Molecular Underpinnings Governing Genetic Complexity of ETS-Fusion-Negative Prostate Cancer

Abstract

Inter- and intra-patient molecular heterogeneity of primary and metastatic prostate cancer (PCa) confers variable clinical outcome and poses a formidable challenge in disease management. High-throughput integrative genomics and functional approaches have untangled the complexity involved in this disease and revealed a spectrum of diverse aberrations prevalent in various molecular subtypes, including ETS fusion negative. Emerging evidence indicates that SPINK1 upregulation, mutations in epigenetic regulators or chromatin modifiers, and SPOP are associated with the ETS-fusion negative subtype. Additionally, patients with defects in a DNA-repair pathway respond to poly-(ADP-ribose)-polymerase (PARP) inhibition therapies. Furthermore, a new class of immunogenic subtype defined by CDK12 biallelic loss has also been identified in ETS-fusion-negative cases. This review focuses on the emerging molecular underpinnings driving key oncogenic aberrations and advancements in therapeutic strategies of this disease.

Molecular Heterogeneity of Fusion-Positive and Fusion-Negative PCa

Prostate adenocarcinoma represents a heterogeneous collection of malignancies with diverse molecular frameworks. Revolutionary advances in high-throughput sequencing technologies revealed extensive molecular heterogeneity at genomic, epigenomic, transcriptomic, and proteomic levels, which laid the foundation of molecular stratification of this disease [1]. Half of the molecular etiology of PCa is driven by recurrent genetic rearrangements involving E26-transformation-specific (ETS) transcription factors, hence classified as ETS fusion positive. Under this category, several members of the ETS family, namely, ERG, ETV1, ETV4, and FLI1 are known to be aberrantly overexpressed [2]. Of these, TMPRSS2–ERG gene fusion involving the 50′-untranslated region of an androgen-regulated transmembrane protease serine-2 (TMPRSS2) and homolog v-Ets Erythroblastosis virus E26-oncogene (ERG) is the most frequent alteration affecting about ~50% of the total PCa cases [3,4].

Comprehensive genomic profiling has further unraveled diverse subclasses of ETS-fusion-positive and ETS-fusion-negative PCa. Molecular alterations involving gene rearrangements, copy number alterations (CNAs, see Glossary), and structural variants such as somatic mutations, insertions/deletion (indels), tandem duplications, and interchromosomal translocations are the key molecular drivers involved in the pathogenesis of this disease [5]. Outlier expression of SPINK1, RAF kinase fusions, deletions in CHD1 and MAP3K7; mutations in SPOP, FOXA1 and IDH1 have been reported to be mutually exclusive to the ETS fusions, and categorized under ETS-fusion-negative subtypes [5–9]. Recently, new subclasses of the fusion-negative subtype, with somatic and pathogenic germline mutations in DNA damage and response (DDR) pathways, epigenetic modifiers, and members of the spliceosome machinery have been added [5,10,11]. Moreover, CNAs involving amplification of genomic loci comprising oncogenes, namely MYC and PVT1; deletions in tumor suppressors such as NKX3.1, PTEN, RB1, and TP53 cooperate with above-mentioned alterations in promoting tumor progression [6,11,12]. In this review, we focus on diverse molecular heterogeneity within the fusion-negative subtype, its clinical significance, and implication in designing novel therapeutic strategies.

Glossary.

Biochemical recurrence: refers to a clinical condition marked by an increase in serum PSA levels in PCa patients who have undergone surgery or radiation therapy.

Bromodomain and extra-terminal (BET) protein inhibitors: small-molecule inhibitors which target the bromodomain of BET proteins hindering their interaction with acetylated histones, thereby inhibiting transcription of BRD target genes.

Copy number alterations (CNAs): structural variations including duplications, insertions/deletions (indels) of genes that impart phenotypic variations or causative events in genetic disorders or cancers.

DNA damage and response (DDR) pathway defects: DDR pathways maintain genomic stability by preventing duplication and propagation of DNA errors to the next generation. These defects initiate neoplastic transformations, neurodegenerative or cardiovascular disorders, and other heritable diseases.

Epigenetic regulators: a group of chromatin-remodeling enzymes such as histone acetyltransferases, deacetylases, methyltransferases, and DNA-methyltransferases that play a crucial role in modulating gene expression by orchestrating the chromatin state. Aberrations in these enzymes alter histone modifications and methylation patterns, which in turn disrupts the cellular homeostasis in human malignancies.

Frameshift mutations: these are caused by insertion/deletion of nucleotides that change the reading frame of the encoded protein, often resulting in a new or a nonfunctional protein.

Hypermutated genomes: individual tumor genome with about ≥300 somatic mutations (an outlier number) compared with its matched normal is considered to be hypermutated. Higher frequencies of mutations arise due to impairment of the DNA mismatch repair pathway.

Immune checkpoint inhibitors: cancer cells exhibit high PD-L1 expression, which interacts with immune checkpoint proteins, namely PD-1 expressed on T cells, resulting in immune escape. This class of drugs/inhibitors targets these immune checkpoint proteins allowing T cell-mediated cytotoxicity.

Neoantigens: newly formed antigens, generally truncated proteins, cause higher degree of antigenicity in tumor cells, hence facilitating immune cells to target and eliminate them. Tumors with high MMR defects or microsatellite instability (MSI) show higher number of neoantigens, making them suitable candidates for immunotherapy.

SWI/SNF nucleosome-remodeling complex: ATP-dependent nucleosome-remodeling complex, comprises a group of proteins that disrupt histone–DNA interactions by ATP hydrolysis, thereby making DNA accessible for transcription machinery and DNA repair.

