Small-Molecule Natural Product Physachenolide C Potentiates Immunotherapy Efficacy by Targeting BET Proteins

The success of cancer immunotherapy in spurring cytotoxic T cell–mediated cancer cell elimination has recently resulted in a paradigm shift in cancer treatment (1, 2). However, resistance of cancer cells to apoptosis is a well-known hallmark of tumor biology (3), yet the relative importance of cancer cell antiapoptotic factors that might limit T cell–mediated cancer cell destruction is poorly understood. Activated CD8⁺ T cells play a critical role in successful immunotherapy of cancer (4, 5). Nonetheless a significant number of patients with cancer do not respond to immunotherapy treatments, and the basis for this unresponsiveness in unclear (6, 7). Furthermore, the molecular basis underlying tumor destruction by CD8⁺ T cells in responding patients is unknown. High local levels of T-cell cytolytic granule proteins, such as perforin and granzyme B, are often correlated with immune cell infiltration and successful responses to immunotherapy (6–8). Nevertheless, it is difficult to determine a direct role of these granule cytolytic proteins, as their elevated levels also correlate with a high level of T-cell activation and cytokine production. Interestingly, some in vivo studies in mouse tumor models have described significant cancer cell destruction in the absence of perforin, a protein that is essential for granule-mediated cancer cell destruction by T cells (9, 10). Under these conditions, cancer cells are killed directly by high levels of cytokines such as IFNγ, TNFα, FasL, or TRAIL (11), possibly due to the preferential activation of cytokine-dependent cytolytic pathways following T-cell recognition of cancer antigens of low avidity (12, 13). In contrast to cytolytic granule proteins, death ligand proteins of the TNF family or agonist antibodies to their death receptors can be administered to patients with cancer. However, TNFα and FasL or agonist antibodies to their death receptors exhibit unacceptable levels of systemic toxicity. Although agonists to TRAIL death receptors are well tolerated, their clinical activity so far has been disappointing (14). Recent studies have suggested that in the absence of a dominant apoptotic response, TRAIL death receptor signaling (in a similar manner to TNF receptor and Fas signaling) can favor tumor progression (15, 16). Therefore, for these death ligands to produce significant anticancer effects, preferential activation of the downstream extrinsic apoptotic signaling pathway in the cancer cells must occur. Although most studies have focused on TNF family members, caspase-8–dependent extrinsic apoptosis signaling can also be triggered by IFNs (17), cell stress (18), and double-stranded RNA viral mimetics like poly I:C (19).

Here, we demonstrate that a natural product (NP), physachenolide C (PCC), with the ability to selectively inhibit proliferation of prostate cancer cells (20) and amplify TRAIL-mediated extrinsic apoptosis signaling in human renal carcinoma cells (21), is also capable of promoting caspase-8–dependent extrinsic apoptosis signaling in melanoma and renal carcinoma cells in response to TNF family death ligands and the immune adjuvant and viral mimetic poly I:C. We determined the molecular mechanism of action involves a reduction of levels of the anti-apoptotic cFLIP. Elevated levels of cFLIP are often associated with poor outcome in a number of cancers, therefore cFLIP remains an important target for cancer therapy (22). PCC, in combination with poly I:C, improves therapeutic benefits in two mouse preclinical cancer models of melanoma. PCC also enhances T-cell–mediated destruction of melanoma cells in response to death ligands produced by the activated T cells, and thus provides significant therapeutic benefit when combined with adoptive cell transfer of human lymphocytes in a mouse human melanoma xenograft model. We identified bromodomain and extraterminal domain (BET) proteins as the critical cellular targets of PCC for apoptosis sensitization.

For details, see Supplementary Materials and Methods.

The ACHN, M14, Malme-3M, SK-MEL-5, SK-MEL-28, UACC-62 cells were all obtained from the Developmental Therapeutics Program (DTP; NCI, Frederick, MD). R331 cells were kindly provided by Dr. Robert Wiltrout (NCI, Frederick, MD). The cFLIP transfectants were generated and characterized by us (A.D. Brooks and T.J. Sayers). B16F10 cells were a kindly provided by Dr. Julio Valencia (CIP, NCI Frederick, Frederick, MD). The melanoma cell lines 888, 1383, and Baldwin were kindly provided by Dr. Steven Rosenberg, NIH, Bethesda, MD. Human foreskin fibroblast (HFF) cells were purchased from ATCC. Normal human renal epithelial cells (HRE) and primary melanocytes cells were purchased from Lonza. Cell lines were maintained as recommended by the source institution. Mycoplasma testing was performed using the PCR test (Thermo Fisher Scientific). Cells were used for a maximal of fifteen passages. Cell line authentication is described at https://discover.nci.nih.gov/cellminer/celllineMetadata.do under the "Fingerprint" header. All cell lines used in animal studies were confirmed to be negative in the Molecular Testing of Biological Materials-Mouse/Rat (MTBM-M/R) test at NCI Frederick (Frederick, MD).

