Regulation of ectodermal Wnt6 expression by the neural tube is transduced by dermomyotomal Wnt11: a mechanism of dermomyotomal lip sustainment

During the development of the avian embryo, the paraxial mesoderm gives rise to segmental epithelial spheres called somites. The ventral part of the somites subsequently becomes mesenchymal to form the sclerotome, which gives rise to the axial skeleton (Christ et al.,2004). The dorsal part of the somite remains epithelial and becomes the dermomyotome, which gives rise to dermis and muscle. In the course of dermomyotomal differentiation, the dorsomedial and ventrolateral lips of the dermomyotome (DML and VLL, respectively) remain epithelial growth zones(Ordahl et al., 2001), whereas the central dermomyotome (CD) de-epithelializes to give rise to the dorsal dermis and subcutis as well as muscle(Gros et al., 2005; Relaix et al., 2005; Ben-Yair and Kalcheim, 2005)(reviewed by Scaal and Christ,2004).

Signals from adjacent tissues play important roles in the patterning of somites along the dorsoventral and mediolateral axes. Commitment of cells in somites occurs after somite formation in response to external cues(Aoyama and Asamoto, 1988; Ordahl and Le Douarin, 1992; Christ et al., 1992). Several lines of evidence indicate that Shh secreted from the notochord and floor plate acts as a ventralizing and medializing signal(Fan and Tessier-Lavigne,1994; Johnson et al.,1994; Chiang et al.,1996; Kos et al.,1998). By contrast, signals from the surface ectoderm and the dorsal neural tube induce the formation and/or maintenance of the dermomyotome(Brand-Saberi et al., 1993; Pourquié et al., 1993; Kuratani et al., 1994; Fan and Tessier-Lavigne, 1994; Spence et al., 1996) (reviewed by Scaal and Christ,2004).

The Wnt family of secreted proteins play many roles during vertebrate development, including cell fate choice, proliferation and survival(Dickinson and McMahon, 1992; Parr and McMahon, 1994; Cadigan and Nusse, 1997; Huelsken and Birchmeier,2001). In somite development, Wnt1 and Wnt3a have been shown to be required for the development of the medial and dorsal regions of the somites,as well as to induce myogenesis (Marcelle et al., 1997; Ikeya and Takada, 1998; Wagner et al.,2000). Moreover, the epithelial structure of the newly formed somite, as well as that of the dermomyotome, are influenced by Wnts, which have been suggested to induce an epithelial morphology via β-catenin(Hinck et al., 1994; Gumbiner, 1996; Linker et al., 2005).

A member of this family, Wnt6, has been cloned in different species, including human, mouse, Drosophila, Xenopus, Amphioxus and chick, and its expression pattern has been described(Cauthen et al., 2001; Janson et al., 2001; Schubert et al., 2001; Itaranta et al., 2002; Schubert et al., 2002; Rodriguez-Niedenführ et al.,2003; Loganathan et al.,2005). In avian embryos, at the level of the segmental plate and the epithelial somite, Wnt6 is expressed in the entire ectoderm overlying the neural tube, the paraxial mesoderm and the LPM. Following somite compartmentalization, which leads to the formation of the dermomyotome and the sclerotome, Wnt6 expression ceases in the medial aspect and becomes restricted to the lateral ectoderm overlying the LPM.

Wnt11 is a member of the non-canonical type of Wnt signaling. Recently, it has been shown to also participate in the canonical Wnt pathway(Tao et al., 2005). In chick,it is expressed in the dorsomedial lip (DML) of matured somites(Marcelle et al., 1997),whereas in zebrafish, Wnt11 is known to be involved in convergence-extension-movements during gastrulation(Heisenberg et al., 2000). The function of Wnt11 in the avian dermomyotome remains unknown.

In this study we examined the regulation of the dynamic pattern of Wnt6 expression in the avian embryonic ectoderm. We show that Wnt6 expression in the surface ectoderm becomes downregulated by Wnt1 and Wnt3a signaling from the neural tube. We present evidence that this inhibiting action of the neural tube is mediated by Wnt11 in the DML, thus providing evidence of a patterning influence of neural tube and paraxial mesoderm on the surface ectoderm. Our results suggest that after the lateralization of Wnt6, Wnt11 takes over the function of Wnt6 to maintain the epithelial state of the medial region of the dermomyotome. We present a model suggesting that Wnt11 in the DML, and Wnt6 in the VLL, maintain the epithelial character of the dermomyotomal margins to enable ongoing growth of the dermomyotomal sheet during epaxial and hypaxial myogenesis.

