Digital gene expression tag profiling of bat digits provides robust candidates contributing to wing formation
© Wang et al. 2010
Received: 5 February 2010
Accepted: 6 November 2010
Published: 6 November 2010
Skip to main content
© Wang et al. 2010
Received: 5 February 2010
Accepted: 6 November 2010
Published: 6 November 2010
As the only truly flying mammals, bats use their unique wing - consisting of four elongated digits (digits II-V) connected by membranes - to power their flight. In addition to the elongated digits II-V, the forelimb contains one shorter digit (digit I) that is morphologically similar to the hindlimb digits. Here, we capitalized on the morphological variation among the bat forelimb digits to investigate the molecular mechanisms underlying digit elongation and wing formation. Using next generation sequencing technology, we performed digital gene expression tag profiling (DGE-tag profiling) of developing digits in a pooled sample of two Myotis ricketti and validated our sequencing results using real-time quantitative PCR (RT-qPCR) of gene expression in the developing digits of two Hipposideros armiger.
Among hundreds of genes exhibiting significant differences in expression between the short and long digits, we highlight 14 genes most related to digit elongation. These genes include two Tbx genes (Tbx3 and Tbx15), five BMP pathway genes (Bmp3, RGMB, Smad1, Smad4 and Nog), four Homeobox genes (Hoxd8, Hoxd9, Hoxa1 and Satb1), and three other genes (Twist1, Tmeff2 and Enpp2) related to digit malformations or cell proliferation. In addition, our results suggest that Tbx4 and Pitx2 contribute to the morphological similarity and five genes (Acta1, Tnnc2, Atp2a1, Hrc and Myoz1) contribute to the functional similarity between the thumb and hindlimb digits.
Results of this study not only implicate many developmental genes as robust candidates underlying digit elongation and wing formation in bats, but also provide a better understanding of the genes involved in autopodial development in general.
Elongation of the posterior digits (digits II-V) in the bat forelimb is initiated at an early stage when digit anlagen initially appear, embryonic Stage 15 . By embryonic Stage 17, the elongation becomes obvious with lengthening of forelimb digits II-V greatly exceeding that of the thumb and hindlimb digits [3, 4]. This rapid elongation of the posterior forelimb digits is thought to results from modification in expression of key developmental genes [5, 6]. In fact, several experimental and empirical studies have found differences in expression of limb and digit associated genes between mouse and bat autopodia. First, a mouse transgenic experiment that replaced an enhancer of the mouse Prx1 - a transcription factor that regulates long bone growth in mouse - with the orthologous putative enhancer sequence from a bat resulted first in an increase in Prx1 expression and second in a small, but significant elongation of the mouse forelimb digits [7, 8]. Second, Shh, a key factor in limb development, exhibits a unique second phase of signaling late in development of the bat wing that is not present in the development mouse autopodia . Next, other developmental genes including Fgf8 and Bmp2/4 are up regulated in bat forelimb digits compared to bat hindlimb or mouse digits [10, 11]. Finally, the distribution Hoxd13 expression is more posteriorly restricted in the bat forelimb relative to mouse fore- and hindlimbs and bat hindlimbs during late embryonic stages [12, 13]. While these results indicate that differences in expression of Hoxd13, Fgf8, Bmp2/4 and Prx1 during forelimb digit development may contribute to bat digit elongation, the complex developmental changes required to generate the bat wing likely require numerous molecular changes . Thus, further examinations of gene expression during development of the bat digits are critical to identify the genes associated with digit elongation in the posterior forelimb digits and those responsible for the morphological and functional similarities between the short thumb and hindlimb digits.
Here, using a DGE-tag profiling approach, we compared patterns of gene expression among three cDNA libraries formed from distinct regions of the fetal autopodia (the thumb, the four posterior forelimb digits and the all five hindlimb digits) in Rickett's big-footed bat, Myotis ricketti. The method of DGE-tag profiling combines the SAGE principle with the massively parallel sequencing technology to study genome-wide gene expression profiles . DGE-tag profiling has three main advantages compared to the conventional microarray technology. First, it is high-throughput sequencing. As a result, data from a single lane of a flow cell (consisting of eight lanes allowing for eight samples) is enough to analyze of genome wide gene expression including weakly expressed genes which cannot be assessed using microarray [15–17]. Next, this type of sequencing data is highly replicable and accurate such that technical replicates have little variation . Finally, sequencing can be performed without a priori sequence information, which is required in microarrays . These features of DGE-tag profiling aided our identification of genes associated with digit morphology in bats. Overall, our examination of differentially expressed genes revealed candidate genes associated with both digit elongation as well as morphological and functional similarities between the thumb and hindlimb digits. Finally, our results revealed genes associated with digit position in both limbs.
