• cell division;
  • growth;
  • morphogenesis;
  • segmentation clock;
  • wavefront


  1. Top of page
  2. Abstract
  7. Acknowledgements

Somite stages were employed as units of intrinsic developmental time to measure cell doubling rate and other cell cycle parameters of chick forelimb level somites. Somite cell nuclei doubled over an interval corresponding to approximately 7+ somite stages (7+ ss; ∼11 hr) and approximately 24 new primary myotome cells are born per somite stage (∼16/hr). FACS analysis of DNA content in dissociated paraxial mesoderm cells indicated that slightly more than half are in G1/G0 phase of the cell cycle and that the average combined length of the S phase and G2 phase intervals is approximately 3 ss (∼4.5 hr). A wavefront of increased mitotic nuclei per segment coincident with somite budding potentially reflects a surge in the number of cells entering S phase 3 ss earlier as each PSM segment becomes unresponsive to FGF signaling as it passes through the determination front. Developmental Dynamics 237:377–392, 2008. © 2008 Wiley-Liss, Inc.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Somites are proto-segments that transiently appear within the paraxial mesoderm strips along both sides of the axial neural tube of vertebrate embryos (Fig. 1A). Somites bud off the rostral end of the presomitic mesoderm (PSM) as a result of segmental collections of mesenchyme cells segregating into morphologically distinct somite segments. Somite budding in chick embryos occurs through epithelialization of the PSM cells comprising one segment, creating an epithelial sphere with an internal cavity, the somitocoel, within which a sub-fraction of mesenchyme (non-epithelial) cells remains (see Fig. 1B). Each somite eventually loses its epithelial organization through cell migration to other body regions, such as the limb (Christ et al.,1974), and through the in situ growth and morphogenesis of cartilage, bone, muscle, endothelia, and dermis (reviewed in Christ and Ordahl,1995), and tendons (Brent et al.,2003), for each body segment (Beresford,1983; Bagnall et al.,1989).

thumbnail image

Figure 1. Somitogenesis and nuclear counting strategy. A: Schematic illustration of a 20-somite chick embryo illustrating the polarity and segmentation of the paraxial mesoderm. The arrowheads indicate the axial positions (somite number, sn) and somite stage (ss) for: somite number 16 at somite stage V (sn16, ssV); somite number 21 at somite stage 0 (sn21, ss0); and sn25 at presomite stage-IV (sn25, ss-IV). Somite stage 0 at the rostral tip of the presomitic mesoderm (PSM) is the transition point for somite budding. The brackets indicate the forelimb level somites and the PSM of the embryo illustrated. B: Schematic drawing of confocal imaging through a forelimb level somite at ssV (illustrated in transverse section) indicating the position of the microscope objective, the confocal z-axis, and the approximate confocal plane (p) for the image in C. C: Confocal microscope image through the coronal plane of a fore limb-level somite at ssV similar to the plane “p” diagrammed in B. In these images, the lateral edge of the neural tube epithelium and the rostral edge of the adjacent ssIV somite can be seen on the upper and left sides, respectively. Note the morphological boundaries identifying the nuclei belonging to the central ssV somite. The right and left panels show the image before and after nuclear marking (magenta dots) for digitized counting. The rostro-caudal (r-c) axis is indicated in the left-hand image by the double-headed white arrow along the somite lateral edge. D: Imaging of presomite segments in the PSM. A composite confocal scan of the PSM of a 16-somite embryo with intervals that are equivalent to the rostro-caudal dimensions of the stage I somite indicated (Pourquie and Tam,2001). The segment interval was 70 μM in the example shown. The nuclei within designated presomite segments were enumerated in the same manner as nuclei in somites. E: Data compilation protocols. Each of the vertical columns represents an idealized unilateral paraxial mesoderm strip from individual embryos whose ages increase from left-to-right by a single somite stage (∼90-min interval) (Borman et al.,1994; Borman and Yorde,1994a). Arrowhead “a” shows the direction of data compilation for nuclear number at varied stages of somitogenesis for a single axial level (ss/sn). Arrowhead “b” compiles numbers of nuclei in somites at a single stage of somitogenesis irrespective of somite axial position (ss alone). Arrowhead “c” compiles nuclei in segments from the paraxial mesoderm strip of a single embryo (as in panel D).

Download figure to PowerPoint

The early development of each PSM presomite segment is governed by a cell-intrinsic “segmentation clock” that oscillates with a 90-min periodicity, the same interval necessary to form one new somite (Palmeirim et al.,1997; recently reviewed in Andrade et al.,2005; Dale et al.,2006; and references therein). In the caudal PSM, the segmentation clock is dependent upon intrinsic FGF signaling, but FGF-dependence and clock function cease as the segment reaches the “determination front,” the point at which rostro-caudal polarity and boundaries of the PSM segment become irreversibly fixed (Pourquie,2003). Fixation of the dorso-ventral and medio-lateral axes in avian embryos occurs during epithelial somite stages I–III (i.e., after somite budding) (Ordahl and Le Douarin,1992; Dockter and Ordahl,2000), and is dependent upon cell-cell signaling (reviewed in Geetha-Loganathan et al.,2006; and references therein).

Because each segment reiterates the morphogenetic changes of the previous segment, the position of a particular segment relative to the point of somite budding within a strip of paraxial mesoderm (Fig. 1A), can be used to index its developmental age expressed in somite stages (ss) (Ordahl,1993; Christ and Ordahl,1995; Pourquie and Tam,2001). One of the experimental goals of this study was to test the proposition that somite staging can be used as a systematic measure of intrinsic developmental time for quantitative analysis of cell doubling rate and cell cycle phases in the growth and morphogenesis of somites and somite-derived tissues, such as the myotome.

Cell doubling time in the paraxial mesoderm of chick embryos was previously estimated to be approximately 9.5–10 hr based upon 3H-thymidine pulse-chase labeling, and several other methods (Primmett et al.,1988,1989) and 11.6 hr, based on mitotic and proliferation indices (Smith and Schoenwolf,1987). Those rates are generally consistent with the average cell doubling time of approximately 10 hr estimated for other mesoderm cell types (Hagopian and Ingram,1971; Weintraub et al.,1971; Minkoff,1984; Minkoff and Martin,1984) and for embryonic mesoderm in general (Sanders et al.,1993).

The very small size of somites presents both significant challenges and advantages for acquisition of quantitative information about cell replication and other growth parameters. An epithelial somite from the forelimb level of a chick embryo is ∼100 μm or less in dorso-ventral diameter (Fig. 1B) and contains an estimated 2,000–3,000 cells (Ordahl,1993). In the new method to measure the growth dynamics presented here, somite cell number is determined directly by enumerating fluorescently-stained nuclei in confocal image stacks of somite and PSM segments in situ within multiple chick embryos (Fig. 1B). The center-points of nuclei were electronically marked manually in each image plane of the entire confocal z-series stack (Fig. 1C, D) and a computer-assisted protocol then used to link contiguous points within the stack that represented individual nuclei and the total number of nuclei within individual segments recorded (see Experimental Procedures section for details). Nuclear numbers were compiled and analyzed according to somite stage (ss) and axial position, designated as somite number (sn) (Fig. 1E; arrows b and a, respectively) to analyze doubling rate and other growth parameters of somite cells. The results of these studies demonstrate that somite staging can be used to measure intrinsic developmental time for the analysis of cell doubling rate and other cell cycle parameters related to somite growth and morphogenesis.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Somite Cell Doubling Rate Estimated by Enumeration of Somite Nuclei

The total number of nuclei was determined for 73 paraxial mesoderm segments (presomite segments in the PSM and formed somites) (Table 1) from lower cervical to upper thoracic segments corresponding to the forelimb level (Fig. 1A). The average forelimb level somite contains between 2,000 and 3,000 cells between ssI and ssVI (Fig. 2A). The exponential solution for the increase in average number of nuclei per somite stage (Fig. 2A) yields an average cell doubling time of approximately 9.24 ss (T½ = ∼14 hr). However, because the averaged data were compiled according to ss alone, they do not account for potential effects that axial level may have on cell number per somite. Comparison of ssI somites at different axial levels (sn) shows that, as a group, rostral forelimb level somites (sn16–18) contained half as many nuclei as caudal forelimb level somites (sn20/21) (Fig. 2B). Somite cell growth rate estimated by this “average/ss” method does not account for the influence of axial level and may explain why the doubling rate estimated by this method is much slower than any previous estimate made by other methods (see Introduction section).