Fertilized chicken and quail eggs were incubated at 38°C, and the embryos were staged according to Hamburger and Hamilton(Hamburger and Hamilton,1951).

A longitudinal slit was made between neural tube and adjacent somites[between somites 12-17 (somite stages V-XI), i.e. corresponding to a length of 4-6 somites] on one side of HH stage 14 chick embryos. An aluminium foil barrier was inserted into the slit and the embryo reincubated from 12 to 18 hours and then processed for in situ hybridization.

Portions of unilateral halves or the whole neural tube of HH stage 12-14 chick embryos were removed at the level of epithelial somites (I-IV) just prior to their maturation as previously described(Christ et al., 1992). Embryos were reincubated from 12 to 18 hours, and then processed for in situ hybridization.

Whole dermomyotomes or medial dermomyotomes (of somites 10-12, at somite stages V-X) of HH stage 12-13 chick embryos were removed at the level of matured somites. Embryos were reincubated from 12 to 18 hours and then processed for in situ hybridization.

Medial lips of dermomyotomes from HH stage 13-14 chick or quail embryos were grafted to the segmental plate of chick embryos of HH stage 11-12. Embryos were reincubated for 8-10 hours, fixed and processed for in situ hybridization and immunohistochemistry.

Wnt3a- and Wnt1-expressing cells were a gift from Andreas Kispert(Medizinische Hochschule Hannover, Germany). CHO B3 cells expressing Noggin protein and DHFR control CHO cells were kindly provided by Richard Harland(University of California at Berkeley). Cell lines were cultured as described elsewhere (Lamb et al., 1993). Confluent cultures were harvested, cells were washed in phosphate-buffered saline (PBS), pelleted and resuspended in a minimal volume of medium. For cell injection, the ectoderm (at the level of somite I-V of HH stage 13-14 embryos)was punctured with a tungsten needle. With the help of a blunt glass needle, a tunnel was made below the ectoderm and concentrated cell suspensions were locally applied with a micropipette along the length of the tunnel. For some embryos, noggin-expressing cells were injected into the neural tube at the level of the epithelial somites. Embryos were reincubated from 12 to 18 hours,processed for whole-mount in situ hybridization. Control cells showed no effect on target genes expression (not shown).

Wnt11 and dnWnt11 RCAS constructs were kindly provided by Philippa Francis-West (Kings College, London)(Anakwe et al., 2003). The electroporation procedures and equipment were used as described by Scaal et al. (Scaal et al., 2004). Electroporation was performed at the level of epithelial somites (I-IV) of HH stage 12-15 chick embryos. Constructs were co-electroporated with GFP plasmids, the latter electroporated alone were used as a control. Embryos were reincubated from 12 to 16 hours, photographed using a fluorescence microscope to visualize the localization of the plasmid, and then processed for whole-mount in situ hybridization.

Embryos were fixed overnight at 4°C in 4% PFA. The embryos were washed in PBT, dehydrated in methanol and stored at 4°C. Whole-mount in situ hybridization was performed as described by Nieto et al.(Nieto et al., 1996). Selected stained embryos were embedded in 4% agar and sectioned with a vibratome at 50μm.

The following probes were used in this study: chick Wnt11 (1000 bp; Christophe Marcelle, Marseille); Pax1 (1.5 kb insert cloned into pBlusescript II KS; Cecilia Ebensperger, Freiburg); Pax3 (a 1543 bp insert cloned in to pGEM 72f; Marianne Bronner-Fraser, Pasadena); full-length Paraxis clone was a gift from Prof. Eric Olson (Dallas). For chick Wnt6, we used the cloned Wnt6 1500 bp fragment(Rodriguez-Niedenführ et al.,2003), as a template. Sense and antisense riboprobes were labeled with digoxigenin RNA labeling kit as recommended (Boehringer, Mannheim,Germany).

Selected embryos after in situ hybridization were used for immunohistochemistry, fixed overnight in 4% paraformaldehyde (PFA), washed in PBS. Following a brief wash in PBS, embryos were incubated overnight with QCPN(anti-quail antibody, DSHB). Embryos were extensively washed in PBS and then incubated overnight in secondary antibody (Cy3-conjugated goat anti-mouse IgG antibody; Jackson ImmunoResearch, 1:100, in PBS). Subsequently, embryos were washed in PBS and stored in 4% PFA.

