Expression of a hindlimb-determining factor Pitx1 in the forelimb of the lizard Pogona vitticeps during morphogenesis

With over 9000 species, squamates, which include lizards and snakes, are the largest group of reptiles and second-largest order of vertebrates, spanning a vast array of appendicular skeletal morphology. As such, they provide a promising system for examining developmental and molecular processes underlying limb morphology. Using the central bearded dragon (Pogona vitticeps) as the primary study model, we examined limb morphometry throughout embryonic development and characterized the expression of three known developmental genes (GHR, Pitx1 and Shh) from early embryonic stage through to hatchling stage via reverse transcription quantitative polymerase chain reaction (RT-qPCR) and immunohistochemistry (IHC). In this study, all genes were found to be transcribed in both the forelimbs and hindlimbs of P. vitticeps. While the highest level of GHR expression occurred at the hatchling stage, Pitx1 and Shh expression was greatest earlier during embryogenesis, which coincides with the onset of the differentiation between forelimb and hindlimb length. We compared our finding of Pitx1 expression—a hindlimb-determining gene—in the forelimbs of P. vitticeps to that in a closely related Australian agamid lizard, Ctenophorus pictus, where we found Pitx1 expression to be more highly expressed in the hindlimb compared with the forelimb during early and late morphogenesis—a result consistent with that found across other tetrapods. Expression of Pitx1 in forelimbs has only rarely been documented, including via in situ hybridization in a chicken and a frog. Our findings from both RT-qPCR and IHC indicate that further research across a wider range of tetrapods is needed to more fully understand evolutionary variation in molecular processes underlying limb morphology.

JM, 0000-0002-9994-6423 With over 9000 species, squamates, which include lizards and snakes, are the largest group of reptiles and second-largest order of vertebrates, spanning a vast array of appendicular skeletal morphology. As such, they provide a promising system for examining developmental and molecular processes underlying limb morphology. Using the central bearded dragon (Pogona vitticeps) as the primary study model, we examined limb morphometry throughout embryonic development and characterized the expression of three known developmental genes (GHR, Pitx1 and Shh) from early embryonic stage through to hatchling stage via reverse transcription quantitative polymerase chain reaction (RT-qPCR) and immunohistochemistry (IHC). In this study, all genes were found to be transcribed in both the forelimbs and hindlimbs of P. vitticeps. While the highest level of GHR expression occurred at the hatchling stage, Pitx1 and Shh expression was greatest earlier during embryogenesis, which coincides with the onset of the differentiation between forelimb and hindlimb length. We compared our finding of Pitx1 expression-a hindlimb-determining gene-in the forelimbs of P. vitticeps to that in a closely related Australian agamid lizard, Ctenophorus pictus, where we found Pitx1 expression to be more highly expressed in the hindlimb compared with the forelimb during early and late morphogenesis-a result consistent with that found across other tetrapods. Expression of Pitx1 in forelimbs has only rarely been documented, including via in situ hybridization in a chicken and a frog. Our findings from both RT-qPCR and IHC indicate that further research across a wider range of tetrapods is needed to more fully understand evolutionary variation in molecular processes underlying limb morphology.

