NF-κB Blockade with Oral Administration of Dimethylaminoparthenolide (DMAPT), Delays Prostate Cancer Resistance to Androgen Receptor (AR) Inhibition and Inhibits AR Variants

This article is featured in Highlights of This Issue, p. 1097

Worldwide, there are approximately 300,000 deaths from metastatic castration-resistant prostate cancer (mCRPC; ref. 1). Strategies that prevent emergence of clones that are resistant to androgen-deprivation therapy (ADT) may augment the efficacy of AR-targeted therapy, whether it be testosterone suppression alone or with abiraterone or new generation potent direct androgen receptor (AR) inhibitors such as enzalutamide, apalutamide or darolutamide. This in turn may decrease the number of men who relapse after adjuvant ADT for localized disease and/or prolong the survival of men with metastatic hormone-sensitive prostate cancer (mHSPC). Advances identifying the aberrant biology driving resistance to therapy and poor clinical outcomes include aberrant AR activation from AR upregulation, AR splice variants such as AR-V7 and non-androgen ligands such as cytokines IL-6 (2–4). There are currently no approved agents that can prevent androgen-mediated AR upregulation or constitutive AR activation in CRPC, and this remains a compelling strategy to prevent therapy resistance.

Activation of nuclear factor kappa B (NF-κB) is involved in a range of biological processes such as cell proliferation, inflammation, and cell invasion, and is strongly associated with oncogenesis (5, 6). NF-κB activation is driven through either canonical or non-canonical pathways. Canonically, NF-κB activation is dependent on the IκB kinase complex (IKK), whereby, IκBα is phosphorylated by IKK in an IκKβ-dependent manner, which results in the degradation of IκBα and the release of the p65/p50 dimer to gather in the nucleus (7). Phosphorylation of p65 at Ser536 is mediated by IKK and is required for canonical NF-κB–dependent cellular activity (8). In the noncanonical NF-kB cascade, RelB/p100 heterodimers are processed to RelB/p52 heterodimers in an IKKα-dependent manner. Activation of the canonical NF-κB pathway is most commonly implicated in oncogenesis.

NF-κB is a prime target for inhibition to augment ADT efficacy. Overexpression of AR-V7 may also feedback and enhance NF-κB signaling resulting in bidirectional positive interactions (9–11). Hence, combination of potent and safe NF-κB inhibitors with anti-androgen drugs is likely to be an effective strategy for the management of CRPC and to prevent or at least delay progression of hormone-sensitive prostate cancer (HSPC) to mCRPC. The relevance of targeting NF-κB signaling is further increased given the development of parthenolide analogues, which are NF-κB inhibitors and have entered clinical testing. Dimethylaminoparthenolide (DMAPT) is a bioavailable parthenolide analogue, which inhibits NF-κB by binding to the cysteine residue of IκKβ and prevents one of its subunits, p65, binding to DNA, thereby inducing cell-cycle arrest and promoting apoptosis of cancer cells (12). DMAPT largely blocks the canonical NF-κB pathway by binding directly to and inhibiting IκKβ (13–15) and has been shown to have no meaningful impact on the non-canonical. Kwok and colleagues (13) showed that inactivation of IκKβ abolished sensitivity of parthenolide (of which DMAPT is a derivative compound). In humans, DMAPT has demonstrated linear pharmacokinetics and no toxicity was observed with doses that were able to achieve low micromolar serum concentrations and decrease NF-κB in circulating blasts. It is also notable that DMAPT has been shown to enhance radiation anticancer activity in prostate cancer (16, 17), and non–small cell lung cancer (18, 19) and chemotherapy in bladder cancer (20) while mitigating toxicity to normal tissue (21, 22).

In this article, we provide novel insights into strategies to block NF-κB–mediated resistance by inhibition of NF-κB and the consequent AR upregulation and AR-V7 with an orally administered agent that is being advanced into prostate cancer clinical trials.

Animal experiments were approved by and performed in accordance with the guidelines of Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee (Animal protocol #102–2015). VCaP-CR cells were implanted subcutaneously into the flank of donor male ICR-SCID mice (Taconic Biosciences) 7–8 weeks of age as a 50% Matrigel/PBS suspension and allowed to grow up to 2,000 mm³ at which point the tumors were harvested and processed into tissue pieces approximately 2–3 mm in diameter. Resected tumor pieces were subcutaneously implanted into the flank of 7-week-old experimental male ICR-SCID mice. Tumors were measured with calipers thrice weekly until tumors reached 200 mm³, at which point mice were randomly sorted into four treatment groups: vehicle, castration (surgical), DMAPT (100 mg/kg in H₂O, by oral gavage Q.D.) or castration plus DMAPT. Mice were treated and tumors measured until mice reached a humane tumor endpoint or the study period ended. DMAPT was obtained from Dr. Peter Crooks (University of Arkansas for Medical Sciences, Little Rock, AR) and stored at −20°C. On the day of treatment, DMAPT was made up to 10 mg/mL in sterile water.

