Targeting Tumor Heterogeneity by Breaking a Stem Cell and Epithelial Niche Interaction Loop

Abstract

Tumor heterogeneity, the presence of multiple distinct subpopulations of cancer cells between patients or among the same tumors, poses a major challenge to current targeted therapies. The way these different subpopulations interact among themselves and the stromal niche environment, and how such interactions affect cancer stem cell behavior has remained largely unknown. Here, it is shown that an FGF-BMP7-INHBA signaling positive feedback loop integrates interactions among different cell populations, including mammary gland stem cells, luminal epithelial and stromal fibroblast niche components not only in organ regeneration but also, with certain modifications, in cancer progression. The reciprocal dependence of basal stem cells and luminal epithelium is based on basal-derived BMP7 and luminal-derived INHBA, which promote their respective expansion, and is regulated by stromal-epithelial FGF signaling. Targeting this interaction loop, for example, by reducing the function of one or more of its components, inhibits organ regeneration and breast cancer progression. The results have profound implications for overcoming drug resistance because of tumor heterogeneity in future targeted therapies.

Keywords: BMP signaling; FGF signaling; drug resistance; individualized medicine; intra‐tumoral heterogeneity; microenvironment; precision medicine; targeted therapy.

Figure 1

Fgfr2 function in basal cells is required for mammary gland regeneration. A) mRNA expression of Fgfr2 in basal and luminal cells at different stages of mouse mammary gland development based on scRNA‐seq. B) Gating strategy for sorting of mammary gland cell populations based on CD24 and Itga6 (CD49f) by FACS. CD24^negCD49f^neg were stromal cells (St), CD24^hiCD49f^low were luminal cells (Lu), CD24^medCD49f^hi were basal cells (Ba). C) Relative expression of Fgfr2 in sorted luminal, basal, and stromal cells. D) Schematic diagram depicting the experimental procedure of sample preparation, adenoviral infection, FACS, and transplantation strategy. Approximately 20 000 cells were injected into the cleared fat pad. E–H) Epithelial network as revealed by Carmine‐staining in recombinant mammary glands derived from E) wild‐type basal and luminal cells, F) Fgfr2 ^∆/∆ null basal and luminal cells, G) Fgfr2 ^∆/∆ null basal and wild‐type luminal cells, H) wild‐type basal and Fgfr2 ^∆/∆ null luminal cells. At least three transplants from each group were analyzed. Scale bars: 1000 µm.

Figure 2

Stromal FGFs are essential for mammary stem cell pool expansion, but not differentiation or survival. A) Schematic diagram depicting the experimental procedure of sample preparation, adenoviral infection, FACS, and transplantation strategy. Approximately 20 000 cells were injected into the cleared fat pad. B–E) Whole‐mount or immunofluorescence on frozen sections F–G’’’) of the epithelial network 2 weeks after transplantation. The EGFP channel allowed visualization of epithelial cells derived from green wild‐type B) and Fgfr2 ^∆/∆ null D) basal cells and the tdTomato channel allowed visualization of red wild‐type luminal cells (C,E). Note that a few wild‐type basal cells gave rise to luminal cells in the middle of the epithelium (inset in F–F’’’), while many Fgfr2 null basal cells also differentiated into luminal cells as they were K8+ (G’’).

Figure 3

Expansion of mammary stem cells depends on luminal paracrine factors. A) Schematic diagram depicting the experimental procedures of organoid preparation, basal and luminal cell sorting, aggregation and in vitro culture methods. Approximately 1000 cells were used to form each aggregate. B–F) in vitro assays in which purified cells were either cultured separately as B–B’’) basal cells, C–C’’) luminal cells, or D–D’’) a mixture containing both basal and luminal at a 1:1 ratio. Aggregate growths, as measured by fold changes of their areas over E) a 4‐day culture and by the number of branches they formed after F) a 7‐day culture, were quantified. G,H) Quantitative analysis of colony‐forming efficiency by wild‐type luminal or basal cells cultured in control medium or conditioned medium from G) the basal or H) luminal cells, respectively. Note that the conditioned medium (CM) from both basal and luminal cells could stimulate growths of the other cell type, suggesting that growth stimulation was, at least partially paracrine in nature. I–L) Basal‐luminal co‐culture experiments in which wild‐type luminal cells were cultured alone (I), or together with either J) wild‐type or K) Fgfr2 ^∆/∆ null basal cells. Seven days after culture their branch numbers were quantified (L). Values shown are the mean ± SD for each data point: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. unpaired, two‐tailed Student's t tests. Scale bars: 50 µm. At least 6 aggregates were used for each data point.