Specimen id and measurements
13 Feb 2008
28 May 2008
13 Feb 2008
6 June 2008
24 April 2009
20 May 2009
24 April 2009
20 May 2009
DEGseq analysis of marker genes as well as those highlighted in this study related to digit development
Tag raw counts
Hand DII-V vs. Hand DI
Hand DII-V vs. Foot
Hand DI vs. Foot
Thumb & Foot
Among the differentially expressed genes, we found 14 genes likely related to digit elongation based on results of previous studies (see Discussion). The expression of Tbx3, Tbx15, Bmp3, Rgmb, Smad1, Smad4, Hoxd8, Hoxd9, Satb1, Hoxa1, Twist1, Tmeff2 and Enpp2 were significantly higher in Library Hand DII-V compared with other two libraries, and the expression of Nog was significantly lower in Library Hand DII-V (Table 2). Moreover, our results hightlighted 7 genes were likely related to the morphological and functional similarity between the thumb and the hindlimb digits. The levels of Tbx4, Pitx2, Acta1, Tnnc2, Atp2a1, Hrc and Myoz1 were significantly higher in Libraries Hand DI and Foot than in Library Hand DII-V (Table 2).
Massively parallel sequencing is a powerful technology enabling extensive investigation of gene expression profiles. In this study, we obtained over 16.5 million sequences of expressed tags from three bat autopodial libraries, including the thumb (Hand DI), the posterior forelimb digits (Hand DII-V), and all five hindlimb digits (Foot). Using these data, we compared gene expression profiles among the libraries to identify candidate genes associated with elongation of the posterior forelimb digits (Hand DII-V). In addition, we examined gene expression patterns associated with the short digit morphology of the thumb and hindlimb digits as well as gene expression patterns associated with the fore- versus the hindlimb digits. The results of DGE-tag profiling were validated by examining gene expression in a subset of genes using RT-qPCR.
To assess the reliability of our gene expression profiles, we first examined expression of several genes with previously established expression patterns. We compared expression of Tbx5 and Pitx1, fore- and hindlimb marker genes, in the forelimb and hindlimb digits (Libraries Hand DI and Hand DII-V and Library Foot). Consistent with previous studies, we found that expression of Tbx5 was significantly higher in the forelimb digits and Pitx1 expression was significantly higher in the hindlimb digits although the embryonic stages used here are much later than those used in previous studies [19, 20]. In addition, we compared expression of Meis2 and Hoxd10 in the fore- and hindlimb digits. Previously published microarray data revealed high levels of expression of these two genes in bat autopodia at Stages 16 and 17 . We found that expression of both genes, Meis2 and Hoxd10, remained significantly higher in Library Hand DII-V in the Fetal Stage (Table 2). Finally, Krt17, a gene expressed in nails and important for the development of nails, exhibited significantly higher expression in the thumb and hindlimb digits . Because only the thumb (Hand DI) and the hindlimb digits have claws, we would expect higher expression of Krt17 in the thumb and hindlimb digits relative to the posterior forelimb digits (Hand DII-V) that make up the wing.
In addition to comparing our gene expression profiles to genes with known expression patterns, we further validated our results by performing RT-qPCR using 14 genes which were identified as being differentially expressed between elongated digits and short digits in DGE-tag profiling. With the goal of identifying genes related to wing formation across bat taxa, we collected samples from a different species which belongs to the other suborder of Chiroptera (H. armiger). We selected 14 genes likely important for wing formation or for the similarity between the thumb and foot digits (see below). All the 14 genes exhibited the same overall expression pattern in the RT-qPCR as in the DGE-tag profiling (Figure 5). This finding not only provides strong support for the DGE-tag profiling method used in this study, but also maintenance of gene expression patterns across these two phylogenetically distant species provides strong support for the role of these genes in digit elongation in the posterior forelimb digits or morphological and functional similarities between the thumb and hindlimb digits.
