DNA methylation differences in noncoding regions in ER negative breast tumors between Black and White women

Introduction: Incidence of estrogen receptor (ER)-negative breast cancer, an aggressive tumor subtype associated with worse prognosis, is higher among African American/Black women than other US racial and ethnic groups. The reasons for this disparity remain poorly understood but may be partially explained by differences in the epigenetic landscape.

Methods: We previously conducted genome-wide DNA methylation profiling of ER- breast tumors from Black and White women and identified a large number of differentially methylated loci (DML) by race. Our initial analysis focused on DML mapping to protein-coding genes. In this study, motivated by increasing appreciation for the biological importance of the non-protein coding genome, we focused on 96 DMLs mapping to intergenic and noncoding RNA regions, using paired Illumina Infinium Human Methylation 450K array and RNA-seq data to assess the relationship between CpG methylation and RNA expression of genes located up to 1Mb away from the CpG site.

Results: Twenty-three (23) DMLs were significantly correlated with the expression of 36 genes (FDR<0.05), with some DMLs associated with the expression of single gene and others associated with more than one gene. One DML (cg20401567), hypermethylated in ER- tumors from Black versus White women, mapped to a putative enhancer/super-enhancer element located 1.3 Kb downstream of HOXB2. Increased methylation at this CpG correlated with decreased expression of HOXB2 (Rho=-0.74, FDR<0.001) and other HOXB/HOXB-AS genes. Analysis of an independent set of 207 ER- breast cancers from TCGA similarly confirmed hypermethylation at cg20401567 and reduced HOXB2 expression in tumors from Black versus White women (Rho=-0.75, FDR<0.001).

Discussion: Our findings indicate that epigenetic differences in ER- tumors between Black and White women are linked to altered gene expression and may hold functional significance in breast cancer pathogenesis.

1 Introduction

Evidence from both epidemiological and large-scale consortium studies supports the hypothesis that estrogen receptor positive (ER+) and negative (ER-) breast tumors derive from distinct etiologic pathways (1, 2). Compared to women diagnosed with ER+ breast cancer, those with ER- tumors in general have a poor prognosis, partly due to their aggressive phenotype and the lack of targeted therapy. ER- breast cancer is more common among Black women than White women (3, 4), but distinct reasons for these disparities remain to be elucidated.

DNA methylation, a major epigenetic mechanism, plays crucial roles in hormone-induced differentiation and tissue remodeling of the mammary gland through the life course (5). Aberrant DNA methylation patterns in breast cancer have been widely observed, and could contribute to differences in breast cancer risk between Black and White women (6–9). Studies have reported that hypermethylation at promoter regions of tumor suppressor genes, such as RASSF1A and CDH13, was inversely associated with gene expression, and that expression levels of these genes were lower in ER- tumors from Black women than those in White women, representing potential underlying tumor biological mechanisms explaining breast cancer racial disparities (10, 11). In addition, our previous research identified differentially methylated loci (DML) by tumor ER subtype and between races, with more DMLs by tumor ER subtype among Black women than that in White women; the number of race-related DMLs identified in ER- tumors were almost twice as those identified in ER+ tumors (6, 12).

Together, aberrant DNA methylation patterns have been used to dissect breast cancer risk by tumor subtype and racial groups, with potential diagnostic and prognostic applications (13, 14). However, previous studies have mainly focused on DNA methylation alterations associated with protein coding genes, with little attention on non-protein coding regions, which constitute more than 98% of the whole human genome.

It has long been known that a large portion of aberrant DMLs in breast cancer is located in intergenic regions (15), which constitute about 50% of the human genome (16). In a genome-wide expression-methylation quantitative loci (emQTL) analysis, Fleischer et al. reported several hundred regulatory elements not associated with protein coding genes whose methylation alterations were associated with different breast cancer lineages (17). Enhancers are critical cis-regulatory elements within non-coding regions, which contain the majority of cancer-associated variants based on genome-wide association studies (18). Studies have also revealed that enhancers are the most consistently differentially methylated regions and that their differential methylation is in a cell-type-specific manner, indicating the importance of enhancer methylation for epigenetic regulation of tumorigenesis (19). The underlying mechanisms could be that certain DMLs overlapping with enhancers can regulate tumor-associated genes and pathways, subsequently playing important roles in cancer (17, 20, 21). In addition, focused on small noncoding RNAs, microRNAs (miRNAs), we previously found that several hundred DMLs that mapped to miRNA genes were differentially methylated by tumor ER subtype and between Black and White women, and that their methylation levels were significantly correlated with corresponding miRNA gene expression (22). In summary, these studies highlight the importance of DMLs occurring within non-protein coding regions in relation to risk of breast cancer, especially their potential roles in explaining breast cancer racial disparities.

As described previously, we conducted genome-wide DNA methylation profiling on breast tumor tissue samples obtained from participants in the Women’s Circle of Health Study (WCHS), a case-control study designed to investigate risk factors for aggressive breast cancer in Black and White women (12, 23). Our initial analysis focused on DMLs mapping within or near protein-coding genes, and revealed that a key pro-luminal transcription factor, FOXA1, was hypermethylated and repressed in tumors from Black women compared to White women (12). Herein, we focus on DMLs mapped to non-protein coding regions due to their biological importance and limited research in the area.

Motivated by the biological significance of the noncoding genome and our research interests in understanding the higher risk of ER- breast cancer in Black compared to White women, we aimed to identify DMLs by race within ER- tumors, with a focus on DMLs located in intergenic and noncoding RNA genomic regions. In addition, we integrated both DNA methylation (Illumina Infinium 450K array) and gene expression (RNA sequencing) data to examine whether DMLs were associated with altered gene expression. Our results were then validated using The Cancer Genome Altas (TCGA) dataset. We further investigated the epigenomic context and molecular features of the top DMLs confirmed in both our and TCGA datasets to determine their regulatory potential and biological functions.

2 Materials and methods

2.1 Study population and tissue samples

Data and breast tumor tissue samples were from participants enrolled in the WCHS. Details on the study design and participant recruitment have been described previously (12). The study protocol was approved by Institutional Review Boards at all participating institutes. In-home interviews were conducted to obtain data on known and suspected risk factors for breast cancer. As part of the informed consent, >95% participants signed a release for their pathology reports and archived specimens in form of formalin-fixed, paraffin-embedded (FFPE) tumor blocks, which were obtained from the pathology departments of the treating hospitals. Data on tumor pathological features, including ER status, were extracted from the pathology reports.