For details, see Supplementary Materials and Methods.

For details, see Supplementary Materials and Methods.

For details, see Supplementary Materials and Methods.

ACHN, M14, or SK-MEL28 cells (1 × 10⁶ cells/well, 6-well plates) were extracted for SDS-PAGE and immunoblot analysis as described in Supplementary Materials and Methods.

For details, see Supplementary Materials and Methods.

For details, see Supplementary Materials and Methods.

Detailed protocols in Supplementary Materials and Methods.

Athymic BALB/c mice were obtained from the Animal Production program (Charles River Laboratories, Inc., Frederick, MD). C57BL/6 mice were provided by the Animal Production Facility of the NCI. NOD/scid/il-2rg^−/− (NSG) mice were kindly provided by Dr. Mary Custer, NIH, Bethesda, MD. NCI is approved by the American Association for the Accreditation of Laboratory Animal Care International and follows the Public Health Service policy for the Care and Use of Laboratory Animals. Animal care was provided in accordance with the procedures outlined in the Guide for Care and Use of Laboratory Animals (National Research Council, Washington, D.C.). For in vivo experiments with R331 cells, BALB/c mice at 6–8 weeks old (25–30 g by body weight) were purchased from Harlan. Mice were housed in filter-topped cages under specific pathogen-free conditions in Meharry Medical College (MMC) Animal Care Facility and cared for in accordance with the procedures outlined in the NIH Guide for the Care and Use of Laboratory Animals and Institutional Animal Care and Use Committee. MMC is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International and follows the Public Health Service Policy for the care and use of laboratory animals. For the human xenograft melanoma model, nude athymic, female mice (6 to 8-week old, n = 8), were implanted subcutaneously with 0.2 mL of M14 melanoma cells (1 × 10⁶ cells/mouse) into the right flank. When tumors reached approximately 100 mm³ (2 to 3 weeks after initial inoculation), mice were randomized into treatment groups where the average size of tumors were equivalent. They were treated either vehicle control (30% Trappsol and 30% DMSO in PBS), PCC (20 mg/kg in 30% Trappsol and 30% DMSO), poly I:C (50 μg/mouse in PBS), or a combination of PCC followed 24 hours later by poly I:C, all intratumorally (i.t.), twice per week for four weeks or as per the experimental schedule. Trappsol was prepared at a concentration of 43% in saline and added to PCC dropwise to yield a final concentration of 30%. For xenograft models of human melanoma in NSG mice, M14 cells (1 × 10⁶ cells/mouse) were implanted into the right flank. When tumors reached approximately 75 mm³ (3 weeks after initial inoculation), mice were injected with 6 × 10⁶ human peripheral blood mononuclear cells (PBMC) or saline intravenously (i.v) as described previously (23). Two weeks post PBMC injection, mice were treated (i.t.) with either vehicle control (30% Trappsol and 30% DMSO in PBS), PCC (10 mg/kg in 30% Trappsol and 30% DMSO) twice weekly for three weeks, and tumor growth was monitored for 50 days after PBMC injection. Mice were euthanized at day 60 when onset of graft versus host disease started to become apparent.

For B16-F10 syngeneic model, C57BL/6 mice male and female mice (6–8) weeks old were used for the study. Mice were implanted subcutaneously with 0.2 mL of B16F10 melanoma cells (5 × 10⁵ cells/mouse) into the right flank. When the tumor reached approximately 100 mm³ (7–10 days after initial inoculation), mice were randomized into groups with equivalent tumor size, they were treated either vehicle control, PCC (10 or 20 mg/kg, i.t) or poly I:C (50 μg/mouse) intraperitoneally (i.p.), or a combination of and PCC (i.t.) followed by poly I:C (i.p.) 24 hours later, twice per week for 3 weeks or as indicated. In some experiments, poly I:C was also administered i.t. For some experiments, single-cell cFLIP KO clones obtained by CRISPR/CAS9 targeting in B16F10 melanoma were used to determine the role of cFLIP on tumor growth. The appearance and growth of tumors were monitored twice per week by an independent observer. The greatest longitudinal diameter (length) and the greatest transverse diameter (width) of a palpable tumor were measured to the nearest 0.1 mm using a caliper. Tumor volume (mm³) was calculated by the ellipsoidal formula [Tumor volume = (length × width²)/2]. Mice were euthanized once control tumors reached the termination point, as indicated by compromised health or tumor size. For survival studies, when the tumors reached greater than 2,000 mm³ or if the tumor had been necrotic for more than 4 weeks with no evidence of regression, mice were painlessly euthanized with CO₂ as outlined in the Frederick National Laboratory for Cancer Research Animal Care and Use Committee guidelines.

