A Sox2-Sox9 signalling axis maintains human breast luminal progenitor and breast cancer stem cells

Abstract

Increased cancer stem cell content during development of resistance to tamoxifen in breast cancer is driven by multiple signals, including Sox2-dependent activation of Wnt signalling. Here, we show that Sox2 increases and estrogen reduces the expression of the transcription factor Sox9. Gain and loss of function assays indicate that Sox9 is implicated in the maintenance of human breast luminal progenitor cells. CRISPR/Cas knockout of Sox9 reduces growth of tamoxifen-resistant breast tumours in vivo. Mechanistically, Sox9 acts downstream of Sox2 to control luminal progenitor cell content and is required for expression of the cancer stem cell marker ALDH1A3 and Wnt signalling activity. Sox9 is elevated in breast cancer patients after endocrine therapy failure. This new regulatory axis highlights the relevance of SOX family transcription factors as potential therapeutic targets in breast cancer.

Fig. 1

Sox9 marks luminal progenitor cells and ALDEFLUOR+cells in the human breast. a Sox9 protein levels in CD49f-EpCAM⁻, CD49f⁺EpCAM⁻, CD49f⁺EpCAM⁺ and CD49f⁻EpCAM⁺ cell populations from three different primary human breast epithelial cell samples were assessed by western blot. b Immunofluorescence analysis of Sox9 expression in CD49f⁻EpCAM⁻, CD49f⁺EpCAM⁻, CD49f⁺EpCAM⁺ and CD49f⁻EpCAM⁺ cell populations sorted from primary human breast epithelial cells from one tissue sample, as representative example. c Sox9 levels in ALDEFLUOR⁻ (indicated as ALDH⁻) and ALDEFLUOR⁺ (indicated as ALDH⁺) cells sorted from three different human breast epithelial cell samples. d A representative example of immunofluorescence analysis of Sox9 expression in ALDEFLUOR⁻ and ALDEFLUOR⁺ cells sorted from human breast epithelial cells

Fig. 2

Modulation of Sox9 levels alters human mammary stem cell phenotype. a ALDEFLUOR assay in primary breast epithelial cells stably transduced with shcontrol (shc) and shSox9 lentivirus (n = 4). b Primary (I MS) and secondary (II MS) mammosphere formation in primary human breast epithelial cells transduced with shcontrol (shc) and shSox9 lentivirus (n = 5). c Colony formation assay on Matrigel of primary epithelial cells stably transduced with shcontrol (shc) and shSox9 lentivirus (n = 4). A representative image is shown. d Colony formation assay on Matrigel of MCF10A cells stably transduced with plenti6.2-GFP c or pLenti6.2-Sox9 (Sox9) (n = 4). A phase-contrast (left) and a confocal immunofluorescence (right) images of acini stained for CD49f-APC (green), Phalloidin (red) and DAPI (blue) are shown. The pie graphs show percentage of MCF10A colonies growing as acini in Matrigel displaying different types of lumen (hollow, half-filled/half-hollow and filled). e ALDEFLUOR assay in MCF10A cells stably transduced with pLenti6.2V.DEST (c) and pLenti6.2-Sox9 (Sox9) (n = 5). f Luminal (keratin 18, K18⁺), myoepithelial (keratin 14, K14⁺) and mixed (K18⁺K14⁺) colonies formed on collagen-coated wells from human primary breast epithelial cells transfected with shcontrol (−) or shSox9 (+). Results are shown as fold change in number of colonies compared to shcontrol cells (n = 3). Representative colony images are shown. g Relative transcript levels of the indicated genes in shSox9 primary human breast epithelial cells compared to shcontrol cells (n = 4/5). Error bars represent standard deviation (SD). *p < 0.05, **p < 0,001, statistical test: two-tail t-test

Fig. 3

Sox9 is highly expressed in human breast tumours. a SOX9 mRNA expression levels in human breast tumour (T) samples compared to their normal (N) counterparts (n = 13). b Immunoblot of Sox9 and β-actin (loading control) in a set of ER-positive and ER-negative breast tumours (T) compared to the corresponding normal (N) and peritumoral (P) tissue (n = 11). c Transcript levels of SOX9 in ALDH⁻ and ALDH⁺ cells sorted from 8 different human primary breast tumours. d Immunofluorescence analysis of Sox9 expression in ALDH⁻ and ALDH⁺ cells sorted from a primary breast tumour, as representative example. e SOX9 mRNA (left) and Sox9 protein (right) levels in ER-positive (MCF7, T47D, and ZR-75–1) and ER-negative (MDA-MB-231 and MDA-MB-468) breast cancer cells, relative to levels in MCF7 cells, set as 1. f Immunofluorescence analysis of Sox9 and ER expression in MCF7, T47D and ZR-75-1 breast cancer cells. *p < 0.05, statistical test: Mann–Whitney test a; *p < 0.05, **p < 0,001, statistical test: two-tail t-test e