Table 1. Total Nuclei Enumerated in Somites and Presomite Segments of the psma
Axial levelSomite stageNucleiAxial levelSomite stageNucleiAxial levelSomite stageNuclei
  • a

    Nuclei column shows the total number of nuclei per segment; Segment axial level is given as somite number (sn) in arabic numerals; Segment somite stage (ss), is shown in roman numerals, with positive values for epithelial (ss I–III) and post-epithelial (ss >III) somites, negative values for PSM pre-somite segments, and ss0 representing budding somites.

thumbnail image

Figure 2. Somite cell doubling rate. A: Average numbers of nuclei per somite stage (ss alone; Fig. 1E, arrow a). Nuclei counts are the average for each somite stage for a total of 73 somite or presomite segments (Table 1). The curve fit reflects exponential growth (y = 1,920.9*e (0.065484x), R = 0.89482) and yields a predicted doubling rate of 9.24 ss (13.86 hr). B: Rostral to caudal increase in nuclear counts for all forelimb-level somites at ssI. Analysis of grouped nuclear counts demonstrated a statistically significant difference (P < 0.01) between somite nuclear numbers at rostral (sn16–18) versus caudal (sn19–21) positions at ssI. The line is the linear fit for all ssI nuclear numbers (y = 2915.8+275.23x, R = 0.78114). C: Nuclei counts for sn17 between ss-IV and ssVII. The predicted doubling rate from the exponential curve fit (y = 1604.3*e (0.10033x), R = 0.78432) is 7.15 ss (10.7 hr). The calculated doubling rate for sn17 after somite budding (ssI to ssVII) was 6.9ss (not shown). D: Nuclei counts for sn 19 from ss-II to ssV. The predicted doubling rate from the fitted exponential growth curve (y-1991.1*e (0.1667x), R = 0.79565) is 7.42 ss (11.13 h).

Download figure to PowerPoint

The similarity in the number of nuclei in ssI somites at adjacent axial levels (Fig. 2B) also indicated that a degree of uniformity exists among individual embryos in the growth rate for specific axial regions, at least with respect to the number of cells incorporated into budding somites. If it is assumed that uniformity in growth kinetics persists during subsequent somite stages, then cell doubling rate should be reflected by compiling the nuclear numbers for a single axial level according to somite stage (arrow a, Fig. 1E). To test that hypothesis, the nuclear numbers for two discrete axial levels (sn17 and sn19) were each compiled separately according to somite stage (Fig. 2C and D). The exponential solution for these data yielded somite cell doubling times of 7.2 and 7.4 somite stages (∼11 hr), respectively, which are in good agreement with previous estimates (see Introduction section). Finally, the number of nuclei incorporated into newly-formed (ssI) forelimb level somites is predicted to approximately double between sn16 and sn22, a span of 7 segments (Fig. 2B line). We conclude, therefore, the nuclear count results for forelimb level somites compiled according to somite staging, and when corrected for axial position, corresponds approximately to the period necessary to form seven somites (∼7 ss) consistent with earlier conclusions (Primmett et al.,1988,1989; Stern et al.,1988).

Myotome Cell Birthrate

Compilation of nuclear numbers independent of axial position (ss alone, Fig. 1E, arrow b) has the potential to measure stage-dependent changes that are common to all somites. Formation of the segmental myotome is a highly-ordered, stage-dependent process (Yang and Ordahl,2006; and references therein), which occurs in all somites, including occipital somites (Huang et al.,1997). Desmin-positive cells are first detectable in the primary epaxial myotome of forelimb somites at ssVII (Kaehn et al.,1988), and the rate of increase in new desmin-positive cells is constant during forelimb level somite development (Borman et al.,1994; Borman and Yorde,1994a). Nascent primary epaxial myotome cells translocate from the dorsomedial lip of the overlying dermomyotome into a layer of rostro-caudally elongated, mononuclear, myotome cells that lie immediately subjacent to the pseudostratified columnar epithelium of the dermomyotome (Fig. 3A). The well-ordered translocation of new myotome cells asymmetrically, at the dorsomedial margin of the myotome layer, results in the expansion of the epaxial myotome in each segment during the next 3+ days of embryonic development (Denetclaw et al.,1997,2001; Denetclaw and Ordahl,2000; Venters and Ordahl,2002; Yang and Ordahl,2006). At the forelimb level, hypaxial muscle precursor cells migrate from the ventrolateral lip of the dermomyotome (VLL) to populate the adjacent limb bud (reviewed in Christ and Ordahl,1995). In somites immediately caudal to the forelimb level, initiation of hypaxial myotome development occurs after the period analyzed here (Denetclaw and Ordahl,2000). Therefore, only nuclei of the epaxial primary myotome were enumerated in this study (Table 2).

thumbnail image

Figure 3. Myotome cell birth rate. A: Schematic drawing of confocal imaging of a ssX+ forelimb level somite. The myotome, dermomyotome, and sclerotome are shown in transverse section view with the relative position of the microscope objective (obj) illustrated. The red dashed box marked “S” indicates the plane of the confocal stack in B and C, which comprised images of nuclei within the entire myotome layer but not the nuclei in the more ventral sclerotome and more superficial dermomyotome layers. B: Confocal stack of z-series scans through the myotome layer of a stage X+ somite. Myotome nuclei were distinguished in each image by three criteria: (1) location, in the myotome layer, immediately subjacent to the nuclei of the dermomyotome epithelium; (2) morphology, elongated along the rostro-caudal axis; and (3) fluorescence pattern, which was more diffuse and reduced in intensity compared to surrounding somite epithelial nuclei. Red-filled arrowheads indicate mitotic nuclei at extreme apical surface of overlying dermomyotome epithelium. C: Same confocal stack as in B except with the red dashed outline around myotome nuclei. D: Myotome nuclear number (N) compiled per ss alone (arrow b, Fig. 1E) between somite stages VII–XIV. The data are fitted with a straight line (y = −135.8+23.755x, R = 0.914), indicating a linear increase of myocyte nuclei of approximately 24 nuclei/ss. Individual values are listed in Table 2.

Download figure to PowerPoint

Table 2. Enumeration of Myotome Cell Nucleia
  • a

    Myonuclei, number of nuclei in primary epaxial myotome layer; other abbreviations as in Table 1.


