Anterior shift in gene expression precedes anteriormost digit formation in amniote limbs


  • Asaka Uejima,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author
  • Takanori Amano,

    1. Mammalian Genetics Laboratory, Genetic Strains Research Center, National Institute of Genetics, 1111 Yata, Mishima, Shizuoka 411-8540
    Search for more papers by this author
  • Naoki Nomura,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author
  • Miyuki Noro,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author
    • Present address: Laboratory for Vertebrate Body Plan, Center for Developmental Biology, RIKEN Kobe, 2-2-3 Minatojima Minamimachi, Chuo-ku, Kobe 650-0047, Japan.

  • Taiji Yasue,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author
    • Present address: Laboratory for Vertebrate Body Plan, Center for Developmental Biology, RIKEN Kobe, 2-2-3 Minatojima Minamimachi, Chuo-ku, Kobe 650-0047, Japan.

  • Toshihiko Shiroishi,

    1. Mammalian Genetics Laboratory, Genetic Strains Research Center, National Institute of Genetics, 1111 Yata, Mishima, Shizuoka 411-8540
    Search for more papers by this author
  • Kunimasa Ohta,

    1. Department of Developmental Neurobiology, Graduate School of Medical Sciences, Kumamoto University, 1-1-1 Honjo, Kumamoto 860-8556, Japan
    Search for more papers by this author
  • Hitoshi Yokoyama,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author
  • Koji Tamura

    Corresponding author
    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai 980-8578
    Search for more papers by this author

  • Present address: Graduate School of Biological Sciences, Nara Institute of Science and Technology, 8916-5, Takayama, Ikoma, Nara 630-0192, Japan.

*Author to whom all correspondence should be addressed.Email:


In tetrapod limbs, an anteriormost digit has common traits of small, short and less-phalange morphology. In this study, we focused on three genes, Mkp3, Sef and Tsukushi (TSK), which have anterior-specific or anterior-prominent expression patterns in the developing limb bud at the autopod-forming stage. The anterior expression is not fixed in the period of limb development, but the expression domains of Mkp3, Sef and TSK change considerably from the distal domain to the anterior domain. This change in expression domains, anterior shift, of these genes involves maintenance of gene expression in the anterior side and downregulation in the posterior side. Manipulated overdose of fibroblast growth factor (FGF) in the presumptive digit 2 region of chick forelimb bud results in elongation of cartilage elements of digit 2, suggesting that attenuated FGF signaling, which Mkp3, Sef, and TSK negatively regulate, provides digit 2-specific traits of morphology. The anterior expression of Mkp3 and Sef but not TSK is conserved also in limb buds of the mouse and gecko, and the anterior shift of these genes, accumulation of their transcripts in the anterior side and appropriate regulation of strength of FGF signaling may control species-specific morphology of the anteriormost digit.


How homologous tissues and organs exhibit such a staggering variety of morphologies is one of the fundamental questions of developmental biology. In vertebrates, for example, teeth, digits, and vertebrae consist of repeated assemblies of modules or units, but each individual module has a different shape depending on its position. It is generally thought that during development, a generic module is initially formed, but acquires a specific identity and eventually a specific morphology based on interactions with adjacent tissues and/or other modules. Module formation and identity acquisition may, however, occur simultaneously.

In tetrapods, each digit of the limb has specific traits, such as the number and length of phalanges. In the flipper of the pantropical spotted dolphin, for example, digits 2 and 3 have six phalanges, whereas digits 1 and 5 have two phalanges, and digit 4 has four. This difference in the number of phalanges is likely due to differences in the timing of apical ectodermal ridge (AER) regression during limb development (Richardson & Oelschläger 2002). However, in mouse and chick embryos, no obvious heterochrony in AER maintenance is observed; even if such heterochrony exists, it does not seem to be the primary cause of digit morphology variation. The mechanism by which the relative position of a module is determined likely involves regional Hox gene expression (reviewed by Zakany & Duboule 2007; and see also references therein), whereas the number of phalanges appears to be regulated by bone morphogenetic protein (Bmp) and fibroblast growth factor (FGF) signaling (Dahn & Fallon 2000; Sanz-Ezquerro & Tickle 2003; Suzuki et al. 2008). Nonetheless, little is known about how each digit module acquires its specific characteristics. The zone of polarizing activity (ZPA) is thought to be responsible for specifying digit identity along the anterior-posterior (A-P) axis in the autopod of the tetrapod limb. Shh, a morphogen secreted by ZPA cells, plays an important role in digit specification and cell proliferation (Harfe et al. 2004; Towers et al. 2008). Many studies in mouse and chick embryos have investigated the role of Shh in digit specification, and several models have been proposed (reviewed by Tickle 2006; Bastida & Ros 2008). However, not all digits are specified by ZPA signals; digit 1, which has two phalanges, is able to form normally in the Shh-deficient mouse limb (Chiang et al. 2001; Kraus et al. 2001). Furthermore, cells that respond to Shh signaling in the mouse are found in the mesenchyme of digits 2 through 5, which have three phalanges, but not digit 1 (Ahn & Joyner 2004; Harfe et al. 2004). In the chick forelimb bud, additional digit 2 in the ZPA implantation is derived from the original digit 2 region, and the ZPA does not induce any ectopic region for digit 2 (Amano & Tamura 2005). These reports suggested that the anteriormost digit in mouse and chick limbs appears to arise in a ZPA/Shh-independent fashion, but little else is known about the developmental process. Understanding the mechanism of anterior digit specification and how it differs from posterior digit formation is necessary for understanding the development of limb autopod and specification of digit identity and offers a fascinating case study of identity acquisition in modules.

