chemotherapeuticsaccordingtotheactionmechanismofPARPinhibitorsonthecombinationstrategies.

**Table**

**2.**

Combination

e

ffects

of

PARP

inhibitors

and

Their synergistic effect on chemoagents by inhibition of catalytic activity is well defined by topoisomerase I inhibitors, topotecan and camptothecin. The synergistic effect of combination therapy with alkylating agents such as MMS is well defined by the PARP-DNA trapping activity of PARP inhibitors (Figure 3) [77–79]. First, alkylating agents generate basic sites that consist of a 1-nucleotide gap with 3 -OH and 5 -deoxyribose phosphate (5 -dRP) groups at the ends of the breaks created by APEX1 endonuclease. PARP1 binds directly to 5'-dRP and recruits BER proteins to induce repair. However, PARP inhibitors bind PARP to the 5-dRP end, trapping and maintaining the PARP-DNA complex so that the accumulation of SSBs leads to DSBs, and ultimately, cell death [60,61,66,78,80]. This synergistic effect based on PARP trapping is the strongest for talazoparib. On the other hand, a topoisomerase I inhibitor causes SSBs by the endonuclease activity of TOP1 and causes DNA damage by trapping TOP1cc covalently bonded with TOP1 at the DNA 3' end. At this time, the PARP inhibitor sustains the trapping of TOP1cc by inhibiting PARP1 recruitment of TDP1 through PARylation, so that TDP1 can remove the covalent attachment of TOP1 by phosphodiesterase activity. At this point, the PARP inhibitor lets the trapping of TOP1cc continue by preventing TDP1 from removing the covalent attachment of TOP1, as PARP1 recruits TDP1 through PARylation. This is due to the regulation of PARylation activity by PARP inhibitors, which has the same effect on all PARP inhibitors developed in the clinic [63,77,79]. Those applying combination therapies of PARP inhibitors and chemotherapy within the clinic should carefully consider the mechanism(s) of action prior to selecting the drug.

**Figure 3.** Principle of combination therapy with chemoagents, based on trapping effects and inhibition of catalytic activity. (**A**) An alkylating agent forms a single-nucleotide gap with 5'-deoxyribose phosphate (5-dRP). PARP1/2 senses and binds it, inducing recruitment of BER molecules. A PARP inhibitor prevents repair by inhibiting dissociation of PARP via trapping the PARP-DNA complex, which is PARP bound to 5-dRP. (**B**) TOP1cc, a covalent binding state of TOP1 induced by Top1 and a DNA 3 -end, is repaired by TDP1 recruited to TOP1cc by PARylation and PAR transferase of PARP1 to induce TOP1-DNA complex excision. Topoisomerase I inhibitors continuously induce TOP1cc. PARP inhibitors inhibit catalytic activity, suppressing the recruitment of TDP1 and TOP1-DNA covalent complex repair, resulting in a synergistic effect.

#### **5. Conclusions**

PARP performs a variety of functions from the transcriptional level, to activation and localization through post-translational modification. In the DNA damage response, PARP contributes to the activation of itself or its target protein through the regulation of PARylation. PARP inhibitors block this catalytic activity of PARP, preventing the activation of normal repair pathways. These PARP inhibitors have demonstrated dramatic anti-tumor effects for tumors with HRD, such as those with BRCA mutations. To date, four PARP inhibitors have been approved by the FDA and applied in clinical practice. However, these four have different effects on the trapping of the PARP-DNA complex, despite inhibiting the common catalytic activity of PARP. This PARP trapping leads to improved cytotoxicity via replication fork collapse, leading to conversion to DSBs. When PARP inhibitors are used in combination with alkylating agents, synergistic effects are achieved. In contrast, inhibition of the catalytic activity of PARP has a synergistic effect when combined with topoisomerase I inhibitors. In other words, the synergies of combination therapies with PARP inhibitors can be induced differently depending on the mechanism of action of individual PARP inhibitors. Understanding the characteristics of each PARP inhibitor to strategically select synergistic partners is an important matter that must be considered to produce maximum antitumor effects.

**Author Contributions:** A.M. and S.-A.I. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** Seock-Ah Im is a recipient of research funds from AstraZeneca Inc., Pfizer and Roche, has consultant and advisory roles for AstraZeneca, Amgen, Eisai, Hanmi Corp, Lilly, MediPacto, Novartis, Pfizer, and Roche. Ahrum Min declared that she has no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Immunomodulatory Roles of PARP-1 and PARP-2: Impact on PARP-Centered Cancer Therapies**

#### **José Yélamos 1,2,\*, Lucia Moreno-Lama 1, Jaime Jimeno <sup>3</sup> and Syed O. Ali <sup>4</sup>**


Received: 30 December 2019; Accepted: 6 February 2020; Published: 8 February 2020

**Abstract:** Poly(ADP-ribose) polymerase-1 (PARP-1) and PARP-2 are enzymes which posttranslationally modify proteins through poly(ADP-ribosyl)ation (PARylation)—the transfer of ADP-ribose chains onto amino acid residues—with a resultant modulation of protein function. Many targets of PARP-1/2-dependent PARylation are involved in the DNA damage response and hence, the loss of these proteins disrupts a wide range of biological processes, from DNA repair and epigenetics to telomere and centromere regulation. The central role of these PARPs in DNA metabolism in cancer cells has led to the development of PARP inhibitors as new cancer therapeutics, both as adjuvant treatment potentiating chemo-, radio-, and immuno-therapies and as monotherapy exploiting cancer-specific defects in DNA repair. However, a cancer is not just made up of cancer cells and the tumor microenvironment also includes multiple other cell types, particularly stromal and immune cells. Interactions between these cells—cancerous and non-cancerous—are known to either favor or limit tumorigenesis. In recent years, an important role of PARP-1 and PARP-2 has been demonstrated in different aspects of the immune response, modulating both the innate and adaptive immune system. It is now emerging that PARP-1 and PARP-2 may not only impact cancer cell biology, but also modulate the anti-tumor immune response. Understanding the immunomodulatory roles of PARP-1 and PARP-2 may provide invaluable clues to the rational development of more selective PARP-centered therapies which target both the cancer and its microenvironment.

**Keywords:** PARP; immunomodulation; tumor microenvironment

#### **1. Introduction**

Poly(ADP-ribose) polymerase-1 (PARP-1) and PARP-2 are two enzymes of the PARP family of proteins that, in response to DNA damage, catalytically cleave β-NAD<sup>+</sup> and transfer ADP-ribose moieties onto specific amino residues of acceptor proteins. This process, termed poly(ADP-ribosyl)ation (PARylation), forms poly(ADP-ribose) (PAR) polymers varying in size and branching, which have diverse functional and structural effects on target proteins [1–3]. The deletion of either PARP-1 or PARP-2 in mice is associated with disturbances of DNA integrity and repair, supporting key shared functions of these proteins that are pivotal to DNA repair [4]. Indeed, combined PARP-1 and PARP-2 deficiency leads to embryonic lethality [5], which is likely due to their central role in the DNA damage response (DDR) [2,4].

Studies based on the role of these PARPs in the DDR in cancer cells have led to the development of PARP inhibitors as new therapeutic tools in cancer, both as adjuvant treatment potentiating chemotherapy, radiotherapy, and immunotherapy and as monotherapy exploiting cancer cell-specific defects in DNA repair, such as BRCA mutations [6–9]. However, the tumor microenvironment is formed from more than just tumor cells, and also includes stromal cells and infiltrating cells of the innate and adaptive immune system, which are likely to also be affected by PARP inhibition. These cells communicate with each other through direct contact and/or indirect signals that can alter the functionality of immune cells so that they either favor or limit tumor growth [10,11]. Emerging evidence supporting the immunomodulatory roles of PARP-1 and PARP-2 has raised the prospect of harnessing PARP inhibition to not only target the cancer itself, but also therapeutically modify its microenvironment.

In this review, we highlight the functions of PARP-1 and PARP-2 in the immune system and how their immunomodulatory roles might impact the response to tumors. We will examine recent data suggesting specific and redundant roles of PARP-1 and PARP-2 in the innate and adaptive immune responses and the immunological potential of PARP inhibitors. Understanding the immunomodulatory roles of PARP-1 and PARP-2 may provide invaluable clues for the rational development and exploitation of more selective anti-cancer PARP inhibitor drugs, both as new monotherapeutic approaches and in combinations with immunotherapy.

