SEARCH

SEARCH BY CITATION

Keywords:

  • avian;
  • chick embryo;
  • mesoderm;
  • neural crest;
  • myogenesis;
  • cell interactions;
  • craniofacial

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Fate maps based on quail–chick grafting of avian cephalic neural crest precursors and paraxial mesoderm cells have identified the majority of derivatives from each population but have not unequivocally resolved the precise locations of and population dynamics at the interface between them. The relation between these two mesenchymal tissues is especially critical for the development of skeletal muscles, because crest cells play an essential role in their differentiation and subsequent spatial organization. It is not known whether myogenic mesoderm and skeletogenic neural crest cells establish permanent relations while en route to their final destinations, or later at the sites where musculoskeletal morphogenesis is completed. We applied β-galactosidase-encoding, replication-incompetent retroviruses to paraxial mesoderm, to crest progenitors, or at the interface between mesodermal and overlying neural crest as both were en route to branchial or periocular regions in chick embryos. With respect to skeletal structures, the results identify the avian neural crest:mesoderm boundary at the junction of the supraorbital and calvarial regions of the frontal bone, lateral to the hypophyseal foramen, and rostral to laryngeal cartilages. Therefore, in the chick embryo, most of the frontal and the entire parietal bone are of mesodermal, not neural crest, origin. Within paraxial mesoderm, the progenitors of each lineage display different behaviors. Chondrogenic cells are relatively stationary and intramembranous osteogenic cells move only in transverse planes around the brain. Angioblasts migrate invasively in all directions. Extraocular muscle precursors form tightly aggregated masses that en masse cross the crest:mesoderm interface to enter periocular territories, while branchial myogenic lineages shift ventrally coincidental with the movements of corresponding neural crest cells. En route to the branchial arches, myogenic mesoderm cells do not maintain constant, nearest-neighbor relations with adjacent, overlying neural crest cells. Thus, progenitors of individual muscles do not establish stable, permanent relations with their connective tissues until both populations reach the sites of their morphogenesis within branchial arches or orbital regions. Developmental Dynamics 235:1310–1325, 2006. © 2006 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

The development of musculoskeletal systems requires both lineage-autonomous and -interdependent programming events. Typically, myogenic and connective tissue progenitors are induced and begin their differentiations independently (reviews, e.g., limbs: Duprez,2002; Anakawe et al.,2003; Venters et al.,2004; head: Francis-West et al.,2003; Noden and Trainor,2005; Noden and Francis-West,2006). However, the subsequent morphogenesis of muscles necessitates especially tight integration with their surrounding connective tissues. Herein, spatial organization is established first within these connective tissues and then imposed by them upon myogenic populations.

In limbs, myogenic populations move into limb buds as positional specification of connective tissue precursors is under way (e.g., Lance-Jones,1979; Kardon,1998; Kardon et al.,2002). In contrast, craniofacial muscle precursors become associated with their connective tissue progenitors, derived from the neural crest, at several different stages (Noden,1983a,1986; Borue and Noden,2004). Throughout branchial arch formation, the two progenitor populations establish and maintain contiguity. However, the stability of specific cell:cell relations across the mesoderm:neural crest interface is unknown. Defining this relation is a necessary prelude to assessing when and by what mechanisms crest cells impart patterning guidelines to myogenic populations.

Fate-mapping studies using approaches such as staining with vital dyes, applied radioactive and fluorescent labels, transplantation of labeled precursors, and reporter gene activation in transgenic mice have identified the precise sites of origin for most craniofacial musculoskeletal and other connective tissues (reviewed in Noden,1988; Le Douarin and Kalcheim,1999; Hall,1999; Francis-West et al.,2003). Detailed mapping studies using quail–chick transplantations have defined neural crest cells as the exclusive source of connective tissue components in the entire midfacial, oral, and branchial regions of the head (LeLiévre and Le Douarin,1975; Le Liévre,1978; Noden,1978,1983a,1988; Couly et al.,1992,1993; Le Douarin et al.,1993; Köntges and Lumsden,1996). Indeed, this process appears to be a common developmental theme in all vertebrates so far examined, including mice (Jiang et al.,2002; Matsuoka et al.,2005; O'Gorman,2005; Gage et al.,2005). These properties have elevated the neural crest to the status of vertebrate synaptomorph (Gans and Northcutt,1983; Kuratani,2003; Santagati and Rijli,2003; Northcutt,2005).

Cephalic paraxial mesoderm gives rise to an array of connective tissues that are histologically identical to those formed by neural crest (but see Kasperk et al.,1995, for functional differences). This mesoderm is the sole source of skeletal muscles in the head and, together with lateral mesoderm, generates all craniofacial plus some cardiac endothelial precursors (Noden,1983b,1986,1991a; Couly et al.,1992,1995). Extraocular and branchial muscles initiate their development within paraxial mesoderm, but complete their differentiation and morphogenesis at peripheral locations populated by neural crest cells (Noden,1983b,1986; Couly et al.,1992; Köntges and Lumsden,1996; Matsuoka et al.,2005; Gage et al.,2005).

The crest:mesoderm interface is a site of critical interactions affecting branchial muscle differentiation (Tzahor et al., 2004) and morphogenesis (Hörstadius and Sellman,1946; Wagner,1949; Noden,1983a,1986; Borue and Noden,2004). The plane of this interface is initially smooth but becomes increasingly irregular. During formation of branchial arches, for example, crest cells initially located superficial to mesoderm move inward, circumscribing the muscle primordia (Trainor and Tam,1994; Trainor et al.,1994; Kimmel et al.,2001). Within the arches, myogenic cells remain segregated until later stages, when crest cells penetrate differentiating muscles to establish endo-, peri-, and epimysial layers (Noden,1983a; McClearn and Noden,1988; Trainor and Tam,1995; Noden and Francis-West,2006). Köntges and colleagues (Köntges and Lumsden,1996; Matsuoka et al.,2005) have shown that, at later stages, the crest-derived progenitors of connective tissues associated with some branchial muscles will distort the original crest:mesoderm interface to reach distant attachment sites.

At early stages, the interface constitutes a selective barrier to mixing of crest and mesodermal cells. Based on analyses of quail–chick chimeras and transgenic mice, skeletogenic populations do not cross the interface (Noden,1983b; Couly et al.,1993; Köntges and Lumsden,1996; O'Gorman,2005). In contrast, angioblasts readily cross the interface, moving invasively from mesoderm into the initially avascular crest population (Noden,1990,1991c; Ruberte et al.,2003). Melanocytes frequently move along interfaces, such as the surface of the dermatome and internal serosal surfaces, but their later presence in epidermal as well as many internal organs indicates an ability to cross boundaries (Tosney,2004).

