It Takes Time to Make a Pinky: Unexpected Insights into How SHH Patterns Vertebrate Digits

See allHide authors and affiliations

Science's STKE  16 Nov 2004:
Vol. 2004, Issue 259, pp. pe53
DOI: 10.1126/stke.2592004pe53


It is widely accepted that the diffusible Sonic Hedgehog (SHH) morphogen signal establishes a spatial gradient that patterns embryonic structures by long-range signaling. In response, cell fates are determined by linear thresholds according to the position of cells within the gradient field. Two recent studies of SHH signaling during vertebrate limb development challenge this spatial gradient model. They establish that a large fraction of limb bud cells patterned by SHH are descendants of cells that have previously expressed Shh. These cells are endowed with a kinetic memory that integrates their exposure to SHH rather than sensing their position in a SHH gradient. In addition, a fraction of cells changes their SHH responsiveness progressively during limb bud pattering, which is indicative of local nonlinear modulation of cell fate specification.

The organization of large cell communities into organs during embryogenesis is a fascinating developmental process. In particular, the morphogen gradient hypothesis for the development of the vertebrate limb proposes that digits are specified by cells responding to a spatial gradient of the diffusible Sonic Hedgehog (SHH) signal. Harfe and co-workers (1) now challenge the morphogen gradient model by establishing that the descendants of Shh-expressing cells themselves form the three posterior of five digits in the mouse. Their identity is specified by a cellular memory that locally integrates the SHH signal over time. In a complementary study, Ahn and Joyner (2) marked the responsive limb bud cells at specific times during morphogenesis by molecular tagging of the SHH transcriptional target Gli1. This analysis reveals the kinetic response to SHH signaling and establishes that the response of posterior cells exposed to the highest cumulative levels of SHH signaling is lowered over time.

Since the seminal experiments of Mangold and Spemann (1924), it has become clear that organizers are universal in orchestrating embryonic patterning. Their special signaling properties enable them to recruit and determine the fates of other cells (3). The zone of polarizing activity (ZPA, also called the polarizing region) is the mesenchymal organizer located in the posterior limb bud that coordinates outgrowth with patterning. In the chicken, ectopic anterior grafts of ZPA cells induce mirror-image digit duplications along the anteroposterior limb axis. Duplication of the most posterior digit 5 (the little finger or pinky) requires the highest polarizing activity, whereas duplication of more anterior digits requires weaker polarizing activity or fewer ZPA cells. At first sight, the results of these and other experimental manipulations can apparently be satisfactorily explained by Wolpert’s morphogen hypothesis (4). His model postulates that ZPA cells produce a morphogen signal, whose diffusion establishes a spatial gradient with its high point in the posterior limb bud mesenchyme. In response, cell fates are specified according to distinct threshold values depending on their distance from the source (Fig. 1A).

Fig. 1.

Schematic representation of Wolpert’s spatial morphogen gradient model (4). The hypothetical threshold values postulated to specify the digits are indicated in red. Digits are numbered so that the most anterior digit (the thumb) is 1 and the most posterior one (the little finger or pinky) is 5. The ZPA in the posterior limb bud mesenchyme is indicated in dark blue, and the limb skeletal elements are shown only schematized.

Ten years ago, SHH was identified as the morphogen signal produced by the ZPA (5). The active SHH signal corresponds to a 19-kD N-terminal peptide generated by autoproteolytic cleavage that is modified by the covalent addition of cholesterol and palmitate (6). Many studies have underscored the general long-range signaling capacity of SHH, and loss-of-function genetics has established essential SHH functions during embryogenesis, maintenance of stem cells, and disease in vertebrates (7). In the developing neural tube, SHH indeed functions as a diffusible long-range morphogen by instructing neuronal identities along the dors-ventral axis in a dose-dependent manner (8).