Here, we examined the relationship between digit specification and morphology in the chick embryo. We focused on three genes, Mkp3, Sef, and Tsukushi (TSK), all of which negatively regulate several intracellular signaling pathways, including FGF signaling. Mkp3 encodes a dual-specificity phosphatase that inhibits mitogen activating protein kinase (MAPK) activity by removing phosphates from serine and threonine residues. Mkp3 is coexpressed with Fgfs, and inhibition of FGF signaling leads to its downregulation (Eblaghie et al. 2003; Kawakami et al. 2003). Sef, which stands for “similar expression to Fgfs”, encodes a transmembrane protein that interacts with FGF receptor (FGFR) and inhibits FGF signaling (Furthauer et al. 2002; Tsang et al. 2002; Harduf et al. 2005). In humans, a cytoplasmic splice isoform of Sef also prevents phosphorylated-MAPK from being transported into nucleus (Preger et al. 2004). TSK, on the other hand, encodes a secreted protein that affects various signaling pathways. It was first reported as an antagonist of BMP signaling that forms a heterotrimer with BMP and Chordin (Ohta et al. 2004; Kuriyama et al. 2006). Other studies in Xenopus embryos showed that TSK can bind to Fgf8b, which inhibits MAPK phosphorylation, thus negatively regulating FGF signaling (Morris et al. 2007).

In this study, we investigated the spatiotemporal expression patterns of Mkp3, Sef, and TSK during autopod formation. At chick stage 25–26, when the prospective autopod region is being formed, Mkp3, Sef, and TSK showed unique patterns of gene expression, with strong expression in the region of the prospective anteriormost digit. These gene expression patterns and their sensitivity to Shh signaling are conserved among amniotes. When excess FGF protein is produced in the presumptive anteriormost digit mesenchyme, the phalanges of digit 2 showed elongation and/or hypertrophy. These findings suggest that differential sensitivity of mesenchymal cells to FGF signaling across the A-P axis, produced by localized expression of Mkp3, Sef, and TSK during digit specification, plays a role in digit morphology.

Materials and Methods

Material egg/mouse/gecko

Fertilized chick eggs were incubated at 38°C, and the embryos were staged according to Hamburger and Hamilton (1992). Mouse embryos at embryonic day (E) 10.5 and E11.5 from wild type (DDY mice, SLC), Hemimelic extra toes (Hx, Masuya et al. 1995), MFCS1+/−, and MFCS1−/− (Sagai et al. 2005) mice were collected after natural overnight matings. For gecko embryos, we used Madagascar ground gecko Paroedura pictus, and the eggs used for staining were obtained from individuals of P. pictus that were hatched and kept in our laboratory and incubated at 28°C, with embryos being staged according to Noro et al. (2009).

Cloning gecko Mkp3, Sef, TSK

We isolated fragments of P. pictus Mkp3 (706 bp) Sef (547 bp) and TSK (615 bp) from cDNA pools prepared from P. pictus embryos at 7 dpo (days postoviposition) (stages based on Noro et al., 2009). We used the following degenerate primers for polymerase chain reaction (PCR) fragment isolation: 5′-TGGAAYGARAAYACNGGNGG-3′ (forward for Mkp3), 5′-YTGNCCCATRAARTTRAARTT-3′ (reverse for Mkp3), 5′-CARGAYTTYTGYGGNTGYGA-3′ (forward for Sef), 5′-TTYTCRARCCARTCNGGYTC-3′ (reverse for Sef), 5′-GGNCCIGGNTAYACIACNYT-3′ (forward for TSK), 5′-GAIARGTYIARYTCYTGNAR-3′ (reverse for TSK).