Background
The integration of the fields of evolution and developmental biology is leading to significant advances in our understanding of the molecular basis of morphological evolution. However, there has been a historical reliance on model organisms in developmental biology, with most studies focusing on relatively few and evolutionarily distant species that are suitable for embryological and genetic manipulation [1], such as mice, chickens, frogs and zebrafish. A case in point is research on the molecular basis of variation in tetrapod limb morphology. Many molecular aspects of limb development have been studied extensively in model tetrapods, mainly chicken and mouse [2]. However, this represents only a small fraction of tetrapod limb diversity and evolution. One group for which very limited research into the developmental and molecular processes underlying limb morphology has been conducted are reptiles. Reptiles form a key evolutionary group in terrestrial vertebrates, both in terms of morphological diversity and the evolution of the tetrapod limb.
Squamates, which include lizards and snakes, are the most speciose group of reptiles, and provide a promising system for examining developmental and molecular processes underlying limb morphology. Limb morphology and especially hindlimb length are known to influence the ecology of many lizard species [3]. Within lizards, the relationship between limb length and ecology is largely a result of increased hindlimb length, which is associated with increased running speed [4][5][6]. Recent work on Anolis lizards has shown that variation in limb length results from changes occurring very early in embryonic development, prior to formation of the cartilaginous long bone anlagen [7]. Anolis lizards serve as an emerging model system for the study of limb development and evolution [7,8], because it is a highly speciose genus that has been well studied, both ecologically and morphologically. Additionally, the Anolis carolinensis genome was the first squamate genome to be sequenced [9]. However, understanding the genetic mechanisms of limb development across a variety of organisms can help elucidate the evolutionary processes that lead to morphological diversity, which ultimately allows terrestrial vertebrates to occupy a vast array of ecological niches [10]. Thus, expanding limb developmental research to other squamate lineages has the potential to provide a particularly powerful system to investigate the developmental and molecular basis underlying tetrapod limb diversity.
Australian agamid lizards (Agamidae: Amphibolurinae) form an ideal study system to examine developmental and molecular processes underlying limb morphology. They are a speciose lineage (more than 72 species) with a wide diversity in limb morphology, and agamids have been shown to exhibit convergent morphological and ecological evolution with the clade incorporating pleurodont iguanians [11], which includes the Anolis lizards. Moreover, the genome of the Australian agamid Pogona vitticeps has recently been sequenced [12]. Agamids, in particular P. vitticeps, constitute ideal laboratory animals, because they are oviparous with medium to large clutches, with a well-established history of captive breeding, and they are readily commercially available. Consequently, Australian agamids provide an exciting new avenue for research into the developmental and molecular processes underlying the evolution and diversity in tetrapod limb morphology.
We integrated a morphological study of limb development during embryogenesis in P. vitticeps with the molecular investigation of gene expression during embryonic limb development. Our study covered stages of limb development from oviposition (egg laying) to 21 days posthatching, and thus all stages at which variation in limb length may originate. Although variation in limb length occurs very early during embryogenesis in Anolis [7], there are four main developmental phases in which variation could occur: limb-bud initiation, limb-bud outgrowth and patterning, morphogenesis (which is the differentiation and development of limb structures) and growth [13]. We sought to characterize limb development during embryogenesis in Australian agamids and to determine at which stage limb length variation occurs. We then undertook the molecular component of this study to investigate gene expression during embryogenesis.
We selected three genes that are known to be involved at different stages of limb development in model organisms, such as mouse and chicken (see the brief review below). We quantified gene transcription in the forelimbs and hindlimbs of P. vitticeps for the paired-like homeodomain transcription factor 1 (Pitx1), sonic hedgehog (Shh) and growth hormone receptor (GHR). We hypothesized, based on previous research in vertebrates, that these genes would each show a peak in transcription levels at different phases of embryonic limb development (i.e. Pitx1 during limb-bud outgrowth and patterning, SHH during morphogenesis and GHR during the growth phase). We tested this hypothesis using RT-qPCR for each gene in both forelimbs and hindlimbs across nine developmental stages in P. vitticeps. We then examined Pitx1 expression in forelimbs and hindlimbs using immunohistochemistry (IHC). To determine whether the pattern of Pitx1 expression in P. vitticeps is representative of Australian agamid lizards, we investigated Pitx1 transcription and expression during morphogenesis in a closely related Australian agamid lizard, Ctenophorus pictus.

Limb development genes: Pitx1, Shh and GHR
Pitx1 is a hindlimb-determining factor expressed early in development [14,15], with Pitx1 deletion causing loss of skeletal hindlimb structures in mice [14,16]. Ectopic expression of Pitx1 in the developing forelimb of mice and chicks generates some bone and soft tissue features similar to that of the hindlimb [16 -18]. More recent work in mice has shown that Pitx1, expressed in the hindlimb bud mesenchyme, is also necessary for normal expression of Tbx4, a transcription factor required for normal hindlimb development [19]. Pitx1 has been found to be enriched on hindlimb cis-regulatory elements but is also strongly associated with many functionally verified limb enhancers [20]. These findings have led to the suggestion that Pitx1 influences hindlimb morphology through the activation of hindlimb-specific enhancers as well as through the hindlimb-specific modulation of enhancers that are active in both sets of limbs [20]. Additionally, in mice, it has been demonstrated that Pitx1 influences the patterning of different tissue types of the limb [17], including influencing morphogenesis of cartilaginous precursors of bone, the organization of myoblasts into muscle bundles, as well as attachment of tendon cells between bone and muscles. This role in morphogenesis is believed to be complete by stage E14.5 in mice [17], which is the stage at which toes and fingers are clearly separated and the distal-most elements of the fingers have formed.
The only study published on the role of Pitx1 in limb development of lizards, which used a micromass culture system, found that Pitx1 transcript levels were maintained in micromasses derived from A. sagrei hindlimb cells compared to those in forelimb cells [8]. While the authors found that the core binding sites for PITX1 were not conserved between Anolis and mammals, they did detect upregulation of the conserved hindlimb transcription factor hoxc11 through ectopic expression of PITX1 in A. sagrei forelimb cells. These recent results suggest that regulation of hoxc11 transcription through PITX1 may be mediated through binding sites that are not conserved between lizards and mammals, providing motivation to further explore the role of Pitx1 in the limb development of squamates.
Shh is known to be important in chicken and mouse for limb outgrowth and extension, and determination of digit identity and number. Shh is integrally involved in limb development and is essential for maintenance of the apical ectodermal ridge (AER), which is required for limb outgrowth and extension [13]. Shh is also required to maintain the zone of rsob.royalsocietypublishing.org Open Biol. 6: 160252 polarizing activity (ZPA) [21] which is responsible for anterior-posterior axis formation, thus ensuring that all limb elements are developed in the correct orientation. One of the few studies examining the role of Shh in limb development of lizards investigated the skink genus Hermiergis [22]. Shapiro and co-workers found that changes in the duration of Shh expression during early development resulted in variation in the number of digits. While it is assumed that the biological role of limb development genes, such as Shh, is highly conserved across species, few studies have looked at limb development genes in non-model species.
GHR encodes a transmembrane receptor that binds to growth hormone (GH) and is thought to stimulate growth, cell reproduction and regeneration. It is also involved in regulation of bone growth and is highly conserved across species [23]. Although GH is obligatory for post-natal growth, early embryonic growth has traditionally been viewed as a 'growth without GH' syndrome [24]. However, it has been shown that GH may act as an autocrine/paracrine factor during early chick embryogenesis [25]. GH coordinates the growth of multiple target tissues during development, including skeletal muscle [26], where it is highly expressed and induces the synthesis of the insulin-like growth factor I (IGF1) [27]. Both GH and IGF1 are important regulators of longitudinal growth [28,29]. GHR has been described as a Z-borne sex chromosome-linked gene in the chicken but an autosomal gene mapped to a contiguous block of chromosome 2 in P. vitticeps [30]. To this date, research into GH and GHR in reptiles is particularly limited, and their role in reptile embryonic limb development remains to be explored.