Lack of SHH during limb bud morphogenesis in various vertebrate species (including humans) results in a complete loss of the ulna (posterior bone) and digits 5 to 2, whereas the most anterior digit 1 (the thumb) is not affected (4). These and other studies establish the absolute requirement of SHH signaling for anteroposterior patterning of the distal limb skeleton. Furthermore, the SHH protein can diffuse from the ZPA to elicit response at a distance in the limb bud mesenchyme (9). Cells responding to SHH signaling activate the GLI1 transcription factor, but genetic analysis in the mouse shows that Gli1 is not essential for limb bud development (10). The related GLI2 protein also functions as a positive downstream mediator of SHH signaling, but is again not essential for limb bud development (11). Rather, SHH signaling seems to enable distal progression of limb bud morphogenesis and formation of the digit arch by inhibiting proteolytic production of the repressor form of another GLI family member, the GLI3 protein (12, 13), which is expressed primarily in cells that do not express Shh. Although many but not all studies concur with the existence of a spatial morphogen gradient, none have analyzed the kinetics of SHH gradient formation and the postulated threshold responses, nor the potential direct contribution of ZPA cells to digit primordia.

Harfe et al. (1) set out to address these issues genetically by marking all Shh-expressing cells and their descendants through permanent activation of β-galactosidase (β-Gal), either from the onset of Shh expression or at specific time points during mouse limb bud development. A first unexpected result of these cell lineage studies was that ZPA cells are not eliminated by apoptosis. Rather, they give rise to descendants that either remain ZPA cells or join a distal-anteriorly expanding population of descendants that no longer express Shh. Together, these two populations of Shh descendant cells give rise to the most posterior digits(5 and 4), parts of digit 3, and the ulna (Fig. 1B). Therefore, digit 2 is the only skeletal element missing in Shh-deficient mice that is not at least partially derived from cells that expressed Shh at some stage. These results challenge the relevance of a spatial morphogen gradient, because only digit 2 depends on long-range SHH signaling. Indeed, genetic reduction of the mobility of SHH across the limb bud affects digit 2, whereas digits 3 to 5 form normally [see also (10)]. Taken together, these results minimize the contribution of long-range SHH signaling to limb bud patterning (Fig. 1B).

Fig. 2.

Schematic representation of the expansion-based temporal SHH gradient (1) that specifies digits 3 to 5. Digit 1 and the radius form independently (as does the humerus; not shown). Digit 2 depends on long-range SHH signaling from the ZPA (indicated in dark green). The ulna and digit 3 are only partly formed from descendants of cells having previously expressed Shh, whereas digits 4 and 5 are completely formed by such cells. The identities of digits 3 to 5 are specified so that cells expressing Shh the longest will be specified as the most posterior digit 5. Cells appear to retain a kinetic memory of the cumulative levels of SHH received over time.

To understand how the identities of digits 3 to 5 are specified, the authors irreversibly marked Shh-expressing cells at defined time points rather than from the beginning of Shh expression. These kinetic studies revealed that the fates of Shh descendants are progressively restricted posteriorly: Descendant cells that do not express Shh and are born early (during embryonic day 9.5) contribute to all three digits, whereas the ZPA cells expressing Shh longest (up to embryonic day 11.5) and their descendants contribute exclusively to digit 5 (Fig. 1B). These and other results establish that limb bud cells somehow acquire a kinetic memory of the SHH signal received. In summary, the cumulative length of time that the expanding population of Shh-expressing cells and their nonexpressing descendants are exposed to SHH signaling is decisive in specifying the three posterior digits. This contrasts sharply with the proposed spatial morphogen gradient model; in fact, long-range SHH signaling is only required to specify digit 2, whereas digit 1 does not depend on SHH at all (Fig. 1, A and B).