For details, see Supplementary Materials and Methods.

For details, see Supplementary Materials and Methods.

For details of synthesis, see Supplementary Materials and Methods.

M14 cells in T150 flasks at about 80% confluence were treated overnight with vehicle (media plus DMSO) or PCC (500 nmol/L). RNA was isolated from M14 cells in three independent experiments by using TRIzol (Thermo Fisher Scientific) and RNEasy Kit (Qiagen) as per the manufacturer's protocol. RNA library preparation and sequencing was performed by ACTG Inc. using their standard procedures (http://www.actginc.com). RNA-sequencing (RNA-seq) data were analyzed using IPA analysis software from Qiagen Inc.

For detailed protocol, see Supplementary Materials and Methods.

For cytotoxicity experiments, results were normalized to cells treated with DMSO alone. Data were obtained from at least three independent experiments and are presented as means ± SEM unless otherwise indicated. Comparisons of mean values between the groups were analyzed using GraphPad Prism 7. Statistical significance of the differences was analyzed by applying two-tailed paired Student t test or by ANOVA with Tukey multiple comparison test or with Dunnett multiple comparison test, with P values <0.05 considered significant.

We previously demonstrated that a small-molecule NP of the 17β-hydroxywithanolide (17-BHW) class, PCC (Fig. 1A), could amplify TRAIL-mediated extrinsic apoptosis signaling in human renal carcinoma cells (21). In patients with cancer, renal carcinomas and melanomas are known to respond best to immunotherapy. Therefore, we also tested the effect of PCC on apoptosis signaling in a panel of human melanomas. PCC was also capable of selectively sensitizing a panel of human melanoma cells to TRAIL-induced apoptosis, whereas normal human melanocyte, human foreskin fibroblast (HFF) and human renal epithelium (HRE) cells were unaffected by this combination (Fig. 1B). Significantly, the M14-VR cell line selected for resistance to molecular targeted therapy using the BRAF (oncogene) inhibitor Zelbora (vemurafenib; Supplementary Fig. S1) was also found to be extremely sensitive to PCC treatment in amplifying death signaling by TRAIL. Thus, the establishment of vemurafenib resistance did not block caspase-8–dependent apoptosis. Because the TLR3 ligand, poly I:C had been reported to promote caspase-8–dependent apoptosis in certain human cancers, we tested PCC for its ability to sensitize human renal carcinoma and melanoma cells to poly I:C. Following exposure to PCC and poly I:C, a significant decrease in cell numbers for both renal carcinoma and most melanoma cell lines, but not normal human melanocyte, HFF and HRE cells was also seen (Fig. 1B). Interestingly, the IC₅₀s of PCC required for TRAIL and poly I:C apoptosis sensitization were very similar (Fig. 1C). One exception was MALME-3M cells, which were sensitized to TRAIL but resistant to poly I:C-mediated apoptosis. Importantly, PCC alone had little effect on survival of both cancer and normal cells at concentrations up to 2,000 nmol/L. Therefore, apoptosis sensitization by PCC was not due to any direct toxic effect of the compound. This poly I:C-dependent reduction of cell number was due to apoptosis as it was completely blocked by the pan-caspase inhibitor zVAD-fmk. Bafilomycin, that blocks TLR3 signaling, also completely abrogated PCC and poly I:C-mediated apoptosis (Supplementary Figs. S2A–S2C), suggesting that the interaction of poly I:C with TLR3 was important for enhancing apoptosis signaling as reported previously (24, 25). It is noteworthy that in contrast to other melanoma cells studied, the resistant MALME-3M cells are reported to lack TLR3 (26). Inclusion of the necroptosis inhibitor Nec-1, IRF3 inhibitor BX795, or neutralizing antibodies to the type I IFN receptor (Supplementary Figs. S2D and S2E), or TRAIL, FasL, and TNFα (Supplementary Fig. S3A and S3B), did not block apoptosis. Therefore, poly I:C-mediated cell death in the presence of PCC was independent of IFN and the TNF superfamily death receptor signaling.