In order to investigate the regulation of Wnt6 expression, we reexamined the Wnt6 expression pattern in the ectoderm during somite development. We found that during early stages of somite formation, at the level of the segmental plate and the epithelial somites, the entire mediolateral extent of the ectoderm overlying the neural tube, the paraxial mesoderm and the lateral plate mesoderm is Wnt6 positive(Fig. 1A,B). However, during subsequent somite compartmentalization forming the dermomyotome and the sclerotome, Wnt6 expression disappeared from the medial ectoderm and became restricted to the ectoderm overlying the VLL and the lateral plate mesoderm (Fig. 1A,C,D). This is consistent with earlier observations(Rodriguez-Niedenführ et al.,2003) and indicates that ectodermal Wnt6 is dispensable during the development of the medial somite.

We examined if the loss of Wnt6 expression in the medial ectoderm is due to an inhibitory action of the axial organs. Axial organs are known to be involved in patterning the medial somite, but so far no patterning influence on the ectoderm has been described(Christ et al., 1992; Dietrich et al., 1997; Marcelle et al., 1997; Münsterberg and Lassar,1995; Stern and Hauschka,1995; Stern et al.,1997; Pourquié et al.,1995; Pourquié et al.,1996). To determine the role of axial organs in the regulation of Wnt6 expression, we separated four to six matured somites and the ectoderm overlying the somites from the neural tube by insertion of an aluminium foil barrier at HH stage 14. After a reincubation period of 12 to 18 hours, we observed an overexpression of Wnt6 at the level of the barrier, with the expression domain extending into the medial ectoderm(Fig. 2A,D; n=8). In a different approach, we microsurgically removed the respective half of neural tube at the same location. Loss of the neural tube also results in a lateromedial expansion of Wnt6 expression(Fig. 2B,E; n=6). This suggests that the lack of Wnt6 expression in the ectoderm covering the medial region of the matured somites is due to inhibitory signals from the neural tube.

In order to identify the signals from the neural tube which inhibit Wnt6 expression, we examined the influence of Wnt1 and Wnt3a on Wnt6 expression. Wnt1 and Wnt3a are expressed in the dorsal neural tube and have been shown to be regulators of medial somite patterning(Münsterberg et al.,1995; Marcelle et al.,1997). After injection of Wnt1- or Wnt3a-expressing cells into the subectodermal space dorsal to the somites, Wnt6 expression was totally inhibited at the site of injection (Fig. 2C,F; n=12; Wnt1; data not shown). This strongly suggests that the loss of Wnt6 expression in the medial ectoderm overlying the maturing somites is due to an inhibitory action of Wnt3a and/or Wnt1 from the dorsal neural tube.

We have shown that Wnt1 and Wnt3a are able to inhibit Wnt6expression in the medial surface ectoderm. However, Wnt1 and Wnt3a both are expressed in the dorsal neural tube already during early stages of paraxial mesoderm development when Wnt6 is still expressed in the entire surface ectoderm, including the medial aspect close to the neural tube (Hirsinger et al.,1997; Marcelle et al.,1997; Cauthen et al.,2001). This paradox could be explained by the hypothesis that Wnt1 and Wnt3a do not have a direct effect on Wnt6 but act via an intermediate signal that is not active before the epithelial somite has developed into dermomyotome and sclerotome.

The development of the dermomyotome depends on interactions with the overlying ectoderm (reviewed by Scaal and Christ, 2004). To test if dermomyotomal signals influence Wnt6 expression, we removed the entire dermoymotome, or medial halves of the dermomyotome, and checked for Wnt6 expression. In both experiments, ectodermal Wnt6 expression was upregulated at the site of surgery, and expression extended to the medial surface ectoderm,demonstrating that the dermomyotome exerts an inhibitory action on Wnt6 expression (Fig. 3A-D; n=18).

