Abstract
Sonic hedgehog (Shh) is necessary both for neural and mesodermal development. By loss and gain of Shh function, and floor plate deletions, we report that Shh released from notochord into sclerotome constitutes a dynamic pool of ligand serving both tissues. Depletion of Shh in sclerotome by membrane-tethered hedgehog-interacting protein or by Patched1, but not by a dominant active version of Patched, reduced motoneuron numbers and also compromised myotome differentiation. These effects were a specific and direct consequence of reducing Shh. In addition, grafting notochords in a basal, but not apical location vis-a-vis the tube, profoundly affected motoneuron development, suggesting that initial ligand presentation occurs at the basal side of epithelia corresponding to the sclerotome-neural tube interface.
Collectively, our results reveal the sclerotome as a previously unknown common pool and potential site of a Shh gradient that coordinates development of mesodermal and neural progenitors.
Introduction
The morphogen Sonic hedgehog (Shh) plays fundamental roles in the development of both neural tube (NT) and somites [1–6]. Its signaling is initiated by binding of the proteolytically processed and lipid modified ligand to the transmembrane receptor Patched (Ptc), that represses the pathway in the absence of ligand [7–10]. Ligand binding to Ptc abrogates its repressive effect on Smoothened (Smo), a key effector essential for canonical Hedgehog signal transduction [11]. The repressive role of Ptc correlates with its localization in the primary apically located cilium, that functions as a signal transduction compartment [12, 13]. Binding of Shh to Ptc removes Smo from the cilium, thereby allowing Smo to enter and propagate the signal further downstream [14, 15] to regulate Gli transcription factor activity [16, 17].
Shh signaling is highly regulated by negative and positive modulators. Ptc1, Hedgehog interacting protein (Hhip) and Gli1 are direct targets of Shh and the former two also inhibit its activity [18, 19]. Sulfatase1 (Sulf1) is a known Shh target [20] and the co-receptors Boc, Gas and Cdo [21, 22] all enhance ligand activities and are expressed in the NT and/or developing mesoderm [23].
It is well established that at early stages following neurulation, Shh secreted by the notochord (No) induces distinct ventral cell identities in the overlying NT by a mechanism that depends on relative concentrations and duration of exposure [24–26]. No-derived Shh is also involved in mesoderm patterning [4, 5]. A ventro-dorsal activity gradient of Shh/Gli signaling in sclerotome was directly visualized using an in vivo reporter in mice [23]. In addition, in chick embryos, Shh activity spreads from the midline through the sclerotome to reach the dermomyotome (DM). There it promotes terminal myogenic differentiation of both epaxial and hypaxial DM-derived progenitors and maintains epitheliality of DM cells [23]. Notably, in both floor plate (FP) and myotome, the activities of Shh are transient and cells become refractory to the ligand, a mechanism that allows dynamic phase transitions to take place within these systems [23, 27].
Because Shh is important for the development of both NT and mesoderm, two functionally interconnected systems, the question arises whether the effects of Shh on either tissue are independent of each other or interrelated. Answering this question is of utmost significance both for better understanding the mechanism of Shh activity and for achieving an integrated molecular view of regional development.
In this study we report that, in addition to affecting muscle development, depletion of Shh in the sclerotome by Hhip1 also significantly reduces motoneuron numbers in NT. Notably, similar effects are monitored when a membrane-tethered version of Hhip1 (Hhip:CD4), unable to be released from cells, is locally electroporated into the sclerotome. The observed phenotypes are a specific and direct consequence of Shh depletion as they are rescued by excess Shh, direct Shh targets such as Hhip1 and Gli1 mRNAs are reduced, and its effects are not mediated through other signaling pathways. Most importantly, the effects of Hhip:CD4 are phenocopied by electroporation of the transmembrane receptor Ptch1 but not by constitutively active PTCΔloop2 which does not recognize the ligand. In addition, by gain and loss of Shh function, and by FP deletions, we show that the sclerotome, but not the NT itself, constitutes a dynamic pool of No-derived Shh that acts both on motoneuron as well as on myotome development. Furthermore, grafting No fragments adjacent to the basal, sclerotomal side of the NT profoundly affects its development when compared to apical grafts. A similar basal grafting with respect to the DM significantly enhances myotome formation, suggesting a general need for initial ligand presentation at the basal side of epithelia. Together, our results uncover the sclerotome as a hitherto unknown common pool of Shh that promotes development of both mesodermal and neural progenitors.
Results
Reduction of Shh in sclerotome by Hhip1 affects both myotome and motoneuron differentiation
To investigate possible interactions between neural and mesodermal progenitors mediated by Shh, electroporations were performed in avian embryos aged 23-25 somite pairs at the level of epithelial somites prior to myotome development. In this region, the NT is composed of proliferative cells [28] and neural patterning is already apparent and ongoing, as evidenced by expression of the positive Shh targets Nkx2.2, Olig2, Nkx6.2 and Nkx6.1 (Supplem. Fig. S1, A-D). However, differentiation into Hb9-expressing motoneurons has not yet occurred at this stage (Supplem. Fig. S1, E) and only starts about 10 hr later at the level of somites 11-12 located rostral to the last segmented pair of somites (Supplem. Fig.S1, F, arrows). Hence, the timing of manipulations of Shh activity corresponds to the transition of specified proliferative progenitors into differentiated motoneurons [29].
