Morpholinos for splice modificatio

Morpholinos for splice modification

Advertisement

Summary

Cardiogenic fate maps are used to address questions on commitment, differentiation, morphogenesis and organogenesis of the heart. Recently, the accuracy of classical cardiogenic fate maps has been questioned, raising concerns about the conclusions drawn in studies based on these maps. We present accurate fate maps of the heart-forming region (HFR) in avian embryos and show that the putative cardiogenic molecular markers Bmp2 and Nkx2.5 do not govern the boundaries of the HFR as suggested in the literature. Moreover, this paper presents the first fate map of the HFR at stage 4 and addresses a void in the literature concerning rostrocaudal patterning of heart cells between stages 4 and 8.

INTRODUCTION

Studies to identify the HFR in the avian embryo have been carried out since the early 1900s (Rawles, 1936; Rawles, 1943; Rudnick, 1948; DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966; Rosenquist, 1966; Rosenquist, 1970; Garcia-Martinez and Schoenwolf, 1993). Briefly, at stage 3, cardiac progenitor cells have been mapped to the anterior two thirds of the primitive streak (PS), excluding Hensen’s node (HN) (Garcia-Martinez and Schoenwolf, 1993). At stages 4 and 5 cardiac progenitor cells are located on either side of HN in the anterior mesoderm (Rawles, 1936; Rawles, 1943; Rudnick, 1948; DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966; Rosenquist, 1966; Rosenquist, 1970). At stage 6, cardiac progenitor cells occupy a crescent-shaped region in the anterior lateral mesoderm, from where they migrate in an anterior and medial direction to fuse at the head fold region by stage 7 (DeHaan and Urspurng, 1965).

Recently, Erhman and Yutzey (Erhman and Yutzey, 1999) placed the lateral border of the HFR (stage 5) at the lateral most region of the mesoderm, adjacent to the extra-embryonic tissues, more lateral than that proposed by earlier studies (Rawles, 1936; Rawles, 1943; Rudnick, 1948; DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966; Rosenquist, 1966; Rosenquist, 1970); and the posterior border at the level of HN. At stage 8, the posterior boundary of the HFR was placed at the level of the first condensing somite (Erhman and Yutzey, 1999), which is more anterior than that previously defined by others (Rawles, 1936; Rawles, 1943; Rudnick, 1948; DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966; Rosenquist, 1966; Rosenquist, 1970). In addition, recent findings by Colas et al. (Colas et al., 2000) describe the heart as arising from paired and separated regions rather than a single crescent, which fuses at the midline between stages 9− and 9. Thus, the boundaries of the HFR (between stages 4 and 8) and the origin of the cells that contribute to the various regions of the developing heart remain unclear. In order to clarify this issue, we have generated cardiac fate maps from stages 4-8 of avian development.

The newly established boundaries generated by us accurately represent the HFR. The cardiogenic fate map of the stage 4 embryo presented in this paper is the first one ever to be generated and reported. Our fate maps were generated by labeling a small group of cells/embryo with DiI (1,1′-dioctadecyl-3, 3,3′, 3′-tetramethyl-indocarbocyanine perchlorate), which, although simple, is more direct and technically accurate than the earlier experiments (Rawles, 1936; Rawles, 1943; Rosenquist and DeHaan, 1966; Erhman and Yutzey, 1999). We also discuss the boundaries of our fate maps in relation to the expression of putative cardiogenic molecular markers Bmp2 (Wozney et al., 1988) and Nkx2.5 (Bodmer, 1993), which have been proposed to govern events during early heart development. Our data suggest that neither of these molecules accurately mark the entire cardiogenic region during early embryogenesis. These newly generated fate maps accurately represent the in vivo cardiogenic region and will serve as a standard for studies on heart development.

