The formation of many organ primordia involves changes in the epithelial cells, which give the organ its characteristic shape. For instance, the neural plate elongates and neural folds elevate to form a tube (Schroeder, 1970), whereas the salivary placode invaginates to form a stalk with a bulbous tip (Spooner, 1973). On the other hand, formation of the lens (Zwaan and Hendrix, 1973), optic cup (Hilfer, 1983), and thyroid (Hilfer, 1973) involves invagination to form cup-shaped structures, which are distinctly different in shape. It seems likely that organ primordia that display characteristic modes of invagination may also display characteristic types of cell movements and cell–cell interactions that contribute significantly to their morphogenesis. For instance, during formation of the neural tube, cells move medially to intercalate with cells along the midline (Burnside and Jacobson, 1968; Schoenwolf and Sheard, 1989). Individual cell movements have also been shown to contribute to notochord formation (Keller et al., 1989) and to gastrulation in both the sea urchin (McClay et al., 1991) and Xenopus laevis (Keller and Danilchik, 1988). In these processes, it is apparent that individual cell movements drive the change in shape of the cell population from a broad sheet to a longer, narrower configuration.
In the thyroid, several observations suggest that development may proceed as a result of cell behaviors that establish cellular domains within the early primordium. The placode evaginates into the space between the ventral aortic roots. As early as stage 14 (Hamburger and Hamilton, 1951), the thyroid possesses concentric circular indentations in its basal surface, which correlate with the position of cells containing longitudinally arranged filaments not found in other cells of the primordium (Hilfer, 1973). These cells above the basal grooves compose a two- to four-cell-wide region of the primordium that does not incorporate tritiated thymidine (Smuts et al., 1978). Moreover, mitotic activity is lowest in the center of the primordium and gradually increases toward the periphery. These regional differences in cytoskeletal structure and mitotic activity point to the possibility that evagination of the thyroid involves cell behaviors that differ in discrete cellular domains.
Although the morphology of the thyroid primordium and the cells that compose it have been studied, little is known about the types of behaviors displayed by these cells (Hilfer et al., 1990) and what, if any, role individual cell movements may play in thyroid morphogenesis. We used the prominent cell apices of the primordium as well as cell markings to track the movements of individual cells through time lapse video recordings, coupled with scanning electron microscopy. Analysis of the images was aimed at identifying the types of behaviors displayed by the cells and any regional differences that exist with regard to those movements. We found that cells of the thyroid placode became rearranged into subgroups, or clusters. Shaping and deepening of the primordium occurred concurrently, involving gradual incorporation of peripheral regions of the pharyngeal floor into the walls of the thyroid pit. We suggest that individual cell movements and interactions between cell clusters are necessary for the thyroid to evaginate into the confined space of the pharyngeal floor.
MATERIALS AND METHODS
Culture and Microscopy of Living Preparations
White Leghorn eggs (Truslow Farms, Chestertown, MD) were incubated at 37°C in a commercial egg incubator for 47 hr (stage 12) to 72 hr (stages 17 and 18). All staging was according to Hamburger and Hamilton (1951). The intact embryos were removed from the yolk, and the region of the pharynx between the mandibular arch and the caudal attachment of the heart was isolated from the embryos. The neural tube and pharyngeal roof were removed, and the isolated ventral pharynx was transferred to a 35-mm Falcon tissue culture dish containing fresh Hank's saline, positioned lumen side up. The preparation was held in place with short segments of 210-μm-diameter platinum wire placed over caudal regions of the pharynx away from the thyroid, allowing the arches to follow their natural tendency to curl up.
This preparation was placed on the stage of a Zeiss Photomicroscope II (Carl Zeiss, 7082 Oberkochen, W. Germany) equipped with a warming device, designed in-house, to keep the tissue at 37°C throughout video recording, photographic sessions, and experimental manipulation, which typically lasted from 1 to 6 hr. Images were captured from phase and Hoffman modulation contrast optics (Modulation Optics, Greenvale, NY). Under these experimental conditions, thyroid development proceeded in a normal manner. At the end of recording sessions, the primordia had the same depth and contours as primordia of similar age developed in ovo, and cell movements did not appear to be impaired.
