Prior to the formation of cutaneous appendages, the embryonic skin is composed of two homogeneous layers, the epidermis and the dermis. In the chick dorsal region, the dermis originates from different regions of the dermomyotome at between 3 and 5 days of incubation (stages HH 20 to 26) (Hamburger and Hamilton,1992), in response to dorsal Wnt signals from the neural tube and the ectoderm (see among others: Olivera-Martinez et al.,2002; Ben-Yair et al.,2003). A densification of the dermal fibroblasts occurs by day 6 (stage HH 29) throughout the area of the future dorsal feather field (spinal pterylae) and by day 7 (stage HH 30) in the other body pterylae (Sengel,1976). A lesser degree of densification occurs later in the inter-pterylae regions, but never in the midventral apterium that results from ventral closure of the body and can be equated to a scar (Dhouailly et al.,2004). Dense dermis formation and maturation are concomitant with expression of the twist-like bHLH transcription factor cDermo-1 in the subectodermal mesenchyme, until the dermal condensation formation (Scaal et al.,2001). The overexpression of cDermo-1 leads to the formation of an ectopic feather tract (Hornik et al.,2005). When the dermis is mature, an initial, ubiquitous and permissive dermal signal (“make an appendage”; Dhouailly,1977), leads to the formation of thickenings of the epidermis, called placodes. The placodes result, in chick embryo, from an elongation of the apico-basal axis of the epidermal cells and a modification of E-cadherin expression (Jamora et al.,2003). The ubiquitous signal from the homogeneous dense dermis has been shown to belong to the Wnt family. The most striking demonstration of this is with the forced expression of β-catenin in mouse or chick epidermis, which leads to the formation of new hair or feather follicles, respectively (Gat et al.,1998; Widelitz et al.,2000). Likewise, the association of an embryonic mouse dense dermis with an adult rabbit corneal epithelium provokes increased β-catenin protein levels in the basal cells, followed by formation of hair placodes (Pearton et al.,2005). Next, molecular dialogue between the placode and the homogeneous dense dermis leads to the formation of dermal condensations. Each placode with its underlying dermal condensation makes up a primordium, the first stage of cutaneous appendage morphogenesis. In the Scaleless chick mutant, which does not form placodes and has no scales and only a few scarce feathers, the dense dermis forms as usual in each pterylae (Viallet et al.,1998; Widelitz et al.,2000; Harris et al.,2004), but the epidermis is defective in signalling (Sengel and Abbott,1963; Dhouailly and Sawyer,1984; Viallet et al.,1998). The missing signal from the epidermis can be replaced by FGF-2 (Song et al.,1996; Viallet et al.,1998). In addition, numerous other signalling molecules are expressed in feather primordia, among them: different Wnts (Chang et al.,2004); cAMP, PKC (Noveen et al.,1995); TGFβ2 (Ting-Berreth and Chuong,1996a); Shh (Ting-Berreth and Chuong,1996b); FGF-10 (Mandler and Neubuser,2004); Ectodysplasin (Houghton et al.,2005); and BMPs (Jung et al.,1998). They have been proposed to be classed as activators (FGF-4, Shh, cAMP, TGFβ2) or inhibitors (BMPs, PKC activators) of feather formation (Jung et al.,1998). The interplay between these diffusible factors (Jiang et al.,2004) appears to create a Turing system (Turing,1990) that allows the emergence of a regular pattern of skin heterogeneity formed by inter-follicular skin and follicles. In dorsal chick embryonic skin, feather primordia are arranged in a hexagonal pattern (Sengel,1976), resulting from the formation of successive, alternate rows from the medial line to lateral parts of the embryo.
Until now it has been thought that this pattern, once formed, is stable, even after dissection and organotypic culture of skin. Indeed, the formation of feather buds in cultured dorsal skin has been considered to be a continuous phenomenon, except in the case where the first midline row was missing (Novel,1973). In contrast, our preliminary studies (reported in Olivera-Martinez et al.,2004) showed that when chick embryonic dorsal skin of stage HH 30 (three first rows of primordia) is cultured, the feather primordia are lost within a few hours and are reformed later. In this report, we take advantage of this fact to study the role of different factors in dermal condensation formation and/or stabilization, whose activity can be easily modified by pharmaceutical manipulation of the culture.
