The somitic involvement in the formation of the vertebral column was examined using the chick-quail chimaera model. Single cervical somites from quail donor embryos were transplanted into similarly staged chick host embryos. Following further incubation, serial sections of variously staged embryos were stained with the Feulgen reaction to distinguish the two cell populations.
Quail cells were generally located within a delimited region in one half of each of the two adjacent vertebrae, as well as in the intervening disc. The horizontal plane of division through each vertebra passed approximately through the centre of the body and divided the neural arch into rostral and caudal halves through the rostral border of the caudal notch. These results give support to the controversial theory of resegmentation, in which it was suggested that there is an apparent realignment of segmentation between the somite stage and the subsequent vertebral stage of development.
While it is well accepted that somites contribute to the formation of vertebrae and the intervening intervertebral discs, the precise relationship between them has long been controversial. Remak (1855) was the first to recognize an apparent realignment of material between the somite stage and the subsequent vertebral stage of development and formulated the theory of resegmentation that considers that a vertebra is formed by the combination of the caudal half of one bilateral pair of somites with the rostral half of the next caudal pair of somites. This theory has received support from several other studies (Von Ebner, 1888; Bardeen, 1905; Piiper, 1928; Dawes, 1930; Williams, 1942; Sensenig, 1943, 1949; Werner, 1971) although the precise contribution of the two pairs of somites to specific parts of the vertebra and disc is not clear (O’Rahilly & Meyer, 1979). The theory of resegmentation has often been challenged, however (most recently by Baur, 1969; Verbout, 1976,1985) and equally plausible alternatives theories have been proposed.
The methodology usually employed in the study of vertebral formation has consisted of serial sections of staged embryos. The inability to identify the cells derived from a single somite by this means has resulted in the development of widely divergent theories and has perpetuated the controversy regarding the details of vertebral formation. Although some limited experimental evidence in support of the theory of resegmentation has been obtained indirectly from studies of tetraparental mice (Moore & Mintz, 1972), cell polarity during development (Trelstad, 1977) and somitic transfers between chick and quail in investigations of-wing muscle determination (Beresford, 1983), there is also some recent experimental evidence that appears to dispute the theory (Noden, 1983; Dalgleish, 1985; Stern & Keynes, 1987). This study was undertaken to reexamine the issue of the somitic involvement in vertebral formation. To this purpose, use of a permanent cell marker system was employed to permit later identification of descendent cells from heterospecific single-somite transplants. We have replaced single somites in chick embryos at 2 days of incubation with those of 2-day quail embryos and observed the subsequent site of the quail cells in the vertebrae of the chimaeras at 10 days of incubation. In this way, we have been able to establish the somitic contribution to vertebral formation.
Materials and methods
Fertile chick (Gallus domesticas) and quail (Coturnix coturnix japonica) eggs were incubated at 37 °C for 42 and 48 h, respectively, to reach approximately stages 11 or 12 according to the system of Hamburger & Hamilton (1951) for chick embryos. A window was made in each of the host chick eggs and a small amount of black drawing ink diluted 1:1 with Pannett and Compton’s saline containing antibiotics (penicillin and streptomycin, 10 000i.u. ml-1, SIGMA) was injected subendodermally to permit visualization of the embryos. The vitelline membrane covering the embryo was then broken. A single somite from the future cervical region, usually within two or three of the last formed somite (see Table 1), was then excised using a tungsten wire needle and a micropipette. Pannett and Compton’s saline was added whenever required to prevent the embryo from drying out.
In preparation of the donor, the quail embryo was removed from the surface of the yolk and placed in Pannett and Compton’s saline. The embryo was of a similar stage to the previously prepared chick embryo (usually within one to two somite pairs [see Table 1]). The equivalent somite to the one removed from the chick (within two or three somites) was isolated from the quail embryo and transferred to the operation site in the chick using a micropipette. It was then manoeuvered into the correct position, using carbon particles as markers of somite orientation, although this method was not always a completely reliable one. The hole in the chick shell was then sealed using adhesive tape and the eggs reincubated to allow further development.
After further incubation ranging from 3 – 8 days, the surviving embryos were fixed in Zenker’s fluid, rinsed in running tap water and stored in 70 % ethanol. The cervical region of each embryo was then dissected from the rest of the body and processed for embedding in paraffin. Serial sections of 8 μm thickness were made in various planes and the sections stained with Feulgen method for demonstration of DNA to distinguish the quail from chick cells (Le Douarin, 1973).
