Bone morphogenetic protein-2 (BMP-2) has been implicated in the polarizing region signalling pathway, which specifies pattern across the antero-posterior of the developing vertebrate limb. Retinoic acid and Sonic Hedgehog (SHH) can act as polarizing signals; when applied anteriorly in the limb bud, they induce mirror-image digit duplications and ectopic Bmp-2 expression in anterior mesenchyme. In addition, the two signals can activate Fgf-4 expression in anterior ridge and HoxD expression in anterior mesenchyme. We tested the role of BMP-2 in this signalling cascade by ectopically expressing human BMP-2 (hBMP-2) at the anterior margin of the early wing bud using a replication defective retroviral vector, and found that ectopic expression of Fgf-4 was induced in the anterior part of the apical ectodermal ridge, followed later by ectopic expression of Hoxd-11 and Hoxd-13 in anterior mesenchyme. This suggests that BMP-2 is involved in regulating Fgf-4 and HoxD gene expression in the normal limb bud. Ectopically expressed hBMP-2 also induced duplication of digit 2 and bifurcation of digit 3, but could not produce the mirror-image digit duplications obtained with SHH-expressing cells. These results suggest that BMP-2 may be involved primarily in maintenance of the ridge, and in the link between patterning and outgrowth of the limb bud.

In the vertebrate limb, a complex network of signalling molecules mediates patterning and outgrowth. It is important to identify the components of signalling pathways, to understand how signals control pattern and growth along the proximo-distal, antero-posterior and dorso-ventral axes of the developing limb, and to determine how individual signalling pathways are coordinated. Pattern across the antero-posterior axis is specified by a group of cells in the posterior mesenchyme of the limb bud, called the polarizing region. Mirror-image digit duplications are produced when polarizing region cells are grafted to the anterior margin of the bud (Saunders and Gasseling, 1968; Tickle et al., 1975). Anterior application of retinoic acid mimics this effect on digit pattern (Tickle et al., 1982), and is believed to induce anterior mesenchyme cells to produce a cascade of polarizing signals (Wanek et al., 1991; Noji et al., 1991). Transcripts of the Sonic hedgehog (Shh) gene are expressed in the polarizing region and are activated in anterior mesenchyme by application of retinoic acid (Riddle et al., 1993). Moreover, both SHH-expressing cells and beads soaked in SHH induce digit duplications when grafted anteriorly (Riddle et al., 1993; López-Martínez et al., 1995). Transcripts of the bone morphogenetic protein-2 (Bmp-2) gene and of members of the HoxD complex of homeobox-containing genes are also expressed in posterior limb bud mesenchyme, and are activated in anterior mesenchyme by application of retinoic acid, polarizing region grafts or SHH (Francis et al., 1994; Izpisúa-Belmonte et al., 1991; Laufer et al., 1994).

The polarizing region and a region of undifferentiated, proliferating mesenchyme cells at the tip of the bud, known as the progress zone, are maintained by signals from the apical ectodermal ridge (reviewed by Tickle, 1995). Fibrobast growth factor (FGF) family members can substitute for ridge signals. FGF-4, expressed in the posterior part of the ridge, is probably the ridge signal which maintains the polarizing activity of the posterior mesenchyme in vivo (Niswander et al., 1993). FGF-8 and/or FGF-2 could be the ridge signal which maintains the undifferentiated cells of the progress zone (Crossley and Martin, 1995; Mahmood et al., 1995). A reciprocal signal from the mesenchyme maintains the ridge, but its identity is not known. Here, we investigate the position of BMP-2 in the polarizing region signalling pathway by expressing BMP-2 in the anterior part of the limb bud.

To understand these signalling pathways we need to work out how interactions of genes are organised, upstream and downstream of SHH, and how signalling pathways are integrated. It is known that FGF-4 from posterior ridge cooperates with WNT-7a, a signal from dorsal ectoderm, to maintain Shh expression in the polarizing region (Niswander et al., 1994; Laufer et al., 1994; Yang and Niswander, 1995), and that SHH maintains Fgf-4 in posterior ridge expression via a feedback loop (Niswander et al., 1994; Laufer et al., 1994). SHH also induces expression of Bmp-2 and HoxD genes (Laufer et al., 1994), but the mechanism and significance of Bmp-2 induction is not known. It is striking that expression domains of Bmp-2, Hoxd-13 and Fgf-4 always appear to be closely linked to each other in the limb. For example, in normal chick limb buds, the mesenchymal expression domain of Bmp-2 is similar to that of Hoxd-13 (Francis et al., 1994) and these domains are close to the posterior ridge expression domain of Fgf-4 (Niswander et al., 1994). When anterior cells are respecified to form posterior structures, ectopic expression of Bmp-2 and Hoxd-13 is activated in similar regions of anterior mesenchyme (Izpisúa-Belmonte et al., 1991; Francis et al., 1994; Laufer et al., 1994), and ectopic expression of Fgf-4 is activated in the anterior ridge (Laufer et al., 1994; Niswander et al., 1994).