Tandem duplications: intra-chromosomal duplication where a segment of DNA is duplicated, and placed on the same chromosomal arm next to the original segment. Tandem duplications are further divided into direct tandem, in which the gene order remains the same as the original, and reverse tandem, where the duplicated segment order is reversed.

Truncating mutations: these are also known as nonsense mutations, caused by a change in DNA base pairs, leading to a premature stop codon that results in a truncated or nonfunctional protein.

Tumor-infiltrating T cells: infiltration and activation of immune cells, especially effector T lymphocytes into the tumor microenvironment. This phenomenon is a predictor for an effective response of T cell-based immunotherapies.

Opportunities and Challenges in Targeting the ETS-Fusion-Positive Subtype

Aberrant expression of ETS fusion activates an invasive transcriptional program, inducing prostatic intraepithelial neoplasia (mPIN) in transgenic mice, and is often associated with poor clinical outcomes [13–16]. Moreover, ERG fusion protein is known to inhibit androgen receptor (AR) signaling and its occupancy on AR target genes, results in increased EZH2-mediated H3K27 methyltransferase activity, which potentiates the stem-cell-like dedifferentiation program [17]. Since ETS fusions are under transcriptional control of AR signaling, androgen-deprivation therapy (ADT) is the preferred therapeutic strategy for fusion-positive cases. Despite recent advances in ADT, most patients develop de novo resistance to second-generation FDA-approved antiandrogens, such as abiraterone and enzalutamide, due to AR amplification, mutations, and AR splice variant (AR-V7), with subsequent progression to metastatic castration-resistant PCa (mCRPC) [18–20]. Although ETS-fusion-positive PCa represents the major molecular subtype, therapies targeting these transcription factors remain challenging. Alternatively, therapeutic modalities against molecular cofactors that modulate the transcriptional activity of ETS factors represent a more viable approach [21].

Poly-(ADP-ribose) polymerase 1 (PARP1) and DNA protein kinases (DNA-PKcs) are known to physically interact with ERG-fusion protein and induce DNA double-strand breaks. Moreover, cell-based assays and preclinical studies using ERG-positive PCa xenografts exhibit an effective response to the pharmacological PARP1 inhibitor (PARPi), olaparib [22]. Unfortunately, the recent clinical trial (NCT01576172ⁱ), using PARPi (veliparib) failed to show any response in fusion-positive PCa patients (Box 1) [23]. Recently, inhibitors against bromodomain and extra-terminal (BET) family proteins (BRD2/3/4, critical AR coregulators), were shown to block BET-protein recruitment to the AR-target gene loci, hence negatively regulating ERG-mediated oncogenesis [24]. Unlike second-generation antiandrogens, BET inhibitors (BETis) regulate RNA processing and AR-V7 alternative splicing, resulting in its downregulation, hence highlighting the clinical utility of BETis for abiraterone- or enzalutamide-resistant castration-resistant prostate cancer (CRPC) patients [25–27]. Currently, Phase I clinical trials (NCT02711956ⁱⁱ and NCT03150056ⁱⁱⁱ) using BETis are underway for CRPC patients, although its clinical efficacy remains a matter of conjecture. Yet another promising therapeutic strategy targeting oncogenic ERG fusion involves ERG-inhibitory peptidomimetics, which results in proteolytic degradation of ERG-fusion protein, and reduced oncogenic activity [28]. Since most small-molecule inhibitors result in drug resistance and restricted therapeutic durability, a newer class of small molecule protein degraders such as proteolysis targeting chimeras (PROTACs), that target hitherto undruggable proteins (such as AR, BET, ERG, and AR-V7) by their selective proteasomal degradation are presently under clinical investigation (Box 2).

Box 1. Clinical Failure of PARPi Therapy for ETS-Fusion-Positive PCa.

Androgen signaling regulates a network of genes involved in DNA repair; thus, ADT could synergize response to PARP inhibition and radiotherapy [94,95]. A recent prospective clinical trial NCI-9012 (NCT01576172ⁱ) evaluating the therapeutic potentials of PARPi veliparib in combination with antiandrogen (abiraterone acetate) for ETS-fusion-positive mCRPC patients failed to show any additive response when compared with antiandrogens alone. Moreover, no significant response such as a decrease in PSA or progression-free survival was recorded in ETS-fusion-positive patients [23]. Furthermore, clinical assessment of tumor biopsies using targeted exon sequencing revealed that ~25% of mCRPC patients harbor mutations in the genes involved in the HR DNA-repair pathway. These patients with DNA-repair defects showed an exceptional response to veliparib with an overall decrease in PSA (≥90%), and a prolonged median progression-free survival in comparison to the wild-type tumors (14.5 vs 8.1 months) [23]. This study implicates a therapeutic strategy that includes PARPis in combination with antiandrogens for patients with DDR defects; however, it failed to predict clinical efficacy for ETS-fusion-positive PCa patients.

Box 2. Clinical Significance of Protein Degraders: PROTACs.

The PROTACs are chimeric bifunctional small molecules with two arms joined by a linker, where one arm recognizes a protein of interest, and the other, E3 ubiquitin ligase to form a ternary complex facilitating proteasome degradation of the target protein through the ubiquitin–proteasomal machinery [96]. Recent studies from multiple groups have utilized PROTACs for targeting oncogenic drivers, such as AR and its coactivator, BRD4 [97,98]. Multiple AR–PROTACs have been designed based on different classes of AR antagonists, E3-ligase degradation systems, and varying lengths of linkers. Of these, the most potent being ARCC-4, ARD-69, and ARV-110 which have been shown to effectively reduce (>90%) expression of AR as well as AR mutants, and attenuate their oncogenic effects in cancer cell lines as well as in enzalutamide-resistant preclinical mouse models [99–101]. Recently the first oral AR PROTAC, ARV-110 has entered a Phase I clinical trial (NCT03888612^ix), which may have potential for mCRPC patients. Notably, a small molecule pan-BET protein inhibitor ARV-771, designed based on similar PROTAC chemistry, has shown a significant decrease in BRD4, AR full length, AR-V7, and ERG proteins. When compared with conventional BET inhibitors, ARV-771 results in a remarkable inhibition of tumor growth in preclinical CRPC mouse models and enzalutamide-resistant AR-V7-positive 22RV1 xenografts [97]. Taken together, these preclinical findings encourage clinical applicability of the PROTACs for advanced stage mCRPC patients.

