Abstract
Somitogenesis starts with cyclic waves of expression of segmentation clock genes in the presomitic mesoderm (PSM) and culminates with periodic budding of somites in its anterior-most region. How cyclic clock gene expression is translated into timely morphological somite formation has remained unclear. A posterior to anterior gradient of increasing PSM tissue cohesion correlates with increasing fibronectin matrix complexity around the PSM, suggesting that fibronectin-dependent tissue mechanics may be involved in this transition. Here we address whether the mechanical properties of the PSM tissue play a role in regulating the pathway leading to cleft formation in the anterior PSM. We first interfered with cytoskeletal contractility in the chick PSM by disrupting actomyosin-mediated contractility directly or via Rho-associated protein kinase function. Then we perturbed fibronectin matrix accumulation around the PSM tissue by blocking integrin-fibronectin binding or fibronectin matrix assembly. All four treatments perturbed hairy1 and meso1 expression dynamics and resulted in defective somitic clefts. A model is presented where a gradient of fibronectin-dependent tissue mechanics participates in the PSM wavefront of maturation by ensuring the correct spatio-temporal conversion of cyclic segmentation clock gene expression into periodic somite formation.
Introduction
Cells in the developing embryo are constantly receiving and integrating information, including mechanical signals generated by the adhesion to neighbor cells and/or the surrounding extracellular matrix (ECM). Cell-cell adhesion molecules and cell-ECM receptors, such as cadherins and integrins, respectively, are linked to the intracellular actomyosin cytoskeleton via intermediate proteins (Campbell and Humphries, 2011; Charras and Yap, 2018; Takeichi, 2014; Wolfenson et al., 2013). These adhesion complexes, called adhesomes, allow cells to perceive and respond to changes in their physical surroundings (Horton et al., 2016; Zaidel-Bar, 2013). Signaling events in adhesomes can impact the actomyosin cytoskeleton through the phosphorylation of non-muscle myosin II (NM II) which binds to actin and converts ATP into mechanical energy (Zaidel-Bar et al., 2015). The resulting actomyosin contractility leads to changes in cell shape and can transmit signals from integrin adhesomes to cadherin adhesomes and vice versa, as well as from the cell surface to the nucleus (Burute and Thery, 2012; Mui et al., 2016; Wolfenson et al., 2019). In this way, through continuous probing of their mechanical environment, cells adjust their shape, functions and behaviors, such as proliferation, differentiation, cell polarity and migration (Burute and Thery, 2012; Mui et al., 2016; Wolfenson et al., 2019). While morphogens have been extensively studied as major chemical regulators of developmental processes (Marek and Kubícek, 1981; Slack, 1987; Tiedemann, 1976), the importance of mechanical forces in embryo development has, until recently, received less attention (Marek and Kubícek, 1981; Slack, 1987; Tiedemann, 1976). It is, however, becoming increasingly clear that the ability of cells to sense and respond to mechanical signals regulates numerous basic developmental processes (e.g. Barriga et al., 2018; Brunet et al., 2013; Hiramatsu et al., 2013; Smutny et al., 2017).
One of the most conspicuous morphogenetic events during early vertebrate embryogenesis is the formation of somites, which are the source of axial skeleton and skeletal muscle precursor cells (Christ et al., 2007). Somites are spheres of epithelioid cells that are formed periodically from the anterior portion of the mesenchymal presomitic mesoderm (PSM), bilateral to the axial structures (Bailey and Dale, 2015). Temporal control of somite formation is dependent on cyclic waves of expression of segmentation clock genes, many of which are targets of the Notch signaling pathway (Dequéant et al., 2006; Masamizu et al., 2006; Palmeirim et al., 1997). These waves periodically sweep the PSM in a posterior to anterior direction (Aulehla and Pourquié, 2010; Bailey and Dale, 2015) and, as they reach the anterior PSM, oscillations slow down and then arrest (Morimoto et al., 2005; Shih et al., 2015). The transcription factor Mesp2/Meso1 is upregulated downstream of the segmentation clock in the anterior PSM, leading to Eph/Ephrin signaling and somitic cleft formation (Barrios et al., 2003; Nakajima et al., 2006; Saga, 2012; Watanabe et al., 2009), followed by progressive cell rearrangements into a somite (Martins et al., 2009; Morimoto et al., 2005; Shih et al., 2015).