The per segment numbers of myotome cell nuclei plotted against somite stage (Fig. 3D) indicate a linear increase in myotome cell number over the 7ss period analyzed (∼11 hr). Approximately 24 new primary myotome cells are born per somite stage (16/hr). The stage-dependent linearity of myotome cell birthrate among somites not indexed for axial level (i.e., collected according to arrow “b” in Fig. 1E) indicates that the primary myotome growth rate is similar among limb-level somites, a conclusion consistent with that made previously by desmin staining (Borman et al.,1994; Borman and Yorde,1994a) and by fluorescent dye lineage tracing (Denetclaw and Ordahl,2000).

Measuring the Relative Lengths of Cell Cycle Phases in Paraxial Mesoderm Cells

The lengths of the individual phases of the somite cell cycle were previously estimated from computational analysis of silver grain peaks from 3H-labeled mitotic figures with an inter-segmental separation of 6–7 segments (Primmett et al.,1989). Fluorescence activated cell sorting (FACS) provides an alternative means of computing the relative length of cell cycle phases (Watson et al.,1987) through analysis of DNA content on a cell-by-cell basis to yield the proportion of cells in G1/G0 (2c DNA), G2/M (4c DNA), and S (>2-<4c DNA) phases. FACS analysis required that paraxial mesoderm strips from multiple embryos be dissected and pooled to provide the hundreds of thousands of dissociated cells necessary to obtain a DNA content histogram of paraxial mesoderm cells (Fig. 4A). These preparations included both somites and PSM and rendering the cell cycle data weighted against PSM cells due to the much lower per-segment cell numbers in the caudal paraxial mesoderm. Computational analysis of the DNA histogram (Watson et al.,1987) indicated that slightly more than half the paraxial mesoderm cells (54%) are in the G1 or G0 phase, while approximately one third (32%) are in S-phase and 14% in G2 or M phase (Fig. 2B). Taking somite total cell cycle time as ∼7 somite stages, the derived cell cycle phase lengths expressed in somite stages are: G1 phase ∼4 ss; S phase ∼2 ss; and G2 phase ∼1 ss, assuming that M phase is rapid, i.e., ∼0.5 hr, or less. The corresponding periods expressed in “real” time are: G1 phase ∼6 hr; S phase ∼3 hr; G2/M ∼2 hr.

thumbnail image

Figure 4. Paraxial mesoderm cell cycle phases. A: FACS histogram of DNA content for paraxial mesoderm cells. PSM and somite cells were dissociated from paraxial mesoderm strips isolated from embryos containing 15–20 formed somites, stained with propidium iodide and analyzed using a Becton-Dickinson FACScan flow cytometer (see Experimental Procedures section for details). The black line shows the DNA content histogram with prominent G1/G0 and G2/M peaks. Colored lines show the fraction of cells computed to be in G1/G0 (red), S (green), and G2/M phases of the cell cycle (Watson et al.,1987). B: Paraxial mesoderm cell cycle phase fractions. The pie-chart shows the estimated average length of each cell cycle phase as a fraction of the average cell cycle period: G1/G0, 0.54; S, 0.32; G2/M 0.14.

Download figure to PowerPoint

In Ovo Measurement of the Length of the G2 Phase of the Somite Cell Cycle

The individual cell cycle phase lengths as estimated above differed substantially from those estimated previously for chick somite cells (Primmett et al.,1989), particularly with respect to the shorter length of the G2 phase reported here. The length of the G2/M phases of the cell cycle in ssI through ssIII somites was, therefore, independently estimated using DNA pulse-labeling to measure the minimum and maximum intervals between S and M phases. After administration of the DNA precursor BrdU at time 0, embryos were incubated in ovo for predetermined intervals and then processed for double-label immunohistochemistry in which mitotic nuclei were identified using anti-phospho-histone H3 antibody while nuclei that had been in S-phase during the time of labeling were identified using anti-BrdU antibodies (Fig. 5). The incidence and coincidence of H3-positive and BrdU-positive nuclei were then recorded for multiple somites at pre-determined intervals (Table 3).

thumbnail image

Figure 5. Measurement of the length of the G2 phase of the somite cell cycle. AI: Confocal images of sections of ssI–III somites collected from embryos harvested 60 (A–C), 90 (D–F), or 180 (G–I) min after administration of BrdU. Sections show immuno-labeling of BrdU (B, E, H), phosphorylated H3 (C, F, I), and both antibodies (A, D, G). At 60 min post-BrdU administration, no pH3-positive nuclei have incorporated BrdU (A–C). After 90 min, some pH3 nuclei have incorporated BrdU (D–F) and after 180 min all pH3-positive nuclei have also incorporated BrdU. Arrowheads mark nuclei that are pH3-positive and BrdU-negative. Arrows indicate nuclei labeled with both BrdU and pH3. The asterisked arrowheads in A indicate nuclei that appeared to be dual labeled but were found, in each instance, to be single labeled in two distinct nuclei. Individual values are listed in Table 3.

Download figure to PowerPoint

Table 3. Somite Cell G-2 Phase Estimate by Anti-BrdU and -Phospho-Histone H3 Antibodiesa
BrdU pulse length (min)Number of somitespH3 alonepH3 + BRDU
  • a

    BrdU pulse length indicates the time between BrdU administration and embryo harvest. Each somite contained many pH3-positive nuclei by confocal microscopy. Their number was not determined. pH3 alone indicates the presence of nuclei that are anti-pH3-positive and anti-BrdU-negative. pH3 + BrdU indicates nuclei present that are both anti-pH3- and anti-BrdU-positive.


H3-only nuclei represent cells that were already in G2 phase at the start of the experiment. H3-only nuclei are present at 60, 90, and 120 min (Table 3). Double-positive mitotic nuclei (BrdU + H3) represent dividing cells that were in S phase at time 0. Double-positive mitotic nuclei were not present in somites that had been incubated with BrdU for 60 min (Fig. 5A–C), indicating that none of the S phase cells at time 0 had yet progressed through G2 phase into mitosis. Double-positive nuclei were detected after 90 and 120 min of incubation in BrdU (Fig. 5D–F). By 180 min of incubation, however, all H3-positive nuclei were BrdU-positive (Fig. 5G–I) and H3-only nuclei were no longer detected.

These results indicate that most epithelial somite cells complete G2 in 90–120 min. These data do not rule out the possibility that some epithelial somite cells may have a G2 interval somewhat shorter or longer than (>60 min and <180 min, respectively) but not longer than 3 hr. The G2 intervals estimated here by FACS and by S phase labeling are substantially shorter than the 4.5–5 hr previously estimated (Primmett et al.,1989). The data presented here provide only a sampling of the somite cells that entered S phase during the 3-hr course of the experiment, and therefore cannot exclude the possibility that some epithelial somite cells may have longer G2 phase intervals because cells early in G1 at the start of the experiment, or non-cycling cells in G0 phase, were not scored. The contribution of nascent myotome nuclei, which may already be post-mitotic and committed to differentiate in ssI to ssIII somites (Williams and Ordahl,1997), is negligible compared to the total number of cells (Fig. 2). These considerations notwithstanding, the results of BrdU-pulse labeling in ovo and of FACS analysis are in agreement in estimating that the “average” duration of G2 phase in cycling cells of epithelial somite cells spans approximately 1 ss (∼1.5 hr).