Mechanistic Insight into Fusion-Negative SPINK1-Positive PCa

The etiology of the ETS-fusion-positive subtype has been well characterized, while little is known about the molecular aberrations driving the ETS-fusion-negative cancers. About 10–15% of PCa patients show outlier SPINK1 expression, exclusively in the fusion-negative subtype [29]. Although several independent studies indicate that increased serine peptidase inhibitor, Kazal type-1 (SPINK1) levels are associated with shorter biochemical recurrence and a rapid progression to the castration-resistant stage [30–32]. However, the prognostic role of SPINK1 in primary PCa remains controversial [33–35].

SPINK1 acts as an autocrine/paracrine factor and mediates its downstream oncogenic effects partly through epidermal growth factor receptor (EGFR) interaction. Monoclonal antibodies against SPINK1 or EGFR (e.g., cetuximab) administered in mice with SPINK1-positive tumor xenografts showed a modest decrease in tumor burden [36], although the clinical significance of EGFR inhibitors in mCRPC patients remains debatable [37,38]. Recently, hepatocyte nuclear factors HNF4G/HNF1A were shown to activate a gastrointestinal-lineage transcriptome in PCa, resulting in upregulation of PCa-associated gastrointestinal genes (PCa-GIs), including SPINK1 [39]. It was shown that HNF1A directly binds to the SPINK1 promoter and positively regulates its expression, while HNF4G upregulates the PCa-GI signature upon androgen deprivation, and thereby supports CRPC progression [39]. Nevertheless, an in-depth exploration of genetic/epigenetic molecular circuitries involved in SPINK1 upregulation and ways to effectively target this subtype remains unaddressed.

A recent study established the molecular basis of SPINK1 upregulation in PCa, wherein miRNA-338-5p/miRNA-421 regulate SPINK1 post-transcriptionally and abrogate SPINK1-mediated oncogenicity [40]. Intriguingly, the majority of SPINK1-positive PCa patients exhibit increased expression of EZH2, a member of the Polycomb repressive complex, which epigenetically represses miRNA-338-5p/miRNA-421 resulting in SPINK1 upregulation. Moreover, SPINK1-positive tumors also exhibit increased CpG-methylation marks on the regulatory regions of miRNA-338-5p/miRNA-421, thereby indicating the utility of epigenetic drugs in restoring the expression of these miRNAs (Figure 1). In concordance with earlier findings, the recent reports showed higher methylation events in fusion-negative compared with fusion-positive PCa [5,41]. Conclusively, the use of epigenetic drugs might have potential for the SPINK1-positive subtype, and reduced miRNA-338-5p/-421 expression might possibly serve as selection criteria for these patients [40,42].

Figure 1. Molecular Circuitries Involved in SPINK1 Upregulation in Prostate Cancer.

A schematic showing (A) Androgen receptor (AR)-mediated transcriptional repression of SPINK1, wherein antiandrogen treatment may result in SPINK1 upregulation (right). (B) SOX2 and HNF1A binds on the SPINK1 promoter and positively regulates its expression (right). (C) Epigenetic repression of post-transcriptional regulators, miR-338-5p and miR-421, results in increased SPINK1 expression and oncogenicity in prostate cancer (left). Abbreviations: ARE, androgen response element; CoR, corepressor; DNMTi, DNA methyltransferase inhibitor; EMT, epithelial–mesenchymal transition; EZH2, enhancer of zeste homolog 2; HDACi, histone deacetylase inhibitor; HKMTi, histone lysine methyltransferase inhibitor; HNF1A, hepatocyte nuclear factor 1-α; miRISC, miRNA-induced silencing complex; REST, RE1 silencing transcription factor; SOX2, SRY (sex determining region Y) box 2; SPINK1, serine protease inhibitor, Kazal type 1; UTR, untranslated region.

An unpublished study shows a possible role of androgen signaling in the transcriptional regulation of SPINK1 in PCa, wherein AR and its corepressor, REST (RE1-silencing transcription factor) may function as a transcriptional repressor of SPINK1, and antiandrogen treatment could relieve AR-mediated repression, leading to SPINK1 upregulation (Figure 1) [43]. Previously, Paju et al. also observed a decrease in SPINK1 expression upon androgen stimulation in 22RV1 cells [44]. Moreover, an inverse association between SPINK1 and AR expression was reported in multiple independent PCa cohorts, and these fidings are in line with the studies, where lower AR expression was observed in SPINK1-positive tumors [33,45]. Furthermore, a possible role of lineage programming factor sex determining region Y (SRY) box 2 (SOX2) in SPINK1 upregulation has been suggested in long-term androgen deprived LNCaP cells [43]. Another possible mechanism of elevated SPINK1 levels in patients who underwent ADT could be attributed to an increase in the expression of the PCa-GI gene signature. Therefore, taking clues from these independent studies, we emphasize that SPINK1 is an androgen-repressed gene, and ADT, using bicalutamide/enzalutamide, might aggravate the clinical outcomes of patients with advanced disease.