Marcelle and co-workers (Marcelle et al., 1997) have shown that the differentiation of the dorsomedial lip of the dermomyotome (DML) depends on Wnt1 and Wnt3a signaling from the neural tube. To test if the hypothesized intermediate signal, which inhibits Wnt6 expression in response to Wnts from the neural tube, originates from the DML, we transplanted DMLs from HH13 quail donor embryos to the segmental plate of HH stage 12 chick host embryos. After 12 hours of incubation, Wnt6 expression in the ectoderm covering the site of DML implantation was downregulated. This shows that the DML has an inhibitory influence on Wnt6 expression (Fig. 3F; n=7).

Wnt11 has been described as a marker gene of the DML, and is expressed in the DML in response to Wnt1 and Wnt3a signaling from the neural tube (Marcelle et al., 1997)(this study see Fig. 3H,I),making it an excellent candidate for the hypothesized intermediate Wnt6-inhibitory signal. Indeed, we found that the DML grafts leading to inhibition of Wnt6 in the ectoderm show robust Wnt11 expression(Fig. 3E; n=3). We compared the expression pattern of Wnt11 and Wnt6 during normal development and found that the onset of dermomyotomal Wnt11expression correlates with the downregulation of Wnt6 in the medial ectoderm (Fig. 1). Together,these data are suggestive of an inhibitory action of Wnt11 on Wnt6expression.

To directly test the effect of Wnt11 on Wnt6 expression, we electroporated Wnt11 RCAS into prospective dermomyotomal cells of the epithelial somite as described by Scaal et al. (Scaal et al., 2004). To facilitate localization of the construct, we coelectroporated GFP-pCLGFPA, which has no impact on target gene expression (Scaal et al.,2004) (own results not shown). After a reincubation period of 12-16 hours, we observed a robust upregulation of Wnt11 expression at the site of electroporation (Fig. 4A,B; n=5). We analyzed the electroporated embryos for Wnt6 expression and found that dermomyotomal overexpression of Wnt11 leads to a total loss of Wnt6 expression even in the lateral ectoderm (Fig. 4C-E; n=7). Conversely, we electroporated a dominantnegative Wnt11RCAS-construct, which inhibits endogenous Wnt11 signaling, into the same location. As expected from our previous results, inhibition of Wnt11 signaling lead to a strong upregulation and medial extension of Wnt6 expression in the ectoderm overlying the electroporated dermomyotome(Fig. 4F-H; n=5). Thus, we have shown that Wnt11 is indeed a negative regulator of Wnt6expression, and we provide evidence that dermomyotomal Wnt11 transduces the Wnt6-inhibitory signals from the neural tube to the ectoderm.

Having found that Wnt6 expression is restricted to the lateral ectoderm by the inhibitory action of Wnt11 in the medial dermomyotome, we sought to determine the functional significance of this regulatory process.

Wnt6 is known to be an epithelialization factor during somitogenesis (Schmidt et al.,2004; Linker et al.,2005). However, in the matured somites, it is expressed only in the ectoderm overlying the lateralmost margin of the epithelial dermomyotome:the ventrolateral lip (VLL). The epithelialization factor of the medial dermomyotome is unknown. We hypothesized that after the loss of Wnt6expression in the medial ectoderm, Wnt11 in the DML could act as a local epithelialization factor. To investigate this, we tested the expression of markers of the epithelial dermomyotome, Pax3 and Paraxis,after overexpression of Wnt11 in the somites by electroporation of a Wnt11 RCAS construct. We found an upregulation of Pax3 and Paraxis in the electroporated somites, which showed an abnormal morphology, indicating an excessive epithelialization of the tissue(Fig. 5A-F; n=11). Conversely, Pax1 expression, which is a marker of the mesenchymal sclerotome, was totally absent (Fig. 5G-I; n=8). These results suggest that Wnt11, like Wnt6,is an epithelialization factor acting on the dermomyotome. By restricting Wnt6 expression to the ectoderm overlying the VLL, and by Wnt11 expression restricted to the DML, the persistent epithelial character of the dermomyotome is limited to the marginal lips, which promote ongoing dermomyotomal growth, whereas the central dermomyotomal domains deepithelialize to form the dorsal dermis(Scaal and Christ, 2004). We,thus, propose an experimental model to explain the maintenance of the dermomyotomal lips in contrast to the early dissociation of the central dermomyotome in the mature somite (Fig. 6).