For a long time (Wessells,1965), it has been known that no cell proliferation is involved in chick dermal condensation formation in vivo. The redistribution of dorsal dermal fibroblasts is apparent when the dermal cell density of a chick embryo is compared between stages HH29 and HH30 (Fig. 1). At stage HH29 (Fig. 1A,B), the dermis is composed by an upper dense dermis and a lower loose dermis. The placode then forms in the epidermis while there is no cell density modification in the dermis (Fig. 1C). At stage HH30 (Fig. 1D–F), the former dense dermis (2.60 nuclei/1,000 μm3) appears to be redistributed between dermal condensations (3.5 nuclei/1,000 μm3) and inter-follicular dermis (1.2 nuclei/1,000 μm3) (Olivera-Martinez et al.,2001). This hypothesis of dermal cell redistribution is supported by in ovo BrdU injections (Rouzankina et al.,2004), in vitro studies (Desbiens et al.,1991) and experiments involving 3[H]Thymidine incorporation in culture (Jiang and Chuong,1992). Moreover, the late expression of Shh in the placode and of Patched, its receptor, in the dermal condensation, a pathway involved in cell proliferation, only correlates with the outgrowth of the feather bud (Jung et al.,1998; Morgan et al.,1998; McKinnell et al.,2004b).
Immunological studies or in situ hybridization show that Fibronectin (Mauger et al.,1982), β1 Integrin (Jiang and Chuong,1992), and Notch-1 (Chen et al.,1997) are expressed in chick skin dermal condensations and thus might be implicated in their formation and/or stabilization. Integrins play a role in a number of cell functions, not only cell adherence, via outside-in signalling. Integrins can, for example, modulate the intracellular Calcium concentration or the activation of protein kinases (Clark and Brugge,1995), which, in turn, control the expression of genes implicated in proliferation and differentiation. Likewise, cell-matrix interactions can be modulated by a regulation of Integrin affinity through an inside-out signal (Bennett and Vilaire,1979). The switch from a low- to a high-affinity state occurs via a conformational change in the extra-cellular domain induced by the cytoplasmic domain (Takagi et al.,2002). It has previously been shown that phosphorylation of the cytoplasmic domain of Integrins by CaMKII induces a switch from the high- to a low-affinity state, while de-phosphorylation of this domain by Calcineurin has an antagonistic effect. Thus, CaMKII can be considered as a repressor whereas Calcineurin can be considered as an activator of the high-affinity state of the Integrins (Hartfield et al.,1993; Leitinger et al.,2000). Consequently, the balance between CaMKII and Calcineurin (two Calcium-dependent proteins) activity might play a role in cell migration. In addition, the activation of Integrin via passive influx of Calcium in cells by Ionomycin (Liu and Hermann,1978; Kuznetsov et al.,1992) is known to enhance their interaction with Fibronectin (Hartfield et al.,1993; Hyduk et al., 2006). Pharmacologically, the addition of Manganese can switch Integrins from a low- to a high-affinity state (Werfel et al.,1996; Olsen and Ffrench-Constant,2005).
Notch signalling has been suggested to play a role either in the binary choice between the two fates of skin, i.e., appendage vs. inter-appendage epidermis, or in the stabilization of primordia (Weinmaster et al.,1991; Viallet et al.,1998; Favier et al.,2000). After fixation of its ligand, the Notch receptor is cleaved by γ-secretase (Pan et al.,2004) and releases the Notch intracellular domain (NICD), which functions as a co-transcription factor leading to the expression of specific target genes. In angiogenesis Notch can interact with Integrin (Leong et al.,2002), leading to a switch to the high affinity state. Moreover, the interaction of NICD with β1 Integrin can modify the localization of the NICD as a function of the extracellular matrix composition (Leitinger and Hogg,2002). A recent study (Campos et al.,2006) has presented more details on the interaction of Notch/β1 Integrin using a neural stem cell model. This interaction apparently occurs only in lipid rafts, which could be involved in a positive regulation of Integrins.
Another class of proteins that has been shown to control a large variety of biological responses that involve cell movement, such as chemotaxis and cytokinesis, is the Rho family GTPases (Etienne-Manneville and Hall,2002). Rho has an impact in specific rearrangements of actin cytoskeleton, an essential process for cell migration (Raftopoulou and Hall,2004). This family of proteins can transduce signal from surface receptors to cytoskeleton, the modification of which is dependent upon the member activated (Nobes and Hall,1995; Hall,1998). Moreover, the Rho-GTPase, RhoB, is expressed in the embryonic chick skin placode and is implicated in the morphological changes that are required for its formation (McKinnell et al.,2004a). Conversely, the inhibition of the Rho family GTPases dependent kinase, ROCK (Maekawa et al.,1999), leads to an absence of placodes.
The aim of this work was to analyse the functions of Rho family GTPases, Integrins, and Notch signalling in dermal condensation formation and/or stabilization using our in vitro organotypic culture system. Although our results were obtained in vitro, we suggest that these factors might also function in the switch that occurs during chick embryonic development that leads to the formation of dermal condensations from a homogeneous dense dermis.
