In higher vertebrates, the paraxial mesoderm is the source of various mesodermal tissues of the body, including the axial skeleton, skeletal muscle, and dorsal dermis. During gastrulation, prospective paraxial mesodermal cells ingress through the cranial primitive streak and form two massive stripes of mesenchymal cells at either side of the neural tube, the segmental plate or presomitic mesoderm (PSM). While at the caudal end of the PSM, cells are continuously added by gastrulation, cells at the cranial end are successively organized into epithelial spheres that bud off from the cranial PSM as somites (Christ and Ordahl, 1995). The epithelial somites mature, dividing into a ventral mesenchymal sclerotome, which gives rise to vertebrae and ribs (Christ et al., 2004), and a dorsal epithelial dermomyotome, which gives rise to skeletal muscle, dermis, and the scapula (Scaal and Christ, 2004). This process is extrinsically regulated by signals produced by structures in the local environment next to the somite. Noggin and Sonic hedgehog from the notochord induce de-epithelialization of the epithelial somite and sclerotomal specification as evidenced by Pax1 and Pax9 expression (Fan and Tessier-Lavigne, 1994; Johnson et al., 1994; Müller et al., 1996; McMahon et al., 1998; reviewed in Christ et al., 2004). Wnt and BMP signals from the neural tube, the ectoderm, and the lateral plate interact to orchestrate dermomyotomal specification and patterning in the dorsal somite (Pourquie et al., 1996; Marcelle et al., 1997; Sosic et al., 1997; Ikeya and Takada, 1998; Wagner et al., 2000; Cheng et al., 2004; reviewed by Scaal and Christ, 2004). Thus, somite patterning and differentiation of somitic cells into various cell lineages and tissues are to a large extent controlled by environmental cues.
The metameric organization of the paraxial mesoderm into somites represents the primary segmentation of the embryonic body (Christ et al., 2000). Even prior to overt segmentation, the PSM displays a metameric pattern corresponding to the prospective somites (Christ et al., 1972; reviewed in Jacobson 1988; Tam and Trainor, 1994). Along the longitudinal body axis, cranial and caudal halves of each somite give rise to segment-specific skeletal elements that differ structurally from one segmental level to the next. For instance, in chick, the caudal sclerotome half of somite 5 up to the cranial sclerotome half of somite 19 give rise to cervical vertebrae that are characterized by Foramina transversaria, whereas caudal sclerotome half of somite 19 up to the cranial half of somite 26 give rise to thoracic vertebrae typically accompanied by ribs (Burke et al., 2000). Likewise, only dermomyotomes of somites 17 to 24 produce cartilage precursor cells that form the scapular blade (Huang et al., 2000). Thus, every somite along the body axis has a specific segmental identity, which reflects the positional information of the cells in that segment (Wolpert 1969, 1996, Burke 2000). The molecular basis of segment-specific positional information is the local combination of Hox-gene expression, which is characteristic of every segment along the craniocaudal body axis, thus creating the segment-specific Hox-code (Kessel and Gruss 1991; reviewed in Kmita and Duboule, 2003).
To realize the developmental programs induced in the dorsal or ventral subdomains by neighboring signaling centers, cells will interpret these signals according to the positional information at their specific segmental level. Morphogenesis of the paraxial mesoderm is, therefore, always the result of non-segment-specific local signaling events interpreted in a segment-specific way according to the positional information of the cells of the mesoderm, thus creating the segmental identity of the developing structures.
