ADP-Ribosylation as Post-Translational Modification of Proteins: Use of Inhibitors in Cancer Control

Among the post-translational modifications of proteins, ADP-ribosylation has been studied for over fifty years, and a large set of functions, including DNA repair, transcription, and cell signaling, have been assigned to this post-translational modification (PTM). This review presents an update on the function of a large set of enzyme writers, the readers that are recruited by the modified targets, and the erasers that reverse the modification to the original amino acid residue, removing the covalent bonds formed. In particular, the review provides details on the involvement of the enzymes performing monoADP-ribosylation/polyADP-ribosylation (MAR/PAR) cycling in cancers. Of note, there is potential for the application of the inhibitors developed for cancer also in the therapy of non-oncological diseases such as the protection against oxidative stress, the suppression of inflammatory responses, and the treatment of neurodegenerative diseases. This field of studies is not concluded, since novel enzymes are being discovered at a rapid pace.


Introduction
Post-translational modifications (PTM) of proteins require the activity of enzyme writers and erasers, and their presence is recognized by readers, protein modules, or structures that bind to specific chemical structures and organize protein-protein interactions. In PTM reactions, the amino acids that are modified may be available or may already be involved in other modifications. Therefore, different types of PTM may interfere with the same amino acids.
The purpose of this review is to update the current knowledge on ADP-ribosylation reactions. There are several types of enzymes, broadly defined as polyADP-ribosylating (PARylating) and monoADP-ribosylating (MARylating) enzymes. In the following paragraph we introduce the polymer-forming writers, the poly(ADP-ribose) polymerases. Then, a short presentation introduces the diphtheria-toxin-like ADP-ribose transferases (ARTDs), the cholera-toxin-like ARTCs, and the MARylating sirtuins. Several ADP-ribosylating enzymes are linked to cancer development, and the review discusses the proposed applications of specific ART inhibitors for the containment of cancers and their therapeutic potential.

Readers
There are several motifs for readers to recognize in MARylation and PARylation. The PAR-binding motif (PBM) is present in proteins involved in the DNA damage response (DDR), chromatin remodeling, or RNA processing. The PAR-binding zinc finger (PBZ) motif is present in proteins involved in DDR, such as the aprataxin PNK-like factor (APLF), and in the regulation of DNA damage-activated checkpoints, i.e., checkpoints with FHA and RING domains (CHFR). The WWE domain is present in two groups of proteins: E3 ubiquitin ligases such as RNF146 and, in a few cases, ARTDs (PARP1, tankyrases, ARTD8/PARP14, ARTD11). The RNF146/Iduna ubiquitin ligase associates with tankyrase, involved in PARylation-dependent ubiquitylation (PARdU) reactions: tankyrases PARylate their substrate proteins, and then RNF146 recognizes the PAR portion of PARylated proteins by its WWE domain and is allosterically activated to ubiquitylate them for proteasomal degradation [27][28][29][30]. Several tankyrase binding motifs (TBM) act in concert to allow the interaction between RNF146 and the ankyrin repeat clusters (ARC) of tankyrases [31]. An Oligonucleotide/oligosaccharide-binding (OB) fold is present in single strand-DNA binding protein 1 (hSSB1), which relocates to sites of DNA damage by the recognition of PAR polymers, promoting the DDR [32]. The exonuclease EXO1 contains a domain named "protein incorporated later into tight junctions (PilT) N terminus" (PIN). The PIN domain that binds to PAR facilitates EXO1 recruitment to DNA and supports the repair of double-strand breaks of DNA [33]. Other domains may contain the RNA recognition motif (RRM), the serine arginine repeats, and the lysine arginine repeats (SR/KR) motifs, or the Forkhead-associated (FHA) domain (binding to iso-ADP-ribose) and the BRCA1 C-Terminal (BRCT) in the Breast Cancer 1 (BRCA1) antioncogene, binding to (ADPR) n units [32,34]. ADPR-binding macrodomains can behave as readers, as in the case of AF1521 from Archeglobus fulgidus, that has been applied to proteomic studies for the enrichment of