Digital dialogues in Dundee: 8th International Conference on Limb Development



Developmental Dynamics, 2005. © 2005 Wiley-Liss, Inc.


Few disciplines in developmental biology have such a long-standing tradition as the study of limb development and regeneration. Since the early 1970s, avid researchers interested in these topics have periodically gathered to review the scientific progress and discuss the major advances in the field. The 8th International Conference on Limb Development and Regeneration held on the 16th–18th of July in Dundee was the most recent addition to this excellent tradition. The true stars of this meeting were the mutants. These are the X-Men (or perhaps the X-Mice) of the scientific endeavor bestowed with the mutations that have the power to shed new light on limb development. Although we still have much to learn from single mutations, in the world of mouse genetics, sometimes loss of a single gene function is not enough. Several of the talks included analysis of double and even triple loss-of-function mutations, whereas conditionals and misexpression also played major roles. These innovations are pushing mouse genetics to its breeding limits and, fortunately, our understanding to a new level.

But of course, the mouse was not the only system discussed. The traditional experimental system for studying limb development is the chick, which was well-represented at the meeting and continues to contribute to our greater knowledge. In addition, progress in dissecting the control of limb development is being made in less-conventional systems. For example, the initial positioning of the limb along the flank of an embryo is being examined in other organisms such as salamanders and even in the common aquaria fish, the cichlid (T. Stephens, Idaho State University). Insect models such as Drosophila (J.P. Couso, University of Sussex) and Gryllus (cricket; S. Noji, University of Tokushima) continue to contribute information. The latter species is making inroads into our understanding of the processes of limb regeneration through the direct application of dsRNAs in developing cricket larva. And perhaps we were also presented with a glimpse into the future in which computer-generated models may soon provide a framework for comprehending gene networks and limb morphogenesis (J. Sharpe, MRC-Human Genetics Unit, Edinburgh).

The quality of the presentations was of the highest standard, and the meeting was jammed full of information with a total of 52 talks and 22 posters. We obviously cannot discuss each presentation within our allotted space, so we have concentrated on a few that highlight the major themes. The abstracts for all the presentations can be obtained on the meeting Web site (


The process of limb bud initiation is invariably the first issue to be discussed at a limb meeting. The questions that motivate this discipline are: What are the genes that initiate the process of limb development, and are these the same that specify the identity of the limb (i.e., whether the limb is an arm or a leg)? Recent dogma suggests the Tbx genes are key: Tbx5 specifies forelimb identity and Tbx4 the hindlimb (Rodriguez-Esteban et al., 1999; Takeuchi et al., 1999). This view is underscored by experiments in the chick in which appropriate misexpression of the tbx genes lead to respecification of limb identity (T. Ogura, Tohoku University). Contradictory evidence in the mouse, however, suggests that the role of Tbx5 and Tbx4 stop short of identity specification and function primarily in the initiation process. In an elegant series of experiments, M. Logan (MRC-National Institute of Medical Research, London) showed that loss of Tbx5 results in the lack of the forelimbs and, therefore, is important in limb initiation. Re-supplying Tbx5-deficient forelimb with Tbx4 rescued the limb, but the forelimb identity remained unchanged. However, Tbx4 coexpressed with Pitx1, also normally found in the hindlimb bud, in the Tbx5-deficient forelimb generates a limb with characteristics of the hindlimb. J. Drouin (Clinical Research Institute of Montreal) complemented these data with the analysis of the hindlimb in Pitx1 mutant embryos. In these mutants, Tbx4 expression is reduced and specification of hindlimb characteristics is lost, leading to mutants in which the knee joint resembles that of the elbow. Clearly, hindlimb identity is dependent on Pitx1, but it is noteworthy that mice mutant for both Pitx1 and Pitx2 are also deficient in hindlimb bud outgrowth without further reduction of Tbx4 expression compared with Pitx1 mutant mice. Thus, both Pitx genes have redundant activity for control of hindlimb bud outgrowth similar to the role of Tbx5 in forelimb buds.

