LIX1 regulates YAP activity and controls gastrointestinal cancer cell plasticity

Abstract Gastrointestinal stromal tumours (GISTs), the most common mesenchymal neoplasm of the gastrointestinal tract, result from deregulated proliferation of transformed KIT‐positive interstitial cells of Cajal that share mesenchymal progenitors with smooth muscle cells. Despite the identification of selective KIT inhibitors, primary resistance and relapse remain a major concern. Moreover, most patients develop resistance partly through reactivation of KIT and its downstream signalling pathways. We previously identified the Limb Expression 1 (LIX1) gene as a unique marker of digestive mesenchyme immaturity. We also demonstrated that LIX1 regulates mesenchymal progenitor proliferation and differentiation by controlling the Hippo effector YAP1, which is constitutively activated in many sarcomas. Therefore, we wanted to determine LIX1 role in GIST development. We found that LIX1 is strongly up‐regulated in GIST samples and this is associated with unfavourable prognosis. Moreover, LIX1 controls GIST cell proliferation in vitro and in vivo. Upon LIX1 inactivation in GIST cells, YAP1/TAZ activity is reduced, KIT (the GIST signature) is down‐regulated, and cells acquire smooth muscle lineage features. Our data highlight LIX1 role in digestive mesenchyme‐derived cell‐fate decisions and identify this novel regulator as a target for drug design for GIST treatment by influencing its differentiation status.

the stomach (50%-60%) and small intestine (30%-35%). More than 80% of GISTs harbour KIT gain-of-function mutations that result in the constitutive activation of the KIT receptor and downstream signalling pathways, leading to spontaneous proliferation and uncontrolled tumour growth. 7 Low-risk tumours can be managed by surgery alone; however, approximately 50% of GISTs relapse or metastasize after surgical resection. Imatinib mesylate, a small-molecular tyrosine kinase inhibitor against constitutively activated KIT, is efficient in adult patients with GIST. 8 However, resistance to imatinib mesylate is increasing and complete remission is rare, highlighting the necessity to improve our molecular understanding of GIST pathophysiology. 5 Cell dedifferentiation is a central mechanism in the initiation of neoplastic transformation and therapeutic resistance. [9][10][11] It involves loss of lineage-specific gene expression and regression from a specialized tissue to a more primitive state of development through expression of genes that govern embryonic cell-fate specification.
Accordingly, developmental biology studies are very useful for the identification of new tumour markers and therapeutic targets. 12 In this context, we previously performed a screen to identify genes with high expression at the earliest stages of stomach development and found that Limb Expression 1 (LIX1) is a novel and unique marker of stomach mesenchymal progenitors. Moreover, we demonstrated that LIX1 stimulates the expression and activity of the Hippo effector YAP1 and that both LIX1 and YAP1 are key regulators of stomach mesenchymal progenitor development. 13,14 Although LIX1 transcripts have been detected in human GIST cell lines, 15 LIX1 role in GIST is unknown.
Here, we found that LIX1 is strongly up-regulated in GIST specimens and that its expression is associated with poor prognosis. LIX1 down-regulation reprogrammes KIT-positive GIST cells towards the SMC lineage, thereby limiting their tumorigenic and malignant potential. Therefore, our study reveals LIX1 key role in GIST pathophysiology as a rheostat for the control of cell identity.

| Cell culture and reagents
The GIST-T1 cell line was from Cosmo Bio. It was established from a metastatic human GIST sample with a heterozygous deletion of 57 bases in exon 11 of KIT. 16 Human gastric SMCs, provided by Innoprot Innovative, were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% foetal bovine serum (FBS) and 1% penicillin/streptomycin. GIST-T1 cells were resuspended in Accutase™ solution (Sigma-Aldrich) before electroporation of different constructs using the Neon Transfection System (Life Technologies), according to the manufacturer's instructions.
GIST-T1 stable cell lines were generated by selection in 500 ng/mL puromycin. All cell lines were routinely tested for the absence of mycoplasma contamination (VenorGeM OneStep Test, BioValley). Verteporfin (Sellekchem) was added to GIST-T1 cells at a final concentration of 2 μmol/L.

