Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Sendai, Japan
Correspondence to: Koji Tamura, Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama 6-3 Aoba-ku, Sendai 980-8578, Japan. E-mail: firstname.lastname@example.org
The chick embryo has long served as a model for amniote development because of its accessibility and ease of manipulation; in addition, the chick embryo shares many morphological features and underlying developmental processes and mechanisms with other vertebrates. Chick embryos were used in various classic developmental biology experiments including studies on gastrulation (Lemaire and Kessel, 1997), somitogenesis (Takahashi and Sato, 2008), and neural tube formation (Colas and Schoenwolf, 2001). Studies on limb development in the chick embryo have also yielded important contributions to our knowledge on organ morphogenesis (Mercader et al., 1999; Tickle, 2006; Suzuki et al., 2008; Zeller et al., 2009). The zone of polarizing activity (ZPA) and the apical ectodermal ridge (AER), which play crucial roles in vertebrate limb/fin development, were originally discovered in the chick embryo (Saunders, 1948; Saunders and Gasseling, 1968; Summerbell, 1979; Riddle et al., 1993). In addition, genetic studies using chick embryos have provided general insights into the molecular mechanisms involved in limb development; the identified genes include fgf10/fgf8 for limb development initiation (Ohuchi et al., 1997), Shh for digit identity (Yang et al., 1997; Tickle, 2006), Hox genes for limb skeletal patterning (Morgan et al., 1992; Yokouchi et al., 1995), and Tbx5/4 for fore- and hindlimb identities (Rodriguez-Esteban et al., 1999; Takeuchi et al., 1999).
The chick limb bud has also been used extensively as a model of limb development in birds, since it gives rise to many characteristic morphologies specific to bird limbs (Verden Berge, 1979), such as the shape and number of bones (Kamiyama et al., 2012; Seki et al., 2012), the intertarsal joint (Yamazaki et al., 2007; Namba et al., 2010), and specific traits of the skeletal muscle and tendons (Kardon, 1998; Coumailleau and Duprez, 2009; Valasek et al., 2011). Among these avian-specific limb characteristics, here we focus on digit formation in the bird forelimb (wing). The avian forelimb contains only three digits, which have been evolutionarily reduced from the common ancestral form containing five digits (Gauthier, 1986; Sereno, 1999; Welten et al., 2005; Larsson et al., 2010). The unique three-digit forelimb is considered a synapomorphy of the modern bird group.
Studies on the chick limb bud have uncovered many common mechanisms involved in digit morphogenesis, including the discovery of the ZPA as the developmental organizing center, and the characterization of Sonic hedgehog (Shh) as the molecular entity of this organizing center determining digit identity (Riddle et al., 1993; Tickle, 2006; Zeller et al., 2009). Despite the common mechanisms of digit formation, the three-digit bird forelimb, which is different from the five-digit tetrapod forelimb, indicates that bird-specific developmental processes are also involved in digit formation in the bird forelimb. Many studies investigating the mechanism(s) regulating the digit number in birds have been conducted. At the early stage of limb development, the ANZ and PNZ (anterior and posterior necrotic zones), apoptotic regions in the anterior and posterior margins of the limb bud, are specifically detected in birds (Zuzarte-Luis and Hurle, 2005; Fernandez-Teran et al., 2006). These bird limb bud-specific apoptotic regions may be involved in regulating digit numbers, because they are missing in talpid3 chicken mutants, which exhibit polydactyly (Hinchliffe and Ede, 1967).
However, there may be additional mechanisms involved in specifying the digit number in bird forelimbs and hindlimbs, which contain three/four digits. It is reported that cartilage condensations and their related gene expressions are detected outside the three-digit primordia in the chick forelimb bud, and disappear during later stages of limb development (Larsson and Wagner, 2002; Welten et al., 2005). These condensations and related gene expressions may thus represent vestigial digit primordia. Interestingly, the anterior lateral condensation is more obvious in the forelimb bud of the ostrich, a member of a primitive clade of birds (Feduccia and Nowicki, 2002). In addition, studies comparing limb development in the chicken and the alligator, an extant species most closely related to birds, have provided a paleontological perspective on digit loss and identity in modern birds (Burke and Feduccia, 1997; Larsson et al., 2010; de Bakker et al., 2013). We previously reported that the chick forelimb bud encounters an early frame shift, in which mesenchymal cells fated to become presumptive digit primordia undergo an anterior shift prior to digit specification (Tamura et al., 2011). Although the cause of the anterior shift remains unknown, it is thought to be an avian-specific event. Despite the many studies alluded to above, the developmental mechanisms giving rise to the three-digit forelimb in birds have not been completely clarified.