Post-oviposition limb development in Pogona vitticeps
We determined embryonic limb development stages in P. vitticeps according to the developmental stages described for Lacerta vivipara [31]. In the following text, the corresponding embryonic stages in A. sagrei are also provided [1]. As in A. sagrei, early embryogenesis in P. vitticeps occurs within the oviducts prior to oviposition, and eggs are laid at the limb-bud outgrowth and patterning stages of development (stage 28-30, Anolis stage 3-5). The seven limb development stages included in our study are illustrated in figure 1 and described in electronic supplementary material, S1. The first sampling period included in our study was 9 days postoviposition, which we determined to be the late limb-bud outgrowth and patterning stage 31 (Anolis stage 6). Morphogenesis in P. vitticeps

Post-oviposition limb growth in Pogona vitticeps
Our sampling regime in P. vitticeps allowed a detailed statistical analysis of embryonic limb growth during embryogenesis. Limb growth was measured across embryonic limb development stages in P. vitticeps for the sampling periods included in our study (table 1). A nested ANOVA comparing total length of limb type (forelimb versus hindlimb) within sampling periods showed that there is a significant difference between forelimb and hindlimb lengths across sampling periods (F 18,119 ¼ 108. 16, p , 0.001). Forelimb and hindlimb lengths were found not to differ significantly during late outgrowth and patterning (day 9 post-oviposition: F 1,4 ¼ 6.84, p ¼ 0.06) and both were shorter than head length (figure 2). Differentiation between forelimb and hindlimb lengths occurred during morphogenesis (figure 2), where hindlimb length was found to be significantly longer than forelimb length by the 12-13 d post-oviposition time period (F 1,12 ¼ 7.36, p ¼ 0.02). The increasing differentiation of forelimb and hindlimb length continued through the morphogenesis and embryonic growth phases. A regression analysis of the relative length of limb elements ( proximal, distal and autopod), with the effects of allometry removed, found that there were significant differences between forelimb and hindlimbs during embryonic development ( figure 3). A generalized linear model (GLM) incorporating the effect of LIMB*TIME found that there was a significant difference between forelimb and hindlimb during embryonic development for all limb elements (proximal: F Regarding the length of hindlimb elements in the first time period for which separate elements could be measured (20 days post-oviposition), relative proximal length was greater than that of both distal and autopod limb elements (figure 3). Then, during late morphogenesis and throughout the growth phase, the relative lengths of distal and autopod elements exceeded that of the proximal hindlimb element. In hatchling lizards, the autopod is relatively longer than the distal and then proximal hindlimb elements. A different pattern of relative growth was observed in forelimbs (figure 3), where relative length of the proximal element remained fairly constant throughout development, while the relative lengths of distal and autopod elements decreased.
We calculated the average percentage growth rate per day from one sampling period to the next (table 1) and found that growth rates in both forelimb and hindlimbs were greatest in the 25 d (embryonic stage 36) to 28 -29 d (embryonic stage 37) post-oviposition period, which corresponds to the transition from the morphogenesis to the growth phases of embryonic development. We used a t-test to compare growth rates to a null hypothesis (H 0 ) of constant growth (forelimb ¼ 13.2%/day; hindlimb ¼ 19.3%/day). For hindlimb, we could significantly reject the H 0 of a constant growth rate, indicating that growth rates vary significantly during limb , with the 28-29 d post-oviposition (embryonic stage 37) period being a significant outlier (studentized residual ¼ 6.67). We were unable to reject the H 0 of a constant growth rate in forelimbs rsob.royalsocietypublishing.org Open Biol. 6: 160252 ; however, the 28-29 d postoviposition (embryonic stage 37) period was identified as a significant outlier (studentized residual ¼ 3.15).      figure 4 (whole limb buds in early embryogenesis and proximal and distal segments in later stages) found that there was a significant difference between transcript levels of GHR between time periods (F 8,135 ¼ 3.30; p ¼ 0.002), but no significant difference between forelimbs and hindlimbs, and no correlation between limb type and time period. A Tukey's post hoc pairwise comparison test indicated a significantly higher level of GHR transcripts in 21 d hatchlings compared with that in any other time period (qPCR time period T9; p 0.02). A two-way ANOVA for the results depicted in figure 5 (autopods) found no significant difference in transcript levels of GHR between time periods or limb type. These results indicate that GHR showed highest transcript levels after hatching in forelimb and hindlimb elements (qPCR time period T9; figure 4), with a similar trend of increasing transcript levels until after hatching in the autopods (figure 5).