ADP-ribosylated proteins [35][36][37]. A PAR-recognizing macrodomain is present in a histone H2A variant named H2A1.1 (mH2A1.1). This macrodomain interacts with PAR and may support the repair of DNA by stabilizing the chromatin. MacroH2A1.1 is recruited to DNA regions rich in PAR, stabilizing their negative charges and slowing down PAR degradation, NAD + consumption, and cell death, allowing for an efficient DNA repair after DNA damage [38]. MARylating sirtuins such as SIRT7 attract the ADP-ribose-binding macrodomain of histone H2A1.1 and promote H2A1.1 accumulation in loci containing metabolic genes [39]. ALC1/CHD1L is a macrodomain protein, a PAR-binding Snf2-like ATPase that has oncogenic activity and is frequently overexpressed in hepatocellular carcinoma (HCC) and other types of cancers. ALC1 is also a chromatin remodeler, and a helicase that supports nucleosomal sliding. In homologous recombination-deficient cells such as in BRCA-mutated phenotypes, the loss of ALC1 sensitizes cells to the PARP inhibitor, in spite of various resistance mechanisms, and reduces the chromatin accessibility to DNA repair proteins [32,40]. Since there are several motifs recognizing specific structures, some of them may compete with other PAR binding proteins, inhibiting their interactions; steric inhibition is another mechanism that may occur to impede the PAR binding, making the region inaccessible to the interacting readers. Other macrodomain proteins behave both as readers and as erasers.
TARG1 performs hydrolase activity toward MAR/PAR as well as O-acetyl ADPR (OAADPr) deacetylase activity. Ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) transforms the last ADPR group, into phosphoribose bound to the proteins leaving 5 AMP. Similarly, nucleoside diphosphates linked to some moiety X (NUDX/NUDT) hydrolases cleave ADP-ribose to produce 5 AMP and ribose-5-phosphate (R5P) [16,52,53]. Various NUDIX/NUDT enzymes are overexpressed in cancers. In particular, NUDT2 and NUDT5 are linked to aggressive adenocarcinomas and homologous recombination (HR)-positive breast cancers, respectively [52], linking them to ADPR catabolism and the NAD + salvage pathway [54,55]. PAR-and MAR-processing enzymes are often deregulated in cancers. MacroD1 is also named leukemia-related protein 16 (LRP16). An aberrant MacroD1 expression has been associated with leukemia and breast, colorectal, gastric, lung, and liver cancers [53,54]. When overexpressed, MacroD2 is associated with estrogenindependent growth and tamoxifen resistance in breast cancer [50,56]. The overexpression of eraser proteins may disrupt the physiology of the signaling role of ADP-ribosylation reactions: following the activity of MARylating enzymes, a pathological accumulation of erasers will reverse the modification rapidly and subvert the requirement of this PTM in a determined timing and cellular localization, preventing the downstream signaling from being transduced.

Various Types of Post-Translational Modification of PARP1 and Competition on Amino Acid Substrates
Concerning PARP1, several post-translational modifications regulate its activity. A SUMOylation on K486 renders this lysine unavailable for acetylation, determining a competition between SUMOylation and acetylation modifications of PARP1 [57]. Sirtuin 6 (SIRT6) directly associates with PARP1 and MARylates PARP1 at lysine 521, stimulating PARylation activity and DNA repair [58]. A similar role is ascribed to PARP3, that MARylates PARP1 on a different amino acid. In PARP1, lysine 508 (K508) can be either acetylated by p300/CBP acetylase or methylated by SET7/9. Thus, when K508 is methylated, PARP1 cannot be recruited to PARylate the p50 subunit of transcription factor NF-kB during the inflammatory response. K508 methylation by SET7/9 can be blocked by PARP1 auto-modification, probably due to steric hindrance and charge effects, making this PTM alternative to methylation. Histone methylase SMYD2 modifies lysine 528 on PARP1, enhancing the PARylation activity in response to oxidative stress [59,60]. Histones are displaced from nucleosomes and chromatin by PTMs. PARP1 ADP-ribosylates lysine residues of the core histone tails. Histone H3 ADP-ribosylation is a PTM competing with H3 methylation by SET7/9. Thus, SET7/9 methylates H1 and H3 in the absence of PARP1. However, the PARylation of H3 stimulates H1 methylation by SET7/9 [61].