Fig. 4

Sox9 expression is repressed by estrogen. a Transcript levels of SOX9 and PS2/TFF1 expression in MCF7 cells after 10⁻⁸ M estrogen (E2) treatment (n = 3). b Immunoblots of Sox9 in MCF7, T47D and ZR-75-1 cells treated for 2 days with 10⁻⁸ M estrogen. c SOX9 mRNA (left) and Sox9 protein (right) levels after 10^–7 M ICI 182,780 treatment in MCF7 cells (n = 3). d Immunoblots for Sox9 in MCF7, T47D, ZR-75–1 and their corresponding tamoxifen-resistant breast cancer cells (parental (c) and TamR, respectively). β-actin, GAPDH or Hsp60 have been used as loading controls, as indicated. Error bars represent standard deviation (SD). *p < 0.05, statistical test: two-tail t-test, compared to control a, c

Fig. 5

Sox9 regulates breast cancer stem cell renewal. Transcript levels of SOX9 and SOX2 in MCF7 (a) and MDA-MB-468 (b) breast cancer cells cultured in adherent (Adh) or suspension conditions, as primary (I MS) and secondary (II MS) mammospheres (n = 3). c Immunoblot of Sox9 and β-actin (loading control) in MCF7 and MCF7TamR cells cultured in adherent (Adh) or suspension (I MS) conditions. d Primary (I MS) and secondary (II MS) mammosphere formation in MCF7TamR cells stably transduced with shcontrol (shc) and 2 different shSox9 sequences (1 and 2) lentivirus (n = 4/5). e Mammosphere formation in MDA-MB-231 and BT549 triple-negative breast cancer cells stably transduced with shcontrol (−) and shSox9 (+) lentivirus (n = 3). f Primary (I MS) and secondary (II MS) mammosphere formation in MCF10A cells stably transduced with control c and Sox9 plasmids (I MS: n = 4; II MS n = 3). g Mammosphere formation in sgRNA binding sense strand only as control (c) and four different CRISPR/Cas9n clones using a pair of sgRNAs for both DNA strands, resulting in Sox9 deletion, derived from MCF7TamR cells (n = 4). h Mammosphere formation in control (c) and Sox9 knockout (clone 1) MCF7TamR cells by CRISPR/Cas9n editing (CRISPR Sox9), as in g. Control and MCF7TamR cells lacking Sox9 were stably transfected with an empty expression vector (vector) or a vector expressing Sox9 (Sox9) and mammosphere formation was quantified, with the control cells set as 1 (n = 3). Error bars represent standard deviation (SD), p-value *p < 0.05, **p < 0.001 compared to control e, statistical test: two tailed t-test a, b, e, f, g, one-way Anova d, h

Fig. 6

Sox9 expression associates with ALDEFLUOR activity. a Immunoblot of Sox9, and β-actin as loading control, in ALDEFLUOR^– and ALDEFLUOR⁺ cells sorted from MCF7TamR cells. b Fold change of ALDEFLUOR⁺ cells in MCF7TamR and T47DTamR shcontrol (−) and shSox9 (+) cells, (n = 4). c Fold change of ALDEFLUOR⁺ cells in sgRNA binding sense strand only, as control (c) and Sox9 knockout (4 different clones) MCF7TamR cells by CRISPR/Cas9n editing (n = 3). d ALDH1A3 mRNA expression in parental MCF7 and MCF7TamR (TamR) cells (n = 3). e ALDH1A3 mRNA expression levels in sgRNA control (c) and Sox9 knockout (4 different clones) MCF7TamR cells by CRISPR/Cas9n editing (n = 3). f ALDH1A3 mRNA expression in MCF10A-GFP (c) and MCF10A-Sox9 (Sox9) cells (n = 5). g ALDH1A3 expression levels in sgRNA control (c) and Sox9 knockout MCF7TamR cells by CRISPR/Cas9n editing. The MCF7TamR cells lacking Sox9 were stably transfected with an empty expression vector (vector) or a vector expressing Sox9 (Sox9). h Chromatin Immunoprecipitation (ChIP) showing Sox9 binding to human ALDH1A3 promoter in MCF7TamR cells at two positions, A (845 bp) and B (1828 bp) upstream from the transcription start site. Data are shown as fold enrichment compared to IgG binding (n = 4). Error bars represent standard deviation (SD). *p < 0.05, compared to control, statistical test: two-tail t-test