Expression of
A two-way ANOVA for the results shown in figure 4 (whole limb buds in early embryogenesis and proximal and distal segments in later stages) found that there was a significant difference between transcript levels of Shh between time periods (F 8,135 ¼ 10.73; p , 0.001), but no significant difference between forelimb and hindlimbs, and no correlation between limb type and time period. A Tukey's post hoc pairwise comparison test indicated that there was a significant higher level of Shh transcript levels in 20 d post-oviposition (qPCR time period T3) embryos compared with any other time period ( p , 0.001). Similarly, a two-way ANOVA for the results shown in figure 5 (autopods) found revealed a significant difference in transcript levels of Shh between time periods (F 4,50 ¼ 5.44; p ¼ 0.001), but not between forelimb and hindlimb autopods, and no correlation between limb type and time period. A Tukey's post hoc pairwise comparison test indicated that there was a significant higher level of Shh transcript levels in the autopods of 28-29 d post-oviposition (qPCR time period T5) embryos compared with later time periods ( p 0.01). These results indicate that Shh transcript levels peaked at 20 d post-oviposition during morphogenesis in both forelimb and hindlimb elements (figure 4), and at 28-29 d post-oviposition in hindlimb autopods (figure 5).

Immunohistochemistry: Pitx1 expression in Pogona vitticeps
We detected PITX1 in both the hindlimb and forelimb sections of P. vitticeps (figures 6 and 7), which parallels Pitx1 transcript

Pitx1 expression in Ctenophorus pictus
Developmental stages of C. pictus embryos at 16 and 30 days post-oviposition (figure 8) were determined to be equivalent to stage 7 and 12-13 of the Anolis staging series [1], stages 31-32 and 37 of L. vivipara [31], and 12 -13 days and 28-29 days post-oviposition in P. vitticeps, respectively (electronic supplementary material, S1). Limb measurements of these embryonic stages are provided in table 2.
Ctenophorus pictus shows strongest transcription levels of Pitx1 (RT-PCR, 30 cycles) in the hindlimb (figure 9), while P. vitticeps expresses Pitx1 (RT-PCR, 30 cycles) strongly in both the forelimbs and hindlimbs (electronic supplementary material, figure S3). Using qPCR, we found that there were high levels of Pitx1 transcripts in the hindlimbs of C. pictus during morphogenesis, but low to nil expression in the forelimbs. Transcription was significantly higher in the hindlimb compared with the forelimb in the 16-day post-oviposition embryo (t2 ¼ 11.7522, p ¼ 0.007). Similarly, transcription was significantly higher in the hindlimb Similarly, in the 30-day post-oviposition hindlimb Pitx1 transcript levels were higher than in the forelimb, but this was not significant (t2 ¼ 2.5715, p ¼ 0.124). IHC results are consistent with our qPCR results in C. pictus (figure 10). Unlike during late morphogenesis in P. vitticeps (figure 6), C. pictus was found to have significant differences in expression of Pitx1 in forelimb and hindlimb sections. In the hindlimb sections of the autopods, PITX1 expression in C. pictus was similar to that in P. vitticeps, with strong DAB staining in the mesenchyme around the digital bones and developing claws, more limited staining in the growth plates of the bone epiphysis, and absence of staining in the chondrocytes of the bone diaphysis ( figure 6).

Limb development during embryogenesis in Pogona vitticeps
We identified eight morphological stages in P. vitticeps, corresponding to distinct developmental stages in L. vivipara [31] and A. sagrei [1]. In addition, we compared staging of P. vitticeps with that of agamid embryonic series, Agama imparlearis [32] and Calotes versicolor [33], although these studies illustrated development of forelimbs rather than hindlimbs. We found strong similarities between these published staging series and development in P. vitticeps. Additionally, in the oviparous P. vitticeps, we found that early embryogenesis occurs within the oviducts prior to oviposition, and eggs are laid at the limb-bud outgrowth and patterning stage of development (stage [28][29][30]. It has been found that multiple-clutching oviparous squamates, such as P. vitticeps, retain eggs in their oviducts and lay them between embryonic stage 20 and 35 depending on the species [34]. Although we were unable to find a published embryonic staging series for P. vitticeps, there has been some comparative work relating embryonic development in this species to other squamates [35]. However, embryos were only examined up to stage 36. Our study, examining limb development past hatching, therefore provides an important foundation to comparative studies in limb development of agamid lizards. In Anolis, it has been shown that post-hatching growth trajectories for trunk-ground versus trunk-crown habitat specialists are consistently the result of changes that occur prior to hatching [7], and differences in limb length are apparent at hatching, with limb long bones elongating in parallel relative to body size in different species after hatching. As such, species-specific morphologies are the result of changes that occur very early in limb development, prior to formation of the cartilaginous anlagen [7]. We found that the greatest phase of embryonic limb growth in P. vitticeps, as a proportion over time, is during the transition from morphogenesis to the growth stage of limb development (stage 36 -37) for both forelimbs and hindlimbs. We also found that onset of differentiation between forelimb and hindlimb length occurred at 20 days post-oviposition, which corresponds to stage 35. At this stage of development, we detected significantly elevated transcript levels of Pitx1 and Shh. However, whether these pre-hatching patterns of limb development in P. vitticeps define the time point when interspecific variation in limb length arises in agamid lizards is yet to be determined. Thus, our study provides a foundation for future comparative work in Agamidae. rsob.royalsocietypublishing.org Open Biol. 6: 160252