In avian embryos, all cells of the paraxial mesoderm, with the exception of the somitocoele cells, undergo an epithelialization process before they differentiate into definitive tissues. The first epithelial cells appear in the superficial layer of the cranial segmental plate(Christ et al., 1972). Upon signaling from the ectoderm, the bHLH transcription factor Paraxis is expressed in these epithelializing cells, together with the tyrosine kinase receptor EphA4 (Burgess et al.,1996; Sosic et al.,1997; Schmidt et al.,2001). During somite formation, the epithelial cells arrange as spheres to enclose the still mesenchymal somitocoele(Kulesa and Fraser, 2002). Studies of segmental plate explants in culture suggest that the epithelialization step is necessary to synchronize expression of segmentation genes (Maroto et al., 2005). Such a community effect might be a general requirement for coordinated gene expression also in later stages of somite development. Following somite formation, the ventral somite halves de-epithelialize under the influence of signals from the notochord and ventral neural tube (reviewed by Christ et al., 2004). In the dorsal compartment, the somitic epithelium remains intact under the influence of the neural tube and the surface ectoderm, thus forming the dermomyotome(Kenny-Mobbs and Thorogood,1987; Christ et al.,1992; Spence et al.,1996). Molecular studies have revealed that the formation of the medial dermomyotome depends on Wnt1 and Wnt3a signaling from the dorsal neural tube (Dietrich et al., 1997; Fan et al., 1997; Marcelle et al., 1997; Wagner et al., 2000). Accordingly, in mouse, Wnt1/3a knockout mice lose the medial aspect of the dermomyotome (Ikeya and Takada,1998). The lateral dermomyotome is known to depend on signals from the surface ectoderm (Dietrich et al.,1997; Fan and Tessier-Lavigne,1994; Fan et al.,1997). In recent studies, Wnt6 has been identified as an epithelialization factor from the surface ectoderm which is required for dermomyotome formation and maintenance(Schmidt et al., 2004; Linker et al., 2005). Wnt6 maintains the epithelial morphology of dermomyotomal cells by promoting Paraxis expression via Frizzled7 and β-catenin intracellular signaling (Linker et al.,2005). Until embryonic day 3 the dermomyotome is a continuous epithelial sheet of approximately rectangular shape. At either margin, the epithelial cells form lip-like structures in which cells de-epithelialize and emigrate to form muscle (Gros et al.,2004) and endothelium (Wilting et al., 1995). Later on, the central region of the dermomyotome(CD) deepithelializes completely to give rise to dorsally emigrating dermal and subcutaneous precursor cells, and ventrally emigrating proliferative muscle progenitor cells and satellite cells(Gros et al., 2005; Relaix et al., 2005; Ben-Yair and Kalcheim, 2005). By contrast, the DML and VLL persist as two separate epithelial proliferation zones which are required for ongoing mediolateral growth of the dermomyotome and its derivatives. At embryonic day 7, when the entire dermomyotome has developed into definite tissues, the DML and VLL disintegrate(Ordahl et al., 2001; Venters and Ordahl, 2002). The molecular basis for the differential timing of dermomyotomal de-epithelialization in the margins and the CD has remained elusive.

In this study, we present functional data to explain the molecular regulation of the de-epithelialization dynamics of the mature dermomyotome. We show that the expression of Wnt6, which keeps the dermomyotome in an epithelial state (Schmidt et al.,2004; Linker et al.,2005), is downregulated in the medial and central paraxial ectoderm by signals from the neural tube. To extend earlier findings that Wnt1 and Wnt3a from the dorsal neural tube pattern the medial dermomyotome(Dietrich et al., 1997; Fan et al., 1997; Marcelle et al., 1997; Wagner et al., 2000), we show here that these signals also pattern ectodermal gene expression and inhibit Wnt6 expression in the ectoderm overlying the medial and central dermomyotome. Further upstream in this signaling cascade, Wnt1/3a signaling depends on BMP4 activity in the neural tube(Marcelle et al., 1997). In confirmation of our results, we observed that inhibition of BMP4 by overexpression of Noggin in the neural tube also leads to an extension of Wnt6 expression to the medial paraxial ectoderm (data not shown). However, as Wnt1/3a are already expressed in early stages of paraxial mesoderm development, while Wnt6 is still expressed throughout the surface ectoderm, we searched for an intermediate regulator restricting Wnt1/3a action on the somite stages following dermomyotome formation. We found that the inhibition of medial Wnt6 expression depends on the presence of the DML, thus identifying a patterning influence of the paraxial mesoderm on the overlying ectoderm. Marcelle and co-workers(Marcelle et al., 1997) have shown that the DML is demarcated by Wnt11 expression, and that Wnt11 expression depends on Wnt1/3a signaling from the neural tube. We therefore tested if Wnt11 is able to inhibit ectodermal Wnt6 expression, and performed localized overexpression of Wnt11 in the dorsal somites. We found that indeed dermomyotomal Wnt11 inhibits Wnt6 expression in the overlying ectoderm, making it an excellent candidate for the intermediate factor restricting the Wnt6 inhibitory action of Wnt1/3a to later stages of somite development.