Fibronectin is essential for somite formation in all vertebrate models studied to date (George et al., 1993; Georges-Labouesse et al., 1996; Goh et al., 1997; Koshida et al., 2005; Kragtorp and Miller, 2007; Rifes et al., 2007; Sato et al., 2007). Fibronectin matrix assembly is a complex cell-dependent process that requires the engagement and unfolding of globular fibronectin by the major fibronectin matrix assembly receptor, the α5β1 integrin, followed by fibrillogenesis involving fibronectin-fibronectin binding (Mao and Schwarzbauer, 2005; Singh et al., 2010). In the chick, a fibronectin matrix starts being assembled around the caudal PSM tissue and then gets progressively denser as the tissue matures (Rifes et al., 2007; Rifes and Thorsteinsdóttir, 2012). This results in the formation of a gradient of fibronectin matrix complexity along the PSM (Rifes and Thorsteinsdóttir, 2012), which correlates with a posterior to anterior gradient in cell density (Bénazéraf et al., 2010; Lawton et al., 2013; Mongera et al., 2018). At the rostral end, fibronectin is required for the polarization of N-cadherin and epithelialization of peripheral cells to form a somite (Martins et al., 2009; Rifes et al., 2007). Interestingly, adhesion to a fibronectin substrate was noted as a regulator of the oscillations of the segmentation clock gene Lfng in cultured mouse tailbud cells (Hubaud et al., 2017; Lauschke et al., 2013). Cell adhesion to fibronectin was linked to dampening and eventual arrest of Lnfg oscillations (Hubaud et al., 2017), reminiscent of what is observed in the anterior PSM prior to somite epithelialization. However, whether the mechanical properties of the PSM tissue play a role in the slowing down of segmentation clock oscillations and their conversion into segments remains unknown.
In this study, we addressed the involvement of PSM tissue mechanics in the regulation of both segmentation clock gene expression dynamics and subsequent somite formation using the chick embryo as a model. First, we experimentally perturbed actomyosin contractility by blocking either NM II ATPase activity or Rho-associated protein kinase (ROCK). We then addressed the role of the fibronectin matrix surrounding the PSM by blocking integrin-fibronectin binding through RGD or by perturbing extracellular fibronectin fibrillogenesis. We found that each one of the four treatments resulted in abnormal segmentation clock oscillations, mis-positioning of meso1 expression in the rostral PSM and perturbations in somite morphogenesis. These results strongly suggest that fibronectin-dependent PSM tissue mechanics play a role in converting segmentation clock oscillations into periodic somite formation.
Results
Intracellular actomyosin contractility is required for timely segmentation clock oscillations and meso1 activation
In the chick embryo, sequential pairs of somites bud off from the anterior PSM every 90 min, which corresponds to the period of segmentation clock oscillations (Figure 1 A). To investigate the involvement of intracellular actomyosin contractility in this process, the expression of the segmentation clock gene hairy1 (Palmeirim et al., 1997) was analyzed in the PSM of embryo half explants cultured in the presence of either Blebbistatin, which directly inhibits the ATPase activity of NM II and consequently all actomyosin contractility (direct inhibition), or RockOut, a chemical inhibitor of ROCK I and II (ROCK I/II) enzymes involved in activating NM II (indirect inhibition) (Figure 1 B, C; Ringer et al., 2017; Straight et al., 2003; Yarrow et al., 2005). The contralateral control sides were cultured with an equal volume of DMSO.
Explants cultured for 6 hours in each experimental condition presented significantly altered hairy1 expression. hairy1 expression was either absent or in a different phase of the cycle relative to the contralateral control in 80% of the Blebbistatin-(n=7/9) and RockOut-treated explants (n= 8/10; Figure 1 D, a-c), suggesting that temporal control of hairy1 oscillations requires the generation of tensional cues mediated by NM II and ROCK I/II activity.
Segmentation clock oscillations are required for the correct spatial and temporal upregulation of Mesp2 in the anterior PSM (Niwa et al., 2011; Saga and Takeda, 2001; Sato et al., 2002), which regulates downstream targets needed for the formation of the somitic cleft (Saga, 2012). The expression of the chick Mesp2 homolog, meso1, was altered in Blebbistatin-(n=9/11) and RockOut-treated explants (n=10/10; Figure 1 D, d-f). meso1 expression was either absent, located more rostrally or presented a different number of bands of expression, clearly indicating that the normal cycles of activation and suppression of meso1 in the rostral PSM were altered. Importantly, meso1 expression was also perturbed after 3 hours in culture with Blebbistatin or RockOut (n=8/9 and 5/5, respectively; Supplementary Figure 1 A-B), corresponding to an effect within two segmentation clock cycles. Furthermore, timely downregulation of dll1 in the anterior-most PSM (Palmeirim et al., 1998), which normally occurs downstream of Meso1/Mesp2 activity (Takahashi et al., 2000; Takahashi et al., 2003), was not observed in either Blebbistatin-(80%, n=8/10) or RockOut-treated (75%, n=6/8) explants after 10.5h of culture (Figure 1 D, g-i). Together these data indicate that interfering with actomyosin contractility perturbs three sequential events: the spatio-temporal expression dynamics of hairy1, timely meso1 expression and the downregulation of dll1 expression in the anterior-most PSM.