The apical ectodermal ridge (AER) appears soon after limb bud initiation and is required for outgrowth. S. Bell (Cincinnati Children's Hospital Research Foundation) described two new genes, Sp8 and Ftl, responsible for the legless (lgl) and footless (ftl) transgene insertional mouse mutations. Gene targeting studies indicate that Sp8 is required for AER formation. A putative downstream target of Sp8 is the Ftl gene. Preliminary studies indicate that both genes play a role in the Wnt signaling pathway. As is evident below, Wnt signaling is important in several developmental and differentiation processes in the limb. However, using animal models to understand the Wnt pathway in limb development will not be straightforward, because functional differences between presumed Wnt orthologues in mouse and chick exist (S. Sasaoka, Kawaski Medical School). The AER is also a source of fibroblast growth factors (FGFs), specifically expressing 4 of the 22 FGFs that are known in mouse (Sun et al., 2002). G. Martin (University of California, San Francisco), using conditional mouse mutations, showed that Fgf8 function predominates in the AER. Whereas Fgf4 can replace this function, Fgf8's importance is due mainly to the fact that it is expressed earliest. This early expression is required to determine limb bud size, and as a consequence, the skeletal progenitor population. Xin Sun (University of Wisconsin) contributed to this discussion by examining the putative feedback loops between the AER and Shh expression. Using a combination of mutations, which included going to the lengths of generating a triple gene loss of Fgf4, Fgf8, and Gli3, she provided support for a Shh independent mode of action for the FGF's during proximodistal limb outgrowth. In contrast, no evidence for an FGF-independent pathway for Shh was found.


SHH signaling is central to skeletal patterning in the limb and has a crucial function in regulating the relative balance of the different forms of the GLI3 protein. Full-length GLI3 is a transcriptional activator, whereas the processed, shorter form acts as a repressor (GLI3R). SHH suppresses the processing of GLI3 such that, in the Shh null limb bud, GLI3R predominates (Wang et al., 2000; Litingtung et al., 2002; te Welscher et al., 2002). The principal theme to several talks examining limb patterning focused on this Shh/Gli3 pathway. M. Ros (Universidad de Cantabria, Santander) examined the relationship between the lack of Shh and the extensive anterior apoptosis that occurs to generate a very prominent narrow limb bud. The resulting increase in GLI3R mediates the induction of bone morphogenetic protein 4 (BMP4) in the distal limb bud, which in turn is responsible for this increase in cell death. U. Ruther (Heinrich-Heine University, Düsseldorf), on the other hand, implicated an important role for the full-length activator form of GLI3. Although GLI3R is essential for AP patterning of the limb, it is the amount of full-length GLI3 that governs the number of digits that are formed.

The 5′ genes of the HoxA and HoxD gene clusters have fundamental roles in patterning the limb (Nelson et al., 1996). Is it possible that these genes interact or coordinate the Shh/Gli3 pathway? Genetic assessment of the role of the Hox gene clusters is complicated by the sheer number of genes involved and not least by the extensive redundancy and synergy between Hox gene products. M. Kmita (University of Geneva) presented the analysis of conditional deletions that removed both the HoxA and HoxD clusters. In the absence of Hoxa and Hoxd gene function, limb development is severely impaired such that the limb is restricted to a small and ill-formed humerus-like structure. Removal of all the 5′ genes in the HoxA;HoxD double mutant leads to down-regulation of Shh showing a functional requirement of these hox genes upstream of Shh. Conversely, late phase expression of the hox genes is down-regulated in the Shh−/− mutant, implicating the presence of a regulatory feedback loop. Further interactions with the Shh/Gli3 patterning pathway were shown by S. Mackem (National Cancer Institute, Bethesda, MD). Hoxd12, and indeed several 5′ Hoxd gene products, directly interacts with both GLI3 full-length and repressor forms. This interaction converts GLI3R to an activator of its target promoters, effectively modulating the functional role of GLI3R and affecting digit identity.

Gremlin, an antagonist of BMP, is another component of the network of genes that regulate Shh expression in the limb bud (Zuniga et al., 1999; Khokha et al., 2003). A. Zuniga (University of Basel Medical School) presented a fascinating relationship between Gre and the ld (limb deformity) mutation. ld is associated with several chromosomal deletions in the C-terminal coding portion of the formin gene. These deletions were shown to result in the loss of a Gre regulatory element situated in an intron of the formin gene. Thus, the ld phenotype is not caused by the disruption of formin, as suspected for so long (Mass et al., 1990), but by alteration of a global control region required for Gre expression. Another unexpected conclusion concerning limb patterning was presented by S. Nissim (Harvard Medical School, Boston, MA) on the genetics of the limb-expressed BMPs. Double mutant embryos, null for Bmp2;Bmp7 and for Bmp2;Bmp4, suggested that BMP signaling is neither required for Shh function nor as a signal for digit pattern formation.


Once the patterning mechanisms have run their course and are long forgotten by the embryo, a great deal of differentiation of chondrocytes and connective tissues, migration of muscle, and innervation continues into later stages. Perhaps, as expected, the Wnt genes play a major role in these differentiation processes in limb. Y. Yang (NIH, Bethesda, MD) presented work that showed the intimate association of the Wnt genes with maturation of chondrocytes (Wnt5a) and in joint formation (Wnt14). During myocyte formation and muscle fiber-type differentiation, the Wnt genes once again are major players (P. Francis-West, King's College London). Wnt 5a, using a noncanonical pathway, is associated with slow fiber-type formation, whereas Wnt11 is associated with fast fiber-type formation. Data were presented showing that Wnts regulate both myoblast number and terminal differentiation. In addition, it was shown that a candidate downstream target of Wnt signaling, Pitx2, regulates myocyte number.

