G-Protein-mediated Signalling Laboratory, Department of Cellular and Translational Pharmacology, Consorzio Mario Negri Sud, Santa Maria Imbaro, Chieti, Italy
M. Di Girolamo, Laboratory of G-Protein-mediated Signalling, Department of Cellular and Translational Pharmacology, Consorzio Mario Negri Sud, Via Nazionale 8/A, 66030 Santa Maria Imbaro, Chieti, Italy
During the development, progression and dissemination of neoplastic lesions, cancer cells can hijack normal pathways and mechanisms. This includes the control of the function of cellular proteins through reversible post-translational modifications, such as ADP-ribosylation, phosphorylation, and acetylation. In the case of mono-ADP-ribosylation and poly-ADP-ribosylation, the addition of one or several units of ADP-ribose to target proteins occurs via two families of enzymes that can generate ADP-ribosylated proteins: the diphtheria toxin-like ADP-ribosyltransferase (ARTD) family, comprising 17 different proteins that are either poly-ADP-ribosyltransferases or mono-ADP-ribosyltransferases or inactive enzymes; and the clostridial toxin-like ADP-ribosyltransferase family, with four human members, two of which are active mono-ADP-ribosyltransferases, and two of which are enzymatically inactive. In line with a central role for poly-ADP-ribose polymerase 1 in response to DNA damage, specific inhibitors of this enzyme have been developed as anticancer therapeutics and evaluated in several clinical trials. Recently, in combination with the discovery of a large number of enzymes that can catalyse mono-ADP-ribosylation, the role of this modification has been linked to human diseases, such as inflammation, diabetes, neurodegeneration, and cancer, thus revealing the need for the development of specific ARTD inhibitors. This will provide a better understanding of the roles of these enzymes in human physiology and pathology, so that they can be targeted in the future to generate new and efficacious drugs. This review summarizes our present knowledge of the ARTD enzymes that are involved in mono-ADP-ribosylation reactions and that have roles in cancer biology. In particular, the well-documented role of macro-containing ARTD8 in lymphoma and the putative role of ARTD15 in cancer are discussed.
During the development, progression and dissemination of neoplastic lesions, cancer cells can hijack normal intracellular pathways and mechanisms. These can include the pathways involved in intercellular communication, control of transcription, and control of protein localization, function and degradation by post-translational modifications, such as ADP-ribosylation, phosphorylation, and acetylation. ADP-ribosylation is a post-translational modification catalysed by enzymes found in organisms ranging from prokaryotes to mammals, and that transfers one or several units of ADP-ribose from βNAD+ to a specific amino acid of the target protein, thus profoundly altering its functional properties [1-4]. Mono-ADP-ribosylation was originally identified as the mechanism of action of several bacterial toxins. The diphtheria, cholera, pertussis and clostridia toxins are mono-ADP-ribosyltransferases, and, after their translocation into mammalian host cells, they act by modifying specific host cell proteins. These toxin targets include elongation factor 2, thus inhibiting protein translation, the α-subunits of the heterotrimeric G-proteins, the small GTPases Rho and Rac, and monomeric actin (see [5-9], and references therein). The first ADP-ribosylating enzyme described in mammals was poly-ADP-ribose polymerase (PARP)1, which can synthesise linear or branched polymers of ADP-ribose [poly-ADP-ribose (PAR)]. PAR is formed from several molecules of ADP-ribose that are linked together via a unique O-glycosidic ribose–ribose bond, and it is transferred onto various protein targets, mainly in the nucleus. PARP1 [also known as diphtheria toxin-like ADP-ribosyltransferase (ARTD)1] is the first and most extensively studied member of the PARP family [10, 11]. PARP1 is activated by DNA strand breaks, and its role in the cellular response to genotoxic and oxidative stress has been widely recognized and studied, with some PARP inhibitors being evaluated in several clinical trials as anticancer therapeutics [12-14].