MATERIALS AND METHODS

DiI labeling and image analysis

Fertilized White Leghorn chicken eggs were incubated at 37°C. Embryos were dissected out at stages 4-8 (Hamburger and Hamilton, 1951) using sterile filter paper rings and cultured ventral side up on 35 mm dishes coated with 0.6% agar:albumin in a 1:1 ratio (Patwardhan et al., 2000). A 5 mg/ml stock solution of DiI (Molecular Probes, USA) was prepared in absolute ethanol containing 10% glycerol and stored at –20oC. A freshly prepared working dilution (1:500) in phosphate-buffered saline (PBS) was used. DiI was injected into the extracellular space around a small number of cells in the mesodermal layer using pulled glass needles and a pressure injector (Narishige, Japan). In some cases, control embryos were labeled with DiI and DiO (3,3′-dioctadecyloxacarbocyanine perchlorate). A 5 mg/ml stock solution of DiO was solubilized in N,N-Dimethylformamide and freshly diluted (1:500) in PBS prior to use. The embryo was repeatedly washed (post-injection) with PBS to ensure that particles of DiI that were not adherent to the labeled group of cells were removed and did not result in nonspecific labeling 20 hours later. We chose to evaluate the DiI and DiO results under fluorescence microscopy, rather than bright field (Erhman and Yutzey, 1999), because the latter technique leads to the underestimation of the final population of labeled cells. DiI and DiO were visualized using a rhodamine and fluorescein optical filter, respectively. Injections were random on either the left or right side of the primitive streak (PS). Many of the injections in a given location were repeated in different embryos to ensure that consistent results could be obtained. A large selection of cells was labeled on both sides of the PS; however, not all areas were labeled on both sides of the PS. Therefore, data from both sides of the PS were considered together to determine the boundaries of the HFR. In the majority of cases, the injected embryos were photographed at ‘0’ hours using the Nikon Eclipse E800 fluorescence microscope and the Optronics digital imaging system. The embryos were then allowed to develop for an additional 20 hours at 37°C, after which they were again photographed. In a few cases, to address the issue of rostrocaudal patterning, embryos were injected with both DiI and DiO, and followed at intervals over 20 hours.

In all cases, embryos were imaged identically to ensure that the original proportions of each embryo was maintained. Fig. 1 shows an example of the imaging process. The fluorescent and bright field images at each time point, ‘0’ and 20 hours post-injection, were superimposed in Adobe PhotoShop 5.5 and the ‘difference’ option in the ‘show layers’ function was used to determine the exact position of DiI label at the two time points separately. The embryo from which the data was derived is identified by a number, which appears on the grids.

Fig. 1.

Imaging and grid to locate position of DiI injections at ‘0’ hours. All images are ventral views of the embryo. DiI labeled cells are seen in red. (A) Fluorescence image of embryo 127 (stage 4+). (B) Bright field image of the same embryo; (C) superimposed images of the embryo shown in A,B with an overlay of the same grid used on the template at stage 4 (Fig. 2A). (D-F) Imaging of the same embryo 20 hours later: (D) fluorescence image; (E) bright field image; (F) superimposed images in D,E showing the location of DiI in the ventricle 20 hours post-injection. HN, Hensen’s node; Ht, heart; arrows indicate superimposed images of A and B in C, and D and E in F.

Sectioning of control embryos

To ensure that cells in the mesodermal layer were labeled, embryos at stages 4 and 6 were injected with DiI or DiO, fixed immediately in 4% paraformaldehyde/PBS, embedded in 2% agar and sectioned (100 μm to 200 μm) using a vibratome.

DiI localization using the grid system

Representative images of embryos at stages 4-8 were used as templates to record injection sites at ‘0’ hour. A grid was superimposed on each template. The axis of the embryo was placed centrally on the grid at column location ‘0’. Equidistant columns to the left and right of the PS were designated from –1 to –7 and +1 to +8, respectively. Equidistant rows (from A-S) intersected the vertical columns to create the grid. At stage 4, the rostralmost point of HN was positioned on the line between rows F and G, and the area opaca/area pellucida (AO/AP) boundary in the anterior region (above HN) was in the middle of row A. At stage 5, the line between rows E and F was positioned in the region of HN, where an asymmetric deflection is seen to either the left or right. The AO/AP boundary was above row A. At stage 6, the beginnings of the head fold was immediately above the line between rows B and C, and the regressed HN in row J. The AO/AP boundary was in row A. At stage 7, the first somite was placed in row G, the regressed HN in row L and the head fold in the row immediately above row A. At stage 8, the first (anteriormost) somite was placed in row H, and the arch above the anterior intestinal portal (AIP) in the row above row A. An identical grid, as used on the template image, was placed over the composite bright field and fluorescence image of each embryo at ‘0’ hours in Adobe PhotoShop 5.5.