VCaP-CR cells were a gift from Stephen Plymate (University of Washington). FVB-MyCCaP cells were purchased from the ATCC (ATCC CRL-3255). VCaP-CR and LNCaP-95 cell lines were cultured in phenol-free RPMI-1640 Medium (Gibco) supplemented with 10% charcoal-stripped FBS (Sigma-Aldrich) and 1x GlutaMAX (Gibco). 22Rv1 cells were cultured in RPMI-1640 Medium (ATCC 30–2001) supplemented with 10% FBS (Sigma-Aldrich). FVB-MyCCaP cells were cultured in DMEM (Gibco) supplemented with 10% FBS (Sigma-Aldrich) and 1x GlutaMAX (Gibco). Cell lines were regularly screened for Mycoplasma contamination using the LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich).

Sub-confluent–treated cells were washed twice with cold PBS, trypsinized, and then lysed in Pierce RIPA buffer (Thermo Fisher Scientific, 89900) with PhosSTOP inhibitor cocktail (Sigma-Aldrich, PHOSS-RO) at 4°C for 30 minutes. Protein concentrations were measured by a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). Proteins were separated by SDS-PAGE (10% Mini-PROTEAN TGX Precast Gel, Bio-Rad, 4561036) and transferred to a PVDF membrane (Bio-Rad, 1620177). The membrane was blocked 3% BSA in TBST for 1 hour at room temperature and then blotted with primary antibodies overnight at 4°C. Full list of antibodies available in Supplementary Data. Blots were washed 3 × 5 minutes with TBST. The blots were then incubated with fluorescent-conjugated secondary antibody (Bio-Rad) 1 hour at room temperature and washed 3 × 5 minutes with TBST. Proteins were visualized with a ChemiDoc MP fluorescent imager (Bio-Rad). Bio-Rad Image Laboratory analysis software was used to quantify protein band density. All western blots were repeated once.

Prostate tumor cells (cell lines or dissociated tumors) were collected, RNA was prepared using TRizol reagent (Invitrogen) and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). cDNA was synthesized (iScript cDNA synthesis kit; Bio-Rad) and used as a template for qPCR reactions using SsoFast qPCR supermix (Bio-Rad). Reactions were performed using a Bio-Rad CFX96 Touch Real-Time PCR thermocycler, using an annealing temperature of 60°C and 40 cycles. List of primer sets available in Supplementary Methods Table S1. Results were analyzed by the 2^−ΔΔCt method relative to β-actin as a control gene.

For IHC, 4-μm thick sections were cut from paraffin-embedded blocks and dried onto positively charged microscope slides. Slides were prepared and stained using the ImmPRESS HRP Anti-Mouse IgG (Peroxidase) Polymer Detection Kit (Vector Laboratories) as previously described (23).

For analysis, tissue sections were imaged using an EVOS M7000 automated microscope (Thermo Fisher Scientific). Images were de-identified and 50 random fields per section were run through QuPath image analysis software (24) to quantify positive DAB-stained cells (as a percentage of total cells).

Prostate tumor cells (cell lines or dissociated tumors) were collected, RNA was prepared using TRizol reagent (Invitrogen) and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). RNA sequencing data from VCaP tissues were aligned with STAR to human reference genome GRCh37 (hg19) and quantified using RSEM to generate gene-level expression in raw counts and transcripts per million (TPM), and isoform-level expression. Two samples were removed due to failure of meeting sequencing and alignment metrics. Count data were first modeled to demonstrate fit with the negative binomial distribution. Samples underwent hierarchical clustering in quality control using principal component analysis, and one outlier sample was manually assessed and removed prior to downstream analysis.

Stable cell lines of LNCaP-95, 22Rv1, and FVB-MyCCaP-overexpressing NF-κB2/p65 were generated by transfection of pCMV4 p65 plasmids and selection of clones after application of selective pressure with antibiotics. pCMV4 p65 was a gift from Warner Greene (Addgene plasmid # 21966; http://n2t.net/addgene:21966; RRID:Addgene_21966). LNCaP-95, 22RV-1 and FVB-MyCCaP cell lines were generated by transfection of pBabe-Puro-IKBalpha-mut plasmids and selection of clones after selective pressure with antibiotics. pBabe-Puro-IKBalpha-mut (super repressor) was a gift from William Hahn (Addgene plasmid # 15291; http://n2t.net/addgene:15291; RRID:Addgene_15291). Target protein expression in generated cell lines was confirmed by western blot.