Periodicity of Segmentation Defects Induced by BrdU

No developmental defects were noted for the ssI to ssIII somites analyzed for G2 phase length as outlined above, but longer re-incubation periods following exposure to BrdU have been shown to induce malformations in chick paraxial mesoderm development (Lee and Kalmus,1978). Stern and coworkers, in a detailed study involving many drugs, showed that chick embryos exposed to extremely low BrdU concentrations (∼2 μM) for 2 hr, and then re-incubated 24 hr in the absence of BrdU, showed defects in paraxial mesoderm development that were spaced at ∼7-somite intervals (Primmett et al.,1989). To re-examine somite defect periodicity induced by BrdU, chick embryo blastoderms were treated essentially as described previously (Primmett et al.,1989) but using somite staging to measure intrinsic developmental time and to identify defect locations (Table 4). BrdU-induced defects in somite development observed here were morphologically consistent with those previously reported (Primmett et al.,1989) and consisted of: (1) increased or decreased somite size, (2) failure to differentiate (persistent epithelial stage somites), and (3) incomplete segmentation between adjacent somite pairs. Unexpectedly, however, BrdU-induced defects were detected at a higher rate than indicated by the previous study, in which 8 single-site defects were recorded for 15 survivors (Primmett et al.,1989) as compared to the present study in which 10 of 11 survivors had one or multiple defects (Table 4).

Table 4. BrdU-Induced Periodic Defects in Somite Developmenta
embryostart #shr BrdU#s stopDuration ΔssDefect locationsDefect spacing
dL som#dR som#ΔdLΔdR
  • a

    start #s, total somite number at start of BrdU administration (0.5 μg/ml); hr BrdU elapsed time (indicated in hours, hr) before BrdU-washout; #s stop, total number of somites at time of harvest; Δss, elapsed time between start and stop expressed in somite stages; dL, defect left segment numbers (sn) affected; dR, defect right segment numbers affected; ΔdL and ΔdR, elapsed times between drug administration and observed defects expressed in somite stages. nd, no defect detected; na, not applicable.

182.5261713–16, 21nd∼7, 14na
4102.0312021, 2725–2611, 17∼15
6122.752815nd17, 22–23na5, ∼11
7122.0301719, 26–2726–277, ∼14∼14
8172.533152817–28110, 11

A significant conclusion reached in the earlier study (Primmett et al.,1989), that BrdU affects paraxial mesoderm patterning during the PSM stage of paraxial mesoderm development, is fully supported by the results of these experiments because in no case was any defect observed that could be attributed to an effect on a somite that was already epithelial at the time of treatment (Table 4; compare ”start#som” to “defect locations”). In some cases, the observed defects were non-repeating, i.e., limited to single segments or a cluster of 2–3 adjacent segments, consistent with the type of defects previously reported for BrdU (Primmett et al.,1989). In the present study, however, BrdU caused repetitive periodic defects within an individual paraxial mesoderm strip in several embryos (nos. 1, 4, 6–8 in Table 4). The interval between BrdU treatment and the appearance of a segmentation defect (“defect spacing” in Table 4) was frequently ∼7 somites, or a multiple thereof (embryo nos. 1–3, 5, 7, 10), but periodic defects with different spacing intervals were also observed (embryo nos. 4, 6, 8, and 9). Some defects were bilateral and symmetric but, in other cases, defects were asymmetric between the left and right sides of an embryo, a finding that is consistent with earlier reports on the effects of BrdU on somitogenesis in mouse embryos (Skalko et al.,1975). These results confirm that BrdU induces segmentation defects with periodicity of ∼7 ss, as previously reported (Primmett et al.,1989), but also demonstrate defects with other spacing intervals. These results also underscore further the extremely high teratogenic potential of BrdU to produce profound segmentation defects in embryos exposed only transiently (<3 hr) to extremely low concentrations (<2 μM) of this drug (Primmett et al.,1989) (see Experimental Procedures section).

Enumeration of Mitotic Nuclei in Paraxial Mesoderm Segments

Mitotic nuclei have a distinctive appearance in confocal stacks (Fig. 3B) allowing their numbers in somites and PSM segments to be quantified (Table 5). Per-segment numbers of mitotic nuclei compiled according to somite stage and independent of axial position (ss alone; arrow b in Fig. 1E) showed that mitotic rate per segment changes during the course of somitogenesis (Fig. 6). The average numbers of mitotic nuclei in PSM presomite segments ss-VI to ss-II were approximately half those for epithelial somites between ssII and ssVI with a sharp increase as somites mature from ss-I to ss0 and ssI (Fig. 6A). The individual mitotic numbers for each individual segment (Fig. 6B) demonstrate substantial variation at each stage, variation that is not evident in the averaged data (Fig. 6A). Statistical analysis of grouped values for numbers of mitotic nuclei showed that PSM segments are significantly lower than epithelial somites and that ss0 is intermediate between the two regions of the paraxial mesoderm (Fig. 6B, and legend).

Table 5. Mitotic Nuclei Enumerated in Somites and Presomite Segments of the psm
ssMitotic nucleissMitotic nucleissMitotic nuclei
thumbnail image

Figure 6. Incidence of mitotic nuclei during somite development. A: Average number of mitotic figures in somites stages -VI through VIII compiled independently of axial level (ss alone; see Fig. 1E, arrow b). For this analysis, the data have been split into 3 groups reflecting developmental stage; PSM (-VI to -II), somite formation (-I to I), and formed somites (II to VIII). The linear solutions for these axial sets are: PSM, y = 47.246+2.366x, R = 0.79771; somite formation, y = 55.21+14.815x, R = 0.8009; formed somites, y = 70.197+0.77429x, R = 0.43448. B: The complete set of mitotic nuclei counts was compiled according to ss (arrow a, Fig. 1E). Open circles, PSM segments -IV to -II; closed circles, segments ss-I, ss0, and ssI; open squares, ss II to VIII. Statistical analysis indicated a significant difference (>99% confidence) when the grouped per segment mitotic counts are compared between: (a) ss-VI to ss-II vs. ssII to ssVIII; (b) ss-VI to ss0 vs. ssI to ssVIII; and ss-VI to ss -I vs. ss0. Mitotic nuclei at ss0 also differ from ssI to ssVIII as a group (>95% confidence level), consistent with it being the transition point. The solid line shows the linear fit only for the ss-I, ss0, and ssI data set (n = 36; y = −51.209 + 14.913x; R = 0.57). Individual values are listed in Table 5. C: Closed squares show mitotic counts for sn 17 (collected according to arrow a in Fig. 1E); open circles show mitotic counts per segment for one embryo (longitudinal image composite as indicated in Fig. 1D and compiled according to arrow c in Fig. 1E). D: Percent mitotic nuclei per somite stage. Closed circles show average percentage of mitotic nuclei per somite stage by combining the data in Tables 1 and 5. The solid line represents the best linear fit through the points at ss-I, ss0, and ssI (y = 2.6 + 0.25x R = 0.945).