In amniote embryos, the crest–mesoderm interface is undetectable without markers. Some boundaries are completely cryptic, e.g., dermis, with no morphological sign of an underlying disparity of origins. For cartilage and bone, the embryonic interface may demarcate sites of suture or other joint formation, but often these embryonic interfaces are also masked by secondary fusions or replacement, for example, at the boundary between squamosal and petrous aspects of the mammalian temporal bone.

To date, most fate-mapping studies have focused on tracking large populations of crest or mesodermal cells as they change locations and differentiate. Chick–quail chimeric studies by Köntges and Lumsden (1996) confirmed Noden's (1988) descriptions of a conserved and constant registration between branchiogenic crest populations and associated muscle precursors during formation of the arches. Precursors of individual muscles and their connective tissues arise at the same axial level and maintain these relations throughout their development (reviewed by Noden,1991b; Noden and Trainor,2005). When myoblasts leave the branchial arch environment, for example the progenitors of the trapezius, both muscle and associated connective tissue primordia remain as a constant, closed compartment (Matsuoka et al.,2005; but see O'Gorman,2005, for exceptions).

Unavailable from previous studies, however, is a clear portrait of the cell:cell relations across the neural crest:mesoderm interface. Defining the stages at which stable, constant relations are established between progenitors of individual muscles and the skeletal tissues with which they will establish connections is an essential prerequisite to dissecting the signaling interactions between them. The primary aim of this research has been to discover when this musculoskeletal partnership is established for specific branchial and periocular tissues. The approach has been to inject β-galactosidase–encoding, replication-incompetent retrovirus at the interface between crest cells and paraxial mesoderm while both are en route to their eventual sites of terminal differentiation. In control embryos, either paraxial mesoderm or neural crest precursors were exposed to retrovirus to the exclusion of the other population.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Validation of the Method

The ability of the replication-incompetent retroviruses used in this study to target avian embryonic cells is well documented (Mikawa et al.,1992; Mikawa and Gourdie,1996; Wei and Mikawa,2000; Evans,2003; Rees et al.,2003). Concerns arising from injections into head mesoderm include the extent of spread and the potential for cross-infection of neural crest cells. Embryos fixed 2–4 days after injection typically showed multiple labeled clusters close to the injection site, with a surrounding halo of single scattered cells. Rarely were more than half the proximate cells labeled. However, analysis at later stages showed that precursors of all mesodermal lineages were equally susceptible to infection.

Unintended infection of crest cells could occur by virus particles reaching the basal surface of the neural fold or, if they spilled over the surface of the embryo, by reaching the apical surface of the neural fold. Alternatively, virus could remain infective until crest cells move over paraxial mesoderm. All mesoderm injections were done >6 hr before the emergence of crest cells from the roof of the mid- or hindbrain regions, and these viruses are quickly inactivated by binding to extracellular matrix and phagocytosis.

Transplantation studies have identified several cell types and also locations within the head that are derived exclusively from either mesoderm or neural crest progenitors; these served as the initial basis for validation of the retroviral infection protocols. Schwann cells and sensory neurons are exclusively of neural crest (and placodal) origin, as are skeletal tissues in branchial and midfacial regions. Conversely, skeletal muscle, endothelial cells, and most skeletal elements in the floor of the neurocranium are exclusively of paraxial mesoderm origin.

Of the 122 embryos included in control paraxial mesoderm infections, we found no cases that deviated from this exclusivity of lineages. In assigning sites of origin for each structure, the mesodermal injection sites from all embryos with a particular muscle or skeletal element labeled were charted, and all other structures labeled in each case were noted. Figure 1 illustrates the center of each injection site that resulted in labeling of selected muscles (dorsal oblique, lateral rectus) and bones (frontal, parietal).

thumbnail image

Figure 1. This illustrates the method used to define sites of individual muscle and bone origin after injection of paraxial mesoderm. Circles indicate the site of each injection relative to fixed landmarks. These are approximately half of the actual volume occupied by an injection bolus. All colabeled muscles are noted. The frequency of colabeling was used to establish which primordia share overlapping domains or are either contiguous or separate. No injections colabeled the frontal and parietal bones, although other neurocranial cartilages were usually also labeled (Table 1). AM, adductor mandibulae; DO, dorsal oblique; LR, lateral rectus; P, pyramidalis; Q, quadratus nictitans; VO, VR, ventral oblique and ventral rectus muscles. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.].

Download figure to PowerPoint

Table 1. Result of Paraxial Mesoderm Injections
inline image

Lineage-Specific Movements of Paraxial Mesoderm Cells

Precursors of each of cell type derived from head paraxial mesoderm display striking, lineage-specific differences in their early movements (Fig. 2). Chondrogenic precursors within mesoderm are essentially stationary. Whereas asymmetric expansions and differential growth may create lateral or dorsal pillars and wings of cartilage, the axial level at which a group of mesoderm cells initiates chondrogenesis beginning on days 5–6 of incubation is identical to its location in the stage 9 embryo. This finding does not change with the later onset of endo- and perichondral bone formation.

thumbnail image

Figure 2. The diverse, lineage-specific behaviors of paraxial mesoderm cells is shown. Medium containing Lac-Z-coding replication-incompetent retrovirus was injected into paraxial mesoderm beside the mid-mesencephalon; the embryo was harvested 2 weeks later, the skin and eyes were removed, and the embryo was processed for β-galactosidase. Some infected cells remained stationary and formed cartilage in the postorbital process (orbital part of the pleurosphenoid). In contrast, osteogenic precursors moved dorsally from the injection site, while myogenic cells aggregated and then moved en masse as cohorts to distant, rostral periocular locations. Angioblasts migrated invasively and in multiple directions, contributing the peripheral and deep endothelial tissues.

Download figure to PowerPoint

In contrast, direct developing osteogenic precursors of neurocranial intramembranous bones egress from their sites of origin and disperse dorsally. Most of these progenitors arise in the ventromedial region of paraxial mesoderm and subsequently move around the lateral then dorsolateral sides of the brain in advance of their overt differentiation. No significant changes in the rostrocaudal location of these osteogenic mesoderm cells was evident, but later growths of each bone were often very asymmetric, which creates the appearance of a shift in their axial location. Also, later expansion of the forebrain changes the identity of neuroepithelial and meningeal tissues underlying the frontal and parietal elements.