#### **2. Impact of PARP-1 and PARP-2 on T Cell Development and Function**

T cell development is a highly regulated process beginning in the thymus from bone marrow-derived lymphoid precursors, and giving rise to mature T cells through well-characterized sequential maturation steps involving a complex transcriptional network orchestrating cell proliferation, survival, and differentiation [12]. The earliest thymic progenitors are named double-negative (DN) cells, comprising four fractions (DN1 to DN4), which are characterized by a lack of CD4 and CD8 surface markers. DN2 and DN3 thymocytes express recombination-activating genes (Rag) and undergo extensive T cell receptor (TCR) β, γ, and δ gene rearrangement to express functional TCR chains. A successful recombination of TCRγ and TCRδ promotes the generation of γδ T cells. In contrast, the generation of αβ T cells requires additional differentiation steps. A successfully rearranged TCRβ chain associates with CD3 chains to form a pre-TCR. The expression of a pre-TCR drives DN4 differentiation into double-positive (DP) thymocytes—the most abundant population in the thymus—expressing both CD4 and CD8 surface markers. During this stage of development, the thymocytes re-express the Rag genes, which allows multiple rounds of TCRα gene rearrangements to increase the probability of forming a functional αβ TCR. DP thymocytes undergo a very strict selection process, such that those that express a TCR which is not able to interact with self-major histocompatibility complex (MHC)/self-peptide complexes die due to neglect. In the same way, the DP thymocytes that bind self-MHC/self-peptide molecules with a high affinity are eliminated by negative selection. Meanwhile, those DP thymocytes expressing TCRs that bind self-MHC/self-peptide ligands with a low affinity are positively selected and differentiated into either CD4<sup>+</sup> or CD8<sup>+</sup> single-positive (SP) thymocytes [12]. At this stage of development, some CD4<sup>+</sup> thymocytes express the transcription factor Forkhead box protein 3 (FoxP3), which confers the cells an immunosuppressive function (Treg) [13]. All kinds of T cells generated in the thymus will seed the peripheral lymphoid tissues (Figure 1).

**Figure 1.** Schematic representation of T cell development depicting the specific stages in which PARP-1 and/or PARP-2 are playing a role. TCR, T cell receptor; DN, double negative; DP, double positive; SP, single positive; DC, dendritic cells; Treg, regulatory T cells.

Although both PARP-1 and PARP-2 proteins are expressed in thymocytes [14], only PARP-2 plays a significant role in thymocyte development. Therefore, PARP-2-deficient, but not PARP-1-deficient, mice show a significant reduction in the number of DP thymocytes. This phenotype is associated with the role of PARP-2 in preventing the accumulation of DNA double-strand breaks (DSBs) and the resulting activation of a DNA damage-induced apoptotic response during TCRα rearrangements [14,15]. In fact, p53 deficiency restores thymocyte populations in PARP-2-deficient mice [15]. In contrast, PARP-1 regulates tTreg development [16], while PARP-2 does not seem to play any role in tTreg development [17] (Figure 1).

Once the T cells in the thymus have matured, they migrate to the peripheral lymphoid tissues forming the naïve T cell pool, where they continue their differentiation to become fully immunocompetent to mount appropriate immune T cell responses to antigen challenge [18]. Naïve T cells proliferate both in situations of lymphopenia (homeostatic proliferation) driven by TCR/self-peptide–MHC interactions, and in response to antigen challenges driven by TCR/foreign-peptide–MHC interactions and co-stimulation, accompanied by differentiation into effector T cells (Th1, Th2, Th17, Treg, and cytotoxic T cells) and the final generation of a memory T cell population [19] (Figure 1). The control of T cell homeostasis is not only mediated by MHC–TCR interactions and cytokine-mediated signals, but also processes which regulate essential T cell functions to maintain genomic stability, such as cell-cycle checkpoints, DNA repair, and apoptosis [20,21].

Although PARP-1 deficiency or PARP-2 deficiency alone does not affect the number of T cells in peripheral lymphoid tissues [14,17], double deficiency in the T cell compartment results in a significant decrease in both CD4<sup>+</sup> and CD8<sup>+</sup> peripheral T cells [17]. The T cell lymphopenia present in mice with double PARP-1 and PARP-2 deficiency indicates that these proteins act in a coordinated manner to prevent the accumulation of unrepaired DNA breaks upon homeostatic proliferation or in response to antigen challenge, but not under basal conditions, avoiding T cell death [17]. PARP-1 and PARP-2 likely act through the principle of synthetic lethality [22], whereby they regulate two independent, but functionally linked, processes. T cell lymphopenia in double-deficient mice for PARP-1 and PARP-2 blunts the anti-viral immune response and the response to other T cell-dependent antigens [17]. Furthermore, double deficiency of PARP-1 and PARP-2 in the T cell compartment in mice affects the T cell response to tumors [23]. Although PARP inhibitors do not achieve the persistence of inhibition as obtained in T lymphocytes with double genetic deficiency of PARP-1 and PARP-2, these pharmacological inhibitors can still impact the T cell compartment and thus the T cell immune response. Indeed, in a mouse breast tumor model induced by the AT-3 cell line, which is sensitive to the PARP inhibitor olaparib, the anti-tumor effect of olaparib is blunted by an intact immune system [23]. As such, it would be interesting to study how PARP inhibitors used in the clinic affect the immune compartment in patients.

Transcriptional activation via different signaling pathways is fundamental to the differentiation of T cells. Among the key transcription factors in T cell development and function are the nuclear factor of activated T cells (NFAT) family of transcription factors (NFAT1 to NFAT5) [24]. After antigenic recognition by the TCR, a signaling cascade is initiated in the T cell, leading to the activation and nuclear translocation of NFAT1, NFAT2, and NFAT4, where, in combination with other transcription factors such as AP1, they regulate the expression of cytokines and lineage-specific transcription factors to control pathways of T cell differentiation into Th1 or Th2 types [24]. Of note, our group has demonstrated that PARP-1 is activated during T cell activation, where it modulates the activity of NFAT through PARylation, as evidenced by PARP inhibitors causing an increase in NFAT-dependent transactivation [25]. Moreover, PARP-1 plays a critical role in the gene expression reprogramming that takes place in T cells upon activation [26]. Indeed, PARP-1-deficiency seems to bias the T cell response to a Th1 phenotype [26] and has been shown to reduce differentiation into Th2 cells in different experimental models [26,27]. The pharmacological inhibition of PARP has led to more controversial results, where, in one case, the PARP inhibitor led to an increase in Th1 cytokine production and a reduction in Th2 cytokines [28], while in another case, the inhibition led to a decrease in Th1 cells [29]. These discrepancies may be associated with the type of inhibitor used or the experimental model.

PARP-1 also plays a role in the generation of Treg cells in the periphery (pTreg) from CD4<sup>+</sup> T cells that express FoxP3, and PARP-1-deficient mice have been found to display an increased number of Treg cells [16]. Moreover, PARP-1 negatively regulates the suppressive function of Treg cells at the posttranslational level through FoxP3 PARylation [30]. PARP-1 can also regulate the generation of Treg cells through its role in regulating the expression of transforming growth factor β receptors (TGFβR) in CD4<sup>+</sup> T cells, and therefore affects TGFβ signaling in T cells [30]. Interestingly, the inhibition of TGFβRI expression by PARP-1 is dependent on PARP enzymatic activity, while the inhibition of TGFβRII expression depends on the interaction of PARP-1 with the promoter of the TGFβRII gene [31]. In contrast, the function of PARP-2 in transcriptional regulation in T cells remains unclear.

#### **3. Impact of PARP-1 and PARP-2 on B cell Development and Function**

As with the development of T lymphocytes, the development of B cells, which takes place in the bone marrow, is also a precisely regulated process that starts from pluripotent hematopoietic stem cells. In the first step, the hematopoietic stem cells differentiate into pro-B cells that transiently express the Rag genes, which mediate immunoglobulin (Ig) heavy-chain gene rearrangements to assemble a VHDJH-Cμ protein. The association of this protein with Vpre-B and λ5 surrogate light chain proteins leads to the formation of the pre-B cell receptor (BCR) complex in the large pre-B cell population. Pre-BCR signaling results in differentiation into small pre-B cells, which re-express the Rag genes, allowing Ig light chain VLJL gene rearrangement. The succeeding association of VLJL with VHDJH-Cμ generates fully functional membrane-bound IgM receptors in immature B cells, which further differentiate into transitional B cells which co-express IgM and IgD receptors on their surface [32] (Figure 2). Despite the role of PARP-1 and PARP-2 in DNA repair, their role in Ig V(D)J gene recombination has remained unclear or unknown [33–35]. Recent data obtained by our group show that mice with dual, but not

individual, PARP-1 and PARP-2 deficiency exhibit a reduced number of B cells in the bone marrow [36] (Figure 2). A possible explanation for this bone marrow hypocellularity is that the V(D)J recombination process is defective in these cells. However, a detailed analysis showed that neither single nor double PARP-1/PARP-2 deficiency affected Ig V(D)J gene recombination [36]. As in the T cell compartment, B cell lymphopenia in dually PARP-1- and PARP-2-deficient mice is associated with an accumulation of unrepaired DNA damage in proliferating B cells leading to cell death, suggesting a potential model whereby coordinated signals from PARP-1 and PARP-2 are required to maintain genomic integrity during lymphoid proliferation. This is consistent with recent data showing that dual PARP-1 and PARP-2 deficiency results in the accumulation of replication-associated DNA damage due to the impaired stabilization of Rad51 at damaged DNA replication forks and uncontrolled DNA resection thereafter [37].