Gene expression
This study confirmed for the first time that key developmental genes are transcribed (Pitx1, GHR and Shh) and expressed (Pitx1) in limbs of the embryos and hatchlings of P. vitticeps, as would be expected based on previous studies on model organisms. In addition, we here documented expression of Pitx1, a hindlimb-specific transcription factor, in the forelimb during morphogenesis. Transcript levels of Pitx1 in the forelimb of P. vitticeps were less abundant than in the hindlimb during early morphogenesis, based on qPCR, and transcript levels were generally less abundant in late morphogenesis but did not differ between the forelimbs and hindlimbs. Expression of Pitx1 during early morphogenesis was concentrated around mesenchymal condensation, and during late morphogenesis in the mesenchyme around digital bones and developing claws, with more limited staining in the growth plates of the bone epiphysis. By contrast, we found significantly lower or no transcription and expression of Pitx1 in the forelimb when compared with the hindlimb of the Australian agamid C. pictus, either in early-or late-stage morphogenesis.
Pitx1 is structurally conserved between the chicken and mouse, showing 92% similarity [14], and plays a fundamental role in limb identity in these model organisms. Additionally, in mice, it has been demonstrated that Pitx1 influences the patterning of different tissue types of the limb during morphogenesis [17], which is completed by stage E14.5. Recently, research on the role of Pitx1 in limb development of the iguanid lizard A. sagrei, using a micromass culture system, found that transcript levels of Pitx1 were maintained in micromasses derived from hindlimb cells compared with those from forelimb cells [8]. By contrast, we here report Pitx1 expression by qPCR and IHC in the forelimb of P. vitticeps during morphogenesis. Expression of Pitx1 in the forelimb has been seen in a whole-mount in situ hybridization of a chick [16]. However, expression was spatially restricted to a small area of the autopod and only found in very early stages. Additionally, Pitx1 expression has been detected at later morphogenic stages in the forelimb of the frog Eleutherodactylus coqui, but not at early limb-bud stages [36]. Authors hypothesized that the forelimb expression of Pitx1 was, in this case, a developmental process peculiar to an organism in which metamorphic and embryonic stages are modified [36], as this species bypasses the tadpole stage and develops limbs within its eggs. With the addition of our results, with Pitx1 expression in the forelimbs during morphogenesis in P. vitticeps, we highlight the need for further research across a wider range of tetrapods needed to more fully understand evolutionary variation in molecular processes underlying limb morphology.
Although Shh and GHR transcript levels did not differ between forelimbs and hindlimbs, we did find that they varied at different stages during embryogenesis. Shh showed greatest levels of transcripts at 20 days postoviposition in proximal and distal limb elements, which equates to morphogenesis (electronic supplementary material, S1). In this stage, the cartilaginous anlagen of the limb bones form the distal tips of digits that are free from digital webbing; digit 4 is notably longer than other digits, the limbs are flexed 908 caudally at elbows, and the digit joints are not yet obvious. This stage of development also corresponds with the period when forelimbs and hindlimbs start to differentiate in length (figure 2). Shh activity is required for the maintenance of growth and patterning of intermediate and distal limb structures. Mice that lack Shh have limbs [37,38], but they are reduced and the skeletal pattern is severely truncated. In the autopods of P. vittceps, Shh showed a peak of transcription at 28 -29 days post-oviposition, which defines the transition from morphogenesis to growth phase and is characterized by significant elongation of all limb elements, including digits. It is known that Shh is involved in autopod patterning [22,39], by acting as both a morphogen as well as a mitogen [40,41]. However, digit identity and patterning has already occurred by 28-29 days post-oviposition. It is possible that Shh is playing a role in extension during this period of embryogenesis. Shh has been shown to be involved in limb muscle formation, Table 2. Morphological measurements for the three developmental stages of Ctenophorus pictus. Values presented are mean length (mm + s.e.) taken from the right limb. Images provided in figure 8. rsob.royalsocietypublishing.org Open Biol. 6: 160252 particularly to regulate directional muscle cell migration in the distal limb elements [42].
In contrast with the other two genes examined, GHR transcripts were most abundant in the limbs of hatchling lizards, in proximal and distal segments. GH coordinates the growth of multiple target tissues during development, including skeletal muscle, and it is believed that GHR exerts specific and direct effects on skeletal muscles [26]. GHR is highly expressed in skeletal muscle, where it induces the synthesis of IGF1 [27]. Both GH and IGF1 are important regulators of longitudinal growth [28]. Our results suggest that posthatching is an important stage in limb growth, especially in the more distal elements. A recent study on bone growth in the brown bat Eptesicus fuscus also indicated that post-natal rsob.royalsocietypublishing.org Open Biol. 6: 160252 development was an important stage for growth [43].
These results indicate that the processes that mediate limb growth could occur at a range of developmental stages and that despite developmental biology focusing on early development, in the case of the agamids, the molecular processes mediating hindlimb length could also occur post-hatching.

