TGF-β2 dictates disseminated tumour cell fate in target organs through TGF-β-RIII and p38α/β signalling

Abstract

In patients, non-proliferative disseminated tumour cells (DTCs) can persist in the bone marrow (BM) while other organs (such as lung) present growing metastasis. This suggested that the BM might be a metastasis 'restrictive soil' by encoding dormancy-inducing cues in DTCs. Here we show in a head and neck squamous cell carcinoma (HNSCC) model that strong and specific transforming growth factor-β2 (TGF-β2) signalling in the BM activates the MAPK p38α/β, inducing an (ERK/p38)(low) signalling ratio. This results in induction of DEC2/SHARP1 and p27, downregulation of cyclin-dependent kinase 4 (CDK4) and dormancy of malignant DTCs. TGF-β2-induced dormancy required TGF-β receptor-I (TGF-β-RI), TGF-β-RIII and SMAD1/5 activation to induce p27. In lungs, a metastasis 'permissive soil' with low TGF-β2 levels, DTC dormancy was short-lived and followed by metastatic growth. Importantly, systemic inhibition of TGF-β-RI or p38α/β activities awakened dormant DTCs, fuelling multi-organ metastasis. Our work reveals a 'seed and soil' mechanism where TGF-β2 and TGF-β-RIII signalling through p38α/β regulates DTC dormancy and defines restrictive (BM) and permissive (lung) microenvironments for HNSCC metastasis.

Figure 1. Growth behaviour of BM- and lung-derived DTCs

(a) Phase contrast and GFP channel images of a single DTC in a BM flush (upper left panel) and a lung metastasis (upper right panel) 3 wks after primary tumour surgery. Scale bar: 40 µm (BM), 400 µm (lung). S= stroma. Lower left panel: fluorescence images of lungs stroma and a micro-metastasis in a freshly resected lung. Lower right panel: laser scanning confocal image of a DTC cluster in lung. Scale bar: 160 µm (left), 20 µm (right). n=16 (BM), 35 (LU) mice. (b) Number of HEp3-GFP DTCs found in lungs and BM after primary tumour surgery (n=5 mice / time / condition). (c) Tumour volume of primary tumour- (PT-HEp3), lung- (Lu-HEp3) and BM-DTC-derived cell lines (BM-HEp3) from mice, after 1 week on CAM. BM-T1= tumorigenic in vivo, BM-D1=dormantin vivo. (n=5 (Lu6), 5 (Lu7), 4 (PT-2), 7 (PT-4), 5 (BM-T1), 5 (BM-D1) tumours). (d) Phospho-Histone3 (P-H3) (left column) and p27 (right column) staining in BM-D1 (upper panels) tumour nodules and Lu-7 (lower panels) tumours grown on CAMs. Scale bar: 40 µm. Arrows = positive cells. Lower panel: quantification of P-H3 and p27 positive cells (n=100 cells assessed/section. 15 sections assessed from 3 different tumour/ condition). FOV = field of view. (e) Weight of BM-D1 tumour nodules on CAMs. BM-D1 cells were inoculated on CAM (5×10⁵ cells/animal) and transplanted into a new CAM every week (n=3 (week1), 5 (week2), 5 (week 3), 5 (week 4), 3 (week 5), 4 (week 6), 2 (week 7), 3 (week 8) tumours/ time point). (f) Number of tumour cells / CAM nodule produced by primary tumour- (PT-HEp3) and BM-DTC cell lines (BM-HEp3) derived from avian after 1 week on CAM (n=4 (BM-D2), 3 (BM-D3), 4 (BM-D4), 10 (PT), 5 (BM-D5), 5 (BM-T2), 4 (BM-T3), 2 (BM-D6), 3 (BM-D7), 3 (BM-D8) tumours). See supplementary table S4 for statistic source data. Data in a, b and d, represent mean ± s.e.m. In b and d, *p<0.05, **p<0.01 and +++p<0.001 by Mann Whitney test. In c and f *p<0.05, **p<0.01 and ***p<0.001 by One-way ANOVA-Bonferroni's multiple comparison test and in e **p<0.01 by Unpaired t test.