Owing to small variations in the size of embryos at comparable developmental stages, an absolute measure of position was not used to mark the location of the labeled cells. Instead, we used the Cartesian coordinate grid system in which the coordinate points in the grid were fixed, but the dimensions of the grid were adjusted proportionately to fit the size of each embryo (Tam and Schoenwolf, 1999). In addition, to correct for the minor variations in size of the embryos, relative distances from the site of injection were measured (1) to HN and the PS for stage 4, (2) from the regressed HN and the PS at stage 5, (3) from HN, the PS and the head fold at stage 6, and (4) from the newly forming somites, the central axis and the head fold at stages 7 and 8. The AO/AP were also used as guides in marking the injection site on templates. This allowed for accurate transposition of injection site location onto the respective stage-specific template. The accurate transposition of the site of injection onto the templates was also preserved because all embryos were carefully imaged using identical microscope and imaging parameters. The movement of the DiI dye 20 hours post-injection was documented and correlated (in terms of location), to the initial site of injection.

In situ hybridization

Immediately after labeling, embryos (stages 5-8) were imaged as described above and processed for in situ hybridization according to Wilkinson (Wilkinson, 1992) with modifications as recommended by Schultheiss (Schultheiss et al., 1995). The digoxigenin-labeled Nkx2.5 antisense mRNA probe was generated by in vitro transcription (Boheringer Mannheim). The Nkx2.5 cDNA was generously provided by T. Schultheiss (Harvard Medical School, Boston, MA). As a positive control, stage 11 embryos were stained for Nkx2.5 expression using the probe described above. The alkaline phosphatase reaction was developed for 24 hours for stages 5 and 6, and for 2 hours for stages 7-8 and 11 embryos. These reaction times were similar to those used by T. Schultheiss (personal communication).

RESULTS

HFR at stages 4 and 5

Mesodermal cells at stages 4 (n=47) and 5 (n=63) were marked by DiI injection (Fig. 1). A summary of all injections at ‘0’ hours for stages 4 and 5 are shown on templates in Fig. 2A,C, respectively. Table 1 provides detailed information on the final location of all DiI-labeled cells for each embryo injected at stages 4 and 5. Embryos included in the data set developed normally. Most embryos (>95%) observed post 20 hour incubation, had beating hearts. Accurate localization of the labeled cells was further facilitated by the fact that the heart tissue was pulsating. The boundaries of the HFR at stage 4 were deduced from the sum of data presented in Fig. 2A. On the stage 4 embryo template, the left-right axis in the region of the AP at the level of HN was divided into 10 divisions (columns –5 to +5) with HN at position ‘0’. The HFR extended from –3 to –5 on the left side of the PS and from +3 to +5 on the right side of the PS. Cells in the region most medial and adjacent to the PS (columns –1 to –2 and +1 to +2) did not contribute to the heart. The anterior boundary was just rostral to HN (row F) and the posterior boundary was caudal to HN at row H. When the HFR at stage 4 (demarcated by black rectangles) was aligned on a stage 4 embryo stained for expression of Bmp2 mRNA (Fig. 2B), the Bmp2 staining was found towards the lateral regions of the HFR but was not restricted to this area. The Bmp2 expression domain extended posterior and anterior to the HFR and was also present in the posterior PS.

Fig. 2.

Fate map of embryos at stages 4-6. (A-C) A composite template of embryos with all injections at ‘0’ hours represented by a circle and number for each embryo, at stages 4 (A), 5 (C) and 6 (F). The color of the circle denotes the region of the heart (yellow, sinus venosus; blue, ventricle; red, atria; black, bulbus cordis or vitelline artery; and white, those that did not enter the heart) to which the DiI-labeled cells became incorporated 20 hours post-injection. The HFR is bilaterally demarcated by a black rectangle (see Materials and Methods) in A,C,F. The HFR at respective stages was superimposed on embryos stained for mRNA expression of Bmp2 at stages 4 (B), 5 (D) and 6 (G), and Nkx2.5 at stages 5 (E) and 6 (H). aip, anterior intestinal portal; hn, Hensen’s node; hp, head plate; ps, primitive streak. (I) The pre-cardiac region from Rosenquist and DeHaan (1966), as shown in Schultheiss et al., 1997. B,G,H reproduced, with permission, from Schultheiss et al., 1997; D,E reproduced, with permission, from Ehrman and Yutzey, 1999.