Chick, mouse and gecko embryos were processed for whole-mount in situ hybridization as described previously (Yonei et al. 1995) using antisense RNA probes for Mkp3, Sef and TSK of the chick, mouse and gecko. Probes for Mkp3 and Sef of the chick and mouse were kindly provided by Y. Kawakami (Kawakami et al. 2003). Mouse Spry2 was kindly provided by G. Martin (Minowada et al. 1999). For analysis of the skeletal pattern, the embryos were fixed in Tyrode’s solution containing 10% formalin, stained in 0.1% alcian blue in 70% acid alcohol, dehydrated in graded alcohols, and cleared in methyl salicylate.

Manipulations on chick limb buds

For barrier insertion, a small piece of thin aluminum foil was inserted into the middle parts of stage 20–21 wing buds that had been nicked with fine-sharpened tungsten needles. For preparing graft tissue of the ZPA, limb buds at stage 21–22 were dissected from the trunk and placed in 0.1% trypsin diluted in Tyrode for 30 min at 4°C. The loosened ectoderm was removed from the underlying mesoderm. A small piece of tissue from the ZPA region was implanted into the presumptive digit 2 region (near the anterior margin of the AER) of a host stage 20–21 limb bud. The implantation site in the host wing bud was prepared by lifting the AER without removing the host tissue. The embryo was reincubated until stage 25–26 for in situ hybridization. Cylclopamine (GR-334, Biomol) was diluted in 45% 2-hydropropyl-β-cyclodextrin (HBC; Sigma) as previously described (Incardona et al. 1998). Embryos were treated with 5 μL of 1 mg/mL solution of cyclopamine three times, once every 12 h starting from stage 19–20. Affigel blue beads (Bio-Rad Laboratories, Ltd.) were used as carriers of FGFs (1 mg/mL for FGF4 and FGF8 from Genzyme/TECHNE).

For each experiment, some embryos were allowed to develop until stage 25–26 to observe gene expression, and others were allowed to develop for up to 9–10 days to observe the skeletal pattern by alcian blue staining. After cartilage staining, digital photographs for all specimens (n = 9 for control, FGF8 and FGF4) were taken with the same magnification. Area of cartilage elements of digit 2 in each digital image was estimated and compared as pixels by the software Image-J. The average value of the area was graphed in Fig. 5D.

Figure 5.

 Excess FGF8 changes the size of digit 2. (A–C) Cartilage skeletal pattern of an embryonic day 10 (E10) wing bud of a control embryo (A), FGF4-bead applied at E3.5 (B) and FGF8-bead applied at E3.5 (C). Arrow in C indicates an elongated digit 2 in the FGF8-applied limb bud. (D) Average degrees of area of digit 2 (n = 9). In all specimens, right wing buds were implanted with fibroblast growth factor (FGF)-beads (black bars, experiment), and the contralateral left wing buds were used as controls (white bars control). All specimens were photographed and analyzed by Image-J. In all cases, bars mean ± standard errors. Data were analyzed by the paired test, and the difference between FGF8-applied digit 2 and the contolateral digit 2 was found to be statistically significant (*P < 0.05).


The Mkp3, Sef, and TSK expression domains undergo an anterior shift

Although Mkp3, Sef, and TSK have been previously reported to show anteriorly biased expression in the developing limb bud, their expression patterns in the context of the A-P axis of the limb bud have not been closely examined (Eblaghie et al. 2003; Kawakami et al. 2003; Ohta et al. 2004; Harduf et al. 2005). Therefore, we performed a thorough analysis of Mkp3, Sef, and TSK gene expression in chick embryos by in situ hybridization, focusing on temporal changes in expression between stages 22 and 28.

At stage 22–23 (embryonic day 3.5, E3.5), Mkp3, Sef, and TSK are expressed uniformly along the A-P axis in the distal mesenchyme beneath the AER (Fig. 1A,D,G). However, by stage 25–26 (E4.5), all three genes were localized to the anterior distal region in partially overlapping patterns (Fig. 1B,E,H). At stage 27–28, expression in the distal mesenchyme extended more posteriorly, but still showed an anterior bias (Fig. 1C,F,I). Mario is a non-coding RNA expressed in the mesenchyme of the prospective anteriormost digit in the chick embryo (Amano & Tamura 2005; Fig. 1J–L). At stage 25–26, the mesenchymal tissue co-expressing Mkp3, Sef, and TSK corresponds to the Mario-expressing domain. Thus, Mkp3, Sef, and TSK undergo an anterior shift in spatiotemporal expression, initially marking all of the distal mesenchymal cells beneath the AER, but later becoming restricted to the cells at the anterior distal limb bud in the presumptive anteriormost digit region.