Conclusion
In conclusion, ours is the first study to demonstrate the transcription (Pitx1, GHR and Shh) and expression (Pitx1) of key limb developmental genes in the agamids. Our study analysed different limb regions across developmental stages and demonstrated that transcripts of the three genes are detectable in both distal and proximal limb regions as well as in the autopod across a wide time frame. Most studies on limb development focus on early embryogenesis. We extended our observations to include hatchlings and showed that Pitx1, Shh and, in particular, GHR continue to be transcribed, with their gene products presumably playing a role in the developing limbs of hatchling lizards. The significant increase in hindlimb length relative to body size between the 25 and 28/29-day post-oviposition stage is further evidence that these later stages of development (embryogenesis) are very important in terms of hindlimb developmental growth. Our results highlight the need for further research in a wide range of tetrapods to fully understand the role of key developmental genes in the evolution of morphological diversity, such as the role of Pitx1 in hindlimb determination and morphology. Our study on P. vitticeps and C. pictus, combined with the recently published P. vitticeps genome [12], illustrates that the Australian agamids can be further used to examine gene expression within and between species, and thus identify mechanisms through which diversity in limb length and morphology could have evolved.

Laboratory animal care
Pogona vitticeps (6 females and 3 males) and C. pictus (10 males and 18 females) were collected at the start of the mating season (September) from the Murray Sunset National Park, Victoria (34850 0 S, 141840 0 E), to produce eggs and hatchlings for this study. Lizards were housed in appropriate laboratory conditions outlined by Uller et al. [44,45]. A light and heat/basking source was provided for each cage, including a 50 W lamp (8.5 h photoperiod) and UV lights (12 h photoperiod). Temperatures in each cage varied between 308C and 408C, which is consistent with spring/summer temperatures in the Murray Sunset National Park. Lizards were fed crickets ad libitum, dusted with calcium and vitamin supplements (Repcal, Los Gatos, California), three times per week. Lizards were sprayed daily with water, in addition, provided with a water dish, and a moistened sand patch was provided for egg laying. Cages were checked daily for eggs, which were then transferred to individual containers half-filled with moist autoclaved vermiculite (1 : 7 autoclaved water: vermiculite). Containers were placed in a Thermoline incubator (Coburg North, Victoria) at 288C.

Collection of embryonic and hatchling limb tissues
Limbs were collected from developing embryos and hatchling lizards to examine gene expression during limb development. Embryos were killed using 0.01 ml of 11.1 mg ml 21 sodium pentobarbitone injected into the egg and hatchlings were killed with an IP injection of 0.1 ml of 32.5 mg ml 21 sodium pentobarbitone. All embryos and hatchlings were photographed with a digital 13-megapixel Canon camera (EOSD5) operated by a computerized system, using DIGITAL PHOTO PROFESSIONAL. Stages of embryonic development were estimated using published embryonic staging in L. vivipara [31] and A. sagrei [1]. In addition, we compared staging of P. vitticeps with that of agamid Ag. imparlearis [32]. Limb length was measured from the digital photographs at each of the developmental stages to quantify limb growth during development, using IMAGEJ (v. 1.38, NIH, USA) to the nearest 0.001 cm. All statistical analyses were undertaken using SYSTAT v. 13 (Cranes Software International). Tissues were collected differently for each developmental stage due to variation in limb size: (i) 12 -20-day postoviposition embryos-whole limb buds were removed in one piece; (ii) 25 -29-day post-oviposition embryos-limbs were removed in two sections (autopods (foot/hand) and limbs); and (iii) 44-day post-oviposition embryos to hatchling lizards. Limbs were removed in three sections: proximal (femur/humerus), distal (tibia and fibula/radius and ulna) and autopods. Limbs were collected and placed immediately on dry ice, then stored in 2808C.

RNA extraction and cDNA synthesis
RNA was extracted from frozen tissue samples (20 -100 mg), homogenized in TRI Reagent (Ambion, Applied Biosystems, Scoresby, VIC) according to the manufacturer's instructions with the Wig-L-Bug crescent shaker (Densply Rinn), with minor modifications [46]. The resulting RNA pellets were washed in 75% ethanol, air dried and resuspended in RNA-Secure water (Ambion). RNA samples were then treated with DNAse I (Ambion) at 378C for 20 min to remove residual genomic DNA. RNA concentrations were measured on the NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA; Biolab, Scoresby, VIC) with an A260 : A280 ratio of more than 1.9. The presence of 18S and 28S ribosomal RNA was confirmed by gel electrophoresis on 1.2% TBE agarose gels. First strand cDNA synthesis used 1 mg total RNA in 30 ml reactions, with Superscript III reverse transcriptase (200 U ml 21 ; Invitrogen) and oligo d(T) (50 mmol, Applied Biosystems) according to the manufacturer's instructions. Samples were incubated at 258C for 10 min, 508C for 50 min, 858C for 5 min to terminate the reaction and then stored at 2208C.