**Figure 2.** Schematic representation of B cell development depicting the specific stages in which PARP-1 and/or PARP-2 play a role. MZ, marginal zone; FO, follicular B cells; TD, T cell-dependent antigen; TI, T cell-independent antigen; SHM, somatic hypermutation.

The differentiation of transitional B cells leaving the bone marrow continues in the peripheral lymphoid organs, which gives rise to mature marginal zone (MZ) or mature follicular (FO) B cells [32]. After their interaction with antigens, mature B cells will either proliferate and differentiate extra follicularly (Bcl-6−) or form germinal centres (Bcl-6+), ultimately generating short-lived antibody secreting plasmablasts or long-lived memory B cells and antibody-secreting plasma cells, respectively [38–40] (Figure 2). Antibody diversity is in part achieved during these responses through DNA editing via both Ig class-switching recombination (CSR) and somatic hypermutation (SHM), which are mediated by activation-induced cytidine deaminase (AID) [41].

While single PARP-1 or PARP-2 deficiency does not affect the B cell compartment in peripheral lymphoid tissues [14,17], combined PARP-1 and PARP-2 deficiency impairs peripheral B cell homeostasis [36]. This lymphopenia does not affect all B cell populations equally, such that only the number of FO B cells is dramatically reduced in mice with double PARP-1 and PARP-2 deficiency, while the number of MZ B cells is not affected. The reason for this is unclear and requires further exploration [36] (Figure 2). Interestingly, double PARP-1 and PARP-2 deficiency impairs antibody responses to T cell-independent, but not T cell-dependent, antigens [36]. In addition, T cell-independent antigens elicit IgG1- and IgG2b-predominant antibody responses in single PARP-1-deficient mice [36,42] (Figure 2). It is important to note that despite their role in DNA repair, neither PARP-1 nor PARP-2 are required for CSR [36,43] and instead, their role in B cell homeostasis underpins their importance for Ig responses to specific antigens. Another role of the PARPs in B cell development is the role of PARP-1 activation in switching off Bcl6 [44]—a transcription factor essential for the formation of germinal centers [45,46]. However, germinal center formation upon immunization is normal in mice with single or dual deficiencies of PARP-1 and PARP-2 [36]. Meanwhile, the role of PARP-1 in SHM is controversial, with some data showing a dispensable role [47], while other data indicate a role of PARP-1 in SHM [48]. Meanwhile, the role of PARP-2 in SHM is unknown.

#### **4. Role of PARP-1 and PARP-2 in the Cellular Components of the Innate Immune System**

In addition to their role during the development and function of cellular components of the adaptive immune system, PARP-1 and PARP-2 have also been involved in different functional aspects of cells involved in the innate immune response, including neutrophils, macrophages, dendritic cells, and natural killer (NK) cells. These innate immune cells serve as the front line of host protection to infection and non-infectious tissue damage. In addition, cells of innate immunity are critical for stimulating subsequent adaptive immune responses [49].

Neutrophils are key players in acute and chronic inflammatory responses through their role in phagocytosis, the recruitment of other immune cells, and the secretion of antibacterial proteins [50]. In cancer, tumor-associated neutrophils are thought to contribute to inflammation in the tumor [51]. Of note, PARP-1 is important in the recruitment and function of neutrophils in different processes related to inflammation [52–55]. Meanwhile, the role of PARP-2 in neutrophil biology remains elusive (Figure 3).

**Figure 3.** Schematic representation of the role played by PARP-1 in cells of the innate immune system.

Macrophages are differentiated from circulating monocytes after extravasation into tissues. Upon differentiation, macrophages are prepared to sense and respond to infection and tissue injuries through the phagocytosis of dead cells, debris, and foreign materials [56]. Besides phagocytosis, macrophages are also important as antigen-presenting cells (APC) to T cells [57]. Macrophages show considerable plasticity, which permits them to adapt their phenotype in response to different microenvironments. There are two major forms of activated macrophages, termed pro-inflammatory M1, which is characterized by the production of pro-inflammatory cytokines, and anti-inflammatory M2, which is characterized by the secretion of anti-inflammatory cytokines [58]. Of note, PARP inhibitors inhibit the expression of LPS-induced proinflammatory cytokines like tumor necrosis factor α (TNFα), interleukin-1 (IL-1), and IL-6 by macrophages [59]. Meanwhile, recent work has shown that PARP-1, but not its enzymatic activity, enhances the transcriptional activity of LPS-induced proinflammatory genes in macrophages [60]. This effect would be mediated by the modulatory role of PARP-1 on the transcription factor NF-κB [61]. In addition, functional interplay between PARP-1 and lysine-specific histone demethylase 1A (LSD1) protects pro-inflammatory M1 macrophages from death under oxidative conditions [62]. Moreover, macrophage recruitment in an airway inflammatory model was severely blocked in PARP-1-deficient mice [63]. Meanwhile, the role of PARP-2 in macrophages remains unknown (Figure 3).

Dendritic cells (DC) are specialized APC which process antigen and present it in the context of self-MHC molecules to T cells. In addition, they also upregulate cell surface receptors, including CD80, CD86, and CD40, which interact with co-receptors on the T cells surface (CD40L and CD28), in order to induce proper T cell activation [64,65]. While the role of PARP-1 in the recruitment of DC to tissues in different pathological situations seems to be well-established, its role in the differentiation and function of these cells is less clear [66–69]. On the other hand, the function of PARP-2 in DC remains unexplored (Figure 3).

NK cells have a wide array of inhibitory and stimulatory receptors on their cell surface that are used for immune surveillance. Upon activation, NK cells show potent cytolytic activity in response to infected or transformed cells by releasing cytotoxic perforin and granzyme and activating apoptotic pathways in target cells through the production of TNFα or via direct cell–cell contact through activation of the tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and Fas ligand (FASL) pathways [70,71]. Recent work has demonstrated important roles of PARP-1 in NK cell biology. For instance, PARP-1 controls NK cell recruitment to the site of viral infection [72,73]. In addition, PARP-1 is involved in the downregulation of NK cell-activating receptor ligands for immune evasion in acute myeloid leukemia [74] (Figure 3).

#### **5. How Could the Immunomodulatory Roles of PARP-1 and PARP-2 Impact the Immune Response to Tumors?**

Tumors contain not only cancer cells, but other cell types, including tissue-resident and peripherally-recruited immune cells, fibroblasts, and endothelial cells, which form the tumor microenvironment. Interactions between these cells—cancerous and non-cancerous—are known to either favor or limit tumorigenesis. Indeed, cancer progression is a dynamic process that, based on those interactions, has been divided into three stages: elimination, equilibrium, and escape [75].

During the cancer elimination phase, a competent immune response takes place in which innate and adaptive immune cells are recruited to the tumor microenvironment, where they exert a strong anti-tumor response [51] (Figure 4). The aforementioned immunomodulatory functions of PARP-1 and PARP-2 would thus be expected to have an impact on the immune response against the tumor. Indeed, we have observed a reduction in tumor growth in PARP-1-deficient host-mice and in PARP-2-deficient host-mice, compared to wild-type specimens, in both a C57 syngeneic tumor model induced by the AT-3 breast tumor cell line [23] and in a Balb/c syngeneic tumor model induced by the LP07 lung adenocarcinoma cell line [76], in which both cancer cells lines are proficient for PARP-1 and PARP-2 proteins. This effect may be associated with their immunomodulatory roles.

**Figure 4.** Schematic representation of the tumor microenvironment indicating the stages in which PARP-1 or its combination with PARP-2 or PARP inhibitors might play a role. In the boxes, we have included further details on the involvement of PARP-1 and/or PARP-2 or PARPi in certain contexts of the immune response. CTL, cytotoxic T cells; DC, dendritic cells; B, B cells; PARPi, PARP inhibitors.

T cells, in particular, CD8<sup>+</sup> cytotoxic T cells (CTL) and CD4<sup>+</sup> Th1 cells, are major contributors to the adaptive host-defense against tumors [10]. Tumor-derived antigens are processed by APCs (mainly dendritic cells), carried to draining lymph nodes and presented to naïve T cells, in order to prime them. Antigen presentation, together with the induction of co-stimulatory signals mediated by the binding of CD28 on the T cell to CD80/CD86 on the APC, leads to the differentiation of naïve CD8<sup>+</sup> T cells into tumor-specific CTLs. These, in turn, migrate to the tumor microenvironment to kill cancer cells through the secretion of perforin and granzyme [77]. The anti-tumor effect of CD4<sup>+</sup> Th1 cells is mediated through the secretion of IL-2, TNFα, and IFNγ, enhancing CD8<sup>+</sup> T cell responses and activating macrophages and NK cells [78–80] (Figure 4). Previous data from our group has indicated that coordinated signals from PARP-1 and PARP-2 are required to maintain T cell homeostasis and for the differentiation from naïve to effector T cells affecting both CD4<sup>+</sup> and CD8<sup>+</sup> lineages [17]. Accordingly, T cell lymphopenia in dual PARP-1/PARP-2-deficient mice can affect the recruitment of lymphocytes to the tumor microenvironment [23]. Moreover, a defect in the ability of dually PARP-1/PARP-2-deficient T cells to differentiate into effector cells could have consequences for the anti-tumor response (Figure 4). On the other hand, PARP-1-deficiency seems to bias T cell responses to a Th1 phenotype [26] that may also impact tumor progression. Similarly, B cell lymphopenia in dual PARP-1/PARP-2-deficient mice can affect the recruitment of B cells to the tumor microenvironment (Figure 4).

