LIX1 defines stomach mesenchymal progenitors
We previously screened for genes that demonstrated higher expression at the earliest stages of stomach development [8] and found LIX1 to be among them. Real-time quantitative PCR (RT-qPCR) analyses on stomach extracts confirmed the dynamic and transitory nature of LIX1 expression during stomach development (Additional file 1: Figure S1A). While high levels of LIX1 transcripts were detected at embryonic day 4 (E4), levels of LIX1 transcripts quickly decreased with the onset of SMC determination (as visualized through the expression of αSMA and SM22), to finally barely detectable levels at E7, when SMC differentiation occurred (as shown by the high level of CALPONIN and CALDESMON expression; Additional file 1: Figure S1A). In parallel, we monitored the levels of BARX1, a marker of stomach mesenchyme [19], as well as SRF and its co-activator MYOCARDIN, which control SMC differentiation [20, 21], and found that, while the onset of MYOCARDIN expression occurs at E5.5, the stage of SMC determination, SRF and BARX1 were detected throughout all examined stages. These results suggest that LIX1 is an early marker of stomach development. We further studied the precise expression pattern of LIX1 in the developing GI tract by in situ hybridization analysis (Additional file 1: Figure S1B). Strong LIX1 expression was detected at E4 throughout the stomach mesenchyme and levels quickly decreased from E5 onwards (Fig. 1a, b). LIX1 transcripts were mainly detected in the pylorus at E5 and in the most posterior part of the stomach at E6 (Fig. 1a, b). When comparing the dynamics of LIX1 expression in the developing stomach with the kinetics of αSMA, the early marker of SMC determination in adjacent stomach sections, we observed that their expression domains appeared mutually exclusive (Fig. 1b). While LIX1 expression was high in stomach mesenchymal progenitors, it progressively decreased with the onset of SMC determination, thus identifying LIX1 as a novel and unique stomach marker, restricted to mesenchymal progenitors (Fig. 1c).
LIX1 silencing impairs mesenchyme determination and decreases YAP1 activity
The complementarity between LIX1 and αSMA expression prompted us to investigate whether LIX1 is required for the process of stomach SMC determination. This was accomplished using the avian replication-competent retroviral (RCAS) transgenesis method that allows in vivo gain- or loss-of-function approaches of specific genes in the stomach mesenchyme (Additional file 2: Figure S2A) [6, 8, 19, 22]. We first performed LIX1 loss-of-function experiments using RCAS(A)-ShLIX1 (short-hairpin RNA directed against LIX1) retroviruses. When injected into the presumptive domain of the developing stomach, RCAS(A)-ShLIX1 retroviruses led to a specific decrease in endogenous LIX1 expression, demonstrated by in situ hybridization and RT-qPCR analyses (Fig. 2a, c). In situ hybridization analysis revealed a decrease in the expression of the SMC determination marker SM22 in E6.5 ShLIX1-expressing stomachs compared to controls (Fig. 2b) upon LIX1 silencing. This was confirmed by RT-qPCR analysis (Fig. 2c). In contrast, injection of unrelated RCAS(A)-ShRNA retroviruses, which do not target LIX1, had no effect on αSMA expression (Additional file 3: Figure S3A). Moreover, when RCAS(A)-ShLIX1 retroviruses were co-injected with RCAS(B)-hLIX1 retroviruses, which induce the expression of the human LIX1 protein insensitive to the chick-specific RCAS(A)-ShLIX1 retroviruses, normal expression of αSMA was restored, demonstrating the specificity of the ShLIX1 construct for LIX1 (Additional file 3: Figure S3B). Levels of BARX1 transcripts were comparable in ShLIX1-expressing stomachs compared to controls, indicating that the patterning of the stomach was unaffected by LIX1 silencing (Fig. 2c). We also observed a decrease in MYOCARDIN expression, while levels of SRF transcripts were not significantly affected in E6.5 ShLIX1-expressing stomachs compared to controls (Fig. 2c). LIX1 silencing induced a smaller determined-SMC territory, as demonstrated by in situ hybridization (Fig. 2b) and immunostaining analyses on ShLIX1-expressing stomach sections compared to controls (Fig. 2d; Additional file 4: Table S1). The diminished expression of SMC determination markers was associated with a 40 % decrease in the rate of cell proliferation in ShLIX1-expressing stomach sections compared to controls, as demonstrated by immunostaining analysis for phosphorylated histone 3-Ser10 (PH3; Fig. 2e), a standard marker of the G2/M transition [6]. These results are in line with a role for LIX1 in regulating cell proliferation, as previously shown in studies on cricket (Gryllus bimaculatus) and mouse that identified homologs of LIX1 as positive regulators of cell proliferation [10, 18]. Lowfat, the arthropod homolog of LIX1, has