Tracking and Analysis of Cell Movements
Several methods were used to record cell movements. (1) Cells of the thyroid exhibit prominent apical blebs, which are clearly visible with modulation contrast. This method permitted imaging of individual cells without intervention. For time-lapse recordings of cell movements, images were taken at 20-, 30-, or 60-sec intervals with a Panasonic solid state CCD camera. The images were stored by using a Panasonic Optical Memory Disk Recorder (OMDR) model 2021. (2) Groups of cells were labeled with the fluorescent carbocyanine dye, 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI). DiI was dissolved in dimethyl sulfoxide (DMSO) (Fisher Scientific, Fair Lawn, NJ) to a concentration of 0.5 μg/ml. The tip of a tungsten needle was used to transfer a small amount of the dye to patches of cells around the periphery of the primordium. The pharyngeal preparations were kept at 37°C on the microscope stage and photographed at half-hour intervals over periods of 3 hr, for stages 13 and 15, and 5.5 hr for stage 14. DiI is stably incorporated into cell membranes (Honig and Hume, 1986; von Bartheld et al., 1990), and both its color (red) and fluorescent qualities allowed for the tracking of cell movements based on movement of the dye in both brightfield and fluorescence. (3) Cell movements were also tracked by labeling groups of cells with carbon particles. India ink, dissolved in Hanks' saline, was transferred on the tip of a tungsten needle to specific regions of the primordium. India ink was used because it could be easily seen in black and white video recordings and photographs, thereby acting as a point of reference as well as allowing for observation of cell movements. In some cases, India ink was used as a reference point in the deepest part of the primordium, whereas DiI was used to label cells at the periphery.
The movements of individual cells were tracked (Hilfer et al., 1990) by using a high-resolution touch screen (Personal Touch, Edmark Corp., San Jose, CA) and image analysis software developed in-house (Silage and Gil, 1982; Sheffield et al., 1986). The X,Y coordinates of individual cell apices or dyespots were traced through succeeding frames of the time-lapse recordings. The data from the image analysis program were processed with a Lotus spreadsheet (Lotus Development Corp., Cambridge, MA). Graphs of the cell movement paths were plotted on Axum (TriMetrix, Inc., Seattle, WA) by using the data from the Lotus spreadsheet.
In some instances, video recordings of unlabeled primordia were reviewed frame by frame to track the movements of individual cells and cell clusters. In these instances, video images of various time points in the recording were made with a Mitsubishi Video Copy Processor (Cypress, CA). These images were used to mark the positions of specific cells as an aid to keeping track of the positions of these cells on the video monitor.
Scanning Electron Microscopy
Embryos were fixed for 30 min in 2.0% glutaraldehyde in phosphate buffer, pH 7.4 (Hilfer et al., 1986). The tissue was post-fixed in 1.0% phosphate-buffered osmium tetroxide for 1 hr, followed by serial dehydration in graded acetone. Samples were then critical-point dried from liquid CO2, sputter coated with gold (Seevac, Pittsburgh, PA), and examined with a Phillips 501 scanning electron microscope (SEM).
Living preparations were photographed by using a Zeiss Photomicroscope II equipped with a 50-watt, 12-V halogen light source, phase, brightfield and Hoffman modulation contrast optics, and epifluorescence. Photographs of living preparations were taken with Kodak 2415 Technical Pan film developed in Technidol or Kodak Plus X 125 film developed with Rodinal. All Scanning EM photography was done with Kodak 2415 Technical Pan film developed in Technidol.
The model was constructed with Amorphium (Play, Inc., Rancho Cordova, CA). A flat, oval solid was formed, and clusters were constructed on its surface with a small distortion tool. Pressure was then exerted on one side to cause evagination. Each stage in the construction was saved before the model was modified with construction of new peripheral clusters and additional pressure applied. The result was a series of images built in stages from the original model.