The aforementioned biological roles of PARP-1 in macrophage biology may impact the response of these cells to tumors. Classically-activated M1 macrophages can kill many tumor cells by mechanisms including the recognition of damage-associated molecular patterns (DAMPs) from dying tumor cells and the production of nitric oxide. In addition, the co-operation of T cells and macrophages through direct contact or through the secretion of cytokines is important in the anti-tumor response [77]. Recent work from Hottiger´s group shows that mice with a conditional loss of PARP-1 in myeloid lineages fail to control tumor growth in an MC-38-induced tumor model of colon cancer, which could be attributed to reduced Th1 and CD8<sup>+</sup> T cell responses [60], suggesting that PARP-1 in macrophages controls Th1 responses to tumors (Figure 4). However, this intrinsic role of PARP-1 in myeloid cells is independent of its enzymatic activity, so would be of limited utility from the point of view of a pharmacological blockade [60].

Intratumor NK cells have been shown to play a very important role in the control of tumor growth [81]. The balance between stimulatory and inhibitory receptor signals determines the activation of NK cells against tumor cells. NK cells may also be activated to kill tumor cells coated with anti-tumor antibodies by antibody-dependent cell-mediated cytotoxicity (ADCC) (Figure 4). Moreover, the tumoricidal capacity of NK cells is increased by cytokines (IL-2, IL-15, IL-12). Although the role of PARP-1 and PARP-2 in modulating NK cell activity against tumors is largely unknown, the aforementioned function of PARP-1 in controlling NK cell recruitment to the site of viral infection [72,73] and its role in the downregulation of NK cell-activating receptor ligands to evade immune surveillance in acute myeloid leukemia [74], may impact tumor progression.

Tumor cells that have escaped the immune response undergo different strategies in order to acquire immune tolerance, including (i) tumor cell-intrinsic modifications, like the loss of human leukocyte antigens (HLA) class I molecules, loss of tumor-associated antigens, and increased resistance to cell killing by immune cells, and (ii) the generation of an immuno-suppressive microenvironment through the recruitment of cells with immunosuppressive activities (Treg, macrophages M2, and myeloid-derived suppressive cells); the expression of inhibitory checkpoint molecules such as PD-L1, PD-L2, and cytotoxic T lymphocyte-associated antigen 4 (CTLA4) [82–84]; the deprivation of nutrients and oxygen; and the secretion of immunosuppressive cytokines (TGFβ, IL-10, and VEGF) [85] (Figure 4).

The previously mentioned roles of PARP-1 in Treg development and function [16,30] may impact the response to tumors. In addition, PARP-1 inhibition leads to the up-regulation of TGFβ receptor expression in CD4<sup>+</sup> T cells that subsequently affects TGFβ signal transduction [31], which may impact the response to tumors (Figure 4).

However, these tumor microenvironment escape mechanisms can be modified by different strategies in order to reactivate the immune response against tumor cells [86]. Indeed, re-activating the normal function of immune cells in the tumor microenvironment is one of the biggest challenges in oncology research. Accordingly, emerging immunotherapeutic strategies aim to reverse immune tolerance either by modulating T cell co-receptor signals or boosting the recognition of tumor-associated antigens by using monoclonal antibodies [10]. In addition to those strategies based on biological approaches, modifying the immune response through a small-molecule approach targeting intracellular signaling pathways, such as with PARP inhibitors, may represent a breakthrough that is complementary to, and potentially synergistic with, immunotherapy [87].

#### **6. PARP Inhibitors as Immunomodulatory Agents**

PARP proteins exert their function through their physical association with or by the PARylation of partner proteins [3]. Although most of the immunomodulatory roles of PARP proteins have been based on studies of mice with the genetic deletion of these proteins (Table 1), PARP inhibitors might induce similar immune cell alterations that will modify their interaction with tumor cells. Indeed, recent work has shown how PARP inhibitors might impact the mechanisms used by tumors to evade immunity, although many of these studies are focused on tumor cell-intrinsic mechanisms. These studies can provide information to rationalize the combined use of PARP inhibitors with other strategies aimed at reactivating the immune system against the tumor.


*Cancers* **2020**, *12*, 392

One of the most successful strategies for reinstating an existing anti-cancer T cell immune response is the use of blocking antibodies against cell surface inhibitory co-receptors like cytotoxic T lymphocyte-associated protein 4 (CTLA4) and programmed cell death 1 (PD-1), which block the engagement of PD-1 or CTLA4 with their ligand (PD-L1 and PD-L2 for PD1; CD80/CD86 for CTLA4), thus avoiding the initiation of signaling pathways leading to the suppression of T cell activation. Of note, PARP inhibitors upregulate the expression of PD-L1 in cancer cells and enhance cancer-associated immunosuppression (Figure 4). This immunosuppression is reversible by blocking the PD-1/PD-L1 interaction [88]. This study established the rationale of combining PARP inhibitors with checkpoint blockade agents [7,89–91] or agents that alter PD-L1 expression [92], which has led to numerous clinical trials (Table 2). Although the result of an early phase II clinical trial combining Durvalumab with Olaparib in patients with relapsed small cell lung cancer did not meet the preset bar for efficacy [93], we are awaiting the results of ongoing clinical trials to better judge their effectiveness. Moreover, the PARP inhibitor Niraparib has been shown to enhance type I interferon signaling and T cell infiltration in the tumor and improve the therapeutic effect of anti-PD-1 [94].


**Table 2.** Clinical trials with PARP inhibitors in combination with check-point blockade agents (www.clinicaltrials.gov).


NK cells can kill cancer cells by inducing death receptor-mediated apoptosis through the expression of FasL or TRAIL [71]. PARP inhibitors have been shown to sensitize cancer cells to death receptor-mediated apoptosis by upregulating death receptor surface expression [95,96] (Figure 4). In addition, the inhibition of PARP-1 upheld the capacity of NK cells to kill myeloid leukemic cells, and restored the proliferation and cytokine production of NK cells and cytotoxic T cells [97].

Recent work has revealed the intriguing link between genomic instability and the accumulation of DNA in the cytoplasm, which triggers the activation of innate immune responses through the cyclic GMP-AMP synthase (cGAS)/stimulator of interferon genes (STING) pathway that evolved to signal the presence of exogenous DNA [98]. Accordingly, it has been shown that PARP inhibitors promote the accumulation of cytosolic DNA, which activates the DNA-sensing cGAS–STING pathway and stimulates type I interferon (IFNs) gene expression to induce anti-tumor immunity independent of the BRCA status, providing a rationale for using PARP inhibitors as immunomodulatory agents [99–101]. Moreover, treatment with PARP inhibitors stimulates the type I IFN response in cells and tumors lacking BRCA2 [102]. Furthermore, PARP inhibition seems to augment cytotoxic T cell tumor infiltration through activation of the cGAS/STING innate immune pathway, leading to increased levels of chemokines, such as CXCL10 and CCL5, that induce the activation and function of cytotoxic CD8<sup>+</sup> T cells [103,104]. The effect of PARP inhibition-induced T cell recruitment to tumors is more noticeable in homologous recombination-deficient compared with homologous recombination-proficient triple negative breast cancer (TNBC) cells [104].

#### **7. Conclusions and Future Prospects**

The promise of PARP inhibitors in cancer therapy was initially based on proposed effects on genomic integrity in the cancer cell itself. Since then, it has been uncovered that PARPs play additional roles in other important aspects of cellular biology which could be of significance for both tumor physiology and its microenvironment. Here, we can see that the immunomodulatory roles of PARP-1 and PARP-2 are complex, with specific and overlapping roles which vary by cellular compartment and context. Future work will be needed to consider how this effect of PARP inhibition on the tumor microenvironment differs by tumor type, grade, and stage. PARP inhibition may serve as an important adjuvant to immunotherapeutic strategies or indeed benefit from the checkpoint blockade itself, but will require further elucidation of the precise mechanism by which it interacts with immune pathways.

**Author Contributions:** J.Y., L.M.-L., J.J., and S.O.A. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by The Fundación Científica de la Asociación Española Contra el Cáncer (AECC), grant number PROYEI6018YÉLA, and the Spanish Ministerio de Economía, Industria y Competitividad, grant number SAF2017-83565-R.

