Similarities and differences in gene expression profiles of BRCA1 methylated and mutated epithelial ovarian cancers

Introduction BRCA1 methylated (BRCA1met ) epithelial ovarian cancer (EOC) is a recently defined and not well-investigated subset of neoplasms. To date, no studies have focused on the transcriptional profiles of BRCA1met cases, and, as a matter of fact, we still do not know if this subset of EOCs is similar, and to what extent, to BRCA1 mutated (BRCA1mut ) cases. Methods We compared a group of 17 BRCA1met cases against 10 BRCA1mut cases using a subset of carefully selected 17 BRCAwt EOCs as a control group. Results First, BRCA1met cases showed a downregulation of the relative transcript, while this association was not observed for BRCA1mut EOCs. The BRCA1met group exhibited a general upregulation of homologous recombination (HR)-related genes, as well as BRCA1mut . Overall, BRCA1met had a different gene expression profile, characterized by diffuse downregulation, whereas BRCA1mut showed a general upregulation (p < 0.0001). Both BRCA1-defective groups showed a slightly activated immune response mediated by interferon (IFN) gamma pathways. Discussion In conclusion, even if the expression profile of many genes related to DNA damage and repair system is shared between BRCA1mut and BRCA1met EOCs supporting that BRCA1met EOCs may benefit from PARPi therapies, our data demonstrate that BRCA1mut and BRCA1met EOCs show different expression profiles, suggesting a different mechanism of carcinogenesis that can be reflected in different responses to therapies and disease recovery.


Introduction
Epithelial ovarian cancer (EOC) is the most common lethal malignancy among gynecological cancers with a poor outcome due to late diagnosis caused by a lack of early signs and symptoms of this disease (1).The standard treatment for EOCs is cytoreductive surgery and platinum-based chemotherapy.Recently, therapy with PARP inhibitors (PARPi) has been proven to be effective for patients with EOC showing homologous recombination deficiency (HRD), which is mainly observed in EOCs harboring germinal/ somatic pathogenetic variants or promoter methylation of the genes involved in homologous recombination (HR) pathway, mainly BRCA1 and BRCA2 genes (2,3).In Italy, PARPi therapy has been approved as a maintenance treatment for patients with HRD EOCs in the front-line setting after chemotherapy as well as in the secondline treatment of recurrent platinum-sensitive disease regardless of HRD status.However, the most important predictive biomarkers for PARPi sensitivity are BRCA1 and BRCA2 pathogenetic variants and more recently HRD status (4)(5)(6)(7).
In recent times, a growing body of research has identified a subgroup of EOCs that exhibit a phenomenon called promoter hypermethylation of BRCA genes.Interestingly, this subgroup is even more prevalent than the subgroup of EOCs with somatic BRCA mutations.Moreover, patients with BRCA-methylated EOCs tend to have a more favorable prognosis compared to those with unmethylated EOCs (8)(9)(10)(11).Despite these findings, the nature of BRCA-methylated EOCs remains poorly understood, particularly in the context of personalized cancer therapy.No studies have yet investigated the expression patterns of these cases in relation to EOCs with BRCA mutations.It is still uncertain whether BRCAmethylated cases share a similar transcriptional profile with mutated cases.Furthermore, from a clinical perspective, it is unclear whether this subset of cases exhibits the same sensitivity to platinum-based and PARPi therapies as cases with BRCA mutations do.
To this aim, the expression profiles of 750 genes were analyzed in a series of 47 EOCs composed of methylated BRCA (BRCA met ), BRCA1-mutated (BRCA mut ), and BRCA wild-type/unmethylated (BRCA wt ) EOCs in order to find out the differences and similarities between epigenetically and genetically BRCA-defective groups and to explore a biomarker signature, if any, that could better predict response and sensitivity to platinum and PARPi therapies in these subsets of EOCs.

EOC series
In this retrospective study, we analyzed a total of 47 samples of formalin-fixed and paraffin-embedded (FFPE) EOCs collected in the Department of Pathology ASST Sette Laghi in Varese (Italy) from 2008 to 2019.All EOCs were tested for somatic and germline BRCA1/2 variants and BRCA1/2 promoter methylation.
All primary EOCs were reviewed and classified by two independent pathology experts in gynecopathology according to the new 2020 World Health Organization (WHO) classification system (14).Ninety-two percent of EOCs were high-grade, and the most frequent histotype was serous (73%).The mean age of onset in the BRCA mut group was significantly lower than in the BRCA met group (49.2 versus 58 years, respectively; p = 0.0083; Supplementary Table 1).
All analyses were performed in agreement with the Declaration of Helsinki, and the study was approved by the Research Ethics Committee of ATS Insubria (ID 238/2018).