2.2 DNA extraction, DNA methylation profiling, and data processing

DNA was extracted from FFPE tumor tissues among 694 women enrolled in the WCHS (Table S1) as previously described (12). Briefly, FFPE samples were deparaffinized in xylene, lysed, and incubated at 56°C with constant rotation until completely digested. Lysates were then heated at 70°C for 20min to inactivate the Proteinase K and stored at 4°C. DNA from a 5ul aliquot of FFPE lysate was purified using the DNA Clean & Concentrator-5 kit. Genome-wide DNA methylation analysis was carried out at Roswell Park Genomics Shared Resource using the Illumina Infinium HumanMethylation450 BeadChip platform, which interrogates > 485,000 CpG dinucleotides per sample at single-nucleotide resolution and covers 99% of RefSeq genes. To minimize the impact of batch effects, DNA samples from tumors were randomized on plates according to age, race, and ER status. The raw intensities from the array were extracted using GenomeStudio, and the data summarized into BeadStudio IDAT files and processed by the minfi R package. The methylation level of each CpG site, calculated as a β value, ranged between (0, 1), with 0 for absent methylation and 1 for complete methylation. In brief, the 450K array data were subjected to rigorous sample and locus specific quality control criteria, SWAN normalization, and correction for batch effects using the ComBat algorithm (24). Low quality probes (probes with detection p value > 0.05 in more than half of samples) and samples with poor detection p values (samples with detection p values < 1 x 10^–5 at more than 75% of CpG loci) were removed using the IMA package (25). We used Bowtie 2 for sequence alignment (26). Probes known to map ambiguously, exhibiting cross-reactivity, and that contain single nucleotide polymorphisms were also removed, leaving the final dataset containing 276,108 CpG loci in 694 tumor samples for final analyses (27, 28).

2.3 Data analysis pipeline

Unlike previous studies that focused primarily on CpGs located near a protein-coding gene, we investigated CpGs mapped to intergenic regions or noncoding RNAs in proximity, and validated results in the independent TCGA breast cancer cohort. We performed functional annotations for top CpGs to determine their potential roles in breast cancer racial disparity by exploring their integrative biologic context from multiple genome and epigenome databases, their associations with target gene expression, and differential gene expression patterns between Black and White women, as shown in Figure 1, the overall workflow.

FIGURE 1

Figure 1 Study flowchart.

2.3.1 Differentially methylated loci in ER- tumors between Black and White women within noncoding regions

Differences in DNA methylation β-values for each probe were evaluated in ER- tumors by race. The Wilcoxon rank-sum test was used to evaluate the statistical significance for each probe in the comparison. To adjust for multiple comparisons, the false discovery rate (FDR) was computed using the Benjamini and Hochberg approach. Differentially methylated loci in ER- tumors between Black and White women (raDMLs) were defined as CpGs with an absolute mean β value difference (|delta β|) at least 0.10 between two race groups and FDR-adjusted p value <0.05. All analyses were performed using R package.

2.3.2 Association between DNA methylation and gene expression

Genes located within a 2 Mb window centered on a CpG site were considered as potential regulatory targets of DNA methylation (29–31). Thus, for each raDML, we adopted a window width of 1Mb on either side of the CpG and assessed the correlations between CpG methylation level and relative RNA expression level of genes within this window. For each CpG/gene pair, the Spearman correlation between DNA methylation (β values) and relative gene expression levels (log counts per million, log CPM) was assessed. An independent collection of 50 fresh frozen breast tumor samples from Roswell Park Pathology Network Shared Resource (PNSR) was used for RNA extraction and relative gene expression analysis as previously described (6, 12). An association was considered significant if FDR-adjusted p value <0.05. Only significant associations were included for further functional annotations, which were then validated by repeating the same analysis using the TCGA breast cancer cohort.

2.3.3 Differential gene expression analysis

Genes with significant correlation between DNA methylation and relative gene expression (CpG/gene pairs) were analyzed for differential expression between Black and White women in ER- tumors using DESeq2 (32). P values were calculated using linear regression function of DESeq2, with adjustment of age at diagnosis and corrections for multiple testing.

2.3.4 TCGA data processing and analysis

To validate our findings, level 3 Illumina HM450 methylation data, Hiseq2 gene expression data, demographic and related clinical features (e.g., age, race, ER status, and tumor stage) of the 1,097 cases included in the TCGA breast cancer cohort were downloaded from the publicly available FireBrowse database (http://firebrowse.org). Validation of race-related methylation difference, association between methylation and associated gene expression, and differential gene expression analysis by race in ER- tumors were conducted following the same pipeline as described above.

2.3.5 Molecular feature annotation

For CpGs showing statistically significant correlation with gene expression confirmed in both our and TCGA dataset, we further investigated their epigenomic context and determined their regulatory potential. Specifically, the regional chromatin landscape in proximity to each CpG site was investigated using multiple publicly available databases. Chromatin state annotations were extracted from the Roadmap Epigenomics ChromHMM on ENCODE (E027 and E119). To facilitate functional interpretation, we focused on six ChromHMM states, including TssA (Active TSS), TssAFlnk (Flanking Active TSS), TssBiv (Bivalent/Poised TSS), EnhG (Genic enhancers), Enh (Enhancers), and EnhBiv (Bivalent Enhancer). DNaseI hypersensitive sites indicative of an open chromatin structure with potential transcriptional activity were similarly identified from the Roadmap/ENCODE reference epigenomes. GeneHancer module under the GeneCards Suite comprises a large collection of enhancer- gene association, which was used to annotate enhancer regions and their associated genes (33). We further inferred super-enhancer regions using a catalogue of super-enhancers in HMEC and MCF-7 cell lines published elsewhere (34). The UCSC Genome Browser was used to visualize genomic features centered on these CpGs. A circos plot summarizing global molecular features of these CpGs was generated using OmicCircos version 1.24.0. All statistical analyses were performed using the R statistical software.