Download figure to PowerPoint

The linear solution to the average number of mitotic nuclei in segments between ss-I and ssI indicates a midpoint transition at ss0 (Fig. 6A) and the linear solution to the un-averaged data for mitotic nuclei at ss-I, ss0, and ssI (Fig. 6B, closed circles; Table 5) also identifies the midpoint at ss0. The mitotic numbers at ss0 as a group are statistically different than those of the younger PSM segments and older somite segments as groups (see legend, Fig. 6B), which is also consistent with ss0 being the transition point of the shift in mitotic rate per segment. Additional observations are consistent with the conclusion that a shift in mitotic rate per segment occurs approximately coincident with somite budding. First, the average per-segment number of mitotic nuclei at ss-I was lower than any other stage of paraxial mesoderm analyzed and the average for ssI lower than any older somite (Fig. 6A). Second, the range in number of mitotic figures per segment was greater at ss-I than at any other stage, with ss0 having the next highest range of values (Fig. 6B). Measurement variability is expected (1) due to operator-dependent variability in assigning somite stage, which has an inherent error potential of ± 0.5 ss (see Experimental Procedures section), and (2) because variability, as it is related to measurement of developmental time, is enhanced during developmental transitions (Ordahl,1986; Wong and Ordahl,1996). Despite these sources of variation, the linear fits for the averaged and un-averaged data for the somite stages between ss-I and ssI (Fig. 6A and B, equations in legend) both demonstrate an increase in mitotic rate per segment that is centered approximately at ss0.

The mitotic numbers per segment plotted in Figure 6A and B are compiled independent of axial level (arrow b, Fig. 1E) indicating that the mitotic rate shift is experienced by every forelimb level segment as it buds off the rostral tip of the PSM. A similar shift is evident when the data for sn17 are compiled according to somite stage (Fig. 6C, closed squares; according to arrow “a” in Fig. 1E). The shift to increased number of mitoses per segment at ss0 is also evident when data from a single embryo is compiled longitudinally (Fig. 6C, open circles; according to arrow “c” in Fig. 1E).

The increase in the number of mitoses between ss-I and ssI reflects an increase in the fraction of dividing cells because the percentage of mitotic nuclei per segment (Fig. 6D, closed circles) is lowest at somite stages -I and increases at ss0 and ssI consistent with a per-segment surge in paraxial mesoderm mitoses that is synchronous with and centered at ss0.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Measurement of Cell Growth Dynamics in Paraxial Mesoderm Segments by Somite Staging

Embryonic development rate is governed by intrinsic time, the mechanisms of which are incompletely understood (Johnson and Day,2000; West et al.,2001). The potential for somite staging to serve as a reliable measure of intrinsic developmental time for paraxial mesoderm development is supported by several lines of evidence. First, the rate of somite formation in avian embryos is temperature-dependent (Bucciante,1935; as cited in Derrick,1937), which indicates that the rate of somite formation is directly dependent upon cellular metabolic rate. Thus, variables, such as temperature fluctuations during the development of one embryo, and/or embryo-to-embryo differences in incubation temperature, which both alter the rate of developmental progression as measured in “real time” (Gillooly et al.,2002), do not affect the intrinsic rate of development as reflected by somitogenesis. Second, detailed analysis of the segmentation clock (Dale et al.,2003; and references therein) demonstrates the cell-intrinsic nature of this oscillator, and its links to intrinsic developmental time and the rate of somite budding (Pourquie,2003). Third, the cell cycle has been implicated in regulating the process and rate of somitogenesis in chick embryos (Primmett et al.,1988,1989; Stern et al.,1988) as well as amphibian embryos (Cooke,1979; Leise and Mueller,2004). Somite staging, because it reflects intrinsic developmental time, was therefore an essential component of the experimental strategy employed here because it made it possible to compile nuclear numbers from multiple embryos and from multiple somite stages to obtain the somite cell doubling time and other somite cell cycle measurements.

The practical use of somite staging to measure the intrinsic developmental clock is limited primarily by the accuracy of assigning stage values to individual somites. Somites assigned the same stage value could vary in intrinsic developmental time by as much as 90 min using current methods of somite staging. Nevertheless, nuclear enumeration data obtained from multiple embryos compiled according to somite stage provided several quantitative measures of somite cell growth that are in good agreement with each other and, in most instances, with similar measures made by other methods. Somite stage subdivisions (see Experimental Procedures section), coupled with larger per-segment nuclear count data sets, could substantially improve overall accuracy of these measurements. The values for the number of nuclei per segment, by contrast, had a lower error rate, estimated to be approximately ± 5% as determined for nuclei enumerated from the same somites independently by multiple operators (see Experimental Procedures section).

A second consideration in the use of somite staging as a measure of intrinsic time was the coordinates with which embryonic somite segments were to be indexed. Segment axial position (sn) is a fixed value reflecting the number of segments rostral to a particular segment in question. Segment developmental age (ss) is a variable value that reflects the number of segments intervening between ss0 and the segment in question. Somite stages are assigned positive or negative values when referring to somites or to PSM presumptive segments, respectively, while the point of somite budding is designated ss0. Indexed nuclear enumeration data can be compiled according to somite stage–dependent growth parameters that apply to: (1) all somites, independent of axial level, or (2) somites from a specific axial level. Both indexing methods yield information regarding the growth parameters of somite cells.

The first method was used to estimate the rate of primary epaxial myotome development by compiling the number of myotome nuclei according to ss alone and independent of axial level (Fig. 1E, arrow b). The results showed that new myotome nuclei appear within forelimb level segments at the average rate of 24 new myotome cells per somite stage or approximately 16/hr (Fig. 3D). Myotome nuclei are post-mitotic and new primary myotome nuclei are born exclusively from the dorsomedial lip (DML) (Ordahl et al.,2001) a region of the dermomyotome epithelium comprising approximately 300–500 cells (not shown). Assuming that cell cycle time within the DML is similar to that of somite cells in general (∼7 ss or ∼11 hr), approximately 150 precursor cells would be required if every new myotome cell results directly from an asymmetric cell division in the DML (Venters and Ordahl,2005) (i.e., one post-mitotic daughter and one regenerated mitotic myoblast mother cell).

Measurement of cell doubling rate within somites required a different approach because nuclear numbers expand logarithmically and because the number of nuclei per somite stage varied according to axial position. Nuclear enumeration data compiled for individual axial levels according to somite stage (arrow a in Fig. 1E) yielded doubling time values of approximately 7+ ss (∼11hr), which are in good agreement with earlier estimates for paraxial mesoderm and other mesoderm cells in comparably-aged chick embryos as outlined in the Introduction section. The present results demonstrate the feasibility of using somite staging to reflect intrinsic developmental time for the quantitative analysis of hyperplasia within the paraxial mesoderm.

Estimating the Duration of Somite Cell Cycle Phases

There is limited information concerning the length of cell cycle phases in the paraxial mesoderm that can be used for comparison to those reported here. The 2.2 hr (∼1.5 ss) duration of S phase previously estimated (Primmett et al.,1989) is reasonably close to the duration estimated here (2 ss/3 hr) (Table 6). By contrast, previous estimates for the lengths of G1 and G2 phases were, respectively, 2-fold shorter and 3-fold longer than those estimated here (Table 6). These differences are likely to result from the different methods used to estimate cell cycle phase lengths. The earlier study employed cell cycle inhibitors, in combination with mitotic indices, to derive the lengths of individual cell cycle phases from inter-somite distances measured between peaks of labeled mitotic figures after a 3H-thymidine-pulse (Primmett et al.,1989). The computer-assisted protocol used here (Watson et al.,1987) measures the average length of G2 + M phases from the fraction of paraxial cells with 4c DNA content (Fig. 4). The doubling time of somite cells (∼7 ss, computed as in Fig. 2) was then used to compute the fractional contribution of G2/M cells expressed in somite stages. An independent estimate of G2, using a BrdU-labeling strategy and “real time” measurements, indicated that G2 phase duration in most somite cells is between 1.5–2 hr (∼1+ ss; Table 3), which is consistent with the estimate derived by FACS methods. The duration of S phase plus G2 phase is approximately 4.5 hr, or 3 ss. That duration is close to the doubling time estimated earlier for mesoderm cells in day-1 chick embryos (Sanders et al.,1993) and that estimated from nuclear counts in squash preparations of occipital somites isolated from day-1 chick embryos (Hultner,2001) (see also Experimental Procedures section). The 4.5-hr period may, therefore, represent the doubling rate of embryonic somite cells with minimal G1 phase duration. In that case, longer doubling times may be largely achieved through extending the duration of G1 phase of the cell cycle.