V. Zuzarte-Luis (Universidad de Cantabria, Santander) highlighted the role of the extracellular elastic matrix during limb skeletogenesis and provided evidence for a function of the amorphous (elastin) and microfibrillar (fibrillins) components during cartilage condensation, at the stage of prechondrogenic core formation and differentiation. A. Kuroiwa (Nagoya University) presented work that suggested a relationship between Six2 and tendon precursor development and concluded that both Hox genes and signals from adjacent tissues involving the Wnt genes act in concert to regulate Six2 expression. B. Mankoo (King's College, London) examined the mechanism by which the elaborate pattern of the limb musculature is obtained. Meox2 not only affects muscle differentiation but formation of some tendons. Similar to Kuriowa's work, Meox is required for Six2 expression in these precursors.


A large number of different limb abnormalities are known in human, some of which are specific to the limb and others with associated dysmorphologies. The identification of the genes involved has been an important step in understanding the developmental mechanisms, which are surprisingly diverse in their activity. For example, M. Hajihosseini (University of East Anglia, Norwich) showed that mice harboring a Pfeiffer-syndrome-type hypermorphic FgfR1 mutation show Digit 1-specific defects, involving perturbed cell death and tissue patterning. Crucially, these defects arise in a threshold-dependent manner, suggesting that the limb anterior normally experiences low levels of FgfR1 activity. Townes-Brocks syndrome results from truncating mutations in Spalt1 (SALL1; Kohlhase et al., 1998), which may be dominant-negative mutations. D. Sweetman (University of East Anglia, Norwich) presented data suggesting that the mutant form of the normally nuclear csal1 is mislocated into the cytoplasm and also relocalizes full-length spalt protein, a possible mechanism for a dominant-negative effect. L. Lettice (MRC-Human Genetics Unit, Edinburgh) presented the basis for preaxial polydactyly (PPD) in human, which is due to misexpression of Shh. Point mutations in the limb-specific enhancer activate ectopic expression of Shh. The point mutations within the enhancer are located 1 Mb away from the gene affected. Identification of the enhancer made possible a targeted deletion (L. Lettice and T. Sagai, National Institute of Genetics, Japan), which generated a phenotype reflecting limb abnormalities in the Shh−/− embryo. Of interest, S. Maas (University of Wisconsin) showed that a spontaneously derived chick mutant, oligozeugodactyly, which has lost specific bones in the wings and legs, resulted from a deletion that corresponds to the mouse Shh enhancer region. Cornelia de Lange syndrome (CdLS) appears to result from more-traditional loss-of-function mutations. CdLS has several features, including variable defects of the upper and lower limbs. E. Tonkin (University of Newcastle upon Tyne) took advantage of a patient chromosomal translocation to isolate a 47-exon–containing gene (called NIPBL) involved in the syndrome. NIPBL, encodes the orthologue of the Drosophila Nipped-B protein involved in both sister chromatid cohesion and developmental gene regulation.


Of long-standing interest to both paleontologists and developmental biologists is the remarkable evolutionary transition from the fins of fishes to the limbs of the land-dwelling tetrapods. Several talks examined important models for understanding the developmental genetics of this transition, including a basal actinopterygian (the paddlefish, B. Metscher, University of Southern Indiana) and a representative of the lineage of fish that are our closest living nontetrapod relatives, the dipnoans (G. Stopper, Yale University). Chondrichthyan limb characteristics recently have become of interest in uncovering the molecular basis of fin evolution. M. Coates (University of Chicago, IL) suggested that modern sharks (elasmobranchs) appeared approximately 200 Myr ago, whereas the chondrichthyans and the bony fishes shared their last common ancestor nearer 500 Myr ago. Many morphological features, such as the out-turned fins, therefore, may be modern elaborations having arisen in the elasmobranchs. Thus, conclusions about fin evolution using extant species should be taken with caution.

At the end of the meeting, there was an excitement about the current state of play in the field. Indelibly, the mouse geneticists left their mark on this meeting. The 2004 Limb conference truly reflected the progress made in deciphering the facets of appendage formation and highlighted the trends in technical approaches that are currently popular in the developmental field. These tools will be exploited and improved, and we look forward to the next meeting in 2 years to share further advances in our understanding of vertebrate limb morphogenesis.