In contrast, the biological function of mono-ADP-ribosylation has been less studied, and is consequently less understood. The mono-ADP-ribosylation reaction is different from the enzymatic transfer of PAR catalysed by PARP1, as it consists of the transfer of a single ADP-ribose moiety to specific amino acid residues of various cellular protein targets. Vertebrate mono-ADP-ribosyltransferase activity was first detected in turkey erythrocytes [15, 16], rat liver homogenates , and Xenopus tissues . Specific enzymes were then cloned from different sources [19-25], and this led to the characterization of the first family of mammalian mono-ADP-ribosyltransferases: the clostridial toxin-like ADP-ribosyltransferase (ARTC) family, which includes five mammalian enzymes, although there are only four in humans, as ARTC2 is a pseudogene  (Table 1). Two of these have been shown to have roles in immune regulation; for example, ARTC1 inhibits T-lymphocyte functions, including their cytolytic activity and proliferation, by ADP-ribosylating arginines of cell surface proteins, including the T-cell coreceptors [1, 27-30]. Although the ARTC family of enzymes has been well characterized, and many proteins that are modified by mono-ADP-ribosylation have been identified, their function, and consequently their therapeutic potential, remains largely unknown. A therapeutic perspective here has been opened by the development of recombinant antibodies that can block mammalian and toxin ADP-ribosyltransferases (see ).
In mammalian cells, both mono-ADP-ribosylation and poly-ADP-ribosylation reactions are also regulated by enzymes that can reverse these post-translational modifications, by cleaving the covalent bond to release the target protein for the next round of modifications. PAR glycohydrolase (PARG) can specifically hydrolyse the ribose–ribose bonds in PAR chains . ADP-ribosyl hydrolase (ARH)1 cleaves the mono-ADP-ribose–arginine bond . Two additional ARH proteins have been identified: ARH3 has been shown to have a PAR glycohydrolase activity [34, 35], whereas the activity of ARH2 remains to be defined. These enzymatic activities have also been linked to human diseases and cancer. Indeed, PARG silencing leads to inhibition of growth of human colon cancer cells via the phosphoinositide-3-kinase–Akt–nuclear factor-κB pathway , and leads to the death of cells deficient in the homologous recombination protein breast cancer type 2 (BRCA2). This suggests that PARG inhibitors could be used to specifically kill BRCA2-negative tumours and other homologous recombination-deficient tumours . Small molecules that effectively inhibit PARG in vitro and in cellular lysates have been recently reported . Moreover, ARH1 deficiency leads to tumour formation in mice . Thus, looking beyond ARTD1/PARP1 in exploring the therapeutic potential of the regulation of ADP-ribosylation reactions will generate a great amount of knowledge in the future, and may provide greater therapeutic possibilities for pathophysiological conditions such as inflammation and cancer.
ARTD family proteins are promising therapeutic targets in cancer
To date, 17 different ARTDs encoded by different genes have been identified in humans . All of these ARTDs are characterized by a PARP domain, and can be classified as poly-ADP-ribosylating, mono-ADP-ribosylating, or enzymatically inactive, on the basis of the presence of an intact or modified histidine–tyrosine–glutamate (H-Y-E) triad within the PARP domain (Table 1). Six of the ARTDs are true PARP enzymes (ARTD1/PARP1, ARTD2/PARP2, ARTD3/PARP3, ARTD4/PARP4, ARTD5/tankyrase1, and ARTD6/tankyrase2), as they can transfer multiple ADP-ribose residues, and even branched PAR, onto target proteins [41-44]. ARTD1/PARP1 and ARTD2/PARP2 are mainly localized in the nucleus, and are directly activated by DNA strand breaks; thus, they are involved in DNA damage repair, chromatin remodelling, and transcriptional regulation (for recent review, see ). Moreover, ARTD1/PARP1, ARTD3/PARP3 and ARTD5/tankyrase1 are localized on centrosomes, and several studies have indicated a role for poly-ADP-ribosylation in the regulation of centrosome behaviour, mitotic spindle integrity, and duration of mitosis [14, 45-48]. Thus, each of these ARTDs has a crucial role in mitosis and cancer, and they represent potential candidates for cancer therapy. Indeed, PARP1 has been at the front line of drug discovery since the 1980s, and the first clinical trial for a