Figure 1.

 Expression patterns of Mkp3, Sef, TSK, Mario and Spry2 in chick limb bud. Chick limb buds showing expression of Mkp3 (A–C), Sef (D–F), TSK (G–I), Mario (J, K) and Spry2 (L–N) from a dorsal view. Anterior-distal specific expression domains are indicated by black arrowheads. Arrows and blank arrowheads indicate specific domains of genes in the middle and posterior, respectively. (O, P) Cartilage skeletal pattern of the autopod region in an embryonic day 10 (E10) chick forelimb (O) and hindlimb (P) with the anteriormost digit towards the top.

Because Mkp3 and Sef are known to be negative regulators of FGF signaling, we also examined expression of Spry2, another negative regulator of FGF signaling in FGF-responsive cells. Like Sef, Spry2 is expressed in the distal mesenchyme of the limb bud, but no anterior shift in expression was observed (Fig. 1L–N).

Anterior shift in other amniotes

The anterior shift in Mkp3, Sef, and TSK expression, as well as the restricted expression of these genes in the Mario-expressing domain that marks the future anteriormost digit, were evident in both the fore- and hindlimb buds of chick embryos (Fig. 1B,E,H). This suggests a correlation between the anterior shift in gene expression, the functions of these genes, and anteriormost digit properties. We next examined whether this dynamic Mkp3, Sef, and TSK expression pattern is conserved among amniote embryos. In E10.5 mouse embryos, in which the limb bud has a dome-like shape (as seen at chick stage 22–23), Mkp3 and Sef are expressed in the distal mesenchyme of the limb bud, with no asymmetry along the A-P axis (Fig. 2A,C). By E11.5, the limb bud adopts a paddle-like shape (as seen at chick stage 25–26), and Mkp3 expression is prominent in the anterior region (Fig. 2B). Sef expression is still seen in the distal mesenchyme, but is strongest in the anterior distal mesenchyme (Fig. 2D). We did not detect TSK transcript in the distal limb bud of the mouse embryo (Fig. 2E,F). Mouse Spry2 is expressed in the distal and middle mesenchyme of the limb bud, but the distal domain did not shift anteriorly (Fig. 2G,H).

Figure 2.

 Expression patterns of Mkp3, Sef and TSK in mouse and gecko limb buds. (A–H) Mouse limb buds showing expression of Mkp3 (A,B), Sef (C, D), TSK (E, F) and Spry2 (G, H). (J–O) Gecko limb buds showing expression of Mkp3 (H, I), Sef (J, K) and TSK (L, M). Anterior-distal specific expression domains are indicated by black arrowheads, and downregulated posterior domains are indicated by brackets. The arrows in G and H indicate Spry2 expression in the middle of the mouse limb bud, and the arrow in N indicates TSK expression in somites of a gecko embryo. (I,P) Skeletal pattern of the autopod region in a mouse forelimb (I) and a gecko forelimb (P) with digit 1 towards the top.

We next examined the expression pattern of these genes in another amniote, the reptile. We examined gecko embryos (Paroedura pictus) at stages 7 dpo and 12 dpo+, when the limb buds resemble those of chick embryos at stages 22–23 and 25–26, respectively (Noro et al., 2009). In the 7 dpo limb bud, Mkp3 and Sef are expressed in the distal mesenchyme, as in the chick and mouse limb buds (Fig. 2J,L). At 12 dpo+, Mkp3 was expressed strongly and broadly in the anterior limb bud, but was repressed in the posterior mesenchyme (Fig. 2K, arrowheads); there was not a complete anterior shift in gene expression. In contrast, Sef showed an anterior shift similar to that observed in chick and mouse embryos (Fig. 2M). At 7 dpo, TSK expression was not found in the limb mesenchyme, even though clear signal was detected in the somites (Fig. 2N). At 12 dpo+, TSK expression was not detected in the distal mesenchyme of the limb bud, but weak expression was observed in the proximal mesenchyme (Fig. 2O).

Effects of tissue interactions and Shh signaling on the anterior shift

Since the fate of distal posterior mesenchymal cells does not correlate with the anterior movement of posterior margin of Mkp3, Sef, and TSK expression domains (Vargesson et al. 1997), the anterior shift in the chick limb bud must include suppression of the posterior domain of gene expression as well as maintenance of anterior expression. For anterior maintenance, Mkp3 and Sef are induced and maintained by FGF signals from the AER. If the AER is removed from limb bud, expression of Mkp3 and Sef is lost (Kawakami et al. 2003; Harduf et al. 2005, and data not shown). TSK expression is also AER-dependent (data not shown).

