Cancer-cell-intrinsic mechanisms shaping the immunosuppressive landscape of prostate cancer

Moderate mutation rate and a subclass with DDR mutations

Neoantigen recognition by cytotoxic T-lymphocytes (CTLs) is believed to be required for CTL-induced tumor cell killing. For most human cancers without a viral etiology, tumor neoantigens form through the various forms of genetic mutations.¹⁰ Tumors with higher somatic mutation load are generally associated with higher neoantigen formation.^10,11 Unlike lung cancer or melanoma, which have mutation rates of around ten somatic mutations per megabase (Mb) or higher, the mutation rate of primary PCa is at the level of one mutation per Mb. Thus, PCa cells are less likely to be recognized by autologous T-cells.^10,12 Metastatic PCa can reach 4.4 mutations per Mb.¹³ A moderate neoantigen load in PCa may be a fundamental reason for the limited infiltration of CTLs into the tumor, which defines an immunologically cold microenvironment.^14,15

A clinically meaningful deviation from the general impression of the low mutational burden of PCa is that a subset of the patients (around 20%) harbor mutations in genes involved in DNA damage repair (DDR).^13,16 DDR gene mutation is associated with a higher mutation burden across cancer types, including PCa.^13,17-¹⁹ Therefore, it is conceivable that PCa patients carrying DDR gene mutations contain a higher number of neoantigens. Of note, patients with DDR mutations exhibit a significantly different TME and show an overall better response to immunotherapy.²⁰-²³ Deficient mismatch repair (dMMR) status was analyzed in 381 advanced PCa samples and was found to be associated with worse overall survival, a higher T-cell infiltration, and more PD-L1 expression.²² Breast cancer gene 2 (BRCA2)-mutated prostate tumors have higher infiltration of immune cells. Nevertheless, the intratumoral CD8 to forkhead box P3 (Foxp3) ratio was lower in BRCA2-mutated tumors suggesting a more suppressed TME.²⁴ However, the case number in this study was low. Thus, further studies with more cases should be performed to validate the findings. Loss of mutS homolog 2 (MSH2), one of the mismatch repair (MMR) genes, was examined in a large cohort (1133 primary prostatic adenocarcinomas and 43 prostatic small cell carcinomas). The results showed that MSH2 loss was associated with high-grade primary tumors, hypermutation, and higher tumor-infiltrating immune cells.²⁵ Overall, the DDR-mutated PCa manifests high immune infiltration, defining a genotype with a higher chance of response to immunotherapy. FDA has approved pembrolizumab for unresectable or metastatic solid tumors that have microsatellite instability-high (MSI-H) or dMMR, which was the FDA’s first tissue/site-agnostic approval.²⁶ Therefore, patients with PCa characterized by dMMR or MSI-H can now benefit from the anti-PD1 immunotherapy.

PARP inhibitors

In the mCRPC patients with DDR gene defects, especially with BRCA1/2 mutation, poly(ADP-ribose) polymerase (PARP) inhibitors olaparib and rucaparib showed impressive efficacy and were both approved by the FDA to treat this type of advanced PCa.^27,28 During DDR, PARP1 (the PARP family member responsible for most mammalian poly-ADP-ribosylation [PARylation]) is activated to catalyze the PARylation of many proteins, including PARP1 itself. PARP inhibitors possess both PARP inhibition and PARP trapping activities.²⁹ PARP trapping prevents PARP1 from dissociating from the DNA, leading to more DNA damage.^30,31 PARP inhibitor-induced PARP trapping and DNA damage lead to double-stranded DNA (dsDNA) leakage into the cytoplasm. Cytosolic dsDNA activates the innate immune signaling through the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway and stimulates the release of type I interferons.^32,33 The higher probability of DNA damage by DDR deficiency may lead to higher neoantigen load and render PCa cells more immunogenic. Besides, PARP inhibitors can upregulate PD-L1 expression at least partly through induction of glycogen synthase kinase 3 beta (GSK3β) phosphorylation.^34,35 PD-L1 upregulation may lead to more pronounced T-cell exhaustion in the TME, but it also provides the immune checkpoint target exploitable for anti-PD1/PD-L1 therapy. These immune-related effects of PARP inhibitors have prompted tremendous interest, and preclinical studies are converging to show that PARP inhibitors can help establish the anti-tumor immunity and promote the efficacy of ICB therapies in both DDR-deficient and -proficient cancers.^32,33,35,36 In PCa, at least three clinical trials are testing the combination of olaparib and anti-PD1 or anti-PD-L1 in advanced PCa (NCT04336943, NCT03834519, and NCT02484404), and the results will soon reveal if this combination is sufficient to sensitize advanced PCa to ICB therapy. We list all the clinical trials mentioned in this review in Table 1.