The extracellular signal-regulated kinases 1/2 (ERK1/2) promote PARP1 activation through the direct phosphorylation of PARP1 at serine 372 and threonine 373 [62]. The phosphorylation sites are located near the beginning of the BRCT portion of the automodification domain, which has been proposed to modulate PARP1 activity [57]. PARP1 and CDK2 specifically interact following hormone treatment via the BRCA1 C-terminal domain (BRCT) of PARP1 and the cyclin-binding domain (CBD) of CDK2, which facilitates the phosphorylation of PARP1 at S785/S786. The phosphorylation of the serines on PARP1 results in an opening of the NAD + -binding pocket within the catalytic domain, increasing PARP1 activity, which leads to auto-PARylation. Furthermore, PARP1 and CDK2/cyclin E are recruited to chromatin, resulting in PARylated proteins and the opening of chromatin through the displacement of histone H1. The receptor tyrosine kinase c-Met associates with and phosphorylates PARP1 at Tyr907 (pTyr907 or pY907) [63]. The enzyme activity of c-Met is required for the PARP1 and c-Met interaction, following H 2 O 2 treatment. The C-Met-dependent PARP1 phosphorylation at Tyr907 increases PARP1 activity and results in a reduced affinity to PARP inhibitors, thereby rendering cancer cells resistant to PARP inhibitors. On the other hand, cancer cells treated with c-Met inhibitors become more sensitive to PARP inhibitors. EGFR is another kinase that phosphorylates PARP1 [63], and the inhibition of both EGFR and PARP1 results in synthetic lethality in highly aggressive triple negative breast cancers (TNBCs).

Cancer Proliferation, Metastasis, Angiogenesis: The Role of ADP-Ribosylating Enzymes According to a Deregulated Expression or Use of ART Specific Inhibitors
Various cellular functions are disrupted by the altered expression of ADPR writers, readers, and erasers. Several enzymes involved in ADP-ribosylation are linked to some form of cancer. Specific enzymes may cooperate in one or several forms of intrinsic hallmarks of cancer: uncontrolled proliferation, evasion of growth suppressors, cell death resistance, genome instability, reprogrammed energy metabolism, and escape from replicative senescence [64][65][66][67]. PARPs generating PAR modifications include PARP1/2 and the tankyrases, both of which are targets for cancer therapy. Finally, a paragraph will be focused on MARylating enzymes with specific involvement in certain types of cancer.
PARP inhibitors were also tested in RNASEH2B deleted cancers, in HER negative cancers, in few cases-as the pretreatment with adjuvant taxanes or anthracycline [73]. Cancers defective in PTEN, as prostate cancers, can be sensitive to PARP inhibitors, due to the downregulation of RAD51. Since lysine methylase SMYD3, often overexpressed in tumors, regulates BRCA1/2 and DNA repair, the inhibition of SMYD3 with BCI-121 promotes cancer inhibition by PARP inhibitors such as olaparib [70]. Olaparib, as well as similar inhibitors, may induce cancer resistance in patients due to the extrusion by multidrug resistance proteins and ATPbinding cassette pumps. PARP1 inhibitor AZD2281 was more effective when co-administered with the P-glycoprotein inhibitor tariquidar [71][72][73]. The use of PARP inhibitors has been approved for treatment in BRCA1 and BRCA2 mutated breast and ovarian cancers, and for therapy in platinum sensitive ovarian cancer patients with defective BRCA1/2 genes [74]. Olaparib resistance may occur in cancers with homologous recombination defects [73]. Re-sistance due to the reversion of mutated BRCA1/2 and other mechanisms is common; these mechanisms include: PARP1 loss of function and the silencing of DNA damage response (DDR) proteins such as 53BP1 and REV7, that reactivate the HR pathway [72][73][74]; selective loss of PARG [75][76][77]; loss of function of proteins involved in destabilization of the replication fork (EZH2, MUS81) [78]; high expression of CCAAT /enhancer-binding protein β (C/EBPβ), upregulating the HR pathway (BRCA1, RAD51) [79]. Ribonucleotide excision repair requires topoisomerase 1 and PARP1 activities: RNASEH2B mutant prostate cancers were found to be hypersensitive to PARP inhibitors [73,80]. In RAD51C-silenced high-grade serous ovarian cancers, RAD51 re-expression was associated with an acquired loss of RAD51 promoter methylation during PARP inhibitors therapy [81]. The deletion or inhibition of eraser enzymes such as PARG may potentiate synthetic lethality drugs in cancers with mutant phenotypes [42,43,75]. The synthetic lethal interaction between ARH3 and PARG may set the basis for new therapeutic approaches [82]. The loss of ARH3 was shown as a mechanism to develop PARP1/2 inhibitor resistance [82]. The application of a DNA methyltransferase inhibitor (DNMTi) may re-sensitize tumors to primary therapies [83]. In cells deficient in PTEN, overexpressing SAMHD1 with 53BP1 loss, PARP inhibitors were ineffective [84]. E2F7 is a member of the E2F transcription factor family. E2F7 and E2F8 are atypical E2F family members and mediate transcription repression, particularly of homologous recombination proteins. E2F7 in BRCA2-deficient cells confers sensitivity to PARP inhibitors, while its depletion counterbalances the chemosensitivity of BRCA2-deficient cells, by promoting both HR and fork stability [85].
PARP3 deficiency or inhibition leads to the growth suppression of BRCA1-deficient TNBC cells [86,87]. The genetic PARP3 knock-down led to decreased survival and in vivo tumorigenicity, results dependent on a defective Rictor/mTORC2 signaling complex, and to ubiquitination and degradation of Rictor. Accordingly, PARP3 MARylates GSK3β, a positive regulator of Rictor ubiquitylation [86]. PARP3 selective inhibitor ME0328 displays higher selectivity over ARTD1 and its nearest homologs [11]. Non-NAD + -like PARP1 inhibitors, binding to sites different from the NAD + binding site, have been studied for prostate cancer treatment [88]. Several conjugates of ADP and morpholino nucleosides target PARP1/2/3 with high specificity [89], while ABT-888 showed high specificity toward PARP2 [90]. A different approach to reverse PARP inhibitors resistance has been proposed through targeting autophagy, and the enzymes controlling autophagy in HR repair proficient breast cancers [91]. Topoisomerase 1 inhibitors have been shown to be effective in cancer therapy of HR-deficient, Schlafen 11-positive cells in association with Olaparib, for the treatment of BRCA1-, BRCA2-, and PALB2-deficient cells [92]. Researchers are continuously developing new scaffolds for the setup of PARP inhibitors [93]. Furthermore, PARP inhibitors' association with HDAC inhibitors such as SAHA or belinostat have shown benefits in prostate cancer [94]. A list of PARP1/2/3 inhibitors applied to cancer treatment is presented in Table 1.

Tankyrases: Cell Functions and TNKS Inhibitors in Cancer Treatment
High levels of TNKS1 and/or TNKS2 expression have been observed mainly in colon, lung, brain, breast, ovarian, and liver cancers [31]. Several mutations altering PARylation activity may occur, depending on large deletions or point mutations. The stability of many proteins is regulated via PARdU: Axin1/2, PTEN, 3BP2, and TRF1 are processed and degraded by PARdU, as well as the actors of this activity, TNKS1/2 and RNF146 [31]. Tankyrases modify the tumor suppressors phosphatase and tensin homolog (PTEN) and axin 1/2, which regulate the levels of phosphatidylinositol 3,4,5 triphosphate and β-catenin, negatively regulating the Akt/PI3K pathway (PTEN) and Wnt signaling (axin). After PTEN or axin1/2 are PARylated by TNKS1/2 on acidic residues, RNF146 ubiquitylates on lysines, destining them to proteasomal degradation. Wnt-driven cancers such as CRC, HCC, and lung cancers may benefit from Tankyrase inhibitors [31, 101,102]. In lung cancer, most often showing TNKS overexpression, tumors possess abnormal Wnt activity, due to APC and β-catenin mutations, as well as to the deregulation of upstream Wnt signaling effectors, such as