The boundaries of the HFR at stage 5 were similarly deduced, from the sum of data presented in Fig. 2C. On the stage 5 template, the region of the AP at the level of HN extended from –4 on the left of the PS to +4 on the right of the PS. The HFR extended from –2 to –4 on the left side of the PS and from +2 to +4 on the right side of the PS. Cells in close proximity (columns −1 and +1) to HN did not contribute to heart tissue. The anterior border of the HFR was just above HN in row E and extended posteriorly to row H. When the HFR (black rectangles) at stage 5 was superimposed on the in situ staining patterns of Bmp2 and Nkx2.5 mRNAs (Fig. 2D,E, respectively), it extended beyond both stained regions, both medially and laterally. As at stage 4, Bmp2 staining extended posterior and anterior to the HFR and was present in the posterior PS and only the anterior border of the HFR overlapped the posterior domain of Nkx2.5 staining (Fig. 2E).

HFR at stage 6

The boundaries of the HFR were deduced from the sum of the data presented in Fig. 2F (n=28). Table 2 provides detailed information on the final location of all DiI-labeled cells for each embryo injected at stage 6. Cells located in the anterior region of the embryo, adjacent to the embryonic axis (columns –1 and +1), and those located to the far lateral regions (columns −4 and +4) of the AP, adjacent to the AO, did not incorporate into the heart. Labeled cells anterior and lateral to the regressing HN in columns –2 and –3 on the left side of the PS and +2 and +3 on the right side of the PS contributed to the heart. The anterior boundary of the HFR was at row E, and the posterior boundary was just posterior to the regressed HN in row J. At this stage, the HFR (black rectangles) overlaps the region stained for Bmp2 mRNA (Fig. 2G) to a greater extent than at stages 4 and 5, but is not completely coincident. At stage 6, the expression of Nkx2.5 mRNA encompasses only the anterior half of the HFR (Fig. 2H).

HFR at stages 7 and 8

The boundaries of the HFR at stages 7 (n=36) and 8 (n=55) were deduced from the sum of data presented in Fig. 3A,D, respectively. Table 3 provides detailed information on the final location of all DiI-labeled cells for each embryo injected at stages 7 and 8. At stage 7, DiI-labeled cells in the medial/paraxial mesoderm (columns –1 and +1) did not incorporate into the heart, but were located in either head or axial mesoderm. Cells anterior to the first forming somite in the lateral mesoderm (columns −2 and –3, and +2 and +3) incorporated into heart tissues. The anterior boundary of the HFR was at row B (below the head fold) and the posterior border was below the first somite at row G. At this stage, the Bmp2 mRNA expression not only overlaps two-thirds of the anterior HFR (black rectangles, Fig. 3B) but also extends anteriorly beyond the HFR and is present in the posterior PS. Nkx2.5 expression encompassed the anterior half of the HFR and extended anteriorly into the head fold region.

Fig. 3.

Fate map of embryos at stages 7 and 8. (A,D) A composite template of embryos with all injections at ‘0’ hours, represented by a colored circle (yellow, sinus venosus; blue, ventricle; red, atria; black, bulbus cordis or vitelline artery; and white, those that did not enter the heart) and number for each embryo at stages 7 (A) and 8 (D). The HFR at stage 7 (black rectangle, see Materials and Methods) was superimposed on a stage 7 embryo stained for Bmp2 expression (B) and Nkx2.5 expression (C). (E) The HFR at stage 8 is represented (on only one side) as a red domain on a stage 8 embryo stained for expression of Nkx2.5 mRNA. B,C reproduced, with permission, from Ehrman and Yutzey, 1999. E reproduced, with permission, from Schultheiss et al., 1995.

The HFR at stage 8 (Fig. 3D) was fused anteriorly above the AIP and extended posteriorly to the level of the fourth somite. Laterally it included cells on either side of the embryonic axis, from columns –2 to –4 and from +2 to +4. At stage 8, a clear rostral to caudal pattern of organization was detected in cells contributing to the various regions of the heart. Cells that will become part of the sinus venosus (SV) were located in the posterior region of the HFR while atrial and ventricular cells were found to be more rostral in position (Fig. 3D). The HFR on the right side (red domain, ventral side-up) was aligned on a stage 8 embryo stained for expression of Nkx2.5 mRNA (Fig. 3E). The HFR in Fig. 3E extended posteriorly, beyond the expression pattern of Nkx2.5 mRNA.

