Our current understanding of avian development is based primarily on studies using chick embryos, and chick development is considered to be representative of all avian species. The ten thousand or so extant avian species constitute a monophyletic clade (Livezey and Zusi,2007; Hackett et al.,2008). Prominent differences among avian species in hatching time (∼11 to 100 days) and hatchling size (∼0.5 to 1,000 g) and maturity (extreme altricial to extreme precocial) have been attributed to variations during late embryonic development (Bergtold,1917; Starck and Ricklefs,1998; Deeming,2002). Early developmental programs and their molecular regulation are assumed to be conserved across the avian class. Centered on this assumption are questions pertaining to both amniote macroevolution (morphogenetic and developmental sequences in ancestral birds and amniotes) and avian microevolution (regulation of developmental speed, tissue size, and developmental heterochrony). Ontogenetic studies of avian development in representative orders are therefore essential for addressing these questions.
In mammals, developmental studies of its basal clades, the monotremes and marsupials, revealed putative ancestral features of primitive mammals. For instance, it has been suggested that many developmental patterns observed in model organisms such as mice are highly derived and that an ancestral mammalian embryo was likely rich in yolk, exhibited a meroblastic early cleavage pattern and had a superficial embryonic disc not covered by trophoblasts (Hughes and Hall,1998; Eakin and Behringer,2004; Selwood and Johnson,2006). Findings from such comparative studies in mammals underscore the importance of investigating developmental patterns of basal clades within the other two amniote lineages, the birds and reptiles.
To date, a complete or near-complete developmental series has been documented in 10 avian species. They are the chick (Gallus gallus; Hamburger and Hamilton,1992), Japanese quail (Coturnix japonica; Ainsworth et al.,2010), turkey (Meleagris gallopavo; Mun and Kosin,1960), pheasant (Phasianus colchicus; Fant,1957), Bobwhite quail (Colinus virginianus; Hendrickx and Hanzlik,1965), Pekin duck (Anas platyrhynchos; Kaltofen,1971; Dupuy et al.,2002), mallard (Anas boschas; Koecke,1958), society finch (Lonchura striata; Yamasaki and Tonosaki,1988), Adelie penguin (Pygoscelis adeliae; Herbert,1967), and lapwing (Vanellus cristatus; Grosser and Tandler,1909). Seven of them belong to the superorder Galoanserae, a narrow clade that includes two orders: Galliformes and Anseriformes. All 10 species are neognaths and developmental ontogeny of the paleognath, the basal avian clade, has not been described. To understand conserved and divergent developmental features within the avian lineage, we carried out morphological and molecular studies of the embryonic development of the emu, a flightless bird and a member of the paleognath. Native to Australia, emus are now farmed worldwide. Studies of prehatching emus haven been focused mainly on egg hatchability for commercial reasons. Its embryonic development has not been described in the literature except for a brief report in late nineteenth century (Haswell,1887).
Emu eggs were purchased from a bird farm in Japan (Fig. 1A,B; Experimental Procedures section). Eggs were incubated at 37°C with 20–40% humidity for 0–44 days to obtain embryos of developmental stages varying from HH1 to HH43. Emu embryos hatch after 56 days of incubation in the wild and 50–51 days in artificial incubators kept at 37°C (Sales,2007). No attempt, however, was made in this study to incubate eggs to the hatching stage. The size of the emu yolk is proportionally larger than that of the chicken yolk (Fig. 1C). The emu vitelline membrane is double-layered (Fig. 1D, white arrows), as in the chick. The inner surface of the inner vitelline membrane layer is rough, more so in the region that covers the developing embryo (Fig. 1D, black arrow). Freeing young emu embryos from the vitelline membrane was difficult after detaching the epiblast edge/vitelline membrane attachment margin, suggesting that the apical surface of the epiblast in the emu interacts with the vitelline membrane. Different methods were used to obtain emu embryos depending on their developmental stages and on whether the extraembryonic region was to be included (Experimental Procedures). The Hamburger and Hamilton (HH) staging system (Hamburger and Hamilton,1992), followed in this work for most of the emu developmental stages, is used to indicate emu equivalents of chick HH stages. The exception is for the period corresponding to the chicken stages HH8–HH16. During this period of emu development, the number of somites and the development of the nervous and cardiovascular systems do not exhibit the type of correlation seen in the chick. A rough adjustment would be to add 4–5 somites for each emu HH stage. To avoid confusion, these stages are referred to in this work by their somite numbers.