3 Results

3.1 DMLs mapped within non-protein coding regions in ER- tumors between Black and White women

In our previous analysis (12), we identified a total of 396 raDMLs that exhibited significant differential methylation between Black and White women within the ER- breast cancer group. Of the 396 raDMLs, 276 CpGs were assigned with at least one protein-coding gene based on the Illumina self-manifestation file (https://support.illumina.com/downloads/infinium_humanmethylation450_product_files.html), leaving 120 CpGs uncharacterized. We further excluded 24 CpGs that were mapped in the protein-coding region based on an updated Ensemble gene annotation file (Ensemble Gene 104, http://www.ensembl.org/). In the end, the remaining 96 raDMLs mapped to intergenic regions or noncoding RNA regions, which were the focus of the current study (Table S2). Validation of observed methylation differences by race at each CpG site was then conducted using the TCGA breast cancer cohort.

Among the 96 raDMLs within non-protein coding regions in ER- tumors between Black and White women, 59 of these CpGs were located in intergenic regions and the remaining 37 CpGs mapped to at least one non-coding RNA gene. Consistent with our previous findings on protein-coding genes (12), there were more hypomethylated CpGs within the non-protein coding genome. Specifically, of the 96 raDMLs, 58 CpGs were hypomethylated and 38 CpGs were hypermethylated in ER- tumors from Black compared to White women.

Out of the 96 raDMLs, data on 59 CpGs were available in the TCGA dataset. Except for one CpG site (dot in red), all other probes showed consistent direction of methylation change by race (dot in black) in the TCGA dataset (Figure S1).

3.2 Associations between DNA methylation and gene expression

The 96 raDMLs were found to be associated with 1,998 unique genes, which correspond to a total of 2,357 unique CpG/gene pairs. Using previously described DNA methylation and RNA-seq data from analysis of an independent group of 50 fresh frozen breast tumor samples (6, 12), Spearman correlation between DNA methylation and gene expression was assessed for each CpG/gene pair. Among the 2,357 unique CpG/gene pairs, analysis identified significant correlations for 39 CpG/gene pairs, corresponding to 23 unique CpGs and 36 unique coding or noncoding RNA genes (FDR-adjusted p value < 0.05, Table S3). Of the 36 unique gene/RNAs, we found six long noncoding RNAs (lncRNAs), including three antisense lncRNAs (SOX9-AS1, HOXB-AS1, and HOXB-AS3), a long intergenic noncoding RNA (LINC01152), and two uncharacterized lncRNAs (LOC102723517, LOC283335). Overall, most CpG/gene pairs (25/39, 64.1%) exhibited positive correlations between methylation and gene expression, while 14/39 (35.9%) pairs showed negative correlations.

Table 1 listed the top 10 CpG/gene pairs, including 5 unique CpGs with associated genes or lncRNAs. Notably, we identified several CpG sites at which methylation was correlated with the expression of multiple genes. Cg20401567 is the top CpG site in proximity to the HOXB gene cluster, with its methylation level inversely, highly associated with the expression levels of multiple members of the HOXB gene family, including HOXB2 (correlation coefficient, rho=-0.74) and HOXB3 (rho=-0.62). Previously, we reported methylation level at cg04932551, a raDML within the gene body of FOXA1, inversely correlated with FOXA1 expression (12). In this study, we discovered another CpG, cg12212453 located at 5 kb downstream of FOXA1, whose methylation level was also strongly, inversely correlated with FOXA1 gene expression (rho=-0.67). We identified a novel CpG, cg05322837, whose methylation level was correlated with expression of two lncRNAs, LOC102723517 and LINC01152 (rho=0.65 and 0.58, respectively), and a Solute Carrier Family 39 gene, SLC39A11 (rho=-0.61). In addition, we found the methylation level of cg05199874 was positively correlated with expression of multiple genes, including the signal peptide-CUB-EGF domain-containing protein 2 (SCUBE2), a novel tumor suppressor gene, which showed inhibitory roles in breast tumor invasion and migration through concerted activities with FOXA1 (35, 36). Moreover, methylation level of cg12821539 was positively correlated with expression of ZIC5 (rho=0.62), which has been implicated as oncogenes in some cancers (37).

TABLE 1

Table 1 Top ten CpG/gene pairs based on methylation-gene expression correlation analysis.

HOXB gene family, FOXA1, SLC39A11, and SCUBE2 were found to play well-established roles in breast cancer development (12, 35, 38), and thus were the focus on our further analysis.

We further validated these findings by repeating the correlation analysis in the TCGA breast cancer cohort. Limited by data availability of HM450 DNA methylation and Hiseq gene expression in the TCGA breast cancer cohort, only 28 out of the 39 significant CpG/gene pairs identified in our analysis were available and thus included in the validation analysis. Nevertheless, 20 out of these 28 CpG/gene pairs were validated with respect to the magnitude and direction of the correlation coefficient and all reached statistical significance (Table S4). The remaining 8 CpG/gene pairs did not reach statistical significance in the TCGA cohort. As shown in Figure 2, we further showed in scatter plots for the top four highly correlated CpG/gene pairs that exhibited the most consistent correlation between methylation and gene expression in both WCHS and TCGA breast cancer cohort, respectively.

FIGURE 2

Figure 2 Scatter plot summarizing correlation analysis of top ranked CpG/gene pairs. X-axis denotes CpG site methylation levels in b value and Y-axis denotes relative gene expression of the associated genes in logCPM. For each pair [cg20401567 and HOXB2 (A, B), cg12212453 and FOXA1 (C, D), cg05322837 and SLC39A11 (E, F), and cg05199874 and SCUBE2 (G, H)], Spearman correlation was used to test the relationship between DNA methylation and gene expression in WCHS (left panel) and the TCGA (right panel) breast cancer cohort with p values and correlation coefficients labeled in the inlet.

