Aged and BRCA-Mutated Stromal Cells Drive Epithelial Cell Transformation

Abstract

The fundamental steps in high-grade serous ovarian cancer (HGSOC) initiation are unclear, presenting critical barriers to the prevention and early detection of this deadly disease. Current models propose that fallopian tube epithelial (FTE) cells transform into serous tubal intraepithelial carcinoma (STIC) precursor lesions and subsequently into HGSOC. In this study, we report that an epigenetically altered mesenchymal stem cell niche, termed high-risk mesenchymal stromal/stem cell (hrMSC), exists prior to STIC lesion formation. hrMSCs are enriched in STIC stroma and contribute to a stromal "field effect" extending beyond the borders of the STIC lesion. hrMSCs promote DNA damage in FTE cells while also fostering FTE cell survival. hrMSCs induce malignant transformation of the FTE, resulting in metastatic cancer in vivo, indicating that hrMSCs promote cancer initiation. hrMSCs are significantly enriched in BRCA1/2 mutation carriers and increase with age. Combined, these findings indicate that hrMSCs can incite ovarian cancer initiation and have important implications for ovarian cancer detection and prevention.

Significance: This work demonstrates a critical role of fallopian tube stromal cells in HGSOC initiation with implications for the pathophysiology of HGSOC formation and the development of prevention and early detection strategies critically needed in this disease. Additionally, the identification of stromal-mediated epithelial transformation has broad implications for understanding pan-cancer initiation. See related commentary by Recouvreux and Orsulic, p. 1093.

Figure 1.

hrMSCs are enriched in the stroma underlying and adjacent to fallopian tube STIC lesions. A, Representative H&E-stained tissue sections of normal fallopian tubes and fallopian tubes harboring STIC lesions. Histology is paired with multispectral IF images of each respective group shown. Black and white images denote cells with the phenotypes of interest. Red x’s indicate hrMSCs (WT1⁺/CD73⁺/CD90⁺/CD105⁺/CD45⁻), whereas yellow x’s indicate nMSCs (WT1−/CD73⁺/CD90⁺/CD105⁺/CD45⁻). B, Schematic showing the criteria for the Vectra spatial quantification. C, hrMSC abundance normalized to the area of ROI. D, hrMSC to nMSC ratios. E, hrMSC abundance (normalized to the area of ROI) in the expanded normal FT patient cohorts consisting of WT (n = 10 young, n = 9 aged), BRCA1 mutant (n = 9), or BRCA2 mutant (n = 6) fallopian tubes lacking STIC/HGSOC. F, Ratio of hrMSC/nMSC in normal FTs. P values were determined by ordinary one-way ANOVA with Tukey’s multiple comparisons analysis. Data points for C–F are reflective of individual ROIs. Statistics were determined on field ROIs for C and D. Statistics were determined with patient averages for E and F. G, Heatmap with unsupervised clustering of STIC stroma vs. normal stroma. H, Enrichment score of DEGs between STIC stroma vs. normal stroma (left: all, right: top 30 genes) applied to stromal location [from normal to STIC distal (dist) to STIC adjacent that contains both STIC contiguous and adjacent regions (STIC adj) to directly underlying STIC to invasive stroma]. I, Volcano plot of significantly differentially expressed genes in STIC stroma vs. normal stroma. J, Enrichment score of WT1 targets in stroma applied to stromal locations as in I. K, Correlation of stromal WT1 enrichment score with epithelial WT1 enrichment score.

Figure 2.

hrMSCs exhibit the epigenetic and phenotypic profile of CA-MSCs. A, DNA methylation array represented as a heatmap. MSCs were taken from benign patient tissues (nMSCs; black circles; n = 20), BRCA1/2^mut carrier patients without cancer (red circles; n = 10), and patients with confirmed invasive HGSOC (CA-MSC; purple circles; n = 14). B, UMAP of MSCs derived from patients in A. C, Relative fluorescence intensity (RFI) of WT1 405 was determined by flow cytometry for WT (n = 5), BRCA1^mut (n = 3), and BRCA2^mut (n = 13) MSC cell lines. WT1 405 RFI was plotted against patient age. Linear regression of WT1 405 MFI vs. patient age (gray line); R² = 0.3745. D, MSCs were categorized into age groups. E, MSCs were classified into nMSCs or hrMSCs based on WT1 405 RFI with a minimum cutoff of 2 RFI for hrMSCs. This categorization was independent of BRCA status and age. F and G, Representative histograms and violin plots of nMSC and hrMSC WT1 expression. H and I, Ovarian tumor cells (TC) were grown independently or in coculture with nMSCs, hrMSCs, or CA-MSCs under adherent (H) or nonadherent (I) conditions. Individual cells and spheroids per low powered field (LPF) were counted and graphed (n = 3). J, Tumor cell adherence to nMSCs, hrMSCs, and CA-MSCs is shown (n = 3). For D–J, P values were determined by ordinary one-way ANOVA with Tukey’s multiple comparison analysis. K and L, WT1 MFI determined by flow cytometry following coculture of nMSCs or hrMSCs in monoculture or coculture with FTE or ovarian cancer cells. M, WT1 KD and scramble shRNA control hrMSCs were generated by lentiviral transduction and validated by flow cytometry. N–P, Relative fold change (FC) of tumor cells after coculture with WT1 shRNA KD or scramble shRNA hrMSCs under (N) adherent or (O) nonadherent conditions. P, Chemoresistance assay on the aforementioned cocultures. Q–T, Similarly, WT1 OE or empty vector (EV) nMSCs were generated using lentiviral transduction. Tumor cells were cocultured with transduced lines, and tumor cell proliferation, sphere formation, and chemoresistance were assessed.