| Human GIST tissue microarrays (TMA) and GIST data set
The SuperBiochips GIST TMA (#DAA2) was purchased from Super BioChips laboratories. It contained 40 formalin-fixed, paraffinembedded (FFPE) human GIST specimens and nine matched normal gut tissue specimens. The clinicopathological features of these patients are in Table S1. The second GIST TMA (#A225, BIOCAT, GmbH) contained 37 FFPE human GIST specimens and four non-neoplastic specimens. The clinicopathological features are in Table S2. For immunohistochemistry, TMAs were rehydrated through Histoclear (Fisher Scientific) and graded alcohol solutions, and then heated at 90°C in 0.01 mol/L citrate buffer (pH 6.0) for epitope unmasking. Spots were evaluated by three examiners blinded to the clinicopathological information. Immunoreactivity was considered positive when signal was above the background signal in the negative control. Staining intensity was scored as negative (−), intermediate (+) or high (++). A clinically annotated gene expression data set of localized, untreated GISTs (n = 60), quantified by microarray, was also used. 17

| Cell proliferation and wound healing assays
BrdU incorporation was performed using the Cell Proliferation

| Reverse transcription and quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted from cell cultures with the HighPure RNA Isolation kit (Roche). Reverse transcription was performed using the Verso cDNA synthesis kit (Thermo Scientific) and qPCR using the LightCycler technology (Roche Diagnostics). PCR primers (Table S3) were designed using the LightCycler Probe Design 2.0 software.
Expression levels were determined with the LightCycler analysis software (version 3.5) relative to standard curves. Data are the mean level of gene expression relative to the expression of the reference genes GAPDH and RPLPO calculated using the 2 −ΔΔCT method.  (Table S4). Nuclei were labelled with Hoechst (Invitrogen).

| Statistical analysis
Data were analysed with the two-tailed or when appropriate, onetailed Mann-Whitney test using the GraphPad Prism 6 software.

| LIX1 expression is a negative prognostic factor in GIST
Gastrointestinal stromal tumours result from the deregulated proliferation of transformed KIT-positive ICCs that share mesenchymal  Figure 1D). LIX1 expression was highest in patients with relapsed tumour (P = .015) ( Figure 1E). These findings identify LIX1 expression as a novel prognostic factor in GIST.

| LIX1 down-regulation decreases YAP1/ TAZ and KIT levels in GIST cells
As increased expression of LIX1 is associated with unfavourable prognosis in GISTs, we determined whether LIX1 regulates the growth of  Figure 2B). Then, to assess the effect of LIX1 silencing on GIST cell migration we performed wound healing assays. Scratch wound closure was slower in GIST-T1-ShLIX1#1 and GIST-T1-ShLIX1#2 cells than in GIST-T1-Scrambled cells ( Figure 2C).
Overall, these results suggest that LIX1 promotes GIST cell proliferation and migration in vitro.
In many GIST specimens (~85%), KIT harbours gain-of-function mutations that cause ligand-independent auto-activation of the receptor. 30 although KIT transcript level was not changed ( Figure S1).
We next investigated the mechanisms by which LIX1 regulates KIT. We previously reported that LIX1 regulates mesenchymal progenitor proliferation and differentiation through the Hippo mediator YAP1, 13 which is also involved in the control of GIST cell proliferation. 35 Therefore, we investigated the effect of LIX1 silencing on the two Hippo effectors YAP1 and TAZ. LIX1 down-regulation resulted in a marked decrease of YAP1/TAZ expression ( Figure 2D) and function, as indicated by the lower levels of CTGF and CYR61, their transcriptional targets ( Figure 2E). Furthermore, reducing YAP1 or TAZ expression (by SiRNA) ( Figure 2F) and activity (using verteporfin, an inhibitor of YAP1-TEAD/TAZ-TEAD interactions) ( Figure 2F) led to a marked decrease of KIT expression.