Dishevelled 3 (Dvl-3), or the downregulation of Wnt antagonists, as Wnt-inhibitory factor 1 (WIF 1). Among the substrates of Tankyrases, AXIN is a key regulator of the canonical Wnt signaling pathway. Wnt activation regulates the level and subcellular localization of the β-catenin transcription factor. Glycogen synthase kinase 3β (GSK3β), in collaboration with axin, APC (adenomatous polyposis coli), and other proteins, phosphorylates β-catenin, that is then degraded by the proteasome [102][103][104]. The activated Wnt complex inhibits the AXIN/GSK3β complex and stabilizes β-catenin, which can translocate into the nucleus. β-catenin protein levels are kept low by phosphorylation, so that β-catenin is ubiquitylated by a complex containing a β-transducin repeat-containing protein (βTrCP). Tankyrases PARylate AXIN and recruit the E3 ubiquitin ligase RNF146 containing the PAR-binding WWE domain, which ubiquitylates AXIN, destining it for proteasomal degradation. In this optic, tankyrase promotes Wnt signaling, while TNKS inhibitors are useful in the treatment of Wnt-driven cancers [31,104]. G007-LK is a selective tankyrase inhibitor [105]: the treatment with this inhibitor induced the loss of expression of MYC and impaired cell growth, with the accumulation of β-catenin degradasomes. IWR1 and AZ-6102 are also selective for tankyrase [106]. TNKS1/2 PARylate PTEN. PTEN contains a Tankyrase Binding Motif (TBM) at the N-terminal region, close to the phosphatase domain. RNF146 recognizes PARylated PTEN and performs its ubiquitylation, destining it to degradation by the proteasome. Endometrial cancer presents the highest rate of TNKS1 and TNKS2 mutations [31], followed by colorectal cancer, bladder, melanoma, esophagogastric, and prostate cancers. High levels of TNKS1 and/or TNKS2 expression have been found in colon, lung, brain, breast, ovarian, and liver cancers. In the first attempts to develop ART inhibitors, few chemical structures were shown to target PARPs and tankyrases: XAV939 is a promiscuous inhibitor of PARP1 and tankyrases, while PJ34 and UPF1069 are broad PARP/tankyrase inhibitors. In recent years, inhibitors were developed specifically for tankyrases: some of them bind to the nicotinamide subsite (NS) (such as XAV939, RK-582, LZZ-02, AZ1366), or to the adenosine subsite (AS) (IWR-1, OM-1700, OD366, G007-LK and K-756) [107][108][109][110][111][112][113][114][115][116]. Dual binders (DB), recognizing both nicotinamide and adenosine subsites, have been found among PARP1 inhibitors (olaparib): for TNKS, several inhibitors showed a potential therapeutic use (CMP4, WIKI4, TNKS656) [31]. K-756 and RK-287107 block cell growth in COLO-320DM and SW403 cancer cells, carrying a truncated form of APC lacking a short stretch of 20 amino acids involved in β-catenin binding. RKO, DLD-1, and HCC2998 cells, possessing a different, long truncated form of APC, that preserves the binding to β-catenin, do not respond to K-756, but are sensitive to LZZ-02. The proposed approach to treat APC mutated cancers is the use of G007-LK, LZZ-02, or RK-582 at low dosage in association with the PI3K (BKM120) and the epidermal growth factor receptor (EGFR) (erlotinib) inhibitors for colon cancer treatment [31,113]. The proteins targeted by TNKS include vinculins, that anchor F-actin to adherens junctions (AJ), structures involved in cancer development, while TNKS inhibitors prevent the assembling of AJ [117].
ARTD10 has tumor suppressor activity by inhibiting myc and by the MARylation of Aurora A kinase, inhibiting its phosphorylation on the same residue: The MARylation is exerted after interacting with the ubiquitin ligase RNF114, which ubiquitylates ARTD10, activating it; a loss of RNF114 in cancers (mutation, deletion, downregulation) may lead to a loss in ARTD10 activity [1,133]. ARTD10 may also have oncogenic activity, being involved in some cancers, promoting the proliferation and restart of stalled replication forks, and alleviating cell survival during replication stress [1,148].