PARP inhibitor was initiated in 2003 with rucaparib (AG-014699) . Since then, eight further inhibitors have entered clinical trials: olaparib (AZD-2281), veliparib (ABT-888), iniparib (BSI 201), INO-1001, MK4827, CEP-9722, GPI21016, and BMN-763. The rationale underlying these studies is to enhance the DNA damage caused by chemotherapy and radiotherapy by inhibiting the mechanisms of repair of damaged DNA through ARTD1/PARP1 inhibition in cancer treatment [12, 50-53]. This therapy was specifically evaluated in breast cancer type 1 (BRCA1)-associated and BRCA2-associated cancer, as BRCA-deficient cells have defects in double-strand break repair, and are thus more dependent on ARTD1/PARP1 activity and base excision repair to maintain genomic integrity. The rationale was to target cells that are genetically predisposed to die by blocking ARTD1/PARP1-dependent base excision repair pathways. Despite some therapeutic benefits of PARP inhibitors in cotreatments with chemotherapy and radiotherapy, phase 3 trials for hereditary BRCA1-associated and BRCA2-associated breast cancer with olaparib were cancelled. A possible explanation for the limited results obtained with PARP inhibitors in BRCA1-deficient or BRCA2-deficient cancer is that resistance to PARP inhibitors and carboplatin might be attributable to the reactivation of the genes encoding these proteins by secondary mutations. PARP inhibitors have also been tested against triple-negative (oestrogen receptor-negative, progesterone receptor-negative, and ERBB2-negative) breast cancer and sporadic serous ovarian cancer, which have some of the properties of BRCA1-deficient and BRCA2-deficient cells. Iniparib failed to prolong survival in a phase 3 clinical trial on metastatic triple-negative breast cancer [54, 55]. However, recent studies have shown that iniparib is not a true PARP1 inhibitor . Thus, despite these setbacks, other PARP inhibitors are likely to be more successful. For this reason, their development on the basis of new chemical scaffolds is still underway [57-60]. Besides ARTD1/PARP1, tankyrases are also promising anticancer targets (see ).
The further 11 members of the ARTD family have lost the conserved glutamate that is crucial for polymer elongation (Glu988 in ARTD1), and thus they are either cellular mono-ADP-ribosyltransferases or inactive proteins (Table 1). Among these, ARTD10/PARP10 was initially isolated as a partner of the oncoprotein c-Myc, and its overexpression inhibited c-Myc-dependent transformation and regulated cell proliferation . Considering that c-Myc controls cell proliferation by modulating the expression of ribosomal proteins and of RNA components that are required for the export of mature ribosomal subunits from the nucleus to the cytoplasm, and that ARTD10/PARP10 contains an RNA recognition motif and a nuclear export sequence, it is highly possible that ARTD10/PARP10 has a role in c-Myc-mediated transport of rRNAs [61Moreover, mutation of the ARTD10/PARP10 nuclear export sequence blocks its export from the nucleus, and also blocks cell proliferation . Thus, in light of this evidence [64, 65], much effort needs to made to fully determine the role of ARTD10 in cell proliferation and cancer biology, particularly as this should open new therapeutic perspectives.
Conversely, the connection between cancer and ARTD8/PARP14 is well established. ARTD8/PARP14 was identified as a signal transducer and activator of transcription (Stat6)-interacting protein, and was shown to be associated with the aggressiveness of B-cell lymphomas in diffuse large B-cell lymphomas (DLBCLs) . ARTD8/PARP14 is also likely to have a mono-ADP-ribosyltransferase activity . More recently, we demonstrated that ARTD15/PARP16 is an active mono-ADP-ribosyltransferase that shows a unique intracellular localization: in the endoplasmic reticulum . We also showed that ARTD15/PARP16 interacts with and modifies importin β1 , a protein that is involved in microtubule regulation and mitosis . Thus, a number of novel intracellular mono-ADP-ribosyltransferases in the ARTD family can also have roles in cell transformation. However, their functions, and consequently their therapeutic potential, remain largely unknown. We review herein the experimental evidence that illustrates the roles that these enzymes can have in cell survival and cancer biology. In particular, the role of ARTD8 in lymphoma and the possible role of ARTD15 in inflammation and cancer will be discussed.
