification in normal and disease states.

The ADP-ribosylation reaction as a potential new drug target

A lack of ART2 expression has been correlated with an enhanced sensitivity to autoimmune disease in several animal models [40,67]. For example, in diabetes-prone BioBreeding (DP-BB) rats, a model for autoimmune insulin-dependent diabetes mellitus (IDDM), a defective expression of ART2 in their T-cells is associated with an increased susceptibility to the disease [39–42]. Conversely, the prevention of IDDM has been described in the same DP-BB rats following a transfusion with ART2-positive T-cells. The development of IDDM has also been observed in diabetes-resistant BioBreeding (DR-BB) rats when they are treated with a monoclonal antibody against ART2 [42]. Thus, ART2 expression appears to confer protection to IDDM in this animal model of the disease [42].

In the ART2.2 natural knock-out NZW mouse, the development of a lupus-like glomerulonephritis has been shown, again supporting the hypothesis that ART2-positive T-cells confer protection against autoimmune disease [68]. Polymorphisms have been reported for ART2.1 in the C57B1/6 mouse, where a stop codon at position 481 leads to an ART enzyme that lacks the GPI-anchor site and that has a reduced transferase activity [69]. However, it is important to stress that in these mouse models, disease development is under the control of several genetic factors, and thus a reduction or absence of ART2 expression is necessary but not sufficient for the onset of autoimmune pathologies. In line with this, both the NZW mouse and other mouse models that are natural knock-outs for ART2.1 (e.g. C57B1/6, BXSB), and the experimentally induced ART2 double knock-out do not show any evident immunological defects [70].

All of the data discussed above are consistent with the hypothesis that NAD-induced cell death via the activation of the P2X7 receptor has a role in immune responses. However, as ART2 knock-out mice show normal numbers and a normal distribution of T-cells [70], this NAD-dependent cell death cannot be crucial in the generation and maintenance of the T-cell. Rather, it is possible that ART2-induced T-cell death has a role during mechanical tissue injury or microbial inflammatory processes with severe cytolysis. Under these circumstances, the massive release of intracellular antigens is combined with high local concentrations of inflammatory cytokines, raising the danger of activation of autoreactive T-cells. Thus ART2-induced T-cell death could provide a safeguard mechanism against the undesirable activation of irrelevant and potentially autoreactive T-cells during an inflammatory response [43]. When extended to the identification of the counterpart in human cells, these findings open the exciting prospect of using NAD and its metabolites to modify the function of the P2X7 receptor and other purinoceptors [43].

Other mechanisms that can benefit from ADP-ribosylation-related drugs have emerged from a number of recent reports. ADP-ribosylation has been coupled to intracellular events that are associated with smooth muscle cell vasoreactivity, cytoskeletal integrity and free radical damage [71,72]. Additionally, there is evidence that ADP-ribosylation is required for smooth muscle cell proliferation [71,72]. Recent data have provided a direct link between mono(ADP-ribosyl)ation and smooth muscle cell proliferation and migration: meta-iodobenzylguanidine (MIBG), a selective inhibitor of arginine-dependent mono(ADP-ribosyl)ation, blocks the stimulation of DNA and RNA synthesis, prevents smooth muscle cell migration, and suppresses the induction of c-fos and c-myc gene expression. MIBG promotes the phosphorylation of the Rho effector PRK1/2, suggesting that mono(ADP-ribosyl)ation participates in a Rho-dependent signalling pathway that is required for immediate early gene expression. Furthermore, expression of the c-fos gene is the earliest proliferative event that has shown sensitivity to MIBG treatment, and it represents a novel mechanism by which mono(ADP-ribosyl)ation can influence cellular processes [71,72].

As the heterotrimeric G-proteins have key roles in cell regulation and the βγ complex is essential in a wide range of G-protein functions, including apoptosis, chemotaxis, secretion and cell proliferation and differentiation, we believe that our finding of the mono(ADP-ribosyl)ation of the endogenous β-subunit identifies a potential target for drug development. Indeed, recent data have shown that cellular invasion induced by src, met and leptin can be abrogated by constitutively activated forms of the Gαo/i subunits, and can be induced by the coexpression of Gβ1γ2 [73]. Moreover, depletion of free Gβγ heterodimers by the C-terminus of the β adrenergic receptor kinase (ct-βARK) results in a remarkable decrease in cellular adhesion and spreading on a collagen matrix [74]. Thus Gβγ dimers can be seen to be positive effectors of invasion pathways that are induced by oncogenes and epigenetic factors.

In line with the proposal that the ADP-ribosylated defensins represent tools for the treatment of pulmonary inflammation and other lung diseases [30,33], other ADP-ribosylated peptides that mimic the modified portions of the various ADP-ribosylation substrates and inhibitors of the ADP-ribosylation reaction itself show potential for the treatment of various pathologies, including autoimmune syndromes and proliferative diseases.


We wish to thank Dr C.P. Berrie for editorial assistance, Ms. E. Fontana for preparation of the Figures and the Italian Association for Cancer Research (AIRC, Milano, Italy), Telethon, Italy (project n. GGP030295) and the MIUR for financial support. N.D. is supported by a fellowship from AIRC.