Previous mapping and gene expression studies have shown that skeletal muscle progenitors form aggregates at or close to their sites of origin within paraxial mesoderm (Noden,1983b; McClearn and Noden,1988; Noden et al.,1999). Retroviral injections confirm that progenitors of extraocular muscles first condense, then move as cohorts toward the mesoderm:crest interface, and subsequently enter the crest-derived periocular environment. For most branchial muscles, their ventral movements occur simultaneously with the translocations of corresponding crest populations. In contrast, the movements of extrinsic ocular muscles are not completed until after crest cells have fully enveloped the optic vesicle.

Endothelial precursors constitute the largest subpopulation within cranial paraxial mesoderm at stage 9. Most of these angioblasts are invasive, and the population originating at each site spreads omnidirectionally. This takes them into neighboring mesodermal regions, and especially into the otherwise avascular neural crest mesenchymal populations (Fig. 2). The dispersal of angioblasts is greater in the rostral and ventral directions, which parallels the eccentric growth of the embryonic head during the days immediately after these injections. Every embryo with labeled myotubes also had labeled endothelial cells, but the two populations were often not congruent.

Mapping Paraxial Mesoderm

Paraxial mesoderm cells located lateral and ventrolateral to midbrain and hindbrain form the walls and floor of the cartilaginous neurocranium (Fig. 1), the large, calvarial part of the frontal bone, and the parietal bone (Figs. 3, 4). Histological analyses confirmed that labeled mesoderm cells in these bones are indeed osteocytes (Figs. 3B3, 4B3), often in addition to endothelial cells.

thumbnail image

Figure 3. Contributions of paraxial mesoderm to neurocranial skeletal tissues. A: Series A illustrates the presence of labeled mesoderm cells in the parietal bone (Par) in an intact head with the skin removed (A1), in dissected frontal (Frn) and parietal bones from which the underlying dura had been removed (A2) and in paraffin section (A3). B,C: Dorsal views of partially dissected skulls, showing the transverse bands of labeling in the floor of the neurocranium; rostral is at the bottom, medial is to the right. In B, after injection beside the caudal mesencephalon, the band of labeled cells extends laterally from the rim of the adenohypophyseal foramen (Pit, adjacent to a loop of the internal carotid artery) to the orbital part of the pleurosphenoid (Opl) and postorbital process (Po). PA, pila antotica. C illustrates labeled chondrocytes in the petrosal (Pet) and surrounding the anterior semicircular canal (ASC), in addition to labeled osteocytes in the parietal bone, after injection of paraxial mesoderm beside rhombomere 3. D: Multiple foci of labeled mesoderm-derived chondrocytes in a section from the rostral side of the wall of the anterior semicircular canal.

Download figure to PowerPoint

thumbnail image

Figure 4. The contributions of neural crest (column A) and mid-mesencephalic paraxial mesoderm (column B) to the frontal bone in dissected specimens from 14-day embryos. A: Crest-derived osteocytes are found in the supraorbital ossification center (arrows, A1; enlarged in A2, in section in A3). B: In contrast, mesoderm cells populate the large calvarial part of the frontal bone (B1, B2). A3 and B3 verify that labeled cells are osteocytes.

Download figure to PowerPoint

Ventral to the brain, the crest–mesoderm cartilaginous interface is located beside the adenohypophyseal foramen. The interface is sharp and abrupt at the site of fusion of the basisphenoid and rostral parasphenoid cartilages, but is irregular with considerable mixing in the later forming perichondral bone that circumscribes the foramen.

Because there is little rostrocaudal movement of skeletogenic precursors, labeled cells in the cartilaginous floor of the neurocranium (pleurosphenoid, basisphenoid, excluding the lateral wings) were distributed in transverse bands that spanned several regions and included bones forming by endochondral ossification (Fig. 3B,C). At sites where mesoderm gave rise to both endochondral and intramembranous bones, many embryos had labeling in both populations, but always within the same transverse band. Thus, if precursors of these two types of bone are spatially segregated, it is by less distance than our injection design can discriminate. The boundary between mesenchymal and somitic paraxial mesoderm corresponds to the junction between sphenoid and occipital structures.

The sites of origin of all intrinsic head and laryngo-glossal muscles have been identified (Table 1; Figs. 2, 5,6). Progenitors for most muscles or muscle groups arise in the same rostrocaudal order as the cranial nerves that will innervate them. The exceptions are the lateral rectus, quadratus nictitans, and pyramidalis muscles. These muscles originate beside rhombomeres 1 and 2, at the same axial level but medial to first branchial arch muscles. However, they are innervated by cranial nerve VI, which emerges from rhombomeres 5 and 6, whereas the mandibular (V) cranial nerve to jaw closing muscles emerges from rhombomere 2.

thumbnail image

Figure 5. Contributions of infected paraxial mesoderm cells to craniofacial skeletal muscles; all except (H) are intact muscles from partially dissected 12- to 15-day embryos. Note: reference to Figure 10A may help with orientation and muscle identification. A: Lateral view of the dorsal oblique. The eye was removed from this embryo before processing to allow visualization of deep orbital structures. B: Lateral view of the palpebral depressor; note the presence of labeled endothelial (endo) cells in this broad, thin muscle. Arrows indicate secondary myotubes, recognized by their short lengths. C: Caudal view of the muscle of the columella, which originates beside rhombomere 3 but shifts location to complete its morphogenesis caudal to the occipital wall of the skull. D,E,F: Muscles in the first, second, and third branchial arches, respectively. D: Ventral view of intermandibularis (IM) and mandibular adductor (MA) muscles attaching to the jaw skeleton. E: Lateral view of the mandibular depressor (MD), stylohyoid (ST), serphihyoid (SR), interceratobranchial (IC), and caudal mylohyoid (CM) muscles. F: Lateral view of the branchiomandibular (BM) muscle. G: Lateral view of the dilator and constrictor muscles of the glottis (DG, CG), showing contributions from somite 1. A small nodule of labeled chondrocytes in the arytenoid cartilage is present; this finding is likely due to the absence of an intact lateral boundary to the transient first somite. Arrows indicate endothelial cells. H: Higher magnification of individual labeled myotubes in a sectioned mandibular depressor muscle. I: A dorsal view of the segment-specific labeling of the biventer muscle; labeled myotubes are from somite 4; cells in the unlabeled segments arise from adjacent caudal somites.