3.3 Molecular features accounting for regulatory effects of aberrant DNA methylation

Chromatin architecture in which DNA methylation occurs provide important clues as to how DNA methylation alterations mediate their effects in disease predisposition. For each of the 23 unique CpGs (out of the 39 CpG/gene pairs) exhibiting significant correlations with its paired gene, we annotated their molecular features globally, including genomic position, ChromHMM state, DNaseI hypersensitive sites, enhancer and super enhancer sites, and their correlations between CpG methylation and RNA expression. As shown in Figure 3, cg20401567, cg12211453, cg05322837, and cg05199874, paired with the HOXB gene cluster, FOXA1, SLC39A11, and SCUBE2, respectively, were highly enriched with candidate cis-regulatory elements (cCREs), characterized by hypersensitive DNase I sites, promoter/enhancer-related ChromHMM segments, H3K4m1/2/3 and H3K27ac histone modification, implicating them as subjects of intensive gene expression regulation through DNA methylation modifications. As we showed in supplementary Figure S2, the Genome Browser plot exhibits regional genomic features of enrichment of DNase I hypersensitive site, ChromHMM segments implicating promoter, and histone marks predictive of open chromatin region around the genomic position of cg20401567. Consistent with our results, the region around cg20401567 includes not only HOXB gene family members, but also HOXB-AS genes. The relative position of cis-elements, cg20401567, HOXB2, HOXB3, and HOXB-AS1 on chromosome 17 was shown in Figure 4. Enhancer, super enhancer, and promoter annotations are obtained from various resources and their genomic positions are overlapped with each other. In addition to enrichment of cis-elements, we also observed high DNA sequence conservation across vertebrates for this region, indicative of important biological functions.

FIGURE 3

Figure 3 Visualization of genomic features of the 23 CpG sites. OmicCircos plot showed overall landscape of these CpG sites. Six tracks were displayed, from outside to inside: chromosome cytobands in black and blue, –log10 transformation of Spearman correlation p values (<0.05) between methylation and associated gene expression in red, the 23 CpG sites in navy blue, ChromHMM features of promoter and enhancer annotation in blue, hypersensitive DNase I sites annotation in brown, and super enhancer annotation of these probes in dark green. The top four CpGs and their associated gene (genes) were labeled in red.

FIGURE 4

Figure 4 Relative genomic positions of cg20401567, cis-elements, HOXB2, HOXB3, and HOXB-AS1 on chromosome 17. The chromosome ideogram was shown on top, with a region spanning this locus highlighted in red. Cg20401567, cis-elements, HOXB family members, and HOXB-AS1 in this locus were displayed at the bottom of the diagram, with genomic coordinates labeled in the middle. Cis-element annotations from various resources around Cg20401567 were labeled, including two promoter/enhancers based on GeneCards (light blue), an enhancer based on ChromHMM (light green), and a super enhancer based on Hnisz D et al. (34) (dark green). In addition, at bottom of the plot, a promoter upstream of HOXB2 was displayed (orange).

3.4 Expression differences on raDML-associated genes in ER- tumors between Black and White women

We further investigated whether the top CpG-correlated genes were differentially expressed in tumor tissues between Black and White women within the WCHS and TCGA breast cancer cohort. Except for SLC39A11, the other three genes (HOXB2, FOXA1, and SCUBE2) exhibited significantly lower expression levels in tumors from Black versus White women in both study cohorts after adjusting for age at diagnosis (Figure 5), which were consistent with observed CpG-gene expression correlation, with higher methylation at cg20401567 and cg12212453 (paired with HOXB2 and FOXA1, respectively), and lower methylation at cg05199874 (paired with SCUBE2) in Black relative to White women. These results suggested that differential DNA methylation between races and their altered gene expression may contribute to breast cancer racial differences.

FIGURE 5

Figure 5 Gene expression levels of HOXB2, FOXA1, SLC39A11, and SCUBE2 in ER- tumors between Black and White women from the WCHS cohort (A) and the TCGA (B) breast cancer cohort. Relative gene expression levels (logCPM) were presented as mean± standard errors (SE). Gene expression differences between Black and White women were tested using DESeq2 after adjustment of age at diagnosis, with adjusted p values labeled on top. Black and grey denotes relative gene expression levels in Black and in White women, respectively.

4 Discussion

Noncoding DNA regions comprise more than 98% of the human genome and play key roles in regulating gene expression through various mechanisms (39). For instance, aberrant methylation in intergenic regions, such as enhancers, has been shown to be associated with altered expression of neighboring genes, including those involved in cell cycle processes, lymphocyte activation and apoptosis (17, 40). Very few prior studies have focused on identifying differentially methylated features, especially in noncoding genome regions, in breast tumors between Black and White women, which may contribute to race-related differential predisposition to aggressive tumor subtypes (6, 10–12). To our knowledge, we are one of the few to globally investigate the role of noncoding region DNA methylation patterns in relation to breast cancer racial disparities (22, 41).

Intergenic regions and lncRNA genes, which comprise most of the non-protein coding genome, are enriched in epigenetic modifications, and have elicited great efforts to systematically annotate the regulatory elements existing within these regions. In this study, we identified several CpGs located within the noncoding genome that were differentially methylated in ER- tumors between races, with some CpGs highly correlated with expression of specific protein coding and/or lncRNA genes. Using multiple publicly available epigenome databases, we further characterized molecular features of these CpGs, such as active histone modifications, chromatin accessibility, and enhancer/super-enhancer sequences. We found that most of the 96 raDMLs mapping to intergenic regions or lncRNA genes were in at least one of these functional elements, with some CpGs enriched by multiple functional elements. The rich content of putative regulatory elements located at intergenic CpGs suggests active regulation of transcriptional activity through DNA methylation modification in these regions. Consistent with our observation, Kamalakaran et al. reported that among featured DNA methylation sites efficiently distinguishing the five major breast cancer subtypes, 70% are within non-protein coding regions, while only 30% of the sites mapped to genes encoding proteins (15). It should be noted, that the high occurrence of these functional elements within intergenic noncoding regions might be also due to the fact that intergenic CpG loci of the original HM450 chip design were preferentially selected toward biologically significant/informative sites, DNase hypersensitive sites, and differentially methylated regions (42, 43). Nevertheless, our findings that these CpGs were differentially methylated between two racial groups and associated with altered gene expression, support their potential roles contributing to breast cancer differences between Black and White women.