RT-PCR
We first established if Pitx1, GHR and Shh were transcribed in the autopods, forelimbs and hindlimbs of P. vitticeps embryos and hatchlings during development. Oligonucleotide primers (electronic supplementary material, table S4) for the target rsob.royalsocietypublishing.org Open Biol. 6: 160252 genes were designed from published chicken sequences (GenBank accession numbers: XM_414626.2, AB012236.1 and NM_204821.1). We sequenced the target genes for the study species and a range of other lizards (in the case of Pitx1 and SHH); alignment files are available in the Figshare digital repository and can be accessed at https://figshare. com/s/9e497ba0790fcb481d0b. RT-PCR was performed on 44-day post-oviposition embryos (n ¼ 2) and 7-day-old hatchlings (n ¼ 2) in 20 ml reactions with GoTaq Flexi (Promega, Annandale, NSW), 25 mM MgCl 2, 100 ng ml 21 forward and reverse oligonucleotide primers (Sigma Aldrich, Castle Hills, NSW) and 1 ml cDNA. A negative control, using water instead of cDNA, was included in each RT-PCR. The RT-PCR programme used for all samples consisted of 808C for 2 min, then 40 cycles of 958C for 30 s, 558C for 1 min and 728C for 1 min, with a final extension step at 728C for 10 min. PCR products (12 ml) were visualized using gel electrophoresis on 1.2% TBE agarose gels with 6 ml SYBR safe stain. Hyperladder IV (Bioline Pty Ltd., Eveleigh, NSW) was included on each gel to estimate size of RT-PCR products. Amplification of the correct gene was confirmed by sequence analysis. The DNA was purified using ExoSAP (GE Healthcare, Ryldamere, NSW), according to the manufacturer's instructions, and sequenced by Macrogen (Korea). A BLAST search (http://www.ncbi.nlm.nih.gov/blast/ Blast.cgi) confirmed that the correct gene had been amplified.

Real-time quantitative PCR (qPCR)
Quantification of Pitx1, GHR and Shh transcript levels in the forelimb and hindlimb of P. vitticeps embryos and hatchlings were measured at each of the developmental stages by qPCR. TaqMan labelled fluorogenic probes with a quencher dye TAMRA (6-carboxyl-tetramethyl-rhodamine) at the 3 0 end and a FAM (6-carboxy fluorescent) reporter dye at the 5 0 end (BioSearch Technologies Inc, CA) and primers were designed from the sequenced PCR products above using REALTIMEDESIGN software (BioSearch) (electronic supplementary material, table S4). Alignment files, including qPCR primers and probes, are available in the Figshare digital repository and can be accessed at https://figshare. com/s/9e497ba0790fcb481d0b, demonstrating the 100% specificity in the study species. Samples were analysed using the relative comparative C T method according to the Stratagene MxPro Protocol (http://www.scribd.com/doc/ 60305516/84/Comparative-Quantitation-Data-Analysis). In relative quantification, the qPCR data are presented relative to another gene, often referred to as an internal control. All qPCR reactions were carried out in triplicate using 96-well optical reaction plates (Bio-Rad Laboratories, Gladesville, NSW) in 20 ml volumes consisting of 1 ml cDNA, 2Â SensiMix dU (Quantace, Alexandria, NSW), 0.8 ml primers (20 uM) and 0.4 ml probe (20 uM) using the Mx3000P qPCR System from Stratagene (Agilent Technologies, Mulgrave, Victoria).
The qPCR experiment was designed to quantify gene expression between all limb elements and different embryonic stages in the hindlimb and forelimb in two separate experiments due to restrictions on plate size. Six individuals (n ¼ 6) for each time stage were included in the experiments. Experiment 1: whole forelimb and hindlimb buds for post-oviposition 12 -13 d, 15 -18 d, 20 d and 25 d forelimbs; whole limb without autopod for 25 d hindlimbs and 28 -29 d forelimbs and hindlimbs; separate forelimb and hindlimb proximal and distal sections (without autopods) for postoviposition 44 d, 53 -55 d embryos; separate forelimb and hindlimb proximal and distal sections (without autopods) for hatchlings 7 dH and 21 dH. Each plate contained the above listed series of hind and forelimb buds, proximal and distal sections. Experiment 2: autopods from forelimb (except 25 d) and hindlimbs of post-oviposition 28 -29 d, 44 d, 53 -55 d embryos and hatchlings 7 dH and 21 dH. Each plate contained the above listed series of hind and forelimb autopods. The genes of interest (GOI) and the endogenous reference gene or internal control (r28S) were assessed in separate qPCR plates and the C T values of each gene of interest were normalized to r28S C T values. The relative difference in expression was calculated using 12-13 d HL (earliest time point) for experiment 1 and 25 d (earliest time point) for experiment 2 as the calibrator. This means that the relative quantity of the calibrator is automatically defined as 1.0 on the graph. The formula used to calculate fold change is