Table 6. Comparison of Somite Cell Cycle Phase Length Estimatesa
  • a

    The data for this study are shown in Figure 4 and Table 3. The estimates from the earlier study (Primmett et al.,1989) are mean values from 17 embryos derived by computerized analysis of pulse-chase labeled mitotic peak intervals. Values in italics are derived from measured values, which are in plain text.

This study    
 FACS (PSM + somites)4 ss2 ss1 ss7.3 ss
 4.5 hr3 hr1.5 hr11.3 hr
 Pulse chase estimate G-2  1+ ss 
   1.5–2 hr 
Primmett et al. (1989)    
 PSM2.3 hr2.3 hr5.1 hr9.5 hr
 1.5 ss1.5 ss3.5 ss6.3 ss
 Somites3.3 hr2.2 hr4.5 hr9.5 hr
 2+ ss1.5 ss3.5 ss6.3 ss

Mitotic Activity in Paraxial Mesoderm Development

Increased mitotic activity has been reported for the region immediately rostral to Hensen's node region of early streak chick embryos (Derrick,1937; Pasteels,1937; Emanuelsson,1961; Ozato,1969; Stern,1979) and for the rostral tip of the PSM (Berrill,1955; Stern and Bellairs,1984; Mills and Bellairs,1989). In those studies, the percentage of nuclei that were mitotic was established from histological sections of individual embryos (i.e., according to arrow c in Fig. 1E.). In the present study, mitotic nuclei were enumerated on a per-segment basis and those numbers compiled according to somite stage and somite number (according to arrows a and b, respectively, in Fig. 1E). The number of mitotic nuclei per segment, when compiled according to somite stage, revealed a coherent shift in the per-segment number of mitotic figures that was centered at ss0, the point of somite budding (Fig. 6). The shift was evident when mitotic numbers are compiled for somites according to somite stage alone (as in arrow b, in Fig. 1E), or for a single axial level according to somite stage (as in arrow b, in Fig. 1E), or along the caudo-rostral axis of an individual embryo (see Fig. 6A–C). A somite stage–dependent increase in the number of mitotic nuclei relative to total nuclei per segment (Fig. 6D) indicates a surge in growth rate coincident with somite budding.

The mitotic rate shift, which occurs between ss-1 and ss I, is consistent with a wave of increased rate of mitotic activity in PSM segments at ss0 as somites bud off. Berrill hypothesized that a wave of mitotic activity in concert with somite budding is an essential feature of segmentation, which represents a continuation of the synchronicity of embryonic cell cycles characteristic of early development of amphioxus embryos (Berrill,1955). If a mitotic wave passes through the segment during the 90-min ss0 interval, then the range in values for number of mitotic figures per segment during the ss-I to ss I interval may reflect its minima and maxima, which would appear (to the observer) as a dearth or abundance (peak) of mitotic nuclei, respectively. Temporal waves of mitotic activity have been correlated with limb segmentation (Lewis,1975) and clustered mitotic activity has been proposed to be responsible for the morphogenetic “hinge point” forming the body wall (Miller et al.,1999). The mitotic increase observed here is unique in that it represents a standing wave that passes through each forelimb-level segment as it matures from ss-I through ss0 to ssI.

Concentrations of mitotic nuclei also appear in the dorsomedial lip (DML) of each somite dermomyotome once myotome formation begins (Williams,1910; Lash and Ostrovsky,1986). The DML is also enriched for mitoses in the apico-basal planes, a characteristic of asymmetric cell division leading to terminal myotome cell differentiation (Venters and Ordahl,2005), whereas the majority of epithelial mitoses in other somite regions are parallel to the plane of the epithelium (Williams,1910; Venters and Ordahl,2005). DML mitoses were not analyzed in this study but the stage-dependent increase in myotome cell nuclei is caused by the surge of progeny from asymmetric mitoses in the DML. Thus, myotome growth and morphogenesis may represent a third mitotic surge restricted to one region of the somite (see Fig. 7).

thumbnail image

Figure 7. Hypothetical relationship between the somite cell cycle and the determination front. The three columns on the left show the developmental history of a single axial segment in somite stages correlated with the incidence of mitotic waves observed in this and other studies. The central column shows the cell cycle phases predicted to precede mitosis at ss0 and ssVII. The next two columns illustrate the hypothesized reduction in pre-S phase cells that occurs during PSM development (open arrow) and the FGF-dependent phase of PSM maturation and segmentation clock function (double-headed arrow), respectively. The right-hand column indicates coincident developmental transitions, including the catastrophic wavefront of entry into S-phase during PSM ss-IV as proposed in the text. df, determination front; Mn, M0, and Md, mitotic waves that occur rostral to Hensen's Node, at somite stage 0 and of myotome differentiation, respectively.

Download figure to PowerPoint

Somitogenesis and Somite Cell Growth Rate

The coincidence between the increased mitotic rate per segment at ss0 (Fig. 6A–C) and the low mitotic index at ss-I (Fig. 6D) is consistent with inhibition of cell cycle progression in the caudal PSM followed by sudden dis-inhibition leading to the concerted increase in mitotic activity at ss0. Cell cycle inhibition followed by a coherent increase in S-phase entry was previously proposed to be coupled to morphogenesis and differentiation in the developing chick limb bud (Solursh and Reiter,1975; Cooke and Summerbell,1980) and segmentation in amphibian embryos has been linked to arrest in the G2 phase of the cell cycle (Cooke,1979). We suggest an alternative and/or additional type of inhibition of cell cycle progression due to the effects of FGF signaling on caudal PSM cells (Dubrulle and Pourquie,2004). Assuming that the cohort of cells that undergo mitosis at ss0 all strictly obey the average cell cycle timing events defined in this study, then the final opportunity for those cells to enter S phase occurs approximately 3 ss earlier at somite stage-IV, a point in PSM development that coincides with the determination front (Fig. 7). PSM cells that enter S phase at ss-III would spend 2 somite stages in S-phase followed by 1 ss in G2 phase before entering mitosis at ss0 (Fig. 7). The 3-ss spacing between the mitotic shift at ss0 and S-phase entry also coincides with the rostral boundary of the FGF signaling gradient (Fig. 7), which has been hypothesized to reflect the wavefront component of the Clock and Wavefront model (Pourquie,2003; Baker et al.,2006a). However, the catastrophic alteration in cell “state” hypothesized to be precipitated by the interaction between the clock and wavefront (Cooke,1998) has not been identified.

Mitotic division is an irreversible and, therefore, catastrophic change in cell state, which is the outcome of irreversible entry into S phase of the cell cycle. A sudden increase in S phase entry in each segment should occur at the point where the density of PSM cells remaining in G1 phase becomes too low to sustain growth factor signaling (Zetterberg et al.,1995). The collapse of FGF-signaling and rush of predicted S phase entry are both located at ss-IV, the location of the determination front (Fig. 7). We suggest that catastrophic changes are inter-related and together constitute a coordinated wavefront of G1-to-S phase transition through which a cohort of PSM cells “pass” (or progress) and that they do so in a rostral-to-caudal direction as predicted in the clock and wavefront model of somite formation (Cooke and Zeeman,1976). We further suggest that imposition of rostro-caudal identity to individual PSM cell may be related to clock oscillations during G1 phase of the cell cycle.