We next investigated whether the anterior mesenchyme plays a role in posterior gene suppression. An aluminum barrier was placed in the middle of the limb bud at stage 20–21 to separate the anterior and posterior halves. It has been reported that a barrier insertion in the middle of the limb bud at the early stage would give rise to the skeletal defects in the anterior side of the barrier (Summerbell 1979). After the barrier insertion, we reincubated the embryos until stage 25–26 and examined gene expression by in situ hybridization. Expression of Mkp3, Sef, and TSK was reduced in the posterior mesenchyme regardless of barrier insertion (Fig. 3B,F,J). Mkp3 expression was totally lost from the posterior mesenchyme (Fig. 3B), whereas Sef and TSK transcripts were still detectable in the region near the barrier (Fig. 3F,J, arrowheads), indicating that barrier insertion had no effects on the expression change of these genes. Thus, the anterior shift does not appear to be influenced by the anterior mesenchyme.

Figure 3.

 Effects of surgical and chemical manipulations on the anterior shift of gene expression. (A, E, I) Expression of Mkp3 (A), Sef (E) and TSK (I) in normal wing buds at stage 25–26. (B, F, J) Expression of Mkp3, Sef and TSK at stage 25–26 after barrier insertion dividing stage 20–21 wing buds into anterior and posterior portions. Expression of each gene decreased in the posterior limb bud. Brackets indicate downregulation of gene expression. (C, G, K) Gene expression after zone of polarizing activity (ZPA) grafting to stage 20–21 host chick wing buds. Brackets indicate ectopic downregulation of the anterior domain of gene expression. (D, H, L) Gene expression after 36 h of cyclopamine treatment at stages 21–22. Blank arrowheads show ectopic posterior expression, and note that expression domain is expanded proximally as well as posteriorly (compare to Fig. 3A, E, I).

To investigate the effect of the posterior mesenchyme on the anterior shift, we transplanted posterior limb bud ZPA tissue into the anteriormost digit-forming region (based on fate mapping studies) at stage 20–21 (Vargesson et al. 1997; Amano & Tamura 2005). This type of transplantation of the ZPA is known to give rise to non-mirror-image duplications, such as 2344334, 234334 and 23434 (Summerbell & Tickel 1977; Tamura et al. 1993; and not shown). Ectopic expansion of the autopod region was observed after the ZPA transplantation. Mkp3 expression was seen in the anteriormost mesenchyme, but in a narrower range than in the control limb bud (Fig. 3C, arrowhead, compare to 3A). Sef expression was divided into two regions by ZPA transplantation (Fig. 3G). TSK expression was also divided into two regions, and the anterior region was extended more anteriorly (Fig. 3K). Expression of Mkp3, Sef, and TSK was greatly reduced in the ectopically expanded mesenchyme (indicated by brackets in Fig. 3C,G,K). The difference in the Mkp3 and Sef/TSK expression patterns (Mkp3 had no posterior domain) might reflect the difference in expression observed in the control limb bud (Sef and TSK have a wider range of expression on the anterior side). It is likely that the ZPA negatively regulates the expression of these three genes.

It is widely known that Shh protein is the main inductive signal of the ZPA. We blocked Shh signaling with cyclopamine (Nissim et al. 2006; Scherz et al. 2007), which inhibits the function of the Shh receptor Smoothened (Cooper et al. 1998; Incardona et al. 1998), to test whether Shh causes a decrease in gene expression in the posterior mesenchyme. The cyclopamine treatment from stage 20 gives rise to a loss of the posterior digit (Scherz et al., 2007; and data not shown). Thirty-six hours after cyclopamine treatment, the gene expression domains were extended posteriorly over the entire distal mesenchyme, and the anterior shift was never observed (Fig. 3D,H,L). These results suggest that Mkp3, Sef, and TSK are downregulated in the posterior mesenchyme in cells that receive Shh signaling. In other words, the region maintaining Mkp3, Sef, and TSK probably does not receive sufficient Shh signaling.