Overall growth rate of emu embryos is shown in Figure 2. A majority of the eggs were incubated to obtain early developmental stages (the first 4 days). Proportional to its total incubation time, an emu embryo takes approximately 2–3 times longer to reach an equivalent chicken stage.
From HH1 to HH7.
Representative embryos at these stages are shown in Figure 3. These embryos were obtained after an incubation period of up to 4 days. On average, HH1 corresponds to 0–1 day of incubation, HH2 to 1–1.5 days, HH3 to 1.5–2.5 days, HH4 to 2.5–3 days, and HH5–HH7 to 3–4 days. Minor variation in embryo stages from eggs incubated for a same period of time did not correlate with difference in the egg collection time, storage duration or embryo quality. The stereotypic morphology of the rudimentary primitive streak seen in HH2 chick embryos, a triangular thickening at the posterior area pellucida, was not observed in HH2 emu embryos. The stereotypic morphology of an HH4 chick embryo, a well-formed primitive streak with a uniform primitive groove along the entire length of the primitive streak, was not observed either in HH4 emu embryos. At stages HH4+ to HH7, gross morphology of the emu embryos is very similar to that of the chick embryos.
From early somite stage to before limb bud formation.
Representative embryos at these stages are shown in Figure 4. Embryos of these stages were obtained from eggs incubated for 3–6 days. This period corresponds to stages HH8–HH16 in the chick and of 3–38 somites in the emu. More somites are generated in the emu than in the chick during this period. Judging from embryo size, central nervous system development and extraembryonic vascular development, an emu embryo with a given number of somites is of a younger developmental state than a chick embryo with the same somite number. A close approximation of emu development in this period would be to add 4–5 somites for the progression of each HH stage before the onset of circulation, and 3–4 somites for each HH stage after circulation. The circulation is well-established at the 32 somite stage, similar in overall embryo morphology to an HH14 chick embryo. At the 26 somite stage, with an overall morphology similar to a late HH12 stage chick embryo, the circulation is still rudimentary in the emu. Formation of the amnion occurs also during this period, starting at approximately the 14 somite stage and leaving only a small opening at the posterior end by the end of this period.
From HH17 to HH43.
Representative embryos at these stages are shown in Figure 5. Embryos of these stages were obtained from eggs incubated for 5–44 days. Initiation of hindlimb outgrowth is visible at the 41 somite stage, corresponding to an HH17 chick embryo. The HH17/18 emu embryo is of similar size and morphology to the same stage chick embryo. Early stages in this period were judged according to the hindlimb morphology, together with the brain and pharyngeal morphology. More advanced embryos were staged based on the feather, eye, beak, and hindlimb morphology. Adult emus have vestigial forelimbs. Emu forelimb buds start to form at HH20–HH21 and have a reduced growth rate compared with that of the hindlimbs. Otherwise, they appear to have a normal initial patterning. Except for a reduction in digit numbers (1–2), bone and cartilage staining of an HH40 embryo (Fig. 6A–C) and cartilage staining of an HH38 embryo (Fig. 6D,E) revealed the presence of all forelimb skeletal elements.
Size and Growth Rate
Unincubated emu and chick embryos are of a similar size (Fig. 7A). This is also true for embryos up to stage HH7 when only the area pellucida is compared (Fig. 7B,C). From the early somite stage to limb bud formation, an emu embryo with a given somite number is smaller in size and is of a younger overall developmental stage than a chick embryo with the same number of somites (Fig. 7D,E). When hindlimb bud appears at HH17/18, emu and chick embryos are again of a similar size (Fig. 7F). After this stage, an emu embryo outgrows a chick embryo (Fig. 7G–I). If the emu stages corresponding to chicken HH8–HH16 are adjusted using a combination of criteria (somite number, head morphology, and cardiovascular development), it can be said that before limb bud formation an emu embryo is of the same physical size to a chick embryo, and that after limb bud formation an emu embryo becomes progressively larger.