Alterations in somite formation were observed concomitantly with the perturbations in hairy1 and meso1 expression. Control explants formed an average of 3.6 somites after 6 hours, consistent with a 90 min periodicity (Figure 1 E), while contralateral RockOut-treated explants formed 3.1 somites, of which only the first two somites were clearly individualized, while subsequent somite-like condensations were poorly defined (Figure 1 E). After 10.5 hours, control explants had formed an average of 7.5 somites, while RockOut-treated explants formed 7.1 somites of which the first two appeared normal, but the remaining ones were ill-defined (Figure 1 E). Importantly, explants cultured with Blebbistatin were unable to form more than 1-2 somite-like aggregates after 6 hours, or even 10.5 hours, of culture (Figure 1 E), evidencing an absolute requirement for NM II ATPase activity in somite formation. These effects were not due to an increase in apoptosis (Supplementary Figure 2 A-D).
Our data reveal a previously unknown role for NM II-and ROCK I/II-mediated cell contractility in the temporal regulation of the segmentation clock, meso1 expression, dll1 downregulation and, consequently, in somite formation.
NM II activity is required for somite cleft formation and cell polarization
Somite formation involves a mesenchymal-to-epithelial transition (MET) of anterior PSM cells (Martins et al., 2009; Saga, 2012). To determine to what extent this process is impaired upon NM II or ROCK I/II inhibition, we performed a detailed analysis of the morphology of S0 to SIII in explants after a 6 hour culture period (regions e-b in Figure 1 A).
In control explants, S0 showed apically enriched N-cadherin (Figure 2 A, D) and some zonula occludens protein 1 (ZO-1) accumulation was observed apically (Figure 2 B, D, arrowheads). Peripheral cell alignment occurred (Figure 2 C, D) and fibronectin matrix was detected in the nascent somitic clefts (Figure 2 E, F, arrows). In contrast, in explants cultured with Blebbistatin, N-cadherin was homogeneous (Figure 2 G, J), ZO-1 immunostaining was absent (Figure 2 H, J) and neither peripheral cell alignment (Figure 2 I, J) nor fibronectin matrix accumulation within the tissue was observed (Figure 2 K, L). Furthermore, the continuous and dense fibronectin matrix observed surrounding the rostral PSM in control explants was disrupted in Blebbistatin-treated explants (compare Figure 2 E and K, arrowheads). Moreover, the characteristic nuclear alignment and F-actin apical enrichment observed in control SI (Supplementary Figure 3 A-C) was absent in Blebbistatin-treated explants and no signs of somitic boundaries could be detected (Supplementary Figure 3 D-F). We conclude that exposure of the S-IV and S-III regions of the PSM (regions e and d in Figure 1 A) to Blebbistatin for 6 hours completely blocks their capacity to form somites.
We next turned our attention to SII and SIII somites after 6 hours of culture. These were at stage S-II and S-I in the PSM, respectively (regions c and b in Figure 1 A), when the explants were placed in culture and had thus already upregulated meso1 (Buchberger et al., 1998). As before, in the presence of Blebbistatin, apical enrichment of N-cadherin failed to occur (Figure 2 S, V), ZO-1 was only detected in a few small foci (Figure 2 T, V) and, although a fibronectin matrix was present, it appeared less dense (Figure 2 W, X). An incipient nuclear alignment was sometimes observed (Supplementary Figure 3 K, arrowheads), but cells did not polarize their F-actin into apically enriched adhesion belts (compare Supplementary Figure 3 G-I with J-L). Epithelial tissues other than somites (e.g. ectoderm and neural tube) did not present significant alterations after incubation with Blebbistatin (Supplementary Figure 4). Altogether, these results point to an indispensable role for NM II activity for the MET underlying somite formation.
RockOut-treated explants also showed perturbations in somite formation, although to a lesser extent (Figure 1 E). When compared to control explants (Figure 3 A-D; Supplementary Figure 5 A-C), RockOut treatment resulted in incomplete somitic clefts, such that S0 shared the somitocoel with SI and sometimes also with SII (Fig 3 E-H, arrows, Supplementary Figure 5 D-F, arrows). In contrast to the accumulation of fibronectin in the nascent clefts in controls (Figure 3 B, arrow), no fibronectin was observed in the incipient somitic clefts of RockOut-treated explants (Figure 3 F, arrows). These results suggest that ROCK I/II activity in the S-IV and S-III regions of the PSM (regions e and d in Figure 1 A) is required for the formation of individualized somites. In contrast, when the rostral-most PSM (stage S-I before culture, region b in Figure 1 A) was exposed to RockOut for 6 hours, it was indistinguishable from control explants, showing apical accumulation of ZO-1 (Figure 3 I, M, arrowhead) and N-cadherin (Supplementary Figure 5, G, J), nuclear alignment (Figure 3, K, O, Supplementary Figure 5, H, K) and a complete, fibronectin matrix-containing cleft (Figure 3 J, N). This indicates that ROCK I/II activity is not required for S-I to develop into a somite.
Altogether, our data indicate that intracellular actomyosin contractility plays a role in periodic somite cleft formation, and that ROCK I/II-independent NM II activity is essential for MET.