The slides were observed and photographed using a Leitz orthoplan photomicroscope. The positions of the quail cells within the vertebrae were noted and recorded for each of the embryos. Specific attention was directed to the quail cell location in each of the vertebral elements (centrum, costal process, transverse process, lamina, pedicle, spinous process, articular processes and the intervertebral disc; Fig. 1).
Reconstructions using a series of contour drawings of the regions that contained quail cells were made of nine embryos that were considered to be representative of the total data.
From a total of 209 experiments, 21 embryos developed to stage 26-36 (Hamburger & Hamilton, 1951) in which some chondrification of the vertebral elements could be distinguished. In these embryos, quail cells were present in the cartilage of the developing vertebrae, in the paraxial musculature, in the spinal cord meninges, in the dermis and in the general connective tissue around the vertebral column. Integration of the quail cells was variable and in three embryos (31/2, 38/6, 52/5) appeared to be minimal. With few exceptions, the quail cells were found in regions only on the operated side of the embryo and in all embryos the quail cell regions were continuous with the adjacent chick cell regions, with no interruptions between the two populations. In all cases, the quail cells had formed portions of two consecutive vertebrae (see Table 2), although considerable variation of quail cell location was observed.
A wide variety of abnormalities was observed in the vertebral elements and in adjacent soft tissues within the quail cell area. In fact, no single embryo exhibited normal morphology of all of the structures and tissues in the operated region. Some of the abnormalities observed are indicated in Table 2 (Ab or F). They generally included intervertebral fusion, appearance of anomalous cartilaginous processes, reduced proportions of elements and complete absence of some parts of the involved vertebrae.
Centrum and disc
Quail cell distribution patterns were found to be most consistent in the centra and intervening intervertebral disc. 20 of the 21 embryos contained quail cells which comprised some or most of the adjacent quadrants of two consecutive centra and most or all of the corresponding half of the intervening disc (Figs 2, 3). This indicates that the central region of the medial sclerotome from a single pair of somites forms the intervertebral disc, and the cells rostral and caudal to the disc region merge with cells from the adjacent somites to form parts of the rostral and caudal centra. Each centrum, then, is a composite of cells from two consecutive somite pairs, each pair making approximately equal contributions and the disc forms bilaterally within one somite pair.
Ten of twenty embryos contained quail cells in the pedicles of two adjacent vertebrae. However, of these ten, only two exhibited relatively normal morphology of the two pedicles and continuous distribution of the quail cells. In two other embryos the quail cell content was high but the pedicles were fused. Five embryos displayed a large proportion of quail cells in the caudal pedicle and only a few quail cells in the rostral pedicle; however, two of these five had fused pedicles. One other embryo contained only a few quail cells in each pedicle. Of the remaining ten embryos, six contained a large proportion of quail cells in the caudal pedicle and none in the rostral pedicle. Few contained quail cells in the rostral pedicle when the caudal pedicle was missing. Therefore, in summary, eighteen of the twenty embryos either contained few or no quail cells in the rostral pedicle, or exhibited a major anomaly of the pedicles (both fused or caudal pedicle absent). In particular, whenever the caudal pedicle was present it contained quail cells (16 of 16 embryos) (Table 2, Fig. 4).
These results suggest that cells from a single somite contribute mostly to the caudal pedicle and to the base of the rostral pedicle (dorsal extension of the centrum). Somitic contribution to the rostral pedicle normally would not extend past the rostral border of the caudal notch of the rostral vertebra, consistent with the pattern seen in other parts of the neural arch.
Quail cells were present in portions of both of the adjacent laminae in all twenty embryos (see Fig. 5).Twelve embryos contained a large proportion of quail cells in both rostral and caudal laminae but seven of these demonstrated at least partial fusion of these elements. In five embryos, the caudal lamina contained numerous quail cells and the rostral lamina only a few. The reverse situation was present in only one embryo. Two embryos contained only a small amount of quail cells in each of the laminae. From the serial reconstructions (Figs 7, 8) of the embryos, it could be seen that the quail cell region extended only a short distance into the more rostral lamina. In particular, quail cells were found in only a few sections (approximately 100 µm or less) rostral to the rostral border of the caudal notch. As this was the original rostral boundary of the somite (position of the spinal ganglion) most likely the normal contribution to the rostral lamina would be relatively small. Conversely, the single somite contribution to the caudal lamina should be relatively large. Cells from a single somite, then, normally should form parts of two adjacent laminae, but a far greater portion of the more caudal lamina.