Organization of the Thyroid Primordium
The pharyngeal floor shows essentially the same morphology, whether it is prepared after fixation of the entire embryo (not shown; Hilfer and Brown, 1984) or fixed after isolation and maintained as a three-dimensional culture (Fig. 1). At stage 13, the thyroid forms a slight depression in the pharyngeal floor at the level of the second pharyngeal arch. This patch of cells is approximately 100 μm in width and is distinguishable from the surrounding pharyngeal tissue by its more extensive apical blebbing and tighter packing of cells (Fig. 1a,b; Shain et al., 1972). The cell apices tend to form linear arrays with a circular bias (Fig. 1b). By stage 14, the thyroid is composed of concentric rings of tissue that form a pit that is more depressed rostrally then caudally (Fig. 1c,d). The walls of the pit take on a jagged appearance by stage 15, due to a series of folds or buckles in the pit walls (Fig. 1e,f). The position and orientation of the folds in the images of the living primordium (Fig. 1e) are remarkably similar when the same specimen is observed with the SEM (Fig. 1f). Narrowing of the thyroid pit creates a cylindrical depression surrounded by elevated rings of tightly packed and apically blebbed cells, which can be seen in both light and SEM micrographs of stage 16 primordia (Fig. 1g,h). When measured at its midpoint, the total cross-sectional width of the primordium increases from 100 μm to 133 μm between stage 13 and stage 16. This size increase includes those cells of the most peripheral ring of tissue surrounding the pit.
Thyroid cells become organized as clusters.
Cells of the thyroid were observed to form small groupings demarcated by deep folds and shadows. These cell clusters were first evident in thyroid primordia at stage 13, in which the clusters were composed of a mixed population of cells having broad flat apical surfaces along with some with blebbed apices (Fig. 2a,b). In light micrographs of living preparations, cell clusters were delineated by spaces between groups of cells (asterisks in Fig. 1a,c,e,g). In many cases, cells tended to form a linear array within clusters along points where bending occurred within the primordium. These cells also displayed prominent apical blebs (Fig. 1b,d). By stage 15, clusters were smaller and composed of cells displaying narrower apical surfaces and more extensive apical blebbing Figure 2c,d). However, the number of cells per cluster varied within each primordium, and no relationship was found between the number of cells per cluster and the age of the primordium (Table 1).
Table 1. Number of cells per cluster in the developing chick thyroid
Staging is according to Hamburger and Hamilton (1951).
Cell counts were taken from scanning electron photomicrographs.
15 ± 6.5
19 ± 9.4
15 ± 8.9
14 ± 4.8
15 ± 2.9
In summary, it was shown that thyroid organogenesis proceeds from a placode that forms a shallow depression in the pharyngeal floor at stage 13 to a deep, narrow pit surrounded by a mound of tightly packed, apically blebbed cells at stage 16. The primordium is composed of cells grouped into clusters. These cell clusters are evident from stage 13 through stage 16 and appear to be the result of creases or folds that segregate one cell group from another. The apical surface area of the cells within the clusters decreases between stages 13 and 15, as the cells become more highly blebbed and tightly packed. Folds form in the wall of the pit, creating a jagged slope from the pharyngeal floor to the bottom of the thyroid pit.
Thyroid evagination is not caused by convergent extension.
The movements of 158 individual cells were tracked from time-lapse video recordings of several different thyroid preparations. Of these cells, 87% moved in a zigzag or circling manner, referred to as random movement, with the remainder (13%) displaying movements that approximated a straight line, referred to as bulk movement (Fig. 3). Moreover, individual cells did not exhibit movements indicative of convergence toward the central pit of the primordium. Forty cells at the margins of three different thyroid primordia were selected to trace directional movement. Only 22% of these cells moved toward the center of the primordium (Table 2). Individual cells within the population that was measured tended to move short distances of two cell diameters (one cell diameter = approximately 3.5μm) or less over time spans that averaged approximately 2 hr. Only 56 of the 158 tracked cells moved farther than two cell diameters over such time spans. These short distance movements resulted in changes in the positions of one or two cells with respect to neighboring cells within a small grouping. These types of cellular rearrangements were observed to occur throughout the primordium, but were more clearly evident in marginal regions.