From the outset, the area opaca of the emu embryo grows faster than the area pellucida. As mentioned above, an unincubated emu embryo is similar in size to a chick embryo in both the area pellucida and area opaca. The embryo proper in the emu and chick grows at a similar rate during early development, but emu area opaca expands much more quickly (Fig. 7J). This can be schematized by the relative areas covered by the area pellucida (white) and area opaca (gray) of young stage emu embryos (Fig. 7K). The expansion of the area opaca has been likened to the epiboly process in lower vertebrates (New,1959). The faster rate of area opaca expansion in the emu thus likely reflects the need to enclose a much larger surface area to complete the epiboly. The area opaca can be divided into the area vasculosa (vascularized region containing all three germ layers) and the area vitellina (nonvascularized region containing only the ectoderm and endoderm). The mesoderm in the area vasculosa is called the extraembryonic mesoderm and it comes from cells located at the posterior primitive streak in the developing embryo proper (Nakazawa et al.,2006; Weng et al.,2007; Shin et al.,2009; Alev et al.,2010; Sheng,2010). Indeed, similar to the chicken extraembryonic mesoderm, emu extraembryonic mesoderm in the area vasculosa expresses the hemangioblast marker Lmo2 (Fig. 8A,A′). The origin of the extraembryonic mesoderm in the emu was investigated by electroporating the posterior primitive streak cells with a green fluorescent protein (GFP) -expressing DNA construct at HH3, followed by culturing the embryo to HH6/7 (Fig. 8G–I). Results from this experiment suggested that, like in the chick, the extraembryonic mesoderm in the emu is derived from the posterior part of the primitive streak. The region occupied by the area vasculosa (Fig. 8A–F) in the emu is similar in size to that in the chick, and reflects the developmental stage of the embryo proper. Thus, although the growth in the area opaca outpaces that in the area pellucida, the area vasculosa within the area opaca grows at a rate similar to that of the area pellucida. This is exemplified by a comparison of HH20 emu and chick embryos (Fig. 8F′). At this stage, the emu embryo proper and area vasculosa together occupy a similar absolute area to that of the chicken counterparts (Fig. 8F′ inset, same scale). Of interest, like in the chick, emu extraembryonic mesoderm cells can differentiate autonomously after their migration to the area vasculosa. As a result, the extraembryonic hematopoietic and vascular differentiation and the size of the area vasculosa are relatively normal even when the embryo proper is severely malformed (Fig. 8J).
Time-Lapse Imaging Analysis of Early Development
To visualize the development of individual embryos, we performed time-lapse imaging analysis. Attempts to image emu development in ovo failed. We therefore used the New culture method and focused on early developmental stages. Emu embryos ranging from unincubated (HH1) to somite stages can be cultured successfully for a period of up to 2 days using emu vitelline membrane and thin albumen. Confirming our earlier observation (Fig. 3), time-lapse imaging analysis showed that unincubated emu embryo takes approximately 1 day to reach HH2 and another half a day to reach HH3 (Fig. 9A, Supp. Movie S1, which is available online). Elongation of the primitive streak at young HH3 occurs in both posterior-ward and anterior-ward directions (Fig. 9B, Supp. Movie 2), with at least a third of the final streak length being added posteriorly. In neither the time-lapse imaging analysis spanning late HH1 and HH2 nor normal stage HH2 embryos was the triangular-shaped accumulation of streak precursor cells at the posterior region of the area pellucida observed. An embryo incubated from HH4/4+ took 7 hr to reach HH5/6, 20 hr to reach 6/7-somite stage, and 26 to reach the 9/10-somite stage (Fig. 10, top). Somitogenesis was more closely monitored from the 6/7-somite stage to the 9/10-somite stage, with images taken every 5 min (Fig. 10, bottom; Supp. Movie S3). The somitogenic periodicity was calculated to be 100–110 min, slightly longer than the 90-min reported for the chick embryos.