Figure 4

BMP7 is a basal‐to‐luminal paracrine candidate regulated by stromal FGFs. A) Diagram depicting the experimental procedures of single‐cell analyses of transplants from mixtures of wild‐type luminal cells with either wild‐type or Fgfr2 ^∆/∆ null basal cells five days after surgery. B) Identification of basal and luminal cells were based on expression of published marker gene sets for these two cell types.^{[ 48 ]} C) GSEA analysis of pathways related to epithelial cell proliferation in the luminal cells of the control group and the experimental group, in which basal cells were wild type or Fgfr2 ^∆/∆ null cells, respectively. D) Comparison of the number of ligand‐receptor pairs between basal and luminal cells based on CellphoneDB analysis in control and experimental samples. E) Select ligand‐receptor pairs between basal and luminal cells based on CellphoneDB analysis. F) Bmp7 mRNA expression in control (Fgfr2 ^+/+) and experimental (Fgfr2 ^∆/∆) basal cells, confirming its reduced experiment in mutant basal cells. G) Western blotting and H) quantification analysis of BMP7 and the internal control GAPDH protein in the absence or presence of FGF2 stimulation using wild‐type and Fgfr2 null primary mammary epithelial basal cells. Ø denotes control group without FGF2 addition. Graph shows mean ± SD. Unpaired Student's t‐test was performed for statistical analysis, n.s., not significant, P ≥ 0.05; *P < 0.05.

Figure 5

Basal derived BMP7 promotes luminal expansion. A) mRNA expression of Bmp7 in basal and luminal cells at different stages of mouse mammary gland development based on mining scRNA‐seq datasets. B) Relative expression of Bmp7 in sorted luminal, basal, and stromal cells. C) Quantification of colonies formed by luminal cells cultured in medium containing with either control PBS or 50 ng mL⁻¹ BMP7. D–F) Luminal cells were cultured with D) either PBS or E) BMP7 added to the medium, with branch number quantified 7 d later (F). G–I) Wild‐type luminal cells were cultured with basal cells transfected with a lentiviral construct expressing G) either control shRNA or H) sh‐Bmp7, with branch number quantified several days later (I). J–L) Wild‐type luminal cells were cultured with basal cells transfected with a lentiviral construct of either J) a control or K) excessive Bmp7 expression, with branch number quantified seven days later (L). M–P) Wholemount mammary gland epithelial network observed under a fluorescent stereoscope 8 weeks after transplantation of M,N) MECs carrying lentiviral shRNAs or O,P) overexpression constructs. At least three glands were used for each data point. Q–S) Branching assay of wild‐type luminal cells cultured together with Fgfr2 ^∆/∆ null basal cells over‐expressing the mRNA of either a Q) control or R) Bmp7. Consistent with Bmp7 working downstream of Fgfr2, the ability of luminal cells to form branches was increased by approximately three‐fold due to S) Bmp7 overexpression. At least 9 aggregates were used for each data point. Values shown are the mean ± SD for each data point. n.s., not significant, *P < 0.05; ** P < 0.01. unpaired, two‐tailed Student's t tests. Scale bars: 50 µm.