Genomic Aberrations Associated with ETS-Fusion-Negative PCa

In the past decade, whole genome, exome, and transcriptome integrative sequencing analysis have paved the way to further categorize primary and metastatic PCa patients based on co-occurring and mutually exclusive genomic aberrations associated with the ETS-fusion-negative subtype. While primary PCa genomes are largely diploid with a lower mutational burden (~0.94 average mutations/Mb), metastatic tumors often show extensive aneuploidy, loss of heterozygosity (LOH), and a significant increase (~4.4 average mutations/Mb) in actionable mutations, which implicate the use of targeted therapeutics either through clinical trials or FDA approval [5,6,8,46]. Here, we discuss the molecular underpinnings of distinct subclasses belonging to ETS-fusion-negative PCa and outline the recent advances in developing therapeutic interventions aimed to target these subclasses (Table 1 and Figure 2).

Table 1. Genetic Alterations and Therapeutic Interventions Targeting Different Molecular Subtypes in PCa.

Molecular subtype	Frequencies	Causal genetic alterations/drivers	Proposed drug therapies	Clinical trials	Refs
ETS-fusion-positive PCa
TMPRSS2–ERG	45–50%	AR and its coactivator BET proteins regulate expression of these oncogenic fusion proteins ETS factor interacts with DNA repair enzyme PARP and DNA-PKs and upregulates DNA repair genes	Antiandrogens PARPis ERG peptidomimetics BET inhibitors PROTACs	PARPi- NCI 9012-NCT01576172i BETi- NCT02711956ii, NCT03150056iii AR PROTACs-NCT03888612ix	[3,5,16,22–24,28,97,101]
ETV-1	~8%
ETV-5	~4%
FLI-1	~1%
ETS-fusion-negative PCa
SPINK1 overexpression	10–15%	AR signaling transcriptionally represses SPINK1, while SOX2 and HNF1A transcriptionally activate SPINK1 SPINK1-positive cases show increased EZH2 expression, which epigenetically silences miR-338-5p/-421 which negatively regulate SPINK1	EGFR inhibitors, Epigenetic inhibitors or miR-338-5p/-421 replacement therapies	—	[36,39,40,43]
RAF rearrangements	2–6%	Upregulates downstream oncogenic MEK and ERK signaling	Sensitive to RAF kinase inhibitor sorafenib, and MEK inhibitors	—	[7]
SPOP mutations (SPOPmut)	6–15%	Increased expression of BET proteins, confers resistance to BET inhibitors Activates AKT and AR signaling.	Sensitive to abiraterone therapy or PI3K inhibition	—	[5,8,50,52,61]
CHD1 deletions (CHD1del)	15–17%	Show increased AR activity Activates NHEJ repair pathway.	Sensitive to DNA damage agents	—	[6,59,60]
SPOPmut/CHD1del	~25%	Show a higher response to AR inhibitors due to increase in AR activity	Sensitive to abiraterone therapy	—	[61]
CHD1del/MAP3K7del	20–25%	Show loss of AR, and increased levels of SYP and CHGA, associates with neuroendocrine and neural features	—	—	[65]
IDH1 mutations	~1%	IDH1 R132 mutations; increase in oncometabolite D-2-hydroxyglutarate	AG-221 and AG-120	NCT02074839iv, NCT02073994v	[5,72,73]
FOXA1 mutations	~4%	Show C-terminal truncating and forkhead domain missense mutations, disrupting its interactions with other cofactors	—	—	[5,8,69]
Epigenetic regulators	~15%	Mutations in KMT2C (7%), KMT2D (6%), KDM6A (3%), KMT2A (1.2%)	—	—	[10]
Chromatin remodelers	~5%	Mutations in SWI/SNF nucleosome-remodeling members ARID1A, ARID4A, ARID2, SMARCA1	—	—	[10]
Splicing pathway	~4%	Mutation in SF3B1 (1.1%), U2AF1 (0.5%), GEMIN5 (0.5%), TCERG1 (0.8%) and PRPF8 (1.2%)	—	—	[10]
Ubiquitin–proteasome and ligase family	~12%	Mutations in USP28 (1.4%), USP7 (1.2%), and CUL3 (1.3%), also SPOP (9%)	—	—	[5,8,10]
DDR pathway alterations	19–23%	Somatic and germline mutations in genes involved in, HR pathway (~19%): BRCA1, BRCA2, RAD51C, FANCD2, ATM. MMR pathway (8–12%): MSH2, MSH6, MLH1	HR defects: PARPi therapy such as olaparib or veliparib MMR defects Immunogenic therapies such as anti-PD-1 and anti-CTLA-4 therapy	HR DDR defects: NCT01682772vi, NCT01576172i, NCT02966587xii For MMR defects: NCT03061539xi, NCT02484404xiii	[5,11,23,77,83–86,103]
CDK12 mutations	1.2–7%	Show frequent focal tandem duplications, increased expression of neoantigens, T cell infiltration, and dendritic cell migration to the tumor microenvironment.	Immune checkpoint inhibitors against PD-1 and CTLA-4	IMPACT-NCT03570619vii	[5,87]

Figure 2. Integrative Clinicogenomic Analysis of Molecular Aberrations in Prostate Cancer.

A schematic representation depicting tissue or liquid biopsies taken from prostate cancer patients subjected to integrative sequencing analysis for molecular stratification, followed by implementation of evidence-based therapeutic interventions. Abbreviations: AR, androgen receptor; ATM, ataxia telangiectasia mutated kinase; BET, bromodomain and extra-terminal protein; EGFR, epidermal growth factor receptor; ERG, homolog v-Ets erythroblastosis virus E26-oncogene; HR, homologous recombination; IDH1, isocitrate dehydrogenase 1; PARPi, poly (ADP) ribose polymerase inhibitor; PD-1, programmed cell death protein 1; PI3Ki, phosphoinositide 3-kinase inhibitor; PROTAC, proteolysis targeting chimera; RPPA, reverse phase protein array.