Blocking integrin-fibronectin binding perturbs segmentation clock oscillations and somitic cleft formation
Our next aim was to address the requirement for the fibronectin ECM surrounding the PSM in regulating segmentation clock oscillations and somite formation. PSM cells bind to the RGD motif of fibronectin through the α5β1 integrin, an interaction that plays a crucial role during somitogenesis (Girós et al., 2011; Yang et al., 1993). αv integrins, which have been described to bind the RGD motif and partially compensate for the absence of the α5β1 integrin in the mouse (Yang et al., 1999), are not detected in the PSM of the chick embryo (Gomes de Almeida et al., 2016). We cultured embryo half explants in the presence of a linear RGD peptide, which competes with fibronectin for integrin binding (Huveneers et al., 2008; Pierschbacher and Ruoslahti, 1984), and compared them to contralateral control explants (Figure 4 A, B). RGD-treated explants formed ill-defined somite-like condensations, although in approximately the same number as the contralateral control (Figure 4 C). This was not due to cell death (Supplementary Figure 2 E-F). Concomitantly, RGD-treated explants displayed alterations in hairy1 (Figure 4 D, a-c), meso1 (Figure 4 D, d-f) and dll1 expression patterns (Figure 4 D, g-i), evidencing that integrin-fibronectin interactions via RGD are required for proper segmentation clock oscillations, meso1 positioning and timely downregulation of dll1 in the anterior PSM.
When compared to the contralateral control, the area corresponding to S0 after 6 hours of culture with RGD (region e in Figure 1 A) showed deficient nuclear alignment (Figure 5 B, E, H, K, arrowheads) and N-cadherin polarization (Figure 5 A, G, arrowheads), accompanied by deficient fibronectin assembly in the nascent cleft (Figure 5 D, J arrowheads). At the level of SII (region c in Figure 1 A), complete somite individualization was impaired in RGD-treated explants (Figure 5 N, Q, T, W, arrowheads) and, although N-cadherin polarization appeared normal (Figure 5 M, S, arrowheads), cleft formation (Figure 5 Q, W, arrowheads; R, X) and fibronectin assembly between adjacent somites was deficient (Figure 5 P, V, arrowheads; R, X).
These findings implicate cell-ECM interactions, mediated by integrin-fibronectin binding via the RGD motif, in temporal control of hairy1 expression, correct positioning of meso1 expression, downregulation of dll1 in the anterior PSM and somite morphogenesis.
Impaired fibronectin matrix assembly results in altered segmentation clock dynamics, meso1 expression and defects in somite morphogenesis
Another way to assess the relevance of the extracellular fibronectin matrix on the events leading up to somite formation is to perturb fibronectin assembly in the PSM and somites. To this end, primitive streak-stage embryos were electroporated with a construct expressing the 70kDa fibronectin fragment, a dominant-negative inhibitor of fibronectin matrix assembly (Figure 6 A, B; McKeown-Longo and Mosher, 1985; Sato et al., 2017). 70kDa-electroporated embryos exhibited a disrupted fibronectin matrix, composed of thinner fibrils when compared to control pCAGGs-electroporated embryos (Supplementary Figure 6 A; also see Figure 7 A, B). These embryos displayed multiple morphological defects, including kinked neural tube and detached tissues as well as perturbations in somite morphogenesis, which are all reminiscent of phenotypes obtained in previous studies interfering with fibronectin matrix deposition and/or with fibronectin-integrin binding (Supplementary Figure 6 B, C; Drake et al., 1992; Drake and Little, 1991; George et al., 1993; Girós et al., 2011; Takahashi et al., 2007). Although the average number of somite-like structures formed in 70kDa-electroporated embryos was similar to the number of somites in controls (Figure 6 C), the former were ill-defined, often appearing fused or crammed (Supplementary Figure 6 B e-f, C), closely resembling the somite-like condensations formed in RockOut- and RGD-treated explants (Figures 3 and 5).
Next, we sought to evaluate the impact of inhibiting fibronectin matrix assembly on the molecular machinery underlying somite formation. Unilateral electroporation of the 70kDa fragment was performed (Figure 6 A) to allow the direct comparison of gene expression patterns within the same embryo. We observed a significant increase in the frequency of perturbations in the expression of embryonic clock genes hairy1 (p<0.01) and hairy2 (p<0.05), as well as in meso1 (p<0.01) and dll1 (p<0.01) expression, when compared with embryos electroporated with pCAGGs alone (Figure 6 D).
