Role of Pitx1 Upstream of Tbx4 in Specification of Hindlimb Identity

See allHide authors and affiliations

Science  12 Mar 1999:
Vol. 283, Issue 5408, pp. 1736-1739
DOI: 10.1126/science.283.5408.1736


In spite of recent breakthroughs in understanding limb patterning, the genetic factors determining the differences between the forelimb and the hindlimb have not been understood. The genes Pitx1and Tbx4 encode transcription factors that are expressed throughout the developing hindlimb but not forelimb buds. Misexpression of Pitx1 in the chick wing bud induced distal expression of Tbx4, as well as HoxC10 and HoxC11, which are normally restricted to hindlimb expression domains. Wing buds in which Pitx1 was misexpressed developed into limbs with some morphological characteristics of hindlimbs: the flexure was altered to that normally observed in legs, the digits were more toe-like in their relative size and shape, and the muscle pattern was transformed to that of a leg.

In most respects the genetic systems controlling limb development act equivalently in the chick forelimb (wing) and hindlimb (leg) buds. However, as is true for the human arm and leg, the chick wing and leg have very different morphologies. Several genes have been described that are expressed in a manner consistent with their playing a role in limb-type specification.Pitx1, an Otx-related paired-type homeobox transcription factor, is expressed in the developing hindlimb bud but not forelimb bud (1–3). Two T-box family transcription factors Tbx5 and Tbx4 are expressed, respectively, in the forelimb bud and hindlimb bud in the mouse (4) and newt (5). Moreover, the correlation between the expression of these genes and limb identity is maintained after various surgical manipulations in the chick (6–10). We focused our initial functional analysis onPitx1 because it is expressed before the Tbxgenes in prelimb stage chick embryos (6) and might therefore act upstream of them in limb-type specification.

We injected a retroviral vector carrying Pitx1 into the wing field of stage 9-10 embryos and examined the expression of various limb-type–specific markers between stages 22 and 27 (11). We observed ectopic induction of Tbx4 in forelimbs where Pitx1 was successfully misexpressed (n = 13). This ectopic expression was largely limited to the distal mesenchyme at stage 22 and its subsequent derivatives in later limb buds (Fig. 1). Although virus is introduced before limb formation, there is a delay of 16 to 18 hours after infection before the transgene is actively expressed. Thus, induction of Tbx4 in response to Pitx1 expression may be restricted to the undifferentiated cells still in the distal progress zone when Pitx1 protein is first ectopically produced. After infection with Pitx1, ectopic Tbx4 induction was also seen in the interlimb flank mesoderm (12), which is capable of giving rise to ectopic limbs after application of fibroblast growth factors (13–16). However, we never sawTbx4 induction in infected limb bud ectoderm and there was no Tbx4 induction in Pitx1-infected somites or paraxial mesoderm.

Figure 1

Ectopic induction of hindlimb-specific genes after ectopic expression of Pitx1. Whole-mount in situ hybridization was carried out with probes forTbx4, Tbx5, HoxC10, andHoxC11, as indicated at the left of each row. The contralateral, uninfected wing is shown in the left column (these images have been reversed in their horizontal plane for easier comparison), the normal expression in the leg is shown in the right column, and the Pitx1-infected wings are shown in the central column. The examples shown for each gene are from the same embryo: Tbx4 and HoxC11, stage 25;Tbx5, stage 24; and HoxC10, stage 26. Red arrows point to domains of ectopic Tbx4, HoxC10, andHoxC11 in Pitx1-infected limbs. Green arrows point to the endogenous posterior domain of HoxC10 and the endogenous distal, posterior domain of HoxC11 in normal limbs, which are recapitulated in the Pitx1-infected limbs.

Because Tbx4 and Tbx5 are normally found in mutually exclusive limb territories, it was possible thatPitx1 might also act to repress the forelimb-specific geneTbx5, either directly or through mutual transcriptional antagonism between Tbx4 and Tbx5. We examinedTbx5 expression in wing buds after Pitx1misexpression but never observed any change in Tbx5expression, even in domains where Tbx4 is up-regulated (Fig. 1; n = 22). Another possible regulatory pathway could lead to autoregulation of Pitx1; however, we found no evidence of endogenous Pitx1 induction in infected wing buds, including its absence in the distal domain where Tbx4was ectopically induced [(12); n = 13]. Thus in a linear pathway, Pitx1 induces expression of the second hindlimb-specific gene Tbx4, whereas the forelimb-specific gene Tbx5 is under independent control. This induction is specific in that Tbx4 was never induced in limb buds infected with the closely related gene Pitx2[(12, 17); n = 9].