To further investigate the relationship between Shh signaling and the anterior shift, we used two mouse mutants, Hemimelic extra-toes (Hx) (Masuya et al. 1995; Blanc et al. 2002) and MFCS1 (Sagai et al. 2005). MFCS1 heterozygotes, which have a reduced level of Shh transcript, have normal autopod patterning, so we used it as a control to show normal expression of Mkp3 and Sef. Hx mutants show ectopic expression of Shh at the anterior end of the limb bud (Blanc et al. 2002), which mimics ZPA transplantation in chick embryo experiments. Hx mutants with extra ZPA in the anterior side give rise to skeletal alternation, including polydactyly in the autopod. In E10.5 Hx+/− embryos, Mkp3 expression was found in the distal mesenchyme mainly in the anterior region (Fig. 4B). At E11.5, Mkp3 expression was reduced in the entire distal mesenchyme except the anterior end, where Shh was ectopically expressed (Fig. 4E, asterisk). It was previously shown that Shh-expressing cells greatly reduce their Shh responsiveness (Ahn & Joyner 2004). Thus, the Mkp3 expression in the anterior Shh-positive region may be due to a failure to respond to Shh signal. In E10.5 embryos, there was no significant difference in Sef expression between control and Hx+/− limb buds (Fig. 4H, compare with 4G). At E11.5, however, Sef expression was reduced in the distal mesenchyme (Fig. 4K). As in chick limb buds, Shh signaling negatively regulates Mkp3 and Sef expression in the mouse limb bud, but responsiveness to Shh signaling seems to differ between these genes.

Figure 4.

 Change in expression of anterior-shifting gene in mouse mutants. Expression of Mkp3 (A–F) and Sef (G–K) in Hx+/− (B, E, H, K) and MFCS1−/− (C, F, I, L) mouse mutants. Dorsal views of right forelimbs at embryonic day 10.5 (E10.5) (A–C, G–H) and E11.5 (D–F, J–L). MFCS1+/− embryos were used as controls (A, D, G, J). Asterisks indicate the domain of anterior-shifted asymmetric expression.

MFCS1−/− mouse embryos lack Shh expression in the limb bud, mimicking the cyclopamine-treated chick embryo. In MFCS1−/− embryos, although the autopod is greatly reduced in size (Sagai et al. 2005), Mkp3 and Sef are expressed throughout the distal mesenchyme in the E10.5 limb bud (Fig. 4C,I), similar to controls (Fig. 4A,G). At E11.5, the reduction in gene expression in the posterior mesenchyme and the resultant anterior shift were not observed (Fig. 4F,L), suggesting that Shh signaling is necessary for the anterior shift. Mkp3 and Sef expression was very weak in MFCS1−/− embryos, likely due to decreased fgf8 expression in this mutant (Sagai et al. 2005). Thus, the mechanism of the anterior shift appears to be conserved among amniotes.

Relationship between the anterior shift and FGF signaling

The anterior domains of these three genes at stage 25–26, after the anterior shift, correspond to the region that will give rise to the anteriormost digit, which is shorter than the other digits and is a common feature of tetrapods (compare Figs. 1O,1P,2I, and 2P). To elucidate the roles of Mkp3, Sef, and TSK in forming the anteriormost digit, we overexpressed each gene in chick embryos by electroporation of expression constructs, but we did not see any significant digit morphology phenotypes (data not shown). We then focused on FGF signaling, since Mkp3, Sef, and TSK all downregulate FGF signaling through the Ras/MAPK pathway (Tsang & Dawid 2004; Morris et al. 2007). FGF administration at early stages of limb development (before the anterior shift occurs) results in additional digit-like protrusions (Riley et al. 1993; Mima et al. 1995), and its application at much later stages (around stage 27–28) gives rise to local effects, including inhibition of tip formation and distal digit fusion (Sanz-Ezquerro & Tickle 2003). Thus, we transplanted FGF4- and FGF8-soaked beads into the anterior margin of the AER at stage 22–23, when the anterior shift begins, in order to examine the effect of excess FGF ligand on the morphology of digit 2. FGF4-treated limbs did not differ from untreated control limbs (Fig. 5A,B), but in FGF8-treated limbs of 10-day chick embryos, the phalanges of digit 2 were elongated and/or hypertrophic, without any additional elements (Fig. 5C, arrow). The area of the metacarpal and phalangeal cartilage of digit 2 was significantly larger than that on the control side of the same embryo (Fig. 5D). Interestingly, excess FGF during specification of the digit-forming region changes the characteristics of digit 2, suggesting that precise regulation of FGF levels controls the size of the anteriormost digit module.


The anterior shift in gene expression and specification of the anteriormost digit

The anterior shift of Mkp3, Sef, and TSK expression in the distal limb mesenchyme was observed by whole mount in situ hybridization (Figs. 1 and 2). In the chick embryo, the anterior domain expressing Mkp3, Sef, and TSK in stage 25–26 limb buds clearly corresponds to or includes the Mario-expressing domain, the presumptive digit 2-forming region in the forelimb and the digit 1-forming region in the hindlimb. The dynamic changes in gene expression pattern between stages 22–23 and 25–26 are likely related to digit specification.