| LIX1 down-regulation induces a SMC-specific transcriptional programme in GIST cells
As LIX1 silencing was associated with loss of KIT expression, we

| LIX1 silencing reduces GIST aggressiveness in vivo
Our previous results demonstrated that in response to LIX1 silencing, GIST cells lose KIT expression, a signature of GIST cell identity, and acquire smooth muscle features, a quality associated with reduced proliferative and invasive capacities. Therefore, we assessed the effect of LIX1 silencing in vivo using the CAM assay, an established in vivo tumour model. 37 After graft of GIST-T1-Scrambled, GIST-T1-ShLIX1#1 and GIST-T1-ShLIX1#2 cells in the CAM of E7 chicken embryos, we allowed tumour growth until E12 ( Figure 4A) and then dissected and analysed them ( Figure 4B). The graft dry weight was lower in GIST-T1-ShLIX1 than GIST-T1-Scrambled cell grafts ( Figure 4B). Moreover, immunohistochemical analysis of the tumour tissue showed that LIX1 silencing significantly decreased cell proliferation (KI67) ( Figure 4C and Figure S4). Conversely, CALPONIN expression was significantly increased and KIT expression reduced in GIST-T1-ShLIX1 compared with GIST-T1-Scrambled cell grafts ( Figure 4D). Collectively, our data indicate that LIX1 silencing promotes a phenotypic modulation of KIT-positive GIST cells towards the SMC lineage and subsequently reduces GIST malignant phenotype.  40 Constitutive activation of YAP1/TAZ has been observed in many human tumours. 41 We found that reducing YAP1/TAZ protein level or activity results in a marked decrease of KIT expression. Although YAP1 regulates the proliferation of GIST cells, 35 to our knowledge, our study is the first to reveal a control of YAP1/TAZ in KIT-mediated GIST development. We found that LIX1

| D ISCUSS I ON
silencing results in a marked decrease of YAP1/TAZ levels and function, and demonstrated that reducing YAP1 or TAZ expression or activity leads to a marked decrease of KIT expression. It is tempting to F I G U R E 4 LIX1 silencing reduces GIST aggressive phenotype in vivo. A, Schematic representation of the approach. GIST-T1-Scrambled, GIST-T1-ShLIX1#1 or -ShLIX1#2 cells were grafted in the ChorioAllantoic Membrane (CAM) of E7 chicken embryos and grafts were allowed to grow for 5 d when individual grafts were removed and analysed. B, Representative images (left panels) and dry weight (right panel) of GIST-T1-Scrambled and GIST-T1-ShLIX1 cell grafts at E12. Scale bars, 2 mm. Values are the mean ± SEM of GIST-T1-Scrambled (n = 11), GIST-T1-ShLIX1#1 (n = 8) and GIST-T1-ShLIX1#2 cell grafts (n = 7). *P < .05 and **P < .01 (one-tailed Mann-Whitney test). C, KI-67 immunostaining of GIST-T1-Scrambled and GIST-T1-ShLIX1 graft sections. Scale bars, 500 μm. Enlargement, 100 μm. D, KIT and CALPONIN expression in GIST-T1-Scrambled and GIST-T1-ShLIX1 grafts. Scale bars, 500 μm F I G U R E 5 Model illustrating how LIX1 controls GIST malignant phenotype hypothesize that LIX1 regulates KIT protein level upstream of YAP1/ TAZ ( Figure 5). Additional studies are now required to precisely determine how LIX1 regulates YAP1/TAZ levels. The human LIX1 gene encodes a 282-amino acid protein that includes a double-stranded RNA-binding domain. This suggests that LIX1 could be involved in mRNA or micro-RNA processing. 39 Accordingly, miR-506 and miR-375 regulate YAP1 expression. 42,43 We found that in response to LIX1 silencing, GIST cells lose KIT expression and subsequently acquire features of differentiated SMCs, a quality associated with reduced proliferative and invasive capacities. These findings provide the proof-of-concept that differentiation therapy could be a potential treatment strategy for GISTs.
In summary, our data demonstrate that LIX1 silencing in GIST cells results in their reprogramming to SMCs, thereby limiting their aggressive potential. These findings highlight LIX1 potential as a drug target in GISTs in the framework of a differentiation therapy strategy for this aggressive digestive mesenchymal malignancy.

ACK N OWLED G EM ENTS
This work was supported by Association contre les Myopathies