In this study, we first focused on the size and width of the digit-forming region in the early limb bud. Despite the different number of digits in the fore- and hind limb, the early chick limb buds have a similar semilunar shape. Several possible mechanisms could account for the difference in digit number in the chicken fore- and hind limbs. One possibility is that the anterior-posterior width of the digit-forming region is different in the fore- and hind limb buds, despite their similar shapes. A second possibility is that one or more developmental mechanisms lead to a reduced number of digits in the forelimb, despite the same width of the digit-forming region. To address this issue, we sought to identify which cell populations of the early limb bud contribute to digit formation in the late limb bud. For this purpose, we prepared fate maps of the early fore- and hindlimb buds, specifically focusing on the mesenchymal cell population at the distal margin along the anterior-posterior axis. Based on these fate maps, we found that the digit-forming region in the forelimb bud is narrower than that in the hindlimb bud, suggesting that the three digits in the forelimb are derived from the same relative positions from which digits 2, 3, and 4 in the hindlimb are derived. We also investigated possible developmental mechanisms that might generate the different cell fates in the fore- and hindlimb buds. Our results suggest that temporal differences in the onset of appearance of the ANZ as well as differences in the position of the anterior edge of the AER are potential factors regulating the number of digits in birds.
Fate Maps of the Early Stage Chick Fore- and Hindlimb Buds, and Contribution of the Anterior Region to Limb Structure
To examine the prospective fates of limb mesenchymal cells along the anterior-posterior (AP) axis in the early stage of chick limb development, we generated detailed and accurate fate maps of distal limb bud cells at an early stage (stage 20, Hamburger and Hamilton, 1951). We defined the Positional Index (PI) as the relative position along the AP axis of the limb bud, where PI 100.0 and PI 0.0 were assigned to the anterior and posterior edges of the stage-20 limb bud, respectively. A small group of mesenchymal cells was labeled by manual injection of DiI and DiO, the PI was calculated for their position along the AP axis, and their fates were traced by observing the position of the labeled cells 3 days later (at stage 30). For example, when the areas at PI 46.4, 30.9, and 18.9 were injected with dye (Fig. 1A), the labeled areas were detected in the digit region 3 days later (Fig. 1B). Dye injected into areas at PI 50.0, 35.2, and 21.1 in the hindlimb bud (Fig. 1C) was also found to be distributed in the digit region 3 days later (Fig. 1D). Repeated labeling analyses (more than 12 experiments for each area, typical examples of which are shown in Fig. 1E–1L) revealed the contribution of each PI point to the structure/region in the later limb bud (Fig. 2A–D). In these maps, we used some of the data that we published in our previous report (for the posterior-most digits; Tamura et al., 2011). The entire fate maps presented here give new insights into our knowledge about developmental differences between fore- and hindlimb buds. The three digits in the forelimb (in green, pink, and orange) were derived from the area spanning PI 49.1–17.7 (Fig. 2C). The four digits in the hindlimb (in gray, green, pink, and orange) were derived from the area spanning PI 63.8–18.9 (Fig. 2D), and the posterior three digits of the four digits (in green, pink, and orange) were derived from the area spanning PI 48.6-18.9, which is equivalent to the area giving rise to the three digits in the forelimb. These numbers for PI (Fig. 2C, D) represent average values of the injections at each site. These results indicate that the three digits in the forelimb and the posterior three digits in the hindlimb are generated from a similar region along the AP axis of the early limb bud.
Our fate maps also demonstrated that the more proximal regions (stylopod, zeugopod, and wrist/ankle in blue, red, and yellow, respectively) have different origins in the forelimb and hindlimb. These three regions in the forelimb were derived from the area spanning PI 78.0–60.3, whereas the same regions in the hindlimb were derived from the area spanning PI 100.0–74.0, indicating that the limb structures in the proximal regions were generated from different areas along the AP axis of the limb buds. Furthermore, the anterior-most region in the forelimb bud (PI 100.0–78.0) did not contribute to any limb structures, and cells in this region were left behind at the anterior basement of the forelimb (or possibly outside of the limb) (Fig. 1I, J). The wrist region just proximal to the digits (in yellow) was derived from the area at PI 60.3 in the forelimb, which was more posterior than the area (PI 74.0 in yellow) generating the analogous region in the hindlimb. The anterior-most digit in the hindlimb was derived from the area at PI 63.8 (in gray); however, in the forelimb there was little space for a digit between the wrist (in yellow) and the other digits (in green, pink, and orange).
In summary (Fig. 2E, F), our fate maps revealed that the three digits in the forelimb were derived from a smaller area in the early limb bud than the region specifying the four digits in the hindlimb, and that the proximal regions in the forelimb were derived from a more posterior area in the limb bud than those in the hindlimb. The differential contribution of the anterior limb bud (compare PI values over 50 in Fig. 2C, D) may result in the development of different numbers of digits in the fore- and hindlimbs. These results suggest that the anterior region in the forelimb bud may have special developmental properties that change cell fate and contribution of the anterior cells to the limb structure, resulting in the narrower space for digits in the forelimb.
Temporal Differences in the Onset of Apoptosis in the ANZ and AER of the Forelimb and Hindlimb
To elucidate the developmental mechanisms underlying the differential contributions of the anterior cells in the forelimb and hindlimb, we first focused on an apoptotic region, the ANZ, which develops at an earlier stage in the limb buds of birds than in those of other amniotes (Fernandez-Teran et al., 2006). We examined the onset, duration, and distribution of the ANZ in the fore- and hindlimb buds (Fig. 3). Apoptosis in the forelimb bud mesenchyme, detected by whole-mount TUNEL staining, was observed as early as stage 18 in an anterior-proximal region (asterisks in Fig. 3A, n=13/17), and mesenchymal apoptosis in the anterior region continued through later stages (Fig. 3B–D, n=5/6, 9/9, and 9/9). On the other hand, apoptosis in the mesenchyme of the hindlimb bud could not be detected until stage 20 (Fig. 3E, F, n=15/15 and n=6/6), and apoptosis in the ANZ had expanded by stage 22 (Fig. 3G, n=7/8). This mesenchymal apoptosis in the anterior region continued until at least stage 23 (Fig. 3H, n=8/8).
We also employed Nile Blue staining as an alternative method for measuring apoptosis. Nile Blue staining showed the differential onset of ANZ in the forelimb (at stage 20, asterisks in Fig. 3J, K, n=7/8 and 8/8, respectively) and the hindlimb (at stage 24, asterisks in Fig. 3O, P, n=8/8 and 13/13, respectively, and not shown), confirming the results for mesenchymal apoptosis by TUNEL analysis shown in Figure 3A–H. We newly found in Nile Blue staining that the onset of apoptosis in the AER was different in the forelimb and hindlimb. Somites 15 and 26 were used as indicators of the anterior border of the limb bud at stage 18. Apoptosis in the forelimb AER was detected in the anterior side at stage 18 (white arrowheads in Fig. 3I, n=6/6), continued until stage 25 (Fig. 3J–L, n=8/8, 8/8, 13/13), and had disappeared by stage 26 (data not shown). In contrast, in the hindlimb, apoptosis in the AER was faintly visible at stage 20 and clearly detected at stage 20+ (Fig. 3M, N, O, n=6/6, 8/8, and 8/8, respectively). This apoptosis continued until at least stage 25 (Fig. 3P, n=10/13), and disappeared by stage 26 (data not shown). Although it is possible that ectodermal apoptosis may start at an earlier stage, our results indicate that AER apoptosis in the forelimb bud begins earlier than in the hindlimb.
The ANZ and apoptosis in the anterior region of the AER have been extensively studied (Milaire and Rooze, 1983; Zuzarte-Luis and Hurle, 2002; Fernandez-Teran et al., 2006). Our new findings suggest that apoptosis in the forelimb bud starts earlier than previously thought. The onset of the ANZ prior to stage 22 in the hindlimb is consistent with data reported by Fernandez-Teran et al. (2006), although the authors did not focus on the onset difference in the forelimb and the hindlimb. Apoptosis in the anterior limb mesenchyme and the AER may be related to each other, since the AER and limb mesenchyme are maintained by a reciprocal feedback signaling loop (Rowe et al., 1982; Ohuchi et al., 1997). Thus, the earlier onset of apoptosis in the anterior mesenchyme and the AER of the forelimb may be related to the difference in anterior cell fates in the forelimb and hindlimb.
Disappearance of the Anterior AER in the Forelimb Bud at Stage 20
Because the anterior mesenchyme and the AER undergo apoptosis earlier in the forelimb than in the hindlimb bud, it is possible that these areas are not maintained functionally or structurally in the forelimb. We examined the AP width of the AER by observing Fgf8 expression that was used as a marker for the AER and its precursor cells (Vogel et al., 1996), focusing on the anterior portion, starting at stage 16. Fgf8-expressing cells at stage 16 are still AER precursor cells, and the AER structure emerges at subsequent stages (Crossley et al., 1996). Somites 15 and 26 were used as indicators of the anterior borders of the fore- and hindlimb buds, respectively (Fig. 4). Weak Fgf8 expression in the forelimb bud was initially detected at 28 somite stage (stage 16; Fig. 4A, n=9/9), had increased and extended to the anterior border of the limb bud by 31 somite stage (stage 17; data not shown, n=8/8), and continued to extend to the anterior border until stage 19 (Fig. 4G, I, n=5/5 and n=9/10). The hindlimb bud showed a similar dynamic pattern of Fgf8 expression, but the onset was delayed to 29 somite stage (stage 17; Fig. 4D, n=8/10, compare to stage 16; Fig. 4B, n=0/9), and the anterior edge of the AER, as indicated by Fgf8 expression, had extended to the anterior border of the limb bud by 32 somite stage (stage 17; Fig. 4F, n=8/8). The agreement between the anterior edge of Fgf8 expression and the anterior border of the limb bud continued until stage 20 (Fig. 4H, J, L, n=5/5, 7/7, and 14/14, respectively).
Clear developmental differences between the fore- and hindlimb buds were detectable by stage 20. In the forelimb bud, anterior Fgf8 expression in the AER had disappeared, and there was a gap between the anterior edge of the AER and the anterior border of the limb bud (Fig. 4K, n=14/14). In contrast, in the hindlimb bud, the anterior edge of the AER still corresponded with the anterior edge of the limb bud at stage 20 (Fig. 4L, n=14/14). At stage 21, anterior Fgf8 expression in the hindlimb had slightly moved posteriorly (Fig. 4N, n=5/5) and no longer extended to the anterior border of the hindlimb bud at stage 22 (Fig. 4P, n=7/7). However, the distance between the anterior boarder of limb bud and the anterior edge of AER is larger in the forelimb bud than in the hindlimb bud (Fig. 4M, O, n=5/5 and n=7/7, respectively; compare to Fig. 4N, P). It is likely that disappearance of the anterior AER (and early stage of apoptosis in the AER) is not due to the difference in developmental timing between the fore- and hindlimb buds, because the onset and extension of Fgf8 expression are almost the same (only a 1-somite-stage gap) (see Fig. 4A and D), and development (growth) of the chick hindlimb is faster than that of the forelimb (Hamburger and Hamilton, 1951). Our findings at a molecular level are consistent with structural observations in previous studies, which showed that the anterior region of the AER has low thickness at stage 20 only in the forelimb bud (Todt and Fallon, 1984, 1986).
These developmental differences in the hind- and forelimb buds corresponded with differences in their fate maps. In the forelimb bud at stage 20, the anterior edge of the AER corresponded to PI 83.3, and the AER spanned the AP range of PI 83.3–PI 7.1 (Fig. 4Q). Fate map analysis (see Fig. 2C) showed that the area just posterior to the anterior edge of the AER contributed to development of the stylopod structure, while the region anterior to the edge of the AER did not contribute to the development of any limb structure. The AER in the hindlimb bud spans the range of PI 100.0 to PI 10.2 (Fig. 4R). According to the fate map of the stage 20 hindlimb bud, the anterior-most region of the limb bud just posterior to the anterior edge of the AER contributed to stylopod development (Fig. 2D). These results indicate that the stylopod is derived from mesenchyme posterior to the anterior edge of the AER at stage 20; however, the position of the edge itself is different in the fore- and hindlimb buds.
Dissociation of the Anterior AER Cells From the AER Region During Early Forelimb Development
We have shown that the anterior AER undergoes apoptosis in the forelimb earlier than in the hindlimb (Fig. 3), and that the AER is narrower in the forelimb (Fig. 4). These results suggest that cells in the anterior AER of the forelimb bud may not remain associated with the AER. Alternatively, the anterior region of the forelimb AER may move posteriorly and remain associated with the AER. To differentiate between these two possibilities, we injected DiI into the anterior-most region of the AER at stage 18 and traced the labeled cells through later stages (Fig. 5). In the forelimb bud, anterior AER cells labeled at stage 18 were not detected in the AER at stage 20, and many labeled cells were distributed in the ventral non-AER ectoderm (Fig. 5A–D, B′, D′). In contrast, anterior AER cells in the hindlimb bud at stage 18 remained associated with the AER up to stage 24 (Fig. 5N–Q, Q′, O′), and few cells were distributed in the non-AER ectoderm (Fig. 5O′). During stages 18–20, the anterior AER cells in the forelimb appeared to dissociate from the AER, resulting in a forelimb-specific regression of the anterior AER.
Further examination showed that the regression also continued after stage 20 (Fig. 5E–M, H′, L′,). Because the anterior edge of the AER in the forelimb bud was evaluated at PI 83.3 at stage 20 (see Fig. 4Q), we labeled cells in the AER posterior to this point at stage 20 and observed these cells at stage 25. Cells injected at PI 81.6, which were located at the anterior edge of the AER at stage 20, were no longer inside the AER by stage 25 (Fig. 5E–H, H′). Cells at PI 73.4, which were initially located at a sub-anterior region of the AER, were repositioned to the anterior edge of the AER at stage 25 (Fig. 5I–L, L′). Furthermore, when cells close to the anterior edge of the AER (at PI 76.8–81.7) were labeled at stage 20, they were no longer located in the AER at stage 25 (shown by red diamonds in Fig. 5M), whereas the sub-anterior region (at PI 67.8–73.4) remained within the AER (shown by blue diamonds in Fig. 5M). In contrast, in the hindlimb bud, the anterior edge of the AER at stage 20 corresponded to the anterior edge of the AER at stage 25 (Fig. 5R–U, U′), and was found to be dissociated from the AER by stage 28 (data not shown).
The schematic representation in Figure 5V shows the cell fate of the anterior edge of the AER, indicating a clear difference between the forelimb and the hindlimb; while the anterior AER in the forelimb bud continued to regress from its initial position early in development, cells at the anterior edge of the AER in the hindlimb bud remained there until at least stage 25 (showed red, light blue, and blue points).
Regression of the Anterior AER Region Results in a Reduced Digit-Forming Region in the Forelimb
During autopod development, the autopod-forming region of the mesenchyme is surrounded by the AER, and the digits form under the region spanned by the AER (Selever et al., 2004; Robert, 2007). Artificial elongation of the AER along the AP axis expands the autopod region and increases the number of digits (Kawakami et al., 2004), suggesting that the AER width is important for defining the number of digits. Our fate map (see Fig. 2) indicated that there was little space for a digit at the anterior side of the forelimb bud, and we also showed that the anterior AER regressed from its original position early in development (see Fig. 5). We next evaluated the positional relationship between the anterior edge of the AER and the anterior-most digit. At stage 20, the presumptive region for digit 1 in the forelimb was located at PI 49.1 (Fig. 2), which corresponds to digit 2 in the hindlimb (PI 48.6). We labeled mesenchymal cells in the regions close to PI 50 and PI 100 at stage 20 (Fig. 6A, B), traced their fates at several time points up to stage 27, and examined their positions relative to the anterior edge of the AER.
We found that in the forelimb bud, the distance between the anterior edge of the AER and the labeled cells at around PI 50 progressively decreased as limb development proceeded (Fig. 6C–F), and was clearly shorter than that in the hindlimb bud (Fig. 6G–J). We evaluated relative positions of the labeled cells at stage 26 and compared them between the fore- and hindlimb buds (Fig. 6K). At injection (stage 20, Fig. 6A, B), average values of PI for the injected points were PI 44.0 in the forelimb (n=9) and PI 50.3 in the hindlimb (n=14). At stage 26, we measured the distance between the anterior edge of the AER and the position where the labeled cells were located. We also measured the length of the whole AER and calculated the proportion of the labeled point relative to the whole AER by percentage. The average values of the proportion were 37.9% for the forelimb and 46.5% for the hindlimb, indicating that the distance between the anterior edge of the AER and the labeled point at around PI 50 is clearly smaller in the forelimb than in the hindlimb. This dynamic change in relative position of the mesenchyme and the anterior edge of the AER in the forelimb, caused by regression of the AER, may result in the reduced space for digit formation and the reduced number of digits in the forelimb.
Altered Cell Fate in the Anterior Region of the Chick Forelimb Bud
The fate maps presented here provide new insights into the cellular mechanism(s) contributing to digit formation in birds. First, we found that the area in the early stage-20 limb bud that gives rise to the three digits in the forelimb was smaller than the corresponding area that gives rise to four digits in the hindlimb, whereas the area that gives rise to each digit is similar in the fore- and hindlimb. The areas contributing to the three forelimb digits or the posterior three hindlimb digits were found to be equivalent, suggesting that the fates for the posterior half of both limb buds were similar at this stage. Shh deletion experiments using mice showed that autopod specification occurs in a short time window after Shh starts to be expressed (Zhu et al., 2008). Furthermore, SHH protein inhibition experiments using cyclopamine in chick embryos showed that digits 1 and 2 in the forelimb and digits 1, 2, and 3 in the hindlimb are formed after SHH inhibition at stage 20 (Scherz et al., 2007). These results suggest that difference in the region for digits between fore- and hindlimb buds starts to be specified at stage 20, when we labeled cells for cell tracing.
In contrast, a clear difference was seen in the fate of the anterior regions of the fore- and hindlimb buds. The anterior area of the stage-20 limb bud contributes to limb proximal structures including the stylopod, zeugopod, and wrist/ankle. In the hindlimb, the anterior-most area near the limb bud border participates in stylopod formation; however, this area in the forelimb was found to have minimal contribution to the formation of any limb structures. Instead, the proximal structures in the forelimb were derived from a more posterior area at stage 20. Accordingly, the width of the prospective limb-forming region in the anterior half of the stage-20 limb bud was smaller in the forelimb than in the hindlimb. As mesenchymal cells build the limb structure in a proximal-to-distal direction, the smaller width in the forelimb may result in fewer mesenchymal cells for the digits.
Our findings from fate mapping analysis suggest that the anterior region of the chick forelimb undergoes an alteration in cell fate that results in the formation of three rather than four digits. Fate mapping, in general, has long contributed to our understanding of developmental processes (Clarke and Tickle, 1999). Vargesson et al. (1997) showed a detailed fate map in the early chick forelimb bud, for example, and our results for the forelimb bud presented here basically correspond to their map, although there was no comparison of cell fate between the forelimb and hindlimb. With respect to digit formation in birds, our fate maps suggest that cell fate alterations result in developmental changes that can generate unique specialized morphologies. However, it is still possible that our findings may just reflect structural differences between the forelimb and the hindlimb, independent of digit number. Similar analyses of the cell fates in five-digit amniotes will provide additional information that can be applied to digit formation in birds. Our preliminary studies suggest that there are minimal differences in forelimb and hindlimb development of mouse and gecko until the digit number is determined (Noro et al., 2009; Kamiyama et al., 2012 and unpublished observation). The developmental differences observed in the chick limb buds could be specific for the bird forelimb, and may also extend to the ancestral group of theropod dinosaurs with three-digit forelimbs.
Putative Mechanisms Contributing to the Altered Cell Fate in the Anterior Region of the Forelimb Bud
The ANZ has been observed in the developing limb bud of several species including the mouse and chicken (Milaire and Rooze, 1983; Zuzarte-Luis and Hurle, 2002, 2005; Fernandez-Teran et al., 2006). In the mouse limb bud, the ANZ appears only at the later stages of limb development in a small part of the wrist/ankle region, suggesting that it plays a role in shaping the joint morphology. In contrast, the ANZ in chick and quail limb buds develops at much earlier stages (Zuzarte-Luis and Hurle, 2005; Fernandez-Teran et al., 2006 and our unpublished observation). Here we found that the onset of ANZ appearance was different in the fore- and hindlimb buds, and was detected in the forelimb at stage 20, corresponding to altered cell fates in the forelimb bud. Moreover, the ANZ in the forelimb at this stage corresponded to an anterior area that showed different contributions to the proximal limb structures in the forelimb versus the hindlimb (see Figs. 2 and 3).
A striking finding was the absence of the anterior one-fifth of the AER at stage 20 in the forelimb bud. The area contributing to the most proximal element, the stylopod, corresponds to the anterior edge of the AER (Fig. 4), indicating that the area outside the AER does not participate in limb structure (fate map in Fig. 2), and that there is little space for a digit in the anterior half of the forelimb bud. Furthermore, continuous regression of the anterior AER in the forelimb bud shortens the distance between the anterior margin of the AER and the region forming the anterior-most digit 1 (Fig. 7). We also found that apoptosis of the AER occurs earlier in the forelimb than in the hindlimb, although it is unclear how apoptosis in the AER impacts limb development and whether the early cell death of the anterior AER is related to the regression of the AER in the forelimb.
Due to technical limitations, we have not yet succeeded in changing the cell fate of the anterior region in the chick forelimb (e.g., by inhibiting cell death); however, mesenchymal cell death and the width of the AER are known to be critical in determining digit number. In polydactyl mutant limbs, absence of the ANZ is sometimes observed (Hinchliffe and Ede, 1967), and mechanical removal of the anterior AER results in loss of the anterior-most digit (Rowe and Fallon, 1981). Furthermore, artificial extension of the AER results in the formation of an additional digit (Kawakami et al., 2004). Thus, two mechanisms, involving cell death and AER length, correlate with the altered cell fate in the chicken forelimb bud, and may contribute to determining the bird-specific digit number in the forelimb.
Although these two mechanisms may independently regulate digit number, it is likely that they work together. Removal of the AER gives rise to extensive cell death (Rowe et al., 1982). Mesenchymal cells maintain the AER structure and function by secreting molecules, including FGF10 (Ohuchi et al., 1997), while FGF proteins such as FGF8 and FGF4 secreted from the AER promote the function of the underlying mesenchymal cells (Fallon et al., 1994; Crossley et al., 1996). These findings suggest that interactions in the AER and limb mesenchyme form a reciprocal feedback loop. Reduced interactions at the anterior region of the early forelimb bud may cause morphological alterations resulting in a reduction in digit number.
Peculiarity of Wing Development in Birds
As mentioned earlier, we found that the three digits in the forelimb are derived from an area equivalent to that specifying the posterior three of four digits in the hindlimb, suggesting that the three digits in the forelimb may be homologous to digits 2, 3, and 4 in the hindlimb. However, zeugopod specification occurs during stage 20, whereas digit identity is not determined until later (Scherz et al., 2007). Thus, the cells at this stage that were labeled are not digit progenitor cells, although the resulting progeny that emerge at the next stage (21/22) are digit progenitor cells. We previously demonstrated that distal mesenchymal cells at stage 20 undergo an anterior shift by stage 21/22 (referred to as the early frame shift; Tamura et al., 2011), resulting in progeny cells (digit progenitors) that are specified as digits 1, 2, and 3. Thus, the forelimb bud generates digits 1, 2, and 3 from the stage-20 limb bud area equivalent to the area that gives rise to the future digits 2, 3, and 4 in the hindlimb (the naming of the bird wing digits is still under discussion) (Tamura et al., 2011; Young et al., 2011; de Bakker et al., 2013).
It is currently unclear whether the early frame shift is related to the altered cell fate of the anterior region of the forelimb bud shown here. However, the anterior digit 1 is known to be formed independently of the morphogenesis of the posterior digits (Litingtung et al., 2002; Ros et al., 2003; Ahn and Joyner, 2004; Harfe et al., 2004; Amano and Tamura, 2005). Since the region investigated here is more anterior than PI 50, the region specifying digit 1, unique developmental events may occur in the anterior region independently of those occurring in the posterior limb bud. Taken together, there may be several wing-specific developmental mechanisms, including the early frame shift in the posterior region of the limb bud and the alteration in cell fate in the anterior region. Elucidation of these mechanisms will increase our understanding of wing development and of the evolutionary processes linking the ancestral group of theropod dinosaurs to birds.
DiI/DiO Administration and Method for Cell Fate Tracing
White Leghorn chicken eggs were incubated at 38°C and staged as previously described (Hamburger and Hamilton, 1951). Limb bud cells were labeled with the lipophilic dyes DiI (1,1-dioctadecyl-3,3,3′,3′-tetramethyl indocarbocyanine perchlorate; Molecular Probes, Eugene, OR) and DiO (3,3′-dioctadecyloxacarbocyanine, perchlorate; Molecular Probes), prepared as previously described (Li and Muneoka, 1999). A small amount of dye was injected manually with a glass micro-capillary tube, rubber tube, and mouthpiece, under a microscope. To determine the position and area injected with the dye, labeled limb buds were observed and photographed under a fluorescence microscope immediately after dye administration. The site of injection was expressed as a Positional Index (PI) according to its location between the anterior and posterior edges of the limb bud, which were assigned values of 100 and 0, respectively (see Fig. 1A as an example). For this evaluation, bright-field and fluorescent images were merged using imaging software (Adobe Photoshop CS4). Three days after injection, the samples were analyzed, and the region and/or structure containing the labeled cell population was identified. In some experiments, the AER of the labeled sample was examined by in situ hybridization of Fgf8, and a fluorescent image before Fgf8 staining and the same angle bright-field image after Fgf8 in situ hybridization were merged.
In Situ Hybridization, TUNEL and Nile Blue Staining
Chick embryos were processed for whole-mount in situ hybridization as described previously (Yonei et al., 1995), using antisense RNA probes for chick Fgf8, a reliable marker for the AER (Ohuchi et al., 1997). To detect apoptosis in the limb bud, whole-mount TUNEL was performed as described previously (Noro et al., 2011). For Nile Blue staining, another method of detecting apoptosis, a 1:10,000 dilution of Nile Blue in PBS was injected into a cavity of the amnioserosa membrane, and the embryos were incubated for 30 min at 38°C.
This work was supported by MEXT and JSPS KAKENHI Grant Number 22124005 to H.Y., JSPS KAKENHI Grant Number 25870058 to H.Y., the Kurata Memorial Hitachi Science and Technology Foundation to H.Y., the Asahi Glass Foundation to H.Y., and the “Funding Program for the Next Generation of World-Leading Researchers” from the Cabinet Office, the Government of Japan Grant LS007 to K.T. N.N. was supported by JSPS Research Fellowships for Young Scientists, Japan, Grant 24 · 6235.