LNCaP-95 and 22Rv1 cells were cultured in 3D in 100% Matrigel droplets. Once tumoroids had established (2–3 days), whole tumoroids were removed from Matrigel by incubation with 2 mg/mL dispase (Thermo Fisher Scientific, 17105041) and re-plated in Matrigel in 96-well tissue culture plates (20–30 tumoroids per well). Cultures were treated for 72 hours with singlet or doublet therapy of DMAPT (5 μmol/L) and/or enzalutamide (10 μmol/L), or DMSO vehicle control. After therapy period, media were removed and tumoroid cultured washed with PBS and dead versus live cells visualized using ReadyProbes Cell Viability Imaging Kit (Thermo Fisher Scientific, R37609). Stained cultures were imaged using an EVOS M7000 automated microscope (Thermo Fisher Scientific) and the percentage cell viability was quantified using CellProfiler image analysis software (25), 10 tumoroids per treatment group were used for analysis.

All statistical calculations were carried out in GraphPad (Prism). Statistical significance was determined by performing an unpaired two-tailed Student t test unless otherwise noted. Comparisons between treatment arms in the Kaplan–Meier analysis of the in vivo study were carried out using a Log-rank (Mantel-Cox) test to compare time-to-1 cm³. A P value of less than 0.05 was considered as significantly different from the null hypothesis across all experiments as indicated by asterisks in all figures.

In vitro, we observed that VCaP-CR cells under pressure from prolonged enzalutamide treatment, AR inhibition showed a marked increase in expression of total and activated NF-κB (phosphorylated-p65) expression (Fig. 1A). This observation matches other examples in the literature where castration resistance is linked to NF-κB signaling (11, 26). This increase in NF-κB activation was counteracted by NF-κB inhibitor DMAPT. Given that activation of NF-κB signaling has been observed to correlate with prostate cancer progression and resistance to AR inhibition, we hypothesized that the combination of DMAPT and castration would provide significant tumor control in vivo. Previous work has documented that DMAPT acts by blocking p65 nuclear translocation and DNA binding (20, 27). VCaP-CR tumor-bearing mice, a prostate cancer cell line with some well-documented castration resistance (CR), were treated with either DMAPT or vehicle control with and without surgical castration (Fig. 1B; Supplementary Fig. S1). DMAPT alone did not significantly slow VCaP-CR tumor growth compared with vehicle treatment (median time for tumors to reach 1 cm³: 34.2 ± 6.5 weeks and 34 ± 6.6 weeks, respectively). Castration provided a moderate level of tumor control (50.2 ± 4.7 weeks); however, the most significant reduction in tumor growth was achieved when mice were castrated in combination with DMAPT (P = 0.001 compared with vehicle, P = 0.0053 compared with castration alone). When treated in this combinatorial manner, remarkably only two of six tumors exceeded 1 cm³ by the end of the study. Given that DMAPT alone provided minimal tumor control, this indicates a strong augmentation of antitumor effect of the combination therapy. IHC analysis (Fig. 1C,–D) of end-point tumors showed that surgical castration alone increased phosphorylated-p65 expression. We also observed significant increases in AR and AR-V7 protein expression following castration, both of which were reduced by DMAPT treatment. In the absence of castration, DMAPT still provided some reduction in AR and AR-V7 expression. The combination of castration and DMAPT treatment also resulted in a significant loss of Ki-67 in tumors compared with vehicle-treated tumors and a trend toward increased cleaved-caspase 3 (Supplementary Fig. S2). Similar decreases in levels of mRNA AR-FL and AR variant expression were observed, with DMAPT treatment particularly providing the more pronounced suppression of AR-V7 expression (Fig. 1E). This may indicate that changes in total AR expression are largely due to AR-V7 changes, rather than AR-FL. RNA-seq performed using end-point tumors validated our IHC results by indicating that castration and DMAPT combination resulted in significantly lower AR-V7–associated gene expression (Fig. 1F; ref. 28). Our observation of NF-κB–driven upregulation of AR-V7 aligns with previous accounts by other investigators that activation of NF-κB can drive the emergence of castration-resistant disease (11, 29).

