Development, maturation, and maintenance of human prostate inferred from somatic mutations

Abstract

Clonal dynamics and mutation burden in healthy human prostate epithelium are relevant to prostate cancer. We sequenced whole genomes from 409 microdissections of normal prostate epithelium across 8 donors, using phylogenetic reconstruction with spatial mapping in a 59-year-old man's prostate to reconstruct tissue dynamics across the lifespan. Somatic mutations accumulate steadily at ∼16 mutations/year/clone, with higher rates in peripheral than peri-urethral regions. The 24-30 independent glandular subunits are established as rudimentary ductal structures during fetal development by 5-10 embryonic cells each. Puberty induces formation of further side and terminal branches by local stem cells disseminated throughout the rudimentary ducts during development. During adult tissue maintenance, clonal expansions have limited geographic scope and minimal migration. Driver mutations are rare in aging prostate epithelium, but the one driver we did observe generated a sizable intraepithelial clonal expansion. Leveraging unbiased clock-like mutations, we define prostate stem cell dynamics through fetal development, puberty, and aging.

Graphical abstract

Figure 1

Whole-genome sequencing of targeted microdissections reveals dynamics of mutational burden in normal prostatic epithelium (A) Workflow to uncover phylogenetic relationships in prostatic ductal tissue. (B) The mutational burden in normal prostatic epithelium increases with age. Points represent the estimated mutation burden of individual microdissections rather than individual clones, colored by the donor from whom they derive. The line is fitted using linear mixed effects models, with shaded area showing the CI_95%. (C) Scatterplot showing the relationship between corrected mutation burden per microdissection (y axis) with distance from the urethral origin (x axis) for the 59-year old donor. The result remains statistically significant when outlying samples with a mutation burden of more than 2,000 are excluded. The R² for the regression model is 0.09. (D) Scatterplot showing the relationship between corrected mutation burden per microdissection (y axis) with telomere length (x axis) for the 59-year old donor. The result remains statistically significant when outlying samples with a mutation burden of more than 2,000 are excluded. WGS, whole-genome sequencing. The R² for the regression model is 0.20. See also Figure S1 and Table S1.

Figure 2

Phylogenetic trees of prostatic epithelium from two distinct glandular subunits Ancestral clones that gave rise to the sampled microdissections were arranged in a lineage tree based on their co-occurrence pattern and cellular contribution to individual microdissections. The number of microdissections in which the corresponding ancestral clone contributed to at least 10% of the microdissected cells is indicated in brackets. Lengths of branches (x axis) indicate the numbers of mutations assigned to that branch. Closed circles represent coalescent events; open circles represent the location of the terminal tip of each branch. Coloring of clades is according to descendants of different embryonic cells. (A) Phylogenetic tree for a glandular subunit on the left side of the prostate. (B) Phylogenetic tree for a glandular subunit on the right side of the prostate. See also Figure S2.

Figure 3

Distribution of embryonic clones mirrors ductal morphogenesis The cellular contribution of four ancestral clones from embryonic development is displayed for the glandular subunit from the left side. Each circle marks a microdissection, and the ductal connection is indicated by gray lines. Microdissections with positive contribution from the clone are circled in black, whereas those with zero contribution are circled in gray. The most urethra-proximal microdissection is marked with an asterisk. The four embryonic clones display a wide and mainly contiguous spatial distribution. Although their general distribution overlaps, the contribution to individual microdissections is largely mutually exclusive. Two examples of microdissections with contribution from multiple clones are marked with arrows. See also Figure S3.

Figure 4

Localized expansion of adolescent clones marks increased ductal complexity during puberty The cellular contribution to the right structure from ancestral clones 73, 103, and 124 that are associated with puberty is displayed. The physical location of each microdissection has been collapsed into two dimensions using multidimensional scaling, each marked with a circle. Ductal connections between microdissections are illustrated with straight black lines. The circle representing each microdissection is colored according to the fraction of cells in that sample that derive from the particular adolescent clone. Microdissections with positive contribution from the clone are circled in black, whereas those with zero contribution are circled in gray. See also Figures S4 and S5.

Figure 5

Ancestral clones marking adult tissue homeostasis are spatially confined The distribution of six clones dated to adult tissue maintenance of the left glandular subunit is displayed. The location of individual ancestral clones is highlighted with an arrow. All adult clones can be found in the corresponding embryonic territory. The embryonic clone represented by cluster 76 simultaneously marks the most recent common ancestor of the six adult clones displayed. Microdissections with positive contribution from the clone are circled in black, whereas those with zero contribution are circled in gray. See also Figure S6.

Figure 6

Rare driver mutations trigger clonal expansion in normal prostatic epithelium (A) The R219S prostate cancer driver mutation in FOXA1 was detected in the right glandular subunit within the ancestral clone that is represented by cluster 118. Three additional clusters dating to adulthood were nested under cluster 118. This was the only detected example of sub-nesting of ancestral clones that must have existed during adult tissue maintenance. Microdissections with positive contribution from the clone are circled in black, whereas those with zero contribution are circled in gray. (B) Exemplary histology of microdissections with the FOXA1 driver mutation. Epithelial structures enclosed in green circles were subjected to WGS and contained the R219S prostate cancer driver mutation in FOXA1. Additional visible structures were not sequenced. Despite the known cancer driver mutation, all epithelial structures are within normal histological limits for an aging prostate. See also Table S2.

Figure 7

Model of clonal dynamics in normal prostatic epithelium