been characterized, through its interaction with the atypical cadherins fat and dachsous, as a component of the Hippo pathway [10, 12]. As the key downstream regulator of the Hippo pathway is the pro-proliferative gene YAP1, we next investigated whether LIX1 regulates the expression of YAP1 during this process. In situ hybridization and RT-qPCR analyses revealed that endogenous transcripts of YAP1 and its transcriptional targets CTGF and CYR61, known to stimulate cell proliferation [15, 23], are abundant during early development of the stomach (E4–5.5; Additional file 5: Figure S4A,B). At this stage, their expression is detectable in both the mesenchymal and epithelial layers of the stomach, as demonstrated by RT-qPCR analyses on layer-dissociated stomach extracts (Additional file 5: Figure S4C). RT-qPCR analysis showed a reduction in the level of YAP1 and its transcriptional targets CTGF and CYR61 in ShLIX1-expressing stomachs compared to controls (Fig. 2f). Although expression data were significant for CTGF, but not for CYR61, the results for both transcripts were consistent. We attribute the lack of significance for the second transcript to low statistical power rather than to absence of an effect. These results indicate that YAP1 activity was decreased in ShLIX1-expressing stomachs compared to controls. Moreover, LIX1 silencing also induced a decrease in the expression of the TEAD transcription factor TEAD1 (Fig. 2f). Taken together, our results show that, when LIX1 expression was silenced in the developing stomach, SMC determination was hindered. This was associated with a decrease in cell proliferation and a decrease in YAP1 transcript levels and YAP1 activity in the developing mesenchyme. Our finding highlights the requirement of LIX1 expression in the stomach mesenchymal progenitors to establish normal proliferation rates and allow proper SMC determination.
LIX1 misexpression expands the determined SMC domain and stimulates cell proliferation and YAP1 activity
We next induced a misexpression of LIX1 in the stomach mesenchyme using RCAS(B)-LIX1 retroviruses (Additional file 2: Figure S2A). This treatment did not drastically affect GI morphogenesis, as the morphology of LIX1-misexpressing stomachs resembled that of control embryos (Additional file 2: Figure S2B). We first observed a premature expression of SMC determination marker SM22 as early as E4.5 in LIX1-misexpressing stomachs, whereas SMC determination had not yet taken place in controls, suggesting that LIX1 misexpression facilitated SMC determination (Fig. 3a, white arrowhead). As a result, we observed at E6 that LIX1-misexpressing stomachs harboured an expanded determined SMC territory at the expense of the adjacent domains, mainly the intermuscular tendons and the submucosa. This was demonstrated both by whole-mount in situ hybridization, which showed a larger expression domain of determined SMC markers SM22 and BAPX1 [24] in LIX1-misexpressing stomachs compared to controls (Fig. 3b), and by αSMA immunostaining on sections, showing that sustained LIX1 expression led to a decrease in the size of the submucosa (Fig. 3c, compare white bars). Accordingly, analysis of the enteric nervous system network using in situ hybridization of SOX10 transcripts revealed that enteric nervous system precursors, which normally colonize the SMC domain specifically [8], had migrated into the adjacent tendon territory, further indicating an expanded SMC domain in LIX1-misexpressing stomachs compared to controls (Fig. 3b, white arrowhead). Further analysis by RT-qPCR confirmed that, compared to control stomachs, LIX1-misexpressing stomachs harboured higher levels of SMC determination marker αSMA and BARX1 transcripts at E6, whereas MYOCARDIN and SRF levels were not significantly affected (Fig. 3d). Taken together, our in vivo results indicate that LIX1 is not only necessary for correct SMC determination, but that it also acts in favour of the process. These changes are associated with an increase in the rate of cell proliferation, as demonstrated by immunostaining analysis for PH3, and consequently to an increase in mesenchymal cell density in E6 LIX1-misexpressing stomachs compared to controls (Fig. 3e). The rate of cell death, however, was comparable in both conditions, as demonstrated by immunostaining analysis of cleaved CASPASE-3 (Additional file 6: Figure S5). Taking into account the positive effect of LIX1 on SMC proliferation and our previous results demonstrating that LIX1 silencing impaired YAP1 expression and activity, we suspected that LIX1 overexpression would stimulate the expression of genes in the YAP1 pathway. In fact, RT-qPCR analysis indicated a significant increase in the expression of YAP1, CTGF and TEAD1, and a slight increase in the expression of CYR61 and TEAD4 in LIX1-misexpressing stomachs compared to controls (Fig. 3f).