Because many melanoma cells exhibit elevated levels of the antiapoptotic cFLIP protein, we assessed the effects of PCC on cellular levels of cFLIP. Both isoforms of cFLIP were reduced by PCC in a dose-dependent manner (Fig. 2A). On subsequent exposure to either TRAIL or poly I:C, a significant increase in effector caspase activation was observed in PCC-treated cells (Fig. 2B; Supplementary Fig. S4A). FLIP-specific siRNAs increased TRAIL or poly I:C–mediated cell death, with both isoforms of cFLIP involved in apoptosis resistance (Fig. 2C and D; Supplementary Fig. S4B). CRISPR/Cas9 editing of the cFLIP gene also resulted in increased sensitivity to poly I:C-mediated apoptosis (Supplementary Fig. S4C). PCC-mediated reduction in cFLIP was posttranscriptional, because no change in mRNA level was observed at any time point (Supplementary Fig. S4D). It is significant that PCC rapidly decreased cFLIP protein in cancer cells, yet this decrease was blocked by the proteasome inhibitor bortezomib, indicating that PCC increases the proteasomal degradation of cFLIP protein (Fig. 2E). In contrast levels of Mcl-1 (another antiapoptotic protein rapidly degraded by the proteasome) were not affected by PCC treatment. Thus, PCC treatment does not result in a generalized increase in proteasomal degradation of all proteins with a short half-life, but displays some selectivity for cFLIP. It has been reported that the proteasome degradation of cFLIP_L and cFLIP_S is mediated via the E3 ubiquitin ligase ITCH in some cells (27). ITCH depletion in the cells tested above had no effect on apoptosis, suggesting that control of cFLIP levels was independent of ITCH (Supplementary Fig. S4E).

Proximal cell signaling pathways in response to TRAIL and poly I:C are quite different. However, downstream assembly of a similar multiprotein signaling complex, the ripoptosome, is critical for the induction of apoptosis (28). RIP1, FADD, caspase-8, and cFLIP are core ripoptosome components. Ripoptosome formation, as assessed by FADD immunoprecipitation, was enhanced by combinations of PCC plus poly I:C (Fig. 2F) or PCC plus TRAIL (Fig. 2G). Procaspase-8 immunoprecipitation also demonstrated increased ripoptosome formation following combination treatment (Supplementary Fig. S5). Ripoptosome assembly is known to be inhibited by both cFLIP and the antiapoptotic inhibitors of apoptosis proteins (IAP; refs. 29, 30). Yet on PCC treatment, no major consistent changes in levels of most IAPs or Bcl-2 family members were observed. Because antagonists to IAPs and the antiapoptotic Bcl-2 family members are in clinical trials, we compared these inhibitors to PCC for their ability to enhance caspase-8–dependent apoptosis signaling. For a panel of renal carcinoma cell lines, PCC was an excellent apoptosis sensitizer to TRAIL-mediated apoptosis. In contrast, IAP antagonists had little activity, while Bcl-2 antagonists were only active at the highest concentration (2,000 nmol/L) tested (Supplementary Fig. S6A). For a panel of human melanoma cells, the pattern was somewhat different (Supplementary Fig. S6B). PCC efficiently sensitized all the cell lines tested. Some of the lines (M14 and SK-MEL-28) were also efficiently sensitized to TRAIL-mediated apoptosis by the IAP antagonist birinapant. Because M14 and SK-MEL-28 cells are quite sensitive to TRAIL as a single agent, it is possible that removing either cFLIP or IAPs is sufficient to induce apoptosis. Other melanoma cell lines, such as 888 and 1383, are much more resistant to TRAIL. For these cells PCC was a superior sensitizing agent than either IAP or Bcl-2 antagonists, suggesting that cFLIP may be the dominant antiapoptotic protein in these melanoma cells. The Bcl-2 antagonist ABT737 had little or no effect on melanoma apoptosis.