One of the interesting observations is that aberrant DNA methylation occurring in CpG islands or within gene body generally correlates with expression levels of gene cluster members spanning an entire continuous region (44–46). Consistent with these findings, we observed that increased DNA methylation at cg20401567 correlated with reduced expression of HOXB gene cluster members and HOXB-AS members. HOX gene clusters are large superfamilies of genes whose members play fundamental roles in cell development (47). Boimel et al. reported that HOXB2 knockdown promoted primary tumor growth in mammary adenocarcinoma cell lines, suggesting that in this context, it functions as a tumor suppressor (48). Intriguingly, there is evidence that in addition to the HOX gene coding regions, the cis-regulatory regions, including intergenic and antisense transcribed regions, provide an extra layer of regulation of HOX expression (49). We observed that the genomic region around cg20401567 is highly conserved and enriched with various cis-regulatory elements in the integrative functional annotation map, which provides direct evidence supporting this model. This CpG locus is not in a gene promoter and its methylation affects at least 7 genes in the HOXB cluster spreading over 50 kb, thus, its mechanism of gene silencing is unlikely to be the same as canonical promoter repression (50). The current study aimed to provide further insights into why Black women are more likely than White women to be diagnosed with aggressive breast cancer, particularly ER- breast cancer. Reduced expression of HOXB2 in breast tumors from Black versus White women represents a novel molecular feature which may be linked to racial differences in breast tumor biology and outcomes. Moreover, our findings support a novel regulatory site for HOXB2 activity, which could be a potential therapeutic target in breast cancer treatment.

FOXA1 is an important transcription factor playing crucial roles in mammary gland development. We previously reported a FOXA1 raDML (cg04932551), located in the gene body, which is annotated as a poised promoter. This site was hypermethylated in tumors from Black versus White women, and its methylation was inversely associated with FOXA1 RNA and protein expression in ER- tumors among Black women (12, 51). We further found that methylation and expression of FOXA1 is associated with parity and breastfeeding, suggesting a potential mechanism that links these reproductive exposures with ER- breast cancer among Black women (12). In the current study, we identified another raDML (cg12212453), located at 120 bp downstream of FOXA1, which was negatively correlated with FOXA1 expression. Our analysis did not show associations (data not shown) of cg12212453 methylation with reproductive factors, suggesting it is less likely to mediate effects of reproductive exposures on cancer predisposition. Nevertheless, this CpG might represent a novel regulatory site for FOXA1 expression and warrants further investigations.

SLC39A11 belongs to a member of a large family of membrane transport proteins participating in wide range of physiological processes (52). Significantly enhanced SLC39A family of proteins are expressed in multiple malignances including colorectal cancer, breast cancer, and esophageal cancer (53, 54). We observed a higher methylation of SLC39A11 (cg05322837) correlated with lower SLC39A11 expression levels in tumors from Black compared to that in White, implicating that differences in methylation and expression of SLC39A11 may contribute to breast cancer racial disparities. Intriguingly, methylation at cg05322837 was positively correlated with expression of two overlapping lncRNAs, LOC102723517 and LINC01152, as well as SOX9-AS1. The cg05322837 is in a putative enhancer region approximately 500 kb upstream of SLC39A11 and about 100kb downstream of the two lncRNAs, thus it is possible that SLC39A11 transcription is targeted by LOC102723517 and LINC01152, with its expression down regulated by methylation at cg05322837.

SCUBE2 is another gene with its expression level positively correlated with DNA methylation at a CpG site (cg05199874) approximately 500kb upstream. It exhibited differential expression by race both in our study population and the TCGA breast cancer cohort. SCUBE2 was reported to work synergistically with FOXA1 as a novel breast-tumor suppressor, driving the reversal of epithelial–mesenchymal transition (EMT) (36). SCUBE2 transcription was epigenetically inactivated by recruitment of DNA methyltransferase 1 onto its CpG islands during EMT while the exact CpGs remain unclear (36). We speculated that the downregulation of SCUBE2 in concert with FOXA1 is part of the EMT program that plays important roles in modulating breast-cancer cell migration and invasion. Our results implicated that EMT could also contribute to racial differences in ER- breast cancer predisposition and revealed that specific cis elements through which DNA methylation may influence two key regulators of EMT. Future studies are warranted to investigate this intriguing finding.

Transcription factors impact gene expression through binding to either positive (such as promoter and enhancer) or negative (such as silencer and insulator) regulatory elements under certain chromatin structure, which is drastically affected by epigenetic modifications including DNA methylation. Thus far, dysregulated DNA methylation in promoter region has attracted much of the research attention. How DNA methylation in other genome regions affects gene expression remains largely unknown. In our study, while the same CpG sites correlated with two genes’ expression levels in diverse direction, we also observed three CpG sites residing a small CpG island correlated with increased expression of the same genes, indicating these three consecutive CpG sites could belongs to the same regulatory element. The multiple correlation patterns between DNA methylation and gene expression suggested complex regulatory mechanisms of these cis-elements.

In summary, unlike previous studies that focused primarily on CpGs mapped to protein-coding genes, this study focuses on aberrant DML located in the noncoding genome regions, with the aim of interrogating biological mechanisms underlying the observed racial differences of high risk of ER- breast tumors in Black women compared to White women. We identified several important genes, being implicated in breast cancer pathology, with their expression correlated with aberrant DNA methylation of CpGs located in noncoding genome regions. The functional potentiality of these aberrantly methylated CpGs were further examined through integrative, molecular feature annotations. These results were subsequently validated in the independent TCGA breast cancer cohort. Our results provide new insights into the contribution of aberrant DNA methylation within the non-protein coding region to breast cancer racial disparities. Further experimental validations will be warranted to confirm these findings in future studies.

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Ethics statement

This study was approved by the institutional review boards of Roswell Park Comprehensive Cancer Center. The patients/participants provided their written informed consent to participate in this study.

Author contributions

CA, ZG and MH conceived and supervised the projects. QH, TK, and SL finished acquisition of all data. JC and QH did all statistical analysis and wrote the original draft. JC, MH, ZG, and CA reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

Funding

This work was supported by the Breast Cancer Research Foundation, the National Cancer Institute (R01 CA133264; R01 CA246688; P01 CA151135), and used Roswell Park Comprehensive Cancer Center’s Data Bank and BioRepository, Genomics Shared Resource, and Bioinformatics Shared Resource (P30 CA016056).