The differences in YAP1 expression and activity observed in LIX1-misexpressing stomachs could be linked to the changes in the identity of the tissue associated with aberrant LIX1 expression, or could be due to a role of YAP1 as a key relay in the establishment of the LIX1 phenotype. We thus performed YAP1 gain-of-function experiments in the developing stomach using RCAS(B)-YAP1 retroviruses. We observed an expanded SM22-positive determined SMC domain in YAP1-misexpressing stomachs compared to control stomachs (Fig. 4a). RT-qPCR analysis indicated that YAP1 misexpression did not affect the endogenous expression of LIX1 (data not shown) and confirmed an increase in the transcript levels of the SMC determination markers αSMA and MYOCARDIN at E6 (Fig. 4b). Levels of BARX1, SRF, TEAD1 and TEAD4 were not significantly affected in YAP1-misexpressing stomachs compared to control stomachs (Fig. 4b, d). Moreover, changes in expression of SMC determination markers were associated with an increase in cell proliferation, as demonstrated by immunostaining analysis for PH3 (Fig. 4c). Our results thus demonstrate that LIX1 stimulates the endogenous level of YAP1 transcripts and YAP1 activity and that sustained YAP1 activity phenocopies LIX1 misexpression regarding stomach mesenchyme determination. Furthermore, when RCAS(A)-ShLIX1 retroviruses were co-injected with RCAS(B)-YAP1 retroviruses, the expression of LIX1 was not rescued (Fig. 4e). However, the restored YAP1 activity (monitored through the expression of CYR61 and CTGF transcripts) rescued the expression of αSMA (Fig. 4e). Altogether, these data demonstrate that YAP1 is a key relay in the establishment of the LIX1 phenotype.
Endogenous LIX1 expression is regulated by the FGF pathway during SMC determination
Collectively, our in vivo loss- and gain-of-function experiments demonstrate that LIX1 expression must be finely regulated in the stomach mesenchyme to control the pool of progenitors required for correct SMC determination, presumably through the regulation of YAP1 activity. It has been shown that aberrant activation of the FGF pathway has a negative impact on stomach SMC determination [8]. As we report that LIX1 silencing impaired SMC determination, we next investigated whether the expression of LIX1 was under the control of the FGF signalling pathway. To address this question, we activated the FGF signalling pathway by misexpressing FGF8 in the stomach mesenchyme using RCAS(A)-FGF8 retroviruses and confirmed by RT-qPCR analysis that mesenchyme determination was hindered, as demonstrated by lower levels of αSMA and SM22 transcripts in FGF8-misexpressing stomachs compared to controls (Fig. 5b). The upregulation of FGF activity was associated with a strong reduction in LIX1 transcript levels compared to control stomachs, which was monitored by in situ hybridization (Fig. 5a) and confirmed by RT-qPCR analysis (Fig. 5b), and this was associated with a decrease in the levels of YAP1 transcripts (Fig. 5b). These results suggest that sustained FGF activity during SMC determination phenocopies LIX1 loss-of-function. Conversely, when using RCAS(B)-sFGFR2b retroviruses, which produce a secreted form of FGFR2b [8, 25], we found that inhibition of FGF pathway activity induced an increase in LIX1 levels at E6.5 compared to control stomachs (Fig. 5c, white arrows). Taken together, these results suggest that the FGF pathway regulates the endogenous expression of LIX1 and thereby maintains the proper levels necessary to ensure correct stomach mesenchyme determination.
Sustained LIX1 expression decreases YAP1 activity and hinders SMC differentiation
To further understand the role of LIX1 in the development of the stomach mesenchyme, we next analysed the consequences of LIX1 misexpression on SMC differentiation, the later step of SMC development. We found a reduction in the level of CALPONIN protein at E7 in LIX1-misexpressing stomachs, indicating that SMC differentiation was impaired (Fig. 6a). A strong reduction in CALPONIN transcript levels was also observed (Fig. 6b). Additionally, we observed a decrease in the expression of MYOCARDIN, while levels of BARX1 and SRF transcripts were not significantly affected (Fig. 6b). The decrease in SMC differentiation markers in LIX1-misexpressing stomachs was also observed later in development at E8.5, suggesting that the reduced level of differentiation markers did not simply reflect a delay in stomach SMC development (Additional file 7: Figure S6). We found that YAP1 misexpression also hindered CALPONIN expression, as demonstrated by immunostaining on stomach sections (Additional file 8: Figure S7A). This suggests that, while LIX1 misexpression and YAP1 stimulation had a positive impact on SMC determination, they hindered SMC differentiation. Surprisingly, we found that, when LIX1 expression was sustained in the developing stomach, the downregulation in the expression of SMC differentiation markers was associated with a lower rate of proliferation (Fig. 6c). Indeed, mesenchymal cell density was comparable in LIX1-misexpressing stomach compared to controls (Fig. 6c). It has been shown that