RAF Kinase Fusions in ETS-Fusion-Negative PCa

Paired-end transcriptome sequencing of ETS-fusion-negative PCa identified actionable genetic rearrangements involving RAF-kinase family members, namely SLC45A3-BRAF and ESRP1-RAF1 recurrent in about ~2% of advanced PCa cases [7]. Ectopic expression of both chimeras in prostate epithelial cells show an increase in oncogenic properties, which exhibits sensitivity to RAF and MEK inhibitors [7], emphasizing that RAF-fusion-positive patients may respond to RAF kinase inhibitors. Unlike Caucasian PCa patients, an increased frequency (~4–6%) of RAF fusions was reported in an Indian cohort [4]. Although of low recurrence, screening of these actionable RAF alterations could be beneficial in disease management of RAF-fusion-positive patients.

SPOP Mutations in PCa

Speckle-type POZ protein (SPOP) is a member of the MATH-BTB protein family and contains an N-terminal meprin and TRAF homology (MATH) domain, which binds to target substrates by recognizing degron motifs and a C-terminal BTB domain responsible for the binding of Cullin 3-RING box-1 protein, to form a functional E3-ubiquitin ligase complex [47]. In localized and advanced prostate tumors, SPOP has been found to be frequently altered in ~15% and ~6% of PCa patients, respectively. Patients harboring somatic mutations in the substrate-binding cleft domain of SPOP (p.Y87C, p.F102C, p.W131G, and p.F133V), show mutual exclusivity to ETS fusions [5,8]. Wild-type SPOP (SPOP^wt) is known to trigger polyubiquitination and proteasomal degradation of target proteins such as oncogenic coactivators TRIM24, chromatin-associated genes DEK [48], c-MYC [49], and AR [48], suggesting its tumor-suppressive role. Recent studies have shown that SPOP^wt tumors are more sensitive to BET-domain inhibitors because of the ubiquitin-mediated proteasomal degradation of BET proteins (BRD2/3/4) [49]. PCa subtypes harboring mutations in SPOP (SPOP^mut) show impaired expression of BET proteins (specifically BRD4), resulting in their accumulation, thus conferring resistance to BET inhibitors [49–51]. Moreover, transcriptome and BRD4 cistrome analyses of SPOP^mut cancer cell lines or organoids revealed enhanced expression of genes involved in AR signaling, cholesterol-biosynthesis pathways, and activation of ATK–mTOR signaling, due to BRD4 stabilization, thereby sensitizing SPOP^mut cells to the AKT inhibitor, ipatasertib [49,52]. Considering the sensitivity of SPOP^mut tumors to AKT inhibitors, several of these are already in clinical trials, hence emphasizing screening of SPOP mutations for guided precision medicine.

CHD1 Deletions Associates with SPOP Mutations ETS-Fusion-Negative PCa

Several studies have revealed that SPOP^mut tumors are enriched for recurrent somatic deletions of 5q21 and 6q21 loci, known to harbor a chromatin-modifying enzyme, chromodomain helicase DNA-binding protein 1 (CHD1) and tumor suppressors Forkhead O transcription factor and B-lymphocyte-induced maturation protein 1 [8,53,54]. In particular, CHD1 deletions (CHD1^del) are recurrent in ~15% of PCa cases, often associated with increased risk of biochemical recurrence, and inversely correlated with ERG fusion [5,8,55]. In normal tissue, CHD1 occupies the promoters of actively transcribed genes and facilitates transcriptional initiation of tumor-suppressor genes by maintaining nucleosome turnover [56,57]. However, recently, the interactome of CHD1 in a normal mouse prostate model established promoter-independent roles, where it occupied prostate-specific enhancers along with AR and its cofactors. Intriguingly, CHD1-deficient PCa cells show a redistribution of AR cistrome to HOXB13-enriched sites, which resembles the AR cistrome of human prostate tumors [58], driving a subtype-specific oncogenic transcriptional network with elevated AR activity; a phenomenon critical for PCa pathogenesis [59]. Moreover, a CHD1 knockout mouse model also showed increased activity of the nonhomologous end joining (NHEJ) DNA-repair pathway, which elicits hypersensitivity to PARPi and DNA-damaging agents [60]. Furthermore, CHD1^del tumors associate frequently with mutations in the SPOP–BTB domain, and inversely correlate with ERG fusion in ~25% cases and hence, are classified as ERG-/SPOP^mut/CHD1^del PCa subtype. Both SPOP^mut and CHD1^del subclasses independently exhibit increased AR activity, therefore, PCa patients positive for SPOP^mut/CHD1^del show enhanced sensitivity to abiraterone therapy [61]. Also, this subtype exhibits elevated levels of DNA methylation, and frequent SPINK1 outlier expression, thereby representing a key subtype of PCa molecular taxonomy [5].

Co-occurrence of CHD1 and MAP3K7 Deletions in ETS-Fusion-Negative PCa

Chromosomal deletion of 6q12-22 loci forms the second most deleted genomic region in ~22% of primary and ~40% of metastatic PCa tumors [9,62]. Genome mapping revealed a marginal deleted 6q15 region (3–5 Mb long), encoding tumor-suppressor MAP3K7 [9]. Heterozygous deletion (~18.48%) of MAP3K7 exclusively in ETS-fusion-negative cases has been associated with a high Gleason score, tumor grade, lymph node metastasis, and shorter biochemical recurrence [9,63]. A comparative analysis of multiple clinical PCa genomic datasets [1,6,8,64] identified codeletion of both MAPK37 and CHD1 in localized (10–20%) and metastatic (20–25%) PCa cases [65], with overall poor disease-free survival. Dual knockdown of MAP3K7–CHD1 in LNCaP xenografts showed increased tumor growth and overall poor survival compared with knockdown of an individual gene. Moreover, cosuppression of Map3k7–Chd1 in mouse prostate critically disrupts normal prostate line-age differentiation, marked by a decrease in luminal cells, with a significant loss of AR, and a trend toward neuroendocrine differentiation marked by increased synaptophysin (SYP) and chromogranin (CHGA) levels. Tumors harboring MAP3K7^del/CHD1^del represent a distinct PCa subtype with neuroendocrine and neural features [63,65], thus therapeutic interventions targeting neuroendocrine PCa [66] might prove effective for this subset of patients.