Cloning of Emu Brachyury, Chordin, and Shh Gene/Gene Fragments
The above observations suggested that, although early embryogenesis in these two species passes through morphologically similar stages, the tempo of developmental progression from one stage to the next varies significantly. To investigate whether this variation is reflected in the expression of genes that are critical for early embryonic patterning, we cloned the emu Brachyury, Chordin, and Sonic hedgehog (Shh) genes and carried out RNA in situ hybridization analyses of these genes during emu early development. The emu Chordin and Shh genes were cloned as partial DNA fragments with a length of 620 base pairs (bp) and 684 bp, respectively. For the emu Brachyury gene, a full-length cDNA sequence was cloned and confirmed (see the Experimental Procedures section; Fig. 11A). Sequences for these three genes have been deposited in the NCBI database with the following accession numbers: HQ336495 (Chordin), HQ336496 (Shh), and HQ336494 (Brachyruy). At the nucleotide level, the cloned fragment of emu Shh is 97% identical to the corresponding region of the chicken Shh (nucleotides 98–781 of NM_204821; Fig. 11C), emu Chordin fragment is 87% identical to the chicken Chordin (nucleotides 903–1528 of NM_204980; Fig. 11B), and emu Brachyury is 91% identical in the 1.3 kb coding region and 87% identical in the 1.8 kb 3′ untranslated region (UTR) to the chicken counterparts (Fig. 11A). In situ probe for the emu Brachyury gene was generated from a 367 bp DNA fragment in the 5′ end of the coding region (Fig. 11A).
Expression of Shh, Chordin, and Brachyury Genes During Early Development
Expression of the emu Shh gene was investigated with only one stage HH7 embryo. Similar to chicken Shh, Emu Shh is expressed in the axial mesoderm and also asymmetrically in the regressing node (Fig. 11D–F). This pattern is very similar to that of chicken Shh (Levin et al.,1995). Expression of the Chordin and Brachyury genes was investigated with embryos ranging from HH1 to HH7. Their expression patterns at HH1–HH5 are shown in Figure 12 and sections shown in Figure 13. Overall, emu Chordin (Fig. 12A–D) is expressed in an identical way to chicken Chordin (Streit et al.,1998; Chapman et al.,2002). It is first detected in late HH1 embryos (Figs.12A; 13A,A′) in the posterior region of the area pellucida. At HH2, it is expressed in the anterior half of the streak (Fig. 12B), and at HH3+ in the anterior third of the streak (Figs. 12C, 13B,C). At HH5−, emu Chordin is expressed in the node and the head process, but is negative in the prechordal plate (Figs. 12D, 13D,E,E′). The emu Brachyury is not expressed in any of the HH1 stage embryos analyzed (data not shown). It is barely detectable in HH2 stage embryos (Fig. 12E). Its expression becomes strong from HH3 (Figs. 12F–H, 13F–L) in a pattern very similar to chicken Brachyury (Knezevic et al.,1997). Thus, the only prominent difference in the expression of these two genes between these two species is that emu Brachyury starts to be expressed at the transition between stages HH2 and HH3, whereas chick Brachyury starts its expression at late HH1.
Haswell gave a brief report of emu early development in 1887 (Haswell,1887), which remains as the only embryological study of the emu in the literature. In that study, the morphology of seven embryos incubated for between 2 and 7 days was described with line drawings. The incubation temperature was not clearly specified in that report and the total incubation period of emu was referred to as 3 months; different from that now, generally agreed to be 50–56 days. In this work, we examined approximately 100 emu embryos ranging in stages from unincubated to a few days before hatching. These stages were analyzed in comparison with the chicken development using the Hamburger and Hamilton (HH) staging system (Hamburger and Hamilton,1992).
The HH staging system was devised to describe chicken embryonic development, and has been adopted in later studies for both Galloanserae (quail, turkey, and duck) and Neoaves (penguin and finch) species. Our studies revealed that emu embryogenesis passes through developmental stages similar to those described for the chicken. The total incubation time for the emu is approximately 2.5 times longer than for the chick. This difference is reflected throughout emu embryonic development. A similar overall decrease in developmental speed was mentioned for ostrich development after the limb bud stages (Ar and Gefen,1998; Gefen and Ar,2001). This broad inverse correlation between the incubation time and developmental speed, however, does not hold true as a general rule. For instance, quail embryos require 4 days (20%) less to hatch than chick embryos, yet their developmental speed is identical up to HH36 (Ainsworth et al.,2010). Mallard duck embryos take 7 days (35%) more to hatch than chicken embryos, yet they take only 1 day more to reach HH36 (Koecke,1958). The society finch, hatching in 17 days, grows more slowly than the chick, with an initial 1 day lag until its growth rate catches up and exceeds the chicken's at HH36 (Yamasaki and Tonosaki,1988). The Adelie penguin hatches in 35 days. But its early development shows a disproportionally slower speed, requiring 4 days for HH3/4, 7 days for HH13, 10 days for HH18, and 24 days for HH36 (Herbert,1967,1969).
These observations suggest that, although avian embryonic development is relatively well-conserved morphologically, growth rates may vary widely. This variability is not restricted to the maturation stages (>HH37), which has been postulated as the major cause for the variation in avian incubation period, but is evident from the earliest stage of embryonic development. Closely related species (e.g., in Galloanserae), however, do seem to have a more comparable developmental speed throughout most of their prematuration stages. Reported small differences in growth rate among Galloanserae during these stages may be explained by minor variations in the duration of in utero development, incubation temperature and other experimental parameters, in addition to intra-specific population variations and genuine inter-specific growth rate differences. It has been speculated that the reason for a minimum of 11 days of incubation in any avian species studied is the strict requirement of completing an invariable prematuration development and a minimal period of maturation to reach the equivalent of HH42 (Nice,1953; Starck and Ricklefs,1998). However, data from the emu (this work) and the penguin (Herbert,1967) suggest prematuration developmental rate can also vary widely. In these two cases and in the ostrich of which the latter half of the prematuration development has been described (Ar and Gefen,1998; Gefen and Ar,2001), the developmental rate is significantly slower than in the chick and its allies. It is conceivable that some avian species/clades, not yet studied, may have a faster prematuration developmental rate than currently known.
In addition to the variation in overall growth rate, minor heterochronous developmental features have also been observed. In the emu, especially during its early development, these features include the time needed to reach the full primitive streak stage (HH4) and from HH4 to HH7, the speed of somite formation, the relationship between the development of the somites, the nervous system and the cardiovascular system, the differential growth of the area pellucida and area opaca, and the onset and rate of the forelimb and hindlimb development. These types of heterochrony are not uncommon and have been well-documented in many avian developmental studies (Chen,1932; Daniel,1957; Bakst et al.,1997; Dupuy et al.,2002; Schneider and Helms,2003; Blom and Lilja,2005; Sellier et al.,2006). A key question in avian comparative developmental studies is whether a given developmental feature in the chick and its regulatory mechanisms revealed from molecular studies in the chick can be extrapolated as representing the avian development in general. Addressing such questions therefore requires comparative study within a narrow developmental window in several representative species, and for this purpose we have chosen emu embryos of the gastrulation stages (<3 days) for more detailed analysis.
In the first 3 days of the emu development, the embryo grows from HH1 to early somite stages. During this period, the basic vertebrate body plan is laid down, including early dorsoventral patterning, primitive streak formation, and mesoderm generation. The primitive streak is an important criterion for staging HH1 to HH4 embryos in the chick. This period takes approximately 18 hr in the chick and 2.5 days in the emu. A rudimentary streak is apparent in the emu after 1 day of development. Elongation of the streak requires another day, and formation of a streak with a primitive groove takes an additional half day. Given that the area pellucida is of the same size in these two species and that cell density and size do not appear to be very different, the slow speed in reaching HH4 in the emu may have a tight genetic-wired regulation. This would serve as a good model for the study of developmental size and rate regulation in general. Future studies combining grafting, labeling, and imaging techniques may be able to shed light on this issue.
The morphology of the young streak at HH2 in the emu is different from that in the chick. This suggests that morphogenetic movement of cells located in the posterior area pellucida and the marginal zone can vary among avians. Emu HH2 streak always forms within the area pellucida, at a certain distance away from the posterior marginal zone and has the shape of a thin and short line rather than the triangular shape at the border of the area pellucida and the marginal zone seen in the chick. The morphology of emu HH3 streak also shows small differences from that in the chick. Posterior part of the emu streak at HH3 is not as morphologically distinct as in the chick, and the elongation of the streak at HH3 has a clear contribution both anteriorly and posteriorly. Emu primitive streak at HH4 appears different as well, with the node not as distinct and the primitive groove not as uniform as in an HH4 chick embryo. From HH4+ onward, the regressing primitive streak in the emu looks very similar to that in the chick.