The acquisition and maintenance of segmental identity have been investigated by embryological transplantation. Transplantation experiments have shown that segmental identity in the paraxial mesoderm is determined as early as the formation of the unsegmented PSM. PSM or epithelial somites transplanted from thoracic to cervical or lumbosacral levels maintain their original developmental potential and form ectopic ribs (Kieny et al., 1972; Jacob et al., 1975). This indicates that the thoracic sclerotome is determined already in the unsegmented paraxial mesoderm to form vertebrae of thoracic morphology, including ribs. The segmental identity of the paraxial mesoderm is not restricted to the sclerotome derivatives. The dorsal dermomyotome also maintains the segment-specific morphology of the epaxial muscle (Murakami and Nakamura, 1991; Nowicki and Burke, 2000) and back dermis (Mauger, 1972). Furthermore, the scapula-forming potential of the dermomyotome is also determined by positional information (Ehehalt et al., 2004). However, somite-derived muscle precursors can be reprogrammed with respect to muscle pattern after segment level change of somites (Christ et al., 1978). These observations give rise to the question of whether all dermomyotomal cells are subject to the axial Hox-code.
In contrast to their early segment-specific determination, PSM and epithelial cells in the dorsal or ventral somitic compartments are plastic with respect to their differentiative responses to extrinsic cues. Dorsoventral rotation of newly formed somites at homotopic locations, or interchanges of dorsal and ventral somite halves, leads to the development of dorsal and ventral derivatives according to their new position, obviously in response to environmental signals at the host site (Aoyama and Asamoto 1988; Christ et al., 1992).
Although segment-specific positional information has been a conceptual paradigm for decades, it is still unknown whether the prospective dermomyotome and the sclerotome become segmentally specified in the same way. To investigate this, portions of the paraxial mesoderm with thoracic segmental identity were rotated relative to their neighboring signaling centers, and transplanted into the cervical region. Rib or scapula formation was used as a marker for the segmental identity of the sclerotome or the dermomyotome, respectively. We show that cells, irrespective of their original dorsal or ventral somitic compartment, do correctly interpret the environmental signals at their new location, to form structures typical of their original position, indicating that somitic cells are identically programmed with respect to segmental identity independently of their differentiation fate.
To determine if all cells from the paraxial mesoderm at a certain segmental level possess the same positional information, irrespective of their affiliation to the different somitic compartments, heterotopic transplantation experiments were performed. Portions of paraxial mesoderm were transferred from thoracic to cervical levels as described previously (Kieny et al., 1972). In addition to this shift in segmental level, the grafts were rotated around the dorsoventral, mediolateral, and craniocaudal axis as well. While Aoyama and Asamoto (1988) and Christ et al. (1992) have shown that such rotation of PSM and somites I to III (somite staging according to Christ and Ordahl, 1995) leads to reprogramming of transplanted cells according to their new orientation in the host embryo, transplantation to a different level as well as rotation would discriminate between two alternative scenarios. If all somitic cells possess the same positional information, transplanted cells differentiate and undergo morphogenesis according to their original segmental position even in the new environment. Alternatively, if positional information differs between somitic compartments, transplanted cells will not be able to interpret the environmental cues at their new position according to their original segmental identity, and will display inappropriate morphogenesis. The following experiments show that the first scenario is correct.
All Paraxial Mesodermal Cells of the Same Segmental Level Have Identical Positional Information
The differentiation of paraxial mesoderm after heterotopic transplantation was studied. Paraxial mesoderm at prospective thoracic levels from chick donor embryos at HH-stages 12 to 14 was excised from caudal PSM, cranial PSM, and epithelial somites to examine the importance of stage of development of the paraxial mesoderm at transplantation (Fig. 1). Accordingly, the caudal half of the PSM was transplanted from HH-stage 12 embryos, the cranial half of the PSM from HH 13 embryos, and the three most recent somites I–III from HH 14 embryos. In all experiments, the grafts were transferred to the cervical level of HH-stage 12 host embryos to replace 3–4 epithelial somites. Prior to transfer, the grafts were dorsoventrally rotated around the craniocaudal axis, so that the formerly dorsal cells came to lie ventrally and vice versa (n = 12). Moreover, the explants derived from the cranial PSM were additionally rotated mediolaterally (n = 4), and the explants derived from the newly formed somites were additionally rotated craniocaudally (n = 2), so that mesodermal compartments had been reoriented in all three dimensions.
Irrespective of the origin of the transplant from caudal PSM, cranial PSM, or epithelial somites, and irrespective of the axis of graft rotation, the transplanted mesoderm was integrated into the host environment and developed according to its original segmental identity. After a re-incubation period of 6 days, the embryos were stained for skeletal elements with Alcian Blue (method 1). At the site of graft implantation, the cervical vertebrae showed rib-like lateral extensions as well as small cartilaginous structures without articular connection to the vertebral column, identified as ectopic scapulae (Fig. 1). This demonstrated that formerly dorsal mesodermal cells normally destined to form dermomyotomal derivatives like muscle and dermis had been induced to develop into sclerotomal derivatives while maintaining their original thoracic identity, thus giving rise to ribs. Conversely, formerly ventral mesodermal cells normally destined to form sclerotomal derivatives had been induced to develop into dermomyotomal derivatives including the scapula (Huang et al., 2000), also according to their original thoracic position. These results demonstrate that all cells of the paraxial mesoderm at a certain segmental level contain the same positional information, and that in responding to environmental cues, this positional information is correctly interpreted by all cells irrespective of their original somitic compartment.
Heterotopically Transplanted Cells Interpret and Realize Their New Developmental Program According to Their Original Positional Information
To confirm that the skeletal structures formed by heterotopic rotated transplants are formed by reprogrammed cells from a newly positioned somitic compartment, the ventral halves of thoracic epithelial somites 20–22 of HH-stage 13/14 quail donors were replaced with the dorsal halves of four consecutive cervical epithelial somites of HH-stage 11/12 chick hosts (n = 17). Likewise, we performed the inverse experiment and replaced cervical ventral somite halves with thoracic dorsal somite halves (n = 6). After transplantation of ventral thoracic somite halves to dorsal cervical locations, ectopic scapulae formed at the site of implantation, but no ectopic ribs formed (Fig. 2). The ectopic scapulae were of quail origin as shown by anti-quail staining of sections, as were the myotome-derived muscle cells in this segment. Similarly, after transplantation of dorsal thoracic somite halves to ventral cervical locations, ribs formed, but no ectopic scapulae (Fig. 3). In sections stained by anti-quail antibody, the corresponding halves of the vertebrae as well as the ribs were of quail origin.
This demonstrates that grafted cells were reprogrammed from a ventral or dorsal fate into a dorsal or ventral fate, respectively, realizing their original morphogenetic positional information in the new environment.
Positional Information of Heterotopic Transplants of Thoracic Origin as Represented by Hoxc6 Expression Is Maintained at Ectopic Locations
Positional information is conveyed by the Hox code, a specific combination of expression of various Hox genes at specific segmental levels (Kessel and Gruss 1991). Hoxc6 is a marker of thoracic positional information, as its cranial limit of expression demarcates the thoraco-cervical transition (Burke et al., 1995). Hoxc6 was used as a molecular marker to confirm that heterotopic grafts from thoracic to cervical levels conserve their positional identity in the new environment. The same experimental procedure as above was used to transfer thoracic segment precursors to cervical regions using chick donor and hosts. After one day of re-incubation, host embryos were subjected to in situ hybridization using a chicken Hoxc6 probe. Cervical dermomyotomes derived from ventral thoracic somite halves, and cervical sclerotomes derived from dorsal thoracic somite halves, both expressed Hoxc6 according to their original segmental identity. This shows that Hoxc6 expression, most likely together with other Hox genes at more caudal levels, is maintained in heterotopic transplants, providing a molecular basis of the maintenance of positional information by heterotopically transplanted mesodermal tissue.
Segmental identity is the developmental potential of a segment according to its positional information along the craniocaudal body axis. Segmental identity is most evident in the vertebral column, as each vertebra has a specific morphology typical for a certain craniocaudal level, making it unique among the other vertebrae. Nevertheless, all vertebrae are formed by the sclerotomes of somites (Christ et al., 2000) as a result of the same signaling events (Christ et al., 2004). The segment-specific morphology of metameric structures results from the combination of a general tissue-differentiation program with a segment-specific patterning plan, which is called positional information (Wolpert 1969, 1996). Here, we investigated if the same positional information is intrinsic to all cells of a segment, notwithstanding their diverse developmental fates.
These results show that all cells of a segment in the paraxial mesoderm contain the same positional information, leading the cells to differentiate according to the morphogenetic pattern of their original segmental identity even after having been placed under the control of different developmental signaling centers in a new somitic environment. This was demonstrated by transplanting portions of PSM or epithelial somites from thoracic to cervical positions after the transplanted tissue was rotated in dorsoventral, mediolateral or craniocaudal directions. The cells transplanted from ventral to dorsal adopt dorsal fate, but form thoracic dorsal structures and express Hoxc6 in the cervical position. Likewise, cells transplanted from dorsal to ventral adopt ventral fate, but form thoracic ventral structures and express Hoxc6 in the cervical position. Additional rotation in mediolateral or craniocaudal direction did not influence these results. Moreover, cells from the caudal PSM, the cranial PSM, and the three newly formed epithelial somites are equally competent to respond to a new somitic environment according to their original segmental position. This demonstrates that positional information is already present, and fixed, in the earliest paraxial mesodermal cells shortly after gastrulation.
These results are in accord with the earlier findings of Kieny et al. (1972), Jacob et al. (1975), and Ehehalt et al. (2004) who showed that heterotopic transplantation of PSM and somites from thoracic to cervical, or thoracic to lumbal levels, leads to development according to the original segmental identity of the grafts. However, in their experiments the relative position of the mesodermal compartments was maintained, replacing prospective cervical sclerotome by thoracic sclerotome, or prospective lumbal dermomyotome by thoracic dermomyotome. It remained unclear whether all cells of the dorsal compartment, including the prospective muscle cells, possess segmental identity. Moreover, because development of the dermomyotome occurs later than that of the sclerotome, it has been unknown whether ventral cells would, therefore, be delayed in the realization of their positional information at dorsal positions, or if dorsal cells would realize their positional information earlier when positioned ventrally. All cells were able to launch their segment-specific developmental temporal program correctly whenever they receive the appropriate environmental signals.
These results contrast with those of Kant and Goldstein (1999) who found plastic segmental identity when occipital somites were transplanted to trunk levels. They found that head-specific positional information is lost because the occipital somites formed vertebrae, without expressing trunk-specific Hox genes. This may be due to the differences in the signaling mechanisms acting on head mesoderm that specify the fate of occipital somite cells (Mootosamy and Dietrich, 2002).
Earlier work has shown that dorsoventral (Aoyama and Asamoto, 1988, Christ et al., 1992) and mediolateral (Ordahl and Le Douarin, 1992) rotation of PSM or newly formed somites leads to normal development of the ectopic tissues according to the new environmental signals, indicating that the transplanted tissue is naive with respect to its differentiation program. These experiments were performed orthotopically, without changing their segmental position. As shown here when rotated, ectopic PSM and epithelial somites from the thorax region are competent to form sclerotomal or dermomyotomal derivatives in response to environmental signals in the neck region by forming patterns characteristic of the thoracic pattern. On the one hand, this illustrates the general potency of the dorsalizing and ventralizing signals in the somitic environment, and on the other hand proves that all cells within a segment have the positional information to form any somitic derivative according to their original segmental identity.