Figure 2. Signalling pathways and genes regulating dormancy of BM-HEp3 Cells

(a–b) Lysates from the indicated DTC-derived cell lines from murine (a) and avian microenvironments (b) were probed by immunoblot (IB) for the indicated antigens. Numbers on top of blots = [ERK/p38] ratio quantification (see Methods). (c) Graph showing percentage of cell lines with high or low ERK/p38 ratio and their phenotypein vivo. (d) Left panel: Heat map showing DEC2 and p53 mRNA expression in the indicated cell lines after 24h in culture. Scale = log2 fold change and upregulation of mRNAs (red) was significant (p<0.05). Middle and right panels: IB for p53 and GAPDH in the indicated cell lines after 24h in culture. (e) BM-D1 cells transfected with either scrambled (scr), p53, p38α or DEC2 siRNAs or Lu-HEp3 cells transfected with empty vector (Ctrl) or a constitutively active p38α construct (p38α-CA-HA) or a DEC2-V5 construct were inoculated in vivo (5×10⁵ BM-D1 cells / CAM and 2.5×10⁵ lu-7 cells / CAM) for 5 days. The graph represents the number of cells per tumour nodule (n=18 (scr), 12 (p53), 6 (p38), 7 (dec2), 20 (Ctrl), 7 (p38CA-HA), 5 (DEC2-V5) tumours per condition). *p<0.05 and **p<0.01 by one-way ANOVA-Bonferroni's multiple comparison test. (f) Growth of BM-D1 cells on CAM after RNAi to DEC2 for two weeks in vivo. BM-D1 cells transfected with scrambled (scr) or DEC2 siRNA were inoculate in CAMs (5×10⁵ BM-D1 cells / CAM). One week after the tumour nodules were mince, cells/tumour nodule were quantified, transfected again with either scrambled or DEC2 siRNAs and reinoculated again in vivo. Graph= mean number of cells / tumour nodule ± s.e.m. (n=4-Si scr week1, 5 Si DEC2 week1, 3-Si scr week2, 5 Si DEC2 week2 tumours per condition). See supplementary table S4 for statistic source data. *p<0.05 and **p<0.01 by Mann Whitney Test. Right panel: p27 and CDK4 expression in BM-D1 cell lines after DEC2 inhibition with siRNAs. Scr=scrambled. (g) p27, p21 and CDK4 immunoblot in T-HEp3 cells transfected with either a GFP or a DEC2-V5 construct for 24h.

Figure 3. Systemic p38α/β inhibition affects HNSCC DTC behaviour in growth restrictive microenvironments

(a) Representative images of Cytokeratin 8/18 (CK), P-H3 (left) and p27 (right) staining in solitary BM DTCs in BM cytospins. Scale bar: 10 µm. Right Graph: percentage of p27 and P-H3 positive DTCs / BM flush, Numbers on top = n for scored DTCs. 6 BM flushes/ 3 independent experiments were assessed (b) Quantification of BM-DTCs in mice treated with DMSO (control) or SB203580 (SB) (10 mg/kg every 48 hrs) for 4 weeks. Left Graph: Alu qPCR quantification of BM DTCs shows the human Alu genomic signal normalized to genomic mouse GAPDH (n=6 (control), 15 (SB) DNA samples were assessed over 3 independent experiments) (see methods for details). Right Graph: quantification of CK positive cells in mice BM cytospins (n=6 BM samples / condition were assessed over 2 independent experiments). *p<0.05 by wilcoxon signed rank test. (c–d) Live imaging of HEp3-GFP DTCs, micro and macrometastasis in spleen (c) and liver (d) of control (scale bar: 80µm) and SB203580 treated mice (Spleen, scale bar: 400µm and liver, scale bar: 80µm). Numbers= GFP positive cells per organ. (e) Quantification of BM-DTCs in chicken embryos’ BM flushes 5 days after inoculation of T-HEp3-GFP cells transfected with either scrambled (si scr) or p38α (si p38α) siRNAs. (n=7 (si Scr), 6 (sip38) BM samples were assessed over 2 independent experiments). Lower panels: representative fluorescence intravital images of BM DTCs in chicken embryo BM flushes. Scale bar: 160µm. (f) Graph: percentage of p27 positive cells / BM flush (n=3 BM flushes were assessed over 2 independent experiments). Lower Panels: representative images of CK (right column) and p27 (left column) staining in solitary BM DTCs in chicken embryos BMs 5 days after inoculation of T-HEp3-GFP cells transfected with scrambled (scr) (upper row) or p38α (lower row) siRNA. Scale bar=40µM. Data in a, b and f, represent mean ± s.e.m. In b (left graph), e and f, *p<0.05, **p<0.01 by Mann Whitney test.