In a previous study, we reported that Shh traversing the sclerotome is necessary for myotome differentiation, as missexpression of the high affinity and selective Shh antagonist Hhip1 in the sclerotome resulted in smaller myotomes expressing desmin accompanied by a corresponding accumulation of Pax7-positive progenitors [[23] and Fig. 1, A,B]. Here, we report that the hemi-NT facing the transfected mesoderm was also affected, exhibiting a significant reduction in the number of Hb9+ motoneurons when compared to controls that received GFP alone (asterisks in Fig. 1A’,B’, C, p<0.001, N= 12 and in both control and treated embryos). Moreover, a ventral expansion of the Pax7-positive domain was frequently observed (arrows in Fig. 1A,B), likely due to a reduced size of the ventral extent of the transfected hemi-NT (see Fig.2).
These results can be explained by the sclerotome constituting a common pool of Shh acting on both mesodermal and neural progenitors. Alternatively, they might result from Hhip1 moving towards the NT. Indeed, in spite of the initial findings that Hhip1 is a transmembrane glycoprotein with cell autonomous functions [18, 30], it has recently been reported that Hhip1 is a secreted Shh antagonist able to exert long range effects on Shh signaling [31, 32]. This is compatible with the possibility that sclerotomal Hhip1 crosses into the NT or at least localizes to its surrounding basement membrane [31].
Differential behavior of secreted Hhip1 compared to membrane-tethered Hhip1:CD4 vis-à-vis Shh
To discriminate between the above possibilities, we produced a Hhip:CD4 plasmid encoding for a membrane-tethered version of Hhip1, unable to undergo secretion [31, 32]. First, we asked whether NT or sclerotomal cells missexpressing either Hhip1 or Hhip:CD4 are able to sequester Shh. To this end, we implemented a plasmid encoding the N-terminus of Shh fused in frame to YFP (ShhN:YFP). ShhN:YFP is able to undergo palmitoylation but not addition of a cholesterol moiety, a property that enables free movement of the mutant protein when compared to native Shh [33]. When electroporated to the sclerotome, secreted ShhN:YFP was apparent along the basement membrane of the NT where it co-localized with laminin, yet no fluorescent signal was detected along the neuroepithelial cells (Supplem. Fig.S2, A-C, arrows).
When double electroporations of ShhN:YFP to the NT and Hhip:CD4 to sclerotome or vice-versa were performed, the Hhip:CD4-transfected sclerotome or NT progenitors, respectively, were decorated with ShhN:YFP, demonstrating that Hhip:CD4 binds and immobilizes the ligand in the expressing cells (Fig. 1D-D”,F-F”). In contrast, no such co-localization could be observed in either sclerotome or NT upon double electroporation of ShhN:YFP and Hhip1 (Fig. 1, E-E”, G-G”) consistent with the notion that native Hhip1 is a secreted protein.
In addition, we monitored the expression of endogenous Shh protein upon transfection of Hhip1 or Hhip:CD4 to the sclerotome. Shh immunoreactive protein was evident both intracellularly as well as associated with the cell membranes of the FP and No likely exposed to their external surface (Supplem. Fig. S2, D-F’). Electroporation of control GFP and of Hhip:CD4 had no effect on Shh immunoreactive protein in either the No or FP (Supplem. Fig.S2, D-E’). In contrast, missexpression of Hhip1 markedly reduced Shh levels in both structures unilaterally adjacent to the transfected cells; this reduction was mainly apparent at their basal sides closer to the source of Hhip1 (Supplem. Fig.S2, arrows in F’). This effect may result from Hhip1 masking antibody binding to the ligand as the 5E1 Shh antibody and Hhip1 bind to the same pseudo-active site on the Shh molecule [30, 34]. Together, the above data confirm that Hhip1 is secreted to adsorb Shh far from the producing cells, whereas Hhip:CD4 acts cell-autonomously.
The effects of Hhip:CD4 in the NT resemble those observed with other Shh inhibitors
Next, we employed the NT to examine the specificity of Hhip:CD4 relative to other, known inhibitors of the Shh pathway. Electroporation of Hhip:CD4, like that of Hhip1, Ptc1 or PTCΔloop2 to hemi-NTs, significantly reduced the number of Hb9+ motoneurons and that of pH3+ mitotic nuclei while enhancing cellular apoptosis. Furthermore, the total area of the transfected hemi-NTs, that reflects overall changes in both proliferation and survival, was significantly smaller in all treatments when compared to control GFP (Supplem. Fig. S3, N=4 for all treatments, *p<0.05, **p<0.03, ***p<0.01). The observed effects on motoneuron numbers could thus result from reduced cell differentiation or, indirectly, from effects on progenitor proliferation or survival. These data confirm that Shh acts both as a mitogen and survival factor in the NT [35, 36]. Most importantly, they demonstrate that Hhip:CD4, which acts like Hhip1, Ptc1 or PTCΔloop2, is a specific tool to abrogate Shh activity.
Loss of Shh activity in either NT or sclerotome inhibits cell differentiation in the adjacent tissue
Local depletion of Shh activity by Hhip:CD4 or Ptc1 in sclerotome inhibits motoneuron differentiation in the NT
Next, we addressed the question whether the effects originally observed across tissues (e.g; between NT and mesoderm) with secreted Hhip1 (Fig.1), can be mimicked by missexpression of two different Shh inhibitors, Hhip:CD4 and the Shh receptor Ptc1, both membrane associated molecules.