Enhanced neoantigen expression for PCa immunotherapy

The majority of PCa patients carry a moderate mutational load and a limited pool of neoantigens. For these PCa patients, improving the neoantigen expression or enhancing the immune system’s reaction to existing neoantigens is a likely promising strategy. The advancement in next-generation sequencing and analysis algorithms now allows the efficient identification of putative neoantigens recognizable by CTLs based on the tumor genome.¹⁰ Once validated experimentally, these neoantigens form the basis for immunotherapies such as adoptive T transfer or vaccines.^11,37,38 Another possible approach is to artificially express the foreign antigens in tumor cells, which may be facilitated by microbial-mediated transgene expression. For instance, highly immunogenic tetanus toxoid (TT) proteins were delivered to pancreatic tumor cells through attenuated Listeria, which induced a strong immunological reaction by reactivating preexisting TT-specific memory T-cells.³⁹

Even though microbial-based immunotherapies are promising, overexpressing neoantigens specifically in the tumor area is limited by the difficulty in directional trafficking of the microbes to the tumor.⁴⁰ Interestingly, intratumoral administration of mRNA may provide a new solution. For example, systemic IL-12 administration is hardly effective, but the intratumor administration of IL-12 mRNA induced a dramatic CD8⁺ T-cell-dependent tumor regression and sensitized the tumors to ICB across multiple syngeneic models.⁴¹ A clinical trial combining IL-12 mRNA administrated intratumorally and anti-PD-L1 durvalumab to treat solid tumors is being investigated (NCT03946800). Therefore, the intratumoral mRNA delivery of foreign antigens and cytokines critical to CTLs and natural killer (NK) cells may be developed as a novel immunotherapy approach to treat advanced PCa.

Effect of androgen ablation on the TME

The effect of androgen in the TME is relevant to both PCa and other cancer types because sex differences in immune functions are well documented.⁴² Both sex chromosome genes and sex hormones, including estrogens, progesterone, and androgens, contribute to the differential regulation of innate and adaptive immune responses between males and females.⁴² Women generally mount a stronger immune response than men.⁴³ In terms of response to ICB therapy, male cancer patients tend to respond better than female patients, but ICB in combination with chemotherapy tends to be more effective in female patients than male patients.^43,44 In patients with PCa, ADT shows a profound impact on immune infiltration, e.g., increased intratumoral T-cells and macrophages.^45,46 Drake et al.⁴⁷ reported that androgen ablation by castration mitigated the tolerance of adoptive T-cell transfer targeting an engineered prostate-restricted antigen in the transgenic adenocarcinoma of mouse prostate (TRAMP) transgenic mouse model of PCa, providing one of the early motivations for exploring the potential of combining ADT and immunotherapy for PCa. In a recent study that conducted single-cell RNA sequencing (scRNA-seq) for mCRPC resistant to enzalutamide and prior to pembrolizumab, AR was found to be expressed in CD8⁺ T-cells and directly bind to the enhancer region of interferon gamma (IFNγ) gene.⁴⁸ In preclinical models, AR blockade in CD8 T-cells prevented T-cell exhaustion and improved the efficacy of anti-PD1 therapy through elevating IFNγ production.⁴⁸ From another study of transcriptome profiling of six paired pre-ADT biopsies with post-ADT PCa lesions, ADT was found to significantly increase the infiltration and activity of CTLs in PCa TME. This immune activation caused by ADT was more pronounced at the early stage of ADT but became very limited when the tumor progressed to the CRPC stage.⁴⁹ This result is consistent with our RNA profiling of the treatment-naïve and fully formed CRPC (tumors progressed to the CRPC stage after surgical castration and enzalutamide treatment) from the phosphatase and tensin homolog/SMAD family member 4 (Pten/Smad4) transgenic mouse model of PCa, where the two tumor stages shared a broadly conserved transcriptome with only modest alterations of a number of collagen-related and immune-related genes including C-C motif chemokine receptor 2 (CCR2).⁵⁰