However, to complete a MARylation cycle, the writers and the eraser should be expressed at similar levels. MAR hydrolases cleave the modifications promoted by ARTD10. MacroD2, expressed in neuroblastomas, and MacroD1, overexpressed in various cancers, can counteract the tumor suppressive role of ARTD10 [1,49]. The MARylation of ARTD10 is recognized by ARTD8/PARP14, that is docked to the MAR structure to form a protein complex. Several molecules have shown the potential to selectively inhibit ARTD10 [14, [130][131][132][133], such as the cell-permeable OU135, a 3,4-dihydroisoquinolin-1(2H)-one that contains a methyl group at the C-5 position and a substituted pyridine at the C-6 position [131], and 4-benzyloxybenzimide derivatives [130]. The targets of ARTD10 that could require a fine-tuning of MARylation are glycogen synthase kinase GSK3β, NEMO/IKK-γ, a subunit of NF-kB transcription factor, histone H3, and the polo-like kinase PLK1 [1]. Targeting ARTD10 could decrease the proliferation of certain types of cancer and HCC cells overexpressing ARTD10. ARTD11 MARylates ubiquitin E3 ligase β-TrCP, regulating the IFN-I-dependent antiviral activity [135]. ARTD11 inhibitors showed the release of ARTD11 from nuclear envelope complexes, where it associates with NUP153 [136] (Figure 3). In addition, ARTD11 was shown to interact with ARTD12 and to inactivate the replication of the Zika virus through the degradation of NS1 and NS3 viral proteins [137]. ARTD14/PARP7 is amplified and highly expressed in squamous cell carcinomas, in ovarian and lung tumors, and associated with poor survival [138][139][140]. The MARylation of Protein substrates such as α-tubulin, by ARTD14/PARP7, has a role in microtubule control in ovarian cancer [141,146,147]. The enzymes are also involved in the maintenance of stemness and pluripotency [142]. PARP-7 MARylates immune-relevant protein targets and modulates cancer-directed host immune responses [143]. RBN-2397's specific inhibition of ARTD14 suppresses cell proliferation and activates the type-I IFN pathway in cancer cells [144][145][146][147]. RBN-2397 is in a phase 1 clinical trial for patients with advanced solid tumors (Identifier: NCT04053673). The ARTD14/PARP7 dependency for proliferation of cancer cells (i.e., lung cancer) and cells with high baseline expression of interferon (IFN)-stimulated genes. Interestingly, RBN-2397 enhanced IFN signaling and induced both cancer-cell-autonomous and immune-stimulatory effects [145][146][147].
ARTD15/PARP16 associates with the endoplasmic reticulum (ER) [149][150][151][152]. It is a single-pass transmembrane protein with the N-terminal region (amino acids 1-280) facing the cytoplasm and a C-terminal tail facing the ER lumen. ARTD15 interacts with the nuclear transport factor karyopherin-1/importin-1 (Kap-1β), MARylating it [150]. ARTD15 automodifies itself and MARylates the double-stranded RNA-dependent protein kinase (PKR)like ER kinase (PERK) and the inositol-requiring enzyme 1α (IRE1α) [150,151]. By this modification, ARTD15 activates PERK and IRE1α, two proteins relevant for the ER stress response and for the unfolded protein response (UPR). During ER stress, ADP-ribosylated PERK and IRE1α increase their kinase activities as well as IRE1α endoribonuclease activity. ARTD15 inhibitors based on latonduine analogs have been validated and tested to correct mutated CFTR by blocking IRE1α modification [153,154]. ARTD15 is associated with UPR-dependent inflammation, involved in inflammatory diseases [6]. Through Kapß1 MARylation, ARTD15 regulates and controls nucleo-cytoplasmic trafficking. Inhibitors targeting the site of ADP ribosylation on Kapß1 and the ARTD15 catalytic site are potential drugs for innovative therapeutic strategies.