Table 2. Direction of movements exhibited by cells at the margin of the early thyroid primordium
Direction of movement
Percentage of total cells displaying movement
Examples of each of these typical cell movements are shown in Figure 4. Directional movement is shown in Figure 4c. The cells were located within a cell cluster in the outermost ring comprising a stage 13 primordium, shown at the start (Fig. 4a) and the end (Fig. 4b) of the recording period. At the start of recording (Fig. 4a,c), the cells were positioned as a curved row in the cranial region of the primordium. Thirty minutes later, the row had straightened due to the movements of the cells at the ends of the row. Between 65 and 75 min, two of the cells in the center of the row had moved in a caudal direction away from the row and at 90 min were partly hidden from view by their inward movement. Two cells in a group at the craniolateral edge of the same thyroid (Fig. 4a) provide an example of circling movement. Although other cells of this group moved to a more central and caudal position in the primordium, cell 7 moved from an initial position cranial to 6 to a caudal position (Fig. 4e).
Clusters move into the thyroid pit.
In addition to the movements of individual cells, cell clusters also were observed to move and change shape. The clusters labeled x, y, and z, near the cranial edge of the central region in Figure 4a, were tracked for as long as they remained in focus. Figure 4d shows enlargements of those clusters at selected time points. Over the 90-min time span, the clusters changed shape and shifted their position such that cluster x moved from a more lateral position, relative to clusters y and z, to a more caudal position. This change in position resulted in the three clusters being more closely apposed. As the clusters rearranged, they annexed adjacent cells and a fold in the wall of the primordium distorted the clusters.
In summary, most individual cells of the thyroid were observed to display short-distance random movements. A small percentage of the cells that were tracked displayed short-distance directional movements that resulted in the repositioning of one or two cells within a grouping. Clusters of cells also changed shape and position within the thyroid. The movement of groups of cells was the most prevalent type of cell movement in the early thyroid primordium.
Shaping the Thyroid
A major force in formation of the thyroid pit is the gradual incorporation of adjacent pharyngeal cells into the deepening primordium. This movement was examined by labeling groups of cells with DiI. As an example, when cells just cranial to a stage 14 thyroid were labeled with DiI and observed over a 2.5-hr time span, the labeled region moved caudally and laterally (Fig. 5). The dye initially was localized to one group of cells within the outermost ring of thyroid and an adjacent group of cells within the pharynx. After a half hour, the inner group of cells had moved to the edge of the tilted region and, by 2.5 hr, was perpendicular to its original orientation and in the wall of the pit. Meanwhile, the labeled pharyngeal cells had become part of a new ring of thyroid cells at the edge of the pit. Concurrent with this movement was a marked increase in the presence of apical blebbing in the labeled region (Fig. 5c–e) along with movement of the caudal-most portion of the labeled region below the plane of focus (Fig. 5e).
The relationship between deepening of the primordium and folding of the tissue was visible in light and scanning electron micrographs of normal embryos. By the end of stage 14, the outer ring had formed a series of elevated projections (Fig. 1c,d). At stage 15, the thyroid had deepened considerably and a series of folds jutted out from its walls (Fig. 1e,f). These folds or buckles in the walls of the primordium were even more evident in SEM micrographs of stage 15 thyroids (Fig. 1f).
The folds also contribute to continued deepening of the thyroid. In video recordings where a single plane of focus within the thyroid was maintained throughout the recording period, groups of cells were observed to move below that focal plane in conjunction with the folding movements. Figure 6 follows this process in the cranial region of a stage 13 primordium, kept in normal orientation, over a 128-min time span. Initially, all cells within the folding region (E and V in Fig. 6) were within the plane of focus forming a shelf above the thyroid pit. After 15 min, this shelf had moved inward several cell diameters. At 95 min, those cells that made up the edge of the craniomedial portion of the folding tissue, along with those cell that were within the V-shaped region created by the folding, were out of the focal plane. The more peripheral region now formed a new ring close to the edge of the pit. At 128 min, the fold was no longer V-shaped, due to convergence toward each other of the two regions that formed the “arms” of the “V” and the cells that were enveloped by the V-shaped fold were clearly below the plane of focus.