**Acknowledgments:** The authors thank C. Ampurdanés, A. Ali., and N. Lutfi. for their helpful discussions.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **PARylation During Transcription: Insights into the Fine-Tuning Mechanism and Regulation**

#### **Zoltán G. Páhi, Barbara N. Borsos, Vasiliki Pantazi, Zsuzsanna Ujfaludi and Tibor Pankotai \***

Department of Oral Biology and Experimental Dental Research, Faculty of Dentistry, University of Szeged, Tisza Lajos krt 83, H-6722 Szeged, Hungary; pahizol@gmail.com (Z.G.P.); borsos.barbara.nikolett@gmail.com (B.N.B.); vasopantazi@outlook.com (V.P.); odianya@bio.u-szeged.hu (Z.U.)

**\*** Correspondence: pankotai@bio.u-szeged.hu or pankotai.tibor@stoma.szote.u-szeged.hu

Received: 5 December 2019; Accepted: 9 January 2020; Published: 11 January 2020

**Abstract:** Transcription is a multistep, tightly regulated process. During transcription initiation, promoter recognition and pre-initiation complex (PIC) formation take place, in which dynamic recruitment or exchange of transcription activators occur. The precise coordination of the recruitment and removal of transcription factors, as well as chromatin structural changes, are mediated by post-translational modifications (PTMs). Poly(ADP-ribose) polymerases (PARPs) are key players in this process, since they can modulate DNA-binding activities of specific transcription factors through poly-ADP-ribosylation (PARylation). PARylation can regulate the transcription at three different levels: (1) by directly affecting the recruitment of specific transcription factors, (2) by triggering chromatin structural changes during initiation and as a response to cellular stresses, or (3) by post-transcriptionally modulating the stability and degradation of specific mRNAs. In this review, we principally focus on these steps and summarise the recent findings, demonstrating the mechanisms through which PARylation plays a potential regulatory role during transcription and DNA repair.

**Keywords:** transcription; PARylation; PARP; DNA damage; transcription silencing

#### **1. Introduction**

#### *The Mechanism of Poly-ADP-Ribosylation and the PARP Superfamily*

PARylation is a reversible post-translational modification (PTM), in which writers, such as poly(ADP-ribose) polymerases (PARPs) as well as erasers, including poly(ADP-ribose) glycohydrolases (PARGs) and ADP-ribosyl hydrolase 3 (ARH3) are involved [1–7]. PARPs are NAD+-dependent enzymes and thus require a source of NAD<sup>+</sup> which is provided by nicotinamide mononucleotide adenylyl transferases (NMNATs) [8]. ADP-ribosylation is a multistep process, involving initiation, elongation, branching, and the release of PAR units. First, PARP binds to nicotinamide adenine dinucleotide (NAD+) and cleaves the nicotine amid unit, catalysing the transfer of the ADP-ribose moieties to the acceptor protein [9]. During initiation, the first ADP-ribose monomer can be covalently linked to Lys, Arg, Glu, Asp, Cys, Ser, or Thr amino acid residues of the acceptor protein [9]. During the branch formation step, 2'-1' ribose-ribose bonds are generated between ADP-ribose units.

In human cells, PARPs are classified based on their enzymatic activity: PARP1, PARP2, PARP5a, and PARP5b catalyse PAR chain formation, while PARPs 3, 4, 6–8, 10–12, and 14–16 have been described as mono-ADP-ribosyl transferases (MARTs) [9]. PARP5a and PARP5b share a high level of similarity and are also called tankyrase 1 and tankyrase 2, respectively, due to their ankyrin repeat region and the sterile alpha motif [10]. Since PARylation is a reversible process, the covalently-attached PAR can be removed by PARGs and ARH3 as well, keeping the PAR levels in the cell under control. While PARG

can efficiently cleave the PAR O-glycosidic bond, ARH3 is mainly responsible for the hydrolysis of protein-free PAR [11].

#### **2. PARylation in Transcription Regulation**

#### *2.1. The Major Regulatory Steps of Transcription Activation*

RNA synthesis requires a well-coordinated regulation of transcription at different levels. As a first step of initiation, transcription factor II (TFII)D-TFIIA-TFIIB binds to the promoter region; then TFIIF, along with RNA polymerase II (RNAPII), is recruited, resulting in the stabilisation of the pre-initiation complex (PIC) [12]. Next, TFIIE and TFIIH join to the core PIC, contributing to its association with the mediator complex [13]. Xeroderma pigmentosum type B (XPB), one of the subunits of TFIIH, induces DNA unwinding around the transcription start site (TSS) and initiates the formation of the transcription bubble [14]. Subsequently, the cyclin-dependent kinase 7 (CDK7) subunit of TFIIH phosphorylates the C-terminal domain (CTD) of RNAPII at Ser5, which is indispensable for transcription initiation. Following the synthesis of approximately 20–60 base pairs of RNA, RNAPII is stopped, which is the so-called promoter-proximal pausing [15]. During this step, negative elongation factors, including the dephosphorylated form of DRB sensitivity inducing factor (DSIF) and negative elongation factor (NELF), bind to the RNAPII, thereby hindering the elongation [16,17]. The cyclin-dependent kinase 9 (CDK9) subunit of positive transcriptional elongation factor b (P-TEFb) promotes the elongation process by phosphorylating DSIF, NELF, and RNAPII CTD at Ser2 [18,19]. Consequently, the phosphorylated NELF complex dissociates, while DSIF becomes a positive elongation factor responsible for recruiting other factors, such as cyclin-dependent kinase 12 (CDK12), which can also catalyse the phosphorylation of RNAPII CTD at Ser2 and by this, supporting the elongation step [20,21].

#### *2.2. PARP1 Plays a Key Role in the Fine-Tune Regulation of Transcription Initiation*

During transcription initiation, PARP1 can PARylate sequence-specific transcription factors, such as nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), Myb-related protein B (B-MYB), organic cation transporter 1 (OCT1), sex determining region Y-box 2 (SOX2), and Krueppel-like factor 8 (KLF8), as well as oestrogen- and retinoic acid receptors, which can either inhibit or enhance the activity of these factors (Figure 1A) [22–29]. Along with PARP1, PARP7 has been identified as an important player in the transcription regulation of pluripotency genes and in their protection from epigenetic repression [30]. Originally, PARP1 was identified as TFIIC, which is capable of facilitating the initiation steps of mRNA synthesis through direct interaction with the basal transcription machinery [31]. Recently, it has been shown that PARP1 can also serve as a scaffold protein by stimulating the recruitment of coregulator complexes, such as p300, NF-κB, and p50 as well as the mediator complex to promoter regions (Figure 1A) [32]. It has been proven that the DNA-binding, rather than the catalytic activity of PARP1, enhances transcription by promoting the early steps of PIC formation [33]. However, only a limited amount of data is available, which suggests that any member of the basal transcription machinery is PARylated [34]. Moreover, PARP1 facilitates not only the recruitment but also the release of specific transcription co-regulators, leading to dynamic exchange between transcription factors, such as TLE family member 1 (TLE1) transcriptional corepressor complex to histone-acetyltransferase (HAT)-containing complex in neurons (Figure 1A) [35]. Additionally, PARP1 is also necessary for maintaining the relaxed chromatin structure of actively transcribed genes, supporting the active transcription of these genes [36]. In contrast, biochemical studies have revealed that PARP1 is not indispensable for the initiation of transcription on intact DNA templates, but single-stranded DNA breaks (SSBs) promote PARP1 binding, resulting in repression of nick-dependent transcription [31].

**Figure 1.** PARylation in transcription initiation. (**A**) As a consequence of the activation of certain signalling pathways, PARP1 catalyses the transfer of ADP-ribose through its binding to NAD<sup>+</sup> cofactor. PARP1 can PARylate numerous sequence-specific transcription factors, such as NF-κB and KLF8, which can facilitate and attenuate transcription initiation, respectively. PARP1 can also act as a scaffold protein by promoting the recruitment of various co-regulator complexes, such as p300 and the mediator complex to the promoter region (P), leading to transcription initiation catalysed by RNAPII. Moreover, PARP1 is implicated in the release of subsequent corepressor complex, such as transducin-like enhancer protein 1 (TLE1), resulting in its exchange to the PARylated HAT complex. PARP1 also participates in the exchange between negative coregulators, (–CRs) to positive ones (+ CRs). TF = transcription factor binding site. (**B**) (Left part) Following transcription initiation, RNAPII is stopped at around +20–+150 bp from the TSS. The promoter-proximal pausing process is induced by negative elongation factors, such as NELF-E and DSIF. During the release of the initial pausing, CDK9/ PTEFb phosphorylates NELF-E, DSIF, and the CTD of RPB1 (the largest subunit of RNAPII) at Ser2. Phosphorylated NELF-E (P-NELF-E) is released from RNAPII, while phosphorylated DSIF (P-DSIF) acts as a positive elongation factor, and the S2P-RNAPII becomes capable of proceeding the elongation step. Right part: Following DNA damage, PARP1 targets P-NELF-E for PARylation, thereby hindering its DNA-binding ability, leading to transcription silencing.