Figure 4. Systemic p38α/β inhibition interrupts DTC dormancy in the lungs

(a–b) Representative images of P-H3 and Vimentin (a) or Caspase 3 (C–C3) and Vimentin (b) staining in lung DTCs (upper panels, scale bar: 20 µm) and lung micro-metastases (micro-MET – a only) 4 weeks after surgery (scale bar: 40 µm (Micro-MET (a), inset (b)). Arrows: marker positive cells. Graphs: % of P-H3 (a) or C-C3 (b) positive cells per lung section, mean ± s.e.m. of three different lungs, 5 sections per lung (n=45 (DTCs), n=416 (Met) DTCs per lungs section (a), n=36 (DTCs) (b)). *p<0.05 (a) and non significant (n.s.) (b) by Mann Whitney Test. (c) Representative lung images containing HEp3-GFP DTCs or metastasis (scale bars: 80 µm, left column and 400 µm right column) after surgery and control or SB203580 (SB) (10 mg/kg every 48 hrs) treatment for 2 and 4 weeks. Lower Graphs: quantification of HEp3-GFP cells in whole lung suspensions. (n=5 (Left graph) and 4 (right graph) lungs per condition). See Supplementary table S4 for the statistic source data of the right graph. *p<0.05 by Mann Whitney Test. (d) Number of lung F3II breast cancer macro-metastasis / lung in mice treated with DMSO (control) or SB203580 (SB). Top numbers: prevalence (%) of mice with more than 3 metastatic nodules per lung. n=25 mice, *p=0.046 by Mann Whitney Test.

Figure 5. Role of TGFβ2 in dormant HEp3 DTCs

(a–b) Q-PCR analysis of TGFβ2 mRNA levels in the indicated cells cultured in serum-free media for 24h (a) in BM and lungs mice organs (b) (n=6 RNA samples per condition were assessed over 3 independent experiments). (c) Q-PCR for TGFβ2 mRNA in T-HEp3 cells treated with serum-free media (SF) or conditioned media (CM) from lung (Lu CM) or BM (BM CM) (24h). (n=6 RNA samples were assessed over 3 independent experiments). (d) Growth after 5 days of T-HEp3 cells on CAMs (2*10⁵cell/CAM) pre-treated in culture for 24h and then on the CAM every day with SF, BM CM or Lu CM (n=11 (SF), 3 (lu CM), 8 (BM CM) tumour nodules per condition). (e) Immunohistochemistry (IHC) for P-H3 (upper row) and p27 (lower row) in T-HEp3 tumour nodule sections after treatment with SF (left column), Lu CM (middle column) or BM CM (right column) for 5 days in vivo, scale bar: 40 µm. Arrows: marker-positive cells. Lower graphs: Quantification of P-H3 (left) and p27 (right) positive cells (n=100 cells assessed/section. 15 sections assessed from 3 different tumour/ condition). (f) Detection of the indicated antigens in lysates of T-HEp3 cells treated with, SF, Lu CM or BM CM control (IP-IgG) or TGFβ2 immunodepleted (IP- TGFβ2) for 24h. Top numbers: [ERK/p38] ratio quantification. (g) Q-PCR for p53 and DEC2 mRNA in T-HEp3 cells treated as in c (n=6 RNA samples were assessed over 3 independent experiments). (h) Top panel: IB against TGFβ2 in the BM CM after IgG or anti-TGFβ2 immunodepletion. Lower panel: Tumour growth of T-HEp3 cells treated with full or TGFβ2-immunodepleted BM CM or with Lu CM supplemented with PBS (CM) or with 2 ng/ml TGFβ2 (n=22 (SF), 15 (IP-IgG), 8 (IP-TGFb2), 7 (lu CM), 5 (Lu-Cm TGFB2) tumours per condition). (i) Effect of TGFβ1 (2 ng/ml) on T-HEp3 and dormant D-HEp3 and BM-D1 tumour growth on CAMs for 5 days in vivo (n=6 (DHEp3-TGFb1), 9 (DHEp3+TGFb1), 10 (BM-D1-TGFb1), 10 (BM-D1+TGFb1), 13 (THEp3-TGFb1), 11 tumours per condition). Data in a, b, c, e and g, represent mean ± s.e.m. In a, b, c, e and g, *p<0.05, **p<0.01 by Mann Whitney test. In d, h and i *p<0.05, **p<0.01 by One-way ANOVA-Bonferroni's multiple comparison test.