Figure 6

Luminal INHBA is a BMP7 signaling target that promotes expansion of basal stem cell pool. A) Differentially expressed genes in response to BMP7 stimulation, with those significantly (fold change >1.2, adjusted‐p‐value <0.05) upregulated genes marked red and down‐regulated genes marked green. Note that several known BMP signaling target genes, including Id1, 2, and 3 were among the upregulated ones. B,C) GSEA analysis of B) the BMP signaling pathway and C) cell proliferation pathway. Both of these two pathways were upregulated in HC11 cells stimulated by BMP7 when compared with control. D) Venn diagram analysis of the 50 upregulated genes in response to BMP7 stimulation and the 20 luminal‐to‐basal ligand‐receptor pairs described above (Figure 4D). E) mRNA expression as detected by qPCR of the BMP7 candidate targets by luminal cells cultured in medium with or without BMP7 stimulation. F) Inhba mRNA expression in luminal cells of transplants derived from wild‐type luminal cells coinjected with either control (Fgfr2 ^+/+) and experimental (Fgfr2 ^∆/∆) basal cells as described above (Figure 4A). Note that its expression is greatly reduced in the experimental transplants. G) Relative mRNA expression of the BMP signaling target genes and Inhba in luminal cells cultured in medium with or without 50 ng mL⁻¹ BMP7 for the durations indicated. H) Western blotting and I) quantification analysis of INHBA and the internal control GAPDH in the absence or presence of BMP7 stimulation. Note that BMP7 significantly upregulated INHBA protein expression. J) mRNA expression of Inhba in the 8‐week developing mammary gland using the above scRNA‐seq datasets from this stage. K) Validation of relative Inhba mRNA expression in sorted luminal and basal cells at the 8‐week stage using qPCR. L) Quantification of colonies formed by 500 basal cells cultured in medium with or without 5 ng mL⁻¹ INHBA. M) qPCR examination of the relative mRNA expression of Inhba in the presence of BMP7 stimulation in luminal cells expressing Bmpr1a and Bmpr2 shRNAs, either individually or in combination. Values shown are the mean ± SD for each data point: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Unpaired, two‐tailed Student's t tests.

Figure 7

BMP7 inhibition blocks triple‐negative breast cancer progression. A) Bmp7 mRNA expression in human breast cancer subtypes using the TCGA database. B,C) Kaplan‐Meier survival curve showing that increased expression of B) Bmp7, and especially C) both Bmp7 and Inhba is correlated with a worse survival probability in patients within ≈150 months post‐diagnosis. D–F) Cell proliferation analysis as assessed by EdU‐incorporation in D) wild‐type and E) Bmp7‐KO 4T1 cells. Data are the mean ± SD. G,H) K14 immunofluorescence (red) in G) wild‐type and H) Bmp7‐KO 4T1 cells. Samples were costained with the nuclear dye DAPI (blue) to reveal both K14‐positive and ‐negative cells. (H) Quantification of the percentage of K14⁺ cells in the total 4T1 cell populations. I,J) Tumor growths from Bmp7 knockout and control 4T1 cells (I), with tumor volume quantified in (J). K,L) Tumor growths from Bmp7 gain‐of‐function and control 4T1 cells (K), with tumor volume quantified in (L). Values shown are the mean ± SD for each data point: *P < 0.05; ****P < 0.0001. unpaired, two‐tailed Student's t tests.

Figure 8

Targeting BMP7‐INHBA signaling loop inhibits progression of luminal breast cancer subtype. A) Percentage distribution of basal and luminal cells in subtypes of human breast cancer. B) Percentage distribution of basal and luminal cells in MMTV‐PyMT tumor at the ages indicated. C,D) Relative mRNA expression of C) Bmp7 and D) Inhba in luminal and basal cells of MMTV‐PyMT mice at the ages indicated. E) Gating strategy for sorting PyMT MECs based on CD24 and Itga6 (CD49f) by FACS. Note that an identical strategy was used to separate luminal cells (CD24^hiCD49f^low) from basal cells (CD24^medCD49f^hi) (see Figure 1B), but it failed to do so here. F–J) in vitro culture of PyMT MECs transfected with a lentiviral construct expressing either F,F’) a control scrambled shRNA or G,G’) individually sh‐Bmp7, H,H’) sh‐Inhba, or in combination I,I’) by which Bmp7 and Inhba expression were knocked down using the shRNA technique. MEC outgrowths, as measured by area fold‐change, were quantified (J), (n >10). K,L) Tumor growths of PyMT cells 8 weeks post‐surgery in nude mice with reduced expression of Bmp7 and Inhba via shRNA knockdown (K), with tumor volume quantified in (L). Values shown are the mean ± SD for each data point: *P < 0.05; **P < 0.01; ****P < 0.0001. unpaired, two‐tailed Student's t tests. M) Schematic diagram of a model depicting the interactions among mammary stem cells and their stromal fibroblastic and luminal epithelial niches. Fibroblast production of FGF ligands, including FGF2, FGF7, and FGF10 are essential for stem cell pool expansion and luminal epithelial morphogenesis. Stromal FGFs are also essential for BMP7 production by MSCs via FGFR2 activation, which activates BMPR1a/2 signaling of luminal niche cells. This in turn upregulates luminal production of INHBA to promote stem cell expansion and completes a positive feedback signaling loop between MSCs and the luminal niche.