While ADT is promising to condition prostate tumors for more effective immunotherapy, evidence shows that AR inhibition or deprivation can induce a more immunosuppressive TME. In a transplanted PCa model, B-cell infiltration in tumors following ADT promotes CRPC through lymphotoxin secretion and inhibitor of nuclear factor kappa-B kinase subunit alpha/signal transducer and activator of transcription 3 (IKK-α/STAT3) activation.⁵¹ Furthermore, castration in PCa-bearing mice increased the infiltration of polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs) to the tumors by elevating IL-8 production from PCa cells to recruit PMN-MDSCs that express the cognate receptor, C-X-C motif chemokine receptor 2 (CXCR2).⁵² Several studies established that PMN-MDSCs infiltrate the PCa microenvironment and promote PCa progression and resistance to ICB therapy through inhibiting the T-cell-receptor signaling, activation, proliferation, and infiltration of CTLs.⁵²-⁵⁷ Antiandrogen therapy also induces infiltration of other malignancy-promoting immune cells, such as macrophages,⁵⁸ and mast cells.⁵⁹ Taken these studies together, it suffices to say that ADT and AR antagonists will remodel the TME by disrupting AR signaling inside both prostate carcinoma cells as well as specific immune cells. Clinical trials combining ant-PD1/PD-L1 with ADT and/or enzalutamide in metastatic hormone-sensitive PCa, non-metastatic hormone-sensitive PCa, or mCRPC are being actively explored (e.g., NCT02312557, NCT03543189, NCT04191096, and NCT01875250). The results from these trials will guide new designs of systemic treatment of PCa.

Considerations for combining ADT and immunotherapy

Due to the complex roles of androgen signaling in shaping the TME and modulating immune cell activities, careful considerations should be given to the design and implementation when combining ADT and immunotherapy in PCa treatment. First, choosing the appropriate ADT regimen is critical. Different ADT methods may have different or opposite outcomes when combined with immunotherapy. Pu et al.⁶⁰ showed that ADT with nonsteroid AR antagonists such as flutamide and enzalutamide significantly inhibited IFNγ production of immune cells and suppressed TLR9 agonist (CpG) immunotherapy by an off-target effect on γ-aminobutyric acid receptor type A (GABA-A) receptor inhibition. Meanwhile, ADT with orchiectomy or androgen synthesis inhibitor abiraterone did not have this off-target effect and significantly enhanced the antitumor effects of immunotherapy.⁶⁰ Second, it is essential to choose the sequential steps that perform better in combination therapy. In a recent clinical study treating nonmetastatic CRPC patients with either poxvirus-based vaccine or the anti-androgen nilutamide, patients who received the vaccine first followed by nilutamide showed a significantly higher survival rate compared with patients who received nilutamide treatment first.⁶¹ Third, foreseeing the possible immunosuppressive mechanisms that might be stimulated due to ADT, an additional targeted therapy may be added to the ADT/immunotherapy combination. For example, ADT can induce IL-8 secretion, thus recruiting PMN-MDSCs to blunt the immunotherapy efficacy.⁵² A phase 1/2 trial based on this finding was initiated to test the combination of nivolumab (anti-PD1), BMS-986253 (anti-IL-8), and ADT drug degarelix to treat men with castration-sensitive PCa (NCT03689699).