ARTD17/PARP6 is an enzyme involved in some cancers. It has been identified as a negative regulator of cell-cycle progression in HeLa cells: ARTD17 overexpression was reported to arrest cells in the S-phase; this activity is dependent on a functional catalytic domain. In colorectal cancer, ARTD17 promotes cancer growth [136]. In addition, ARTD17 exerts proliferative effects, while its inhibition induces the formation of multipolar spindles in breast cancer [155][156][157][158]. By screening a library of compounds for the ability to induce mitotic defects, researchers identified AZ0108 as a potent ARTD17 inhibitor, showing antitumor effects in vivo and inducing cell death in breast cancer cells in vitro [144,157].
Concerning cholera-toxin-like ARTs, ARTC1 MARylates the Grp78/BiP chaperone, required for protein quality control in endoplasmic reticulum [159]; this modification is relevant for the ER stress response system. Either alterations in UPR or in ER stress response can have consequences in normal cells and in cancer development. ARTC1 regulates the RhoA/ROCK/AKT/β-catenin pathway in colon carcinoma, possibly through the regulation of kinase signaling pathways by the direct ADP-ribosylation of key kinases [160][161][162][163].
A list of MARylating enzymes and their potential inhibitors applied to cancer treatment is presented in Table 4.

Sirtuin Inhibitors Addressing the Roles of MARylating Sirtuins in Cancer
SIRT1, SIRT2, SIRT3, and SIRT6 are blocked by a sirtuin inhibitor selisistat (Ex527) targeting the NAD + binding pocket; the sirtuins are not inhibited by PARP inhibitors such as olaparib and similar compounds, PJ34, and XAV939, due to differences in their NAD + binding site [166,167].
SIRT4 is localized in the outer mitochondrial membrane. SIRT4 MARylates GDH [17] and deacylates pyruvate dehydrogenase (PDH), two enzymes that regulate the tricarboxylic acid (TCA) cycle. The MARylation of GDH inhibits the conversion of glutamate to α-ketoglutarate (α-KG), decreasing the glutamine uptake: α-KG upregulates H3/H4 histone acetylases and stimulates histone lysine demethylases; on the other hand, during the TCA cycle, the accumulation of fumarate and succinate leads to the inhibition of α-KG-dependent enzymes, while an increase in citrate inhibits α-KG-dependent N6methyladenine demethylase (ALKBH5). SIRT4 was also associated with negative impacts on the mitochondrial quality and with aging [27].
SIRT6 was found to be overexpressed in skin cancer and non-small cell lung cancer (NSCLC) [168,169], osteosarcoma, colon carcinoma, serous ovarian cancer, and clear cell renal cell carcinomas (ccRCC) [169][170][171][172][173], showing a poor prognostic value. However, in certain types of cancer, SIRT6 is classified as a tumor suppressor. Since SIRT6 deacetylase activity has important effects in cells and in cancers, and since NAD + -binding inhibitors block both deacetylation and MARylation reactions, it is not possible to discriminate between the two activities in SIRT6 studies [174]. SIRT6-silenced melanoma cells showed a considerable anti-proliferative effect both in vitro and in vivo [175]. SIRT6 has been involved in genome integrity, DNA repair, energy metabolism, and inflammation. SIRT6 levels decrease during aging and cell senescence [176].
A study on the structure of bacterial SIRT bound to the acetylated +2 arginine peptide showed that the arginine entered in the active site and reacted through a deacetylation reaction intermediate, yielding an ADP-ribosylated peptide [167]. New compounds, such as diketopiperazine-containing 2-anilinobenzamides, simultaneously targeting the "selectivity pocket" substrate-binding site and the NAD + -binding site [180], effective in SIRT2 inhibition, have been set up as the basis to develop new specific inhibitors [191,192].
SIRT7 has been involved in genome integrity and non-homologous end-joining (NHEJ) DNA repair [96]. SIRT7 has auto-MARylation activity [39]. SIRT7's auto-MARylation occurs on several sites: proteomic studies identified 7-8 MARylated peptides. Auto-MARylation modifies SIRT7's chromatin distribution. The ELHGN catalytic motif that is conserved among sirtuins, H187, is involved in deacetylation and faces the NAD + -binding pocket and the catalytic site, together with the flanking residues E185 and N189. In SIRT6 and SIRT7, E185 and N189 are in the opposite direction, facing the surface of the cavity, and the two residues interact forming a loop. These residues are important for their role in the ADP-ribosylation reaction: E185 is the catalytic residue that starts the reaction, while N189 is the first acceptor of the ADP-ribosyl moiety [39]. In Table 5 a list of candidate sirtuin inhibitors is presented.