Consistent with these data, 70kDa-electroporated embryos have deficiencies in somite morphogenesis which, in severe cases, leads to incomplete somitic clefts (Figure 7 A, B k, w, arrows). Peripheral cells of nascent somites of 70kDa-electroporated embryos did, however, accumulate ZO-1 apically (Figure 7 B e, h, i, l) which was maintained as the somites matured (Figure 7 B q, u, t, x). Nevertheless, these somites were abnormal in shape and appeared smaller in severely affected embryos (Figure 7 B, i-I, u, x). In fact, the SI of the electroporated sides of 70kDa-treated embryos were significantly smaller in width than those of the contralateral control, while SV was significantly shorter in length (n=151; Supplementary Figure 7). In addition to defects in somite morphology, the ectoderm and endoderm were separated from the paraxial mesoderm in 70kDa-electroporated embryos, indicating that their fibronectin matrix was insufficient to hold these tissues together (brackets in Figure 7 B g, k, s, w).
We conclude that proper fibronectin matrix assembly in the PSM is required for timely clock gene expression dynamics, positioning of meso1 expression and dll1 downregulation in the rostral-most PSM, as well as for the complete separation and morphogenesis of somites.
Discussion
Fibronectin-dependent tissue mechanics coordinate segmentation clock dynamics and cleft formation
We have identified fibronectin-dependent tissue mechanics as a regulator of segmentation clock gene expression and the positioning of the presumptive somitic cleft in the chick embryo. Four independent treatments interfering with the mechanical properties of the PSM (Figure 8 A) consistently lead to asymmetric patterns of hairy1 expression as well as incorrect positioning of meso1 expression on the experimental versus the control sides of the same embryo. The similarity of the phenotypes obtained in experiments impairing fibronectin fibrillogenesis, cell-fibronectin interactions via RGD and blocking ROCK (Figure 8 B) suggests that the fibronectin matrix, the RGD-binding α5β1 integrin, and ROCK-dependent actomyosin contractility are part of the same pathway. The α5β1 integrin can transduce mechanical signals by ROCK activation (Schiller et al., 2013) and this requires the binding of α5β1 to two sites of fibronectin, namely the RGD and the synergy site (Friedland et al., 2009). Surprisingly, although the RGD site of fibronectin is crucial for somitogeneses (Girós et al., 2011), removing the synergy site of fibronectin (Fn1syn/syn) had no effect on mouse embryonic development (Benito-Jardón et al., 2017). However, cells expressing αv-integrins form strong adhesions on fibronectin lacking the synergy site and can compensate for the inability of α5β1 to mediate adhesion strengthening on this fibronectin (Benito-Jardón et al., 2017), and αv-integrins can partially compensate for the complete absence of α5β1 during mouse somitogenesis (Yang et al., 1999). Alternatively, ROCK can be activated indirectly, for example through cadherin engagement and subsequent adherens junction formation (Burute and Thery, 2012; Schwartz and DeSimone, 2008), which occurs upon fibronectin-induced polarization of peripheral PSM cells (Martins et al., 2009).
It is well established that Notch signaling plays a role in the segmentation clock and is also required for timely meso1 activation (Saga, 2012). Hence, the mechanical environment may be regulating Notch signaling in the PSM. In agreement with our results, chicken embryos electroporated with RNAi constructs against integrin β1 showed alterations in hairy2, lfng and meso1 expression in the PSM (Rallis et al., 2010). Mouse embryos where the fibronectin RGD site has been substituted with an RGE sequence (Fn1RGE/RGE) also showed asymmetric and/or dampened expression of Lnfg and Hes7 in the PSM (Girós et al., 2011) and EphA4, a direct target of Mesp2 in the anterior PSM (Nakajima et al., 2006), was diffusely expressed or absent (Girós et al., 2011). Finally, combined roles of integrin α5β1 and Notch are required for zebrafish somitogenesis (Jülich et al., 2005).
Exactly how tissue mechanics regulate Notch signaling is unknown. It is becoming increasingly clear that mechanics play a crucial role in Notch activation (Gordon et al., 2015; Luca et al., 2017; Meloty-Kapella et al., 2012) and sustained Notch signaling in the Drosophila notum requires actomyosin contractility in both signal sending and receiving cells (Hunter et al., 2019). Intriguingly, clock oscillations in mouse tailbud PSM explants cultured on fibronectin are sustained in the presence of a ROCK inhibitor, suggesting that ROCK activity must be low for the maintenance of clock oscillations and that an increase in ROCK activity normally stops segmentation clock oscillations in this system (Hubaud et al., 2017). Moreover, in the same study, cell adhesion to fibronectin was linked to nuclear localization of Yes-associated protein (YAP), an intracellular sensor of cell mechanics, and dampening and eventual arrest of Lnfg oscillations was found to be YAP-dependent (Hubaud et al., 2017). ROCK-mediated actomyosin contractility is known to promote the nuclear localization of YAP in several cell types (Piccolo et al., 2014) and YAP-null mouse mutants (Morin-Kensicki et al., 2006) have a phenotype very similar to that of integrin α5-null mutants (Yang et al., 1993) and Fn1RGE/RGE embryos (Girós et al., 2011), suggesting that they contribute the same processes during early embryo development. Thus, it is conceivable that increased fibronectin-dependent tissue cohesion may translate into increased ROCK activity and actomyosin contractility, promoting sustained Notch signaling and nuclear localization of YAP, leading to the dampening and eventual arrest of clock oscillations. Further studies are needed to test this hypothesis.