Thus, in the mature dermomyotome, the epithelializing impact of Wnt6 is restricted to the VLL, whereas the CD is devoid of epithelializing signals and becomes mesenchymal to form muscle and connective tissue(Gros et al., 2005; Ben-Yair and Kalcheim, 2005). In addition, the cranial and caudal dermomyotomal lips, which give rise to myotomal cells and persist longer than the CD(Ordahl et al., 2001; Gros et al., 2004), are also exposed to prolonged Wnt6 activity as Wnt6 expression is maintained in the ectodermal ridges extending into the intersegmental clefts between the adjacent somites until embryonic day 4(Rodriguez-Niedenführ et al.,2003) (our observation). How, then, is the DML kept epithelial? We found that overexpression of Wnt11 leads to an overexpression of the epithelial markers Pax3 and Paraxis, whereas the marker of sclerotomal fate, Pax1, is absent. Moreover, the affected somites displayed an abnormal morphology and reduced size, indicating an excessive epithelialization including ventral somitic cells(Schmidt et al., 2004), thus preventing sclerotome formation. Thus, we could show that Wnt11 is a mesoderm-intrinsic epithelialization factor that, upon induction from the neural tube, maintains the epithelial state of the DML while restricting Wnt6 expression to the ectoderm overlying the VLL. In the literature,Wnt11 has been described to be involved in convergence extension movements during zebrafish gastrulation (Heisenberg et al., 2000). Here, we present a novel dual role of Wnt11 in downregulation of ectodermal Wnt6 expression and maintenance of the epithelial state of the DML. It remains to be elucidated if Wnt11 is furthermore required for the cell movements during myotomal cell recruitment in the DML. An important aspect of our results is that Wnt11 represents the first known mesodermal epithelialization factor acting on its cells of origin in a paracrine fashion, thus establishing a fundamental difference between the regulation of epithelialization in the DML and the other dermomyotomal regions, which are thought to depend entirely on ectodermal signals.

Taken together, our data lead us to propose a conclusive model explaining both the dynamic expression pattern of Wnt6 and the regulation of the epithelial morphology of the dermomyotome(Fig. 6). Prior to somite compartmentalization, Wnt6 is expressed in the entire surface ectoderm and promotes epithelialization of the early somites viaβ-catenin and Paraxis (Schmidt et al., 2004; Linker et al.,2005). Following dermomyotome formation, Wnt1 and Wnt3a, which are secreted by the dorsal neural tube upon local BMP signaling, induce Wnt11 expression in the DML(Marcelle et al., 1997). Wnt11 inhibits Wnt6 expression in the ectoderm overlying the medial and central dermomyotome, thus restricting Wnt6 expression to the lateral ectoderm overlying the VLL. As both, Wnt6 and Wnt11, are epithelialization factors, the DML (via Wnt11) and the VLL (via Wnt6) remain epithelial and contribute to dermomyotomal growth, whereas cells in the CD de-epithelialize to differentiate. Importantly, the intermediate activity of Wnt11 as transducer of neural tube-derived signaling provides a timing mechanism restricting this process to the fully formed dermomyotome.

We thank Dr Christophe Marcelle for providing the Wnt11 probe, Dr Philippa Francis-West for Wnt11 RCAS and dnWnt11 RCAS constructs, Dr Andreas Kispert for the Wnt3a- and Wnt1-expressing cells, and Dr Richard Harland for CHO cells expressing noggin. We thank Dr Marc Rodriguez-Niedenführ for his excellent suggestions and help provided during this study. We also thank Ulrike Pein for excellent technical assistance. This study was supported by the Deutsche Forschungsgemeinschaft (SFB-592, A1 to B.C. and M.S.) and the European Network of Excellence, MYORES (to B.C. and M.S.).