In summary, pharyngeal cells directly adjacent to the thyroid are gradually incorporated into the primordium. Shaping of the early thyroid primordium is effected by folding movements that initially create buckling of the tissue. This is followed by the repositioning of cell clusters resulting in a deepening of the pit within the primordium.
We have shown that cells of the early thyroid primordium are arranged in subgroups or clusters and that form is acquired through the incorporation of peripheral cells into the walls of the primordium concurrent with cell movements that deepen the thyroid. Cell rearrangements of the type observed in sea urchin (McClay et al., 1991) and Xenopus gastrulation (Keller and Danilchik, 1998), neural tube formation (Burnside and Jacobson, 1968; Schoenwolf and Sheard, 1989), and notochord elongation (Keller et al., 1989) were not observed in this system. However, shape changes in and translocation of clusters of cells (Figs. 4, 6) in conjunction with the short distance movements of individual cells (Figs. 3, 4) were observed.
The folds that jut out into the lumen of the thyroid pit at stages 14 and 15 (Fig. 1c,e) are no longer present at stage 16 (Fig. 1g,h), at which time the former conical pit is a narrow tube-shaped structure (Shain et al., 1972; Hilfer, 1973). Cell rearrangements observed in this system were sparse and involved only a few cells within a small grouping. Figure 5 shows cell apices that rearranged to form a somewhat linear array in conjunction with changes in the shape of the cell cluster. This change in cluster shape is expected to facilitate cluster movements in much the same way that change in cell shape facilitates cell movements. Clusters of epithelioid cells studied in vitro have been shown to change shape and to exhibit directionality of movement (Kolega, 1981), indicating that clusters of cells can act in a manner consistent with active participation in the morphogenesis of an epithelium (Fristrom, 1988; Honda et al., 1982). Deepening of the thyroid pit was concurrent with changes in the positions of clustered groups of cells (Figs. 5, 6). Moreover, on closer inspection of SEM micrographs of stage 15 primordia (as in Figs. 1f, 2c,d) a small cluster of cells can be seen in the indentation between two clusters that appear to be squeezing in closer together. The effect of this type of cluster rearrangement would be narrowing of the circumference of the buckled ring of cells, while simultaneously deepening the primordium due to the stacking effect of cluster rearrangements that occur in a vertical plane. This behavior is similar to the boundary shortening procedure described by Honda et al. (1982) for rearrangement of individual cells, only viewed looking down on a vertically oriented sheet of cells.
This restructuring of the thyroid could also be a passive response to compressive forces acting to move cranial and caudal regions of a buckled cell population closer together, forcing cells positioned between the two regions to a lower plane in the thyroid wall. This type of mechanism acting alone would be expected to produce bending (Bard, 1990) in the plane of least resistance. However, this is not consistent with the observed morphogenesis of this system. Cells, that formed the v-shaped region where bending was occurring actually relocated to a lower plane within the wall of the thyroid pit (Fig. 6), which suggests a cellular rearrangement process is operative rather than simple passive bending.