To investigate the therapeutic potential of a combination of PCC and poly I:C in vivo, we established mouse xenograft models of the SK-MEL-28 and M14 human melanomas. Although an initial experiment demonstrated that PCC plus poly I:C combination treatment showed a superior therapeutic benefit in SK-MEL-28 xenografts, these xenografts were very difficult to establish in immunodeficient mice and grew extremely slowly, thus M14 xenografts were used for subsequent experiments. Intratumoral administration of PCC or poly I:C as single agents showed some benefit in M14 xenografts, but a combination of PCC and poly I:C provided a superior therapeutic effect, resulting in complete tumor regression in 90% of the mice (Fig. 3A). This regression was due to extensive melanoma cell death as assessed by TUNEL staining of tumor tissue sections (Fig. 3B; Supplementary Fig. S7A). Furthermore, staining of sequential tissue sections for caspase-8 demonstrated a strong overlap between terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining and enhanced cytoplasmic staining of caspase-8 (Fig. 3C). In addition, hematoxylin and eosin (H&E) staining of tissue sections showed large areas with characteristic morphologic features of apoptosis in tumors of mice treated with the PCC plus poly I:C combination (Supplementary Fig. S7B). On whole-slide image analysis of all tumors from each group, a good correlation was seen between TUNEL staining and cytoplasmic caspase-8 staining. Thus, tumor cell death in vivo occurs by apoptosis, accompanied by high levels of cytoplasmic caspase-8 activity, consistent with the previous in vitro data. Although differences in tumor volume between PCC and PCC plus poly I:C-treated mice were not significant at 7 weeks after treatment, a prolonged follow-up for over 3 months of mice treated with PCC and poly I:C showed no further signs of tumor, consistent with a complete and sustained tumor regression. In these immunodeficient mice, it is likely that the observed tumor regression was predominantly due to a direct induction of apoptosis in the human melanoma cells.

To determine therapeutic effects of PCC in mice with an intact immune system, we utilized the B16F10 melanoma model in B6 mice. PCC led to a decrease in cFLIP levels in B16F10 cells in vitro (Fig. 4A) as seen with human melanoma cells. To our surprise, in contrast to most of the human melanoma lines, B16F10 cells were refractory to PCC and poly I:C-mediated cell death in vitro (Supplementary Fig. S8A). Interestingly, we could not detect TLR3 expression in B16F10 cells by Western blotting. Nonetheless, following PCC treatment, B16F10 were rendered very sensitive to apoptosis in response to a combination of the immune cytokines IFNγ and TNFα (Fig. 4B). Because poly I:C treatment can result in the immune activation and increased cytokine production in vivo, we investigated the effect of the PCC and poly I:C combination in the B16F10 melanoma model. Intratumoral injection of PCC followed administration of poly I:C either intraperitoneally (Fig. 4C; Supplementary Figs. S8B and S8C) or intratumorally led to more significant tumor regression in the combination group. Because B16 F10 mouse melanoma cells are resistant to the direct apoptotic effects of the PCC plus poly I:C combination, the therapeutic benefit of poly I:C in vivo is likely indirect, depending on enhanced immune cytokine production occurring following either intratumoral or intraperitoneal administration of poly I:C. The relatively good response to PCC alone may be the result of low levels of immune cytokine production in these mice. To identify a critical role for cFLIP, we reduced cFLIP levels in B16F10 cells using CRISPR/Cas9. Gene editing efficiency measured by indel frequency was 70% in C7, and close to 100% in C50 cells (Fig. 4D; Supplementary Fig. S9A). This genetic analysis suggests that C50 cells were derived from a monoclonal population where all the three copies of the cFLIP gene found in B16F10 melanoma cells were successfully disrupted. In contrast, C7 cells are likely derived from a heterogeneous oligoclonal population where in some cells not all the cFLIP genes were disrupted. Western blotting confirmed significant decreases in cFLIP proteins in both C7 and C50 cells (Fig. 4E). Furthermore, this cFLIP depletion was associated with an increased susceptibility to apoptosis on exposure of cells to the combination of IFNγ and TNFα (Supplementary Fig. S9B), while editing of cFLIP had no major effect on cell proliferation in vitro (Supplementary Fig. S9C). Following injection into B6 mice, the partially edited C7 cells could form melanomas in a similar manner to control unedited C71 cells, likely due to the preferential outgrowth in vivo of the cells in which cFLIP disruption was incomplete. It is notable that even in the absence of any therapy, C50 cells were less efficient in establishing a melanoma in B6 mice. Also, the melanomas that developed following injection of C50 cells responded very well to immunotherapy (Fig. 4F), and 4 of 5 mice (80%) were still tumor free at 21 days posttherapy (Supplementary Fig. S9D). This would all be consistent with cFLIP levels in B16F10 cells being very important for tumor establishment, progression and resistance to immunotherapy in B6 mice.