Specificity of labeling cells in the mesodermal layer

To ensure that cells in the mesoderm were labeled, stages 4 (Fig. 4) and 5 (data not shown) embryos (n=10) injected with DiI or DiO were immediately fixed post-injection, sectioned using a vibratome and viewed using fluorescence microscopy to locate DiI- or DiO-labeled cells. In eight out of the ten cases, labeled cells were detected solely in the mesoderm. In one case, labeled cells were detected in the mesoderm and ectoderm and in another embryo they were detected in the mesoderm and endoderm.

Fig. 4.

Specificity of labeling technique. (A) Superimposed bright field and fluorescence image immediately post-injection of a stage 4+ embryo, ventral side upwards, injected with DiO (green). (B-1,B-2) 150 μm section at the level shown in A (×100). Because of the thickness of the section, B-1 shows an image where the ectoderm and mesoderm layer are in focus and B-2 shows the mesoderm and endoderm in focus. (C) 150 μm section at the level shown in A (×200). The arrow denotes the location of the primitive groove. Ec, ectoderm; En, endoderm.

Rostrocaudal patterning in the avian embryonic heart

Data derived from our fate maps (Figs 2, 3) suggested that the previously discussed (Fischman and Chien, 1997) rostrocaudal pattern in the embryonic HFR is not present between stages 4 and 6. To confirm our data, we double labeled embryos (n=20) at stages 4 and 5, and followed them at intervals over a 22-24 hour period. An example of one such embryo is shown in Fig. 5. The point is best made by focusing on the DiO (green)-labeled cells on the right side of the embryo (ventral side upwards). The cells proliferated and migrated to give rise to part of the ventricle anteriorly (arrow) and the SV posteriorly (arrowhead). Also, at ‘0’ hour, although the DiI-labeled cells were more rostral to the DiO-labeled cells, they intermingled during migration and did not retain their initial rostral and caudal positions, respectively.

Fig. 5.

A definitive rostrocaudal pattern was not present in the HFR at stages 4-6. Stage 4+ embryo (number 758) injected with DiI (red) and DiO (green) at ‘0’ hours was photographed at 4, 8, 10, 15 and 22 hours during development. Images shown are the fluorescent DiI and DiO images superimposed on each other and on the bright field image at each time point in Adobe PhotoShop 5.5. Arrow denotes DiO-labeled cells in the ventricle and arrowhead denotes the DiO-labeled cells in the sinus venosus.

Nkx2.5 does not mark the entire HFR

To confirm that Nkx2.5 expression does not mark the entire HFR, cells (stages 5-8) in the HFR expected to form the heart but that did not express Nkx2.5 according to Schultheiss (Schultheiss et al., 1995) and Erhman and Yutzey (Erhman and Yutzey, 1999) were tagged with DiI. A subset of these embryos (n=30) were processed for in situ hybridization to detect Nkx2.5 mRNA expression immediately post-labeling. The embryos were re-imaged after the in situ hybridization process, and the pre- and post-in situ hybridization images were compared (two examples are shown in Fig. 6A,B). These data confirmed that Nkx2.5 expression did not overlap the entire HFR, because the injection site was outside the Nkx2.5-stained region and these labeled cells incorporated into the heart. The remaining embryos (n=10) were allowed to develop for an additional 20 hours and imaged to ensure that the tagged cells did indeed incorporated into the heart. These embryos were fixed, embedded in agar and sectioned using a vibratome. DiI-labeled cells were located as predicted, in the heart, head mesenchyme and pharynx (Fig. 6D-G).

Fig. 6.

Nkx2.5 expression was not detected in the entire HFR. All views of whole-mount embryos are ventral except in B2, where a dorsal view is presented. (A1,B1,E,G) The fluorescent image was superimposed on the bright field image. (A1) Embryo (three pairs of somites) injected with DiI (red dot) in the HFR, based on the HFR fate map from Fig. 3. Arrow indicates level of first somite. (A2) Whole-mount in situ hybridization of the embryo in A1 with digoxigenin-labeled probe of antisense Nkx2.5 (blue). Red dot indicates position of DiI injection, as in A1. Arrow indicates level of first somite. (B1) Embryo (four pairs of somites) injected with DiI (red dot) in the HFR, based on the HFR fate map from Fig. 3. Arrow indicates level of first somite. (B2) Whole mount in situ hybridization of the embryo in B1 with digoxigenin labeled probe of antisense Nkx2.5 (blue). Red dot indicates position of DiI injection, as in B1. Arrow indicates level of first somite. (C) Control stage 11 embryo stained with digoxigenin-labeled probe of antisense Nkx2.5. Nkx2.5 was expressed specifically in the heart. (D) Stage 5 embryo (number 888) injected with DiI (red dot) in the HFR, based on the fate map from Fig. 2C. Arrow shows site of injection. (E) Same embryo as in D, 20 hours post-injection, developed to stage 10. (F) Bright field image of a 400 μm section at level shown in E (white * indicates neural tube; black *, ventricle). (G) Superimposed bright field and fluorescent image of the same section as in F.