Overall, comparisons of gene expression profiles between digit morphologies and limbs identified hundreds of differentially expressed genes. Several interesting patterns have emerged from this data. Specifically, we highlight 21 genes likely related to wing formation or to morphological and functional similarities between thumb and hindlimb digits (Table 2). First, we found 14 genes that are likely associated with digit elongation in bats - two Tbx genes (Tbx3 and Tbx15), five genes from the BMP pathway (Bmp3, Rgmb, Smad1, Smad4 and Nog), four Homeobox genes (Hoxd8, Hoxd9, Satb1 and Hoxa1) and three other gene (Twist1, Tmeff2 and Enpp2) related to either digit malformation or cell proliferation. Next, we identified seven genes (Tbx4, Pitx2, Acta1, Tnnc2, Atp2a1, Hrc and Myoz1) that are likely associated with the morphological and functional similarities between the thumb and hindlimb digits. Expression patterns of these genes in the three distinct digit libraries as well as known function of these genes are described in detail below.
In general, Tbx3 and Tbx15 belong to T-box gene family and are involved in skeletogenesis [23, 24]. In digit formation specifically, Tbx3 is thought to modulate posterior digit identity through Shh and BMP signaling, and Tbx15 insufficiency causes Cousin Syndrome which includes congenital dwarfism, moderate brachydactyly (or shortening of the digits) and leg shortening [25–27]. Mutants of Tbx15 have reduced bone size and changes in bone shape in the forelimb skeleton indicating an important role in the skeletal development of forelimbs .
Bmp3, Rgmb, Smad1, Smad4 and Nog are all components of the BMP pathway, a pathway critical for bone formation . Bmp3 belongs to the transforming growth factor-beta (TGFβ) superfamily and stimulates cell proliferation and differentiation in a concentration dependent fashion . Overexpression of Bmp3 in the chick wing bud has be shown to result in an increase in cell proliferation resulting in lengthening of skeletal elements . Rgmb, also named as Dragon, enhances BMP signaling by directly binding to BMP proteins including Bmp2 . Overexpression of Bmp2 and Bmp4 in the developing chick limb leads to a dramatic increase in the volume of cartilage elements . In addition, Bmp2/4 has been shown to be up regulated in the bat forelimb embryonic digits relative to both mouse forelimb digits and bat hindlimb digits . This up-regulation is correlated with increased rate of cartilage proliferation . Beyond up-regulation of BMPs themselves, Smad1 mediates BMP signaling. Specifically, in its phosphorylated state Smad1 forms a complex with Smad4 and then accumulates in the nucleus. This SMAD complex regulates transcription of several target genes . Embryonic manipulation of Smad1 has confirmed that its expression at the phalanx-forming region influences digit identity in the chick limb . In addition, cartilage-specific knockout of Smad1 and Smad4 reduce chondrocyte proliferation and increase apoptosis resulting in defects of limb elements in the mouse [34, 35]. While all the components of the BMP pathway mentioned above stimulate cartilage formation, Nog is an antagonist of BMPs and as such inhibits cartilage formation . Mutations of Nog disturb the balance of BMP signaling leading to brachydactyly . Thus, our results - exhibiting up regulation of Bmp3, Rgmb, Smad1, Smad4 and down regulation of Nog in the elongated bat forelimb digits - suggest a role for BMP signaling in the lengthening of the posterior bat forelimb digits.
Hoxd8, Hoxd9, Hoxa1 and Satb1 are homeobox genes which specify the anterior-posterior axis and have been implicated as digit identity determining genes . Although Hoxd8 and Hoxd9 mainly contribute to proximal limb development and are only mildly expressed in the developing digits, we found that Hoxd8 and Hoxd9 were highly expressed in bat wing digits II-V relative to the thumb and hindlimb digits . Our results also indicate that the homeobox gene Satb1 and Hoxa1 were highly expressed in bat wing digits II-V relative to the thumb and hindlimb digits. These two genes are highly expressed in cancer cells promoting tumor growth [40, 41]. To our knowledge, there are no studies examining expression of Satb1 and Hoxa1 in limb development making them interesting genes for future investigation.