the Hippo pathway acts as a sensor of cell density [16, 17], thus mediating the relationship between cell proliferation and cell contact inhibition of proliferation. As cell density becomes higher, the Hippo pathway is activated, resulting in an inhibitory phosphorylation of YAP1 and thus a decrease in cell proliferation [26]. Interestingly, we observed a decrease in YAP1 activity in YAP1-misexpressing stomachs at E7 by western blot analysis monitored through an increase of the inactive phosphorylated form of YAP1 compared to controls (Additional file 8: Figure S7B). The decrease in YAP1 activity was confirmed by RT-qPCR analysis, which showed lower transcript levels of CTGF in YAP1-misexpressing stomachs at E7 compared to controls (Additional file 8: Figure S7C). These results indicate that, while YAP1 misexpression in the stomach stimulated YAP1 transcriptional activity at determination stages, a decrease in YAP1 activity was observed later on at differentiation stages. One possible explanation is that sustained LIX1 expression led to a decrease in YAP1 activity consecutive to cell contact inhibition of proliferation, as a consequence of the early stimulation of mesenchymal progenitor proliferation, and this could be inhibitory for SMC differentiation. In line with this hypothesis, western blot analysis revealed an increase of the inactive phosphorylated form of YAP1 compared to controls (Fig. 6d). The decrease in YAP1 activity in LIX1-misexpressing stomachs at E7 was further confirmed by RT-qPCR analysis, which showed lower transcript levels of YAP1 targets CYR61 and CTGF in LIX1-misexpressing stomachs compared to controls (Fig. 6e). Additionally, we observed a decrease in TEAD1 transcript levels in LIX1-misexpressing stomachs compared to controls (Fig. 6e). These data indicate that Hippo signalling was activated as a result of sustained LIX1 expression at E7. Altogether, our results demonstrate that LIX1 has an early role in the process of stomach SMC determination, through the regulation of YAP1-induced mesenchymal progenitor proliferation. However, as stomach development proceeds, sustained LIX1 expression has a negative impact on further SMC differentiation and this is associated with a decrease in YAP1 activity.
The ability of LIX1 to regulate cell proliferation is dependent on cell density
These results prompted us to investigate the role of LIX1 in regulating both proliferation and contact inhibition of proliferation in heterologous cell cultures. DF-1 chicken fibroblasts were infected with empty RCAS(A) (control) or RCAS(B)-LIX1 retroviruses and cultured for 5 days to ensure homogeneous expression. Then, cells were seeded at low density (Fig. 7a). As expected according to our in vivo results demonstrating a positive effect of LIX1 overexpression on the expression of YAP1 (Fig. 3f), after 1 day in culture (day 1) LIX1-expressing cells demonstrated a higher expression of YAP1 transcript (Fig. 7b) and protein levels (Fig. 7c) compared to control cells. This greater expression was associated with higher transcript levels of YAP1 target genes CTGF and CYR61 (Fig. 7b) and an increase in cell proliferation (Fig. 7d). Interestingly, when LIX1-expressing cells were treated with verteporfin, an inhibitor of the YAP-TEAD interaction that abrogates YAP activity but not its expression [27, 28], levels of CTGF and CYR61 transcripts (Fig. 7e) and rates of proliferation (Fig. 7f) were comparable with those of control cells. Analysis of cell death in these cultures confirmed that this result was not due to a cytotoxic effect of verteporfin (Fig. 7g). These data demonstrate that, at low density, LIX1 regulates cell proliferation through modulation of YAP1 activity. After 3 days in culture, LIX1-expressing cells had grown faster than control cells (Fig. 7h, i). However, although YAP1 expression in LIX1-expressing cells remained higher than in controls (Fig. 7j, k), the levels of CTGF and CYR61 transcripts were similar to control levels. In addition, we observed an increase of the inactive phosphorylated form of YAP1 compared to controls in LIX1-expressing cells (Fig. 7k), indicating that YAP1 activity was downregulated at day 3 compared to day 1 (compare Fig. 7b with Fig. 7j). These data suggest that, under the influence of LIX1, a compensatory response to growing cell density took place. Indeed, while LIX1 acts to promote cell proliferation at low cell density, its pro-proliferation activity is abolished when cells had grown, suggesting that its ability to regulate cell proliferation is dependent upon cell density. In line with this hypothesis, when cells were seeded at high density, levels of CTGF and CYR61 transcripts, YAP1 activity, and rates of proliferation were comparable in controls and LIX1-expressing cells after 1 day in culture (Additional file 9: Figure S8). The overexpression of LIX1 in vitro thus recapitulates the effects we had observed under misexpression of LIX1 in vivo during stomach mesenchyme development, suggesting that LIX1 drives an increase in cell density that feeds back on the system to block the activity of YAP1 and further proliferation.