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2023.1167815/full#supplementary-material

Supplementary Figure 1 | Validation of raDMLs identified in current (WCHS) study using the TCGA breast cancer cohort. Scatter plot of delta beta value of raDMLs from WCHS (X-axis) versus TCGA (Y-axis). Fifty-nine out of the 96 raDMLs were available in the TCGA breast cancer dataset. Fifty-eight raDMLs identified in both data, with a consistent direction of methylation changes (delta beta) and FDR-adjusted P<0.05, were plotted as black dots, whereas one CpG site showing an inconsistent methylation change was labeled in red. Cg20401567 was highlighted and labeled in green.

Supplementary Figure 2 | An integrative annotation showing molecular features of the genomic regions of Cg20401567/HOXB/HOXB-AS gene clusters. From top to bottom, the tracks showing: HOXB and HOXB-AS gene clusters; the Cg20401567 site highlighted in vertical cyan line; measurements of evolutionary conservation from alignments of 100 vertebrate species; conserved transcriptional binding sites; the hypersensitive DNaseI sites profile of HMEC and MCF cell lines of the ENCODE project; the HMEC ChromHMM tracks indicating putative active (bright red) promoters, strong enhancer (orange), strong transcript (green), as well as putative weak enhancers (yellow); histone modifications of H3K27ac, H3K4me1/3 in various breast cell types.

References

1. Garcia-Closas M, Chanock S. Genetic susceptibility loci for breast cancer by estrogen receptor status. Clin Cancer Res (2008) 14(24):8000–9. doi: 10.1158/1078-0432.CCR-08-0975

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Anderson WF, Rosenberg PS, Prat A, Perou CM, Sherman ME. How many etiological subtypes of breast cancer: two, three, four, or more? J Natl Cancer Inst (2014) 106(8):dju165. doi: 10.1093/jnci/dju165

PubMed Abstract | CrossRef Full Text | Google Scholar

3. DeSantis CE, Fedewa SA, Goding Sauer A, Kramer JL, Smith RA, Jemal A. Breast cancer statistics, 2015: convergence of incidence rates between black and white women. CA Cancer J Clin (2016) 66(1):31–42. doi: 10.3322/caac.21320

PubMed Abstract | CrossRef Full Text | Google Scholar

4. Amend K, Hicks D, Ambrosone CB. Breast cancer in African-American women: differences in tumor biology from European-American women. Cancer Res (2006) 66(17):8327–30. doi: 10.1158/0008-5472.CAN-06-1927

PubMed Abstract | CrossRef Full Text | Google Scholar

5. Ivanova E, Le Guillou S, Hue-Beauvais C, Le Provost F. Epigenetics: new insights into mammary gland biology. Genes (Basel) (2021) 12(2):231. doi: 10.3390/genes12020231

PubMed Abstract | CrossRef Full Text | Google Scholar

6. Ambrosone CB, Young AC, Sucheston LE, Wang D, Yan L, Liu S, et al. Genome-wide methylation patterns provide insight into differences in breast tumor biology between American women of African and European ancestry. Oncotarget (2014) 5(1):237–48. doi: 10.18632/oncotarget.1599

PubMed Abstract | CrossRef Full Text | Google Scholar

7. Davis MB, Newman LA. Breast cancer disparities: how can we leverage genomics to improve outcomes? Surg Oncol Clin N Am (2018) 27(1):217–34. doi: 10.1016/j.soc.2017.07.009

PubMed Abstract | CrossRef Full Text | Google Scholar

8. Lara OD, Wang Y, Asare A, Xu T, Chiu HS, Liu Y, et al. Pan-cancer clinical and molecular analysis of racial disparities. Cancer (2020) 126(4):800–7. doi: 10.1002/cncr.32598

PubMed Abstract | CrossRef Full Text | Google Scholar

9. Song MA, Brasky TM, Marian C, Weng DY, Taslim C, Dumitrescu RG, et al. Racial differences in genome-wide methylation profiling and gene expression in breast tissues from healthy women. Epigenetics (2015) 10(12):1177–87. doi: 10.1080/15592294.2015.1121362

PubMed Abstract | CrossRef Full Text | Google Scholar

10. Mehrotra J, Ganpat MM, Kanaan Y, Fackler MJ, McVeigh M, Lahti-Domenici J, et al. Estrogen receptor/progesterone receptor-negative breast cancers of young African-American women have a higher frequency of methylation of multiple genes than those of Caucasian women. Clin Cancer Res (2004) 10(6):2052–7. doi: 10.1158/1078-0432.CCR-03-0514

PubMed Abstract | CrossRef Full Text | Google Scholar

11. Wang S, Dorsey TH, Terunuma A, Kittles RA, Ambs S, Kwabi-Addo B. Relationship between tumor DNA methylation status and patient characteristics in African-American and European-American women with breast cancer. PloS One (2012) 7(5):e37928. doi: 10.1371/journal.pone.0037928

PubMed Abstract | CrossRef Full Text | Google Scholar

12. Espinal AC, Buas MF, Wang D, Cheng DT, Sucheston-Campbell L, Hu Q, et al. FOXA1 hypermethylation: link between parity and ER-negative breast cancer in African American women? Breast Cancer Res Treat (2017) 166(2):559–68. doi: 10.1007/s10549-017-4418-y

PubMed Abstract | CrossRef Full Text | Google Scholar

13. Chiu AM, Mitra M, Boymoushakian L, Coller HA. Integrative analysis of the inter-tumoral heterogeneity of triple-negative breast cancer. Sci Rep (2018) 8(1):11807. doi: 10.1038/s41598-018-29992-5

PubMed Abstract | CrossRef Full Text | Google Scholar

14. Stirzaker C, Zotenko E, Song JZ, Qu W, Nair SS, Locke WJ, et al. Methylome sequencing in triple-negative breast cancer reveals distinct methylation clusters with prognostic value. Nat Commun (2015) 6:5899. doi: 10.1038/ncomms6899

PubMed Abstract | CrossRef Full Text | Google Scholar

15. Kamalakaran S, Varadan V, Giercksky Russnes HE, Levy D, Kendall J, Janevski A, et al. DNA Methylation patterns in luminal breast cancers differ from non-luminal subtypes and can identify relapse risk independent of other clinical variables. Mol Oncol (2011) 5(1):77–92. doi: 10.1016/j.molonc.2010.11.002