FOXA1 Mutations Define a Distinct Molecular Subtype

Multiple independent cohorts identified recurrent mutations in the forkhead protein, FOXA1, in ~4% of primary and metastatic PCa, which exhibits a mutually exclusive pattern with other genomic alterations [5,6,8,11]. FOXA1, being a pioneering factor facilitates AR recruitment by opening up compact chromatin, and plays a critical role in mediating androgen signaling [67,68]. Cases with FOXA1 mutations show several truncating mutations adjacent to the C-terminal transactivating domain and missense mutations in the forkhead domain, affecting the FOXA1 winged-helix DNA-binding domain. These forkhead mutations occur in the residues that do not directly interact with DNA; therefore, they do not alter its DNA-binding functions, but rather disrupts its interactions with other chromatin-bound cofactors [5]. Overexpression of C-terminal FOXA1 mutants has been shown to increase cell proliferation, migration, and mouse xenograft growth [6,68]. While earlier studies focused only on FOXA1 protein-coding domains, two recent independent studies using whole genome and untranslated regions (UTRs) sequencing revealed tandem duplications (~14%) and 3′-UTR indels (~12%) of FOXA1 in mCRPC cases. However, the detailed functional and clinical significance of these aberrations remains unexplored [69,70]. Furthermore, the screening of aberrations associated with FOXA1 might be useful in prognosis and clinical decision-making.

Mutation Burden in the Epigenetic Regulators of ETS-Fusion-Negative PCa

Overall increased DNA-methylation events have been reported in advanced-stage and fusion-negative PCa patients, when compared with benign and fusion-positive cases, respectively [41,71]. The epigenetic profiles associated with primary PCa show substantial heterogeneity [5,41,71]. Recently, The Cancer Genome Atlas (TCGA) identified a distinct genomic subtype (~1%) exclusively present in ETS-fusion-negative PCa cases, defined by hotspot IDH1 R132 mutations associated with lower frequency of CNAs and elevated genome-wide hypermethylation [5,72]. Isocitrate dehydrogenase 1 and 2 (IDH1/2) are multifunctional enzymes that maintain cellular epigenetics and metabolism by catalyzing isocitrate to α-ketoglutarate (α-KG) [73]. However, the mutated form of IDH1 catalyzes the formation of oncometabolite D-2-hydroxyglutarate (D-2HG) from α-KG, which in turn deregulates cellular epigenetics and blocks differentiation, thereby driving the progression of multiple human malignancies [74]. Small-molecule inhibitors, namely AG-221 and AG-120 targeting IDH1 mutation have shown promising preclinical results, and are currently in Phase I clinical trials (NCT02074839^iv, NCT02073994^v) for the advanced stage hematological and solid cancers [75]. Furthermore, these mutations are not observed in any mCRPC cohorts [11,46], suggesting their role in early onset of disease. Therefore, examining IDH1 mutations or detecting D-2HG oncometabolite levels in plasma will be instrumental in identifying PCa patients with these aberrations.

Of late, a new subclass of ETS-fusion-negative tumors defined by truncating mutations in genes coding for chromatin remodelers (~20%) and spliceosome-complex members (~4%) were identified in a large cohort of primary and metastatic PCa cases [10]. Mutations in chromatin remodelers include epigenetic modifies (~15% of the cases) such as KMT2C, KMT2D, or KDM6A, and members of the SWI/SNF nucleosome-remodeling complex (~5%) such as ARID1A, ARID4A, ARID2, and SMARCA1 were identified. Notably, ~4% of the cases show hotspot mutations in genes involved in spliceosome machinery such as SF3B1 and U2AF1. Moreover, ~12% of the PCa patients harbor mutations in ubiquitin–proteasome and ligase family members, such as USP28, USP7, CUL3, and SPOP, which affects proteasomal degradation of several oncogenic regulators such as AR and its splice variants and coactivators [10]. Although the functional and clinical relevance of these mutations remains to be elucidated, the preliminary findings from this study implicate an essential role of these epigenetic modifiers and spliceosome machinery in the pathogenesis of this disease.

Aberrations Associated with Novel Classes of DNA-Repair Pathways

Evolution has selected interconnected networks of DDR pathways, which ensures genomic integrity and cell survival after genotoxic insults. These critical pathways include the base excision repair (BER), nucleotide excision repair (NER), mismatch repair (MMR), and homologous recombination (HR), and NHEJ, which are typically involved in the removal of mismatches, indels, and double-stranded breaks induced during DNA replication. The somatic/germline aberrations, including SNPs and indels in DDR-pathway genes, have now been redefined as a ‘double-edged sword’ since an increase in the mutational burden of DDR pathways serves as a mechanism by which tumors generate secondary oncogenic drivers. In contrast, these mutations also make tumors more vulnerable to specific therapies targeting DNA-repair genes [76].