Figure 6. Effect of TGFβ signalling on dormancy and self-renewal markers

(a) IB for the indicated antigens on PT-HEp3 (PT), BM-D1 and BM-T1 cell lysates after TGFβ2 (2 ng/ml) or TGFβRI inhibitor (LY-364947, 5 µM) treatment for 24 h in SF media. (b) IB for the indicated antigens on T-HEp3 cell lysates after treatment with TGFβ2 (2 ng/ml) for 24 h in SF media. (c–d) QPCR measured DEC2 (c) and p53 (d) mRNA levels in the indicated cells after treatment with 2 ng/ml TGFβ2 or LY-364947, 5 µM for 24 h in SF media (n=6 RNA samples were assessed over 3 independent experiments). (e) BM-D1 growth on CAMs for 4 days after TGFβ2 knockdown. BM-D1 cells were transfected with either scrambled (scr) or TGFβ2 siRNA and inoculated onto CAMs (5*10⁵ cells/CAM) for 4 days (n=20 (scr), 19 (TGFβ2) tumours per condition). (f) BM-D1 growth on CAMs for 4 days after TGFβRI inhibition with LY364947 (5µM). BM-D1 cells were pre-treated with TGFβRI inhibitor (LY-364947, 5 µM) for 24 h in SF media and then inoculated onto CAMs (5*10⁵ cells/CAM) for 5 days. Tumour nodules were treated with LY-364947 (5 µM) ever day in the CAMs (n=9 tumours per condition). Data in c and d represent mean ± s.e.m. In c, d, e and f *p<0.05, **p< 0.01, ***p<0.001 by Mann Whitney Test.

Figure 7. TGFβ2/TGFβ receptor III signalling regulates DTC dormancy

(a–b) IB for the indicated antigens on T-HEp3 cell lysates after treatment with either TGFβ1 or TGFβ2 (5 ng/ml) for 10, 30 min and 2 h (a) and 24 h (b) in SF media. (c–d) QPCR for DEC2 (c) and p27 (d) mRNA levels in T-HEp3 cells treated with either TGFβ1 or TGFβ2 (5 ng/ml) for 6 (d) and 24 h in SF media (n=6 RNA samples were assessed over 3 independent experiments) *p<0.05 by Mann Whitney Test. Error bars denote s.e.m. (e) IB for the indicated antigens on cell lysates of T-HEp3 treated with 5 ng/ml TGFβ2 for 24h in SF media with or without SB203580 (SB) (5 µM) (f) T-HEp3 tumour growth on CAMs for 4 days of T-HEp3 cells with TGFβR-III knockdown (siRNA TGFRβ3) treated with 5 ng/ml TGFβ2 (n=16 (scr, SF), 11 (scr, TGFb2), 15 (TGFbR3, SF), 17 (TGFbR3, TGFb2)), *p<0.05 by One-way ANOVA-Bonferroni's multiple comparison test. (g) IB for the indicated antigens on cell lysates of T-HEp3 treated with TGFβ2 (5 ng/ml) for 24 h in SF media and after TGFβR-III knockdown (siRNA TGFRβ3). (h)Tumour growth on CAMs for 4 days after treatment with TGFβ1 (2 ng/ml) in T-HEp3 cells after type III TGFβ receptor knockdown (siTGFβR3) (n=6 (scr, SF), n=7 (scr, TGFb1), n=6 (SiTGFBR3, SF), n=6 (SiTGFBR3, TGFb1)), p=0.3246 by One-way ANOVA-Bonferroni's multiple comparison test.

Figure 8. Effect of TGFβRI inhibition on DTC burden

(a) Prevalence of HEp3 DTCs (%) per mouse in control (n=20) and LY364947-treated (n=32) animals (2 independent experiments) (b) Alu qPCR quantification of HEp3 DTCs in BM of mice treated either with DMSO or with LY364947 10 mg/kg every 48 h for 2 weeks. Graph: mean ± s.e.m. of Alu amplification signal normalized to mice GAPDH (n=6 DNA samples from 6 different BM samples were assessed over 2 independent experiments), *p<0.05 by Mann Whitney Test. (c) Fluorescence images of HEp3-GFP DTCs in mouse lungs, 2 weeks after control (scale bar: 80 µm) and LY364947 treatment (scale bar: 120 µm). Lower graph: quantification of HEp3-GFP DTCs in lungs of control and LY364947 treated animals. n=4 mice per condition, *p<0.05 by Mann Whitney Test. (d) Scheme of the proposed mechanism for TGFβ2-induced dormancy. Binding of TGFβ2 to TGFβ receptor III (TGFβRIII) recruits TGFβ receptor II (TGFβRII) and TGFβ receptor I (TGFβRI) into the complex and activates TGFβ2 signalling which in turn activates Smad1/5 and induces p27. In addition, TGFβ2 also activates p38α in a TGFβRIII independent way. In response to TGFβ2, p38α activates SMAD2 and DEC2, which induces p27 and inhibits CDK4. All these signals integrate and contribute to TGFβ2 induction of quiescence.