PubMed Abstract | CrossRef Full Text | Google Scholar

16. Francis WR, Worheide G. Similar ratios of introns to intergenic sequence across animal genomes. Genome Biol Evol (2017) 9(6):1582–98. doi: 10.1093/gbe/evx103

PubMed Abstract | CrossRef Full Text | Google Scholar

17. Fleischer T, Tekpli X, Mathelier A, Wang S, Nebdal D, Dhakal HP, et al. DNA Methylation at enhancers identifies distinct breast cancer lineages. Nat Commun (2017) 8(1):1379. doi: 10.1038/s41467-017-00510-x

PubMed Abstract | CrossRef Full Text | Google Scholar

18. Sur I, Taipale J. The role of enhancers in cancer. Nat Rev Cancer (2016) 16(8):483–93. doi: 10.1038/nrc.2016.62

PubMed Abstract | CrossRef Full Text | Google Scholar

19. Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, et al. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature (2009) 462(7271):315–22. doi: 10.1038/nature08514

PubMed Abstract | CrossRef Full Text | Google Scholar

20. Bell RE, Golan T, Sheinboim D, Malcov H, Amar D, Salamon A, et al. Enhancer methylation dynamics contribute to cancer plasticity and patient mortality. Genome Res (2016) 26(5):601–11. doi: 10.1101/gr.197194.115

PubMed Abstract | CrossRef Full Text | Google Scholar

21. Cho JW, Shim HS, Lee CY, Park SY, Hong MH, Lee I, et al. The importance of enhancer methylation for epigenetic regulation of tumorigenesis in squamous lung cancer. Exp Mol Med (2022) 54(1):12–22. doi: 10.1038/s12276-021-00718-4

PubMed Abstract | CrossRef Full Text | Google Scholar

22. Gong Z, Chen J, Wang J, Liu S, Ambrosone CB, Higgins MJ. Differential methylation and expression patterns of microRNAs in relation to breast cancer subtypes among American women of African and European ancestry. PloS One (2021) 16(3):e0249229. doi: 10.1371/journal.pone.0249229

PubMed Abstract | CrossRef Full Text | Google Scholar

23. Ambrosone CB, Ciupak GL, Bandera EV, Jandorf L, Bovbjerg DH, Zirpoli G, et al. Conducting molecular epidemiological research in the age of HIPAA: a multi-institutional case-control study of breast cancer in African-American and European-American women. J Oncol (2009) 2009:871250. doi: 10.1155/2009/871250

PubMed Abstract | CrossRef Full Text | Google Scholar

24. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics (2012) 28(6):882–3. doi: 10.1093/bioinformatics/bts034

PubMed Abstract | CrossRef Full Text | Google Scholar

25. Wang D, Yan L, Hu Q, Sucheston LE, Higgins MJ, Ambrosone CB, et al. IMA: an r package for high-throughput analysis of illumina's 450K infinium methylation data. Bioinformatics (2012) 28(5):729–30. doi: 10.1093/bioinformatics/bts013

PubMed Abstract | CrossRef Full Text | Google Scholar

26. Langmead B, Salzberg SL. Fast gapped-read alignment with bowtie 2. Nat Methods (2012) 9(4):357–9. doi: 10.1038/nmeth.1923

PubMed Abstract | CrossRef Full Text | Google Scholar

27. Chen YA, Lemire M, Choufani S, Butcher DT, Grafodatskaya D, Zanke BW, et al. Discovery of cross-reactive probes and polymorphic CpGs in the illumina infinium HumanMethylation450 microarray. Epigenetics (2013) 8(2):203–9. doi: 10.4161/epi.23470

PubMed Abstract | CrossRef Full Text | Google Scholar

28. Zhang X, Mu W, Zhang W. On the analysis of the illumina 450k array data: probes ambiguously mapped to the human genome. Front Genet (2012) 3:73. doi: 10.3389/fgene.2012.00073

PubMed Abstract | CrossRef Full Text | Google Scholar

29. Jia Y, Chng WJ, Zhou J. Super-enhancers: critical roles and therapeutic targets in hematologic malignancies. J Hematol Oncol (2019) 12(1):77. doi: 10.1186/s13045-019-0757-y

PubMed Abstract | CrossRef Full Text | Google Scholar

30. Lettice LA, Heaney SJ, Purdie LA, Li L, de Beer P, Oostra BA, et al. A long-range shh enhancer regulates expression in the developing limb and fin and is associated with preaxial polydactyly. Hum Mol Genet (2003) 12(14):1725–35. doi: 10.1093/hmg/ddg180

PubMed Abstract | CrossRef Full Text | Google Scholar

31. Yip KY, Cheng C, Bhardwaj N, Brown JB, Leng J, Kundaje A, et al. Classification of human genomic regions based on experimentally determined binding sites of more than 100 transcription-related factors. Genome Biol (2012) 13(9):R48. doi: 10.1186/gb-2012-13-9-r48

PubMed Abstract | CrossRef Full Text | Google Scholar

32. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol (2014) 15(12):550. doi: 10.1186/s13059-014-0550-8

PubMed Abstract | CrossRef Full Text | Google Scholar

33. Fishilevich S, Nudel R, Rappaport N, Hadar R, Plaschkes I, Iny Stein T, et al. GeneHancer: genome-wide integration of enhancers and target genes in GeneCards. Database (Oxford) (2017) 2017:bax028. doi: 10.1093/database/bax028

PubMed Abstract | CrossRef Full Text | Google Scholar

34. Hnisz D, Abraham BJ, Lee TI, Lau A, Saint-Andre V, Sigova AA, et al. Super-enhancers in the control of cell identity and disease. Cell (2013) 155(4):934–47. doi: 10.1016/j.cell.2013.09.053

PubMed Abstract | CrossRef Full Text | Google Scholar

35. Cheng CJ, Lin YC, Tsai MT, Chen CS, Hsieh MC, Chen CL, et al. SCUBE2 suppresses breast tumor cell proliferation and confers a favorable prognosis in invasive breast cancer. Cancer Res (2009) 69(8):3634–41. doi: 10.1158/0008-5472.CAN-08-3615