DISCUSSION

The DiI-labeling technique is widely used (Darnell and Schoenwolf, 1997; Psychoyos and Stern, 1996) to study migrating cell populations and to generate highly accurate fate maps (Garcia-Martinez and Schoenwolf, 1993; Hatada and Stern, 1994; Lumsden et al., 1991; Wilson and Beddington, 1996), because one small group of cells/embryo (as we have done) can be tagged and followed to their final destination. It is therefore preferred over the use of explant cultures of blastoderm (Rawles, 1936; Rawles, 1943) or labeled graft experiments (Rosenquist and DeHaan, 1966) that generate prospective potency maps or fate maps, respectively. Our approach is more direct (because we specifically tagged the mesodermal cell layer that contains cardiac progenitor cells) as opposed to that used by others in which the HFR has been defined by labeling the endodermal cell layer. This approach, which is based on the premise that endoderm and mesoderm cells co-migrate during early embryogenesis (Bellairs, 1953; DeHaan, 1963a; Erhman and Yutzey, 1999), is indirect. However, to ensure that we were accurately tagging cells in the mesodermal layer, embryos injected with DiI and DiO were sectioned. Embryonic sections confirmed that labeled cells were located in the mesoderm at ‘0’ hours. In addition, in a few embryos (embryos 26 and 29 in Fig. 2A, and embryos 34 and 376 in Fig. 2C) we stripped off the endoderm and placed DiI directly on to the mesoderm. Results from these embryos were identical to those in which the endoderm was not removed.

Small groups of cells labeled with DiI were followed and their collective fate was observed 20 hours post-injection. The data showed that in some cases, all labeled cells in a group were confined to only one region of the embryo (e.g. stage 8, embryo number 107; Fig. 3D), whereas in other instances, the labeled descendants were visualized in multiple regions of the embryo (e.g. stage 8, embryo number 570; Fig. 3D).

Our studies clearly showed that, at stage 4, the anterior border is just above the level of HN and extends to ∼one quarter the distance of the PS, below HN. The medial border was ∼0.3 mm from the primitive groove (PG), and the lateral border extended almost to the AO/AP boundary. This is the only published fate map of the HFR generated at stage 4 in the chicken embryo. The ‘cardiogenic specification marker’; Nkx2.5 (Lough and Sugi, 2000), has not been reported to be detected by in situ hybridization until stage 5 (Schultheiss et al., 1995), and therefore cannot be used to ascertain the location of cardiogenic cells prior to stage 5. However, Bmp2 expression can be detected at stage 4, but only in the lateral endoderm at the level of the anterior PS (Schultheiss et al., 1997) and, conversely, at low levels in the mesodermal layer (Andree et al., 1998). The discrepancy in location of Bmp2 at stage 4 in these two studies remains unresolved. In addition, the expression of Bmp2 was not restricted to the HFR, but was also detected in the head fold and posterior PS (Schultheiss et al., 1997; Andree et al., 1998). Our data clearly show that Bmp2 expression is not coincident with the HFR, thereby suggesting that Nkx2.5 and Bmp2 cannot be used to define the HFR.

Another observation shown in Fig. 1 is that labeled cells on the right of the PS (ventral side up) appear to contribute to the ‘C’ shaped wall of the ventricle. Our data are consistent with the findings of Stalsberg (Stalsberg, 1969). A few sentences from her paper are relevant: ‘She stained the presumptive heart regions of amphibian embryos with vital dyes and stated that the ventricle was later seen to be formed predominantly by material from the left side. She concluded that the primary loop of the heart is caused by an over growth of the left-sided material in the ventricular region’ (Stalsberg’s description of the work of Wilens, 1955). ‘At stage 10-, significantly more cells have been contributed to the epimyocardium from the right side than from the left side. This relation is maintained in the cephalic part of the bulboventricular loop at later stages. In the caudal part of the bulboventricular loop, the relation is reversed, with significantly more cells originating from the left side. The asymmetry in cell numbers may be the expression of a primary difference between the developmental potencies of right and left heart primordia through which the laterality of the heart loop formation is determined’ (Stalsberg, 1969). Therefore, our data are consistent with those of Wilens and Stalsberg, such that most of the ventricular loop labeled with DiI originated from cells on the left side of the embryo (Fig. 1A is the ventral view, therefore the right side).