Whole exome and targeted deep-sequencing of advanced and metastatic PCa revealed hypermutated genomes, harboring mutations in MMR genes and associated microsatellite instability (~12%), with a few patients displaying complex structural rearrangements in MSH2 and MSH6 (MutS protein homologs) and focal homozygous deletion of MSH2 and BRCA2 [6,77,78]. Moreover, using multiomics platforms, loss-of-function mutations associated with several genes involved in DDR pathways, mostly affecting the HR pathway, were reported [5]. Of these aberrations, germline frameshift mutations in BRCA1, C-terminal truncating mutations (K3326*) in BRCA2, homozygous focal deletions in CDK12, heterozygous deletion in RAD51C, nonsense mutations in ATM and FANCD2, a key regulator of the Fanconi anemia pathway were found in ~19% of primary prostate tumors [5]. While an increased mutational burden, comprising pathogenic germline (~8%) and somatic aberrations (~23%) in DDR genes have been associated with mCRPC. Of these, BRCA2 is the most altered gene in tumors with germline (~5.3%) and somatic biallelic (~12.7%) losses, followed by biallelic loss of ATM and somatic mutations BRCA1, CDK12, FNACA, and RAD51B/C [46,79].

Since aberrations associated with HR-defective genes are shown to be actionable, tumors with such defects are clinically more vulnerable to PARPi or platinum-based chemotherapy [80–82]. The recent Phase II clinical trials using the PARPis olaparib (NCT01682772^vi) and veliparib (NCT01576172ⁱ) for advanced stage mCRPC showed impressive clinical outcomes with 20–30% of patients who harbored DDR defects responding to these therapies (Boxes 2 and 3) [23,83]. A recent follow-up clinical trial using olaparib on the other side identified divergent evolution of resistant tumor subclones with secondary mutations (Box 3), as a resistance mechanism to overcome PARPi-therapy-induced selective pressure [84]. This emphasizes that management strategies involving sequential monitoring of patients are critical during clinical trials to detect any therapy-induced early genetic changes (Box 3). Furthermore, emerging lines of evidence indicate that MMR-deficient tumors possess an active immune microenvironment, with robust immune-checkpoint ligand expression, such as, that of programmed death-1 (PD-1), programmed death ligand-1 (PDL1), cytotoxic T-lymphocyte-associated protein-4 (CTLA-4), increased tumor-specific neoantigens, and tumor infiltrating T lymphocytes, forming the basis of clinical investigation of immunotherapies for this subset (Box 3) [85,86].

Box 3. Clinical Studies Targeting DNA Damage and Response Defects in PCa.

The clinical outcomes of the recently conducted Phase II trial (TOPARP-A: Trial of PARP inhibition (PARPi) in prostate cancer, NCT01682772^vi) showed an impressive response to PARP inhibition. Advanced stage mCRPC patients (n = 50), who failed to respond to prior therapies such as antiandrogens, docetaxel, or cabazitaxel were enrolled; of these, 16 patients (~33%) responded to PARPi therapy. Importantly, 88% of these responders harbored homozygous deletions or mutations in BRCA2 (seven of 16), ATM (four of 16), FNACA (2 of 16), and CHEK2 (one of 16) [83]. Subsequently, FDA approved olaparib as a breakthrough therapy designation for treatment of mCRPC patients harboring BRCA1/2 or ATM gene mutations [102]. However, a recent follow-up study of this trial, using cell-free DNA (cfDNA) isolated from the blood samples serially collected during the course of the trial, identified evolution of PARPi-therapy-resistant subclones.Whole-exome cfDNA sequencing of six responders, where samples were collected during disease progression and at resistance stage, identified two patients with prior germline BRCA2 frameshift mutations, and newly acquired frameshift somatic deletions, hence restoring the reading frame of BRCA2. Additionally, two patients with nongermline BRCA2 and PALB2 mutations in the pretrial tumor biopsy showed a new inframe somatic deletion, thereby restoring BRCA2 and PALB2 reading frames, and a patient with HDAC2 biallelic loss before treatment exhibited new TP53 and TSC2 mutations in the resistant-tumor biopsies during the trial [84]. These secondary mutations were acquired as a result of PARP inhibition, which indicate clonal evolution of resistant tumors to overcome therapy induced drug pressures.

Recent studies also indicate that mCRPC patients with hypermutated genomes are enriched for MMR defects, which warrants clinical exploration and treatment responses to immune-checkpoint inhibitors in MMR-deficient mCRPC patients [103]. Phase II clinical studies (NCT01876511^x) evaluating the effect of immune-checkpoint inhibitor, pembrolizumab (anti-PD-1) for metastatic colorectal carcinoma patients with MMR deficiency have shown remarkable response with prolonged progression-free survival [85]. Phase I/II clinical trials (NCT03061539^xi) evaluating combinatorial effects of anti-PD-1 and anti-CTLA-4 immunotherapies for mCRPC patients with immunogenic MMR- and DDR-mutational signatures are currently ongoing.

Biallelic Loss of CDK12 Represents a Distinct Immunogenic Class of ETS-Fusion-Negative PCa

Comprehensive sequencing analysis of multisite biopsies of 360 metastatic PCa patients (CRPC360) and a primary TCGA dataset identified a novel subtype, marked by the biallelic loss of CDK12, recurrent in ~1.2% and ~7% of primary and mCRPC cases, respectively [87]. This subtype shows mutual exclusivity to tumors harboring ETS fusions, DDR deficiencies, and SPOP mutations, thus pointing toward distinct tumor evolutionary patterns. Previously, CDK12 has been implicated in regulating the expression of several DDR genes [88,89], and was shown to be mutated in advanced CRPC patients [6,11,90]. Comparative genome analysis of CDK12-mutation-positive tumors suggests that CDK12 mutants are largely diploid, with the highest number of differentially expressed genes, higher gene-fusions incidences, and focal tandem duplications. Additionally, these CDK12-mutant tumors show increased expression of chemokines such as CCL18 and CXCL8, which support dendritic cell migration into the tumor microenvironment and an overall increase in neoantigens and T cell infiltration, marked by increased CD3 levels [87].