Figure 3.

hrMSCs increase FTE proliferation and stemness, induce DNA damage in FTE, and enhance the recovery of FTE following oxidative stress. A, Primary FTE were grown in coculture with matched nMSCs, hrMSCs, or alone for up to 4 days. Each day, the total number of viable cells was determined (n = 3). B, Primary FTE were grown under nonadherent conditions with or without nMSCs and hrMSCs. Spheroids were grown for 7 days and manually counted (n = 3). C, The percentage of ALDH⁺ primary FTE cells or FT190 control cells was determined by flow cytometry after 5 days of adherent coculture with or without nMSCs or hrMSCs. The percentage of total cells that are ALDH⁺ is shown as well as representative flow plots (n = 3). Primary FTE and immortalized FT190 control cells were cocultured for 24 hours alone or with nMSCs or hrMSCs. D−F, Representative IF images of FTE cells (D), γH2AX fluorescence intensity (F) and 53BP1 foci per nucleus by fluorescent microscopy (n > 3) (F). Individual data points reflect individual nuclei. G, Primary FTE and FT190 control cells were treated with 50 μmol/L hydrogen peroxide for 10 minutes in FBS-free media and supplemented with either nMSC conditioned media (CM) or hrMSC CM. G, Cell viability was determined by MTS assay (n > 3). H, Representative multispectral images of FTE with DNA DSBs (53BP1 foci) overlying either nMSC dense or hrMSC dense stroma in WT or BRCA1^mut fallopian tubes. For A–G, P values were determined by t test or two-way ANOVA with Tukey’s multiple comparison analysis.

Figure 4.

hrMSCs induce malignant transformation of primary FTE cells. Primary FTE cells were cultured under nonadherent conditions alone or with either nMSCs or hrMSCs. Cocultures were grown for 4–10 weeks and then injected into the fat pad of NSG mice. A, Total number of mice that developed tumors. B, Macroscopic representative images of primary tumors and secondary metastases. C, Tumor volume per mouse was graphed over the course of 5 months. D, Tumor cells from the primary and metastatic tumors were excised, dissociated, and passaged for up to 6 days ex vivo. E, Tumor cells were treated with 1 μg/mL cisplatin, and cell growth was determined up to 5 days (n = 3). In D and E, 12 wells per cell line per time point were analyzed, and statistics were determined by two-way ANOVA with Bonferroni correction. All three transformed lines compared with the primary epithelial line had a P value < 0.0001 in both panels. F, IVIS imaging of a representative mouse after weeks 2 and 3 of secondary initiation (n = 1). G, Gross anatomy of the tumor initiation site. Tumors were allowed to grow for 2 weeks before IVIS imaging began. H, Tumors were excised and processed for H&E and p53 IHC staining.

Figure 5.

Transformed, metastasized P53^null FTE harbor mutational hallmarks of HGSOC. Whole-genome sequencing was used to characterize mutations present in transformed P53^null FTE originating from P53^null FTE/hrMSC organoids. A, Oncoprint including mutations in genes commonly mutated in HGSOC. This includes correlations to COSMIC single- and double-base signatures that are shown [single-base substitution (SBS) and double-base substitution (DBS), respectively]. B, Summary statistics of mutational analysis including somatic synonymous and nonsynonymous mutations. C, Representative graph of single-base mutations for one sample exhibiting the pan-cancer mutational signature (signature 5). D, Gene set enrichment analysis on STIC stroma relative to normal stroma from Fig. 1 DSP. E, Enrichment score of stromal genes associated with oxidative stress–induced senescence derived from our DSP dataset. F, Gene set enrichments shared between STIC stroma and STIC epithelium. G, IF and (H) flow cytometry of hrMSCs stained with the general oxidation probe CellROX Green and CellROX DR, respectively. Representative cells are shown at 20× magnification with a 100 μm scale bar. In G and K, individual data points correspond to individual nuclei. More than 3 fields per group were taken to analyze 150–200 individual cells. In H–J, individual data points correspond to separate wells. I, Flow cytometry analysis of MSC CellROX DR after treatment with the antioxidant Trolox. J, Flow cytometry analysis of carboxyfluorescein succinimidyl ester (CSFEL)-labeled FTE cells cocultured with MSCs (n = 3). MFI data were normalized for cell number. FTE CellROX values are displayed. K, 53BP1 foci per FTE nuclei after 24-hour coculture with hrMSCs ±10 μmol/L Trolox. L, Simple linear regression correlating WT1 relative fluorescence intensity (RFI) with CellROX DR determined by flow cytometry. Individual data points represent single cells. M, CellROX DR MFI following lentiviral overexpression of WT1 in nMSCs. N, CellROX DR MFI following lentiviral shRNA KD of WT1 in hrMSCs. O, FTE or ovarian cancer cells were cocultured with either CellTrace-labeled (O) nMSCs or (P) hrMSCs. MSCs were assessed for changes in CellROX DR fluorescence. For G–P, P values were determined by the Student‘s t test. Q, WT1 overexpression induces MSC and FTE oxidative stress, resulting in increased FTE DNA DSBs.