Altogether, our results show that perturbation of the normal PSM tissue mechanics leads to a dysregulation of segmentation clock oscillations and the mispositioning of the segmental border, indicating that the mechanical properties of the PSM modulate Notch signaling and coordinate the translation of clock oscillations into periodic segmental border formation.
Somite cleft formation and cell epithelialization have different mechanical requirements
In the rostral PSM, Mesp2/Meso1 activates the expression of EphA4, which interacts with EphrinB2 in cells rostral to the presumptive cleft, causing cell-cell repulsion and the formation of an incipient cleft (Nakajima et al., 2006; Watanabe et al., 2009). Then, fibronectin matrix assembly within the cleft stabilizes it (Jülich et al., 2015; Rifes and Thorsteinsdóttir, 2012) and promotes the epithelialization of cells rostral to the cleft (Martins et al., 2009). This can be defined as the first step of morphological somite individualization. The second step is defined as the complete epithelialization of the remaining cells of the nascent somite and lasts until SII, when all somitic cells have acquired a spindle-like shape and are organized into a rosette (Martins et al., 2009).
RockOut-, RGD- and 70kDa-treated embryos (Figure 8 A, B) all show perturbations in segmentation clock gene expression, abnormal positioning of meso1 and defects in the somitic clefts. Although somite morphology is also perturbed, the acquisition of the spindle-shape cell morphology which occurs as S0 develops into SII does not appear to be significantly perturbed. Thus, the first step of morphological somite formation is affected but the second step is not. In contrast, Blebbistatin-treated explants not only have the defects listed above, but cells that had already upregulated meso1 before the addition of the drug and formed an incipient cleft during culture, were completely unable to epithelialize. In fact, Blebbistatin-treated explants formed only 1 or 2 somites, and their cells did not acquire the elongated, spindle-shape typical of SI and SII somites. Hence, Blebbistatin affects both steps of morphological somite formation.
RockOut targets NM II activity indirectly by inhibiting ROCK I and II, two of the kinases that activate NM II (Newell-Litwa et al., 2015). In contrast, Blebbistatin directly targets the NM II ATPase. Our results thus raise the possibility that the acquisition of the spindle-shaped morphology of cells may be dependent on another NM II activator. Interestingly, Ca++/calmodulin signaling can activate NM II and inhibiting calmodulin was shown to block the acquisition of this morphology during chick somitogenesis (Chernoff and Hilfer, 1982).
A gradient of fibronectin-dependent tissue mechanics as a player in the PSM wavefront of maturation
While the waves of Notch oscillatory activity travel through the entire length of the PSM, they are only translated into segments in the anterior-most region of the tissue. Opposing gradients of Fgf/Wnt and Retinoic acid (RA) in the PSM are thought to define the wavefront, which marks the region where PSM cells become competent for somite formation (Hubaud and Pourquié, 2014). Rostral to the wavefront, segmentation clock oscillations progressively slow down until they reach a halt and cells become part of a somite (Palmeirim et al., 1997).
Herein, we propose a model (Figure 8 C) where a posterior to anterior gradient of fibronectin-dependent tissue cohesion is interpreted by the PSM cells as an increasing gradient of mechanical tension (Figure 8 C). We propose that the fibronectin-dependent tensional state of the PSM at the level of the wavefront acts as a threshold that activates a mechanotransduction signaling cascade, ensuring the correct spatio-temporal conversion of the cyclic expression of segmentation clock genes into periodic meso1 expression, which in turn defines the next somitic cleft (Figure 8 C). In support of this hypothesis, a gradient of increasing paraxial mesoderm stiffness from the tail to rostral somites has been identified in the chick embryo (Bénazéraf et al., 2017; Marrese et al., 2019). During its maturation, the PSM thus integrates a combination of chemical and mechanical signals, namely gradients of Fgf/Wnt and RA (Aulehla and Pourquié, 2010) and, simultaneously, a progressive increase in fibronectin-dependent tissue cohesion (Figure 8 C). Fibronectin matrix-dependent tissue mechanics would thus be a key contributor to the PSM wavefront of maturation in that it regulates where and when the next somitic cleft is positioned in the anterior PSM (Figure 8 C). Hence, we propose that a mechanotransduction pathway downstream of fibronectin plays a major role in the translation of cyclic waves of expression of segmentation clock genes into the periodic morphogenesis of somites.
Materials and Methods
Embryos
Fertilized chicken (Gallus gallus) eggs were obtained from commercial sources (Sociedade Agrícola Quinta da Freiria or Pintobar Exploração Avícola, Lda, Portugal) and incubated at 37.5°C in a humidified chamber until the desired HH stage (HH4 or HH11-14; Hamburger and Hamilton, 1992). Somite nomenclature is according to Pourquié and Tam (2001).