RNA isolation and NanoString nCounter ® PanCancer IO360
Total RNA was obtained after manual microdissection from three representative FFPE sections (8 µm) of 47 ovarian tumor samples using the Maxwell ® RNA FFPE Kit and Maxwell 16 system (Promega, Madison, WI, USA) according to the recommendations of the manufacturer.RNA was quantified using Qubit ™ RNA XR Assay Kit (Invitrogen-Thermo Fisher Scientific, Waltham, MA, USA) and conserved at −80°C until use.
Gene expression analysis was conducted on the NanoString nCounter ® gene expression platform (NanoString Technologies, Seattle, WA, USA) using the NanoString PanCancer IO 360 ™ code set, as we previously described in Bolzacchini et al. (15).NanoString nCounter ® PanCancer IO360 code set included 20 reference genes and 750 target genes that were grouped in 13 categories as listed in Supplementary Table 2, such as Release of Cancer Cell Antigens (74 genes involved in DNA damage response and repair), Tumor-Intrinsic Factors (155 genes), and Common Signalling Pathways (162 genes included in Wnt, Hedgehog, TGF-beta, NF-kB, Notch, PI3K-Akt, and RAS-MAPK pathways).In detail, a total of 250-300 ng of RNA per sample was mixed with a 3′ biotinylated capture probe and a 5′ reporter probe tagged with a fluorescent barcode.
The probes and target transcripts were hybridized overnight at 65°C for 12-16 h, according to the manufacturer's recommendations.The hybridized samples were run on the NanoString nCounter ® preparation station using the high-sensitivity protocol and scanned at high resolution (280 fields of view (FOVs)) on the nCounter Digital Analyzer.The resulting data file in RCC format was used for data analysis.

NanoString nCounter ® PanCancer IO360 analysis and statistics
The RCC data for each sample were normalized to internal controls by using nSolver 4.0 software.The obtained counts were then normalized to the geometric mean of 20 endogenous housekeeping genes followed by log2 transformation.Gene expression signatures were calculated as a weighted linear average of the constituent genes (15).The weighted scores used for the calculation of the signatures are NanoString ® intellectual property.Normalized gene counts and signature scores were compared to molecular and clinical features.The log2 fold-change, Wald-type confidence interval, and p-value were calculated for each gene and signature.
STRING database of known and predicted interaction was used to identify any possible protein-protein interactions between differentially expressed genes obtained by NanoString ® analysis (16).The interactions included direct (physical) and indirect (functional) associations (from computational prediction, knowledge transfer between organisms, and interactions aggregated from primary databases).
Gene list enrichment analysis was performed on April 2023 using ToppGene Suite (17) (updated 8-ago-2022), a one-stop portal for gene list enrichment analysis and candidate gene prioritization, by using the ToppFun tool.The analysis uses 14 annotation categories including gene ontology (GO) terms, pathways, protein-protein interactions, protein functional domains, transcription factor-binding sites, microRNAs, tissue-specific gene expression, and literature.The enrichment significance cutoff level was set to 0.05.GraphPad v.5.0 (GraphPad Software Inc., San Diego, CA, USA) software was used for the statistical analysis.In detail, Student's ttest and ANOVA followed by the Bonferroni test, Pearson's chisquare test, and linear regression analyses were performed, and a pvalue of <0.05 was considered significant.
Interestingly, BRCA1 expression evaluation considering the BRCA1 mut group did not highlight differences versus BRCA wt EOCs, indicating that BRCA1 mutated cases did not display a downregulation of BRCA1 expression levels.No differences were recorded considering truncating or stop variants (n = 8) versus missense variants (n = 2, Supplementary Table 1), although very few cases were available for this analysis.As reported in the literature (18), the majority of EOCs have a complex chromosome constitution showing several polysomies and of consequence several BRCA1 alleles.On this ground, it is expected that BRCA1 mut transcripts are comparable with BRCA wt transcripts.In addition, NanoString ® probes detecting BRCA1 and BRCA2 mRNAs map on exon 13-14 junction of BRCA1 and on exon 10 of BRCA2 transcripts.
Then, BRCA2 expression was compared in the four BRCA2methylated cases (one of those showed both BRCA1 and BRCA2 methylation) with respect to the BRCA wt group, but, unexpectedly, no significant differences between the two groups were observed (6.27 vs. 6.03 mean log2 counts, respectively; Figure 1B).This finding may suggest that the BRCA2 methylation test did not address informative BRCA2 promoter regions-if any-that are associated with gene silencing, casting doubts about the performance of the BRCA2 methylation test used.Thus, to keep data consistency and exclude any possible bias, the three BRCA2 methylated cases were eliminated from further analyses (Supplementary Table 1, ID# 28, 29, 39).Surprisingly, BRCA1-defective classes (10 BRCA1 mut and 17 BRCA1 met ) showed a significant upregulation of BRCA2 mRNA versus BRCA wt class (7.24 and 7.21, respectively versus 6.03 mean log2 counts, p = 0.0122 and p = 0.0039, Figure 1B).