Discussion and Perspectives
In this review the most studied players of MARylation and PARylation have been reviewed. The main objective was to relate the proteins targeted by these post-translational modifications and their interactions with specific proteins toward a disruption of their roles in the maintenance of cell functions through MARylation/PARylation inhibitors. These disfunctions lead to cell proliferation, anomalies in growth control, loss of cell-to-cell contacts, and tumor development. Furthermore, worthy data have been produced through clinical trials on several types of PARP1/2/3 inhibitors, on tankyrase/PARP5a/b inhibitors, and on MARylation inhibitors targeting ARTD14/PARP7. The deregulated cell functions may derive from the MAR/PAR cycling control, stemming from the overexpression of eraser enzymes that have been associated with various cancers. Approaches using PARP1/2 inhibitors or broad PARP inhibitors have been proposed and tested not only in cancer treatment [98][99][100], but also for non-oncological diseases: for instance, in the protection against oxidative stress, in the suppression of inflammatory responses, in the treatment of ALS and neurodegenerative diseases, and in TDP-43 dependent neurological disorders [194][195][196]. Thus, PARP inhibitors are seeing a new wave of medical exploitation both in cancer and in inflammatory diseases, since it can be applied in the therapy of non-oncological diseases.
Similarly to MAR erasers, several RNA viruses express nsp3, a macrodomain protein with eraser activity that subverts the stress granule structure and re-establishes the translation and function of stalled mRNAs [197].
Recently, new data have been produced on MacroD1, MacroD2, and TARG, uncovering their interactomes and their cell specificity [198,199]. Among these findings, the specific cellular localization of the three eraser proteins, the mitochondria for MacroD1, the nuclei, nucleoli and stress granules for TARG1, and the nuclei and cytoplasm for MacroD2, were found to be expressed mainly in tissues of neural origin.
Furthermore, compounds able to inhibit and discriminate between ARH3 and ARH1 have been developed and could be tested to protect cells from the deregulated MAR/PAR hydrolytic activity or the inhibition of the nsp3 macrodomain of SARS-CoV2 [199,200].
The perspectives for ADP-ribosylation's future studies lead in the direction of antiviral therapies and the development of more specific inhibitors of the MAR/PAR modifications and their control, for the amelioration of oncological and non-oncological diseases.
Finally, a new enzyme with supposed PARylating activity has been associated with mitochondria: Neuralized-like protein 4 (NEURL4) possesses an H-Y-E-like ART domain in the C-terminal domain, in addition to six Neuralized Homology Repeat (NHR) domains, functioning as protein-protein interaction scaffolds [201][202][203]. The PARylating activity associated with mitochondria is lost in cells lacking NEURL4. One main protein target for PARylation is mitochondrial Ligase III (mtLIG3), required for mtDNA stability. Furthermore, LRRC9-ART has been characterized at the expression level, and is predicted to interact with ZFP36L2, an RNA-binding protein that controls the cell cycle and is involved in pancreatic cancer [201,202]. New macrodomain proteins have also been identified, such as C12ORF4, a cytoplasmic protein predicted to have eraser activity, involved in mast cell degranulation [203]. C12ORF4 is highly expressed in glioblastomas, adenocarcinomas of the ovary and pancreas, and lymphomas. For a more rapid circulation of the new findings within the PARP/ART community, it is important to update the databases on the enzymes involved in this PTM, such as ADPriboDB 2.0 [204]. Just as importantly, also in the plant kingdom, a sirtuin enzyme has been reported to possess MARylating activity [205]. New findings will support MAR/PAR studies for a deeper understanding of their functions.
Further studies analyzing 3D structures of MARylation/PARylation enzymes should help develop new inhibitors. It is envisaged that in the next year new therapies will be tested via the exploitation of PARylation and MARylation inhibition, the safety of the administration of new drugs, and the recovery to a healthy state.