The prospective HFR at stage 5 as defined by Rawles (Rawles, 1936; Rawles, 1943), by donor/host chorio-allantoic membrane transplant experiments, showed that the HFR was localized on either side of the PS extending from the anterior most point of the head process down to half the length of the PS (Fig. 8 from Rawles, 1936) (Fig. 6 from Rawles, 1943). The medial boundary was ∼0.13 mm from the primitive groove (PG) and extended almost to the AO/AP boundary. Studies using labeled grafts (Rosenquist and DeHaan, 1966) identified the HFR as being similar to that defined by Rawles, although the medial border was more lateral, ∼0.3 mm from the PG. More recently (Erhman and Yutzey, 1999), molecular markers (Nkx2.5 and Bmp2; ventricular myosin heavy chain 1 (VMHC1) (Bisaha and Bader, 1991); and antibody against sarcomeric myosin (Bader et al., 1982)) were used to identify the HFR at stages 5 and 7. In Erhman and Yutzey’s study, the lateral boundary of the HFR was placed adjacent to the extra embryonic tissue and the medial boundary ‘at the lateral border of the mesoderm overlying the prospective neural plate’, a more laterally located HFR than previously described (Rawles, 1936; Rawles, 1943; Rudnick, 1948; DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966; Rosenquist, 1966; Rosenquist, 1970). The posterior boundary was determined to be at the level of HN (Erhman and Yutzey, 1999). These experiments (Erhman and Yutzey, 1999), because of the techniques and marker (VMHC1) used, more appropriately define the ventricle-forming region rather than the entire HFR. Our in vivo studies at stage 5 mapped the anterior border of the HFR anterior to HN, and the posterior border ∼one quarter the distance of the PS below HN. The medial border was ∼0.3 mm from the PG, similar to that identified by Rosenquist (Rosenquist and DeHaan, 1966) but inconsistent with that of Rawles (Rawles, 1936; Rawles, 1943) and the lateral border was close to the AO/AP border. Our data do not agree with the recent mapping study (Erhman and Yutzey, 1999), in that the anterior border of the HFR in our study was coincident with the posterior border of Nkx2.5 with minimal, if any, overlap. To confirm that cells within our HFR but outside of the Nkx2.5 expression domain contribute to the heart, we injected stage 5 embryos (outside of the Nkx2.5 expression domain but within the HFR shown in Fig. 2C) and either processed them for in situ hybridization at ‘0’ hours or sectioning 20 hours post-injection (Fig. 6). Very low levels of Nkx2.5 mRNA were detected at stage 5 (confirmed by T. Schultheiss, personal communication). These data confirmed that the HFR extended posterior to the region of Nkx2.5 expression (Fig. 6A,B) and that the labeled cells outside the Nkx2.5 expression domain contributed to the heart (Fig. 6D-G).

According to the classical mapping studies, cardiac cells at stage 6 should lie in a crescent-shaped region on either side of a central embryonic axis, anterior and lateral to the regressed HN, in the lateral mesoderm (DeHaan and Urspurng, 1965; Rosenquist and DeHaan, 1966). At stage 7, the cells move closer towards the head fold region and fuse above the AIP between stages 6 and 7 (DeHaan, 1963a). We identified HFR cells above the AIP at stage 8, but not at stages 6 and 7, which suggests that the heart primordia fuse above the AIP between stages 7 and 8. The posterior border of the HFR at stage 8 extended to the fourth somite, which is more caudal than previously described (Erhman and Yutzey, 1999; DeHaan and Urspurng, 1965). This location of the HFR is further corroborated by our data on chamber specification (Patwardhan et al., 2000). The discrepancy in the location of the posterior boundary of the HFR at stage 8, can be explained by the fact, that we looked at cells contributing to all regions of the heart, while others (Erhman and Yutzey, 1999) have focused on ventricle-forming cells alone. We would place the posterior border of the ventricle-forming region at the same level as Erhman and Yutzey (Erhman and Yutzey, 1999), and this boundary is coincident with the posterior boundary of Nkx2.5 expression domain (Fig. 3D,E), suggesting that Nkx2.5 may better define the ventricle-forming region than the HFR.