Download figure to PowerPoint

thumbnail image

Figure 10. The results of double-labeling experiments. A: The avian skull with tissues derived from paraxial mesoderm in red and those from neural crest in blue; muscles included in B–H are indicated. Red and blue dots indicate the locations of labeled mesoderm and neural crest cells, respectively. The central sketch portrays a stage 10 embryo from dorsal perspective, identifying the sites of injection corresponding to examples B–H; all injections were done on the right side. Cartilages and endochondral bones: Ar, retroarticular process; basihyoid (rostral basibranchial); Bs, basisphenoid; Cb, ceratobranchial; Eb, epibranchial; En, endoglossum (paraglossum); Eo, exoccipital; Ios, interorbital septum; Lpl, lateral wing of the pleurosphenoid; Mc, mandibular (Meckel's) cartilage; Nc, nasal capsule; Otic, otic capsule, which includes the petrous bone; Po, postorbital process of the pleurosphenoid; Qd, quadrate; Rps, rostral pleurosphenoid; So, supraoccipital. Intramembranous bones: Ang, angular; Den, dentary; Eth, ethmoid; Frn, frontal; Jug, jugal; Max, maxillary; Nas, nasal; Pal, palatine; Pfr, prefrontal (lacrimal); Pmx, premaxilla; Ptr, pterygoid; Qju, quadratojugal; San, supra-angular; Spl, splenial; Sqm, squamosal; Vom, vomer.

Download figure to PowerPoint

thumbnail image

Figure 6. Summary map, showing the sites of origin within paraxial mesoderm of avian skeletal muscles (right side) and skeletal elements (left side). The ventral, medial, and dorsal recti were occasionally labeled, but not in adequate numbers to allow accurate mapping.

Download figure to PowerPoint

There is additional topographical organization of individual muscle precursors within each of the large populations of branchial arch muscle progenitors. This finding is best seen in the distinct foci for the palpebral depressor (Fig. 5B) and columella (Fig. 5C) muscles within the larger branchial arch 1 and 2 groups, respectively. The distinctness of these foci was evident in embryos in which these muscles were fully labeled, but their accompanying muscles, which share the same innervation, were sparsely labeled.

Endothelial precursors showed the greatest dispersal from each injection site (Figs. 1, 7) and are the only cell lineage that readily crosses the midline, especially dorsal to the brain. Excluding cartilage, every tissue in the head is well vascularized by day 14 of incubation, including loose and dense connective tissues (Fig. 7A,B) and also skeletal muscles (Fig. 5B). Close to the site of injection, plexuses of labeled endothelial channels were seen both within meningeal layers on the surface of the brain and deep within neuroepithelial tissues (Fig. 7C,D). At sites further distant from the injections, chimeric vessels formed by angioblasts from multiple regions were more predominant.

thumbnail image

Figure 7. A–D: Contributions of Lac-Z-expressing paraxial mesoderm cells to endothelial tissues in periocular feathers (A), in the squamosal bone (B), on the dorsal surface of the cerebellum (C), and penetrating the floor of the hindbrain (D). A–C are intact tissues; D is a section. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Neural Crest Labeling

A different strategy was used to label neural crest cells. Retroviruses were washed over the surface of embryos beginning at stage 6, when the neural plate is open and neural crest progenitors within neural folds are exposed. Surface washes frequently labeled cells within the brain (Fig. 8A) and spinal cord and also those derived from placodes in, for example, the trigeminal and vestibuloacoustic ganglia, the membranous labyrinth (Fig. 9A), or along the olfactory nerves. These are not confounding with respect to musculoskeletal structures. With the titers available, the infected crest progenitors were often scattered and separate. This distribution typically resulted in a mosaic of labeled and unlabeled cells in each crest-derived skeletal structure. Collectively, in the 66 cases of crest-labeled embryos examined in this control series, all cranial ganglia and branchial skeletal structures were labeled, as were a variety of loose connective tissues. Equally important, no skeletal myotubes, cardiomyocytes, or endothelial cells were labeled in this series.

thumbnail image

Figure 8. Labeling patterns in embryos whose surface ectoderm, including neural crest progenitors, was washed with Lac-Z retroviruses. A: This 12-day embryo was cut in a parasagittal plane to reveal labeled cells in the floor of the brainstem (arrows) and also tracts in the telencephalon. Labeled crest cells are in several midfacial skeletal structures. B: A stream of labeled crest cells fills the right first branchial arch in this stage 13 embryo, shown in ventral view. C: An intact 7-day (stage 30) embryo with labeled cells throughout the frontonasal and second branchial arch regions. This embryo was washed at stage 9+, by which time the caudal midbrain and metencephalic neural folds had already fused, leaving precursors of the maxillary and mandibular arch crest populations inaccessible to infection. D: An anterior view of the eye, showing labeled crest cells (arrow) invading the space between the lens and the anterior corneal epithelium. In E, a section through a 14-day embryonic cornea, crest-derived corneal stromal cells are evident. F,G: The pyramidalis and quadratus nictitans muscles and their shared tendon (arrows) after mesoderm (F) and neural crest (G) infection. Note the opposite labeling patterns, with mesoderm forming muscle cells and crest forming the connective tissues both within and beside these muscles. H: A surface view of the eyelid, showing labeled dermal fibroblasts. I: A ventral view of a 7-day embryonic heart. Crest cells envelope the aortic arches, including the root and arch of the aorta and the pulmonary trunk, and extend into the truncal part of the outflow tract. J: Labeled crest cells contributing to smooth muscle tunics of the brachiocephalic arteries. K: A dissected brachiocephalic and carotid artery; as expected, labeled cells are absent from the subclavian artery, whose enveloping smooth muscle is derived from mesoderm.

Download figure to PowerPoint

thumbnail image

Figure 9. Examples of labeling within skeletal tissues after exposure of neural crest progenitors to LacZ retroviruses. A: A median view of a bisected 14-day head, showing the broad distribution of crest-derived cells in the frontonasal, maxillary, palatal, mandibular, and branchial regions. The arrow shows the boundary of contributions by crest cells to roofing elements of the skull. Labeled cells are visible in the trigeminal ganglion (Trig.) and as Schwann cells along the associated ophthalmic nerve, and also in the placode-derived posterior semicircular duct (asterisk). B: A dissected mandible, in ventrolateral view, with labeled crest cells in the dentary (Den), surangular (San), and splenial (Spl) bones. C: Labeled chondrocytes in the interorbital septum (IOS) and osteocytes in the pterygoid (Ptr) and palatine (Pal) bones. D: The hyobranchial skeleton, with labeling in the rostral basibranchial (rBb, also called the basihyoid), ceratobranchial (Cb), and epibranchial (Eb) cartilages, which are derived from crest cells in branchial arches 2/N4. E,F: Sections showing labeled crest cells as chondrocytes and perichondrocytes in the interorbital septum and osteocytes in the maxilla.

