Why ARNT Prostate Tumors Responding to Enzalutamide?

Although localized prostate cancer is often treated successfully with radiation, surgery, or active surveillance, men who progress on these therapies typically receive systemic androgen deprivation therapy (ADT). ADT targets the androgen receptor (AR), a hormone-responsive transcription factor essential for prostate tumor cell proliferation. Almost all prostate tumors are initially sensitive to ADT, but ADT exposure selectively favors cells that harbor somatic changes that allow them to grow in an androgen-starved environment (1–3). AR, therefore, remains a target in ADT-resistant prostate cancer, a condition termed castration-resistant prostate cancer (CRPC). The small-molecule drug enzalutamide targets the AR protein, preventing it from translocating to the nucleus. Enzalutamide is approved for the treatment of localized and metastatic CRPC and hormone-sensitive metastatic prostate cancer. Enzalutamide is now under evaluation as neoadjuvant therapy in high-risk localized prostate cancer, prior to radical prostatectomy. Resistance to enzalutamide and other AR-targeted therapies inevitably arises. Delaying this resistance and identifying novel targets in the enzalutamide-resistant setting are essential to improving the lives of men with prostate cancer.

In this issue of Cancer Discovery, Linder and colleagues identified a transcription factor that may play an important role in the development of enzalutamide resistance (4). A sizable minority of CRPCs treated with AR-targeted therapy develop neuroendocrine features reminiscent of small cell carcinoma (5, 6). These treatment-induced neuroendocrine prostate cancers (NEPC) have a particularly poor prognosis. The most consistent hallmark of NEPC is low or absent AR expression, indicating a dramatic change in the tumor's primary driver of proliferation has occurred. The NEPC transition, sometimes termed lineage plasticity, is associated with inactivating mutations in the tumor suppressors TP53 and RB1 (7), but these alterations are not inevitably present in NEPC and can also be observed in adenocarcinoma. It is clear that epigenetic alterations are a key driver of prostate tumor progression and NEPC (8, 9). Recent studies have demonstrated transcription factors associated with neuronal development, such as ASCL1, may be important in the transition to NEPC (10). Linder and colleagues studied paired biopsies from the DARANA phase II trial of enzalutamide in neoadjuvant prostate cancer (ClinicalTrials.gov identifier: NCT03297385). They identified a novel dependency in enzalutamide-resistant cells on the key circadian regulator ARNTL (aryl hydrocarbon receptor nuclear translocator like, also known as BMAL1), a transcription factor not previously associated with AR-targeted therapy resistance or prostate cancer.

The investigators profiled 56 men in the DARANA trial who had high-risk localized prostate cancer. Patients received 3 months of enzalutamide therapy, followed by radical prostatectomy. Pretreatment tumor samples were obtained by MRI-guided needle biopsy, whereas posttreatment biopsies were obtained from resected tissue guided by palpation and reference to the baseline MRI scan. Biopsies were subjected to chromatin immunoprecipitation followed by sequencing (ChIP-seq) targeting AR, the pioneer transcription factor FOXA1, and H3K27ac, a histone mark found in active DNA enhancer regions. The authors also performed RNA sequencing (RNA-seq), assessed copy-number alterations, and stained for protein expression by IHC in 51 posttreatment samples.

Differential RNA-seq analysis showed that after treatment, tumors had a reduced expression of genes associated with AR, mitosis, and MYC signaling and increased expression of genes associated with TNFα signaling, IFNγ response, and epithelial–mesenchymal transition. They then clustered the expression data, restricting analysis to a gene set previously shown to distinguish three subtypes of primary treatment-naïve prostate cancer: ERG fusion–positive, ERG fusion–negative, and NEPC. Most tumors moved into the NEPC cluster after treatment, which was associated with lower AR signaling. However, these samples showed modest expression of NEPC-associated genes, and most lacked increases in protein markers of NEPC such as chromogranin A or synaptophysin. These results were consistent with the expected effect of a potent AR-inhibiting therapy and could indicate that a lineage switch was in progress, but they did not fully recapitulate key features of the NEPC phenotype previously observed in metastatic CRPC.