Electroporation of Hhip1 to sclerotome caused significant effects in the NT adjacent to the transfected mesoderm, as expected from a secreted molecule. These included a reduction in the number of Hb9+ motoneurons (Fig. 2, N=8, p<0.01), a decreased number of pH3+ mitotic nuclei (N=4, p<0.03), enhanced apoptosis (N=4) and an overall decrease in the area of the respective hemi-NT (N=4, p<0.03) when compared to controls (N=5, Fig.2 B,B’,G,L compared to A,A’,F and K, and see quantification in P).
Notably, transfection of both Hhip:CD4 and Ptc1 significantly affected motoneuron numbers (N=4, p<0.03 and N=7, p<0.01, respectively, Fig. 2C,C’D,D’,P), yet had a mild but non statistically significant effect on cell proliferation (N=4 and 7), survival or total hemi-NT area (N=4 and 6, respectively) opposite the treated sclerotomes (Fig.2 H,M,I,N,P). Thus, inhibition of Shh activity in mesoderm mainly affects motoneuron differentiation, contrasting with Shh abrogation in the NT where all parameters were considerably compromised (Supplem. Fig.S3). The finding that motoneuron differentiation is more sensitive to a reduced amount of ligand, indicates that progenitor proliferation, survival and motoneuron differentiation are separable processes that depend upon different Shh concentrations.
As a control for Ptc1 activity, we electroporated PTCΔloop2 that is unable to bind circulating Shh and acts cell autonomously to inhibit its signaling. PTCΔloop2, like Hhip1, Hhip:CD4 and Ptc1 adversely affected the size of the electroporated sclerotomes when compared to the intact contralateral ones (Fig.2Q, N=5, p<0.01), altogether demonstrating that Shh signaling is necessary for proliferation and/or survival of sclerotomal progenitors. In contrast, PTCΔloop2 had no significant effect on either proliferation, survival or total area of adjacent NT cells (Fig.2E,E’,J,O,P). As expected, unlike Ptc1, PTCΔloop2 had no effect on motoneurons (Fig.2E,E’,P) suggesting that reduced sclerotomal mass is not sufficient to account for the observed loss of motoneurons.
It is worth mentioning that Hhip1 also caused a slight ventralization of Pax7 expression (arrows in B) in association with a reduced size of transfected hemi-NTs. This effect was less apparent or not evident upon electroporation of Hhip:CD4, Ptc1 or PTCΔloop2, which did not significantly affect hemi-NT size (Fig.2, C,D and E).
The specificity of Hhip:CD4 was further tested by examining Olig2, an earlier marker of motoneuron progenitors. The expression domain of Olig2 was reduced adjacent to the electroporated sclerotome (Supplem. Fig. S4). This indicates that depletion of Shh in sclerotome also affects ongoing specification of motoneuron progenitors, a process that already begun by the time electroporations were performed (Supplem. Fig.S1).
In addition, while sclerotomal missexpression of Hhip:CD4 significantly affected the number of Hb9+ cells, Shh alone moderately but not significantly (p=0.056), enhanced their differentiation, and co-treatment of Hhip:CD4 with Shh rescued the effect of Hhip:CD4 back to control levels (Fig. 3, N=12, 26, 8 and 7, for control GFP, Hhip:CD4, Hhip:CD4+Shh and Shh alone, respectively, p<0.001).
Local depletion of Shh activity by Hhip:CD4 or Ptc1 in NT inhibits myotome differentiation in mesoderm
Next, we examined whether depletion of Shh in the NT influences myotomal size. Electroporation of control GFP to hemi-NTs had no effect of myotome size (N=14) whereas Hhip1 and Hhip:CD4 caused a significant decrease in the size of the desmin+ myotomes adjacent to the transfected side of the neuroepithelium (N=3,10, respectively, p<0.001, Fig. 4, A-C, F). Electroporation of Ptc1 exhibited a similar effect (N=16, p<0.001); in contrast, PTCΔloop2 revealed no reduction in myotome size (N=6, Fig. 4D,E,F).
Together, our data suggest that depletion of Shh in the sclerotome promotes a concomitant loss of effective Shh in the NT and vice-versa. Since Hhip:CD4 and Ptc1 are membrane-bound and not secreted, the present results could be explained by the existence of a common pool that supplies Shh to both tissues.
The effects of Hhip:CD4 are a direct consequence of Shh depletion
The similarity between the effects of Hhip:CD4 and Ptc1 (Fig. 2) and the rescue of motoneuron numbers by co-electroporation of Shh along with Hhip:CD4 (Fig. 3) suggests that the effects of Hhip:CD4 are specifically mediated by ligand depletion. To further investigate the possibility of direct versus indirect effects, we examined the expression of Gli1 and Hhip1, two transcriptional targets of Shh and compared it to Gli3 mRNA expression, which is not directly regulated by Shh [37]. Electroporation of control GFP or of Hhip:CD4 to sclerotome or NT had no effect on expression of Gli3 mRNA in the same or adjacent tissue (Supplem. Fig. S5, A-C). In contrast, similar transfections of Hhip:CD4 to NT or sclerotome reduced Gli1 and Hhip1 mRNAs in both the transfected and in adjacent tissues when compared to the respective contralateral sides (Supplem. Fig. S5, D-I). These results further support the notion that the observed effects on motoneuron and myotome development specifically and directly result from ligand depletion.