Download figure to PowerPoint

Figures 8 and 9 illustrate examples of labeled crest cells contributing to a variety of loose and dense connective tissues as well as perivascular smooth muscle cells. Within the frontal bone, osteocytes derived from the neural crest were found in the rostral, supraorbital region but not the larger calvarial part of this bone (Fig. 4A). This fully complements the pattern of frontal bone labeling found after mesoderm infection. In a few cases, multiple positive foci were grossly visible along the inner surface of the frontal bone. Careful dissection revealed that this labeling was limited to the dura, which secondarily arrives beneath the frontal bone as the cerebral hemispheres expand caudally. Labeled crest cells were not observed in the parietal bone. Also, no crest labeling was found within the cricoid or arytenoid cartilages (birds do not have a thyroid cartilage).

Double Labeling of Mesoderm and Neural Crest Cells

Virus was injected at the interface of mesoderm and neural crest cells in embryos having 8–11 somites (stages 10 to 11), during which time the crest population expands over the surface of cephalic paraxial mesoderm. This approach permits analyses of the congruence between crest and mesodermal populations during the formation and growth of branchial arches. Of particular interest in this study, it enabled an assessment of nearest-neighbor relations among initially adjacent crest and mesodermal cells. As shown schematically in Table 2, some injections targeted the interface near the leading edge of the neural crest, others were more medial and targeted regions where the interface was well established. This series included 49 embryos in which both crest- and mesoderm-derived tissues were labeled.

Table 2. Double Labeling of Neural Crest and Paraxial Mesoderm
inline image

Mesodermal tissues labeled in this series correspond closely to those labeled after mesoderm injections at earlier stages. The absence of labeling in frontal and parietal bones confirms their origins from deeper, ventromedial mesoderm progenitors that were not targeted in these interface injections. These data confirm the results of control neural crest washes, which did not identify crest contributions to these roofing elements of the skull.

Focal injection studies allow for positional mapping of the neural crest (Table 2), which was not possible in whole embryo washes. The results confirm previous crest mapping studies (Noden,1978,1983a; Köntges and Lumsden,1996), showing that the individual elements of the tongue and hyobranchial skeleton do not strictly correspond to crest cells associated with each branchial arch. The most rostral element, the entoglossum (paraglossum), receives contributions from both first and second arch crest cells, and often small clusters of labeled crest cells, likely derived from single progenitors, were present in adjacent hyobranchial cartilages (Fig. 10E). Similarly, small clusters of labeled cells frequently spanned the joints between the mandibular (Meckel's) cartilage and the quadrate (Fig. 10C), including labeling of the joint capsular connective tissues.

Crest cells destined to populate the first branchial arch arise along a greater rostrocaudal territory than do the mesoderm-derived muscles that will attach to skeletal structures derived from them. Rostrally, crest cells initially overlying the dorsal oblique precursors are not destined for periocular regions (Fig. 10B). Rather, these crest cells will contribute to maxillary and distal mandibular connective tissues. In addition to labeling second branchial arch structures, some injections targeting second arch crest cells generated labeled osteocytes in the squamosal bone. There are no comparable data from previous quail–chick transplants or labeling studies in mice (O'Gorman,2005). It is likely that viral medium placed near the boundary of the second arch population may have spread across the superficial crest-free zone and infected caudal first arch crest cells.

Double-labeling results confirm that precursors of first branchial arch muscles always arise within the rostral and caudal limits of the corresponding neural crest population. This conserved axial registration persists through later stages for most branchial muscles, but there are exceptions. The palpebral depressor muscle, for example, is first evident as a distinct population of myotubes located proximally within the first arch muscle mass and partially wrapped around the mandibular motor nerve. These cells subsequently break away and shift rostrally into the lower eyelid, whose connective tissues arise from more rostral sites (Fig. 10E).

The registration of muscle and skeletal progenitors destined for branchial arches 2 and 3 is initially less tight. These myogenic precursors are dispersed over a greater rostrocaudal expanse than are the corresponding neural crest cells, most of which emigrate from rhombomeres 4 and 6. As crest populations move laterally, they condense to form transversely oriented tongues of cells with distinct gaps between them. The gaps are greater in size than the spaces between muscle progenitors, suggesting the need for additional alignment of these two populations.

Analysis of β-galactosidase–positive embryos 1 to 2 weeks after injections reveals that several clusters of labeled neural crest cells were typically found in the same or adjacent skeletal elements (Figs. 10, 11). These results indicate that changes in the relative positions of individual neural crest cells and their progeny are modest as the crest populations expand and shift ventrally to form branchial arches. The same conservatism is present in paraxial mesoderm cells during this period, excepting, of course, angioblasts.

thumbnail image

Figure 11. The absence of retained nearest-neighbor relations between myogenic (mesoderm) and connective tissue (neural crest) progenitors. A: Labeling is in the mandibular adductor myotubes and in the periocular loose connective tissues (neural crest, arrows). B: Ventral view of the jaw, showing labeled intermandibularis muscle fibers plus crest-derived osteocytes in the quadratojugal (Qju) and angular (Ang) bones and chondrocytes in the mandibular cartilage (MC; note: this is the embryo shown in Fig. 10C). C: Lateral view, showing a similar case with crest labeling in the mandibular cartilage, quadratojugal and angular bones. In this embryo, the mandibular adductor complex is labeled, but few cells at the sites of myotube attachment were labeled. D: The one case in which both a muscle (mandibular depressor) and its crest-derived attachment site (squamosal) were labeled. This embryo had an unusually large and widely dispersed number of labeled crest cells, so the specificity of this colabeling is uncertain. Qd, quadrate cartilage. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

In contrast, crest and mesoderm cells that once were close neighbors across the interface typically move apart and occupy disparate locations; their earlier spatial relations were transient. In some cases, labeled crest cells are present in the correct bone or cartilage but are not at the muscle attachment sites (e.g., Figs. 10D,G, 11B). More often, however, labeled crest cells are not within the appropriate skeletal element (e.g., Fig. 10C,E,H). These results indicate that the separation of former nearest neighbors is both common and extensive. In only two cases (Fig. 10D,E) were muscle cells and their associated connective tissues both labeled; one of these involved the proximal attachments of the mandibular adductor externus onto the mesodermal postorbital process.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Critique of Methods

Targeted infection using replication-incompetent retroviruses has several advantages over quail–chick transplantation for lineage analyses in avian embryos; it is less invasive, avoids species differences in rates of development, and permits both whole embryo and histological analyses. This approach has been used to examine contributions of embryonic cell populations to the development of a wide range of tissues/structures including the heart, ribs, and limb muscles (Mikawa et al.,1992; Mikawa and Gourdie,1996; Wei and Mikawa,2000; Evans,2003; Rees et al.,2003).