PTEN/phosphatidylinositol 3-kinase (PI3K) pathway

Nearly 20% of primary PCa and 50% of CRPC patients harbor PTEN loss-of-function alterations, leading to hyperactivation of the PI3K pathway and cell-autonomous mechanisms of tumor cell proliferation and metastasis.⁸² Accumulating evidence shows that PTEN/PI3K pathway is also involved in the immune evasion of PCa. Prostate epithelial-specific Pten knockout in mice generated prostate adenocarcinoma with abundant infiltration of Gr-1⁺ CD11b⁺ MDSCs through upregulation of inflammatory cytokines and chemokines.⁸³ In the more aggressive Pten/Smad4 double knockout mouse model of metastatic PCa, MDSCs were recruited through the CXCR2 chemokine pathway to the tumor bed and constituted over 50% of all the live cells, forming a formidable barrier for the antitumor immunity by T-cells and NK cells.⁵⁴PTEN loss increased CXCL8 secretion and sustained tumor cell survival.⁸⁴ Higher CXCL8 led to the infiltration of tumor-associated macrophages (TAMs) in PCa.⁸⁵ A study with transcriptome-based immune deconvolution followed by validation with clinical sample staining showed that PTEN deficiency was correlated with increased indoleamine 2,3-dioxygenase 1 (IDO1) expression and FoxP3⁺ regulatory T-cells in PCa.⁸⁶ Given the high prevalence of PTEN loss in advanced PCa and the evidence supporting its consequence in shaping the immunosuppressive TME, pharmacological targeting of the PI3K pathway is expected to elicit dual effects: inhibiting the survival and proliferation of prostate neoplastic cells, and sensitizing mCRPC to ICB therapy and perhaps other modalities of immunotherapy.

TP53 mutations

Like in other cancer types, TP53 is also one of the most frequently mutated genes in advanced PCa. In primary PCa, 8% of patients harbor TP53 gene alteration. In mCPRC patients, however, the mutation frequency of TP53 dramatically increases to 50%.¹⁶TP53 mutation is a critical underlying mechanism for senescence bypass, metastasis, and resistance to second-generation AR antagonists.⁸⁷ In other solid tumors, loss of function of TP53 has been shown to induce immunosuppression through upregulating PD-L1,⁸⁸ reprogramming TAMs to be immunosuppressive through exosomal miR-1246,⁸⁹ and activating the nuclear factor-kappa B (NF-κB) signaling to promote chronic inflammation.⁹⁰ In PCa, on the genetic backdrop of Pten loss, co-deletion of Trp53 in the mouse prostate led to faster tumor formation and increased MDSC infiltration through CXCL17 secretion.⁹¹ A truncated TP53 isoform Δ133TP53β was found elevated in PCa, and patients with high Δ133TP53β expression had an immunosuppressive TME featured by a higher proportion of immune cells positive for PD1, PD-L1, and colony stimulating factor 1 receptor (CSF1R) as well as CD163⁺ macrophages.⁹² Although TP53 missense mutation was associated with increased tumor-infiltrating T-cells in prostate primary tumors,⁹³ it is very likely that the infiltrated T-cells are rendered exhausted and dysfunctional by the severely immunosuppressive TME. Nevertheless, the T-cells in the prostate tumors harboring TP53 missense mutations may be a targetable population for combinatorial immunotherapy to waken the antitumor cytotoxicity.

Other oncogenic pathways impacting the TME

MYC is a major driver of PCa tumorigenesis and progression.^94,95MYC is amplified at chromosome 8q24 in about 4% of primary PCa and 10% of mCRPC.⁸¹ Han et al.⁹⁶ developed an MYC inhibitor MYCi361 which induced proteasome-mediated MYC degradation in cancer cells through MYC phosphorylation on threonine-58. Inhibition of MYC with MYCi361 in PCa models suppressed immune infiltration, upregulated PD-L1 on tumor cells, and sensitized the tumors to anti-PD1 immunotherapy.⁹⁶

In a prostate adenocarcinoma model driven by the loss of Pten and Smad4, MDSCs were the primary infiltrating immune cell type in the tumor. Mechanistically, the transcription factor YAP was translocated to the nuclei of cancer cells to drive CXCL5 upregulation. CXCL5 recruited CXCR2-expressing MDSCs to the tumor area, blockade of CXCL5-CXCR2 signaling with CXCR2 antagonist triggered a robust antitumor response and prolonged mice survival.⁵⁴