Next, we examined the possibility of secondary effects across tissues mediated by Shh depletion. Two candidates are the BMP pathway which acts antagonistically to Shh [1, 38, 39] and retinoic acid from the somite that affects NT development [40, 41]. To this end, control GFP or Hhip:CD4 were electroporated into the sclerotome. If the effects on the NT of Shh deprivation in sclerotome are mediated by BMP, it is predicted that the extent of BMP signaling is expanded. No such change of pSmad 1,5,8 immunoreactivity, a readout of BMP activity, was measured upon Hhip:CD4 treatment (Supplem. Fig.S6, N=4 per treatment). As an internal control for Hhip:CD4 activity, we observed a reduced number of Hb9+ motoneurons in the ventral NT adjacent to the transfected side (Supplem. Fig.S6B, arrow).
In addition, control GFP or Hhip:CD4 were electroporated into the sclerotome or NT and RARE-AP, a specific reporter of retinoic acid activity was transfected into the adjacent tissue, respectively. The specificity of RARE-AP was first tested by co-electroporating it with a dominant negative receptor plasmid, that abolished RARE-AP signal (Supplem. Fig. S7A-B”). No apparent changes in RARE-AP were observed in either tissue upon depletion of Shh in the neighboring one when compared to control GFP (N= 4 for each treatment, Supplem. Fig. S7, C-F).
Taken together, these results show that the observed effects are a direct consequence of ligand depletion. Thus, inhibition of Shh in the sclerotome by Hip:CD4 or Ptc1 results in a corresponding reduction of Shh ligand in the NT and vice-versa.
Gain of Shh function in sclerotome enhances motoneuron differentiation but Shh missexpression in NT has no effect on myotome
Our loss of function results would be consistent with the possibility that Shh can translocate bidirectionally between mesoderm and NT. To gain additional insight into the directionality of Shh effects across NT and mesoderm, we adopted a complementary gain of function approach. Control GFP or full length Shh were electroporated into sclerotome. A day later, an increase of approximately 40% in the number of Hb9+ motoneurons was monitored in the NT of Shh-treated embryos ipsilateral to the treated side when compared to controls (p<0.001, N=9 for each treatment, Fig. 5A-C). In addition, desmin+ myotomes were enlarged at the expense of Pax+ DM progenitors (Fig. 5B).
Reciprocally, control GFP or full length Shh were electroporated into hemi-NTs and the size of adjacent desmin+ myotomes was examined. Although the overall size of the transfected hemi-NT increased, no significant change in myotomal size was measured and no apparent effects on DM or sclerotome were observed (N=6 and 5 in control and treated embryos, respectively, Fig. 5D-F).
Hence, our gain of function data are inconsistent with the simpler possibility that Shh moves bidirectionally between mesoderm and NT, that is based solely on data from loss of Shh activity. While excess Shh in mesoderm profoundly affects NT development, the observation that gain of Shh in NT has no effect on myotome development is in line with Shh being transported into the neuroepithelium but not outside into mesoderm. In this regard, our finding that Shh depletion in the NT inhibits myotome differentiation (Fig. 4) is consistent with this procedure causing a corresponding lower effective concentration in mesoderm. This could be accounted for by enhanced uptake of the ligand into NT cells via directional baso-apical transport from the sclerotome. This suggests that the sclerotome constitutes a common pool of No-derived Shh that serves both tissues.
Sclerotome, but not FP-derived Shh is necessary for motoneuron differentiation
Classical data support the view that the development of various cell types in the NT depends upon a local ventrodorsal gradient of Shh emanating from the No and FP [42, 43]. Our present results, show that No-derived Shh secreted into the sclerotome is also necessary for motoneuron development (Figs. 4,5). Hence, we examined the relative contribution of the FP compared to sclerotomal Shh to the differentiation of Hb9+ neurons.
Control GFP or Hhip:CD4 were electroporated dorsoventrally to attain the FP of the NT. Control GFP had no effect on FP integrity or on expression of Shh protein (N= 7, Fig.6A). Surprisingly, electroporation of Hhip:CD4 caused the total disintegration of the FP and consequent loss of FP-derived Shh (N= 10, Fig. 6, D,E); an effect that enabled us to accurately monitor the contribution of the FP to motoneuron development. Because the effect is bilateral, in order to prevent variability between embryos and obtain a reliable measurement of motoneuron numbers, the proportion of Hb9+ neurons was measured as the ratio between the neurons present at the flank level lacking a FP to the intact neck level of the same embryos. In spite of FP disappearance, the proportion of flank-level motoneurons was unaltered when compared to control embryos that received GFP only (N=4 for each treatment, Fig. 6, B,E,H). Similar to what was observed with Hhip:CD4, dorsoventral electroporation of Ptc1 also compromised FP integrity while having no apparent effect on motoneurons (N=5, Fig.6G). In contrast, inhibition of Shh in the sclerotome by Hhip:CD4 exhibited a visible reduction in ventral motoneurons (Fig. 6, C,F arrow, and see also Figs. 2 and 3).
Sclerotomal Shh is not likely to act by affecting ligand levels in the producing cells because inhibition of Shh by Hhip:CD4 in sclerotome had no effect on Shh expression in either FP or No (Supplem. Fig. 2, E,E’). Thus, we conclude that Shh traversing the sclerotome plays a significant part in motoneuron differentiation whereas the FP itself has no apparent contribution at least at the stages examined in our study (see Discussion).
A basal, but not apical, presentation of Shh is required for ligand activity on both NT and DM/myotome
Based on our finding that Shh transiting through the sclerotome is needed for NT development, we predicted that the NT would be more sensitive to Shh presented from its basal side abutting the sclerotome than from its apical side. To examine this hypothesis, fragments of No were grafted into either the lumen of the NT (apical) or between somites and NT (basal grafts). A day later, the expression of Pax7 in the dorsal NT was not affected by the luminal grafts yet was dorsally restricted when facing the basal grafts (Fig. 7A,B). The latter also caused a characteristic bending of the NT, previously reported to represent an ectopic FP-like structure [44](N= 6 and 6 for apical vs. basal grafts, Fig. 7, B,D). In addition, the amount of Hb9+ motoneurons was unchanged by the apical grafts, yet was markedly increased in the basal grafts in which the No’s were similarly localized in a dorsal position vis-vis the NT (N= 6 and 6 for apical vs. basal grafts, Fig. 7, C,D), in agreement with classical No graft experiments [45].
Likewise, implantation of No fragments at epithelial somite levels and at an apical position with respect to the DM had only a mild effect on the subsequent development of desmin-positive myotomes with no apparent alteration in Pax7 expression in the DM. In striking contrast, equivalent grafts performed basal to the prospective DM, produced large myotomes expressing desmin and a radical in situ differentiation of the DM into muscle at the expense of Pax7-positive progenitors (N= 5 and 7 for apical vs. basal grafts, Fig. 7, E,F). Therefore, an initial basal presentation of the ligand vis-à-vis the target epithelium is required for the activity of Shh. This is consistent with our results showing that Shh emanating from the sclerotome and reaching the NT from its basal side, is necessary and sufficient for aspects of NT differentiation.
Discussion
In the present study we uncover a previously unknown domain, the sclerotome, as being an important “en passant” pool of Shh that influences not only DM and myotome development, as previously shown [23] but also aspects of NT differentiation (Fig. 7G). Our loss of function data show that local depletion of Shh in either NT or sclerotome results in major defects across tissues such as reduced myotomal size and less motoneurons, respectively. Reciprocally, only gain of Shh function in the sclerotome significantly enhances motoneuron differentiation while missexpression of the ligand in NT has no effect on myotomal size. Taken together, these results suggest that there is a common pool of Shh ligand present in sclerotome that supplies both tissues.
In contrast to the reduction in motoneurons observed upon inhibition of Shh in sclerotome, we show that Hhip:CD4-mediated ablation of the FP has no short-term effects on motoneuron numbers. This is consistent with the development of a normal ventral pattern in Gli2 mutants that lack a FP [46, 47]. Likewise, loss of Shh in FP did not dramatically alter ventral neural patterning, yet altered gliogenesis at a later stage [48]. Furthermore, abrogating Shh in FP under the regulation of Brn4, revealed a normal short-term expression of both Nkx2.2 and Olig2, but a reduction at later stages suggesting a continuous need for FP-derived Shh in development of the ventral NT [24]. This initial phenotype is consistent with our results yet we did not analyze later stages. Moreover, it is possible that in the absence of a FP, the No which although separated from the NT is still closely adjacent to it, compensates for the loss of FP-derived Shh as suggested for Gli2 mutants [reviewed in [49]]. In comparison, abrogation of Shh ligand in mesoderm, performed at the same stage and for a similar duration, revealed a significant NT phenotype even in the presence of both Shh-producing axial structures. Taken together, we propose that a major fraction of Shh operating on the neuroepithelium stems from the sclerotome which, at the stages examined, seems more active that the FP in promoting motoneuron development.
Furthermore, how can Shh from sclerotome be more important than Shh from FP, given that the ligand is readily detected in FP but not in mesoderm by immunostaining? Shh in the synthesizing cells of the FP and No is intracellular and membrane-bound [[46] and Supplem. Fig. S2)] whereas in the sclerotome it is expected to be extracellular and/or included in organelles (e.g, exosomes, etc). Most protocols used for tissue processing are likely to keep only the former type of immunoreactive protein and wash away the extracellular ligand in sclerotome. Notably, when using a method that allowed proteoglycan/glycosaminoglycan preservation and/or perhaps also a different antibody, a previous study showed a sclerotomal localization of Shh immunoreactive protein [50].
It is noteworthy that different procedures that perturb the production of Shh protein or its signaling have different effects on FP integrity. Whereas Gli2 mutants lack a FP [51], deletion of Shh in FP does not appear to be necessary for maintenance of this structure [24]. In both chick and mouse, Shh was suggested to be necessary for initial FP induction but later on, during somitogenesis, the FP becomes refractory to the ligand [52]. In our experiments, we implemented a membrane tethered version of the high affinity Shh inhibitor Hhip and also the Shh receptor Ptc1, both resulting in the death of FP cells. This might be accounted for by a combination of ligand depletion with accumulation of Shh-Hhip or Shh-Ptc1 complexes at the cell membrane altogether compromising the structural integrity of this epithelium.
One open question stemming from our results is how is Shh transported through the sclerotome. Possible models could be packaging of the ligand in No-derived exosomes [53], diffusion of Shh released by matrix metalloproteinases in a lipid-free form [54]; secretion of Shh as multimeric complexes of various molecular compositions [see for example [55]] and/or via carrier-mediated transport through the extracellular space [56]. The precise mechanism responsible for Shh transport in the present context remains to be unraveled.
If provided to the NT from the sclerotomal domain, it is inferred that neurepithelial cells sense Shh from their basal pole that faces the mesoderm. Consistent with this notion, grafting No fragments in a basal, but not apical position with respect to the NT, profoundly affects NT shape, motoneuron differentiation and Pax7 expression. Indeed, during normal development, the No underlies the basal domain of the NT, further supporting the idea that No-derived Shh can only reach the NT via a basal route. In line with the above, it was reported that lipidated Shh enters the cells of the imaginal disc in Drosophila only through its basolateral surface [57]. A similar phenomenon was observed in high density human gastruloids, that self organize into an epithelium. In these cultures, cells were responsive to BMP4 or Activin ligands only when presented from the basal side and this correlated with the localization of BMP receptors at the basolateral domain of the cells [58]. Although it is unclear if Shh receptors are also localized at the basal aspect of neuroepithelial cells, or if Shh needs first to be transported baso-apically in order to act on apically localized receptors, this and our results suggest that cell polarization controls ligand response.
The above findings are interesting in light of the proposed concept that apically-localized cilia serve as antennae to sense and transduce a Shh signal [59, 60]. Based on our data, we suggest that, initially, Shh is presented from the basal side of epithelial cells from which it may be transported to the apical domain where cilia are localized. Since an apical presentation of the No and associated Shh was without a significant effect in our implant experiments, we propose that cilia act primarily as transducers of a Shh signal, but not as the primary antennae sensing the presence of the ligand. Moreover, the observation that a direct apical presentation of Shh is without effect, further suggests that basal reception followed by baso-apical transport might be necessary for the activity of Shh arriving at the cilia. Along this line, growing evidence suggests cilia-independent Shh reception that occurs through basally localized cytonemes [reviewed in [61] and refs therein]. Likewise, in the retina neuroepithelium, Shh and its coreceptor Cdo colocalize at the basal side of the cells where filopodia-like structures are present [62]. An extreme example is provided by some cell types where Shh signaling takes place even in the absence of cilia [reviewed in [61] and refs. therein].
In the NT, direct visualization of the behavior of labeled Shh, revealed that the ligand from the No concentrates in association with the apically localized basal bodies from which cilia stem, while forming a dynamic gradient in the ventral NT [46]. In light of the present results, the above observed graded distribution of ligand could be explained as being the end point of a transport process that begins at the No, travels through the sclerotome forming there a ventro-dorsal gradient, then reaches the neuroepithelium through its basal side finally concentrating in the apical cilia (Fig. 7G). In such a model, the mechanisms mediating the baso-apical transport of the ligand associated with possible biochemical changes of the protein that make it availabile to cilia for signaling remain to be unraveled. A microtubular network spanning the extent of neuroepithelial cells could be involved in this process, as previously suggested [46].
An initial presentation of Shh from the basal side of an epithelium seems to be of general significance as basal grafting of a No also elicited robust in situ differentiation of DM progenitors into myocytes when compared to an apical implant that exhibited only a mild phenotype. How can this differential effect be explained given that the endogenous ligand apparently arrives from the apical sclerotomal direction?. Initially, pioneer myotomal cells in the epithelial somite face the No from their basal aspect [63]. With ongoing development and upon sclerotome dissociation, the epithelial DM is consolidated and becomes composed of a central sheet and four inwardly curved lips pointing towards the sclerotome. In previous studies, we and others showed that the four lips of the DM are the main sources of myotomal cells [[64] and refs therein] and it is their basal domain that points towards the sclerotome from which Shh arrives. Next, myotomal progenitors enter the nascent myotome and a basement membrane begins assembling at the interface between myotome and sclerotome [65]. At this time, the central DM sheet also contributes myotomal progenitors by direct translocation into the myotome [66], which then differentiate in a Shh-dependent manner [23]. These precursors were shown to exhibit an inverse apicobasal polarity when compared to the central DM cells [66]. Hence, both the DM lips and these translocating progenitors point with their basal surfaces towards the source of endogenous Shh, likely being the main targets for its activity and also accounting for the partial effect of the apical grafts.
As discussed above, the basal domain of the DM/myotome and NT epithelia are characterized by the presence of a surrounding basement membrane. An association between Shh and the basal lamina has been shown. For instance, in cerebellar granule cell precursors, the laminin-containing basement membrane binds and locally enhances Shh signaling [67]. Similarly, in the mouse somite, Shh induces the activation of Myf5 in DM. Myf5+ cells then translocate to the myotome and upregulate α6β1 integrin and dystroglycan, allowing a myotomal basement membrane to be assembled using primarily laminin α1 produced by sclerotomal cells, and laminin α5 produced by the dorsomedial lip of the DM [reviewed in [65]]. This is further confirmed in Shh-deficient mice, which fail to form a myotomal basement membrane, and in which myotomal cells do not properly exit the cell cycle, maintain Pax3 expression and delay the differentiation program [68, 69]. Additionally, Shh immunoreactive protein was found to localize in the basement membrane surrounding the NT [50]. Taken together, these results suggest that in both NT and DM/myotome, a feedforward mechanism may exist whereby Shh controls laminin expression and the assembly of a basement membrane. This could allow a local concentration of the ligand and/or signal stabilization. However, the basement membrane alone is unlikely to serve as the common pool, as much of the ligand is present in the mesoderm itself (our data and see also [50]).
Our present findings provide an additional argument in support of the importance of NT-somite interactions which are pivotal for the normal patterning of trunk components. For instance, opposite gradients of retinoic acid and Fgf8 in mesoderm are required for early NT development [40]. In addition, the nascent DM controls the timing of neural crest delamination by modulating noggin mRNA and BMP activity levels in the NT [70]. Reciprocally, Bmp4 and/or Wnt1 from the dorsal NT pattern the somite-derived DM and affect myogenesis [71, 72]. Formation of the dorsal dermis from the DM is influenced by NT-derived Wnt1 and by neurotrophin-3 [71, 73, 74]. Furthermore, interactions between neural and somitic cells control neural crest migration and segmentation of peripheral ganglia and nerves as well as specific aspects of myogenesis [75, 76]. The present results raise the intriguing possibility that the dual activity on both motoneurons and myotome of Shh released into sclerotome, serves to couple and coordinate development of the neuromuscular system.
Materials and Methods
Embryos
Chick (Gallus gallus) and quail (Coturnix japonica) eggs were from commercial sources (Moshav Orot and Moshav Mata, respectively).
Expression vectors and electroporation
Expression vectors were: pCAGGS-GFP, pCAGGS-RFP [77], Ptc1 [78], PTCΔloop2 [23, 78], a retinoic acid reporter fused to alkaline phosphatase (pRARE-AP, from J. Sen) [79], a dominant negative pan-retinoic acid receptor (RAR403dn-IRES-GFP, from S. Sockanathan) that abrogates retinoic acid signaling [80], full length Shh [23], cholesterol deficient Shh (mShh-N-YFP, from V. Wallace) [33] that was subcloned into pCAGGS, and Hhip1 [23]. To produce Hhip1:CD4, a membrane-tethered version of Hhip1, the transmembrane and intracellular domains of mouse CD4 were fused to the C-terminal domain of Hhip1 lacking amino acids A679-V700, as previously described [31, 32] and further subcloned into pCAGGS for electroporation.
For electroporations, DNA (1-4 µg/µl) was microinjected into the center of flank-level epithelial somites (somites 20-25) of 23-25 somite-stage embryos. Electroporations were performed to the ventral half of epithelial somites (prospective sclerotome). To this end, the positive tungsten electrode was placed under the blastoderm in a location corresponding to the ventro-medial portion of epithelial somites on a length of about 7 segments, and the negative electrode was placed in a superficial, dorso-lateral position with respect to the same somites. [23, 66, 81, 82]. For hemi-NT electroporations, DNA was microinjected into the lumen of the NT. One tungsten electrode was placed underneath the blastoderm on one side of the embryo and the other electrode was placed in a superficial position at the contralateral side. For FP electroporations, the positive electrode was inserted under the blastoderm near the midline and the negative electrode was placed over the dorsal NT. In some cases double electroporations to the hemi NT and sclerotome were performed sequentially. A square wave electroporator (ECM 830, BTX, Inc.) was used. One pulse x12V for 5msec was applied.
No grafts
No fragments comprising a length of 7-8 segments were enzymatically excised from donor embryos aged 25 somite pairs as previously described [35] and kept in cold phosphate buffered saline until grafting. For grafting at the apical side of the NT, the ectoderm and dorsal NT of host embryos were cut and the No piece placed in the NT lumen. A day following implantation, the grafts were usually found in the dorsal portion of the NT facing its cavity. For basal grafting with respect to the NT, a slit was performed between somites and NT and the No fragments were inserted. A similar, but more profound slit, was performed to reach the ventral sclerotome abutting the apical side of the DM. To reach the basal domain of the DM, the ectoderm was cut to precisely accommodate the length of the No fragment. Following microsurgery, embryos were reincubated for additional 24 hours.
Immunohistochemistry
Embryos were fixed overnight at 4°C with 4% formaldehyde in phosphate-buffered saline (PBS) (pH 7.4) followed by washings in PBS. Most immunostainings except for desmin were performed on whole embryo fragments. Immunolabeling for desmin was performed on tissue sections, as described [83, 84].
For wholemount immunostaining, antibodies were diluted in PBS containing 1% Triton X-100 and 5% newborn calf serum and tissues were incubated overnight at 4°C on a rotatory shaker. Next, they were washed twice in a large volume of PBS/1% Triton X-100 first for 10 min and then for 2 hours at room temperature. Secondary antibodies were similarly diluted in PBS/ 1% Triton X-100/ 5% newborn calf serum and incubated overnight followed by repetitive washings. Embryo fragments were dehydrated in increasing ethanol solutions (30%,70%, 90% and 100%, 10 minutes each) followed by toluene (2 times, 10 minutes each), then embedded in paraffin wax and sectioned at 8μm. Paraffin was removed in Xylene and slides were rehydrated in decreasing ethanol solutions.