To confirm that the observed co-upregulation of p65 NF-κB and AR/AR-V7 was not limited to the VCaP-CR cell line, we cultured mouse and human prostate cancer cell lines (FVB-MyCCaP, LNCaP-95, and 22Rv1, respectively) that are known to express AR-V7 or AR-variants (Fig. 2A; refs. 30, 31) in enzalutamide. All three enzalutamide-treated lines showed significantly increased phosphorylated-p65 expression compared with the respective parental lines, along with increased AR and AR-variant expression. As was observed in prior VCaP-CR experiments, DMAPT treatment reversed AR and AR-V7 expression. Because DMAPT is also known to have biological effects that are not NF-κB related, including induction of reactive oxygen species and STAT inhibition (32), we compared the effect of DMAPT treatment with genetic inhibition of NF-κB by generating cell lines with stable expression of an IκBα super repressor (33), which sequesters NF-κB in the cytoplasm. Both DMAPT treatment and the IκBα super repressor slowed cell proliferation relative to parental cell lines to a similar extent by 72 hours (LNCaP95 parental: P = 0.0068 compared with DMAPT and P = 0.0027 compared with IκBα SR; 22Rv1: P = 0.0052 compared with DMAPT and P = 0.0011 compared with IκBα SR; FVB-MYCCaP: P = 0.0207 compared with DMAPT and P = 0.0194 compared with IκBα SR, Fig. 2B) and there was no significant difference in AR-variant mRNA expression between the chemical and genetic inhibitors (Fig. 2C). This indicates that the anti–AR-variant effect of DMAPT is largely NF-κB dependent. Given that the canonical NF-κB pathway, rather than non-canonical, is most commonly activated in cancer pathology, we hypothesize that NF-κB–driven upregulation of AR-variants is largely occurring through this canonical signaling cascade. As such we set out to test the reliance of this mechanism on p65.

To validate the dependence of the action of DMAPT on p65, we performed rescue experiments by generating cell lines stably expressing p65 using human LNCaP-95 and 22Rv1 prostate cancer lines. In both cell lines, constitutive overexpression of p65 resulted in significant upregulation of AR-V7 protein expression compared with the parental and empty vector–containing cell lines (Fig. 3A). In addition to upregulating AR-V7 protein expression, constitutive expression of p65 resulted in increased mRNA expression of variant and full-length transcripts (Fig. 3B). The ability of DMAPT to reduce AR-V7 was significantly blunted in the cell lines constitutively overexpressing p65, indicating a potentially crucial role for p65 in NF-κB–driven castrate-resistant disease. Expression of AR-V7 and AR-FL was not significantly different in the empty vector–containing cell lines compared with the parental lines (Supplementary Fig. S3). EDN2 and ETS2 have previously been shown to be specifically upregulated by AR-V7 (34). We compared basal expression of both genes with and without p65 overexpression (Fig. 3C). In both LNCaP-95 and 22Rv1 lines, constitutive p65 expression resulted in upregulation of END2 and ETS2. Interestingly, overexpression of p65 resulted in almost a 10-fold increase in expression of both genes in 22Rv1 cells, which might be due to the cell line having significantly higher basal expression of AR-V7, compared with LNCaP-95. In R1881-stimulated cells, DMAPT treatment significantly reduced expression of AR-target genes, FKBP5 and KLK3 (Fig. 3D).

Constitutive expression of p65 inhibits the antitumor action of DMAPT alone or in combination with castration.

Using the constitutively expressing p65 LNCaP-95 and 22Rv1 prostate cancer cell lines cultured in 3D, we performed in vitro therapy experiments where tumoroid cultures were treated with DMAPT and/or enzalutamide as singlet or doublet therapy (Fig. 4; Supplementary Fig. S4). In empty vector–containing cell lines, DMAPT and enzalutamide singlet treatment reduced cell survival by 47% to 66% and combination therapy reduced cell survival by 73% to 77% compared with DMSO vehicle. This result in two in vitro prostate cancer tumoroid culture models matches the VCaP-CR in vivo data showing significant reduction in tumor growth in mice treated with DMAPT and surgical castration (Fig. 1B). Although some reduction of cell survival was still observed, constitutive overexpression of p65 significantly blunted antitumor activity of both singlet and doublet therapies. This result confirms the importance of the p65 NF-κB subunit for modulating the restoration of castration-sensitivity by DMAPT in prostate cancer cells.