The role of macro-ARTDs in lymphoma
Within the PARP/ARTD family, there are three members that are characterized by their two or three N-terminal ‘macro domains’ in addition to the C-terminal region with sequence homology to the PARP catalytic domain. ARTD9 (BAL1/PARP9) and ARTD7 (BAL3/PARP15) have two macro domains, whereas ARTD8 (BAL2/PARP14) contains three of these modules [40, 69]. The macro domain is an evolutionarily conserved protein module of 130–190 amino acids that was originally identified in the histone variant macro-H2A (mH2A), which is associated with repression of transcription and X-chromosome inactivation [70, 71]. The macro domain of mH2A was shown to affect the interaction of the nucleosome-remodelling complex SWI/SNF with chromatin, thus impeding the binding of transcription factors, and hence repressing transcription . Later, it was demonstrated that the macro domain recognizes monomeric and/or polymeric forms of ADP-ribose , and we showed that this domain interacts not only with free ADP-ribose molecules, but also with both mono-ADP-ribosylated and poly-ADP-ribosylated proteins . The fact that these macro domains specifically recognize this post-translational modification on cellular proteins has led to the hypothesis that this module, which is often present in addition to other protein domains, is relevant in bringing together specific proteins into multiprotein functional complexes. This is the case for mH2A.1.1, which interacts specifically with ARTD1/PARP1 through its macro domain. This interaction leads to inhibition of ARTD1/PARP1 activity and gene silencing; thus, upon association with mH2A, ARTD1/PARP1 becomes part of a transcriptional repression complex [75, 76]. Appropriate conditions can reactivate transcription: heat shock was reported to affect the interaction between mH2A1.1 and ARTD1/PARP1, thus allowing ARTD1/PARP1 to ADP-ribosylate itself and nearby histones. The ADP-ribosylated histones then leave the Hsp70.1 promoter, thus allowing transcription . It is of interest that at least two macro-containing ARTDs have been shown to have roles in transcriptional repression in a cellular context . These proteins can sterically block the access of transcription factors and coactivators to specific chromatin regions, similarly to mH2A1.1.
The gene encoding ARTD9/BAL1 was originally identified in a genome-wide search for genes related to the risk for DLBCL, the most common non-Hodgkin lymphoma, and it was found to be the most abundant B-aggressive lymphoma (BAL) family member . When stably overexpressed in DLBCL cells, ARTD9 stimulates cell migration, thus suggesting that ARTD9 promotes malignant B-cell migration and dissemination in high-risk DLBCL . Then, in a pilot series of primary DLBCLs, ARTD9 expression was shown to be significantly higher in more aggressive and chemoresistant tumours than in low-risk tumours . Specifically, the higher expression of ARTD9 and of its interactor BAL1 binding partner (an E3 ubiquitin ligase) was shown in malignant B cells of lymphomas characterized by an active host inflammatory response. The inflammatory interferon (INF)-γ, which is known to be secreted by host-activated tumour-infiltrating T lymphocytes, can induce both ARTD9 and BAL1 binding partner expression in DLBCL cells. In turn, this induced ARTD9 upregulates the INF-γ-dependent genes . Thus, ARTD9 positively regulates tumour genes in an inflammatory environment, possibly by inhibiting the host immune response against the lymphoma.
ARTD9 and the other macro-containing PARPs have complex and not yet completely defined roles in the regulation of transcription; in addition to participating in transcriptional repression, both ARTD9 and ARTD8/BAL2 are transcriptional coactivators, although the relevance of their catalytic activity remains to be clarified. ARTD9 does not show ADP-ribosyltransferase activity, in contrast to ARTD8 and ARTD7/BAL3 . Thus, we can imagine that the ADP-ribosyltransferase activity of the active ARTDs, which is translated into either auto-ribosylation or hetero-ADP-ribosylation (e.g. of nearby histones), might have roles in the switch from transcriptional repression to transcriptional activation.
