MRTF-A controls myofibroblastic differentiation of human multipotent stromal cells and their tumour-supporting function in xenograft models

Abstract

Tumour growth and metastatic colonization is strongly influenced by the tumour stroma, including cancer-associated fibroblasts (CAF). Multipotent mesenchymal stromal cells (MSC) are a possible source of CAF following myofibroblastic differentiation, and we have previously shown that MSC support tumour growth. Triggered by tumour cell-derived factors like transforming growth factor β1 (TGF-β1), myofibroblastic MSC differentiation is associated with the increased expression of markers including alpha smooth muscle actin (α-SMA). Here we show that myocardin-related transcription factor A (MRTF-A) plays an important role in myofibroblastic differentiation of primary human MSC in vitro and their tumour-supporting function in vivo. Recombinant TGF-β1 or tumour cell conditioned medium (TCM) elevated α-SMA, calponin 1 and collagen 1 A1 (COL1A1) amount on mRNA and protein level in MSC. This correlated with increased MRTF-A activity during MSC differentiation. MRTF-A knockdown by siRNA or shRNA impaired TGF-β1 and TCM induction of α-SMA and calponin 1, but not of COL1A1. Mixed xenograft experiments using HCT8 colorectal carcinoma cells and primary MSC of different donors revealed a significant reduction in tumour weight and volume upon MRTF-A knockdown in MSC. Our study suggests that MRTF-A is involved in the functional differentiation of MSC towards a tumour-promoting CAF phenotype in vivo.

Figure 1

TGF-β1 and TCM induce expression of myofibroblastic markers in MSC. Bone-marrow derived MSC were isolated from the iliac crest of individual patients and treated with TGF-β1 (TGF, 10 ng/ml) (a) or tumour cell conditioned medium (TCM) (c) for 24 h. Relative mRNA of calponin 1, α-SMA and COL1A1 was quantified by real-time RT-PCR and normalized to ALAS and GAPDH. Western blots of α-SMA and calponin 1 in MSC after 48 h of TGF-β1 (TGF) (b) or TCM (d) treatment. Relative protein amounts were normalized to the α-tubulin signal. All data were normalized to the value for the starved control (Ctrl.) which was set to 1. (e) Relative collagen deposition determined by Sirius Red staining normalized to cell accumulation following treatment with TGF-β1 for 96 h. Error bars correspond to SD (n = 3). Asterisks indicate significant differences *p ≤ 0.05 according to an unpaired Student’s t test.

Figure 2

TGF-β1 and TCM cause morphological and cytoskeletal changes in MSC. (a) Phase contrast images of the morphology of untreated proliferating MSC (cycling), quiescent MSC (starved) and MSC differentiated by TGF-β1 or TCM for 48 h. (b) Immunofluorescence images of α-SMA in MSC. Cycling, starved and treated (TGF, TCM) MSC were stained with anti-α-SMA and DAPI after 48 h of treatment. 20 x magnification, scale bars 20 µm.

Figure 3

Tumour cell conditioned medium induces TGF-β1-like signalling in MSC. (a) TGF-β1 amounts in medium conditioned by HCT8 colorectal carcinoma cells for 72 h (TCM), and identically treated control medium (Ctrl.). The + and – signs below the bar chart indicate whether samples were acid-treated for the release of active TGF-β1. Human recombinant TGF-β1 was used for calibration. n.d., not detectable. Error bars correspond to SEM (n = 4). (b) Western Blot of phosphorylated SMAD2 (pSMAD2) and total SMAD2 (tSMAD2). Cells were treated with TCM or TGF-β1 (TGF) for 1 h, respectively. For quantification, pSMAD2 was normalized to total SMAD2 signal. (c) Western Blot of SMAD2 phosphorylation upon TGF-β1 and TCM stimulation after pre-treatment with 200 nM Repsox for 1 h. (d) Calponin 1 induction in MSC by 48 h of TGF or TCM stimulation in the presence of RepSox. Quantification normalized to tubulin. Error bars correspond to SD (n = 3). Asterisks indicate significant differences (*p ≤ 0.05) compared to the control sample (Ctrl) according to an unpaired Student’s t test.

Figure 4

Role of MRTF-A activity during myofibroblastic MSC differentiation. (a) MSC stably infected with lentiviral MRTF-SRF dependent luciferase reporter constructs were analysed during MSC differentiation induced by TCM or TGF-β1 (TGF) for the indicated time points (7 h, 24 h, 48 h). Relative luciferase activities normalized to constitutive Crimson Red expression are displayed. Error bars correspond to SD (n = 5). (b) Intracellular localization of MRTF-A. Cells were immunostained with anti-MRTF-A antibodies and DNA counterstained with DAPI following 48 h of treatment with TGF-β1 (TGF) or TCM. 20 x magnification, scale bars 20 µm. (c) Quantification of MRTF-A displaying the change of nuclear to cytoplasmatic MRTF-A signal in 30 cells (each) from three independent MSC donors. Asterisks indicate significant differences (*p ≤ 0.05) according to an unpaired Student’s t test.

Figure 5

Impaired MSC differentiation by MRTF-A knockdown. (a) Expression of MRTF-A and myofibroblastic marker genes upon TGF-β1 treatment (24 h) following transient knockdown by MRTF-A siRNA (siMRTF-A) or control siRNA (siCtrl.). The relative mRNA amounts were determined by quantitative RT-PCR, normalized to the starved siCtrl. Error bars correspond to SEM (n = 4). Asterisks indicate significant differences (*p ≤ 0.05) according to an unpaired Student’s t test. (b) Western blot of MRTF-A, α-SMA and calponin 1 in lentivirally transduced MRTF-A knockdown MSC of three individual donors. Control infected cells (shCtrl.) as well as MRTF-A knockdown cells obtained by three independent shRNA constructs (#41, #77, #78) were treated with TCM or TGF-β1 (TGF) for 48 h. The relative protein amounts were normalized to α-tubulin signals. For quantification, data were normalized to the value for the starved shCtrl., which was set to 1. Shown are averages of 3 individual donor MSC. Error bars correspond to SD (n = 3). Asterisks indicate significant differences (*p ≤ 0.05) according to a one-way ANOVA with Dunnett’s multi comparison test (post-hoc).

Figure 6

MRTF-A knockdown partially impairs tumour-promoting effects of MSC on HCT8 xenograft growth in mice. 3 × 10⁶ HCT8 cells were coinjected with or without 7.5 × 10⁵ MSC s.c. in athymic nude mice. (a) Tumour volumes were determined using calliper measurement at day 7 (d7). (b) Mice were killed and the tumours were extracted and weighted after 25 days (d25). One-way ANOVA with Tukey’s multiple comparison test (post-hoc). (c) In vivo acquisition of multispectral images of DsRed-expressing tumours in athymic nude mice on d25. Pictures were taken via 2.2 CRi Maestro in vivo fluorescence imaging system (CRi, Woburn, MA, USA) with Maestro software (2.22). Exemplary grayscale images were overlayed with respective fluorescence images (intensity weighted pseudocolor mode, scale bar displayed beneath).