PubMed Abstract | CrossRef Full Text | Google Scholar

36. Lin YC, Lee YC, Li LH, Cheng CJ, Yang RB. Tumor suppressor SCUBE2 inhibits breast-cancer cell migration and invasion through the reversal of epithelial-mesenchymal transition. J Cell Sci (2014) 127(Pt 1):85–100. doi: 10.1158/1538-7445.AM2014-595

PubMed Abstract | CrossRef Full Text | Google Scholar

37. Satow R, Aiga Y, Watanabe T, Ishizuka N, Yoneda A, Fukami K. Zic family member 5 promotes survival in human pancreatic cancer and cholangiocarcinoma cells. Biochem Biophys Rep (2022) 31:101289. doi: 10.1016/j.bbrep.2022.101289

PubMed Abstract | CrossRef Full Text | Google Scholar

38. de Bessa Garcia SA, Araujo M, Pereira T, Mouta J, Freitas R. HOX genes function in breast cancer development. Biochim Biophys Acta Rev Cancer (2020) 1873(2):188358. doi: 10.1016/j.bbcan.2020.188358

PubMed Abstract | CrossRef Full Text | Google Scholar

39. di Iulio J, Bartha I, Wong EHM, Yu HC, Lavrenko V, Yang D, et al. The human noncoding genome defined by genetic diversity. Nat Genet (2018) 50(3):333–7. doi: 10.1038/s41588-018-0062-7

PubMed Abstract | CrossRef Full Text | Google Scholar

40. Almamun M, Kholod O, Stuckel AJ, Levinson BT, Johnson NT, Arthur GL, et al. Inferring a role for methylation of intergenic DNA in the regulation of genes aberrantly expressed in precursor b-cell acute lymphoblastic leukemia. Leuk Lymphoma (2017) 58(9):1–12. doi: 10.1080/10428194.2016.1272683

PubMed Abstract | CrossRef Full Text | Google Scholar

41. Gong Z, Wang J, Wang D, Buas MF, Ren X, Freudenheim JL, et al. Differences in microRNA expression in breast cancer between women of African and European ancestry. Carcinogenesis (2019) 40(1):61–9. doi: 10.1093/carcin/bgy134

PubMed Abstract | CrossRef Full Text | Google Scholar

42. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, et al. High density DNA methylation array with single CpG site resolution. Genomics (2011) 98(4):288–95. doi: 10.1016/j.ygeno.2011.07.007

PubMed Abstract | CrossRef Full Text | Google Scholar

43. Sandoval J, Heyn H, Moran S, Serra-Musach J, Pujana MA, Bibikova M, et al. Validation of a DNA methylation microarray for 450,000 CpG sites in the human genome. Epigenetics (2011) 6(6):692–702. doi: 10.4161/epi.6.6.16196

PubMed Abstract | CrossRef Full Text | Google Scholar

44. Kulis M, Esteller M. DNA Methylation and cancer. Adv Genet (2010) 70:27–56. doi: 10.1016/B978-0-12-380866-0.60002-2

PubMed Abstract | CrossRef Full Text | Google Scholar

45. Jadhav RR, Ye Z, Huang RL, Liu J, Hsu PY, Huang YW, et al. Genome-wide DNA methylation analysis reveals estrogen-mediated epigenetic repression of metallothionein-1 gene cluster in breast cancer. Clin Epigenet (2015) 7(1):13. doi: 10.1186/s13148-015-0045-9

CrossRef Full Text | Google Scholar

46. Smith RG, Hannon E, De Jager PL, Chibnik L, Lott SJ, Condliffe D, et al. Elevated DNA methylation across a 48-kb region spanning the HOXA gene cluster is associated with alzheimer's disease neuropathology. Alzheimers Dement (2018) 14(12):1580–8. doi: 10.1016/j.jalz.2018.01.017

PubMed Abstract | CrossRef Full Text | Google Scholar

47. Afzal Z, Krumlauf R. Transcriptional regulation and implications for controlling hox gene expression. J Dev Biol (2022) 10(1):4. doi: 10.3390/jdb10010004

PubMed Abstract | CrossRef Full Text | Google Scholar

48. Boimel PJ, Cruz C, Segall JE. A functional in vivo screen for regulators of tumor progression identifies HOXB2 as a regulator of tumor growth in breast cancer. Genomics (2011) 98(3):164–72. doi: 10.1016/j.ygeno.2011.05.011

PubMed Abstract | CrossRef Full Text | Google Scholar

49. De Kumar B, Krumlauf R. HOXs and lincRNAs: two sides of the same coin. Sci Adv (2016) 2(1):e1501402. doi: 10.1126/sciadv.1501402

PubMed Abstract | CrossRef Full Text | Google Scholar

50. Fukushige S, Horii A. DNA Methylation in cancer: a gene silencing mechanism and the clinical potential of its biomarkers. Tohoku J Exp Med (2013) 229(3):173–85. doi: 10.1620/tjem.229.173

PubMed Abstract | CrossRef Full Text | Google Scholar

51. Cheng TD, Yao S, Omilian AR, Khoury T, Buas MF, Payne-Ondracek R, et al. FOXA1 protein expression in ER(+) and ER(-) breast cancer in relation to parity and breastfeeding in black and white women. Cancer Epidemiol Biomarkers Prev (2020) 29(2):379–85. doi: 10.1158/1055-9965.EPI-19-0787

PubMed Abstract | CrossRef Full Text | Google Scholar

52. Lin L, Yee SW, Kim RB, Giacomini KM. SLC transporters as therapeutic targets: emerging opportunities. Nat Rev Drug Discov (2015) 14(8):543–60. doi: 10.1038/nrd4626

PubMed Abstract | CrossRef Full Text | Google Scholar

53. Barman SK, Zaman MS, Veljanoski F, Malladi CS, Mahns DA, Wu MJ. Expression profiles of the genes associated with zinc homeostasis in normal and cancerous breast and prostate cells. Metallomics (2022) 14(8):mfac038. doi: 10.1093/mtomcs/mfac038

PubMed Abstract | CrossRef Full Text | Google Scholar

54. Liu L, Yang J, Wang C. Analysis of the prognostic significance of solute carrier (SLC) family 39 genes in breast cancer. Biosci Rep (2020) 40(8):BSR20200764. doi: 10.1042/BSR20200764

PubMed Abstract | CrossRef Full Text | Google Scholar