DNA damage and repair system genes are upregulated in BRCA1-defective EOCs
Based on what was observed for BRCA2 gene expression and in order to understand if other genes with overlapping functions may compensate for the putative impairment of BRCA1, we compared the expression of 61 genes involved in DNA damage and repair in BRCAdefective cases with BRCA wt EOCs (Supplementary Table 3, all genes).We observed that, beyond BRCA1 and BRCA2 genes, the other 17 genes involved in DNA damage and repair were differently expressed (DE) in BRCA1-defective cases with respect to BRCA wt (Table 1, Supplementary Table 1 and Supplementary Figure 1).In detail, in both BRCA1defective groups, HELLS, MRE11, UBE2C, and UBE2T were upregulated with respect to BRCA wt EOC.Nine genes (BRIP1, DTX3L, H2AFX, MSH2, PARP12, PARP9, RAD51, TYMS, and WDR76) were upregulated in BRCA1 mut only, whereas EXO1, FANCA, and MGMT were upregulated exclusively in BRCA1 met cases versus BRCA wt EOCs.Curiously, in BRCA1 met , we found solely one gene downregulated with respect to BRCA wt EOCs, namely, the DDB2 gene.Interestingly, the DDB2 promoter region contains a response element for p53-BRCA1-E2F downstream of the transcription initiation site, and it has been demonstrated that BRCA1 downregulation leads to the downregulation of DDB2 expression (19,20).
Although all these DE genes are known to be involved in DNA damage and repair systems, such as HR, base excision repair (BER), nucleotide excision repair (NER), mismatch repair (MMR), nonhomologous end joining (NHEJ), and chromatin remodeling (CR) (Table 1), we wondered if crosstalk between these systems would exist.Supplementary Figure 1 shows that these DE genes are connected by shared function or are part of a physical complex.In addition, when considering the direct interactions only (Supplementary Figure 1B), STRING analysis identified a main group including 12 interacting proteins, whereas HELLS-WDR76, MGMT, DDB2, and DTX3L-PARP9-PARP12 do not directly bind a BRCA1-associated complex.