The relationship between streak formation and mesoderm induction and patterning was also investigated in this work. The primitive streak, serving as a morphological criterion of developmental progression, is composed of precursor cells that will later on leave the epiblast and become true mesoderm cells after undergoing epithelial to mesenchymal transition (Nakaya and Sheng,2008; Nakaya et al.,2008). It also serves as a site for early dorsal/ventral patterning of these mesoderm precursor cells (Alev et al.,2010). Chordin expression marks the future dorsal side before the formation of the streak and the dorsal mesoderm precursors during streak elongation (Streit et al.,1998; Chapman et al.,2002). Our expression analysis of the emu Chordin gene indicates that these two processes are conserved between the emu and chick. Expression analysis of the emu Brachyury gene, however, revealed a surprisingly late onset of its expression. Unlike the chick, in which the Brachyury gene starts to be expressed at late HH1, in the emu it starts to be expressed after the appearance of a rudimentary streak. This suggests an uncoupling of two distinctly regulated developmental events: a morphogenetic movement leading to the formation of the primitive streak and an inductive event leading to allocation of some epiblast cells as mesoderm precursors. Other important developmental processes that take place before HH2, such as the formation of the Koller's sickle and the hypoblast (Eyal-Giladi and Kochav,1976; Stern,1990; Callebaut and Van Nueten,1994), have not been investigated in this work.
Another example of early developmental heterochrony is somitogenesis. In the chick, somites form with a periodicity of 90 min, whereas in the emu, our study suggested a periodicity of 100–110 min. Future experiments with a larger size of the samples, a longer duration of the time-lapse imaging and variations in the incubation temperature (∼36–39°C) will be needed to confirm this observation. The presumed small decrease in the speed of somite formation and a greater decrease in the overall developmental speed result in a relatively faster rate of somite formation in the emu than in the chick. The chick embryo generates 55 pairs of somites in total (Gomez and Pourquie,2009), with 50 of them formed before the end of HH22. The total number of somites in the emu is unclear. In an emu specimen of stage HH21 (shown in Fig. 5), we have counted 54 pairs of somites. An emu neonate was reported to have 51 vertebrae (Starck,1996), whereas a newly hatched chicken has approximately 40. Before limb bud formation at HH17/18, the somite number is a major staging criterion in all the avian species that have been studied. This does not seem to apply to the emu. An adjustment of the somite numbers (more somites in the emu for a given stage between HH8 and HH16), while still using other criteria in the HH staging system, will allow a more accurate comparison among avian species during this developmental period.
Aside from these differences, our analysis revealed many similarities between these two species. For instance, the Shh is asymmetrically expressed in the node in both species, suggesting a similar mechanism for the L/R specification. The specification of extraembryonic mesoderm, contributing to the vascular and hematopoietic systems in the area vasculosa, appears to be conserved. The size of the embryo proper is strikingly similar in these two species during early development, despite the extreme variation in the size of the area opaca. The size of the area vasculosa within the area opaca is also well conserved, and shows a proportional correlation with that of the embryo proper, rather than of the area opaca. The emu is among the largest in extant birds, second only to the ostrich. It is, therefore, interesting to notice that embryonic size difference is manifested only after the limb bud stages. This suggests that molecules regulating embryonic tissue size in birds start functioning only after the completion of basic body patterning. It is unclear whether this regulation is exerted cell-autonomously or through physiological changes in the levels of cytokines and growth hormones. Studies using interspecific tissue grafting among Galloanserae suggest both cell-autonomous and non–cell-autonomous mechanisms may be at work (Ohki-Hamazaki et al.,1997; Schneider and Helms,2003; Tucker and Lumsden,2004; Eames and Schneider,2005,2008). Emu/chick or emu/quail cross-specific graft experiments in the future may be able to address this question.
As a ratite, one of the defining features of the emu is its lack of functional wings. Despite this obvious relevance, forelimb development was not focused in this study. Nevertheless, gross morphological changes during emu development indicate the forelimbs are formed with a slight delay in developmental timing and a severe reduction in growth rate (Figs. 5, 6). In summary, our analysis of the emu embryogenesis suggests that extant avian species have an overall conservation in developmental ontogeny, a wide variation in developmental speed and prominent cases of developmental heterochrony. Chick embryos remain a powerful model for avian developmental studies. A broad spectrum of basic description in representative avian species may help data interpretation of molecular and cellular analyses in the chick.