Embryo explant culture and chemical treatments
Explant tissues of HH11-14 embryos were collected and cultured as previously described (Palmeirim et al., 1997; Rifes et al., 2007). Embryos were bisected along the midline and then cut transversally rostral to somites IV and Hensen’s node. The two contralateral halves thus retained half of the neural tube and notochord as well as the first four somites and the PSM, with all remaining neighboring tissues intact. Explants were placed on top of a polycarbonate filter floating on M199 medium supplemented with 10% chick serum, 5% fetal calf serum and 100 U/ml of penicillin and streptomycin (Palmeirim et al., 1997). Explants were then cultured at 37°C with 5% CO2 from 6 to 12 hours.
InSolution™Blebbistatin (Calbiochem) and RockOut (Calbiochem) diluted in DMSO were used at a final concentration of 50 µM in culture medium. Equal volumes of DMSO (Sigma-Aldrich) were used as control for both drugs. The linear RGD peptide (GRGDS - G4391, Sigma) was diluted in culture medium and used at 0.9 mM, while control explants were cultured in medium only. RGD peptide efficiency was confirmed in a cell adhesion assay (Danen et al., 2002; Pierschbacher and Ruoslahti, 1984) before using it on explants.
Embryo electroporation and ex ovo culture
HH4-5 embryos were electroporated on one (randomly selected) side of the primitive streak in the presumptive PSM and/or ectoderm and cultured ex ovo using the Early Chick culture method (Chapman et al., 2001). The electroporation mixture contained plasmid DNA at 0.5-1 µg/µl mixed with 0.4% Fast Green for visualization. Embryos were submerged in an electroporation chamber filled with Tyrode’s saline and three pulses of 6-9 V, 50 ms each, at 350 ms intervals were applied. Control embryos were electroporated with pCAGGs containing a GFP reporter (pCAGGs-GFP; abbreviated pCAGGs). pCAGGs-70kDa qFN1 was kindly provided by Yuki Sato (Sato et al., 2017) and was co-electroporated with the pCAGGs-GFP plasmid in experimental embryos (treatment abbreviated 70kDa). Electroporated embryos were screened for GFP after fixation to select embryos with an intense signal on only one side to process for whole mount morphological analysis, in situ hybridization experiments and transverse sectioning. For morphological analysis in sagittal sections, the embryo side electroporated with pCAAGGs was compared to the 70kDa-electroporated side of other same stage embryos.
Cryosectioning and immunohistochemistry
Cryosectioning was performed on embryo explants and whole embryos fixed in 4% paraformaldehyde in 0.12 M phosphate buffer containing 4% sucrose and processed for cryoembedding. Fixed samples were washed in 0.12 M phosphate buffer with 4% and 15% sucrose and then embedded in 7.5% gelatin in 0.12 M phosphate buffer containing 15% sucrose, frozen on dry ice-chilled isopentane and stored at −80°C until sectioning. Cryostat sections (10-30 µm) were processed for immunofluorescence as previously described (Gomes de Almeida et al., 2016). Permeabilization of sections was performed with 0.2% Triton-X100 in phosphate buffered saline (PBS). 5% bovine serum albumen (BSA) or a combination of 1% BSA and 10% Normal Goat Serum (NGS) in PBS were used for blocking depending on the presence or absence of anti-fibronectin antibodies, respectively. Primary and secondary antibodies were diluted in 1% BSA in PBS. Sections were incubated with primary antibodies overnight at 4°C and with secondary antibodies for 1 hour at room temperature. For whole-mount immunodetection, explants were fixed in 4% paraformaldehyde in PBS and processed as previously described (Martins et al., 2009; Rifes and Thorsteinsdóttir, 2012). 1% Triton-X100 in PBS was used for permeabilization and 1% BSA in PBS was used for blocking and antibody dilution. Antibody incubation was performed overnight at 4°C.
The following primary antibodies were used: anti-ZO-1 (Zymed, #40-2200, 1:100 or Invitrogen, #33-9100, 1:100); anti-N-cadherin (BD Biosciences, #610920, 1:100); anti-fibronectin (Sigma, #F-3648, 1:400), anti-activated caspase3 (Cell Signaling, #9661, 1:1000) and anti-GFP (Invitrogen, #A11122, 1:100). For F-actin staining we used Alexa 488-conjugated phalloidin (Invitrogen, 1:40) and for staining DNA we used ToPro3 (Invitrogen, 1:500) in conjunction with ribonuclease A (Sigma, 10 µg/ml), 4% Methyl Green (Sigma, diluted 1:250; Prieto et al., 2015) or 4′,6-diamidino-2-phenylindole (DAPI, 5µg/ml in PBS with 0.1% Triton-X100). For detection of the primary antibodies the adequate secondary goat anti-mouse and anti-rabbit Alexa 488-, Alexa 568-or Alexa 546-conjugated F’ab fragments from Invitrogen were used (#A-11017, #A-21069, #A-11071, #A-11019, #A-11070, 1:1000). Immunohistochemistry was performed on at least 6 different explants/embryos and the respective controls for each treatment (Blebbistatin n=13; RockOut n=15; RGD n=13; 70kDa n=7/pCAGGs n=6).