To confirm the underlying effects and critical role of cFLIP reduction in PCC-mediated responses, we utilized R331 murine renal carcinoma cells. PCC-mediated responses were compared between cFLIP-overexpressing R331 cells (R331-FLIP) and vector control cells (R331-VC), both in vitro and in vivo. PCC treatment dramatically reduced numbers of R331-VC, but not R331-FLIP cells, in response to murine TNFα in vitro (Fig. 4G). Furthermore, in a lung metastases model, treatment of R331-VC tumor-bearing mice with PCC or PCC plus poly I:C significantly reduced numbers of lung metastases. R331-FLIP generated fewer metastases likely reflecting their slower growth rate in vitro, nonetheless neither PCC nor PCC plus poly I:C treatment had a significant effect on numbers of metastases (Fig. 4H). These data confirm that PCC reduction of cFLIP is crucial for its molecular mechanism of action both in vitro and in vivo.

To directly examine whether PCC could provide any benefit in other immunotherapies where CD8⁺ T cells play a major role (such as in adoptive or CAR T-cell transfer), we analyzed the effect of PCC on an in vitro T-cell cytotoxicity assay. Melanoma cells treated with PCC were incubated together with anti-CD3 and CD28 activated T cells, after which, residual viable adherent melanoma cells were quantified by crystal violet staining. PCC sensitized the melanoma cells to T-cell–mediated killing (Fig. 5A). Direct contact between the T cells and melanoma cells was not critical, because cell-free supernatants from the activated T cells were equally effective in reducing melanoma cell numbers (Fig. 5B and C). Addition of neutralizing antibodies to TNFα to the cell-free T-cell supernatants reduced or completely blocked the cytotoxic effects. Encouragingly, PCC did not affect survival of either resting or activated T cells, even in the presence of high in vitro concentrations used in these experiments (Fig. 5D). Thus, PCC selectively sensitized the melanoma cells to undergo apoptosis in response to TNFα and possibly other death ligands produced by activated T cells. To directly examine the relevance of these in vitro findings, we established an in vivo model using the injection of M14 melanoma cells into the immunodeficient NOD/Scid/Il-2rg^−/− (NSG) mice (Fig. 5E). Assessment of tumor volume at 7 days after the final PCC treatment showed that administration of PBMC or PCC alone had little effect on tumor progression, whereas the combination treatment demonstrated a highly significant therapeutic benefit (Fig. 5F). Therefore, PCC administration likely enhances the immune-mediated destruction of the melanoma cells by the transferred PBMCs.

To gain more insight into the molecular mechanism of action of PCC several biotinylated versions of PCC were synthesized. Introduction of a biotin moiety at the C4 position to generate 4β-biotinylpentylaminocarbamoyloxy-PCC (Bt PCC; Fig. 6A; Supplementary Fig. S10A) had no adverse effect on biological activity (IC₅₀ = 85 ± 4 nmol/L) compared to that of PCC (IC₅₀ = 91 ± 6 nmol/L). This Bt PCC was bound to streptavidin beads that were then incubated with total cell extracts of M14 cells. Specific binding of 3 major proteins to the Bt PCC-modified streptavidin beads was observed (Fig. 6B). Mass spectrometry analysis of these 3 protein bands identified BRD2, BRD3 and BRD4 as the major protein components (Fig. 6C). Western blotting further confirmed the identity of the major individual BET proteins associated with each band (Fig. 6D). To determine how the interaction of PCC with BET proteins might affect their biological activities, we performed RNA-sequencing analysis of M14 cells treated with PCC. Interestingly, PCC treatment had no detectable effect on mRNA expression of BRD4, yet the expression of mRNAs of multiple genes reported to be regulated by BRD4 were significantly decreased (Supplementary Fig. S10B and S10C). Thus, it is highly likely that PCC treatment inhibits the biological activity of the BRD4 protein. As expected, no major change of cFLIP mRNA was observed in the RNA-sequencing analysis. However, analysis of the mRNAs of various IAPs showed a significant reduction of Livin (Supplementary Fig. S10D), and a decrease in Livin protein, following PCC treatment, was also confirmed by Western blotting (Supplementary Fig. S10E). Thus in some melanoma cells, PCC treatment could have the additional benefit of enhancing caspase-8–dependent apoptosis signaling by decreasing both cFLIP and Livin. To validate a role for BET proteins, specific siRNAs to BRD2, BRD3, and BRD4 as well as cFLIP (as a positive control) were introduced into cells. Interestingly, siRNA to BRD4 significantly sensitized the melanoma cells to poly I:C-mediated apoptosis (Fig. 6E and F), whereas siRNAs to both BRD2 and BRD4 significantly sensitized ACHN cells to TRAIL-mediated apoptosis (Fig. 6G). Western blotting demonstrated an efficient reduction of cFLIP protein levels in all cell lines, while siRNAs for both BRD2 and BRD4 resulted a partial reduction in protein levels (Supplementary Fig. S10F). A known BET inhibitor, (+)-JQ1, could also effectively sensitize renal carcinoma and melanoma cells to caspase-8–dependent apoptosis, albeit with a higher IC₅₀ than PCC (Supplementary Fig. S10G). Furthermore, (+)-JQ1 treatment also resulted in decreased levels of cFLIP as indicated by Western blotting, which correlated well with apoptosis sensitization of these cells (Supplementary Fig. S10H).