The distribution patterns in the spinous processes were inconsistent, possibly related to the high incidence of abnormalities associated with these elements and to the lack of chondrification in ten embryos. In nine embryos, quail cells were not present in either of the two spinous processes or their precursors. Seven other embryos, however, contained a large proportion of quail cells in both spines or precursors but of these, four embryos displayed some degree of fusion of the two spinous processes and the other three had not yet formed complete spinous processes on the involved vertebrae. In three embryos, quail cells were present to a large extent in the region of the caudal spine but only minimally, or not at all, in the rostral spine. One embryo contained just a few quail cells in each of the two spinous processes. On the basis of these results no firm distribution pattern of quail cells in the spinous processes could be established although, perhaps, it should be closely associated with that of the lamina, as each spinous process seemed to be an extension and fusion of the two bilateral laminae.
Fifteen of the twenty-one embryos contained quail cells in each of the costal processes of the two adjacent vertebrae. Of these fifteen, however, eleven cases were found in which the rostral costal process contained quail cells only in the caudally projecting tip. In the remaining six embryos, four were missing one of the two costal processes (the existing costal process contained a large number of quail cells) and two exhibited quail cells only in the caudal costal process. These results suggest that cells from a single somite make up almost the entire caudal costal process and also make a small contribution to the caudal tip of the rostral costal process.
The situation here was very similar to that already described for the pedicles. Adjacent transverse processes contained at least some quail cells in ten of twenty cases; but in one of these embryos the transverse processes were fused. In contrast, seven of the remaining ten embryos demonstrated quail cells in only the caudal transverse process, although the rostral process was present, while each of the other three embryos were missing the transverse process of the caudal vertebra. Therefore, as in the pedicles, in fifteen of twenty embryos examined, either few (five cases) or no (seven cases) quail cells were found in the rostral process, or the caudal transverse process was absent (three cases) and whenever the caudal transverse process was present, it was found to contain quail cells (17 of 17 embryos). Furthermore, it was noted in many embryos that the quail cell distribution in the transverse processes frequently matched that of the pedicles, possibly indicative of developmental synchrony in these two elements. It would appear, therefore, that the contribution from a single somite is mainly to the caudal transverse process with a minimal contribution, if any, to the rostral transverse process.
Fifteen of twenty embryos contained quail cells in both of the adjacent articular processes (i.e. the caudal articular process of the rostral vertebra and the rostral articular process of the caudal vertebra) (see Fig. 6). However, in twelve of these cases, the involved elements demonstrated at least partial fusion of the elements. Surprisingly, in six embryos, quail cells were also noted in the caudal articular process of the caudal vertebra, but were not present in the rostral process of the rostral vertebra in any of the embryos. Of two embryos without quail cells in adjacent articular processes, one exhibited an unusual pattern of quail cell distribution in the centra and the other contained no quail cells in any of the articular processes. The continuity of the cell distribution across the facet joint was also apparent from the composite reconstructions and it would appear that adjacent articular processes, as well as all the tissues that form the joint surfaces and joint capsule, are formed of cells from a single somite.
The quail cell distribution was readily apparent from the contour drawings made from the serial sections of nine of the embryos. Examples of these reconstructions from two of the embryos are shown in Figs 7-10. Fig. 7 shows caudal views (A,C), a rostral view (D), and two caudal oblique views (E,F) of the caudal vertebra of two adjacent vertebrae that contained quail cells. It also shows a rostral view of the region within the vertebra occupied by the quail cells (B). The corrresponding views of the rostral vertebra of the same embryo are shown in Fig. 8. Fig. 7 clearly shows the quail cells contained only within a rostral quadrant of the vertebral body and not extending over the midline (C,D). The quail cells can be seen occupying all of the pedicle adjacent to the centrum (F) and are distributed further into the transverse process (C,E). All of the accompanying costal process can be seen to contain quail cells apart from a small area at its tip (E). As the quail cell distribution enters the lamina, it gradually becomes more dorsal in the rostral sections, although only a few quail cells in this specimen can be seen in the spinous process (E). The relative proportion of quail cells in each section of the lamina also becomes less as the sections become more rostral (E). There are no quail cells to be found in the caudal articulating process (E,F) whereas the rostral articulating process is formed entirely of quail cells (E).