A number of other genes show limb-type–specific expression patterns, for example HoxC10 and HoxC11 in the hindlimb buds (18). These genes are unlikely to play a primary role in establishing limb identity, as they are expressed in limited subdomains of the hindlimb bud and are expressed relatively late during limb development. Nevertheless, they are excellent candidates for genes involved in mediating aspects of hindlimb-specific morphogenesis. We found that both HoxC10 andHoxC11 were induced ectopically in infected wing buds (Fig. 1; HoxC10, n = 15; HoxC11,n = 7). Only the distal and posterior expression of their respective leg domains were recapitulated inPitx1-infected wings. Thus, at a molecular level,Pitx1 induces hindlimb-specific changes when misexpressed in wing buds.

To examine the morphological consequences of Pitx1misexpression, we infected the wing territories of stage 8-9 chick embryos with the Pitx1 virus, and the embryos were allowed to develop until stages 35 to 37. Immediately obvious, at a gross morphological level, was a change in the overall flexure of the limb (Fig. 2, B and D). The autopod (hand and digits) of the normal chick wing is posteriorly flexed relative to the zeugopod (lower arm). The digits remain in the anterior-posterior plane in which they formed (Fig. 2A). In contrast, in the normal hindlimb there is no equivalent posterior flexure at the ankle; the distal elements of the leg are maintained in a straight orientation. In addition, the autopod is rotated 90° such that the digits are in a horizontal position for walking, with the former anteriormost digit becoming medial and the more posterior digits becoming sequentially more lateral in position (Fig. 2C). In the chick, this rotation begins around stage 30 and is complete around stage 34 (19). ThePitx1-infected wings were often held in a leglike position in both respects, showing no posterior flexure at the wrist and having the digits rotated 90° (Fig. 2, B and D). None of the examples in which Pitx1 was successfully misexpressed had normal posterior flexure at the wrist. The rotation of the digits 90° was a more variable phenotype. Although all the examples showed at least some degree of rotation, only a minority were rotated fully 90° (n = 4/11). There are also distinctions between the ectodermal derivatives in the wing and leg. The wing uniquely has a flap of skin, the patagium, extending between the flexed upper and lower arm (Fig. 2A). In all Pitx1-infected limbs, the patagium was absent or severely reduced in size (Fig. 2, B and D). Because after infection the flexure of the upper limb is unchanged, the absence of the patagium is not a simple consequence of the angle of the limb, but a direct result of the patterning alterations due toPitx1.

Figure 2

Gross morphology of normal andPitx1-infected limbs. (A) Normal wing, stage 36. (B) Pitx1 -infected wing, stage 35. In thePitx1-infected wing, the normal flexure of the wing has changed, no patagium has formed, and the ectoderm is smooth, with no evidence of formation of nascent featherbuds. The blue arrows indicate the areas of interdigital cell death at the distal extreme of the limb, as normally occurs in the leg. (C) Normal leg, stage 36. Red arrows are discussed in the text. (D) An example of a different embryo (stage 35) in which Pitx1 has been misexpressed in the right wing. The overall morphology of the infected wing has been altered, and in addition, the digits have rotated 90° relative to their normal position (compare with the uninfected, left wing) and are now aligned similarly to the digits of the leg.

Another ectodermal difference between fore- and hindlimbs in the chicken is that the wing forms feathers whereas the distal leg forms scales. Wings in which Pitx1 was misexpressed always showed at least some loss of normal featherbud formation (Fig. 2) (12), although the degree of loss was variable. Affected areas were always distally restricted and were correlated with regions of the wing bud in which the normal morphology had been most obviously grossly affected. Infected wings, however, did not develop scales but rather remained smooth on their surface. Although it is not clear why scales did not form, in principle the smooth ectoderm could have resulted as a nonspecific suppression of ectodermal appendage formation when Pitx1 was ectopically expressed in the ectoderm. However, Pitx1 never disrupted scale formation when misexpressed in the hindlimb, and importantly, no suppression of featherbud formation was observed after Pitx1 misexpression in the flank (12). This indicates that Pitx1expression per se is not incompatible with feather formation. Rather, we propose that the loss of feathers in infected wings is a patterning defect due to the misexpression of this hindlimb-specific gene in the wing.