#### *2.3. PARP1 Mediates Promoter-Proximal Pausing and Transcription Elongation*

Following transcription initiation, RNAPII promoter-proximal pausing can be observed between the +20 and +150 region around the TSS [37]. In the early elongation step, RNAPII activity is temporarily paused by the contribution of NELF-E and DSIF [21]. During the promoter-proximal pause release, P-TEFb phosphorylates DSIF, NELF-E, and the CTD of RPB1 (the largest subunit of RNAPII) at Ser2, making it competent for transcription elongation and preventing its inhibition by DSIF and NELF-E

(Figure 1B) [37,38]. It has been also demonstrated that PARP1 directs RNAPII for PARylation shortly after DNA damage, facilitating the recruitment of NELF-E to RNAPII and by this, NELF-E plays a potential role in transcription silencing at the DNA break sites (Figure 1B) [39].

The *Drosophila* orthologue of PARP1 plays an important regulatory role during the elongation phase of heat shock-induced transcription [23,40]. Two members of the NELF complex, NELF-A and NELF-E, have been shown to be targeted by PARP1 [41]. Moreover, the phosphorylation of NELF-E catalysed by CDK9/P-TEFb, is indispensable for its subsequent PARylation, resulting in the attenuation of its DNA-binding ability (Figure 1B) [41]. Additionally, inhibition of either PARP1 or CDK9/P-TEFb results in reduced phosphorylation of RNAPII CTD at Ser2, thus leading to promoter-proximal pausing. The genome-wide distribution of PARylation is mostly enriched at actively transcribed regions, where high levels of NELF-B, RNAPII and H3K4me3 can be observed [41]. Based on these data, PARylation plays a potential regulatory role throughout the entire transcription process. Nevertheless, further investigations are needed to highlight the proper PARP-mediated regulatory mechanisms.

#### *2.4. PARylation Regulates Transcription Responses during DNA Damage*

Genome integrity is being constantly challenged by various genotoxic stresses, which can result in different types of DNA lesions, including SSBs and double-stranded breaks (DSBs). DNA damage repair requires tight regulation, since inappropriate repair can lead to genome instability and tumourigenesis. In this regard, eukaryotes have evolved various mechanisms addressing DSBs, including homologous recombination (HR) and non-homologous end-joining (NHEJ) [42]. Although different factors are involved in these DNA repair pathways, crosstalk may occur between them. When DSBs arise within an actively transcribed unit, NELF-E and -A are rapidly accumulated at the break sites through a PARP1-dependent manner, leading to the silencing of this transcription unit [39]. Additionally, in vitro PAR-binding assays have also revealed that NELF-E has elevated binding capacity to PAR moieties, and the interaction between PARP1 and NELF-E is weakened following exposure to ionising radiation [41]. Additionally, the broken DNA region is PARylated by PARP1, facilitating the recruitment of NELF-E and silencing of transcription at the break site [39]. Furthermore, at sites of laser-induced DNA damage, PARP1 can also indirectly regulate transcription through its interaction with the TIMELESS protein. It has been also established that the PARP1–TIMELESS complex plays an essential role in HR [43].

The dual inhibition of PARP1 and PARP2 can lead to reduced binding of nucleosome remodelling deacetylase (NuRD) complex to sites of DNA damage [44]. NuRD can facilitate the recruitment of protein kinase C-binding protein 1 (ZMYND8) to the PARylated DNA damage sites [45]. Lysine-specific demethylase 5A (KDM5A) plays a key role in demethylation of H3K4 to regulate the binding of ZMYND8-NuRD complexes to the DSB (Figure 2A,B) [46]. Consequently, PARP1 can regulate transcription silencing not only by recruiting chromatin remodellers but also demethylases, which can remove the methyl groups from H3K4me3, responsible for transcription activation. During DNA damage-induced transcription silencing, PARP1 facilitates the recruitment of the polycomb repressive complex 1 and 2 (PRC1 and PRC2), ensuring the proper chromatin structure and transcription arrest at the damaged sites [44,47]. Moreover, the chromodomain Y like (CDYL) protein interacts with PRC2 in a PARP-dependent manner [48,49]. During DSB-induced transcription silencing, PARP1 promotes the recruitment of KRAB-associated protein-1 (KAP-1), heterochromatin protein 1 (HP1), and suppressor of variegation 3–9 homolog 1 (SUV39H1) (Figure 3A) [50]. SUV39H1 methylates H3K9 around the DSB, leading to the recruitment of additional KAP-1/HP1/SUV39H1 complexes thereby contributing to the spreading of H3K9me<sup>3</sup> signal (Figure 3A,B) [50]. It results in the activation of histone acetyltransferase 5 (KAT5), which acetylates ataxia-telangiectasia mutated (ATM) (Figure 3C) [51,52]. ATM phosphorylates KAP-1, leading to the dissociation of the KAP-1/HP1/SUV39H1 complex from the chromatin, which allows the activation of the ATM-dependent HR repair pathway (Figure 3C,D) [50].

**Figure 2.** PARylation in transcription repression. (**A**,**B**) PARP1 is also involved in transcription silencing by recruiting demethylases, such as KDM5A, catalysing the removal of methyl groups and being responsible for transcription activation. KDM5A promotes the recruitment of NuRD and ZMYND8 to the lesion site by demethylating H3K4me3, thereby contributing to transcription silencing.

By contrast, during NHEJ, the Ku70/Ku80 heterodimers bind to the broken DNA ends and recruit the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), a key player in transcription silencing processes. Additionally, PARP1 has been shown to interact with both DNA-PKcs and Ku70/80 [53]. Furthermore, even a single DSB can lead to DNA-PKcs-dependent transcription silencing, suggesting that PARP1 could also be involved in this pathway. At the end of this process, the HECT E3 ubiquitin ligase, WWP2 directs the stalled RNAPII complex to proteasomal degradation [54].

A recent systematic analysis has demonstrated that the TFIID complex member, TBP-associated factor 15 (TAF15), is bound to laser-induced DNA damage sites in a PAR-dependent manner [55]. Moreover, a proteome-wide mass-spectrometry analysis has revealed that the RNA-binding protein, RNA-binding motif protein X-linked (RBMX), is PARylated following exposure to genotoxic stress [56,57]. The DNA damage-induced appearance of PARylated transcription factors suggests an uncommon and transient transcription bursting, which generates the so-called DDR RNAs (DDRNAs) or DSB-induced RNAs (diRNAs) [58]. Furthermore, PARylation can indirectly facilitate transcription silencing through diRNA-mediated chromatin compaction [58,59].

The diRNAs presumably contribute to the recruitment of DDR factors and chromatin modifiers or participate in transcription silencing at DNA break sites [60]. The presence of DDRNAs at the damage sites is critical for the activation of DDR, since in the case of hindering the formation of double-stranded RNAs by silencing either Drosha or Dicer, tumour suppressor P53-binding protein 1 (53BP1) foci formation is highly reduced. Additionally, it has been also shown that breast cancer type 1 (BRCA1), implicating in HR and competing with 53BP1 and DNA repair protein RAD51 homolog 1 (RAD51) foci formation, is dramatically reduced upon silencing Drosha regardless of the cell cycle phase [61].

This finding might be explained by the fact that in the absence of sister chromatids, the HR pathway can only be activated in the presence of an RNA strand, which is ensured by DNA damage induced *de novo* transcription [62]. On the contrary, the recruitment of DDRNAs is not influenced by 53BP1 but rather it is highly dependent on the presence of RNAPII. This suggests a potential role of RNAPII in DDR activation by synthesising damage-induced long non-coding RNAs (dilncRNAs) at the site of DNA damage. It seems that the broken DNA ends can serve as promoters for the

transcription of dilncRNAs, while simultaneously with this process, the transcription of coding regions is prevented. The transcription of dilncRNA is bidirectional, since it can be initiated towards both directions from the break site. However, DNA:RNA hybrids should be resolved by helicases or RNase H enzymes, since these can prevent further recruitment of the repair factors to the site of DNA damage [63]. Recently, it has been demonstrated that besides the interaction with RNAPII, BRCA1 also contributes to the resolution of DNA:RNA hybrids and preserves genome integrity through the recruitment of DNA/RNA helicase senataxin (SETX) to the terminal regions of genes [64,65]. Since HR is the most accurate DSB repair pathway, the appropriate and controlled recruitment of BRCA1/2 to the damage site is indispensable for the efficient repair. At this point, the inhibition of PARPs can counteract with the ongoing repair processes, resulting in genome instability [66]. Additionally, in BRCA1/2-deficient tumour cells, higher sensitivity to PARP inhibition can be observed [67,68]. These results as well as the preclinical trials highlight the substantial role of PARP inhibitors in cancer therapy [67,69–71]. Additional factors, including ATP-dependent RNA helicase A (DHX9), PARP1, scaffold attachment factor B2 (SAFB2), multiple myeloma SET domain (MMSET) and DNA-PKcs have been recently identified as proteins that interact with RNA/DNA hybrids [72]. Furthermore, DHX9 along with PARP1 plays a remarkable role in preventing R-loop accumulation and facilitating transcription termination [72]. PARP1 contributes to the enhancement of DHX9 helicase activity [73]. DHX9 interacts with a large number of proteins related to transcription, including RNAPII, suggesting that DHX9 travels with the elongating RNAPII and contributes to the resolution of R-loops in a PARP1 dependent manner [72].