There are limitations and potential drawbacks. Retroviruses do not concentrate well. Titers above 2 × 106 CFU/ml are difficult to obtain, and each injection typically results in the labeling of some but not all nearby progenitor cells. As such, each individual case presents a sample of the precursor populations present. Fortunately, it is much easier to obtain multiple replicates using this approach than by performing transplants. Thus, many cases can be assessed to define the totality of fates by cells in a given region and to determine the boundaries of specific progenitor populations. The volumes of media introduced transiently occupy a volume approximately 100 μm in diameter, which is slightly greater than the size of a somite. However, the number of virions near the boundaries of the bolus was too low to result in frequent infection.

All experimental mapping studies must ensure that the labeling strategy used is limited to the intended target population. In the present study, injections of paraxial mesoderm rostral to the somites were made when the embryo had fewer than six somites and, based on previous descriptive and immunocytochemical analyses (Noden,1975; Tosney,1982; Cochard and Coltey,1983), lacked any neural crest cells. The complimentarity and exclusivity of labeling resulting from control crest and mesoderm infections validates the experimental design used. No known neural crest derivatives were labeled in any control embryo that received a mesodermal infection, whereas injections at the same sites only a few hours later consistently labeled both mesoderm and neural crest derivatives.

Viral washes extensively labeled neuroepithelium. However, these embryos showed no labeling of cardiomyocytes, striated muscle, or endothelial cells. This finding confirms the results of Yaneza et al. (2002) and Boot et al. (2003), who were unable to confirm the presence of a late-emerging population of multipotential ventral hindbrain (VENT) cells (Sohal et al.,1998).

Spatial Relations and Lineage Determination

Neural crest cells and paraxial mesodermal cells move and interact extensively during craniofacial morphogenesis, producing a diverse array of tissues in a spatially organized and integrated manner. These interactions are bidirectional, with signals from crest cells regulating the rate of muscle differentiation (Tzahor et al., 2004) and patterns of muscle morphogenesis (Noden,1986; Olsson et al.,2001; Borue and Noden,2004), and mesoderm participating in the positional specification of nearby crest populations (Trainor and Krumlauf,2001; Trainor et al.,2002). Both components also receive organizing influences from surface ectoderm and underlying pharyngeal endoderm (Couly et al.,2002; Rubin et al.,2003; Graham et al.,2004).

Previous mapping studies have not established when myogenic mesodermal cells and skeletogenic precursors within neural crest populations establish intimate, stable relations. An early and sustained association would allow for the simultaneous and synchronous modulation of muscle differentiation and muscle morphogenesis by signals from adjacent tissues, including but not limited to the neural crest (Tzahor et al.,2003). Alternatively, precise, nearest-neighbor relations may be unnecessary during the early differentiation stages of myogenesis. Indeed, transplantation data show that the initial stages of myoblast induction in the head are very site-specific (Borue and Noden,2004), but that morphogenetic specification of muscles occurs over a longer period and, especially for extraocular muscles, at a later time (reviewed by Noden and Francis-West,2006).

Results from our studies confirm that populations of neural crest and myogenic mesoderm cells bound for specific branchial arches arise and develop in close longitudinal registration, confirming the description by Köntges and Lumsden (1996) that crest populations maintain a “persistent coherence.” However, within these boundaries, individual crest and myogenic mesodermal cells do not establish stable relations until both have populated the branchial arches.

Once established, this partnership with connective tissue primordia is maintained during subsequent stages of muscle individuation, segregation, and the formation of attachments (McClearn and Noden,1988). Descriptive accounts in avian (Köntges and Lumsden,1996; Noden and Francis-West,2006) and mammalian (Matsuoka et al.,2005) embryos indicate that crest–muscle associations are often maintained even when muscles move out of the immediate branchial region. Whether these crest cells are determinative with respect to selecting sites of muscle attachment or are only necessary to provide connective tissues at these sites has not been resolved.

There are some exceptions to this rule of congruency. The avian palpebral depressor muscle arises within the first branchial arch domain, but its attachments are formed by periocular crest cells. Similarly, the mammalian mimetic (facial expression) muscles originate from myoblasts within the second arch (Gasser,1967; Carvajal et al.,2001), but fully exit this connective tissue environment as they spread throughout midfacial and jaw regions (O'Gorman,2005, and personal communication).

Individual mesodermal injections typically labeled several lineages. Our mapping studies do not establish whether the disparate behaviors exhibited by myogenic and skeletogenic lineages represent an early response to commitment or, conversely, reveal stochastic movements that bring otherwise uncommitted cells into novel inductive environments.

Our results further suggest that adjacent mesoderm cells follow different developmental pathways. However, given the area occupied by the initial injection bolus, we cannot exclude the presence of a cryptic compartmentalization in head mesoderm comparable to that found in somites. The classic model of a rigid, tripartite compartmentalization of each somite has been replaced in recent years by a more dynamic model in which boundaries between lineages are constantly reorganizing (e.g., myogenic and fibroblastic, Burke and Nowicki,2003; myogenic and dermoblastic, Tosney,2004). Our injection data support the early delineation of chondrogenic and osteogenic precursors in a location that would be comparable to the sclerotomal part of a somite. Whereas this location places them in proximity to hedgehog- and noggin-producing tissues such as the notochord, their broad contacts with both pharyngeal endoderm and a segmentally organized hindbrain likely provide unique signaling environments.

Both somitic sclerotome cells and their medial counterparts in head paraxial mesoderm express similar markers of chondrogenic differentiation, e.g., type II collagen (Thorogood,1993) and Sox9 (Eames and Helms,2004). However, their upstream signal/response pathways are very different. Markers of early sclerotome cell commitment, e.g., the hedgehog receptor patched and the transcription factor Pax1, are not expressed in head paraxial mesoderm before the onset of chondrogenesis (Noden, unpublished data).

It is known that some mesoderm cells have already committed to the angiogenic lineage at the time of viral infection (reviewed by Noden,1991c). Information on the other lineages is based on the earliest appearance of known lineage-specific transcription factors. The earliest signs of muscle commitment are evident at stage 13.5 (Noden et al.,1999), and most head muscles initiate aggregation and molecular differentiation close to the sites of their stage 9 progenitors. Craniofacial cartilage and bone do not show molecular evidence of lineage commitment, e.g., expression of Sox9 and Runx, until day 5 of incubation (Eames and Helms,2004; also Jaskoll et al.,1981), which is coincidental with the onset of precartilage mesenchymal condensation.

In their seminal paper of 1983, Gans & Northcutt imbue neural crest cells with the unique ability to form intramembranous (dermal, exoskeletal) bone, and list this as a key feature in vertebrate evolution. The data by Couly et al. (1993) reinforced this model. In contrast, Huang et al. (1997,2000) identified a mesodermal origin for the intramembranous caudal parasphenoid in their transplants. Based on the current report and several previous quail–chick transplantation papers (LeLiévre and Le Douarin,1975; Noden,1978,1983a), the ability to form intramembranous bone is not a unique property of neural crest-derived mesenchyme, but rather whatever mesenchyme is present overlying the midbrain and hindbrain regions can adopt this lineage. This lack of germ layer restriction has been confirmed recently by Jiang et al. (2002) and Matsuoka et al. (2005).