Fig. 7

Sox9 expression is implicated in tumorigenicity in vitro and in vivo. a Soft agar colony formation assay in MCF7TamR shcontrol (shc) and shSox9 cells with different concentrations of tamoxifen (10⁻⁹–10⁻⁷ M), (n = 4). b Soft agar colony formation assay in MCF7TamR, T47DTamR and MDA-MB-231 shcontrol (−) and shSox9 (+) cells (n = 3). c Cell invasion assay of MDA-MB-231 shcontrol (−) and shSox9 (+) cells invading through Matrigel in Transwell plates (n = 3). d Representative images of MDA-MB-231 spheroids grown in Matrigel at the indicated time points from wild type (wt), sgRNA binding sense strand only as control (c) and a CRISPR/Cas9n clone using a pair of sgRNAs for both DNA strands, resulting in Sox9 deletion (CRISPR) cells. Below each photograph the analysis of the invaded area by ImageJ is shown and their quantification represented in the graph (n = 3). Arrows indicate areas of invasion. e Soft agar colony formation assay in wild type (wt), sgRNA control (c) and 4 different CRISPR/Cas9n-mediated deletion of Sox9 clones (1–4) in MDA-MB-231 cells (n = 3). f Soft agar colony formation assay in MCF7 cells stably transduced with an empty vector (v) or a Sox9 expression vector (Sox9) and treated with 10^–9 M or 10^–8 M tamoxifen (n = 4). g Tumour volumes from mammary tumours from each cohort (sgRNA control (c) or CRISPR/Cas9n-mediated deletion of Sox9 (CRISPR Sox9) in MCF7TamR cells) collected 18 weeks after injections into mammary fat pad four in NSG female mice in the presence of an exogenous slow oestrogen supplement and a tamoxifen pellet (n = 4–6 tumours/group). 100 cells, p = 0.0028; 1000 cells, p = 0.0423. Mann–Whitney test was used. Statistic test: t-test (a, b, c) and one-way Anova (e, f). Error bars represent standard deviation (SD). P, p-value: *p < 0.05, **p < 0.001

Fig. 8

A Sox2–Sox9 axis regulates Wnt activity in breast cancer cells. AXIN2 a and Fzd4 b mRNA expression levels in MCF7TamR and MDA-MB-231 cells stably transduced with shcontrol (−) and shSox9 (+) lentiviral vectors (n = 3/4). c Relative change in the proportion of ALDEFLUOR⁺ cells in MCF7TamR and T47DTamR cells treated during 48 h with IWP-2 or IWR-1 Wnt inhibitors (n = 3). d Mammosphere formation in sgRNA binding sense strand only as control (c) and two different CRISPR/Cas9n clones with Sox9 deletion, derived from MCF7TamR cells in the absence (carrier containing CHAPS) or presence of Wnt3a (n = 4). e Sox9 protein expression levels in control (c) and Sox2 overexpressing (Sox2) MCF7 cells, with β-actin as control. f Sox2 and Sox9 mRNA expression levels in MCF7TamR, BT549 and MDA-MB-231 cells transiently transfected with sicontrol (−) or siSox9 (+) sequences (n = 3). g SOX9 and h AXIN2 mRNA expression levels in MDA-MB-231 shcontrol (shc) and shSox9 cells transiently transfected with sicontrol (−) and siSox2 (+) sequences (n = 3). i Chromatin Immunoprecipitation (ChIP) showing Sox2 binding to the human cyclin D1 and SOX9 promoters (left) and Sox9 binding to the human ALDH1A3 and SOX2 promoters (right). IgG control binding is set as 1 (n = 3). Error bars represent standard deviations (SD). *p < 0.05, **p < 0.001 compared to control. Statistic test: two-tail t-test (a, b, c, d, f, g) and Anova (i). j Model shows reciprocal regulation between Sox2 and Sox9 leading to activation of ALDH1A3 in breast cancer cells. Dashed arrow shows regulation of Wnt target genes by Sox2 [12], which may involve Sox9 (this report). ER negatively regulates both Sox2 and Sox9. The Sox2–Sox9 axis contributes to increased tamoxifen resistance