Eggs, Egg Incubation, and Embryology
Fertilized emu eggs were purchased from Kakegawa Kachoen (Kakegawa, Japan) during the 2008 and 2009 mating seasons. The facility maintains a colony of two dozen adult emus for exhibition. The mating season in Japan typically lasts from late November to February, and approximately 10 eggs per week were collected during this period. In total, approximately 200 eggs were obtained, with data from approximately 100 embryos used in this study. Eggs collected in the facility were stored at room temperature before shipping, and eggs received in the lab were either incubated immediately or stored at room temperature for later incubation. Total duration of room temperature storage before incubation, varying from 1 to 3 weeks, had no obvious correlation either with embryo quality or with observed variation in embryonic development after a fixed period of incubation. An emu egg before incubation measures on average 134 ± 5 mm (long axis) by 90 ± 4 mm (short axis), and weighs 598 ± 61 g (n = 44).
Emu eggs were incubated at 37°C with horizontal 90-degree rotation every 2 hr and with the humidity kept between 20 and 40%. Fertility rate, judged by the egg having a well-formed embryo at the time of shell-opening, is over 90%. To obtain late stage embryos (after HH40), the egg shell was cracked open mechanically. For early stage embryos, two small holes were drilled first at opposite ends of the long axis. Approximately 20–30 ml of albumen were removed using a syringe. A window of the size of a large coin was opened at one end using a circular drill and excess albumen was removed. This was followed by opening a second and larger window to remove the top third of the egg shell. The embryo was located by rotating the yolk carefully and cut out with scissors. In some cases, the entire yolk with the embryo was poured into a tray of Pannett-Compton solution to obtain embryos with intact area opaca or area vasculosa.
Embryos were either fixed immediately or used for time-lapse imaging. Embryo pictures were taken either before or after fixation. For time-lapse imaging, embryos were cultured ex ovo in the New culture setting. The method is identical to the one used for chick embryo culture, except for use of emu vitelline membrane and thin albumin in this case. In a few test cases, emu embryos were shown to be able to grown reasonably well on chicken vitelline membrane. Both still and time-lapse images were taken using an Olympus ZX12 microscope mounted with an Olympus DP70 camera. For large embryos, a metal stand and an Olympus PEN EP1 camera was used for picture taking. A home-made warm chamber was used during time-lapse imaging. In the case of the time-lapse movie shown in Supp. Movie S1, several short movies were used with no time interruption for the assembly of the long 36-hour movie with the Adobe Photoshop CS4 software. In one stage HH3 emu embryo, posterior streak cells were electroporated with a GFP expressing DNA construct using chick electroporation settings.
Cloning of Emu Genes, RNA In Situ Hybridization of Emu Embryos and Histology
Fragments of emu Brachyury, Chordin, and Shh were amplified from a stage HH7 emu embryo cDNA preparation, using primer sets based on sequence comparison of available vertebrate orthologous genes. For the emu Brachyury gene, additional PCR-based amplifications, 5′ rapid amplification of cDNA ends (RACE) and 3′ RACE were used to assemble the full-length sequence. The final Brachyury sequence was confirmed by additional independent PCRs from cDNA. All three emu gene sequences have been deposited in the NCBI database with the following accession numbers: emu full-length Brachyury (HQ336494), emu Chordin fragment (HQ336495), and emu Shh fragment (HQ336496).
For RNA in situ hybridization, the cloned fragments of emu Chordin and Shh genes were used to generate digoxigenin (DIG) -labeled probes. An emu Brachyury fragment, instead of the full-length sequence, was used for making emu Brachyury in situ probe, corresponding to nucleotides 43–409 of the deposited full-length emu Brachyury sequence (HQ336496). For Lmo2 in situ, chicken Lmo2 probe corresponding to nucleotides 1–399 of AF468789 was used. This region is 93% identical to zebra finch Lmo2 at the nucleotide level. Lmo2 gene is well conserved in vertebrates, with 100% protein level identity among the human, chick, and zebra finch orthologues. Chicken protocols were followed for in situ hybridization, histology, and skeletal staining.
We thank Mr. Junichi Otsuka at the Kakegawa Kachoen for the supply of emu eggs, Mr. Hideshi Mizuno at the Kakegawa Kachoen for the photo shown in Figure 1A, Ms. Kanako Ota in the Lab for Early Embryogenesis for coordinating the egg shipment, and members of the Lab for Early Embryogenesis and the Lab for Sensory Development for tolerating the temporary invasion of the emu.