Our data address another important aspect, which is rostrocaudal patterning during heart development. There is no direct evidence in the literature to suggest that cardiogenic cells continue to be ordered in a rostral-to-caudal direction at stages 4 and 5, as they are in the PS at stage 3 (Garcia-Martinez and Schoenwolf, 1993). Schoenwolf created quail-chick chimeras at stage 3 and observed the location of cells in the heart at stage 10, although this does not tell one exactly what happened at stages 4-9. The findings of Garcia-Martinez and Schoenwolf (Garcia-Martinez and Schoenwolf, 1993) and those of DeHaan (DeHaan, 1963a; DeHaan, 1963b) have been used by others to conclude that the cardiogenic cells, after gastrulation, remain in the same organized pattern between stages 3 (Garcia-Martinez and Schoenwolf, 1993) and 12 (DeHaan, 1963a; DeHaan, 1963b, reviewed by Fischman and Chein, 1997).

On the contrary, we found that between stages 4 and 6, the cardiogenic cells are not organized in a rostral to caudal pattern. Similar findings of DeHaan (DeHaan, 1963a) where iron oxide particles attached to the endoderm were used to track the movement of ‘migrating’ cells between stages 5 and 12 were described as ‘At stage 5 the regions of the thickened mesoderm (LHFR,RHFR) are broad and rather diffuse; however, they match nicely the areas shown to have heart forming capacity as explants. If the movements of individual clusters are traced on film through subsequent stages, it is noted that for the first few hours during stages 5-6, the direction of migration of a given cluster may bear no relation to its later “goal” at the AIP. Clusters move at different speeds and in different directions from their neighbors…With development through stages 6 and 7, the anterior medial border of each mass of heart-forming cells extends forward and mesiad forming the crescent of cardiogenic material which arcs rostral to the precordal plate. At this time a pattern of organization of the clusters within the crescent emerges, relating their position with their later differentiation’ (DeHaan, 1965). DeHaan also detected clumps of cells that started in rostral positions but ended up in a caudal position (DeHaan, 1963a).

Our data are similar to those of DeHaan’s, except that we would say that this pattern begins to form at stage 7 and a definitive pattern re-established by stage 8. We confirmed this trend by labeling four groups of cells (two on either side of the embryo) in the HFR with DiI and DiO, and followed them at intervals over 20 hours. The rationale for using multiple dyes was that in our experience it is misleading to inject a single embryo in multiple sites with the same tag. This process introduces error as it is difficult to monitor and identify each tagged group of cells over time. We observed that an overall rostrocaudal pattern prevailed but there was considerable overlap of the proliferating cell populations within that pattern. This can be seen in Fig. 5 where the general rostrocaudal relationship of the red and green labeled cells is maintained though a considerable overlap of the caudal cells (green) into the rostral cells (red) was observed.

In summary, our studies have established accurate boundaries for the HFR. Moreover, our data present evidence that the proposed putative cardiogenic markers Bmp2 and Nkx2.5 are not coincident with the HFR between stages 4 and 8. Therefore, even though these well characterized ‘early cardiac regulators’ may have a functional role in cardiac development (Erhman and Yutzey, 1999; Schultheiss et al., 1995; Schultheiss et al., 1997), they fall short of demarcating the entire HFR during early developmental stages. Furthermore, neither of these molecules has been shown to be sufficient for cardiac commitment or differentiation, making Nkx2.5 and Bmp2 erroneous candidates as HFR markers. Therefore, we believe that it is inaccurate to use these markers to identify the HFR. Finally, our highly accurate fate maps, including the stage 4 HFR fate map (reported for the first time), can now be used as a benchmark for studies involving cellular and molecular events during cardiac development.

Acknowledgments

The authors thank Dr John Lough for his careful review of the manuscript and his valuable suggestions. The scientists that reviewed this manuscript provided excellent suggestions that helped us address key issues in the manuscript. We thank them for their thoughtful and positive critique. We thank Dr Mary Barbe for her patient and professional help with the sectioning of the embryos. This work was supported by the National Institutes of Health (NIH) grant (HL52052) to J. L.

Footnotes

    • Accepted March 26, 2001.

References

View Abstract