We have identified PCC as a potent NP-based agent that can sensitize various human melanoma and renal carcinoma cells to undergo apoptosis in response to TNF family death ligands and the viral mimetic poly I: C. Increased degradation of cFLIP protein levels by PCC seems sufficient to explain its apoptosis-sensitizing effects. Apoptosis in response to both death ligands and poly I:C signaling converges on a common downstream protein complex, the ripoptosome, that signals caspase-8–dependent cell death. Antiapoptotic proteins such as cFLIP and IAPs are known to inhibit ripoptosome formation, thus effectively blocking apoptosis signaling from this multiprotein complex. Elevated levels of antiapoptotic cFLIP, IAPs, and Bcl-2 family proteins are described in many cancers, and several reports suggest apoptosis-sensitizing effects of specific antagonists of antiapoptotic IAPs and Bcl-2 family members (31–33). Thus, for some cancers a reduction of two or more antiapoptotic proteins may be required for optimal sensitization (34, 35). PCC was not effective in sensitizing a panel of normal nontransformed cells to undergo caspase-8–dependent apoptosis in response to either poly I:C or TRAIL. For nontransformed cells, lower levels of proapoptotic proteins and high levels of redundancy in apoptosis resistance pathways may protect them from caspase-8–dependent apoptosis. Encouragingly, even when two antiapoptotic proteins are reduced, normal nontransformed cells still remain resistant to apoptosis (36), suggesting a more complex control of extrinsic apoptosis signaling in nontransformed cells.

We did not observe any overt toxicities following administration of PCC and poly I:C, at doses that provided a therapeutic benefit in two mouse models of melanoma. There was no significant weight change in treated mice, and no pathologic changes were observed on examination of multiple organs. Nonetheless, to date only limited pharmacologic studies have been performed with PCC. TLR3 expression is often elevated in human melanoma cell lines and melanoma cells isolated from single-cell suspensions obtained from tumor biopsies (37). Thus, it is tempting to speculate that for these melanomas exhibiting elevated levels of TLR3, local administration of PCC and poly I:C could provide a therapeutic benefit by directly promoting melanoma cell death. In addition, in these melanomas, cell death could also be enhanced indirectly by increased levels of IFNγ and death ligands in response to the immunoadjuvant effects of poly I:C. For cancer cells resistant to the direct effects of poly I:C, therapeutic benefit of PCC plus poly I:C would be dependent on the enhanced levels of death ligands such as TRAIL, FasL, or TNF produced in vivo following exposure to poly I:C. With regard to poly I:C, both local and systemic delivery may offer some additional therapeutic benefits (38–40). There has recently been renewed interest in intratumor delivery of poly I:C, and a clinical trial of intratumor administration of a nanoplexed poly I:C (41) is currently underway. Our in vitro and in vivo studies also suggest that PCC can enhance cancer cell apoptosis in response to multiple TNF family death ligands produced by activated immune cells, as observed following adoptive cell transfer of activated human PBMCs. Therefore, administration of PCC together with multiple immunotherapeutic strategies may prove beneficial.