The results reported here apply to forelimb somites and the extent to which the conclusions may apply to somites at other axial levels remains to be established. The nuclei in occipital somites double more rapidly than forelimb level somites (4.5 hr vs. 11 hr) during the brief period before they begin to dissociate (see Experimental Procedures section, Squash preparations) and as expected for mesoderm cells in day-1 chick embryos (Sanders et al.,1993). In addition, the appearance of desmin-immunoreactivity is delayed initially in cervical somites rostral to forelimb level somites (i.e., snzz<15) but, after desmin expression is initiated in these more rostral somites, it accumulates at a rate even more rapid than that of forelimb level somites (Borman et al.,1994; Borman and Yorde,1994a). Finally, the hyperplastic growth surge at the point of somite budding reported here may also be compatible with recent mathematical and conceptual modeling of somitogenesis (Schnell and Maini,2000; Collier et al.,2000; Kaern et al.,2000; Kerszberg and Wolpert,2000; Jaeger and Goodwin,2001; Andrade et al.,2005; Baker et al.,2006b) as well as observations reported much earlier (Berrill,1955; Schwartz,2007).


  1. Top of page
  2. Abstract
  7. Acknowledgements

Preliminary Somite Cell Counting Studies

Determination of the number of cells within somites is difficult because of their small size and the low number of cells they contain. A preliminary study (Ordahl,1993), in which individual somites were dissociated by trypsinization and counted in a hemocytometer, indicated that newly formed fore-limb-level somites contain approximately 2,000–3,000 cells. The dilution/multiplication factor of 100 involved in hemocytometer counting, however, precluded more accurate counts. Occipital somites (sn1–6) are transient in appearance (Huang et al.,1997) but have fewer cells. Nuclear numbers in the six rostral-most somites, as determined from isolated somite squashes, indicated that these somites contain approximately 850 cells at the time of their formation and that cell doubling time is ∼4.5 hr (Hultner,2001). The more rapid cell cycle duration in these occipital somites is consistent with shorter cell cycle period estimates for mesoderm at this early stage in chick embryos (Sanders et al.,1993).

Somite Cell Counting by Nuclear Enumeration

The enumeration of somite nuclei by confocal scanning, the approach employed for this study, had several attributes that all contribute to accurate counting of all the nuclei within individual somite segments. First, somite nuclei were enumerated in situ, using image stacks of fixed tissue. This allowed the intrinsic morphological boundaries to be used to discriminate between nuclei within the individual somite or presomite segment and those outside. Elimination of the need to isolate somites by dissection also eliminated potential for loss of somite cells, or the adventitious inclusion of non-somite cells in the sample. Second, because all of the nuclei within a particular somite segment are enumerated, and image stacks can be analyzed multiple times, reproducibility can be objectively determined (see Counting Controls section, below).

The embryos used in this study were from a single commercial source and comparisons between different sources of eggs may demonstrate more variability in the nuclear numbers per segment per stage. The source of greatest variation in the present studies was the indexing of somite stage. Improved resolution of somite stage into sub-stages, possibly by indexing the formation of precocious segmentation fissures (Rhee et al.,2003; Sato and Takahashi,2005), may improve resolution in the measurement of segment growth rate by this method.

Embryo Preparation for Confocal Analysis

Fertilized chicken eggs obtained locally (Petaluma, CA) were incubated at 38°C for pre-determined periods, and the embryos excised and sorted according to their total number of fully formed somites (Ordahl and Christ,1997). Embryos were washed in Tyrode's solution and PBS before fixation in ice cold 4% paraformaldehyde in PBS for 45 min. Fixed embryos were incubated in 1M HCl for 1 min followed by 0.1M sodium thiosulfate and, after Feulgen-staining with Schiff's Reagent (Fisher) for 30 min, washed in 0.1M sodium thiosulfate and PBS. Feulgen-stained embryos were optically cleared (Kardon,1998) and then mounted on glass slides in mixture of Benzoic Acid:Benzyl benzoate for confocal imaging of fluorescent nuclei.

Confocal Microscopy and Imaging

A Z-series set of confocal images was collected for each somite or presomite segment in this study by capturing the red fluorescence of the Feulgen-stained nuclear DNA using either a Zeiss LSM or a Nikon PCM confocal microscope. The Z-series stack was collected in a dorsal-to-ventral direction with the X-Y plane of each image parallel to the coronal plane and the Z-axis of the image set parallel to the dorsal-ventral axis or, for myotome nuclei, perpendicular to the plane of the dermomyotome epithelium. Optical sections were collected at 2–3 μM such that each nucleus was represented in multiple adjacent images. The limitation of the confocal microscope comes from the loss of resolution as the section depth increases. The brightness and the specificity of the Feulgen staining, along with optical clearing, yielded a practical depth limit of 100 to 120 μM into the specimen.

Somite Staging

Somite counting and staging is based on the methods outlined previously (Ordahl,1993; Christ and Ordahl,1995; Pourquie and Tam,2001). In practice, this involves assignment of somite stage I as the most caudal somite that is fully epithelial, regardless of the status of somite stage 0, which may be anywhere between 1–89 min from its own final epithelialization event. Thus, among the somites assigned to stage 0, some were slightly younger and some slightly older than estimated. This also affects somites at later stages that also may be at the beginning or end of one stage. Therefore, the “horizontal error bars” on the somite stages reported here are approximately ± 1/2 ss. Measurement variability can often reflect uncertainties in the measure of intrinsic developmental time that occurs during transitional periods in development (Ordahl,1986; Wong and Ordahl,1996).

3D Enumeration of Nuclei in Segments of the Paraxial Mesoderm

Nuclear counts were limited to ss IX and younger somites and presomite segments in the rostral half of the PSM. In these regions, nuclear densities permitted discrimination of every nucleus throughout the dorso-ventral Z-axis of the PSM (Fig. 1). At later somite stages (somite stage X and higher), de-epithelialization in several regions of the somite, and overall somite size, made enumeration of every somite nucleus not feasible for the present study. Early myotome nuclei, which lie immediately subjacent to the dermomyotome epithelium, could be resolved in somite stages greater than somite stage VII using a confocal z-axis perpendicular to the myotome layer (Fig. 3).

A comprehensive count of nuclei for each segment was conducted manually by digitizing the (x,y) position of the center of each cell nucleus in each image of the serial set using copyrighted Surfdriver 3D reconstruction software (Hultner,2001). Manual marking was necessary because inter-nuclear distances were too small to be discriminated by automated systems (data not shown). A 3D cluster analysis was then performed to determine the position of the center point of each nucleus in 3-dimensions within the segment (Fig. 1C). Cluster analysis was performed by Perl scripts written to process the (x,y,z) coordinates from SurfDriver. A distance-based cluster analysis algorithm was used to find clusters of points corresponding to individual nuclei. The discriminator of the algorithm relied on two assumptions regarding the position of the nuclei: (1) That center-points belonging to one nucleus have a root-mean-square distance that is less than points that belong to another nucleus; and (2) that all the points in one nucleus are contiguous along the Z-axis. The algorithm first finds contiguous points in the Z-direction and then searches for points within a radius less than a discrimination threshold. Points are contiguous if the closest point on an adjacent XY plane is within a threshold distance determined by calculating the minimum distance between points on the adjacent image plane. If an acceptable point cannot be found in an adjacent image plane, the cluster search is halted in the plus Z-direction and the search is conducted in the minus Z-direction. The cluster search is concluded when both Z directions have been searched. The central (x,y,z) coordinate of the cluster is then calculated to represent the center of one nucleus. This is repeated for each nucleus in the segment and the center-points total for each somite and PSM segment was recorded as the number of nuclei per segment (Table 1).