Figure 6.

Increased oxidative stress in hrMSCs results in increased lipid peroxide breakdown products capable of inducing DNA DSBs in FTE. 4-HNE and MDA were compared between nMSCs and hrMSCs by fluorescence microscopy (n = 3) and flow cytometry (n > 3; A–J). Representative images at 20× magnification are shown. More than three fields were imaged and analyzed per condition. Silicon rhodamine (SiR) actin stain was used for cell masking. In B and G, each data point represents an individual cell, whereas in D and I, each data point represents the MFI of individual patient cell lines (n = 3 nMSC lines and n = 7 hrMSC lines). C and H, Stacked histograms of nMSCs, hrMSCs, and hrMSCs with 10 μmol/L Trolox treatment as a negative control. E and J, Flow cytometry was utilized to conduct a simple linear regression correlating WT1 expression with either MDA or 4-HNE. Individual points represent single cells. K, Representative images of MDA in hrMSCs (CTV⁺) that are cocultured with FTE (CT CSFE⁺). White arrowheads depict MDA puncta that are visible within nanotubule connections between cells (n > 3). L and M, Analysis of BODIPY C-12/CellTrace DR double-positive FTE after coculture with BODIPY C12-stained hrMSCs. hrMSCs were treated with 5 μmol/L Lat B or ethanol control overnight preceding coculture. Individual data points represent individual wells. Black circles represent FTE alone, and green circles are FTE with hrMSC coculture. %Double positive indicates FTE that received hrMSC-derived lipids. Quantification of (N) MDA-APC and (O) 4-HNE-PE MFIs in FTE cell monocultures or cocultured with nMSCs or hrMSCs. P, FTE were treated with 5–10 μmol/L 4-HNE and assayed for 53BP1 foci. Student t test was used to determine significance in all panels except P, where P values were determined by ordinary one-way ANOVA with Tukey’s multiple comparison analysis. Q, Excessive stromal lipid peroxidation contributes to increased DNA damage in FTE.

Figure 7.

Aging-associated AMPK derepression of the JNK/c-JUN/WT1 axis results in the hrMSC protumorigenic phenotype. A, DNA methylation in nMSCs, hrMSCs, and CA-MSCs at the promoter region of the PRKAG1 gene. B, qRT-PCR expression data of AMPK-related genes in nMSC vs. hrMSC. C, Hypothesized AMPK/JNK/c-JUN/WT1 axis. D–F, Characterization of phospho-AMPKα1 and total AMPKα1, JNK, and c-JUN in nMSCs and hrMSCs. Representative Western blots are shown with corresponding densitometries. Patient samples were grouped, and statistical significance was determined using the Student’s t test. For both nMSCs and hrMSCs, N > 3 patients. G and H, Simple linear regressions correlating p-JNK and p-c-JUN with WT1 expression on a per-sample basis via densitometry. I, Quantification of AMPK Western blot levels in nMSC vs. hrMSC at passage 4 vs. passage 8. J, Western blot of hrMSCs treated with increasing doses of the JNK inhibitor SP600125. Quantified band intensities are shown. K and L, Western blot and quantification of AMPKα1 and pAMPKα1 following BC1618 treatment for 24 hours. M, BC1618-treated hrMSCs were analyzed by flow cytometry for MDA fluorescence. P values were determined by ordinary one-way ANOVA with Tukey’s multiple comparison analysis. M–P, Quantification of p-JNK, p-c-JUN, and WT1 bands following treatment of hrMSCs with either BC1618 or SP600125. Q, Representative 20× images of FTE that were cocultured at a 1:1 ratio with hrMSC AMPKα1-m-GFP or empty vector (EV) transduced. The percentage of cells with >9 53BP1 foci was quantified by fluorescence microscopy. More than 3 fields per condition were taken to analyze 150–200 individual cells. The Student’s t test was used to determine the significance of Q. R, Aging-associated loss of AMPK phosphorylation and expression results in the derepression of JNK phosphorylation, resulting in increased WT1 expression, oxidative stress, and FTE DNA DSBs.