A small percentage of the total set of ChIP-seq peaks for AR, FOXA1, and H3K27ac were recurrently observed among the biopsies, suggesting that considerable heterogeneity was present. Peaks that overlapped in at least 10 samples were present in ∼5% of AR, ∼20% of FOXA1, and 25% of H3K27ac ChIP-seq experiments. The investigators assigned consensus peaks for each assay, defined as peaks present in at least 25% of all sites. AR binding patterns were not significantly different in paired analysis after enzalutamide treatment, though the investigators’ power to detect differences in AR binding may have been limited by the low number of paired AR–ChIP samples that passed quality control checks (10 pretreatment, 12 posttreatment, and 4 paired samples). FOXA1 signals were significantly different before and after enzalutamide treatment. Few of the FOXA1 consensus peaks enriched before treatment overlapped with the H3K27ac peak. These sites tended to overlap with AR binding sites that lacked active or AR-inducible enhancer activity in a previously published self-transcribing active regulatory region sequencing (STARR-seq) reporter assay experiment. FOXA1 sites enriched after treatment, in contrast, were more likely to overlap with H3K27ac enhancer sites and were enriched for proximity to essential genes. These observations were consistent with a model wherein FOXA1 takes on a more active role in promoting cell survival after 3 months of enzalutamide exposure. However, posttreatment FOXA1 binding sites did not tend to overlap with previously published NEPC-specific FOXA1 binding sites, and FOXA1 did not preferentially bind at genes previously associated with NEPC in metastatic disease.

An unbiased binding enrichment analysis of posttreatment FOXA1 peaks identified ARNTL as the second most enriched transcription factor after FOXA1 itself. This was not an expected observation, as ARNTL is a core component of the circadian rhythm regulatory system, and it has no prior documented role in prostate cancer. ARNTL gene expression and protein levels were higher in biopsies after enzalutamide treatment. The investigators then characterized how manipulating ARNTL levels affected cell proliferation in LNCaP cells, an AR-sensitive model of CRPC, and in LNCaP-42D, an enzalutamide-resistant LNCaP derivative. They showed that 48 hours of enzalutamide exposure recapitulated FOXA1–ChIP binding site shifts they had observed in the DARANA clinical samples and showed that enzalutamide treatment increased ARNTL protein levels, dependent on FOXA1.

Physical interactions between ARNTL protein and its binding partners were assessed using rapid immunoprecipitation mass spectrometry of endogenous protein (RIME) targeting ARNTL. The data were consistent with a physical association between ARNTL and FOXA1, as well as the expected circadian rhythm proteins. Supporting the hypothesis that FOXA1 acts as a pioneer factor for ARNTL, ARNTL binding significantly decreased upon FOXA1 knockdown only at regions cobound by FOXA1 and ARNTL. To test whether ARNTL drives tumor cell proliferation during enzalutamide exposure, the investigators knocked down or overexpressed ARNTL in LNCaP and LNCaP-42D cells. ARNTL knockdown significantly suppressed the growth of enzalutamide-resistant LNCaP-42D cells in both enzalutamide and vehicle conditions, but no significant effect was observed in LNCaP cells. These experiments were validated in two other enzalutamide-resistant prostate tumor cell models by knockdown and CRISPR-mediated knockout of ARNTL. LNCaP-42D–derived xenografts with ARNTL knockout grew more slowly than ARNTL-intact cells upon enzalutamide exposure. ARNTL overexpression did not increase proliferation in either cell line model.

Studying solid tumor biopsies in advanced prostate cancer is logistically and technically challenging, but it is an essential complement to preclinical investigations and studies of liquid biopsies. This report presents evidence that ARNTL, best known as a circadian protein, may play a role in the development of NEPC. Localized prostate tumors exposed to enzalutamide for 3 months are distinct genetically and epigenetically from metastatic NEPC; whether these tumors represent a precursor to NEPC or a distinct entity is not yet clear. Characterization of ARNTL in the metastatic setting and in models of localized prostate cancer will be important to clarify this point. ARNTL knockdown slowed the growth of enzalutamide-resistant CRPC models in in vitro and xenograft experiments but did not affect the growth of enzalutamide-sensitive LNCaP cells. The investigators suggest this may be because 48 hours was not long enough to “achieve full epigenetic reprogramming” required for a response in hormone-sensitive cells. LNCaP is a model of metastatic rather than localized prostate cancer, and this may also have affected these observations. Further study of prostate cancer biopsies will be essential to define the epigenetic phenotype of this disease and to develop effective therapies for patients with enzalutamide-resistant prostate cancer.