Furthermore, a retrospective clinical analysis of MI-ONCOSEQ enrolled patients revealed that four of 11 CDK12-mutant cases show a modest decrease in prostate-specific antigen (PSA) levels after anti-PD1 monotherapy. Additionally, one patient with prior lymph node metastases showed a robust decrease in CD3⁺ staining in the lymph node metastatic biopsy and a concomitant decline in disease burden [87]. A prospective Phase II clinical study (NCT03570619^vii) evaluating the combined effects of anti-PD-1 and anti-CTLA-4 immunotherapies for CDK12-mutant mCRPC patients is ongoing. Yet another, Phase III clinical trial (NCT02054806^viii) with anti-PD1 monotherapy, revealed an increase in progression-free and overall survival in advanced PD-L1-positive mCRPC patients [91]. Taken together, these clinicogenomics trials using evidence-based sequencing approaches and combinatorial therapeutic strategies will support the development of precision therapies and better disease management [92].

Concluding Remarks

The rapid advancement in sequencing technologies and integrative data analyses has advanced the field of PCa by revealing hitherto unknown molecular key players specific to the ETS-fusion-negative subtype. Comparative molecular studies of metastatic and primary PCa cohorts have identified a plethora of actionable and meaningful genetic mutations associated with AR signaling, epigenetic regulators, DDR pathway, immunogenic subclasses, polyubiquitin, and spliceosome machinery. Despite the detailed molecular taxonomy of the PCa genome, ~26% of all tumors appeared to be driven by still occult molecular abnormalities/aberrations, which need future investigations [5]. Nonetheless, most of these studies are limited to a single tumor focus, whereas most prostate tumors that are multifocal demonstrate molecular heterogeneity across different foci within the prostate gland. Such intratumoral heterogeneity should be considered for future genomics or clinical studies, as well as for designing gene signature panels and novel therapeutic strategies (see Clinician’s Corner). Moreover, improved diagnostic technologies are required to detect pre-existing resistant mutations in heterogeneous tumors, followed by evidence-based tailored therapeutic approaches, which might circumvent the growth of drug-resistant clones. As stated by Adam Dicker, ‘there is a lot more we need to figure out in the realm of prostate cancer, but it’s the beginning of a road map for precision oncology, and before this, we didn’t have a road map’ [93]. Therefore, future cancer research should focus on integrating critical information pertaining to genetic aberrations and mutational profiles with artificial-intelligence-aided digital pathology and automated-image analysis to predict the clinical and therapeutic outcomes in a subtype-specific manner (see Outstanding Questions). Taken together, these integrative transdisciplinary efforts will enhance our understanding of the pathobiology of prostate cancer and facilitate the development of sensitive and robust predictive biomarkers for precision cancer medicine.

Clinician’s Corner.

Multiclonality and evolution of resistant tumor clones with secondary mutations due to druginduced selective pressures remain challenging and must be considered while designing therapeutic strategies and clinical trials.

Clinicogenomic studies using noninvasive liquid biopsies collected pre-treatment, on therapy, and resistant stages offer opportunities for early detection of molecular events that might rationalize resistance-induced changes in the tumors.

Preclinical and clinical evaluation of targeted therapies involving PROTACs, which utilizes endogenous proteasomal degradation machinery, represents a novel class of small molecule inhibitors that might effectively target oncogenic transcriptional activators/coactivators.

Screening of clinically relevant genomic aberrations involving actionable RAF fusions, SPINK1 overexpression, SPOP, FOXA1, or IDH1 mutations, deletions of CHD1 or MAP3K7, DNA repair defects, or immunogenic CDK12 mutations would guide the development of precision medicine for these distinct molecular subtypes.

The use of evidence-based sequencing approaches or combinatorial therapeutic strategies should be employed for patients harboring DNA repair defects along with immune activation (such as patients with mismatch repair defects) would predict better treatment response.

Outstanding Questions.

How can genomics and functional genomics approaches be systematically integrated to fully understand the role of novel somatic or genetic molecular events in tumor initiation, progression, and therapy-induced drug resistance?

Not all molecular aberrations are drivers, many of these are mere passenger alterations, thus, how can we accurately identify these events for their clinical utility?

Can we effectively predict the clonal evolution of resistant tumor clones by employing serial evaluation of tumor biopsies collected during the course of therapeutic trial using sensitive techniques and integrative analysis, such as single cell sequencing?

Can artificial-intelligence-based approaches through the integration of digital pathological features with genomics, transcriptomics, and functional genomics data help to develop precision medicine for diverse cancer molecular subtypes?

How could unsuccessful clinical trials be utilized to improve our understanding and provide ways to overcome the mechanisms of therapy resistance?

Highlights.

Several distinct subtypes marked by germline or somatic mutations, gene fusions, focal deletions, and amplifications have been identified in the ETS-fusion-negative subtype.

Novel regulatory mechanisms underlying increased SPINK1 levels, include: (i) AR-mediated transcriptional repression of SPINK1; (ii) EZH2-mediated epigenetic silencing of miR-338-5p/miR-421; and (iii) negative post-transcriptional regulators of SPINK1.

Aberrations in epigenetic regulators or chromatin remodelers, ubiquitin, and spliceosome machineries represent novel subclasses of ETS-fusion-negative PCa.

Mutations in DNA damage and repair genes sensitize PCa cells to PARP inhibition, increases neoantigens and T cell infiltration hence sensitizing them towards immune checkpoint inhibitors.

Sequentially collected liquid biopsies support identification of therapy-resistant clones, and reveal the mechanism involved in tumor evolution.