Our data also provide insight into the diversity and conservation of Mkp3, Sef, and TSK expression patterns. TSK, which shows a clear anterior shift in expression in the chick limb bud, is not expressed in the distal mesenchyme of mouse and gecko limb buds (Fig. 3). It is possible that our probe was not sensitive enough to detect a low level of transcription, but β-gal staining of TSK+/LacZ mouse embryos did not show activity in the distal region (unpublished observation, Asaka Uejima, Kunimasa Ohta and Koji Tamura, 2008), suggesting that TSK is not expressed in the distal limb bud. It is possible that the anterior shift in TSK expression is avian-specific, since the anteriormost digit in bird limbs has peculiarly small and short phalanges, particularly the metacarpal and metatarsal elements (compare Figs. 1O 1P,2I, and 2P). During evolution, birds may have co-opted molecular mechanisms, such as TSK inhibition of FGF signaling, in order to develop traits particular to the anteriormost digit. Interestingly, homologues of Mario, a marker gene for the most anterior digit (Fig. 1J,K), are found only in avian genomes (Amano & Tamura 2005, and our unpublished data, Naoki Nomura and Koji Tamura, 2009).

The anterior shift in Mkp3 and Sef expression, though not as obvious compared with that observed in chicks, was conserved in mouse and gecko embryos. Thus, the developmental mechanism used for anteriormost digit formation appears to be conserved among amniotes. The bird wing has three digits, 2, 3, and 4, and the leg has four digits, 1, 2, 3, and 4. Wing digit 2 and leg digit 1 are therefore both anteriormost digits that share a common developmental mechanism. This mechanism is adapted to digit 1 of the mouse and gecko, which have five digits, 1, 2, 3, 4, and 5.

This raises a question that has been a continuing controversy for over a hundred years between developmental biologists, who have claimed that the digits of the wing of birds are digits 2, 3, and 4, and paleontologists, who say that they are 1, 2, and 3 (Burke & Feduccia 1997; Wagner & Gauthier 1999; Feduccia & Nowicki 2002; Kundrat et al. 2002; Larsson & Wagner 2002; Welten et al. 2005). Recently, some reports suggested that the digits of bird wing show a 1-2-3 pattern, based on the expression of 5′ HoxD genes during development (Vargas & Fallon 2005; Vargas et al. 2008). Our results partially support this idea, but more investigation is required in order to definitively answer this question, since the hypotheses proposed to date, including the frame-shift hypothesis (Wagner & Gauthier 1999), are based on too many assumptions. Our findings and the above HoxD gene evidences complementarily suggest that digit 2 in the chick wing and digit 1 in the chick leg and in other amniote limbs might share a common mechanism of gene expression specific to that region.

The anteriormost digit is formed independent of Shh signaling

Mario has a fixed anterior-specific expression domain in the developing chick limb bud (Amano & Tamura 2005), whereas the genes analyzed here show dynamic patterns of expression, with a shift from distal to anterior expression. Cell fate mapping indicates that the posterior portion of the distal mesenchyme undergoes proliferation to expand along the A-P axis (Vargesson et al. 1997; and our unpublished observation, Naoki Nomura and Koji Tamura, 2009), but this cannot completely account for the anterior shift. Downregulation of gene expression, likely through inhibitory signals from the posterior margin, must also be involved. When we inserted an aluminum barrier into the middle of the limb before the anterior shift occurred, we observed no effect on posterior reduction of Mkp3, Sef, and TSK expression (Fig. 3B,F,J), suggesting that the repression in the posterior mesenchyme occurs independently of the anterior mesenchyme. Importantly, ZPA transplantation experiments suggest the existence of the inhibitory effects in the ZPA. Moreover, cyclopamine treatment (Fig. 3D,H,L) strongly suggests that Shh in the posterior mesenchyme inhibits these three genes.

Since Mkp3, Sef, and TSK are maintained in the anteriormost region of the distal mesenchyme and are repressed by Shh signaling, it is likely that the anterior expression domains of these three genes are formed in an area where the Shh signal cannot reach (Fig. 6). Interestingly, the anterior domains re-expand posteriorly by stage 27–28 (Fig. 1C,F,I), suggesting that the range of the Shh-responsive area narrows at later stages. The idea that the anteriormost digit is produced independently of Shh signaling and in the area outside the range of Shh signaling has been proposed by studies in both mouse and chick (Ahn & Joyner 2004; Harfe et al. 2004; Amano & Tamura 2005). However, it is still possible that the limb mesenchyme requires Shh for cell proliferation to form the anteriormost digit, probably at the early limb bud stage before the anterior shift occurs. This is supported by the fact that transgenic mice and a chick mutant (Ozd) that do not express Shh sometimes contain no digit-like structures in the forelimb (Chiang et al. 1996; Kraus et al. 2001; Ros et al. 2003; Sagai et al. 2005).