**Figure 3.** *Cont*.

**Figure 3.** PARylation in transcription silencing during DNA damage. (**A**) As a response to DNA damage-induced transcription silencing, PARP1 facilitates the recruitment of SUV39H1, KAP1, and HP1. (**B**) Subsequent SUV39H1–HP1–KAP1 containing complexes are recruited, resulting in the spreading of H3K9me<sup>3</sup> signal. (**C**) KAT5 is activated, which acetylates ATM, being responsible for the phosphorylation of KAP-1. (**D**) Phosphorylation of KAP1 contributes to the dissociation of SUV39H1–HP1–KAP1 from the chromatin.

Moreover, it has been shown that topoisomerase I and II play an indispensable role in the RNAPII pause release [74]. Elongating RNAPII induces torsional and topological stresses in the super-helical DNA, which should be resolved by topoisomerase I and II [75–78]. Topoisomerases take part in the activation of DDR during transcription activation and elongation processes [74]. Topoisomerase I is involved in the prevention of R-loop formation, which is one of the major sources of transcription-coupled genome instability, by removing negative supercoiling structures behind the RNAPII [78,79]. Upon transcription blockage, spliceosome displacement could result in R-loop formation, leading to ATM activation in a DSB-independent manner [80,81]. As a consequence of RNA/DNA hybridisation, ssDNA strands are formed, which are more susceptible to different kinds of DNA damage. Additionally, the formation of R-loops can lead to genome instability by interfering with DNA replication [82,83]. In addition to topoisomerases, R-loop formation can also be inhibited by RNase H, RNA/DNA helicases, and suppressors of proteins promoting R-loop formation [84,85]. Although Topoisomerase I is responsible for alleviating the torsional stress in the DNA, it gets trapped and is accumulated in close proximity to the DNA lesions. This leads to failures in DNA repair, genome instability, and tumourigenesis [86]. Such phenomenon occurs during camptothecin (CPT)-induced topoisomerase I inhibition, resulting in the accumulation of antisense RNAPII transcripts and then R-loop formation at actively transcribed regions [87–89]. Following CPT treatment, PARP1 interacts with topoisomerase I in both nucleolar compartments, playing a role in eliminating covalent topoisomerase I–DNA complexes through PARylation and recruiting repair factors to these sites [90,91]. In addition to PARP1, through the PARylation of topoisomerase I, PARP2 also plays a pivotal role in the removal of the stalled enzyme. By this mechanism, both PARP1 and PARP2 have a positive impact on preserving genome stability. Furthermore, PARP enzymes interfere with the actions of CPT, resulting in drug resistance. Therefore, combining PARP inhibitors with CPT may enhance therapeutic efficacy [92].

#### **3. PARylation in the Regulation of DNA Damage-Induced Chromatin Structural Changes**

Several studies suggest that both reduced structural constraints and altered nucleosome occupancy influence the accessibility of chromatin in response to DNA damage. Additionally, as a consequence of persistent DNA damage, the unfolding and spatial expansion of certain chromatin regions can be observed. Following DNA damage, robust chromatin decondensation occurs in a PARP-dependent manner [93,94]. The addition of the highly negatively charged PAR chains to histones and to other chromatin-associated proteins results in an electrostatic repulsion with the negatively charged DNA, leading to chromatin relaxation. Following DNA damage, PARP1 can also initiate chromatin conformational changes through interaction with other chromatin-modifying factors. Therefore, the activity of PARP1 seems to be indispensable for chromatin decondensation, contributing to initiation of DDR signalling and the recruitment of repair factors [55,95]. Following laser-micro irradiation, PARP1 is recruited to the site of DNA damage within seconds, while PARP2 binds only 30 s later. This finding supports that PARP1 is mainly responsible for the transient reorganisation of the chromatin structure [96,97]. Furthermore, kinetic analyses have shown that the binding of PARP1 is necessary for the recruitment of MRE11–RAD50–NBS1 (MRN) to the DSB sites (Figure 4A) [96]. Following DSB recognition, ATM, recruited by the MRN complex, phosphorylates H2A.X at S139 (referred to as γH2A.X), which triggers the recruitment of ring finger protein 8 (RNF8) and ring finger protein 168 (RNF168) ubiquitin ligases, participating in the K63-linked polyubiquitylation of H1 histones and in K13 and K15 ubiquitylation of H2A histones, respectively [98,99]. PARP1 facilitates the recruitment of the SWItch/sucrose non-fermentable (SWI/SNF)-related matrix-associated actin-dependent regulator of chromatin subfamily A member 5 (SMARCA5/SNF2H) to the sites of DNA damage and promotes the interaction between SMARCA5 and ADP-ribosylated RNF168 (Figure 4B) [100]. These results have confirmed that PARylation is a crucial step both for chromatin reorganisation and RNF168-mediated ubiquitylation of H2A, being responsible for the recruitment of additional DDR factors [100]. Hence, a functional link can be recognised between PARylation and ubiquitylation during DNA repair.

Subsequently to the auto-activation of PARP1, amplified in liver cancer 1 (ALC1), which interacts with PARP1 and histones, is recruited via a similar kinetic to PAR and catalyses nucleosome sliding in an ATP-dependent manner [101–103]. This recruitment requires its C-terminal PAR-binding macrodomain, which recognises PARylated PARP1. Recently, it has been shown that the interaction between the ATPase catalytic domain and the C-terminal macrodomain of ALC1 is necessary to keep ALC1 in an inactive state under physiological conditions. However, the activation of PARP1 disrupts the interaction of the aforementioned domains and allows the stimulation of the remodelling and PAR-dependent binding activity of ALC1 [104,105]. Nevertheless, the interaction of ALC1 with histones is dependent on the presence of the histone chaperone aprataxin and PNK-like factor (APLF), which are localised at the site of DNA lesion in a PAR-dependent manner and are PARylated by PARP1 [106,107]. It has been still unclear how the ALC1 contributes to the relaxation of the chromatin structure around the break site; therefore, further investigations are needed to reveal novel chromatin-associated interaction partners of it.

Additionally, a strong interaction has been demonstrated between APLF and macroH2A.1.1 following hydrogen peroxide-induced DNA damage, thus indicating a potential role of APLF in chromatin rearrangement [107]. Furthermore, this histone chaperone recognises branched PAR chains, catalysed by PARP2, and mediates histone H3 removal during DNA repair [97]. With regards to histone H3, the incorporation of the histone variant H3.3 has been linked to the PARP1-mediated accumulation of the chromodomain helicase DNA binding protein 2 (CHD2), triggering chromatin expansion (Figure 4B) [108]. Although CHD2 cannot recognise PAR moieties, earlier PAR-dependent events, including the rapid localisation of ALC1 and relaxation of the chromatin structure can trigger the accumulation of CHD2 near DNA breaks.

Several studies have highlighted that transient chromatin relaxation precedes chromatin compaction for protecting regions around the DNA break, resulting in the reveal of the DSB site that needs to be restored by the contribution of DDR factors [109]. Another hypothesis concerns the inhibition of replication and transcription at DSB-flanking regions in order to prevent interference with the repair machinery and mediate faithful repair. MacroH2A acts as a tumour suppressor and is a significant player in the maintenance of the heterochromatic structure as well as in the inactivation of the X chromosome, during which macroH2A inhibits the enzymatic activity of PARP1 [110–113]. The PAR-binding ability of the macrodomain modules has been recently demonstrated, underlining the PAR-capping of macroH2A.1.1, its capability of sensing PARP1 activation, and the subsequent reorganisation of chromatin structure by establishing a compacted chromatin environment [114]. Furthermore, only macroH2A.1.1 suppresses PARP1 activity, preventing the formation of an open

chromatin structure. By contrast, all three histone variants of macroH2A (macroH2A.1.1, macroH2A.1.2 and macroH2A.2) retain the ability to stabilise condensed chromatin structure via their common linker region [115]. Subsequently, macroH2A.1.1 participates in recognizing and binding PAR chains to inactivate PARP1 (Figure 4C), whereas the linker region, being present in all three isoforms, may play an additional role in chromatin compaction. Together with PARP1 inhibition, the linker region contributes to the stabilisation of this architecture.