Fig. 8 shows the corresponding views of Fig. 7 of the rostral vertebra of the two adjacent vertebrae that contained quail cells. In this example, the quail cells can be seen to occupy only a caudal quadrant of the vertebral body (C,F) with few or no quail cells being found in the pedicle and transverse process (B-E). However, there are a few quail cells to be found in the tip of the accompanying costal process (E). As the quail cells continue around the neural arch, they populate the caudal half of the lamina entirely, including the spinous process (E,F). The caudal articulating process is also composed entirely of quail cells (E,F).
Fig. 9 is an expanded caudal oblique view of the two vertebrae shown in Figs 7 and 8 when combined. The distribution of the quail cells can clearly be seen to be continuous between the two vertebrae when the vertebral elements are aligned, particularly in relation to the lamina and articular processes. Also conspicuous is the small contingent of quail cells found in the tip of the costal process of the rostral vertebra.
Fig. 10 is a contour drawing of coronal sections of another embryo. The area of quail cell distribution is shaded and clearly demonstrates that it is restricted to one side of the vertebral column and is almost equally divided between adjacent rostral and caudal halves of two vertebral bodies. The distribution can also be seen in the adjacent halves of the corresponding neural arches.
The results of this study clearly support the concept of a resegmentation between the somitic stage of development and the resulting vertebrae as first proposed by Remak (1855) and supported by a number of other investigators, including Piiper (1928), Williams (1942) and Sensenig (1949). A single somite has been shown to contribute to the caudal half of one vertebral body, the intervening disc and the rostral half of the next vertebral body. This division applies not only to the vertebral bodies but also to the neural arches with single somites contributing to caudal and rostral portions of adjacent neural arches, although the division is not quite so clear as with the vertebral bodies. In the neural arch, the major somitic contribution is to the caudal arch with contribution to the rostral arch being limited to a transverse plane situated approximately through the rostral border of the caudal notch. Fig. 11 summarizes these conclusions and extrapolates them to the expected contribution made by a single somite to ossified vertebral formation based on our results. It is interesting to note that the single somitic contribution (the quail cell distribution) has clearly defined limits which appear to follow the geometry of the vertebral column rather than the anatomy. This is particularly evident where quail cells are found in the tip of the costal process of the rostral vertebra. It suggests that quail cells differentiate according to their spatial position and warrants further study, particularly of the associated intervening tissues that are somitically derived.
The methodology employed in this study is not without problems which may have affected the results and their interpretation. For example, it has been shown by Strudel (1966) and ourselves (Bagnall et al. 1986) that the embryo can recover completely from the removal of single somites and produce a normal vertebral column. The site and mechanism of this regulation is unknown and, therefore, it could not be determined whether the methodology employed here produces a vertebra composed of quail somitic cells and chick cells derived from regulation. Furthermore, it has also been reported (Kieny et al. 1972) and shown (Bagnall et al. 1986) that the procedure for removal of somites often results in small pieces of somitic material being left behind inadvertently. Further studies (Bagnall et al. 1987) have shown these pieces to be capable of reorganization into a somitelike unit with a reduced volume. While every attempt was made to remove all of the chick somite and replace it solely by an intact quail somite, there is no guarantee that this was achieved in every case. The consequences of potential cell interactions between remnant chick somitic tissue and extraneous quail somitic tissue are unknown.
Kieny et al. (1972) using somite transfer techniques demonstrated that the somites are regionally determined prior to the time of their formation. The degree of regional determination was restricted in their study to differences between cervical and thoracic vertebrae, with vertebrae having cervical characteristics being produced in the thoracic region and vice versa. Whether or not this determination can be applied to consecutive vertebrae is unknown but does have implications for this study. The levels of the extirpated and donor somites were generally within two or three of each other. The effect of such a small difference in vertebral level on future development is not known, but we considered that it would be minimal in the midcervical region where most of the surgical procedures were performed. Transitional differences between consecutive vertebrae at this level are slight compared to those differences between more widely separated levels and between the different regions of the vertebral column.