We have previously suggested that the basal grooves could act as hinges that allow bending within the thyroid in response to exogenous forces (Hilfer, 1973). A role for basal grooves as hinges is not ruled out by the results of this study. During the period from stage 13 through stage 15, the thyroid is confined to a space that is bounded laterally by the developing pharyngeal arches, pouches, and associated arterial structures (Shain et al., 1972; Hilfer, 1973) and is immediately surrounded by a more mitotically active population of pharyngeal cells (Smuts et al., 1978). The force exerted by these tissues would be expected to contribute to bending that accentuates the curvature of the primordium and hinges could act to permit this bending to occur within the confines of the allotted space. The dual role of basal groove as hinge and cells that form the groove as barriers to expansion could act in shaping the thyroid in the following manner. In response to size increase due to cell division and to force exerted by surrounding tissues, a population of cells located between two basal grooves, prevented from expanding beyond the grooves, is expected to buckle in a manner similar to what has been suggested for rhombomere formation (Guthrie et al., 1991). In the manner proposed for neural fold elevation (Schoenwolf and Smith, 1990), the basal grooves, acting as hinges, would bend in response to force exerted by surrounding tissues, shifting the buckled ring of cells to a position facing the thyroid lumen. This finding is consistent with the morphogenetic movements observed in this system (see Fig. 5).
It is possible that the basal grooves and the cells that form them act only to stabilize the morphology of the developing thyroid. The longitudinally arranged microfilament bundles of these cells (Hilfer, 1973) resemble the stress fibers of cultured cells, which are associated with stabilizing effects (Bard, 1990). However, the temporal and spatial arrangement of the grooves argue for the more active role in thyroid morphogenesis. The first groove begins to form in the cranial region of the primordium at stage 14 (Hilfer, 1973), a time when the first ring of pharyngeally derived cells is evaginating (Fig. 5). Moreover, it was observed in this study and suggested by earlier work (Shain et al., 1972; Hilfer, 1973) that the thyroid develops in a cranial to caudal manner. Elucidation of the mechanism by which the basal grooves form could provide a clearer understanding of their role in thyroid morphogenesis.
A Model for Thyroid Morphogenesis
This study has provided some insight into the cell movements that shape the early thyroid and forms the basis for a model of early thyroid morphogenesis (Fig. 7). At stage 13, the thyroid consists of the original placode (zone 1), formed at stage 12, surrounded by a ring of cells (zone 2) undergoing conversion from pharyngeal to thyroid. This conversion process involves subdivision of the cell sheet into clusters of cells beginning with the cells closest to the placode. Cell division within the placode and force exerted against it by the more mitotically active neighboring cell population, cause the placode to buckle ventrally. Bending at the first basal groove contributes to buckling of zone 1 and elevates zone 2.
By stage 14, a third ring of pharyngeal cells is undergoing conversion to thyroid (zone 3). Cluster rearrangements have started to narrow the circumference of the placode, which accentuates bending at the first basal groove. Bending occurs at the second basal groove in response to population pressure from the more rapidly dividing zone 3. This bending positions the buckled zone 2 so that it faces the lumen and it elevates zone 3. Cluster rearrangements in zone 2 have already begun to narrow its circumference. These cell movements also deepen the primordium, because the rearrangement occurs in a vertical plane. Zone 3 buckles and bends in a manner similar to zone 2 and cluster rearrangements begin to reshape this zone. Cells pack tightly within the wall of the primordium as a result of elongation coupled with loss of apical size and continued growth through cell division. Toward the end of this period, a fourth zone of cells begins to be added to the primordium and bending at the third basal groove elevates zone 4. Force exerted by surrounding tissues coupled with bending at basal groove three positions zone 4 to face the thyroid lumen. A fifth zone of cells begins to convert from pharyngeal to thyroid. This activity gives the stage 16 thyroid the appearance (from the dorsal surface of the pharyngeal floor) of being composed of a tube surrounded by two concentric rings of cells (Fig. 1g).
In conclusion, this study provides direct evidence that growth of the thyroid occurs, in large part, by way of annexation of pharyngeal cells. We have also shown that cell clusters act in shaping the early thyroid. It would be of interest to know whether cell cluster movements are involved in the morphogenesis of other internal organs. Liver and pancreas would seem to be good candidates for such a morphogenetic mechanism. Both are offshoots of gut endothelium and both are lobed in their final form, characteristics that they share with thyroid. The finding of cell clusters in these organs would implicate this as a general mechanism for the formation of a class of internal organs.
S.R.H. received support from the National Science Foundation and a Grant-in-Aid from Temple University. G.M.K. received a Future Faculty Fellowship from Temple University.