Oncogenic wingless-type MMTV integration site family (WNT)–β-catenin signaling was first revealed to reduce T-cell recruitment and cause resistance to ICB therapy in melanoma.⁹⁷ In mCRPC, mutations of APC and catenin beta 1 (CTNNB1) occur at 8.7% and 4.0%, respectively, indicative of hyperactivated WNT–β-catenin signaling.¹³ Linch et al.⁹⁸ explored the intratumoral heterogeneity of PCa by sequencing multiple regions in human PCa biopsies and identified that patients with dMMR and patients with activating mutations of β-catenin harbored high inflammatory infiltration. Consistent with the findings from melanoma, patients with activating mutations of β-catenin exhibited a low CD8⁺/FoxP3⁺ ratio, a potential surrogate marker for immune evasion.⁹⁸

Genomic profiling identified a subtype of primary PCa characterized by missense mutation of SPOP (approximately 11%) and homozygous deletion of chromodomain helicase DNA-binding protein 1 (CHD1, approximately 8%), both encoding proteins involved in genome integrity and DNA repair through homologous recombination.^12,16,99-¹⁰² Using genetic mouse models that carried the loss of function of the two genes individually or simultaneously, we found that SPOP and CHD1 synergistically promoted repair of naturally occurring or chemically induced DNA damages in prostate epithelial cells.¹⁰³ It will be important to determine whether this PCa subtypes is particularly sensitive to PARP inhibition and ICB therapy.

Frequent nonsynonymous mutations of SPOP seem to be unique to PCa, with the underlying reasons still unresolved. Interestingly, SPOP mutations were enriched in primary PCa relative to mCRPC, whereas other gene mutations were more enriched in metastases.⁸¹SPOP as the substrate adaptor for the cullin 3-RING E3 ubiquitin ligase was stabilized through CDK4-mediated phosphorylation. The CDK4/6 inhibition or SPOP loss-of-function mutations impaired PD-L1 degradation and increased PD-L1 protein levels.¹⁰⁴ Therefore, in PCa patients with SPOP mutation, SPOP function was lost, leading to higher PD-L1 and lower tumor-infiltrating lymphocytes.¹⁰⁴

Interestingly, deletions of CHD1 and PTEN display a mutually exclusive pattern in the PCa genome because CHD1 is essential to PCa cell survival when PTEN is deficient.¹⁰⁵ Using genetic mouse models of PCa, Zhao et al.¹⁰⁶ further found that in PTEN-deficient PCa, CHD1 fostered the immunosuppressive TME through sustaining IL-6 production and promoting MDSC recruitment.

An integrative genomic analysis of mCRPC samples identified cyclin-dependent kinase 12 (CDK12) biallelic inactivating mutations defining a distinct PCa subtype.¹⁰⁷CDK12 loss was mutually exclusive with DNA repair deficiency, ETS fusion, and SPOP mutations. Further, CDK12 loss was associated with genomic instability, higher neoantigen burden and more T-cell infiltration. These suggest that patients with CDK12 loss might have higher response to ICB treatment.

Polycomb repressor complex 1 (PRC1) and PRC2, acting in tandem to silence target genes, were reported to promote dedifferentiation and stemness during development and cancer. Using AR⁻/NEPC⁻ double-negative PCa models, Su et al.¹⁰⁸ found that overexpression of PRC1 in PCa cells induced CCL2 secretion and promoted the recruitment of M2-like tumor-associated macrophages (TAMs) and regulatory T-cells (Tregs). PRC1 inhibitor cooperated with ICB to abrogate the metastasis of double-negative PCa.¹⁰⁸ The histone methyltransferase, enhancer of zeste homolog 2 (EZH2), is a catalytic subunit of PRC2 for tri-methylation of histone H3 at Lys 27 (H3K27me3). EZH2 is overexpressed in mCRPC and promotes PCa progression.¹⁰⁹ Inhibition of EZH2 was found to activate the dsRNA-STING-interferon stress axis and significantly increase the infiltration of CD8⁺ T-cells, and M1-polarized TAMs in the PCa models.¹¹⁰ The combination of EZH2 inhibitor and anti-PD1 demonstrated combinatorial benefit in preclinical models.¹¹⁰