Muscle Origins

The progenitors of extraocular muscles associated with each of the three associated cranial motor nerves are segregated within head mesoderm. Injections lateral to the isthmus/metencephalic region frequently, but not always, labeled both extraocular (lateral rectus) and first branchial arch muscles. This finding suggests that the medial (extraocular) and lateral, more superficial (branchial) populations of muscle progenitors are closely apposed but separate foci.

Some individual branchial muscle precursors also appear to occupy segregated domains within larger, arch-specific progenitor pools. The results of transplantation studies (Noden,1983a,1986; Borue and Noden,2004) suggest that these segregated foci do not represent myoblasts with actual commitment or restriction to a specific muscle. It is not known when during the extended period of morphogenesis each muscle acquires its positional and molecular identity. Examination of early projections by motor axons suggests that cranial nerves III and VI can discriminate among alternate targets when muscle primordia are still at their early condensation phases (Wahl et al.,1994), suggesting that identity may be established within or around myoblast aggregation sites, as has been suggested for appendicular muscles (Duprez et al.,1999; Kardon et al.,2003).

Skeletal muscle fibers, including those within head muscles, are known to develop in two temporally distinct waves, with primary myotubes forming first and acting as scaffolds for later forming secondary myotubes (McClearn and Noden,1988; Wigmore and Evans,2002). During our investigation, both primary and secondary myotubes in individual muscles were labeled, indicating that the precursors of these different myotubes arise from the same or overlapping progenitor populations. Many positive mononucleated cells interspersed among myofibers were typically present, but we did not specifically assay for satellite cells.

The locations of extraocular and branchial muscle precursors identified in this study correlate well with the sites of activation of muscle-specific transcription factors, e.g., Myf5 and MyoD (Hacker and Guthrie,1998; Noden et al.,1999; Mootoosamy and Dietrich,2002). They are concordant with transplantation-based maps by Noden (1983b) and add finer definition to the boundaries of individual myogenic foci. The maps are similar to those presented by Couly et al. (1992) for the dorsal oblique and first branchial arch muscles but not the lateral rectus. Couly et al. (1992) placed this precursor beside the mid-mesencephalon, rostral to first arch muscle, whereas the present data confirms earlier studies by Noden (1983b) and Mootoosamy and Dietrich (2002) that the lateral rectus originates ventrolateral to the mesencephalic–metencephalic junction and rostral rhombomeres. The origins of cartilaginous components of the neurocranium (e.g., pleurosphenoid, basisphenoid) are generally consistent with previous transplantation studies.

Couly et al. (1992) identified the second branchial arch muscle primordia as extending caudally to the level of the first somite. We map its location to the level of rhombomere 4, at the same axial level as both its innervation (cranial nerve VII) and the neural crest cells with which these muscles will populate the second branchial arch. This congruency among all musculoskeletal and neural components destined for each branchial arch is an essential prerequisite for the coordinated morphogenesis of these tissues (Noden,1988,1991b; Noden and Trainor,2005; Kuratani,2005).

There is agreement that both of the first two somites contribute to the intrinsic laryngeal musculature (Noden,1983b; Couly et al.,1992; Huang et al.,1997,1999,2000), and that one (possibly both) of these participate along with several additional occipital somites in the genesis of the tongue musculature. We found no evidence that unsegmented head mesoderm contributes to either set of muscles.

Calvaria and Laryngeal Cartilages

The location of the crest:mesoderm boundary in roofing elements of the skull remains enigmatic, both evolutionarily and developmentally. Reports place this interface at frontal:parietal junction in mammals (Morriss-Kay,2001; Jiang et al.,2002) and further caudally at the frontoparietal:occipital boundary in Xenopus (Gross and Hanken,2005). Based on analyses of quail–chick transplantation chimeras, some investigators defined the boundary within the frontal bone, at the junction of supraorbital and calvarial parts of this bone (LeLiévre,1978; Noden,1978,1983a; Köntges and Lumsden;1996), whereas others find it further caudally, at the parietal:occipital junction (Couly et al.,1993).

In the present study, neural crest cells made no osteogenic contribution to the calvarial region, with the supraorbital component of the frontal bone being their most caudal location within the roof of the skull. In avian embryos, this supraorbital region arises from a separate ossification center than the rest of the frontal bone (Erdmann,1940). It is possible that these centers, which fuse to form a single bone in the chick, do not unite in mammals, with the coronal suture maintaining this separation. If this is the case, the nomenclature of bones in the avian and mammalian calvariae is misleading (see Stone and Hall,2004).

Alternatively, the location of the crest:mesoderm interface may have changed during vertebrate evolution. Formation of the skull roof, similar to the branchial arch musculoskeletal system, requires a consortium of interactions in which each component brings a unique history and responsiveness to the enterprise (Thompson,1993; Noden and Trainor,2005). In the calvaria, evolutionary shifts in the location of the crest–mesoderm interface may not be of irresolvable consequence if the local signaling environment is strongly directive. As an extension, assigning homologies based on embryonic ancestry (Wagner,1989; Kuratani,2004; Stone and Hall,2004) may not be relevant in this situation.

In contrast, neural crest cells that populate each of the branchial arches do carry positional biases (Hörstadius and Sellman,1946; Noden,1983a; Cerny et al.,2004a), which in large part define their responses to signals from neighboring tissues (Trainor and Krumlauf, 2000; Graham et al.,2004). In this context, slight shifts in the location of the crest–mesoderm interface are likely to have profound primary effects on craniofacial skeletogenesis (Schneider,1999; Helms and Schneider,2003; Kuratani,2005) and secondary consequences for the organization of muscles and other soft tissues.

The positive identification by Couly et al. (1992) of neural crest cells in the frontal and parietal bones is difficult to reconcile with these and earlier transplantation results. During avian neurulation, paraxial mesoderm is excluded from elevating and fusing neural folds throughout the mesencephalic and metencephalic regions, in large part because of the width of these brain regions that allows an extensive zone of close neuroepithelial-surface ectoderm apposition. However, this exclusion does not occur at more caudal regions, where the neural tube is narrower. At stage 8 (three somites), the hindbrain neural folds are vertical and underlying paraxial mesoderm is tightly adherent and, therefore, difficult to exclude from excised pieces of neural fold tissue without the use of proteolytic enzymes to separate epithelial from mesenchymal populations.