In the tumour context, ARTD8 and its ADP-ribosyltransferase activity have crucial roles. ARDT8, which is also known as collaborator of Stat6, regulates interleukin (IL)-4-dependent and Stat6-dependent transcription, inducing IL4-responsive genes . IL-4 is a cytokine that is produced by activated T lymphocytes and regulates the differentiation, proliferation and apoptosis of lymphocytes and other haematopoietic cells. IL-4 is responsible for the homeostasis of the immune system, and has anti-inflammatory properties under normal conditions. However, in the tumour environment, IL-4 might act through the Akt pathway to promote the survival of cancer cells, with Stat6 being responsible for mediation of this antiapoptotic action of IL-4. Two types of cell surface receptor for IL-4  are expressed on a wide variety of haematopoietic and non-haematopoietic cells, and these mediate the biological activities of IL-4. Both types have a common subunit, which is the functional receptor chain (IL-4R/CD124). The interaction of IL-4 with its receptor results in receptor dimerization and activation, which leads to Janus kinase activation and the consequent tyrosine phosphorylation of the cytosolic tails of the receptor. The phosphorylated tyrosines then serve as docking sites for a number of adaptor or signalling molecules, including Stat6 . IL-4-activated Stat6 dimerizes, translocates to the nucleus, and binds to specific DNA sequences, thus transcriptionally activating genes that are responsive to IL-4 [80, 82]. Members of the Stat transcription factor family are specifically activated by cytokines, and each Stat mediates its biological effects by trans-activating a unique profile of target genes [82, 83]. INF-γ signalling also leads to the activation of Janus kinase and the subsequent phosphorylation of Stat1, which homodimerizes, translocates to the nucleus, and induces the transcription of INF-γ-stimulated genes. Importantly, Stat6 recruits a number of coactivators at the promoter to activate transcription efficientlyp; among these is the protein p100, which interacts with RNA polymerase II, and functions as a bridging factor between Stat6 and the transcription machinery . In contrast to ARTD9, ARTD8 is enzymatically active and can catalyse both auto-modification and hetero-modification of p100 . In line with this, a catalytically inactive mutant of ARTD8 did not enhance Stat6-mediated transcription, and ADP-ribosyltransferase inhibitors blocked IL-4-dependent transcription . Thus, ARTD8 provides an example of transcription regulation that requires ADP-ribosyltransferase activity. Recently, a possible mechanism to explain how the ADP-ribosyltransferase activity of ARTD8 is involved in transcriptional regulation has been defined. Indeed, as well as its function as a transcriptional coactivator, ARTD8 was initially reported to function as a repressor by recruiting histone deacetylase 2 (HDAC2) and histone deacetylase 3 (HDAC3) to IL-4-responsive promoters. However, in the presence of IL-4, the ADP-ribosyltransferase activity of ARTD8 is activated, ADP-ribosylated HDAC2 and HDAC3 are released from the promoter, and efficient binding of Stat6 is induced, with consequent transcription . Thus, the ADP-ribosyltransferase activity of ARTD8 is needed to relieve its repressive function, and can be viewed as the transcriptional switch for Stat6-dependent gene induction (Fig. 1). Furthermore, altered proportions of B-cell subsets in the spleen have been reported in ARTD8-deficient mice, and it has been shown that IL-4-induced B-cell suppression of apoptosis after irradiation or growth factor withdrawal is mediated by ARTD8 through the induction of B-cell survival factors ; this is in line with a role of ARTD8 in B-lymphoid oncogenesis. Another important feature of cancer development is its requirement for a major supply of cellular energy in order to sustain continuous cell growth and proliferation. In B cells, the IL-4-mediated enhancement of glycolysis requires ARTD8, and its deficiency counteracts B lymphomagenesis driven by the Myc oncoprotein , thus suggesting that ARTD8 is at the centre of prosurvival signalling that is dependent on IL-4. Finally, the discovery that ARTD8 stabilizes phosphoglucose isomerase/autocrine motility factor, a cytosolic enzyme that is involved in tumour progression and metastasis, by inhibiting its ubiquitination , offers a new therapeutic target for inhibition of cancer cell migration and invasion during metastasis.
Our increasing molecular understanding of the heterogeneous subsets within DLBCL should improve the current therapy for DLBCL by identifying rational therapeutic targets in specific disease subtypes.