Similarly to Satb1 and Hoxa1, Tmeff2 and Enpp2 are up regulated in carcinomas [42, 43]. Previous study showed Tmeff2 contributes to cell proliferation. Enpp2 which is expressed in precartilaginous condensations can be strongly activated by Hoxa13 and Hoxd13 which are crucial factors for digit identity and development . Finally, knowledge of the role of Twist1 and the digit development comes from pathological studies . Mutations Twist1 have been found in patients with Saethre-Chotzen syndrome which is associated with digit malformation . Overexpression of Twist1 inhibits osteoblast differentiation and prevents premature fusion of skeletons .
In addition to genes related to bat digit elongation, we found several genes that may contribute to the morphological and functional similarities between the bat thumb and hindlimb digits. First, Tbx4 and Pitx2 are primarily associated with hindlimb development. Tbx4 has also been shown to regulate bone size. Specifically, in the developing mouse forelimb, overexpression of these genes results in shortening of the forelimb skeletons [47, 48]. Next, we found several genes that are likely associated with the distinct musculature and function of the short digits. Because the clawed thumb of bats is used to cling to the cave while roosting and the clawed hindlimb digits are used for hanging upside down these digits should be richer in skeletal muscles than the elongated digits. In fact, expression of several skeletal muscle associated genes - Acta1, Tnnc2, Atp2a1, Hrc and Myoz1 - was significantly higher in the thumb and hindlimb digits then in the elongated posterior forelimb digits .
Using genome-wide sequencing method we have identified many genes differentially expressed in the developing digits of the bat autopodia. Several of these genes are highly related to digit formation and thus likely play important roles in the development of bat wing. Our results provide robust candidate genes for future functional tests and molecular evolution comparison with other mammals to fully understand the mechanisms of wing formation and the evolution of bat flight.
All procedures were carried out in accordance with the Policy on the Care and Use of Animals, approved by the Ethical Committee, East China Normal University. Myotis ricketti, Rickett's big-footed bat, belongs to suborder Yangochiroptera and is widely distributed in China . Their feet - which are used to forage on fishes - are larger than those of other bat species (Figure 1A) . As in many species of bats, copulation in Myotis ricketti occurs in autumn and spermatozoa are stored in the female's uterus and oviducts through the winter hibernation. Ovulation and subsequent fertilization begin in the spring with parturition occurring in summer . Because gravid M. ricketti could not be found during the summer breeding season, we capitalized on sperm storage by females and captured 10 hibernating female M. ricketti from a cave (39°42'N, 115°43'E) in Beijing on February 13, 2008. In addition to Myotis ricketti, we captured three gravid Hipposideros armiger from a cave (30°20'N, 117°50'E) in Anhui province of China on April 24, 2009 to use for quantitative real-time PCR confirmation sequencing results. H. armiger, the great leaf-nosed bat, belongs to the other suborder Yinpterochiroptera and is widely distributed in subtropical and tropical zones of Asia .
After capture, bats were kept in a large cage (1.2 × 1 × 1 m) covered with wire netting allowing them to hang. The cage was kept in a dark, temperature controlled room. The temperature was maintained between 18 and 24°C. Water was freely available and meal worms mixed with a powdered multivitamin and calcium tablets were provided daily.
M. ricketti in the captive colony were euthanized by decapitation on May 28 and June 6, 2008 (Table 1). H. armiger were euthanized on May 20, 2009 (Table 1). Fetuses from the gravid females were put into ice cold PBS. The thumb, the remaining forelimb digits (metacarpi, phalanges and dactylopatagium) and all five hindlimb digits (metatarsi and phalages) were dissected from the fetuses (Figure 1B and 1C) and stored in liquid nitrogen until use. All other components of the fetus were fixed overnight in Bouin's fluid, washed with several changes of 70% ethyl alcohol and stored at room temperature. Crown rump length (CRL) and forearm length (FA) were measured for further estimation of the embryonic stages.