Couly et al. (1993) based their conclusions on the presence of cells containing the quail nuclear marker (Le Douarin,1973), as identified using Feulgen staining, but did not include assays using the QH1 antibody to detect quail endothelial cells. Angioblasts are ubiquitous within early mesodermal populations (Noden,1988,1991; Feinberg and Noden,1991; Borue and Noden,2004), absent only from notochordal and prechordal mesoderm (Noden,1990). Previous studies have found that the quail nuclear marker is unstable in angioblasts (Noden,1984,1991a). Therefore, Feulgen-based assays often fail to detect these mesodermal cells in chimeric tissues.

We occasionally observed small clusters of labeled chondrocytes within the cricoid cartilage after retroviral injection of somite 1. The boundary between the unsegmented mesoderm and the first somite is a notoriously unreliable and debated landmark in avian embryos. As described by Meier (1979), cells along the caudal margin of unsegmented head paraxial mesoderm often partially epithelialize, and many investigators define this as the first somite (reviewed by Huang et al.,1997). In our embryos, this partial epithelium breaks apart between stages 9–11, often forming one or two small vesicles that grossly appear fully epithelialized.

Caudal to this, the first true somite is small and also transient. With respect to origins of muscles and axial skeletal structures from this and the other occipital somites, our results generally agree with previous studies (Couly et al.,1993; Huang et al.,1997,1999,2000). Quail–chick transplantations identified lateral mesoderm adjacent to somites 1 and 2 as the source of laryngeal cartilages (Noden,1984). Given the small size and incomplete epithelialization of somite 1, it is likely that virus injected into the somitocoele of this somite leaked into adjacent mesenchyme. However, recent molecular labeling studies in mice suggest a neural crest origin for laryngeal cartilages (Matsuoka et al.,2005), as has long been postulated by comparative anatomists. Because both the methods and species are very different, it is not possible to reconcile these differences.

SUMMARY

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

As efforts continue to unravel the molecular dialogues between myogenic and skeletogenic populations in the head, and especially to identify similarities and differences between extraocular, branchial, glossal, epaxial, hypaxial, and appendicular muscular populations, the relative stability or changeability of tissue relations must be considered. The present results show that conserved relations between myogenic and skeletogenic populations are not paralleled by the maintenance of nearest-neighbor relations among individual cells. While these analyses do not indicate when signals directing the anatomical organization of individual muscles are initiated, they suggest that interactions occurring after myogenic mesoderm and skeletogenic crest cells arrive at their definitive branchial and periocular locations are critical for muscle morphogenesis. The molecular basis for these interactions remains elusive.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

Fertilized chicken eggs (Gallus gallus, White Leghorn strain) were obtained from Cornell University and Poyndon Farm (Herts, UK), then incubated at 38°C in a humidified incubator. Three sets of in ovo experiments were designed: (1) inject retrovirus into paraxial mesoderm at stages 8–9; (2) wash retrovirus over the surface the embryo, beneath the vitelline membrane, at stages 6 to 9.5; and (3) inject retrovirus at the interface between paraxial mesoderm and overlying neural crest cells at stages 10 to 11. Injections from sets 1 and 3 are site-specific, but the washes in set 2 do not target crest progenitors at any specific axial level.

Replication-deficient spleen necrosis retrovirus was produced in canine packaging cells designated CXL-D17 (Mikawa et al.,1991). In vitro titers of 5 × 105 to 2 × 106 CFU/ml were obtained. Initial studies used embryos from a line of SPF chickens maintained at Cornell University, but subsequent screenings revealed that K-strain birds from our open flock were free of endogenous viral antibodies, lacked potential helper virus sequences, and were equally susceptible to infection.

Embryos were exposed, stained with 0.05% neutral red, and staged according to routine procedures (Hamburger and Hamilton,1951; Noden,1983a). For injections (sets 1 and 3), the vitelline membrane was opened directly over the targeted site and one to five boluses of viral medium ranging in volume from 0.05 to 0.1 μl were introduced using pressure injection. We calculate that each embryo typically received fewer than 100 virions per bolus. Generally, a small, transient expansion of the embryonic tissue was visible at the site of injection. In some cases, a few crystals of fast green or India ink were added to the retroviral medium to assess the extent of spread from the injection site. These additions confirmed that the bolus remained localized close to the injection site.

The location of the injection site was mapped relative to identifiable landmarks in the central nervous system and somites, and measurements were taken using an eyepiece graticule. The exact depth reached by infective viruses was difficult to determine because the ejection force from the micropipet could not be defined. However, injections targeting superficial cells did not result in infection of progenitor populations known to be located further ventrally within paraxial mesoderm.

Injections into cervical somites were carried out in a similar way with the tip of the micropipette penetrating the outer epithelial layer of the somite. The virus did not spread to adjacent somites during the manipulation. Sham injections were identical, except for lacking virus. For surface washes (set 2), a small hole was made in the vitelline membrane overlying the area pellucida caudal to the embryo. A micropipet was inserted, and several microliters of viral medium containing 100 μg/ml of polybrene (Sigma) were introduced.

Mesoderm injections were made at all locations from the level of the rostral mesencephalon to somite 4. A few embryos were harvested at 1.5–6 days after injection, but most were maintained to stages 32–41 (9 to 15 days incubation). Embryos were rinsed in cold phosphate buffered saline then fixed for 1.5–4 hr in cold 2% paraformaldehyde. Older embryos were usually bisected and partially dissected to enhance penetration. Tissues were incubated overnight in the dark at 37°C in 0.1% X-gal (Molecular Probes, Inc.) with dimethyl sulfoxide at pH 7.4, and subsequently re-fixed in 2% paraformaldehyde. Each specimen was dissected and analyzed in whole-mount for the appearance of the blue reaction precipitate, with paraffin embedding and histology performed on representative tissues. No endogenous β-galactosidase activity was found in the regions of study in normal embryos (uninjected yet stained for β-galactosidase) or in embryos that received a sham-injection.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES

We thank Dr. Takashi Mikawa (Cornell University Medical College) for the generous gift of the CXL packaging cell line and guidance in generating retroviruses. The authors are indebted to Celia S. Fieler, Julie Fixman, Krista Halling, Gwyn Thomas, Sian Thomas, Michael Vilenchik, and Christina Wahl for technical assistance and also to Edith Koehn, Maria Laux, and Derek Scarborough for help with the histology. D.J.R.E. was funded by the Wellcome Trust; D.M.N. was funded by NIDCR and the NEI. We are indebted to the anonymous reviewers at Developmental Dynamics for their extensive constructive critiques and suggestions.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. SUMMARY
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES