The fibula, a zeugopod bone in the hindlimb, exhibits various morphologies in tetrapod species. The fibula in some species has a similar length with the other zeugopod element, the tibia, while other species have obvious differences in the sizes of the two elements. In the avian hindlimb, for example, the fibula is extremely short, thin, and truncated. Basic morphology of the fibula is established during development, and cartilage primordium of the bone emerges in a certain region defined by a distinct combination of expression of Hox genes (Hox code). In order to elucidate how the different morphologies are produced from a region that is defined as the fixed Hox code, we examined spatial and temporal patterns of Hoxd11/Hoxd12 expression in the developing limb bud, which defines the region from which the fibula emerges, in comparison with the sites of precartilaginous mesenchymal condensations representing regions for cartilage formation among chick, mouse, and gecko embryos. We found that in the chick hindlimb, expression of Hoxd11/Hoxd12 decreased and disappeared from the presumptive zeugopod region before cartilage formation. This heterochronically early decline of expression of Hox genes is strongly correlated with the peculiar trait of the fibula in the avian hindlimb, since in the other species examined, expression of those genes continued after the onset of cartilage formation. This is morphological phenotype-related because the early disappearance was not seen in the chick forelimb. Our results suggest that temporal change of the Hox code governs diversification in morphology of homologous structures among related species.
The zeugopod, forearm/lower leg, is the middle segment of tetrapod limbs that contains two long bones, the radius and ulna in the forelimb and the tibia and fibula in the hindlimb. The basic pattern of the skeleton, presence of two long bone elements in the zeugopod, is highly conserved in tetrapods (Tamura et al. 2008), whereas the morphology of each element varies among species. Some tetrapod species have two zeugopod elements that are similar in morphology, while others have two elements that exhibit greatly different traits in some dimensions (length, width, and shape). For example, in the chick embryo, the tibia is long and very large, whereas the fibula is short, thin and truncated, eventually becoming vestigial and fusing with the tibia in the adult chicken (Müller & Streicher 1989).
Basic morphology of the bone reflects the shape of cartilage element that emerges in a certain region during the developmental process. The process of fibula cartilage formation serves as a good model for morphogenesis during development because differentiation of an element in a certain region is the simplest model of morphogenesis. Moreover, how nonequivalence of bone morphology is established among species is a good example to illustrate diversification of morphology. In chick embryo, despite the difference in their final morphology, precartilaginous mesenchymal condensations for embryonic rudiments of the fibula and tibia initially show similar lengths (Archer et al. 1983), and the length is secondarily modified in the later process of development. In this process, the distal epiphysis of the fibula cartilage becomes detached from the diaphysis of the cartilage and integrated into the distal part of the tibia (Archer et al. 1983). This rearrangement of the fibula epiphysis allows the tibia to elongate distally, and the proximal part of the fibula remains short and left behind at the knee. The truncated shaft of the fibula is eventually fused with the proximal part of the tibia, making an appearance of a long single bone element in the zeugopod. The developmental process, which seems peculiar to the avian hindlimb, involves many cellular events, including cell proliferation, cell movement, matrix secretion, and cell differentiation (Hampe 1960; Archer et al. 1983; Goff & Tabin 1997), although the mechanism upstream of these cellular events remains unresolved.
One possible candidate is a mechanism mediated by Hox genes. Many Hox genes are known to be expressed in the developing limb bud in spatially and temporally regulated manners. In the earlier stages of limb development, each area corresponding to each bone element (called “compartment” in this paper) can be definable by an overlapping but distinct pattern of expression of Hox genes (Yokouchi et al. 1991; Haack & Gruss 1993; Small & Potter 1993; Nelson et al. 1996). The zeugopod is classified as a compartment that exclusively expresses Hoxa11 (Haack & Gruss 1993; Small & Potter 1993). The zeugopod compartment can further be divided into two parts corresponding to two bone elements as being the Hoxd11/Hoxd12-negative domain (for the anterior element, the radius and tibia) and Hoxd11/Hoxd12-positive domain (for the posterior element, the ulna and fibula) (Yokouchi et al. 1991; Nelson et al. 1996; Suzuki & Kuroiwa 2002). Gain- and loss-of-function analyses (Davis & Capecchi 1994, 1996; Davis et al. 1995; Goff & Tabin 1997; Wellik & Capecchi 2003; Boulet & Capecchi 2004) suggested that in the developing zeugopod, Hox genes are involved in growth and morphogenesis of the compartment and skeletal morphology that emerges from the compartment. Thus, the posterior compartment in the zeugopod possesses an identity of Hoxa11/Hoxd11/Hoxd12-positive domain. Considering that the basic expression pattern of these Hox genes is common among tetrapod limb buds, the posterior zeugopod can generally be identified and defined as the same “Hox code” (Tabin 1992; Graham 1994).
Although the fibula is derived from a compartment defined by a specific Hox code, the final morphology of the element exhibits difference among species. Simple application of the Hox code cannot explain the difference. One possible explanation may be that there is limb type-specific expression of other genes. The limb type-specific expression of genes such as Hoxc9, Hoxc11 (Nelson et al. 1996; Papenbrock et al. 2000), Pitx1, and Tbx4 (Logan & Tabin 1999; Rodriguez-Esteban et al. 1999; Takeuchi et al. 1999) may regulate the difference in morphology between the ulna and fibula, providing the peculiar traits of the fibula in the chicken. Even this interpretation, however, cannot account for the interspecies difference in fibula morphology, and, moreover, an avian-specific gene, expression of which is only seen in the avian hindlimb zeugopod, is not thought to exist. In this study, we focused on the timing of decline of Hoxd11 and Hoxd12 expression in the developing hindlimb zeugopod and compared it with emergence of precartilaginous mesenchymal condensation among mouse, chick and gecko embryos. We found a good correlation between duration of Hox expression and morphological difference in the fibula, suggesting that the temporal shift of Hox expression plays a role in diversification of traits in a compartment that is defined as a “Hox code” domain.
Materials and methods
Collection of eggs and embryos
Fertilized chicken (Gallus gallus) eggs were incubated at 38°C, and the embryos were staged according to Hamburger & Hamilton (1951). Mouse (Mus musculus) embryos from wild-type mice (Slc:ddY mice) were collected at embryonic day (E) 10.5 to E12.0 after natural overnight mating. For gecko embryos, we used Madagascar ground geckos (Paroedura pictus) that were hatched and kept in our laboratory (Noro et al. 2009), and the eggs used for staining were incubated at 28°C. The embryos were staged according to Noro et al. (2009).
Cloning of Hoxd11 and Hoxd12 from the chick, and cloning of Aggrecan1 and Hoxd12 from Paroedura pictus
We isolated fragments of chicken Hoxd11 and Hoxd12 from chick embryos at stage 22. Polymerase chain reaction (PCR) fragments of P. pictus Aggrecan1 and Hoxd12 were cloned from cDNA pools prepared from P. pictus embryos at 10 dpo (days postoviposition). We used the following primers for PCR fragment isolation: 5′-GCGAAAAGGAAGACATGACC-3′ (forward for chicken Hoxd11), 5′-TCCTCCTCAAAACAAGGGATT-3′ (reverse for chicken Hoxd11), 5′-TTTCCGCTTATTCCCTGTTG-3′ (forward for chicken Hoxd12), 5′-ATTTGGTAACCGCTTTGTGC-3′ (reverse for chicken Hoxd12), 5′-GATGCCACCYTGGAAATCAAG-3′ (forward for P. pictus Aggrecan1), 5′-CTCTTCYACTGAGCTCACCTC-3′ (reverse for P. pictus Aggrecan1), 5′-GTGTGAKCGCAGTCTCTAC-3′ (forward for P. pictus Hoxd12), 5′-GTTCTGRAACCARATTTTG-3′ (reverse for P. pictus Hoxd12).
For whole-mount in situ hybridization, chick, mouse and gecko embryos were processed as described previously (Yonei et al. 1995; Yonei-Tamura et al. 1999; Noro et al. 2009) using antisense RNA probes for chicken Aggrecan, Hoxd11 and Hoxd12, and mouse Col2a1, Sox9, Hoxd11 and Hoxd12, and gecko Aggrecan1 and Hoxd12. Probes for mouse Col2a1, Sox9 and Hoxd12 were kindly provided by Y. Kawakami, and the probe for mouse Hoxd11 was kindly provided by D. Duboule. In situ hybridization of frozen sections was performed essentially by the same method as that described by Yoshida et al. (1996). The data on the same stage of forelimb buds and hindlimb buds with different probes were obtained as serial sections from the same embryo with the same staining condition at the same time. This procedure allowed us to compare the signals more precisely and reliably. Moreover, digital photographs in each figure were taken under the same conditions. For analysis of the skeletal pattern, embryos were fixed in 10% formalin, stained with Alcian blue and Alizarin red, and stored in glycerol.
Morphology of zeugopod elements in mouse, gecko and chick
We first observed the morphology of the zeugopod elements in mouse, gecko, and chick. In the mouse (a mammal), two bone elements in the forelimb zeugopod are similar in width (Fig. 1A,A′), while two elements in the hindlimb zeugopod have similar lengths but different widths (Fig. 1B,B′). Zeugopod bones of the gecko (a reptile) have similar lengths and widths in both forelimbs and hindlimbs (Fig. 1C,C′,D,D′). On the other hand, the two bone elements in the chick hindlimb zeugopod exhibit extreme difference in size (Fig. 1F,F′), while the two in the chick forelimb are not so different (Fig. 1E,E′).
In this study, we use the gecko as a model of reptiles in addition to the mouse and chicken. Mammals and birds have mainly been used as representatives of amniote animals in the study of experimental embryology. In fact, they are markedly different and distant from each other, and it is therefore valuable to have a species of reptiles for this study as an additional amniote model. Furthermore, this is the first report that shows the detailed expression pattern of Hoxd genes in the reptile limb bud.
Expression of Hoxd11 and Hoxd12 in the developing chick limb bud examined by whole-mount observation
Details of the expression pattern of Hoxd11 and Hoxd12 in the developing chick limb bud have been repeatedly reported (Izpisua-Belmonte et al. 1991; Mackem & Mahon 1991; Nohno et al. 1991; Yokouchi et al. 1991; Mackem et al. 1993; Nelson et al. 1996). Nelson et al. (1996) stated that “in the lower leg the expression of all of these genes has faded significantly by stage 25, such that only low levels of expression are detected along the posterior margin of the lower leg,” suggesting that difference in Hox gene expression between the forelimb and the hindlimb might be consistent with the morphological differences that exist between them. Similar suggestions were previously made by Mackem and colleagues (Mackem & Mahon 1991; Mackem et al. 1993). However, a comparative analysis of hindlimb expression pattern of these genes remains to be done among different species that have different morphologies in the zeugopod. Thus, we decided to examine the expression pattern of Hoxd11 and Hoxd12 in the hindlimb zeugopod in comparison between chick, mouse and gecko embryos. We also compared expression pattern of Hoxd11 and Hoxd12 to the sites of precartilaginous mesenchymal condensations for cartilage formation.
At chick stage 22–23, a clear signal for expression of both Hoxd11 (Fig. 2B,E,H,K) and Hoxd12 (Fig. 2C,F,I,L) was detected at the posterior portion of the limb bud. The intensity of the signal at the posterior-proximal domain of the hindlimb bud at stage 23 (brackets in Fig. 2K,L), however, was less than that at the distal-most domain and at the comparable, posterior-proximal domain of the forelimb bud (compare with Fig. 2H,I). In this domain of the forelimb bud, clear expression of Hoxd11 and Hoxd12 continued to stage 25 (Fig. 2N,O) and stage 27 (Fig. 2T,U). In the hindlimb bud, on the other hand, expression of both Hoxd11 and Hoxd12 in the presumptive posterior zeugopod was very weak at stage 25 (brackets in Fig. 2Q,R). At stage 27, almost no expression could be observed in the proximal region in the hindlimb bud, whereas distal expression was still discernible (Fig. 2W,X). The signal in the distal hindlimb bud as well as in the forelimb bud largely decreased by later stages 29–30 (not shown).
To compare these expression domains of Hox genes with the sites of cartilage formation, we selected some early and reliable markers for cartilage formation, Aggrecan (encoding a primary structural protein of cartilage), Col2a1 (collagen type II alpha1) and Sox9 (SRY-related high mobility group-box gene 9), all of which are known to be expressed in precartilaginous mesenchymal condensation in the developing limb bud (Chimal-Monroy et al. 2003; Kawakami et al. 2005; Sato et al. 2007). In the chick zeugopod, the first visible stage of precartilaginous mesenchymal condensation was stage 25 by whole-mount observation of Aggrecan gene expression (Fig. 2G,J,M,P and not shown). In Aggrecan expression domains that represent the radius and ulna, the posterior domain is overlapped with Hoxd11 and Hoxd12 expression domain in the forelimb bud at stages 25 and 27 (arrowheads in Fig. 2M,S, compare with 2N,O,T,U), suggesting that the ulna cartilage appears in the Hoxd11/Hoxd12-positive compartment as described in previous reports (Yokouchi et al. 1991; Nelson et al. 1996). In the hindlimb at stage 25, Aggrecan expression was also clearly detected as a pair of domains with similar lengths, representing the tibia and fibula (Fig. 2P), and the similar lengths of expression domains continued to stage 27 (Fig. 2V). The posterior domain of Aggrecan expression for the fibula was seen outside the Hoxd11 and Hoxd12 expression domain (compare Fig. 2P with 2Q,R and Fig. 2V with 2W,X).
Expression of Hoxd11 and Hoxd12 in mouse and gecko limb buds examined by whole-mount observation
We next examined the relationship between Hox expression and cartilage formation in other amniotes that have similar lengths of the tibia and fibula. We used two species, mouse and gecko, as representative species (see Fig. 1).
In the mouse limb bud, cartilage rudiment or precartilaginous mesenchymal condensation for the ulna and fibula (shown by arrowheads in Fig. 3) was visible from embryonic day 11 (E11.0) by Col2a1 (Fig. 3A,E) and Sox9 expression (Fig. 3B,F) (Kawakami et al. 2005). Hoxd11 and Hoxd12 were expressed in the proximal-posterior region, corresponding to the posterior zeugopod, from E11.0 through E11.5 to E12.0 (brackets in Fig. 3C,D,G,H). The expression in the posterior zeugopod at E12.0 was weaker than that in the autopod. We could not observe a clear difference in expression of these genes between the forelimb and hindlimb. Hoxd11 and Hoxd12 appear to continue being expressed during cartilage formation of the fibula and the expression domains seem to be overlapped with the site of fibula formation (compare Fig. 3G,H with 3E,F).
Hoxd12 expression was also examined in the developing limb bud of the gecko, Paroedura pictus (Fig. 4). Hoxd12 was expressed in the posterior and distal limb bud with similar patterns in the forelimb and hindlimb buds at 12 dpo (Fig. 4B,D). Expression of Hoxd12 continued until stages at 14 dpo and 16 dpo, and the posterior-proximal domain of expression also remained discernible (Fig. 4F,H,J,L). Aggrecan1 expression could be detected from 12 dpo, and posterior domain of Aggrecan1 expression, representing the ulna and fibula, was included in the domain of Hoxd12 expression (compare Fig. 4A,B,G,H). However, at 16 dpo, the posterior domain of Aggrecan1 expression was outside the domain of Hoxd12 expression (compare Fig. 4I,K with 4J,L).
These results suggest that in mouse and gecko embryos, the domain of Hox expression in the hindlimb bud includes precartilaginous mesenchymal condensation for the fibula and continues during early stages of cartilage formation with similar patterns in the forelimb and hindlimb.
Expression of Hoxd11 and Hoxd12 in the limb buds examined by observation of sections
Whole-mount in situ observations of gene expression, which are the standard method for assessing a gene expression pattern in the vertebrate embryo, provided some important insights into Hox expression in the presumptive zeugopod region. (i) In mouse and gecko embryos, the basic expression pattern of Hoxd11 and Hoxd12 genes in the presumptive zeugopod region is similar in the forelimb and hindlimb. (ii) In mouse and gecko limb buds, precartilaginous mesenchymal condensation for the ulna and fibula begins prior to decline of the expression of Hoxd11 and Hoxd12 in the posterior zeugopod region. (iii) In the chick embryo, expression pattern of Hoxd11 and Hoxd12 in the presumptive zeugopod region is greatly different between the forelimb and hindlimb, and expression of Hoxd11 and Hoxd12 is very weak in the hindlimb zeugopod. (iv) In the chick embryo, the decline of Hoxd11 and Hoxd12 expression in the hindlimb is prior to cartilage formation for the fibula, whereas in the forelimb, expression of Hoxd11 and Hoxd12 continues during the process of cartilage formation. However, the data obtained from whole-mount observations were not conclusive because they did not allow us to compare gene expression pattern directly in the same embryo, and the overlap of sites between Hox gene and cartilage formation was not convincing. We further examined the gene expression pattern in sections in order to determine whether precartilaginous mesenchymal condensation for the fibula appears inside the Hoxd11 and Hoxd12 expression domain or not.
Chick limb buds
At stage 25 in the forelimb autopod, precartilaginous mesenchymal condensation was not detectable by Aggrecan expression (Fig. 5A), while two foci of the condensations were observed in the zeugopod region (anterior condensation for the radius being small and weak, Fig. 5G). A clear boundary of Hoxd11/Hoxd12 expression was seen in the middle of the limb bud at both levels of the autopod (Fig. 5B,C) and the zeugopod (Fig. 5H,I). Comparing Figure 5G with 5H,I, Aggrecan-positive mesenchymal condensation for the ulna is localized inside the Hoxd11/Hoxd12-positive domain (see also Fig. 8).
In the hindlimb zeugopod at stage 25, transcripts of Hoxd11 and Hoxd12 were detected only at the peripheral margin of the posterior-dorsal region (Fig. 5K,L). On the other hand, clear expression of Hoxd11 and Hoxd12 was detected throughout the hindlimb autopod with the exception of the anterior small portion (Fig. 5E,F). A comparison of the positions of precartilaginous mesenchymal condensation detected by Aggrecan (two clear foci for the tibia and fibula, Fig. 5J) shows that the expression of Hoxd11 and Hoxd12 was not overlapped with precartilaginous mesenchymal condensation for the fibula at this stage.
At stage 27 in the forelimb and hindlimb, both genes were widely and predominantly expressed in the autopod with the exception of the anterior small portion (Fig. 5N,O,Q,R). In the forelimb bud, Hoxd11 and Hoxd12 were highly expressed in the posterior half with a boundary at the middle of the zeugopod (Fig. 5T,U). The zeugopod domain of Hoxd11 and Hoxd12 surrounded precartilaginous mesencymal condensation for the ulna detected by Aggrecan expression (Fig. 5S), confirming that expression of these Hox genes is maintained during the process of cartilage formation in the chick forelimb bud. On the other hand, in the hindlimb zeugopod, Hoxd11 and Hoxd12 were expressed only in a very thin layer of the posterior-dorsal periphery (Fig. 5W,X), without overlapping with Aggrecan expression (Fig. 5V).
Taking all of the results shown in Figures 2 and 5 together, we obtained evidence that in the chick hindlimb bud, precartilaginous mesenchymal condensation for the fibula is not overlapped with the Hoxd11 and Hoxd12 expression domain and that expression of these two genes disappears from the posterior mesenchyme long before cartilage formation.
Mouse limb buds
In the mouse autopod at all stages we examined (E11.0, E11.5, and E12.0), Hoxd11 and Hoxd12 were widely expressed in similar patterns with the exception of the anterior small portion in both forelimb and hindlimb buds (Fig. 6C,D,G,H). Expression in the zeugopod was relatively weak but had a domain in the posterior portion with a clear boundary at the middle of the limb bud at E11.0 and E11.5 (indicated by arrowheads in Fig. 6C,D,G,H). Indication of precartilaginous mesenchymal condensations in the zeugopod appeared at E11.0, and two foci for the condensations were clearly detected at E11.5 (Fig. 6A,B,E,F). These results coincide with those obtained from whole-mount observations (Fig. 3). Overlapping of expression between Hoxd11/Hoxd12 and col2a1/sox9 was definitive in the presumptive fibula region at E11.0 and E11.5 (Fig. 6E,F,G,H). Expression of Hoxd12 in the hindlimb zeugopod, however, was decreased at E11.5 (Fig. 6H). Both Hoxd11 and Hoxd12 genes ceased to be expressed in the zeugopod region by E12.0, but reminiscent expression could be seen at a thin layer in the posterior periphery both in the forelimb and hindlimb buds (Fig. 6C,D,G,H), the expression pattern of which resembled that in the chick hindlimb buds at stages 25 and 27 (Fig. 5).
These results indicated that in the mouse limb bud, Hoxd11 and Hoxd12 expression shows similar patterns spatially and temporally between the forelimb and hindlimb bud and that these genes continue to be expressed during the precartilaginous mesencymal condensation stage (E11.0 and E11.5). We conclude that fibula cartilage condensation appears in the Hoxd11/Hoxd12-positive domain in the mouse.
Gecko limb buds
At 14 dpo, two foci of zeugopod condensations were discernible (Fig. 7E,G), and the posterior focus representing the ulna and fibula was inside the Hoxd12 expression domain both in the forelimb and hindlimb (Fig. 7F,H). At 16 dpo, Hoxd12 expression decreased, and the domain was restricted to a peripheral thin layer in the zeugopod region (Fig. 7N,P), whereas Hoxd12 expression was seen throughout the autopod (Fig. 7J,L). At the later stage, the focus for the ulna and fibula was outside the Hoxd12 expression domain (compare Fig. 7M,O with 7N,P, respectively). As seen in the mouse limb bud (Fig. 6), the precartilaginous mesencymal condensation for the fibula definitively originated in the Hoxd12 expression domain.
Difference in temporal expression of Hoxd11/Hoxd12 between the forelimb and hindlimb
We showed spatial and temporal patterns of Hoxd11/Hoxd12 expression in the developing limb bud and compared the expression patterns in three species that belong to different groups of amniotes, particularly focusing on their expression in the presumptive zeugopod region both in the forelimb and hindlimb. A previous study on expression of Hox genes in the chick limb bud (Nelson et al. 1996) suggested that Hox gene expression in the developing limb bud has three phases: phase one for forming the proximal-most segment, phase two corresponding to the middle segment, and phase three corresponding to the distal segment. The phases we examined in this study correspond to the second phase (for the zeugopod) and third phase (for the autopod). A general remark from our observations is that the spatial expression pattern of Hoxd11/Hoxd12 is well conserved in the three animals examined; they are expressed in the presumptive regions of the posterior zeugopod and the autopod with the exception of the anterior-most portion.
Although a basic pattern for the second and third phases of Hoxd11/Hoxd12 gene expression is well conserved, temporal change of Hox expression for the second phase was different between the chick forelimb and hindlimb (Fig. 8B). In the chick hindlimb zeugopod, Hoxd11/Hoxd12 gene expression starts to decrease at stage 23, and the expression is greatly diminished by stage 25, with the exception of expression in the dorsal-peripheral margin. Stage 25 is the stage when cartilaginous mesenchymal condensation becomes visible, and condensation for the fibula is not included in the Hoxd11/Hoxd12 expression domain. On the other hand, condensation for the ulna in the forelimb buds is inside the Hoxd11/Hoxd12 expression domain at stage 25 and stage 27. At stage 27, Hoxd11/Hoxd12 expression in the hindlimb zeugopod becomes faint and restricted, whereas they are highly expressed in the autopod region, indicating that the early disappearance of Hox gene expression is specific in the hindlimb zeugopod. Our results confirmed results of previous studies (Mackem & Mahon 1991; Mackem et al. 1993; Nelson et al. 1996) that suggested hindlimb-specific downregulation of these Hox genes, and our results further reveal that the downregulation occurs prior to cartilage formation and is “hindlimb zeugopod-specific.”
Interspecies difference of Hoxd11/Hoxd12 expression in the hindlimb zeugopod
In fact, the early disappearance of Hoxd11/Hoxd12 gene expression is “chick” hindlimb zeugopod-specific. Both in the mouse and gecko embryos, precartilaginous mesenchymal condensation for the fibula emerges inside the Hoxd11/Hoxd12 expression domain, and then expression of the two genes begins to decrease in the hindlimb zeugopod region as in the forelimb (Fig. 8). In the hindlimb zeugopod, mouse Hoxd12 expression begins to decrease at E11.5, while Hoxd11 remains clearly expressed, suggesting that total numbers of Hoxd11/Hoxd12 transcripts are less than those in the forelimb.
We propose that expression pattern of Hoxd11/Hoxd12 genes in the zeugopod region can be categorized into three types (Fig. 8). The first type is continuous expression of Hoxd11/Hoxd12 genes in the posterior zeugopod during the early period of cartilage formation of the ulna and fibula. This type of expression can be seen in mouse and chick forelimbs, and probably gecko forelimbs and hindlimbs. The second type is a long period of Hox gene expression in the zeugopod region during the early period of cartilage formation but with the expression of one of the pair disappearing earlier. The mouse hindlimb zeugopod has this type of expression, and Hoxd12 expression disappears earlier in this case. We cannot exclude the possibility that expression in the gecko might be categorized into the second type because Hoxd11 expression remains unknown in gecko embryos (we have not yet succeeded in cloning Hoxd11 gene from the gecko for unknown reasons). The third type is that seen in the chick hindlimb zeugopod, in which expression of both genes disappears long before the zeugopod cartilage formation begins. There is a good correlation between the type of Hoxd11/Hoxd12 gene expression pattern and final morphology of skeletal element of the posterior zeugopod; as the timing of decline of Hoxd11/Hoxd12 gene expression is moved forward, bone morphology of the posterior zeugopod element becomes thinner and shorter (Figs 1, 8). This correlation infers that heterochronic shift of Hox gene expression among species plays a role in diversification of skeletal morphology.
Spatial regulation of expression of Hoxd11/Hoxd12 genes in the zeugopod is known to be governed by some regulatory elements, including global (long range) and local (short range) enhancers, both of which are necessary to build up the complex expression pattern of Hoxd genes during limb development (Beckers et al. 1996; Herault et al. 1998; Spitz et al. 2001) These elements are highly conserved among tetrapods (Beckers et al. 1996; Herault et al. 1998) but temporal regulation (cessation in particular) of expression of these genes remains largely unknown. It will be interesting to exchange zeugopod enhancers of Hoxd11/Hoxd12 among species in order to investigate cis- and trans- factors/elements responsible for species-specific regulation of temporal expression of Hoxd genes as examined for the axial skeleton (Gerard et al. 1997) and digit (Schneider et al. 2011).
Possible role of Hoxd genes in fibula morphology
It is important to understand the mechanism by which Hoxd11/Hoxd12 expression leads to the diversity in pattern of the zeugopod skeletal elements. It has been proposed that Hox genes determine the pattern by controlling the rates of cell proliferation in regions where they are expressed (Duboule 1995). Misexpression of Hoxd11 at an early stage of cartilage formation induces an extra segment in digit 1 of the chick hindlimb (Morgan et al. 1992; Goff & Tabin 1997), and misexpression of the gene at a later stage of bone formation affects bone length in the zeugopod (Goff & Tabin 1997). The authors suggested that Hox genes can contribute in two ways to both cartilage condensation and bone growth. Although the hindlimbs of the Hoxd11-mutant or Hoxd12-mutant mouse appear fairly normal (Davis & Capecchi 1994), and Hoxd11-13 knockout mice still have fibula (Sheth et al. 2007), some combinations of mutations, including Hoxa11-/Hoxd11- double mutant (Davis et al. 1995; Boulet & Capecchi 2004) and Hoxa11-/Hoxd12- double mutant (Davis & Capecchi 1996), show severe phenotypes in the zeugopod (shortened or almost entirely eliminated zeugopod elements). These phenotypes are generally mild in the hindlimb, and it is possible that there is a greater degree of redundancy of function in formation of the hindlimb by paralogous genes (Davis & Capecchi 1994), Hoxc9/Hoxc10/Hoxc11, all of which are expressed in the hindlimb but not in the forelimb (Peterson et al. 1994; Nelson et al. 1996). Mutants lacking all Hox11 genes (Hoxa11/-c11/-d11-triple mutants) indeed exhibit a severe phenotype of the zeugopod elements both in the forelimbs and hindlimbs (Koyama et al., 2010). An artificial overdose of Hoxc11 transcripts in the mouse hindlimb results in the absence of the fibula and reduced chondrocyte proliferation, suggesting that Hoxc11 negatively functions in zeugopod formation (Papenbrock et al. 2000). All of these results strongly suggest that a combination of Hox genes plays a pivotal role as an upstream mechanism in the formation of skeletal morphology and pattern in the zeugopod. However, no studies have confirmed the function of Hox genes in the formation of diverse morphology of zeugopod elements in various tetrapods. Further functional analyses of Hox genes should lead to elucidation of their function in bone morphogenesis and the causal relationships between different Hox gene expression and final morphology. In addition to Hoxd11/d12, it is further important to examine expression patterns of Hoxa11/c9/c10/c11, Tbx4 and Pitx1 as candidate genes up- and down- stream of Hoxd11/d12 (Nelson et al. 1996; Logan & Tabin 1999; Rodriguez-Esteban et al. 1999; Takeuchi et al. 1999; Papenbrock et al. 2000) during fibula formation and compare them among species.
How can early decline of Hoxd11/Hoxd12 expression contribute to the peculiar morphology of the fibula in the chick hindlimb? Regulation by Hox genes can be classified into two phases: one is early function in formation of a compartment (making a mesenchymal condensation unit for each skeletal element), and the other is morphogenesis in a compartment (growth and differentiation of bone). If Hoxd11/Hoxd12 expression participates in the early phase for compartmentation, the early decline of expression of these genes would give rise to incomplete formation of the fibula compartment, resulting in a larger, unusual compartment that is Hoxd11/Hoxd12-negative. Hampe (1960) showed that the insertion of a mica plate in the middle of the hindlimb bud, dividing the presumptive fibula and tibia region, sometimes resulted in an elongated fibula, and he suggested that there is competition between these two regions for a definitive number of mesenchyme cells (see also discussion in Archer et al. 1983). This might be due to incomplete compartmentation in the hindlimb zeugopod. If the second phase of Hox gene function on bone morphogenesis is the case, distal truncation of the chicken fibula may be due to insufficient growth and/or differentiation of the bone. Hox genes generally play a role in cell proliferation, particularly during chondrogenesis (Yokouchi et al. 1995; Goff & Tabin 1997; Boulet & Capecchi 2004), and early decline of Hox gene expression may prevent the distal epiphysis of the fibula from elongating as a part of the fibula, resulting in the distal epiphysis detaching from the diaphysis of the fibula cartilage (Archer et al. 1983). Both early and late functions of Hox genes might contribute to the peculiar morphology of the chick hindlimb zeugopod.
Heterochronic shift of gene expression in the developing limb bud
There are some good examples of temporal change in the duration of gene expression giving rise to morphological change and interspecies diversity in morphology of limbs. In Australian lizards with reduced numbers of digits, expression of the patterning molecule Sonic hedgehog (Shh) in limb buds ceases earlier than in limb buds of related lizards with more digits, and the duration of expression is consequently shortened and is correlated with decreased cell proliferation (Shapiro et al. 2003). Relationships between heterochronic shift of shh expression and morphological change have also been reported in developing pectoral fins, locomotor organs in fishes homologous to tetrapod forelimbs (Sakamoto et al. 2009). shh starts to be expressed in the fin buds at the timing of onset of hoxd10a expression in zebrafish, and the timing of onset of shh expression is critical for pectoral fin morphology (Sakamoto et al. 2009). Interestingly, if the onset of hoxd10a expression is shifted earlier or later, the timing of onset of shh expression is consequently and accordingly shifted, suggesting that a specific level of Hox gene expression leads to the heterochronic shift of shh expression in the pectoral fin bud.
These studies imply that temporal change of Hox gene expression may affect the duration of Shh expression also in the chick hindlimb bud. It has been reported that Shh expression disappears from the chick forelimb at around stage 29 (Riddle et al. 1993; Scherz et al. 2004) and from the hindlimb at around stage 27 (Dahn & Fallon 2000). Although these stages are actually later than the timing of decline of Hoxd11/Hoxd12 expression (stages 23–25), it is still possible that Shh signaling is also involved in peculiar morphogenesis in the chick hindlimb zeugopod. It should be interesting in the future to explore the relationships among molecular cascades for regulation of temporal gene expression, gene functions, cellular events on morphogenesis, and diversity in skeletal morphology.
We thank S. Yonei-Tamura and T. Yano for helpful comments on the manuscript. We are grateful to D. Duboule and Y. Kawakami for providing probes for in situ hybridization. This work was supported by research grants from the Ministry of Education, Science, Sports and Culture of Japan, KAKENHI (Grant-in-Aid for Scientific Research), the Global Center of Excellence Program (J03), Grant-in-Aid for Scientific Research on Innovative Areas and the “Funding Program for the Next Generation of World-Leading Researchers” from the Cabinet Office, the Government of Japan.