In conclusion, pharmacologic reduction of cFLIP in cancer cells by the small-molecule natural product PCC may provide a therapeutic advantage for various immunotherapies by preferentially promoting caspase-8–dependent apoptosis in cancer cells. It is noteworthy that to date no highly selective small-molecule antagonist of cFLIP has been described. Indeed, specific targeting of the antiapoptotic cFLIP without adversely affecting the expression of structurally related proapoptotic caspase-8 may prove difficult (42). Remarkably, high levels of BRD4 expression are associated with a poor prognosis in patients with melanoma (43). Also, inhibition of BRD4 has been reported to decrease proliferation and increase apoptosis in various cancer cells (44–46), and there is currently much interest in utilizing BET inhibitors in cancer therapy (47). The BET inhibitory activity and selective cFLIP reduction in cancer cells, together with the encouraging preclinical data, warrant development of this class of small-molecule NPs for use in combination with various immunotherapeutic approaches for cancer treatment.

P. Tewary reports a patent for 10849912 issued and a patent for US20210047363A1 pending. A.D. Brooks reports a patent for 10849912 issued and a patent for US20210047363A1 pending. Y. Xu reports a patent 10849912 issued, a patent US20210047363A1 pending, and a patent 20110230551 pending. E. Wijeratne reports a patent for 10849912 issued and a patent for US20210047363A1 pending. C.J. Henrich reports a patent for 10849912 issued. A. Gunatilaka reports grants from Arizona Biomedical Research Center during the conduct of the study; other support from Sun Pharma Advanced Research Co. Ltd., India and grants from Regulonix, LLC., outside the submitted work; in addition, A. Gunatilaka has a patent 10849912 issued and a patent for US20210047363A1 pending. T.J. Sayers reports a patent for 10849912 issued and a patent for US20210047363A1 pending. No disclosures were reported by the other authors.

P. Tewary: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. A.D. Brooks: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. Y.-M. Xu: Formal analysis, investigation, methodology, writing–original draft. E.M.K. Wijeratne: Formal analysis, investigation, methodology, writing–original draft. A.L. Babyak: Data curation, formal analysis, investigation, writing–original draft. T.C. Back: Data curation, formal analysis, validation, investigation, writing–original draft. R. Chari: Resources, data curation, formal analysis, investigation. C.N. Evans: Resources, data curation, formal analysis, investigation. C.J. Henrich: Resources, data curation, formal analysis, validation, investigation. T.J. Meyer: Resources, data curation, formal analysis, investigation. E.F. Edmondson: Resources, data curation, formal analysis, investigation. M.T.P. de Aquino: Data curation, formal analysis, Investigation. T. Kanagasabai: Data curation, formal analysis, investigation. A. Shanker: Resources, data curation, formal analysis, writing–review and editing. A.A.L. Gunatilaka: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing. T.J. Sayers: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing.

This study was funded in whole or in part with federal funds from the NCI, NIH, under contract HHSN26120080001E (awarded to T.J. Sayers). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the U.S. Government. This work was also supported by a grant from the Arizona Biomedical Research Center (grant number ADHS-16–162515; awarded to A.A.L. Gunatilaka) and (in part) by the Intramural Research Program of NIH, Frederick National Lab, Cancer and Inflammation program (CIP), Center for Cancer Research (CCR), and by funds from the NIH grants (grant numbers U54 CA163069 and SCI CA182843; awarded to A. Shanker). The authors thank Dr. Julio Valencia (CIP, NCI Frederick) for his scientific suggestions, discussions, and valuable comments, Ms. Anna Trivett for her technical assistance, Ms. Roberta Matthai for FACS single-cell sorting of cells, Ms. Nirmala Sharma for her assistance with cFLIP blots, and Ms. Megan Karwan for her assistance with in vivo studies. They would also like to thank Drs. Maggie Cam and Jack Chen (CCR Collaborative Bioinformatics Resource, NCI) and Steven Anderson (CIP, NCI Frederick) for their help with the analysis of RNA-sequencing data, Drs. Thorkell Andresson and Sudipto Das (Protein Characterization Lab, FNL) for their assistance in the mass spectrometric identification of cellular target proteins. Mass spectrometric and proteomics data for identification of cellular target proteins were also acquired by Drs. Krishna Parsawar and Cynthia David of the University of Arizona Proteomics Consortium (supported by NIEHS grant ES06694 to the SWEHSC, NIH/NCI grant CA023074 to the UA Cancer Center and by the BIO5 Institute of the University of Arizona). The authors are also grateful to Drs. Marc Hennequart and Benoit van den Eynde, Ludwig Cancer Institute, Brussels, Belgium, for their helpful advice on adoptive cell transfer in NSG mice, Dr. Esteban Celis (Georgia Cancer Center, GA) for his critical reading of the manuscript, and Andrew Sayers for his assistance with the artwork.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.