To enumerate the nuclei in presomite segments, caudal regions of the PSM were scanned by confocal microscopy to create multiple z-series stacks of overlapping register along the length of the PSM (Fig. 1D). A section interval of 1.5 μM was used to collect the image data and for counting purposes. A composite of the scans was developed using points of overlap that were identified and correlated with a scale factor relating axial length (in microns) to pixels in the image sets. Each composite scan was then digitized and processed as described above such that the positions of each nucleus were collected into a separate file on the computer. The cell data were then queried as a function of axial interval to determine the number of cells in each somite equivalent along the segmental plate. A somite equivalent is a segment of the PSM consisting of cells that will contribute to one somite segment after epithelialization. This interval is taken as the equivalent to the rostro-caudal dimensions of the stage I somite (Pourquie and Tam,2001) (shown in Fig. 1D).

Counting Controls

The potential for human error in the enumeration of the cells in somites is important to control. The computer-assisted counting method relies on the operator to discriminate individual somite nuclei and mark their centers in successive images in a z-series stack using a computer application. To determine the potential magnitude of this interpretive error, the cells of one somite were counted multiple times by one operator and again by multiple operators. The average single operator variability was approximately 50 cells out of 1,500 or less than 4% (n = 5). The inter-operator variation was 78 cells out of 1,500 or 5.2% (n = 4). The accuracy of the per-segment nuclear enumeration, therefore, far exceeds that of other variables such as somite staging and contributions from inter-embryo variability.

Counting Mitotic Nuclei

Mitotic nuclei were scored on the basis of the characteristic appearance of condensed chromosomes in prophase through metaphase nuclei. The whole-mount method used here does not provide subcellular detail and this tends to exclude nuclei in early M-phase prior to full chromosome condensation. Because this should affect all samples equally, it should not affect the comparisons of paraxial mesoderm segments reported here, all of which were conducted using the same whole-mount method.

Statistical Analysis

Mathematical analysis of nuclear enumeration data in Figures 2, 3, and 6 were performed using SigmaPlot to provide the curve fits indicated in the figure legends. For purposes of curve-fitting, all somite stages were given positive values. The statistical significance of nuclear and mitotic counts compared between grouped sets of somites was tested using an unpaired t-test.

Fluorescence-Activated Cell Sorting Using Paraxial Mesoderm Tissue

Fluorescence-activated cell sorting (FACS) requires mono-dispersed cells in large numbers (Watson et al.,1987). To obtain sufficient cell numbers, the paraxial mesoderm (PSM + formed somites) was dissected en bloc from ∼5 dozen ED2 chick embryos containing 15–20 formed somites (Ordahl and Christ,1997). The dissected paraxial mesoderm fragments were then treated with 0.25% trypsin STV for 30 min at 37°C followed by placing the trypsinized tissue onto a 10-ml column of 10% Fetal Bovine Serum and allowing tissue fragments to descend by gravity to inactivate trypsin. After aspirating the supernatant, the cells in the pellet were dissociated by gently vortexing for 10 min in 1 ml 10% FBS and the dissociated cells immediately fixed in 70% ethanol in PBS at room temperature for 1 hr, followed by washing in PBS and digestion of RNA by treatment with 10 μg/ml RNaseA (Sigma) for 30 min at 37°C. The dissociated paraxial mesoderm cells were then stained with 50 μg/ml propidium iodide (Molecular Probes) in PBS for a minimum of 15 min prior to being analyzed with a Becton-Dickinson FACScan flow cytometer (UCSF Cancer Center). Scattergrams were analyzed with CellQuest 2.1 software. Cell cycle fractions were estimated with ModFit 1.0 software to integrate peaks and estimate overlaps of the curves according to the Watson Pragmatic model (Watson et al.,1987). These calculations assume that all cells are cycling and the presence of G0 cells affects that assumption. The forelimb level somites in 20 somite embryos included for FACS were no older than ssV so differentiated myotome cells were not yet present in these segments. However, differentiated myotome cells have been shown to be present in more rostral cervical segments in embryos containing 20 somites or less (Borman et al.,1994; Borman and Yorde,1994a,b). In addition, small numbers of committed, post-mitotic but not yet differentiated, myotome precursor cells are present in forelimb somites at epithelial stages (Williams and Ordahl,1997). The fractional contribution by post-mitotic cells, and potentially other cells in G0 phase of the cell cycle, is therefore unknown but probably small (<10%, estimate) and their contribution is considered to be negligible for the estimation of the relative duration of cell cycle phases made here.

BrdU Labeling of Mitotic Nuclei in Somites

E2–E2.5 chicken embryos were treated with 50 μl of 10 mM BrdU in PBS pipetted over the surface of the embryo, after which they were reincubated in ovo for 30, 60, 90, or 180 min, and then fixed in 2% paraformaldehyde for 2 hr at room temperature. These embryos were processed for sectioning and immunohistochemistry as described previously (Venters and Ordahl,2002) using antibodies against BrdU (Dako) and phosphorylated H3 (Upstate Biotechnology) diluted to 1:500. Antibody labeling was revealed using goat anti-mouse conjugated with alexa488 and goat anti-rabbit conjugated with Alexa555 secondary antibodies (Molecular Probes) diluted to 1:500. Z-series of sagittal or longitudinal sections through epithelial somites (ssI–ssIII) were collected using a Nikon PCM confocal microscope. Images were compiled using Adobe Photoshop and mitotic nuclei at the apical epithelium surface (phosphorylated H3 positive) scored for incorporation of BrdU.

Induction of Segmentation Defects by BrdU Treatment

The method for BrdU treatment outlined below was based on a previously published method (Primmett et al.,1989). Embryos were staged and BrdU (0.5 μg/ml in Tyrodes) was injected under the blastoderms of one dozen staged chick embryos (Ordahl and Christ,1997) after which the eggs were re-sealed with parafilm and returned to incubation for a period of 2 to 2.75 hr. After timed exposure to BrdU, the treated embryos were excised from the egg, washed in BrdU-free Tyrodes solution, and explanted into New culture (New,1959) for development in the absence of BrdU. After overnight incubation, 11 surviving embryos were examined under a dissecting microscope and the number of somites counted to determine intrinsic elapsed time in somite stages. That and the location of defects in the left and right paraxial mesoderm strips were recorded and each embryo was then fixed in 4% paraformaledehyde in PBS for preservation and re-examination. Results are given in Table 5.


  1. Top of page
  2. Abstract
  7. Acknowledgements

The authors thank Fiona Deegan and Rebecca Argent for expert technical assistance, Duane Coughlan for expert assistance with statistics, and the students who participated in the somite counting project. This work was supported, in part, by research grants to C.P.O. from NIH and MDA.


  1. Top of page
  2. Abstract
  7. Acknowledgements