In situ hybridization
In situ hybridization using DIG-labeled RNA probes was performed as described previously (Henrique et al., 1995) with minor alterations (Gomes de Almeida et al., 2016). RNA probes were synthetized from linearized plasmids: dll1 (Henrique et al., 1995), meso1 (Buchberger et al., 1998), hairy1 (Palmeirim et al., 1997) and hairy2 (Jouve et al., 2000).
Statistical analysis
Paired Student’s t-tests were performed to assess for differences in the number of somites formed in Blebbistatin-, RockOut- and RGD-treated explants relative to the respective controls, and in embryos electroporated with pCAGGs only and pCAGGs + 70kDa. Differences in the frequency of morphological and gene expression phenotypes found in 70kDa-electroporated embryos compared to pCAGGs-electroporated control embryos was tested through a Chi-square test. Differences in somite size between pCAGGs- or 70kDa-electroporated sides compared to the control (non-electroporated) sides of embryos was tested through a nested ANOVA. The side in which the embryo was electroporated (left or right) was nested in the treatment (non-electroporated, pCAGGs-electroporated or 70kDa-electroporated) to account for a potential variability between the two sides. Statistical significance was set at p<0.05. Statistical analyses were performed in Statistica 10 (https://statistica.software.informer.com/10.0/), Graphpad Prism 5 (https://graphpad-prism.software.informer.com/5.0/) and RStudio (https://rstudio.com/).
Sample preparation and imaging
Whole mount explants were gradually dehydrated in methanol and cleared in methylsalicylate (Sigma-Aldrich) as described previously (Martins et al., 2009; Rifes and Thorsteinsdóttir, 2012), except for phalloidin-labelled embryos and explants, where a shorter series of ethanol dehydration series was used. Cryostat sections were mounted in Vectashield (Vector Laboratories) or in 5mg/ml propyl gallate in glycerol/PBS (9:1) with 0.01% azide. Immunofluorescence images were taken on a confocal Leica SPE microscope, following imaging acquisition steps described previously (Rifes and Thorsteinsdóttir, 2012). Imaging of electroporated embryos and explants processed for in situ hybridization was performed using a Zeiss LUMAR V12 Stereoscope coupled to a Zeiss Axiocam 503 color 3MP camera. Image analysis was performed using Fiji v. 1.49 (https://imagej.net/Fiji) software. Image histogram corrections and, when appropriate, maximum intensity projections of immunofluorescence confocal stacks were produced in Fiji and exported as TIFF files. When applicable, contiguous images were stitched together into a single image using the pairwise stitching Fiji plugin (Preibisch et al., 2009). For the analysis of in situ hybridization patterns along the PSM explants, the Fiji plugin Straighten (Kocsis et al., 1991) was used.
Competing interests
The authors declare no competing or financial interests.
Author contributions
Conceptualization: P.G.A., P.R., R.P.A., S.T.; Methodology: P.G.A., P.R., R.P.A., S.T.; Validation: P.G.A, P.R., A.P.M.J., G.G.P., R.P.A., S.T.; Formal analysis: P.G.A.; Investigation: P.G.A., P.R., A.P.M.J., G.G.P.; Resources: R.P.A., S.T.; Writing - original draft: P.G.A., S.T.; Writing - review & editing: P.G.A, P.R., A.P.M.J., G.G.P., R.P.A., S.T.; Visualization: P.G.A., P.R., R.P.A., S.T.; Supervision: R.P.A., S.T.; Project administration: P.G.A., R.P.A., S.T.; Funding acquisition: R.P.A., S.T.
Acknowledgements:
We thank Dr. Yuki Sato for generously sharing the pCAGGs-q70kDa construct and Inês Fragata for help with the statistical analysis. This work was supported by Fundação para a Ciência e a Tecnologia (FCT, Portugal) projects PTDC/SAU-OBD/103771/2008, PTDC/BEXBID/5410/2014, UID/BIA/00329/2013, UID/BIM/04773/2019 CBMR, and FCT scholarships SFRH/BD/86980/2012 (PGA) and SFRH/BD/37423/2007 (PR). Imaging and image analysis were done in the Microscopy Facility at the Faculty of Sciences of the University of Lisbon and the Light Microscopy Unit of CBMR-UAlg, nodes of the Portuguese Platform for BioImage (reference PPBI-POCI-01-0145-FEDER-022122). Finally, we are grateful to Isabel Palmeirim and all members of our laboratories for their support and helpful discussions.
Footnotes
The text of the manuscript has been slightly revised.
References:
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