PARP1 inhibition elicits immune responses against non-small cell lung cancer.

High levels of intracellular poly(ADP-ribose) (PAR) resulting from an elevated activity of PAR polymerase-1 (PARP1) correlate with poor infiltration of non-small cell lung cancers by cytotoxic T lymphocytes and dismal patient prognosis. Preclinical experimentation now demonstrates that PARP1 inhibition in cancer cells mediates strong immunostimulatory effects.

Poly(ADP-ribose) (PAR) is a polymer that is synthesized by the enzyme PAR polymerase-1 (PARP1), which is activated by DNA damage. PAR is covalently attached to chromatin-binding proteins and facilitates the recruitment of repair enzymes to the damaged double strain.[1,2] For this reason, upregulation of PARP1 may reduce the susceptibility of cancer cells to DNA-damaging chemotherapeutics such as cisplatin, while its inhibition sensitizes to cisplatin.[3-5] The mRNA levels transcribed from the PARP1 gene are more elevated in non-small cell lung cancers than in adjacent normal tissues, perhaps reflecting an intrinsic resistance of malignant cells against DNA damage.[6] The levels of PAR can be easily assessed by immunohistochemistry. Both in non-small cell lung cancer (NSCLC) and in cervical cancer, high levels of PAR indicate poor prognosis, correlating with a reduced infiltration of the tumors by CD8+ T lymphocytes.[7,8] Indeed, there is an expanding literature indicating that scarce infiltration of NSCLC by CD8+ T cells is associated with dismal prognosis, as well as with poor responses to immunotherapy.[9] It is important to note that there is little if any correlation between the levels of PARP1 protein expression and the abundance of intracellular PAR,[8] likely reflecting the fact that PARP1 is usually enzymatically inactive and that it is only its activation that leads to PAR accumulation. Indeed, the expression of PARP1 protein has no prognostic value in NSCLC.[8]

Recently, we decided to determine the mechanism through which PAR expression anticorrelates with CD8+ T cell infiltration in NSCLC. Theoretically, the activation of PARP1 yielding PAR in malignant cells could result from T lymphocyte-mediated immunoselection. Alternatively, or in addition, exaggerated PARP1 activation might have local immunosuppressive effects. For this reason, we decided to modulate PARP1 expression/activation in NSCLC cells and to investigate the capacity of PARhigh and PARlow NSCLC cells to elicit T cell-mediated immune responses in mice.[6]

First, we created two PARhigh NSCLC cell lines (Lewis lung cancer [LLC] and tissue culture number one [TC1]) by selecting them by long-term culture in the presence of low-dose cisplatin (Figure 1a). The resulting PARhigh NSCLC cell lines maintained their elevated PAR level even after several months of culture in the absence of cisplatin, likely due to an increased level of spontaneous DNA damage.[6] PARhigh NSCLC cells were cloned to reduce their heterogeneity and then subjected to the knockout (KO) of PARP1 by CRISPR/Cas9 technology, yielding cells that lacked any signs of PAR accumulation (which demonstrates that PARP1 is the sole PARP isoform generating PAR in these cells) and became sensitive to cisplatin (which demonstrates that PARP1 was indeed responsible for chemotherapy resistance). Of note, PARhigh cells exhibited strong responses to the cytostatic and cytotoxic effects of niraparib, a specific PARP1 inhibitor which is clinically approved.[10] Conversely, PARP1KO cells became resistant to niraparib, as expected (Figure 1b).

Figure 1.

Modulation of PARP1 activity and immune response in mouse NSCLC models. A. Multistep generation of PARhigh and PARP1KO NSCLC cell lines. LLC and TC1 PARhigh cell lines were obtained by exposure to sublethal cisplatin doses. PARhigh cells were cloned and transfected with a CRISPR/Cas9 vector that lacks a guidance DNA or that target PARP1. Then, LLC and TC1 clones were derived that possess a high PARP1 activity or that lack PARP1 expression and activity. B. Characteristics of PARhigh and PARP1KO NSCLC cells in vitro. While PARhigh cells were largely resistant to cisplatin but died in the presence of niraparib, PARPKO cells died when treated with cisplatin but only showed a marginal response to niraparib. C. Immunosurveillance of PARP1KO NSCLC in vivo. PARhigh cells proliferated similarly in immunocompetent and T cell-deficient mice while the proliferation of PARPKO cells is strongly controlled by T cells. D. Niraparib effects on PARhigh and PARP1KO NSCLC in vivo. Whereas PARhigh tumors implanted in mice reduced their growth in response to niraparib in both immunocompetent and T-cell deficient mice, niraparib decreased PARPKO tumor growth only in the presence of T cells. LLC, Lewis-lung cancer; NSCLC, non small cell lung cancer; PAR, poly(ADP-ribose); PARP1, Poly(ADP-ribose) polymerase-1; s.c., subcutaneous; TC1, Tissue culture number one.

Next, we inoculated PARhigh and PARP1KO NSCLC cells into immunodeficient mice, from which T lymphocytes had been depleted by injection of antibodies specific for CD4 or CD8. PARhigh and PARP1KO LLC or TC1 cells indistinguishably formed tumors in these T cell-depleted animals. In sharp contrast, PARP1KO TC1 cells failed to develop tumors in immunocompetent histocompatible mice, in conditions in which PARhigh TC1 cells readily proliferated, forming tumors with similar growth kinetic in immunocompetent and immunodeficient mice. When immunocompetent recipient mice had rejected the inoculation of PARP1KO TC1 cells, they subsequently became resistant against PARhigh TC1 cells implanted into the opposite flank. These findings indicate that PARP1KO TC1 cells cause a durable protective anticancer immune response mediated by T cells. We also found that PARP1KO LLC cells came under immunosurveillance. PARhigh LLC cells similarly grew on T cell-depleted and control mice.[6] However, PARP1KO LLC cells experienced a considerable growth disadvantage when implanted into immunocompetent mice. The immune infiltrate of such PARP1KO LLC cells contained more dendritic cells and more activated, ICOS-expressing CD8+ T cells, but less activated regulatory T cells, coupled to the downmodulation of the exhaustion marker PD-1 on CD4+ and CD8+ T cells.[6] These findings confirm the capacity of active PARP1 to locally suppress the anticancer immune response (Figure 1c).

Next, we investigated whether pharmacological (rather than genetic) inhibition of PARP1 would stimulate an anticancer immune response. Of note, PARhigh LLC tumors similarly responded to niraparib treatments in immunocompetent and T cell-deficient mice, suggesting that niraparib effects on cells possessing the niraparib target (which is activated PARP1) do not require any major contribution by the cellular immune system. In sharp contrast, PARP1KO LLC cancers only responded to niraparib when T lymphocytes were present in the system. PARP1KO LLC tumors evolving in T cell-depleted mice did not respond to niraparib. However, PARP1KO LLC tumors developing in immunocompetent mice reduced their growth upon systemic injection of niraparib (Figure 1d). These findings point to the possibility that PARP1 inhibitors can act through the stimulation of an anticancer immune response even when the malignant cells lack this target. However, the mechanisms of this immunostimulatory effect remain to be elucidated.