Figure 6.

 Model for mechanism of the anterior shift and specification of digit 2 in the chick wing bud. In the early-stage limb bud (upper left), Mkp3, Sef and TSK are expressed in the entire distal limb bud under control of the apical ectodermal ridge (AER). At this stage, the zone of polarizing activity (ZPA) releasing Shh protein has no effect on the expression of these genes, whereas the posterior half of the limb bud is thought to receive Shh signal. At a later stage (lower left), when the autopod region is being specified, the posterior mesenchyme of the distal limb bud that receives the Shh signal (indicated in blue as expression of Ptc2, a downstream target of Shh signal) diminishes the expression of Mkp3, Sef and TSK and the anterior-side expression is maintained by FGF8, resulting in the anterior shift of expression of these three genes. Mkp3, Sef and TSK are negative regulators of fibroblast growth factor (FGF) signaling, and the relatively weak FGF signal in the anterior side gives rise to the digit 2-specific traits (light green in lower right).

The effect of the anterior shift on digit identity

It has been very difficult to determine the role of Mkp3, Sef, and TSK in digit formation during limb bud development because they are feedback regulators of FGF signaling that attenuate the strength of signaling to some extent, but do not repress it completely. Many attempts have been made to examine how loss of function of these genes affects development in mice, including limb bud development. Heterozygous of Mkp3 knockout mice show overall dwarfism but no digit loss or change in digit identity (Li et al. 2007). Sef knockout mice do not show any significant limb phenotype (Lin et al. 2005), nor do TSK knockout mice (unpublished observations, Asaka Uejima, Kunimasa Ohta and Koji Tamura, 2008). Overexpression of these genes by electroporation in chick embryos did not result in digit pattern phenotypes either (data not shown). Taken together, these results suggest that the signaling pathways these molecules are involved in are complex and redundant. Therefore, it may be necessary to perform simultaneous knockout or knockdown or gain-of-function experiments targeting multiple molecules. It is still possible, therefore, that these three proteins (and possibly others) together specify anteriormost digit character.

Mkp3, Sef, and TSK are known to negatively regulate FGF signaling, which is a well-known signaling mechanism essential for various events during limb development (reviewed by Martin 2001). Since loss of FGF signaling by specific inhibitors represses limb development before digit formation, we applied FGF-soaked beads to limb mesenchyme to see the effect of FGF overexpression on digit morphology. In the case of FGF8, the phalanges of digit 2 were elongated, perhaps due to an increase in cell proliferation of the digit-forming mesenchyme. There were no extra elements, and all three phalanges of digit 2 appeared enlarged and elongated, suggesting that during normal development, the levels of FGF signaling in mesenchymal cells must be tightly controlled in order to produce the proper shape of digit 2. Mkp3 is a negative regulator of the mitogen-activated protein kinase (MAPK)/extracellular signal-regulated protein kinase (ERK) pathway, and Sef also negatively works on this pathway. The intracellular response to FGFs is mediated by several signal transduction cascades, including the PI(3)K/Akt pathway as well as the MAPK/ERK pathway. Since Mkp3 expression is induced through the PI(3)K/Akt pathway in the developing limb (Kawakami et al. 2003), and it is possible, therefore, that a high level of phosphorylated Akt and Mkp3 expression plus a low level of phosphorylated Erk may be constituted in the anterior-distal mesenchyme. We further speculate that different intracellular response to FGFs along the AP axis in the limb bud differentially regulates digit morphology. It is also interesting to speculate about the role of FGF signaling in the context of digit morphology diversification among species. As discussed above, the anteriormost digit (referred as digit 1 in the five digits, 1, 2, 3, 4, and 5) is almost always shorter and smaller than the other digits in all classes of tetrapods (even in bat limbs whose digits, 2, 3, 4, and 5, have extremely long phalanges), and avian limbs have a particularly shorter anteriormost digit both in the forelimb (wing) and hindlimb (leg). It is possible that phalange elongation caused by FGF signaling is prevented in the phalange elements in digit 1 via inhibition or modulation of FGF signaling.


We thank Dr Y. Kawakami (Minnesota University) and Dr G. Martin (University of California, San Francisco) for the kind gift of plasmids. This work was supported by research grants from the Ministry of Education, Science, Sports and Culture of Japan, KAKENHI (Grant-in-Aid for Scientific Research) on Priority Areas “Comparative Genomics”, and Toray Science Foundation.