In a series of experiments, Stern & Keynes (1986, 1987) have shown that the somite is further differentiated into presumptive rostral and caudal sclerotome halves early in its formation and have suggested that this difference is determined at the time of segmentation (Stern & Keynes, 1986). Furthermore, in experiments in which similar somite halves (rostral or caudal) have been situated adjacent to each other, Stern & Keynes (1987) have found that sclerotome cells from like halves mix with each other while those from unlike halves do not. This suggests that correct orientation of the donor quail somite is important, since incorrect orientation with subsequent fusion of cells from similar sclerotome halves might produce abnormal vertebrae perhaps similar to some that we encountered. However, the stage of development at which somitic cells become committed to forming specific vertebral elements is not known, nor is the stage at which they become committed to forming specific vertebrae. In addition, Gallera (1966) showed that the somite is not yet differentiated into sclerotome and dermatome at the time of its formation, although it is possible that determination of rostral and caudal halves of the somite on the one hand, and dermatome and sclerotome, on the other, are separate events. While every attempt was made to maintain the correct orientation of the transferred quail somite, it is uncertain and unlikely that this occurred in every instance. The development of approximately normal vertebrae, assuming incorrect orientation of the somitic axes, raises questions concerning the cell determination that has already occurred and warrants further study.
Chevallier et al. (1977), when investigating muscle function in the wings of chicks using chimaeras, found a small but significant difference in results when transferring donor chick to quail host as compared with transferring donor quail to chick host. These differences were attributed to differences in the developmental rates between chick and quail. The more mature quail cells might behave differently in a less-mature chick environment. In addition, Sanders (1986) found differences in cell surface properties (adhesiveness) between homologous cells from these two species and suggested that they could possibly explain some of the occasional inconsistent results using this methodology. The effects in the present study of potential disparity in developmental rates and cell adhesiveness are uncertain.
Kieny et al. (1972), in their experiments to investigate regional transfer of somites, suggested that, in those cases where completely normal regional development had occurred, the transplanted somite had been rejected and that the normal regulatory process of chick embryos had been activated. The question of selfregulation following injury and removal of somite by the embryo has been discussed earlier, but the question of rejection of the transplanted somite, either in whole or in part, by the host must be considered, even though it was not assessed in this study. The one case (53/2) of complete absence of donor quail cells in the developing chick embryo might demonstrate complete rejection of the transplant by the host and subsequent regulation. Differing degrees of rejection, in whatever form, must be considered to have affected the results.
A final problem relating to the validity of the development of chimaeric vertebrae, centres on the general trauma associated with surgical intervention of this kind. It is conceivable that alteration of the embryonic environment by the surgical procedure, particularly that which is local to the site of extirpation, could affect subsequent vertebral development.
From the preceding discussion, it is clear that there are many problems associated with the methodology employed that could affect the validity of the results. However, we believe that the general consistency of our results indicate that they were valid and that the methodological uncertainties could explain the relatively small differences found in quail cell distribution, the abnormalities in vertebral formation that were found and the variation in the number of quail cells present.
The phenomenon of resegmentation between the obvious somitic segmentation in the early embryo and the vertebral segmentation in the more mature embryo is intriguing, particularly as many authors consider this phenomenon to be common to all vertebrates. The resegmentation of a fundamental, underlying metameric pattern has been studied in other species. Lawrence and coworkers (Lawrence & Martinez-Arias, 1985; Martinez-Arias & Lawrence, (1985), studying segmentation patterns in the fruit fly (Drosophila melanogaster), have found that the most obvious embryonic segmentation pattern is not the true underlying metameric pattern. They have found that the more obvious segmental embryonic pattern is shifted by half a segment in relation to the metameric pattern. If this concept is applied to somites and the subsequent formation of vertebrae, then a remarkable similarity becomes apparent (Stern & Keynes, (1986). The most obvious embryonic segmental pattern (the somites) appears to be one half segment out of alignment with the adult segmental pattern (the vertebrae). Perhaps our attention should be shifted one half segment from the somites to focus on the intersomitic clefts coupled with a cluster of cells lying rostral and another cluster lying caudal as being representative of the metameric pattern in the mature animal. In this way, a somite would contain cells from two separate segments and the appearance of the intersomitic cleft would be highlighted as it simultaneously defines the caudal half of one somite that has just formed and the rostral half of the next one (Stern & Keynes, 1986, 1987). Application of this concept would mean that the vertebrae would not be out of alignment with the underlying metameric pattern of the embryo and the conceptual difficulty of ‘resegmentation’ will have been circumvented. The implications of this concept are numerous and perhaps warrant further study particularly if the field of genetic expression is based on segmentation patterns.
This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (K.M.B.), and the Medical Research Council of Canada (E.J.S.).