Impact of Organotypic Skin Culture on Pre-Existing Feather Primordia
Chick skin pieces, corresponding to the entire dorsal spinal pterylae, were dissected from stage HH-30 embryos and cultured at the air/liquid interface. Chick dorsal skin had three complete rows of feather primordia when initially placed in culture. In order to follow what happens during the initial time period, we cultured the explants and fixed them every 3 hr, to a total of 24 hr, to perform histology and in situ hybridization. At time zero (Fig. 2A), well-defined primordia consisting of a placode and a dermal condensation are present. The inter-follicular region consists of a loose dermis underlying a flat epidermis. After 6 hr, the primordia are less defined (Fig. 2B). The dermal fibroblasts cannot have proliferated, and the apparent increase in density results from a global contraction of skin components, including the extracellular matrix, during the first hours of culture. After 12 hr, the skin is homogeneous as the primordia have disappeared (Fig. 2C). After 18 hr in culture, dermal condensations reappear (Fig. 2D). This lability of embryonic skin organization leads to a temporary loss of the periodic pattern. The disappearance and reappearance of feather primordia can be followed via time-lapse video (see Supplementary Video, which can be viewed at www.interscience.wiley.com/jpages/1058-8388/suppmat), which shows a dispersal of the fibroblasts of the dermal condensation. To confirm the scattering of fibroblasts to the inter-follicular region, DiI was microinjected into dermal condensations (Fig. 2E). The skin pieces were then placed on culture grids in order to follow the behaviour of labelled cells via time-lapse videography: some of the cells that dispersed from feather primordia (Fig. 2F) during the dermal disorganization phase are found in the inter-bud region after reformation of the condensation.
To analyze the impact of the culture method on the expression of factors that are known to participate to feather primordia formation, we performed in situ hybridization to detect the presence of BMP-2 (which is found in both dermal condensation and placode), FGF-10 (in dermal condensation), and Wnt-7a (in placode) messages. At time zero, the three rows of feather primordia express these three markers (Fig. 3A–C). Their expression is lost concurrently with the disorganization of the primordia, Wnt-7a expression is lowered, but not completely lost, after only 6 hr of culture, whilst expressions of BMP-2 and FGF-10 disappear after 8 and 9 hr, respectively (Fig. 3D–F). The disorganization of the primordia, which leads to the loss of the periodic skin pattern, is thus correlated with a down regulation of Wnt-7a expression (number of cases (n) = 6/7) and the loss of BMP-2 (n = 68/73) and FGF-10 (n = 6/8) expressions in the placodes, dermal condensations, or both.
After the disorganization phase, which is complete by 10 hr of culture, Wnt-7a, BMP-2, and FGF-10 begin to be re-expressed in a pattern resembling that of time 0. The skin starts to reform defined feather primordia (Fig. 3G–I) (n = 5/5, 37/43, 7/8, respectively). The patterning of the skin explant is not yet completed by 18 hr of culture and can show some irregularities (Fig. 3J–L) (n = 6/6, 28/31, 4/5), but is complete after 24 hr (data not shown). Via time-lapse video (see Supplementary Video), it is clear that the new pattern is established de novo, as it does not coincide with the original.
As migration is strongly suspected to play a major role in condensation formation and in the dermis reorganization, we then studied the migratory properties of dermal fibroblasts in different culture conditions, as dissociated cells in adherent or hanging drop cultures and in organotypic cultures. As cell migration can be altered via the action of Rho family GTPases on the actin cytoskeleton, or Integrins interacting with Fibronectin, we investigated the effect of different factors known to interact with these systems.
Dermal Fibroblast Migration Abilities in Adherent Cell Culture
We used a wound-healing assay with freshly dissociated dermal cells to analyze the behaviour of dermal fibroblasts in adherent cell culture. At time zero, 750-μm wounds were made in a confluent primary dermal cell culture (Fig. 4A) in which cell proliferation was inhibited by Thymidine treatment. In the control condition, cells had migrated and covered the wounds, only a narrow space (around 75 μm) remaining cell-free by 20 hr after wounding (Fig. 4B) (n = 5/5).
It was previously shown (McKinnell et al.,2004a) via in situ hybridization that RhoA is not expressed in HH-30 embryonic chick skin, and RhoB only in the placode. Using a RT-PCR (Fig. 4C) on separated epidermis and dermis, we confirm that RhoA expression is absent in both the epidermis and the dermis. However, we detect the RhoB expression in both skin components. By comparing the ratio of RhoB/actin in each tissue, we show that this expression is 4 times higher in the epidermis than in the dermis. RhoC is also expressed in both compartments. The quality of our tissue dissociation, and hence purity of the cell samples, is attested by the fact FGF10 expression is found only in the dissected dermis.
Having shown dermal expression of both RhoB and RhoC, we then added Y-27632, a specific inhibitor of the Rho-associated protein kinase (ROCK) (Uehata et al.,1997) in order to perturb cell migration. In the wound-healing assay, ROCK inhibition significantly delays cell migration, a 300-μm gap remains cell free (Fig. 4D) (n = 5/5). This delay is even more drastic with activation of Integrins by activation of Calcium signalling via addition of Ionomycin after wounding, which slows down cell migration even further, the gap being around 700 μm (Fig. 4E) (n = 5/5). Another way to activate Integrins is to add Manganese in the culture medium. This non-specific activation also leads to an arrest of dermal cell migration in the wound-healing assay (data not shown) (n = 4/5).
In order to distinguish an eventual cell spreading from a true migration in the wound healing assay, we have quantified the effect of these molecules on cell migration through a membrane in various conditions (Fig. 5). This assay was done with or without 10% of FBS, as an attracting factor, to emphasize cell migration. Each measure was compared to the highest migration that occurs under control condition with 10% FBS. Without this attracting factor, cell migration is decreased by 71%. In the presence of FBS, the inhibition of Rho-associated protein kinase decreases the number of migrating cells by 39%. This decrease is even more notable with the addition of Mn2+ (−56%), with the strongest effect occurring with the activation of Calcium signalling via Ionomycin (−71%).
Dermal Fibroblast Aggregation in Hanging Drop Culture
In order to mimic the formation of dermal condensations in vitro by dispersed fibroblasts, we used the micromass formation method (Stott and Chuong,2000) with 25,000 cells in a 10-μl hanging drop, to follow cell aggregation. Two steps can be distinguished. First, aggregations of cells to form a condensation, where cells outline are clearly distinct. Then, a second step, called the compaction phase, where the aggregate condenses so that the individual cell outlines are no longer distinguishable. After 24 hr, untreated dermal fibroblasts have undergone cell aggregation (Fig. 6A) (n = 251/275), whereas ROCK inhibition leads to the formation of a loose cell network (Fig. 6B) (n = 239/275), which supports a role for Rho family GTPases in cell aggregation. Modification of Integrin regulation by Ionomycin activation of Calcium signalling leads to a more rapid aggregation than in the control with compaction occurring after only 24 hr (Fig. 6C) (n = 237/275) instead of 48 hr. Surprisingly, non-specific activation of Integrins in the high-affinity state by Manganese leads to a defect in aggregation of dermal cells (data not shown) (n = 112/150). To investigate the role of Fibronectin, one of the possible Integrin ligands in skin, we added a Fibronectin-blocking antibody B3/D6 to the culture medium (Harrisson et al.,1993). As expected, this prevented a normal fibroblast aggregation. A low cell density aggregation is formed, many cells remaining outside of this aggregation (Fig. 6D) (n = 87/100).
Thus, formation of dermal fibroblast aggregation in vitro appears to rely on dermal fibroblast aggregation via Integrins on Fibronectin, as well as on ROCK activation. Is this also the case for the formation of dermal condensations in vivo? In order to study this, we used short-term organotypic skin culture.
ROCK Inhibition Delays the Dermal Cell Spreading and Leads to a Disorganized Pattern in Organotypic Culture
As the Rho-associated protein kinase may be implicated in the fibroblast migration in the dermal condensation, we used its inhibitor (Y-27632) in organotypic culture to analyze its effect while the dermis is undergoing reorganization. At time zero, the skin fragment displays three rows of feather primordia (Fig. 7A). After 8 hr of culture with inhibition of ROCK, there is no loss of BMP-2 expression (Fig. 7B) (n = 8/10), which occurs in the control condition (Fig. 3E). After 12 hr of culture, there is a loss of BMP-2 expression in the lateral rows while the medial row still expresses BMP-2 (Fig. 7C) (n = 6/9). After 24 hr of culture, primordia are present all over the skin piece, but their pattern is irregular (Fig. 7D) (n = 8/9). After 48 hr, primordia have heterogeneous sizes, shapes, and location throughout the skin fragment (Fig. 7E) (n = 9/9). This pattern perturbation is striking when it is compared to the control (Fig. 7F) (n = 5/5).
Integrin and Fibronectin Involvement in Dermal Condensation Formation in Organotypic Culture
Fibronectin expression is high in dermal condensations and in the upper layer of dense dermis, prior to dermal condensation formation (Fig. 8A,B). β1 Integrin expression was previously shown to be increased in dermal condensations (Jiang and Chuong,1992), potentially implicating this Fibronectin receptor in this formation. To confirm its role, we cultured skin with the Fibronectin-blocking antibody (B3/D6). We added the B3/D6 antibody after the disorganization step (8 hr of culture) to analyze only the action of Fibronectin in dermal condensation formation. After 12 hr of culture, whereas dermal cells start to aggregate in control condition (Fig. 3H), after the addition of the B3/D6 antibody no dermal condensations can be seen (Fig. 8C) (n = 5/7). After 18 hr, dermal condensations are present, although they are not well defined in the caudal part of the skin piece (Fig. 8D) (n = 3/4). Moreover, dermal condensation formation is delayed compared to the control (Fig. 3K). This is confirmed by the fact that after 48 hr with the B3/D6 antibody, some condensations are present, but their outlines and pattern are not well-defined (Fig. 8E) (n = 7/9), while in the control (TROMA III antibody), feather primordia had formed and even begin to grow up to the bud stage (Fig. 8F) (n = 9/9).
To determine the role of Integrins in dermal condensation lability, we activated Integrins to the high-affinity state by the addition of Manganese to the culture medium. At time zero, the skin piece has three rows of feather primordia (Fig. 9A). After 8 hr in control culture conditions, the expression of primordium specific factors is lost (Fig. 3E). In contrast, when Integrins are maintained in the high-affinity state with the use of Mn2+ the dermal condensations are stabilized and do not disappear (Fig. 9B) (n = 7/9). Another method used to test the proposed function of Integrins in the stabilization of dermal condensations was to activate Calcium signalling in dermal cells with Ionomycin. We performed 8-hr organotypic cultures with Ionomycin. Moreover, to focus on the role of Calcineurin, a Calcium-dependent phosphatase, we added Cyclosporin A, a Calcineurin inhibitor (Fruman et al.,1992), to certain of the cultures. As with other experiments, at time zero the skin explant had three rows of feather primordia. The activation of Calcium signalling with Ionomycin leads to stabilizations of BMP-2 expression and feather primordia (Fig. 9C) (n = 37/43), whereas in the control condition both BMP-2 expression and feather primordia are lost (Fig. 3E). The activation of Calcium signalling via Ionomycin with the concurrent inhibition of Calcineurin (Calcium-dependent phosphatase) (Fig. 9D) (n = 19/27) leads to a lowering of BMP-2 expression and none of the stabilization of the feather primordia observed with the Ionomycin treatment only.
To confirm the effect of Ionomycin on feather primordium development, we cut skin pieces along the medial row, the left half being cultured in control condition and the right half with Ionomycin. After 48 hr of culture, the control skin piece had round, almost flat feather primordia (Fig. 9E) (n = 6/7), whereas in the presence of Ionomycin, the feather primordia had progressed to elevated feather buds oriented in the caudal direction (Fig. 9F) (n = 5/7). Thus, in Ionomycin-treated skin fragments, the stabilization of dermal condensation allows the continued outgrowth of feather primordia, while in the control the feather primordia formation resets to the beginning, resulting in a delay of their outgrowth.
Preventing Notch Signalling Leads to an Alteration of Patterning in Organotypic Culture
To assess the role of Notch signalling in the patterning of feather primordia, we inhibited γ-secretase, which is required for Notch activation, by addition of L-685,458 (van Es et al.,2005) in organotypic culture medium. At time zero, chick embryo dorsal skin exhibits three rows of primordia (Fig. 10A). After 18 hr, that is, at the end of the reorganization phase, rectangular primordia have formed, which, in the medial part of the skin explant, are the result of lateral fusions of prospective round primordia, or in the lateral part, of fusions of primordia from the fifth to seventh rows on the rostro-caudal axis (Fig. 10B) (n = 11/15). After 48 hr, the fused primordia gave rise to giant feather buds (data not shown). Thus, the absence of Notch signalling does not inhibit primordia formation, but instead leads to a fluidity/lability of their boundaries. Consequently, the Notch system is not in control of the initial events in feather morphogenesis, i.e., the formation of the placode, then of the dermal condensation, but rather it is involved in boundary formation and maintenance.
We have developed a model to study the initial steps of cutaneous appendage primordium formation in organotypic culture. The loss and the reinitiation of differentiation in vitro show that this process is autonomous as it occurs in skin dissected from the embryo, thus isolated from any influences of the underlying organs. This phenomenon allow us to study the impact of different factors by modifying the culture medium, and it could be also useful to establish the order of expression of different genes. Altogether, our results, although obtained in vitro, strongly suggest that feather primordium in vivo morphogenesis might also be due to cell rearrangement and migration rather than differential proliferation.
Wnt-7a has been implicated in the growth of cutaneous appendages as it is expressed throughout the feather placode and then in the posterior half of the feather bud epidermis (Chuong et al.,1996; Widelitz et al.,1999). Wnt-7a is also uniformly expressed where the future placode will form, before the thickening of the epidermis, as shown for the chick scale development (Prin et al.,2004). In our present work, Wnt-7a expression is the first of those factors analyzed to decrease. It might, therefore, be involved in the determination of the placode itself as well as later its proximodistal orientation and growth. Epidermal cells of the placode express factors that promote dermal condensation formation (Cadi et al.,1983; Dhouailly and Sawyer,1984; Chuong et al.,1996). These factors may have direct or indirect target genes, some of them being expressed in dense dermis, then in the dermal condensation, and they work together to create the microenvironment of the primordium. The decrease of placodal gene expression in our culture of skin fragments appears to be succeeded by the loss of gene expression in dermal condensation. Thus, BMP-2, which is expressed both in placode and dermal condensation (Jung et al.,1998), and then FGF-10 (known to promote cell proliferation; Tao et al.,2002) in the dermal condensation, disappear, apparently sequentially. Nevertheless, this proposed sequence needs to be confirmed by other studies.
After the loss of this microenvironment, the isolated and cultured dorsal skin redifferentiates to form feather primordia autonomously and de novo. Reformation of dermal condensations that contain cells from both the previous dermal condensations and the inter-follicular domain, shows the plasticity of dermal cells at stage HH30. This result is supported by a separate previous experiment of Dr. Chuong's laboratory (Jiang et al.,1999) in which dermal cells were completely dissociated followed by re-association, with the result that cells that had previously been found in the dermal condensation could be found in the interfollicular dermis after reformation of the condensations. Moreover, this new pattern can show some irregularities even in the control conditions. However, striking differences occur only when the dermal cell migration or the stabilization is disturbed (see below).
A previous work (McKinnell et al.,2004a) showed that there is no RhoA or RhoB expression in the dermis, while RhoB is expressed in the epidermis, and that the use of a specific ROCK inhibitor prevents the formation of the placode. The implication of RhoB in the maintenance of the morphogenetic program leading to feather bud development is thus well demonstrated. Our RT-PCR results show that, firstly, RhoB is expressed at a low level in the dermis, maybe not enough to be detected by in situ hybridization in comparison to its strong expression in the placode; and, secondly, that RhoC is expressed in both the dermis and the epidermis. Specific inhibition of ROCK, thus, not only alters the formation of the placode, but also can have an impact on fibroblast migration in cell culture. The disrupting of primordia formation thus could be due to a combination of effects on both the epidermis and the dermis. The dense dermis organization, thus, could require fibroblast migration via the action of Rho family GTPases on the actin cytoskeleton.
This migration also involves the dynamic regulation of Integrin affinity. In cell migration assays, as well as in micromass formation and in organotypic culture, modifications of Integrin affinity, as of Rho family GTPases activity, lead to perturbations of dermal fibroblast properties, i.e., their capacities for migration and aggregation. Integrins, thus, play an important role in dermal condensation formation. Positive regulation of Integrins by intracellular Calcium accelerates micromass formation, even though switching to the high-affinity state via Mn2+ activation leads to a defect of cell aggregation.
In the cell migration assay, we showed that the activation of Calcium signalling induced a decrease in cell migration; this result was confirmed in the organotypic culture, but not in the micromass formation. These apparently opposing effects might arise from the fact that in organotypic culture, we started with formed dermal condensations, while in the micromass method, we induced this formation from a homogeneous cell suspension.
When Fibronectin is blocked in organotypic culture, dermal condensations are more diffuse than in control conditions. Our results are in accordance with a previous study (Jiang and Chuong,1992), which has shown that the inhibition of the interaction between β1-Integrin and Fibronectin leads to defects of dermal condensation formation. So, cells might use this Integrin ligand to migrate during dermal condensation formation. With activation of Integrins by intracellular Calcium, or with Manganese, the lability of the dermal condensations is not observed. The preformed rows do not disappear and the development continues without the microenvironment being lost. Thus, the maintenance of cell adhesion is a key factor for the maintenance of expression of the different genes, such as BMP-2, Wnt-7a, and FGF-10, that are involved in cutaneous appendage formation. The activation of Integrins plays a direct or indirect signalling role during activation of these genes, as well as for the maintenance of their expression. Contradictory results are obtained by the activation of Integrins via Calcium or Manganese. In the micromass formation, the Calcium activates this formation whereas the Manganese inhibits it. This difference might come from the mechanism of Integrin activation: the Manganese, by its direct fixation on the MIDAS domain (Leitinger et al.,2000), switches Integrins in a high-affinity state, whereas activation of calcium signalling modifies the regulatory pathway of Integrin affinity. In the organotypic culture, we obtained the same result with Calcium and with Manganese. In both methods, the complete switch in the high-affinity state of Integrins fixed the initial situation (dispersed cells for the micromass formation and maintenance of dermal condensations in organotypic culture). The Integrin activation leads to the maintenance of the feather primordium.
A recent study using conditional knockout mice showed that deletion of Notch-1 in the epidermis leads to abnormal hair development (Vauclair et al.,2005). In our culture model, Notch signalling is inhibited in both epidermal and dermal cells. Due to the contact-dependent, trans-acting nature of the Notch ligands/receptor complex, the cellular reorganization of dense dermis, and consequent loss of cell contact, the Notch signal might be lost in the first hours of culture. The inhibition of this signal, by use of a γ-secretase inhibitor, does not inhibit primordium formation, but leads to a perturbation of pattern establishment by fusion of some primordia. Fusion of primordia might reflect a dynamic requirement of Notch signalling for the maintenance of the dermal condensation. The most common hypothesis of the role of the Notch system in boundary formation and in segregation of one cell type from another has been previously described in other models such as axial skeletal formation (Bulman et al.,2000) and chondrogenesis (Fujimaki et al.,2006). Here we show that Notch signalling appears to be implicated not in the initial choice between the two fates of embryonic skin, but in a later step to allow stabilization of the newly formed pattern. To go further, a new hypothesis could be developed in regard to the interaction between the NICD and β1 Integrin. Indeed, to enhance the efficiency of Integrins signalling (inside-out and outside-in), there could be a clustering of β1 Integrin and Notch-1 to facilitate the interaction of the NICD and β1 Integrin. A study on neural stem cells has proposed a model of differentiation based on free concentration of the NICD (Campos et al.,2006). They have shown an interaction between the NICD and the intracellular tail of β1 Integrin and have proposed that this interaction could modify NICD localization as a function of the extracellular matrix composition. With Fibronectin, there is strong interaction between the NICD and β1 Integrin, resulting in a low pool of free NICD. This would prompt cells to migrate and/or to commit to differentiation. So, interaction of Notch and Integrins might lead to inside-out and outside-in signalling via modifications of Integrin affinity, and intracytoplasmic concentration of the NICD. This could be a good hypothesis to explain many aspects of dermal condensation formation. On the dermal condensation domain, characterized be a high amount of Fibronectin, there could be a strong NICD/β1 Integrin interaction, leading to a switch of Integrins to a high-affinity state, necessary for dermal condensation stabilization, with low nuclear NICD, which promotes dermal cell differentiation.
In conclusion, Integrins might play at least two important roles in dermal condensation formation: firstly, in the migration of dermal fibroblasts to the appendage domain via engagement of Integrins on Fibronectin, and, secondly, in their stabilization via interaction with Notch intracellular domain. Furthermore, the stability of the feather dermal condensation boundaries allows, via cell adhesion and outside-in signalling, cell differentiation by sending signals required for at least the maintenance, and perhaps even the activation, of the expression of several genes known to be essential for cutaneous appendage formation. Although there are many studies showing similarities of feather formation between in vivo and organotypic culture (Novel,1973; Sengel,1976; Jung et al.,1998), our results, having been obtained in vitro, should be generalized with care, and confirmed with new in vivo experiments.
Chick Embryos and Materials
Fertilized chicken eggs (JA957 strain, SFPA, St Marcellin, France) were incubated at 38°C until the embryos reached Hamburger Hamilton stage 30 (corresponding to 3 rows of feather primordia) (Hamburger and Hamilton,1992). Ionomycin, γ-secretase inhibitor (L-685,458), ROCK Inhibitor (Y-27632), and Cyclosporin A were purchased from VWR International (France).
Seven-day chick embryo (stage HH 30) dorsal skins were dissected as a single piece starting from the wing to the femoral level, or cut in two symmetrical fragments. They were placed on a liquid-permeable plastic sheet (cellophane) in organ culture dishes (Dominique Dutcher, France) at the air-liquid interface and cultured with DMEM (Gibco) containing 10% FBS and 0.2% Gentamycin (Sigma-Aldrich) for the times indicated in the figure legends.
The B3/D6 hybridoma (monoclonal antibody specific for avian fibronectin) and the TROMA-III hybridoma (monoclonal antibody specific for rat cytokeratin 19, used as an isotype control for B3/D6), developed, respectively, by D.M. Fambrough, W. Halfter, and R. Kemler, were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. Primary antibody was detected with a secondary antibody directed against the appropriate species labelled with Alexa Fluor-546 (Molecular Probes, Eugene, OR). Nuclei were routinely counterstained with DAPI (Sigma-Aldrich).
Skin explants were fixed in ethanol 95%/acetic acid 5%, and embedded in paraffin. Sections (7 μm) were stained with Hematoxylin/Scarlett Biebrich.
Microinjections of 0.2 to 1 μl of DiI (2 μg/ml in DMSO) (Sigma-Aldrich) were performed in dermal condensations at the femoral level in the medial row of HH 30 chick embryo dorsal skin with a Transjector 5246 (Eppendorf). DNA was stained with Hoechst 33342 (Sigma-Aldrich) to identify high nuclei density regions such as dermal condensations. Explants were then cultured for the times indicated in the figure legend.
In Situ Hybridization
The chick FGF-10 probe was cloned via PCR into the pGEM-T Easy vectors (Promega, France) (based on its published sequence; Ohuchi et al.,1997). The chick Wnt-7a probe was a gift from Dr. C. Logan (U.K.). The chick BMP-2 probe was a gift from Dr. A.-H. Monsoro-Burq (Monsoro-Burq et al.,1996). Alkaline phosphatase-labelled in situ hybridizations were carried out as previously described (Wilkinson and Nieto,1993).
Dorsal skin (HH30) pieces were separated into dermis and epidermis after 5 min of Trypsin 2.5%/Pancreatin 4% (Sigma-Aldrich) treatment. Tissues were mechanically dissociated into single cell suspension. RNAs were isolated with the High Pure RNA tissue kit (Roche). Reverse transcription of mRNA was performed with the SuperScript First-Strand synthesis system (Invitrogen, La Jolla, CA). Polymerase Chain reaction was done with the Taq Polymerase (Promega, Madison, WI). Reactions have migrated in a 1% agarose gel and pictures were analyzed with the ImageJ software (NIH). Primers used, cActine sense: AGACCTTCAACACCCCAGC; cActine antisense: TGATTTTCATTGTGCTAGG; cFGF10 sense: GTTTCAGTACAATGTGCAAATGG; cFGF10 antisense: AGTTCCTTCATCTATGACATTAC; cRhoA sense: TCAATCGATAGTCCTGATAGT; cRhoA antisense: TATAAGAGAAGGCACCCGGAC; cRhoB sense: GCGCAACGACGAGCACGTGCG; cRhoB antisense: TATAGGACCTTGCAGCAATTG; cRhoC sense: ACTACATCGCCGACATTGAGG; cRhoC antisense: TCACAGCAGCGGGCAGCCCCG.
A fresh single cell suspension of dermal fibroblasts was prepared. Cells were cultured at 780 cells/mm2 dish with DMEM, 10% FBS, 0.2% Gentamycin, and 2 mM Thymidine (VWR International) to arrest cell proliferation. Wounds were made with a plastic tip after 12 hr of culture. Pictures were taken 20 hr after wounding.
Cell Migration Assay
A fresh single cell suspension of dermal fibroblasts was prepared. For each condition, 250,000 cells were used in the Innocyte cell migration assay, 24-well plate (VWR International) (Lauffenburger,1996). The migration was measured after 10 hr of migration by the fluorescence due to the Calcein-AM dye. Fetal Bovine Serum was used as an attracting factor for cells.
A fresh single cell suspension of dermal fibroblasts was prepared. Ten microliters of a 2.5 × 106 cells/ml suspension, with DMEM 10% FBS (Fibronectin-depleted), 0.2% Gentamycin, were spotted as hanging-drop cultures. Pictures were taken after 24 hr of culture.
Seven-day chick embryos (stage HH30) were fixed in Paraformaldehyde 4%, 30 min at 4°C. They were cryoprotected with an infiltration of 30% Sucrose overnight at 4°C and then embedded in OCT-compound, sectioned (7 μm), and stored at −80°C.
The authors thank Dr. D.J. Pearton for helpful comments and critical reading of the manuscript, Dr. I. Olivera-Martinez for chick embryo skin photographies, Mrs. B. Peyrusse for the iconography, and Miss P. Betton for technical assistance. F. Michon was the recipient of a doctoral fellowship from the French Ministère de l'Education Nationale et de la Recherche. This work was supported by a grant from the Laboratoire de Biologie Cellulaire Cutanée, Institut de Recherche Pierre Fabre, to Pr. D. Dhouailly.