BRCA1met EOCs show a distinctive gene expression profile
The comparative analysis between EOC groups was extended to all 750 genes.Remarkably, a total of 196 genes were differentially expressed in BRCA1-defective cases with respect to BRCA wt EOC (Supplementary Table 3).Interestingly, among all the pathways activated or repressed in BRCA1-defective EOCs compared to BRCA wt cases, both the BRCA1 met and BRCA1 mut groups shared significant upregulation of proliferation signature (p = 0.02232 and p = 0.01051, respectively), suggesting a more aggressive behavior of these EOCs in comparison to HR proficient ovarian cancers (Table 2).Regarding anti-tumor immune activity, no difference in Tumor Inflammation Signature (TIS) was observed among all classes of EOCs, suggesting a general "cold" immunological profile.Of note, 53 out of 196 genes resulted in DE in both BRCA1 met and BRCA1 mut EOCs in comparison to BRCA wt cases, possibly indicating a common pathogenetic profile driven by BRCA1 deficiency.Twenty-seven genes were downregulated, while 26 genes were upregulated in both BRCA1-defective groups.The 27 downregulated transcripts were significantly enriched by genes involved in cell attachment to extracellular matrix (ECM), cell death, and apoptosis by regulating p53, WNT, RAS, or PI3K/AKT pathways, whereas the 26 upregulated transcripts were enriched by genes involved in cell cycle regulation, DNA damage and repair, chromosome organization, and immune response (Supplementary Table 3).
Interestingly, a total of 58 out of 196 genes were DE exclusively in BRCA1 mut EOCs (Supplementary Table 3, Figure 2A): only 13 genes were downregulated, while 45 genes (78%) were significantly upregulated.In detail, the 13 downregulated transcripts were significantly enriched by genes involved in cell-cell adhesion and regulation of apoptotic process (such as ARG2, SERPINB5, PVR, CASP9, and WNT5B genes); instead, the 45 upregulated genes were significantly enriched by functions as innate immune response and cell cycle regulation (Supplementary Table 3, Figure 2C).In fact, both MHC2 and interferon (IFN) DOWNSTREAM signatures were significantly upregulated in this subset of EOCs (Table 2).
Finally, 85 out of 196 genes were DE exclusively in the BRCA1 met group (Supplementary Table 3, Figure 2B): 73 of these genes (86%) were significantly downregulated, whereas only 12 genes were upregulated in comparison to BRCA wt EOCs.The 73 downregulated  transcripts were enriched by genes involved in the regulation of cell death, cytokine-cytokine interaction, and development, while the 12 upregulated transcripts were enriched by genes of regulation of the immune system such as ICOSLG, BID, CXCL16, and VTCN1 and DNA repair genes (Supplementary Table 3, Figure 2D).In detail, in BRCA1 met EOCs, a significant downregulation of MAST CELLS, TH1 CELLS, and NK CD56DIM signatures were identified, whereas a significant upregulation of IFN GAMMA signature was observed (Table 2).
In spite of a cold immune signature background, both BRCA1defective groups show a slightly activated immune response mediated by IFN gamma pathways.
Overall, comparing the expression profiles of the BRCA1 mut and BRCA1 met groups, BRCA1 met EOCs show a significant strong gene downregulation with respect to BRCA1 mut cases (chi-square p-value < 0.0001, Figures 2C, D), suggesting that, irrespective of the DE gene functions, remarkably different oncogenic mechanisms underlying the tumor initiation phase of BRCA1 mut or BRCA1 met EOC exist.

