Frequent mutation of the FOXA1 untranslated region in prostate cancer

Abstract

Prostate cancer has a low somatic mutation rate but non-coding regions remain underexplored. We sequenced the untranslated regions (UTRs) of 72 established driver genes in 428 patients with metastatic prostate cancer and identified FOXA1 3'-UTR mutations in 12% of patients. The mutations were predominantly insertions or deletions, covered the entire UTR without motif enrichment, and were not detected in other cancers. FOXA1 lies in head-on orientation with the androgen-regulated non-coding gene AL121790.1, resulting in strong prostate lineage-specific bidirectional transcription across the FOXA1 3'-UTR. This suggests transcriptional activity as a cause for the localized hypermutation. The indel-dominant pattern of somatic mutation extends into the FOXA1 coding region, where it is shaped by clonal selection to yield a cluster of non-frameshift indels inside the forkhead domain. Somatic FOXA1 3'-UTR mutations may prove useful for diagnostic and screening approaches, given their high frequency and lineage specificity.

Fig. 1

Somatic mutation rates in the untranslated regions of prostate cancer driver genes. a Comparison of untranslated region coverage between our 72-gene panel and the Agilent SureSelect Human All Exon panel used by the Cancer Genome Atlas prostate adenocarcinoma (TCGA PRAD) working group. Coverage threshold was 200× for the 72-gene panel (sufficient for mutation detection from ∼10% ctDNA sample) and 50× for the Agilent SureSelect Human All Exon panel (sufficient for mutation detection from tissue with ∼30% cancer fraction). Incomplete coverage of BRAF, GNAS, and PIK3CB UTRs was due to differences in annotated UTR length between RefSeq (used for panel design) and Ensembl (used in this study). b Comparison of somatic mutation rate in the coding vs untranslated regions covered by our 72-gene panel. Mutation rate was defined as the total number of somatic mutations, divided by the number of genomic positions with >200× coverage (in megabases), divided by the number of cfDNA samples with ctDNA >2%. c Bar plot showing 15 genes with the highest number of somatic mutations, broken down by mutation type and region. Percentage of samples carrying one of the included mutation types is shown on the right. d Bar plot showing 15 genes with the highest 3′-UTR mutation rate. Mutation count was normalized by number of ctDNA-positive samples (n = 439) and 3′-UTR length in megabases. Percentage of samples carrying a 3′-UTR mutation is shown on the right. e Distribution of somatic mutations along the FOXA1 exonic regions, in our cohort (top) and in published whole-exome sequencing cohorts (bottom). For consistency with the coding region, indels in the 3′-UTR were also colored based on length, although they cannot result in frameshifts. Gray silhouettes indicate sequencing coverage

Fig. 2

A localized hypermutation process affects the entire FOXA1 3′-UTR. a Tracks showing the 72-gene panel sequencing coverage, phylogenetic conservation score (PhastCons20, 20 mammal species), and GC percentage calculated using a 20 bp sliding window. b Average GC percentage at different distances around FOXA1 3′-UTR indel mutations. c Scatter plot showing the frequency of 4-mers within 20 bp neighborhoods around FOXA1 3′-UTR indel mutations (y-axis) and the frequency of 4-mers across the entire FOXA1 3′-UTR. d Beeswarm plot showing ctDNA%-corrected allele fractions of coding region and 3′-UTR mutations in the 439 cfDNA samples with ctDNA. Allele fractions are shown separately for FOXA1 and the other 71 genes. e Evolution of FOXA1 mutation allele fractions across time. Each group of bars represents one of 17 patients that had multiple cfDNA timepoints with ctDNA > 5%. Heights of gray bars represent estimated ctDNA%. Dots represent somatic mutations, with y-axis position representing allele fraction. Lines connecting dots indicate that the same FOXA1 mutation was observed at multiple timepoints

Fig. 3

FOXA1 3′-UTR mutations are prostate cancer specific and associated with high bidirectional transcriptional activity. a Barplot showing the number and type of FOXA1 mutations in 33 TCGA cancer types (data obtained from cBioPortal); ordered by mutation rate. Bottom barplot shows FOXA1 expression level. b Scatter plot showing correlation between FOXA1 and AL121790.1 expression across 33 TCGA cancer types. c Scatter plot showing correlation between androgen receptor (AR) activity score and AL121790.1 expression across 33 TCGA cancer types. d Genomic visualization illustrating transcriptional overlap between FOXA1 and AL121790.1, inside the FOXA1 3′-UTR. Red and blue tracks show strand-specific RNA-seq coverage in prostate adenocarcinoma tissue. RNA-seq coverage was normalized by library size (reads per million, RPM). Transcription start sites (TSS) are indicated with arrows. At the bottom, AR ChIP-seq data from prostate cancer cell line LNCaP treated with synthetic androgen R1881 suggests presence of AR binding site 3 kb downstream of the AL121790.1 TSS. Location of RNA-level fusion junction for ETV1 fusions in the TCGA prostate adenocarcinoma cohort is also shown