To analyze the changes in skeletal pattern induced by Pitx1misexpression, we cleared and stained infected wings with Alcian blue (n = 12) (20). In the normal leg, digits II, III, and IV, which form the forward-pointing toes, are of approximately equal length, and there is an additional small anterior digit I that ultimately rotates to point toward the back of the foot (Fig. 3C). In infected wings, digits II, III, and IV were similar in size (Fig. 3B). In some cases there were modified wing digits II, III, and IV, with digit II being notably extended (12). In other examples, there was a fourth digit, often formed by bifurcation of wing digit III (Fig. 3D), whereas the anteriormost wing digit II remained short, approximating the normal leg digit I (n = 4/12). Although the number of phalangeal elements in each digit remained unchanged afterPitx1 misexpression, the shapes of many of the individual bone elements in the digits were more similar to those found in the leg (Fig. 3B). Moreover, at the tips of the these digits, claws formed, which are normally seen only on hindlimbs (Fig. 3B, red arrow;n = 6/12). In addition, an apparent change occurred in the pattern of interdigital cell death. In the normal leg, interdigital cell death results in a separation of each of the digits (Fig. 3C), whereas in the wing, digits III and IV remain together surrounded by soft tissue (Fig. 3A). In infected wings there was at least partial separation of the distal ends of digits III and IV (Figs. 2B and 3B) and in some cases the digits separated completely (Fig. 3D). Finally, the change in wrist flexure was also clear in the skeletal preparations. For example, the characteristic downward flexure at the wrist was lost in the Pitx1-infected wing and instead the digits pointed straight out in a similar fashion to the digits of the foot at the ankle (Fig. 3B). On the basis of the molecular analysis, it is not surprising that Pitx1 only induced a partial forelimb-to-hindlimb transformation, because the normal expression of the forelimb-specific gene Tbx5 is maintained in addition to the ectopic expression of the hindlimb-specific gene Tbx4. However, a more complete transformation was observed when the soft tissues were analyzed.

Figure 3

Alcian blue staining of the skeletal elements of normal and Pitx1-infected limbs. (A) A normal wing, (B) a Pitx1-infected wing, and (C) a normal leg at stage 38. The digit patterns are indicated with roman numerals, wing II to IV and leg I to IV. The digits in the Pitx1-infected wing (distinguished with an asterisk) are more uniform in length and contain some phalangeal elements most similar to those in the foot. A claw on digit III* is indicated by the red arrow. In addition, four digits have formed instead of three. The extra digit often forms by a bifurcation of wing digit III, as can be seen in an infected wing harvested at stage 33 when the cartilage elements are forming (D). The blue arrows indicate areas of interdigital cell death around the transformed digits.

The limb-type–specific patterning of the soft tissues is, at least to some extent, independent of the skeletal patterning, because wing-specific muscles form even in the absence of any skeletal elements (21). Because Pitx1 is expressed throughout the hindlimb mesenchyme, it seemed likely that Pitx1 might, either directly or indirectly, modulate the patterning of the musculature in addition to the skeleton. Moreover, because the undifferentiated myoblasts migrate into the limb bud after proximal skeletal elements have been specified, it was possible thatPitx1 infection might be able to alter muscle patterning at a more proximal level than observed for the bones. Indeed, gross morphological examination showed that the soft tissue of the infected zeugopod (level of the radius and ulna) bulged dorsally like the shank of the leg (level of the tibia and fibula) and unlike the morphology of that region of normal wings (Fig. 2, red arrows).

To directly examine the effect of Pitx1 on muscle patterning, we stained infected and control stage 32-36 limbs with a muscle-specific antibody to myosin (MF20) (20). The zeugopod of a normal stage 32 wing contains 6 muscles dorsally and 7 muscles ventrally (Fig. 4, A and B) (22), whereas the leg contains 4 muscles dorsally and 12 muscles ventrally (Fig. 4, E and F) (23, 24). Importantly, each of these muscles is uniquely defined by characteristic shape, position, pattern of fiber orientation, origin, and insertion. Together, these characteristics allow each muscle in the wing and leg to be unambiguously distinguished. InPitx1-infected limbs the overall muscle pattern underwent a dramatic transformation (n = 9/11). This was particularly apparent on the dorsal side (Fig. 4, C, D, and H). The extensor digitorum longus (EDL) (Fig. 4, E and F, in detail I) is uniquely identifiable by its central dorsal position, narrow fusiform shape, and characteristic bipennate fiber pattern (fibers in the anterior and posterior of the muscle are arranged at opposing angles;Fig. 4G). Moreover, this muscle is attached to a large central tendon that extends to each of the hindlimb digits (it is the muscle controlling release of the digits from a flexed, perching position). There is no equivalent muscle, either in morphology or in tendon attachment, in the wing. In infected wings there was a clearly identifiable EDL (in 2/11 the transformation to an EDL was complete in all respects) (Fig. 4, D and H). A similar muscle transformation was seen in the formation of a leg-specific fibularis brevis (FB) in infected limbs, which is distinguished by its distal lateral position, short rectangular shape and unipennate fiber pattern (Fig. 4, C and D) (25). Two other muscles seen in Pitx1-infected wings were similar in shape and position to the normal hindlimb tibialis cranialis and fibularis longus (Fig. 4, C and D).