A role for ARTD15 in cancer can be hyphothesized
We provided the first demonstration that ARTD15/PARP16 is a novel ADP-ribosyltransferase with a new intracellular location . ARTD15 is the only known ARTD with a putative C-terminal transmembrane domain, and we have shown, using immunofluorescence and electron microscopy, that ARTD15 associates with the membranes of the nuclear envelope and the endoplasmic reticulum. Moreover, we showed, using protease protection assays, that it is a tail-anchored protein with a cytosolic catalytic domain. Importantly, we also identified importin β1/karyopherin β1 (Kapβ1) as a molecular partner of ARTD15  (Fig. 1). Kapβ1 is a soluble receptor protein with a pivotal role in nuclear transport and the shuttling of proteins with nuclear localization signals through the nuclear pore complex [90-93]. During this process, Kapβ1 binds directly to karyopherin α (Kapα), which, in turn, binds to the nuclear localization signal of target proteins, leading to the formation of a trimolecular complex [94-96]. This complex tethers to and passes through the nuclear pore complex through Kapβ1 binding to the nucleoporins. Once in the nucleoplasm, the complex releases the cargo protein, and Kapβ1 and Kapα are exported back to the cytosol to reinitiate a new import cycle [91, 92]. Dysregulation of this import process can lead to the mislocalization of both oncogenic and tumour-suppressor proteins, with the consequent loss of their functions [97-99]. Specifically, alterations in the expression levels of components of the nuclear transport machinery appear to be key determinants of development, differentiation, and transformation . High levels of Kapβ1 and Kapα and of the exportin Crm1 (Crm1 recognizes the leucine-rich nuclear export signal of proteins that need to be carried from the nucleus into the cytoplasm) have been shown in cervical cancer tissue as compared with normal cervical epithelium . It is of note that the inhibition of Kapβ1 expression in cervical cancer cells leads to cell death via apoptosis, which suggests that cervical cancer cells become functionally dependent on Kapβ1 overexpression. High levels of Kapα have been associated with apoptosis suppression and cancer development in a wide variety of tumours [101-104]. As the importer and exporter components of nucleo-cytoplasmic trafficking are upregulated in cancer cells, this might result in more efficient transport to sustain the higher proliferation rate of cancer cells. Kapβ1 molecules can also transport cargoes independently of Kapα. In addition, the structures of several different karyopherins have been solved, making these proteins attractive therapeutic targets. Both Crm1 and Kapβ1 have the potential to be regarded as biomarkers and therapeutic targets. Whereas normal epithelial and fibroblast cells are unaffected by Kapβ1 inhibition, cancer cells die as a consequence of Kapβ1 inhibition . Thus, targeting the proteins that are involved in nuclear transport can provide a powerful approach for controlling cancer cell growth. As we better understand the regulation of the transport of tumour suppressors and oncoproteins, new ways to perturb these pathways can be discovered: for instance, methods for redirecting mislocalized proteins to the correct cellular compartment can be developed. Here, through Kapβ1 mono-ADP-ribosylation, ARTD15 might represent a novel, crucial element in the regulatory mechanism of nucleo-cytoplasmic trafficking. In this respect, both the site of ADP ribosylation on Kapβ1 and the ARTD15 catalytic site can be thought of as potential targets for an innovative therapeutic strategy. In this respect, knowledge of the three-dimensional structure of ARTD15  is expected to speed up pharmaceutical research on specific drugs, which can then be evaluated by means of high-content, cell-based and high-throughput small-molecule screening. Such screening approaches might lead to the discovery of novel compounds that have anticancer activities, as well as compounds that provide new insights into the mechanisms of nucleo-cytoplasmic trafficking and cancer-regulatory pathways.
The endoplasmic reticulum is responsible for protein synthesis and folding, and has an important role in sensing cell stress and initiating the signalling pathway known as the unfolded-protein response (UPR) [106, 107]. ARTD15 has also been reported to have a role in the UPR, through its ability to ADP-ribosylate two components of the UPR: dsRNA-dependent protein kinase-like endoplasmic reticulum kinase and inositol-requiring 1, but not activating transcription factor 6 . These data link ARTD15 to inflammation; indeed, it has been shown that the UPR can initiate inflammation and is involved in the pathogenesis of inflammatory diseases . It is of note that Kapβ1 can also have a role in the regulation of endoplasmic reticulum protein quality control, through its ability to promote endoplasmic reticulum-associated degradation . Thus, considering ARTD15 as a drug target has the potential to aid in the development of therapies targeting the UPR and inflammation.
We thank E. Fontana for preparation of figures, and C. P. Berrie for editorial assistance. We acknowledge the financial support of the Banca Popolare di Lanciano e Sulmona.