In a complementary study, Ahn and Joyner (2) marked the mouse limb bud cells responding to SHH signaling by analyzing transcriptional activation of Gli1. Gli1 served as a target to allow permanent activation of an inducible β-Gal lineage marker at specific time points in cells responding to SHH. These studies established beyond doubt that the mesenchymal cells giving rise to digits 5 to 2 and the ulna respond to SHH signaling, because they showed activation of the Gli1 promotor (Fig. 1C). This first major conclusion is in agreement with the phenotype of limbs lacking SHH and nicely complements the studies by Harfe and co-workers (1) (Fig. 1B). However, although the cumulative SHH response was, as expected, highest in the posterior mesenchyme (digit 5) and progressively lower toward the anterior, no specific thresholds of response were found as predicted by the morphogen gradient model (Fig. 1A). In particular, the responsiveness of the most posterior cells (fated to digit 5), which are exposed to SHH the longest (through autocrine or paracrine signaling, or both), was reduced with time. The mechanism underlying this dynamic modulation of response to SHH signaling is unknown, but it seems that cells exposed to high cumulative levels of SHH signaling lose their responsiveness with time.

Fig. 3.

The Gli1-mediated positive response to SHH signaling is dynamically modulated (2). The ulna and digits 2 to 5 are largely composed of cells that positively respond to SHH signaling (indicated by the green gradient domain). Over time, the responsiveness of the most posterior cells is reduced, so that the gradient of responsiveness to SHH signaling is neither stable nor linear along the anterior-posterior axis. In wild-type (Wt) limb buds, the graded distribution of GLI1 activator is opposed by an anterior-to-posterior gradient of GLI3R protein that is established by SHH-mediated inhibition of GLI3R production (12). In Gli2-deficient limb buds, the SHH-mediated positive response as measured by Gli1 expression is altered in a complex manner, as is schematically indicated by the loss of GLI1 activator in the posterior and anterior mesenchyme. However, because the limb and, in particular, digit identities are specified normally (11), these results corroborate proposals that SHH specifies digit identities mainly through inhibiting GLI3R production and establishment of an anterior-to-posterior gradient of GL3R (12, 13).

Finally, the number of Gli1–β-Gal–marked cells is reduced and their distribution is altered in limb buds of mouse embryos lacking the Gli2 gene (Fig. 1C), despite the fact that Gli2-deficient limbs develop completely normally (11). These results, together with the analysis of Gli3-deficient limb buds (2), indicate that it is not the positive response to SHH as mediated by GLI1 and GLI2 (10), but its inhibitory effects on GLI3 repressor (GLI3R) formation that determines digit identities (12, 13). In particular, the most anterior digit 1 is specified in the absence of SHH by high levels of GLI3R, whereas cumulative high levels of SHH response effectively repress GLI3R formation (12) and thereby specify the most posterior digit 5 (Figs. 2 and 3).

Taken together, these studies establish that the vertebrate limb bud is patterned mainly by a kinetic memory integrating the cumulative length and strength of autocrine and paracrine SHH signaling that cells receive. This temporal gradient patterns the posterior digits 3 to 5 and the ulna, whereas only digit 2 requires long-range SHH signaling. These processes work in combination with modulation of cellular responsiveness and inhibition of GLI3R formation over time (12, 13). Therefore, it is important to identify the molecular timekeepers that regulate the temporal kinetics of SHH signaling and modulation of mesenchymal SHH responsiveness. One candidate is the bone morphogenetic protein antagonist Gremlin, because genetic analysis in the mouse has shown that Gremlin regulates the temporal kinetics of Shh expression (14). In particular, Gremlin is essential to establish the epithelial-mesenchymal feedback signaling between ZPA and the apical ectodermal ridge (AER). In Gremlin-deficient limb buds, Shh expression is prematurely lost and mesenchymal response to SHH signaling is altered, which interferes severely with the specification of posterior digits (15, 16). Furthermore, the expansion of ZPA descendant cells over time eventually terminates Gremlin-mediated feedback signaling between ZPA and AER, which reveals a self-regulative aspect of the kinetics of morphogen signaling and response (17). Finally, it will be important to determine whether descendants of Shh-expressing and organizer cells contribute in general as substantially to embryonic tissues as the descendants of ZPA cells do during mouse limb development.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
View Abstract

Stay Connected to Science Signaling

Navigate This Article