The following antibodies were used: rabbit anti GFP (1:2000, Invitrogen, Thermo Fisher Scientific, A6455) and mouse anti-desmin (1:200, Molecular Probes, 10519). Monoclonal antibodies against Pax7 (PAX7-s, 1:20), Shh (5E1, 1:20) and Hb9 (1:200) were from DSHB, University of Iowa). Phosphorylated Smad 1-5-8 (PSmad, 1:1000) was a gift from Ed Laufer. Anti-Histone H3 (phospho S10) was from Abcam (mAbcam 14955, 1:500). Detection of DNA fragmentation was done by TUNEL (ab66110, Abcam) according to manufacturer’s instructions. Nuclei were visualized with Hoechst.
In situ hybridization
Embryos were fixed in Fornoy (60% ethanol, 30% formaldehyde, 10% acetic acid), then dehydrated in ethanol/toluene, processed for paraffin wax embedding and sectioned at 10 μm. Slides were rehydrated in toluene/ethanol/PBS, treated with proteinase K (1µg/ml, Sigma Aldrich P2308) at 37°C for 7 minutes, and then fixed in 4% formaldehyde at room temperature for 20 minutes. Next, slides were washed in PBS followed by 2X SSC and hybridized in hybridization buffer (1X salt solution composed of 2M NaCl, 0.12M Tris, 0.04M NaH2PO42H2O, 0.05M Na2HPO4, 0.05M EDTA, pH7.5], 50% formamide, 10% dextran sulfate, 1mg/ml Yeast RNA, 1X Denhardt solution) containing 1μg/ml DIG labeled RNA probes (prepared with a DIG RNA labeling mix, Roche, 11277073910) for overnight at 65°C in a humid chamber. Post-hybridization, slides were rinsed in a rotating incubator with 50% formamide, 1X SSC, 0.1% Tween 20, until coverslips dropped and then an additional wash for 1 hour followed by 2 washes in MABT (10% Maleic acid 1M pH 7.5, 3% NaCl 5M, 0.1% Tween 20) and preincubation in MABT/ 2.5% FCS. Anti-DIG-AP antibody (1/1000, Roche 11093274910) diluted in MABT+2% BBR+20% FCS was then added for overnight at room temperature. This was followed by rinsing in MABT and then in NTMT (2% NaCl 5M, 10% Tris HCl 1M pH9.5, 5% MgCl2 1M, 0.1% Tween20), and then incubation in NTMT + 1:200 NBT/BCIP Stock Solution (Sigma-Aldrich, 11681451001) at 37°C until the AP reaction was completed.
The following probes were employed: Hhip1 (from J. Briscoe) , Nkx2.2, Nkx 6.1, Nkx 6.2, and Olig2 (from J. Ericson), and Gli1, Gli3 from A.G. Borycki.
Data analysis and statistics
Four to 26 embryos were analyzed per experimental treatment. The number of Hb9-positive motoneurons or phospho-histone H3 (pH3) + nuclei was counted in 5-10 alternate sections per embryo.
Myotomes were defined by desmin staining. The area occupied by desmin+ myotomes was measured in alternate sections of 3 to 16 embryos per experimental treatment. Sclerotomes of 4 to 6 embryos were defined in the mediolateral aspect as the tissue between myotome and NT, and in the dorsoventral extent as the mesenchyme between the dorsomedial lip of the DM up to the dorsal border of the cardinal vein and aorta. The surface area of hemi-NTs was monitored in 4 sections per embryo. Myotomal, sclerotomal and hemi-NT areas were measured using Image J software (NIH) and expressed as the mean ratio between treated and control sides±SEM. All results are expressed as the mean proportion of positive cells or area in treated compared to control contralateral sides±SEM.
Images were photographed using a DP73 (Olympus) cooled CCD digital camera mounted on a BX51 microscope (Olympus) with Uplan FL-N 20x/0.5 and 40x/0.75 dry objectives (Olympus). For figure preparation, images were exported into Photoshop CS6 (Adobe). If necessary, the levels of brightness and contrast were adjusted to the entire image and images were cropped without color correction adjustments or γ adjustments. Final figures were prepared using Photoshop CS6.
Significance of results was determined using the non-parametric Mann–Whitney test. All tests applied were two-tailed, and a P-value of 0.05 or less was considered statistically significant. Data were analyzed using the IBM SPSS software version 25. The number of embryos analyzed for each treatment (N) is detailed in the Results Section. P-values can be found both in the Results Section and in the corresponding Legends.
Author Contributions
NK and CK conceived the project and designed the experiments; NK performed the experiments; CK and NK wrote the paper.
Declaration of Interests
The authors declare no competing interests.
Acknowledgements
We thank Mordechai Applebaum and Dina Rekler for help with figure preparation and Tali Bdolach for assistance with statistics. We thank James Briscoe and Johan Ericson for helpful comments during the course of this study, and Joel Yisraeli for critical reading of the manuscript. This work was supported by the Israel Science Foundation (#97/13) to CK.
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.↵