ADT has been the main treatment strategy for HSPC but unfortunately after an initial response to AR inhibition, progression to the highly aggressive state of mCRPC is often inevitable. Several mechanisms have been identified as key driving forces in resistance to ADT and the development of mCRPC, including, AR amplification, phosphorylation of AR, and expression of AR-V7 (35, 36). It is well documented that treating mCRPC with single-agent therapies has limited durable benefit. Interestingly, it has previously been demonstrated that activation of the AR and splice variant signaling axis by p52 (non-canonical NF-κB signaling) can induce enzalutamide resistance (37), whereas here we have demonstrated that AR inhibition with enzalutamide or castration drives canonical NF-κB activation and subsequent AR-V7 upregulation. Recent work by Liang and colleagues (38) has shown that IκKβ blockade prevented the emergence of enzalutamide resistance in prostate cancer in vivo models. The work by Liang and colleagues identify both BCL-2 and IκKβ dependencies in clinically relevant enzalutamide-resistant prostate cancer cells in vitro and in vivo and furthermore indicate the greater relevance of IκKβ upregulation in the progression of human castrate-resistant prostate cancer. This further highlights the therapeutic benefit of DMAPT, which acts largely through the canonical NF-κB pathway, in reversing resistance to castration.

Significantly, our data demonstrate that the expression of AR-V7, and to a lesser degree AR-FL, was significantly reduced in CRPC cells following DMAPT treatment, suggesting that DMAPT-mediated reduction in AR-V7 expression restores CRPC responsiveness to ADT. Therefore, we would suggest that using a treatment plan combining DMAPT and AR antagonism might increase therapeutic efficacy and decrease resistance to AR inhibition in both HSPC and mCRPC. It is worth noting that DMAPT, and its less soluble parental compound parthenolide, have both also previously been demonstrated to be a novel inhibitor of prostate cancer growth (22, 39). NF-κB has been an appealing target for prostate cancer therapy for many years and numerous studies have shown some anticancer efficacy of NF-κB inhibitors. However, low bioavailability and toxicity at higher concentrations of such phytochemicals and proteasome inhibitors is a major challenge for clinical translation (40–42). In this regard, DMAPT is of particular benefit as it has been developed to have increased bioavailability and has low toxicity in mice and humans at biologically relevant doses and is thought to be one of the only promising orally dosed clinical NF-κB inhibitors. The DMAPT clinical data to date in a 28-day pilot trial patients with hematological malignancies clinical trials have shown good PK-PD data, and no adverse events as serum levels above 10 μmol/L with inhibition of NF-κB in circulating blasts were observed and this bodes well for a planned prostate cancer trial.

These data highlight the key importance of p65 in the restoration of sensitivity to AR inhibition in mCRPC. Cell lines with constitutively active p65 did not respond to treatment with DMAPT, either alone or in combination with enzalutamide. This loss of therapeutic benefit closely aligned with high expression of AR-V7 or AR-FL. The results of this study would recommend the use of a combination therapy approach, targeting the androgen–AR axis and NF-κB/p65 signaling pathway to treat mCRPC and mHSPC by mitigating a cause of resistance to AR-targeted therapy whether it be by suppression of androgens or direct AR inhibition. Given the encouraging early clinical data for DMAPT, it is our belief that this study provides strong preclinical justification for the use of DMAPT in the treatment of mCRPC and mHSPC and suggests the need for clinical trials combining DMAPT and AR antagonists in such patients with prostate cancer.

C.J. Sweeney reports grants from Janssen, Astellas, Pfizer, and Bayer, as well as personal fees from Janssen, Astellas, Pfizer, Bayer, Genentech/Roche, Lilly, and grants from Dendreon during the conduct of the study; and reports a patent for TTP as a biomarker of lethal prostate cancer licensed to Dana Farber Cancer Institute and a patent for DMAPT in cancer licensed to Leuchemix. No disclosures were reported by the other authors.

K.L. Morel: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. A.A. Hamid: Formal analysis, methodology, writing–review and editing. J.G. Clohessy: Resources, investigation, writing–review and editing. N. Pandell: Resources, investigation. L. Ellis: Supervision, writing–original draft, writing–review and editing. C.J. Sweeney: Conceptualization, resources, supervision, writing–original draft, writing–review and editing.

We would like to thank Stephen Plymate for the gift of the VCaP-CR cell line and Michelangelo Fiorentino for support with in vivo tissue analysis, as well as the Molecular Biology Core Facility (DFCI) for assistance with Illumina RNA sequencing. This work was supported by a U.S. Department of Defense (DOD) Idea Award (W81XWH-19–1-0564; to C.J. Sweeney).

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.