Figure 4

Whole-mount immunohistochemistry with an antibody to myosin that identifies the limb musculature. Dorsal views are shown. For improved clarity, most of the underlying ventral muscles have been removed. (A) A schematic of the muscles of the normal wing zeugopod (radius and ulna region). The entire limb, from stylopod to autopod (elbow to digits) is shown in (B). (C) A schematic of the muscles in aPitx1-infected wing zeugopod. The entire limb, from stylopod to autopod is shown in (D). (E) A schematic of the muscles of the normal leg zeugopod (tibia and fibula region). The entire limb from stylopod to autopod is shown in (F). Overall, the muscle pattern in the Pitx1-infected wing is altered. In particular, a leg extensor digitorum longus (EDL) is present, as well as fibularis brevis (FB), tibialis cranialis (TC), and fibularis longus (FL) [(C) and (D)]. The muscles can be unambiguously identified by shape, origin and insertion, and patterns of fiber orientation. In the normal EDL [shown schematically in (G) and in detail in (I)], the fibers angle distally and medially to the thick central tendon (arrow), which ultimately splits to control the extension of each digit. A detail of the transformed EDL is shown in (H). The muscles shaded in light gray in (C) and (E) have a ventral origin and were not included in the comparison. In the wing schematic in (A), the muscles shaded in gray do not send tendons across the wrist and are not described. EMR, extensor metacarpi radialis; EIL, extensor indicis longus; EDC, extensor digitorum communis; EML, extensor medius longus; ANC, anconeus; EMU, extensor metacarpi ulnaris.

On the ventral side, specific transformation of muscle type was harder to classify. There was, however, an increased number of muscles (9), which is more characteristic of the leg (12). Moreover, there was a change in the attachment of the flexor digitorum superficialis in the wing. This characteristic muscle normally crosses the wrist joint in the wing, whereas no equivalent muscle crosses the ankle joint in the leg. After Pitx1infection there was no extension of any muscle crossing the wrist joint (12). In the autopod, the muscle pattern was also altered in the infected wings, but the similar morphology of these muscles in the wing and leg makes definitive identification difficult (Fig. 4, B, D, and F). In the stylopod, there appears to have been no transformation of the muscle pattern (12), consistent with the fate of this region being the first to be fixed during development of the limbs and with the lack of affect on gene expression in the proximal wing bud.

The effects of Pitx1 on muscle organization were highly specific and in detail appear to have been clear pattern transformations; therefore, the muscle alterations do not appear to have been secondary consequences, for example, of changes in the size of the limb bud. Indeed, after Pitx1 infection the length of the zeugopod in the infected wing and in the contralateral wing were identical (Fig. 4). It should be noted that muscles can affect skeletal development, and therefore the transformation of muscle identity inPitx1-infected wings could be at least partly responsible for the transformation of skeletal morphology observed. Furthermore, the transformation of muscle and distal tendon attachments could cause the apparent footlike flexure we describe above (Fig. 2). As with the skeletal transformations, the effect of Pitx1 on muscle pattern is specific to the hindlimb gene Pitx1 because parallel misexpression of a highly related control gene,Pitx2, did not have an effect on soft tissue morphology, in spite of the fact that this gene had a high biological activity in other contexts (17). Because comparatively little is known about the molecular basis for muscle patterning, the ability ofPitx1 to respecify muscle identity will provide a useful tool in this regard, in addition to being evidence for a role ofPitx1 in the specification of hindlimb patterning.

In summary, we found that Pitx1 acts upstream ofTbx4 and regionally expressed Hox genes in a pathway that regulates limb-type identity. The correct induction of the forelimb- or hindlimb-specific genes in the respective limb fields must depend on upstream genes that regionalize the rostral-caudal body axis, which is likely ultimately dependent on axial expression ofHox genes. Further gain- and loss-of-expression studies with other genes in this pathway will provide additional insight into the mechanism by which the common aspects of limb patterning are modified to produce limb-type–specific morphologies.

  • * To whom correspondence should be addressed. E-mail: tabin{at}


Stay Connected to Science

Navigate This Article