To reduce sex-specific gene expression patterns, we pooled the samples from a male and a female fetus of M. ricketti (Bat M1 and Bat M2 respectively, see Table 1). Tag libraries were prepared using the DGE-Tag Profiling for NlaIII Sample Prep kit from Illumina according to the manufacturer's instructions. In brief, total RNA was extracted from forelimb digit I, forelimb digits II-V and all five hindlimb digits using TRIzol® Reagent (Invitrogen, Cat. No. 15596026, US), treated with DNase I (Roche, Cat. No. 04716728001, Germany) and converted to double stranded cDNA using Oligo dT beads. Subsequently, the cDNA samples were digested using the restriction enzyme NlaIII, which recognizes and cuts the most 3' "CATG". Once digested into fragments, cDNA samples were ligated to Illumina specific adapter A, which contains a recognition site for enzyme MmeI. The Illumina specific adapter B was ligated after MmeI digestion (Illumina, Cat. No. FC-102-1005, US). Using this technique, three DGE libraries (Libraries Hand DI, Hand DII-V and Foot) from the three samples (Figure 1B and 1C) were constructed such that each molecule in the library is a 21 bp tag derived from a single transcript with Illumina specific adapters attached to both ends. Next, the 21 bp tags were enriched using PCR primers that anneal to the adaptors and clusters were created on a flow cell using a Solexa Cluster Station following the manufacturer's instructions (Illumina, Cat. No. FC-103-1004, US). Finally, massively parallel sequencing-by-synthesis was performed on the Illumina Genome Analyzer. Image analysis, base calling, extraction of 21 bp tags, and tag counting were performed using the Illumina pipeline. The number of times that a unique tag sequence is detected represents the quantitative expression level of the corresponding transcript in the tissue. The raw data, including tag sequences and counts, were deposited in NCBI Gene Expression Omnibus (GEO) database with the accession number [GEO: GSE20038]
We analyzed saturation of each of the three libraries by examining the addition of new distinct tags associated with the increase of sequencing depth (frequency = 1, > 1 and all distinct tags). Because little gene annotation exists for the bat genome, all distinct tags were aligned against the well annotated NCBI human and mouse Refseq cDNA sequences using SOAP2 [54, 55]. Taking into account the differences between species, one base pair mismatches were allowed during the mapping of tags to cDNA sequences. Tags that matched to more than one gene were not used in analyzing differential expression among the libraries. The identified cDNA sequences were mapped to NCBI Entrez GeneID. Differential gene expression analysis was performed using DEGseq software . We compared libraries Hand DII-V with Hand DI and Foot to identify differentially expressed genes between elongated digits and short digits. We also compared libraries Hand DII-V and Hand DI with Foot to identify differentially expressed genes between the forelimb and hindlimb digits.
Total RNA were extracted from forelimb digit I, forelimb digits II-V and all five hindlimb digits of two H. armiger (Bat H1 and H2, Table 1), using TRIzol® Reagent (Invitrogen, Cat. No. 15596026, US) and treated with DNase I (Roche, Cat. No. 04716728001, Germany). cDNA synthesis and qPCR was performed using the SYBR® PrimeScript® RT-PCR Kit (TaKaRa, Cat. No. DRR063A, Japan) on a Applied Biosystems 7300 Real-Time PCR System (ABI, US). RT-qPCR for each H. armiger (Bat H1 and H2, Table 1) was performed in duplicate. β-actin was taken as a reference gene and it was amplified by a pair of previously published primers . We designed 14 pairs of degenerate primers to amplify 14 target genes (Meis2, Hoxd10, Tbx3, Tbx15, Bmp3, Rgmb, Twist1, Hoxa1, Tmeff2, Enpp2, Krt17, Tbx4, Acta1 and Tnnc2). PCR products of each gene were ligated into a pMD®19-T vector (TaKaRa, Cat. No. D102B, Japan), cloned, and sequenced using the Big Dye Terminator kit on an ABI 3730 DNA sequencer (Applied Biosystems, US). Using the obtained sequences, we designed gene specific primers for each target gene for qPCR. 2-ΔΔCT method was used to quantify gene expression, data are expressed as mean ± SD . Student's t-test was performed to examine the difference of gene expression between two different samples. Sequences of forward and reverse primers for normal PCR and RT-qPCR of each gene are shown in additional file 4, Table S4. The partial CDS of the genes were deposited at GenBank under accession number HM777024 - HM777038.
Crown rump length
We thank Junpeng Zhang, Yinan Wang and Qian Yao for bat capture and care. This work was funded by a grant awarded to S. Zhang under the Key Construction Program of the National "985" Project and "211" Project.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.