**Figure 4.** PARylation is required for chromatin structural changes following DNA damage. (**A**) PARP1 facilitates the binding of the MRN complex, which is known as the first DNA damage sensor, being responsible for the recruitment of ATM. Following PARP1 auto-activation, amplified in liver cancer 1 (ALC1) is implicated in nucleosome sliding by interacting with histones in a PARP- and aprataxin and PNK-like factor (APLF)-dependent manner. (**B**) ATM phosphorylates H2A.X at S139, resulting in the recruitment of RNF8 and RNF168. RNF8 catalyses the K63-linked poly-ubiquitylation of H1, while RNF168 is responsible for the ubiquitylation of H2A at K13 and K15. PARP1 promotes the recruitment of SMARCA5 to DSB sites and its subsequent interaction with RNF168. PARylated APLF, recognizing branched PAR chains catalysed by PARP2, participates in H3 removal during DNA repair. Furthermore, PARP1-mediated accumulation of chromodomain helicase DNA binding protein 2 (CHD2) leads to dynamic exchange of H3 to H3.3. (**C**) Members of the NuRD and polycomb complex, involving metastasis associated protein 1 (MTA1), CHD3, CHD4, polycomb group ring finger 2 (PCGF2), polycomb group ring finger 4 (BMI1), and enhancer of zeste homolog 2 (EZH2) take part in transcription inhibition through a PARP-mediated pathway.

Factors involved in the NuRD complex, along with members of the Polycomb complex, play an indispensable role in the PARP1-associated DDR [44,55,116]. Particularly, metastasis associated protein 1 (MTA1), chromodomain helicase 3 and 4 (CHD3, CHD4), and all members of the NuRD complex are rapidly PARylated and exhibit enhanced binding to the site of the DSB. At laser-micro-irradiated sites, PARP1-mediated complete loss of transcription can be observed [44]. While Polo and colleagues revealed similar recruitment kinetics for CHD4 and PAR, other groups demonstrated that CHD3 and CHD4 require the initial PAR-dependent chromatin relaxation, potentially mediated by ALC1. Consequently, their accumulation appears at a later time point [116]. Recruitment of polycomb group ring finger 2 (PCGF2), polycomb group ring finger 4 (BMI1), and components of the polycomb repressive complex 1 (PRC1) are largely abrogated upon inhibition of PARP1. Interestingly, enhancer of zeste homolog 2 (EZH2), belonging to the polycomb repressive complex 2 (PRC2) and associated with transcription repression (H3K27me3 formation), is recruited depending on the enzymatic activity of PARP1 (Figure 4C) [44,47]. Recent findings have indicated that PARP1 inhibits the histone methyltransferase activity of EZH2, thus resulting in a more relaxed chromatin structure at the damaged regions [117,118]. Nevertheless, these opposing observations may arise from different kinds of DNA damage source, inducing different repair pathways or even from diverse phases of the repair process.

#### **4. Role of PARP1 in RNA Metabolism**

In addition to transcription, the abundance and decay of mRNAs can be post-transcriptionally regulated by controlling splicing, polyadenylation and nuclear export. A recent study has demonstrated that PARP1 is recruited to specific nucleosomes localised at the exon/intron boundaries, corresponding to specific splice sites. Additionally, PARP1 inhibition results in changes in the alternative splicing. Emerging data have indicated that PARP1 stimulates the recruitment of U2 snRNPs (small nuclear ribonucleoproteins), therefore positively influencing the exon recognition and the further splicing procedure [119]. Moreover, heterogeneous nuclear riboproteins (hnRNPs) can tightly bind to PAR chains, promoting the dissociation of hnRNPs from RNA and subsequent intron splicing [120]. So far, 11 human hnRNP proteins have been demonstrated to be capable of recognising and binding to PARylated targets. Affinity-purification mass spectrometry assays (AP-MS) combined with Gene Ontology classifications (GO) have shown that not only PARP1, but also PARP2 plays a role in intron splicing and interacts with hnRNPs [121]. It has been also shown that PARP10 is able to PARylate specific RNA substrates, leading to the protection of RNA ends which can act as a platform for recruiting other proteins [122]. Following stress responses, auto-PARylated PARP1 can bind to several nuclear proteins and initiate their transport to Cajal bodies, contributing to the regulation of either the assembly or disassembly of transcription- and splicing-related complexes [123]. Additionally, genome-wide data suggest that PARP1 can oppositely influence the transcription elongation by altering the elongation speed [124]. PARP1 also affects the assembly of human pre-mRNA 3 -processing complex. During transcription termination, PARylation is involved in hindering polyadenylation by catalysing the ADP-ribosylation of polyadenylate-polymerase (PAP), resulting in its reduced binding to mRNA transcripts [125]. Furthermore, PARP14 seems to be involved in the posttranscriptional regulation of mRNA stability since it can promote the degradation of specific transcripts by interacting selectively with tristetrapolin (TTP) [126]. Finally, PARylation could affect transcription by regulating the transport of specific mRNAs, since the nuclear export of mature mRNAs is also regulated by PARP1. In lipopolysaccharide-treated cells, PARP1-dependent PARylation of embryonic lethal abnormal vision-like 1 (ELAV-like protein 1) triggers the RNA nuclear-cytoplasmic shuttling, leading to the enhanced stability of mRNA [127]. These data suggest that following stress responses, PARP enzymes could affect the mRNA maturation at multiple levels. PARylation may be involved in most of the RNA metabolism-related processes, such as splicing, polyadenylation and mRNA maturation. Nonetheless, further investigations are needed to the more precise understanding of these regulatory pathways.

#### **5. Discussion**

In addition to DNA damage response, PARylation regulates various processes, such as chromatin remodelling, transcription activation and repression, ubiquitylation, RNA metabolism as well as cellular stress responses. PARylation can have the following effects on these processes: (1) it can ensure a surface for protein interactions, and (2) certain proteins, such as Imitation SWI (ISWI) and ALC1, possessing a PAR-binding domain (consisting of a PAR-binding zinc finger (PBZ), a PAR-binding motif (PBM), and a WWE), can be recruited to DNA through this process. Additionally, PARP1 can act as a scaffold protein by regulating the recruitment of transcription co-regulator complexes, such as p300 or the mediator complex. These results also support that PARylation is implicated in various mechanisms by promoting complex assembly [32]. On the other hand, during DNA repair, PARPs have an indispensable function in identifying DNA breaks and participating in DNA repair pathway choice. In the absence of PARylation, the insufficient activation of proteins involved in DNA repair can result in the malfunction of the repair mechanism. Furthermore, inappropriate activation of any DNA repair pathway can contribute to genome instability, leading to tumourigenesis. For instance, in HR-deficient tumour cells, in the absence of BRCA1/BRCA2, PARP1 inhibition has been shown to have cytotoxic side-effects. The putative mechanism of this hypersensitivity can be explained by the fact that PARP inhibitors can disturb the recruitment of BRCA1 to the damaged sites, resulting in inadequate HR activation. Although PARP1 is a well-characterised protein, the precise function of other PARPs in regulating other cellular processes has remained unclear. PARP inhibitors may alleviate the speed of DNA repair, leading to the collapse of the replication fork and high therapeutic efficacy during tumour therapy [71]. Moreover, PARylation can interfere with the early recruitment of both BRCA1 and BRCA2, contributing to HR deficiency [128–131]. Interestingly, during HR, PARPs not only recruit MRE11 and NBS1 to the damage sites, but also play a regulatory role during transcription. In HR-related de novo transcription, DHX9 interacts with PARP1, thereby regulating the transcription [72,73]. Moreover, BRCA1-RNAPII interaction contributes to the resolution of DNA:RNA hybrids [64,65]. These results highlight that PARPs can act as transcription regulators in various processes, which can reveal new possibilities in applying PARP inhibitors in clinical trials [132,133]. Although several clinical reports have already demonstrated that PARP inhibitors could be beneficial during tumour therapy, we have to mention that the exact biochemical mechanisms regulated by these PARP inhibitors still remained unexplored. Therefore, further investigations are required to uncover these PARylation-mediated mechanisms to reduce the off-target effects of PARP inhibitors.

#### **6. Conclusions**

In this review, we address the role of PARylation to understand their function during the transcription-coupled cellular responses. We summarise the possible canonical mechanisms by which PARylation exerts its regulatory roles during the transcription responses. However, despite the increasing knowledge on the related topic, the extent of the contribution of PARylation needs to be elucidated. Depending on the cellular context of this PTM, it can exert opposite effects on the same cellular processes. During transcription, it was shown that PARylation can determine the transcription state either by activating or inhibiting the transcription of different sets of genes, leading to distinct biological outcomes. According to this, it is worthwhile performing studies able to address the effect of PARP inhibition on transcription responses. Further efforts should be initiated for the better understanding of the underlying mechanism of actions to achieve more effective therapeutic benefits with minimal side-effects.

**Author Contributions:** Wrote the paper: Z.G.P., B.N.B., V.P., Z.U., T.P. All authors have read and agreed to the published version of the manuscript

**Funding:** This work was supported by the National Research, Development and Innovation Office grants GINOP-2.3.2-15-2016-00020, GINOP-2.3.2-15-2016-00036, GINOP-2.2.1-15-2017-00052, and NKFI-FK 132080, and the Tempus Foundation.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### MDPI

St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Cancers* Editorial Office E-mail: cancers@mdpi.com www.mdpi.com/journal/cancers

Academic Open Access Publishing

www.mdpi.com ISBN 978-3-0365-8157-6