Discussion
Much clinical evidence demonstrated that BRCA pathogenetic variants are positive predictors of response to platinum-based and PARPi therapies (3,11,(21)(22)(23).HRD status resulting from the disruption of genes involved in the HR pathway is currently considered a useful clinical tool to detect PARPi-eligible cases (5).However, HRD status embraces a class of heterogeneous tumors showing a wide spectrum of pathogenetic mechanisms, including Top genes that are differentially expressed in BRCA1 mut and BRCA1 met EOCs.Volcano plots display each gene's fold change (x-axis) and significance (pvalue, y-axis) between (A, C) BRCA1 mut (B group) and BRCA wt (A group) samples and (B, D) between BRCA1 met (C group) and BRCA wt (A group) samples.Genes that have greater statistical significance will produce points that are both larger and darker in hue, in addition to appearing higher on the plot.Genes that have greater differential expression versus the baseline group (BRCA wt , A group) appear further from the center of the plot.somatic, germline, or epigenetic alteration of several HR genes.Thus, HRD status might not a suitable marker to predict specific therapy responses and patient outcomes.Despite this, the HRD test is spreading through many laboratories, without a consolidated knowledge about the overlap between the HRD test and, in particular, BRCA methylation that characterized approximately 20% of BRCA wild-type EOCs (9).To date, no studies have focused on the transcriptional profiles of BRCA methylated cases, and, as a matter of fact, we still do not know if this subset of EOC is similar, and to what extent, to BRCA mutated cases.
In this study, we compared the expression profile of BRCA1 met and BRCA1 mut EOCs with respect to carefully selected EOC groups without germline or somatic defects in HR genes (namely, BRCA wt EOCs) using a highly sensitive and precise transcript quantification method (24).
As a preliminary step, we verified that BRCA1 met cases had a downregulation of BRCA1 transcript, irrespective of the degree of promoter methylation.This is in line with what was reported in our previous work by using a real-time approach on a small subset of BRCA1 met EOCs (8).Of note, the setting of a cutoff of methylation with a clinical meaning is a tricky issue, even in the light of the chromosome instability with chromosome 17 aneuploidies that characterize virtually all EOCs (3,25,26), the tumor purity of each analyzed sample, the type of method used for methylation quantification, and the size of validation cohorts (27, 28).On the contrary, in BRCA1 mut cases, the gene dosage was not altered, neither in cases showing a truncating mutation, as it is expected that the loss of function is due to a pathogenetic non-functional protein variant.Indeed, epigenetic and genetic BRCA1 alterations are mutually exclusive as extensively demonstrated in the literature (3,9,28,29).
Compared to BRCA wt EOC, both BRCA1 mut and BRCA1 met cases display an upregulation of many genes related to DNA damage and repair system, among them BRCA2 transcript, supporting the concept, better explained elsewhere (30), that the perturbation of a particular gene's function may alter the expression of other genes within the same network as a mechanism to try to maintain cellular wellness (31, 32).Moreover, both subsets show the upregulation of pathways related to cell proliferation and interferon gamma.These findings were extensively demonstrated for BRCA1 mut cases; in fact, several authors proved that BRCA1 loss leads to transcriptional reprogramming in tumor cells involving type I IFN signaling (33)(34)(35).
Considering all the 196 DE transcripts, the analysis of single genes highlighted a prevalent "upregulation profile" in BRCA1 mut and a prevalent "downregulation profile" in BRCA1 met (Figure 2).This finding suggests that the pathogenetic mechanisms of BRCA1 mut and BRCA1 met may be very different and that widespread methylation may characterize the initial tumorigenesis of the latter group.Interestingly, many authors have reported that a subset of ovarian cancers can be characterized by extensive gene hypermethylation, suggesting the existence of a CpG Island Methylator Phenotype (CIMP) in the ovarian site (36)(37)(38)(39).This phenotype has been deeply characterized and illustrated in colorectal cancer (CRC), where CIMP-CRC has been classified as a distinctive entity, with MLH1 methylation as the flag of a diffuse methylator pattern (40).Transposing this concept to the ovarian site, BRCA1 methylation could be the hint to identify such cases.Moreover, the presence of widespread methylation might prompt further investigation to identify epigenetic drugs, as well as to address potential different clinical outcomes as reported for other tumor sites (41).
Curiously, despite that the genes deregulated in the BRCA1 mut and BRCA1 met EOCs are remarkably different, the enrichment analysis suggests that, overall, the two groups share the same perturbed pathways.This finding, derived from the study of a limited proportion of the transcriptome, i.e., 750 genes grouped in 48 pathways, deserves to be further investigated by global expression analysis.In fact, BRCA1 mut and BRCA1 met EOCs may result as different entities, presenting even more remarkable differences than those highlighted by this study.
Furthermore, these findings indicate the potential expansion of this approach to other tumor sites where BRCA is implicated, aiming to explore whether methylation is a prevalent characteristic across different types of tumors.
In summary, this study represents the first instance where a correlation has been established between promoter methylation, as identified through a BRCA1 methylation test developed in-house, and transcript downregulation.Moreover, BRCA1 met EOCs are characterized by diffuse gene downregulation, suggesting that this subset shows a pathogenetic mechanism remarkably different with respect to BRCA1 mut cases.Even if the expression profile of many genes related to DNA damage and repair system is shared between BRCA1 mut and BRCA1 met EOCs, supporting that BRCA1 met EOCs may benefit from PARPi therapies, our data demonstrate that BRCA1 mut and BRCA1 met EOCs have different expression profiles underlying different mechanisms of carcinogenesis that can influence the type of responses to therapies and disease recovery.
Genes further to the right indicate an increase in expression, and genes further to the left indicate a decrease in expression relative to the baseline group.Horizontal lines indicate 0.01 and 0.05 p-values.Blue dots indicate adjusted p-values lower than 0.05.Green dots indicate adjusted p-values higher than 0.05.The lower panels focus only on the 196 significantly DE genes in BRCA1 mut and BRCA1 met EOCs: red-dotted square indicates upregulated genes in BRCA1 mut ; bluedotted square indicates downregulated genes in BRCA1 met .EOC, epithelial ovarian cancer; DE, differentially expressed.

TABLE 1
Significantly upregulated/downregulated DNA damage and/or repair genes in BRCA-defective cases (BRCA1 mut and BRCA1 met ) compared to BRCA wt EOCs.