The molecular basis of positional information is the segment-specific combination of Hox-gene expression, the Hox code (Kessel and Gruss, 1991). Hox genes contain a 183-bp homeobox and code for transcription factors regulating pattern formation in various tissues (McGinnis et al., 1984). In the paraxial mesoderm, the cranial expression boundaries of Hox genes demarcate the transition between different regions of the vertebral column. In chick, the cranial expression boundary of Hoxc6, which is situated at somites 19/20, marks the transition from thoracic to cervical vertebrae (Burke, 1995). Therefore, Hoxc6 can be used as a marker of thoracic versus cervical positional information. It is demonstrated that paraxial mesoderm transplanted from thoracic to cervical locations maintains Hoxc6 expression in the ectopic environment, substantiating the notion that positional information is stably determined (Nowicki and Burke, 2000). Importantly, it appears that Hoxc6 expression is kept in its normal position within ectopically developing structures following rotation and heterotopic transplantation, because cervical ribs formed from primordial dermomyotome precursor cells, and cervical scapulae formed from primordial scapula precursor cells. Why Hox gene expression is so stable, and how Hox-based positional information is transformed into pattern, is still unknown.
In summary, our results demonstrate that dorsally and ventrally located cells of a paraxial mesodermal segment share the same segmental identity and are able to form segment-specific skeletal patterns from both somitic compartments, the sclerotome and dermomyotome. It is additionally shown that the intrinsically controlled acquisition of segment-specific morphogenetic information precedes the extrinsically induced cell differentiation program.
Fertilized eggs of the white Leghorn (Gallus gallus) and the Japanese quail (Coturnix coturnix) were incubated at 80% relative humidity and 37.8°C. The embryos were staged according to Hamburger and Hamilton (1951), henceforth referred to as HH-stage.
Donors were incubated up to HH-stage 12, 13, or 15. After removal of surface ectoderm, portions of caudal or cranial PSM corresponding to the level of somite 19–24 (thoracic), or three consecutive epithelial somites 20–22, were isolated by use of electrolytically sharpened tungsten needles. They were marked with Nile-blue for orientation, rotated as desired, and transferred by a mouth-controlled glass capillary into the prepared hosts of HH-stage 11/12. At the cervical level of the hosts, the ectoderm was cut open, 3–4 somites were removed and replaced with donor tissue. For transplantation of half-somites, somite halves were dissected from donor and host embryos by use of tungsten needles and processed as described in the Results section.
In Situ Hybridization
In situ hybridization was performed as previously described (Nieto et al., 1996) using a chicken Hoxc6 probe (Burke et al., 1995). Selected embryos were vibratome-sectioned and mounted for bright field microscopy.
Immunohistochemistry of histological sections was performed as described previously (Huang et al., 2000). Quail cells were detected with a monoclonal QCPN-antibody (Developmental Studies Hybridoma Bank, Iowa City, IA). A polyclonal anti-desmin antibody (Sigma, Deisenhofen, Germany) was used for identification of differentiated muscle cells. Choice of second antibodies and color reactions led to a brown signal for desmin-including cells and a blue signal for quail cell nuclei.
Two different whole-mount alcian blue staining methods were used to investigate the skeletal pattern of the host embryos. (1) To increase the permeability of the tissue, specimens were fixed in ethanol overnight and then kept overnight in acetone. Subsequently, specimens were stained for several days with 0.017% alcian blue and 0.006% alicarin red in 68% ethanol and 5% acetic acid, digested in 1% KOH in 20% glycerine, cleared and stored in 100% glycerine (Wallin et al., 1994). (2) Specimens were stained with 0.015% alcian blue in 80% ethanol and 20% acetic acid for 1 to several days, fixed and dehydrated in ethanol for 1 day, cleared and stored with 100% methylsalicylate (Kant and Goldstein, 1999). Using method (2), it is possible to combine paraffin sectioning with immunohistochemistry. 4
We are indebted to Dr. Cliff Tabin for the chick Hoxc6 probe. We thank Developmental Studies Hybridoma Bank for supplying QCPN antibody. We thank Mrs. E. Gimbel, Mrs L. Koschny, Mrs. U. Pein, and Mr. G. Frank for excellent technical assistance. This study was supported by a grant from the Deutsche Forschungsgemeinschaft (Hu729-3) to R.H.