Immunohistochemistry
To detect Pitx1 expression in various stages of the developing embryonic limbs, the following time points were collected with n ¼ 5 per time point: 15 -18-day post-oviposition forelimb and hindlimb buds; 28 -29-day post-oviposition forelimb and hindlimb limbs and autopods; 53-55-day post-oviposition hind and forelimb proximal, distal and autopod sections; and 7-day hatchling hind and forelimb proximal, distal and autopod sections. The limbs and buds were embedded in paraffin, and transverse sections were cut and mounted on SuperFrost slides. Two sections from the same individual were mounted on the same slide, one for application of the primary antibody (Pitx1) and one as a negative control. Slides were de-waxed with a series of histolene, xylene and ethanol washes followed by antigen retrieval using Tris/EDTA/EGTA ( pH9.0). A peroxidase block (30% hydrogen peroxide and 1ÂTBS) was used to block endogenous peroxidase for an hour followed by two 1ÂTBS washes. The sections on the slide were circled with a PAP pen (Abcam), which provides a hydrophobic barrier around the specimen. The sections were then blocked with a universal background Sniper (Biocare Medical) for 30 min with the excess blotted away. The primary polyclonal antibody Pitx1 (Bioworld Technologies) and a negative control (rabbit IgG) were added onto the sections at a 1 : 300 dilution (in PBS with BSA) each and incubated overnight at 48C in a humidifying chamber. The secondary antibody, MACH4 Universal HRP (Biocare Medical), was added on the following day for 30 min after a series of TTBS washes. The slides were blotted and incubated with DAB (Vector Laboratories) for 2-10 min. Slides were viewed to check for brown colour development, then washed with distilled water. This was rsob.royalsocietypublishing.org Open Biol. 6: 160252 immediately followed by a DAB enhancer step, and then each slide was incubated for 10 min in 0.05 M sodium bicarbonate. The slides were then washed with MilliQ water, stained with haematoxylin, dehydrated through a series of ethanol and histolene washes, and then mounted in Cytoseal (ThemoScientific) and dried at 408C overnight.

Pitx1 expression in Ctenophorus pictus
To investigate whether the pattern of Pitx1 expression in P. vitticeps is typical across Australian agamid lizards, we selected a species, C. pictus, from a closely related genus of agamids [48]. We compared Pitx1 transcription in the forelimbs and hindlimbs of C. pictus and P. vitticeps embryos using RT-PCR, which was undertaken at 16 days and 30 days post-oviposition for C. pictus, and 20 days, 29 days and 44 days post-oviposition for P. vitticeps, using the same protocol detailed above. In addition, the RT-PCR was repeated using 30 cycles to reduce the intensity of the PCR product.
We subsequently undertook qPCR of Pitx1 at development stages of limb morphogenesis, one in early morphogenesis (16 days post-oviposition) and one in late morphogenesis (30 days post-oviposition), using the methods, primers and probes detailed above for P. vitticeps. All qPCR reactions were carried out in triplicate using 96-well optical reaction plates (Bio-Rad Laboratories, Gladesville, NSW). Four individuals (n ¼ 4) for each time stage were included in the experiments. Samples were then analysed using the relative C T standard curve method according to the Applied Biosystems User Bulletin #2 (http://www3. appliedbiosystems.com). A serial dilution, using distal hatchling limb tissue, was used to generate standard curves for Pitx1 (50 ng ml 21 , 10 ng ml 21 , 5 ng ml 21 , 1 ng ml 21 , 0.5 ng ml 21 ) and the housekeeping gene (5 ng ml 21 , 1 ng ml 21 , 0.5 ng m 21 , l 0.1 ng ml 21 , 0.05 ng ml 21 ). The known initial concentration of RNA was plotted against the C T value: the cycle at which the intensity of fluorescence, indicative of the amount of PCR product, crosses an arbitrary threshold, on a logarithmic scale. 33 ng ml 21 of cDNA was used for the analysis of gene of interest and 2 ng ml 21 for the housekeeping gene, r28S. All PCR reactions were carried out in triplicate using 96-well optical reaction plates (Bio-Rad Laboratories, Gladesville, NSW) in 20ml volumes consisting of 1 ml cDNA, 2Â SensiMix dU (Quantace, Alexandria, NSW), 0.8 ml primers (20 mM) and 0.4 ml probe (20 uM) using the DNA Engine Opticon 2 System (MJ Research; Bio-Rad Laboratories). The experimental design to quantify differences in Pitx1 expression between the forelimb and hindlimb contained tissue samples (16dHL, 16dFL, 30dHL, 30dFL, 30dHA and 30dFA) on one plate. Pitx1 and the endogenous reference gene (r28S) were assessed in separate PCRs and the gene of interest C T values were normalized to r28S C T values. The relative difference in expression was calculated using 30HL as the calibrator.
IHC assays in C. pictus were undertaken at the time point (30 days post-oviposition), where there was least difference in Pitx1 transcription between forelimbs and hindlimbs in P. vitticeps. We used the sample protocols in C. pictus as those in P. vitticeps, as detailed above, with n ¼ 4 individuals.
Ethics. Animals were collected under permits issued by the Victorian Department of Sustainability and Environment and Parks Victoria (no. 10004214). All experimental procedures were approved by the Museum of Victoria Ethics Committee (MVAEC20070501).
Data accessibility. Primer and probe sequences, additional IHC results, and RT-PCR results can be found in the electronic supplementary material. The sequence alignment files supporting the results of this article are available in the Figshare digital repository and can be accessed at https://figshare.com/s/9e497ba0790fcb481d0b.