To further understand the link between Bmp-2, Fgf-4 and HoxD gene expression, we expressed BMP-2 ectopically in anterior mesenchyme of the early limb bud and monitored changes in expression of Fgf-4, Hoxd-11, Hoxd-13 and Shh. The results show that BMP-2 can activate ectopic expression of Fgf-4 in anterior ridge and of Hoxd-11 and Hoxd-13 in anterior mesenchyme. In addition, ectopically expressed BMP-2 induced duplication and bifurcation of digits.

Chick embryos

Fertilized White Leghorn chicken eggs were obtained from Poyndon Farm, Waltham Cross, Hertfordshire, UK. Fertilized 0 line and C line chicken eggs were obtained from the BBSRC Poultry Research Station, Compton, Berkshire, UK. Eggs were incubated at 38±1°C and staged according to the method of Hamburger and Hamilton (1951).

Construction of a recombinant retroviral plasmid encoding human BMP-2

The complete human BMP-2 (hBMP-2) coding region, and 40 nucleotides of 3′ non-coding region were excised from plasmid PS P65 (Wozney et al., 1988) by digestion with SalI, and inserted into the SalI site of adaptor plasmid Cla12 (Hughes et al., 1987). The inserted sequences were then excised by digestion with ClaI and inserted into the ClaI site of the replication defective retroviral vector pCRNCM (de la Pompa and Zeller, 1993). Clones containing hBMP-2 coding sequences in the sense (hBMP-2/pCRNCM) or antisense (Control/pCRNCM) orientation were selected, the latter for use as a negative control. The structure of hBMP-2/pCRNCM is shown in Fig. 2A.

Production of BMP-2-expressing cell lines

Infectious hBMP-2/pCRNCM and Control/pCRNCM viruses were produced in the quail packaging cell line Q2bn (Stoker and Bissel, 1988). Q2bn cells were grown in 40% (v/v) Dulbecco’s modified Eagle’s medium (DMEM), 40% (v/v) Ham’s F12 medium, 8% (v/v) fetal calf serum, 2% (v/v) chicken serum and were transfected with 10 μg of hBMP-2/pCRNCM or Control/pCRNCM DNA using Lipofectin as described by the manufacturer (Gibco/BRL). Stable transfectants were selected using 800 μg ml-1 G418 and cloned as described by de la Pompa and Zeller (1993). The titre of virus particles produced by each selected clone was determined as described by de la Pompa and Zeller (1993). The hBMP-2/pCRNCM clone and the Control/pCRNCM clone that produced the highest titres of retrovirus particles (approximately 105 ml−1) were chosen for further work.

Micromass culture of limb bud mesenchyme

Whole limb buds were dissected from stage 21/22 White Leghorn chick embryos. The ectoderm was removed and high density micromass cultures were prepared from mesenchymal cells as described by Ahrens et al. (1977). Cells were plated at 2×107 ml−1 in Ham’s F12 medium containing 10% (v/v) fetal calf serum and cultured overnight. The cultures were then supplemented with supernatant from hBMP-2/pCRNCM-or Control/pCRNCM-transfected Q2bn cells. After two more days of incubation, the micromass cultures were fixed in 2% (w/v) paraformaldehyde for 15 minutes and stained for cartilage with Alcian blue.

Grafting of BMP-2-expressing cells into chick limb buds

Spherical cell aggregates were prepared as described by de la Pompa and Zeller (1993). hBMP-2/pCRNCM-or Control/pCRNCM-transfected Q2bn cells were grown to approximately 90% confluence in 90 mm culture dishes and were trypsinized, seeded into 90 mm bacteriological Petri dishes and incubated overnight in 40% (v/v) DMEM, 40% (v/v) Ham’s F12 medium, 8% (v/v) fetal calf serum, 2% (v/v) chicken serum, without G418. The cells formed small spherical aggregates of different sizes which were then selected under the microscope. Unless otherwise stated, aggregates with a diameter of approximately 100 μm, containing 1,600±100 cells, were grafted into stage 19-20 White Leghorn or C line chick wing buds. In some experiments large aggregates with a diameter of approximately 400 μm, containing 12,000±300 cells, were used. In other experiments, cells were pelleted by centrifugation as described by Riddle et al. (1993) and then grafted. For examination of skeletal pattern, embryos were harvested on embryonic day 10, fixed in 5% (w/v) trichloroacetic acid overnight, stained for cartilage with alcian green and cleared in methyl salicylate (Tickle et al., 1982). Other embryos were harvested 24, 40 and 48 hours after grafting, fixed overnight in 4% (w/v) paraformaldehyde and processed for whole mount in situ hybridization as previously described (Francis et al., 1994).

Grafting of beads soaked in FGF-4 to the anterior margin of chick wing buds

Beads were soaked in FGF-4 as described by Niswander et al. (1993) and grafted to the anterior margin of stage 19-20 chick wing buds.

In situ hybridization to whole mounts

In situ hybridization of digoxigenin-labelled RNA probes to whole mounts was performed as previously described (Francis et al., 1994). Probes specific for chicken Bmp-2 (Francis et al., 1994), human Bmp-2 (Wozney et al., 1988), chicken Hoxd-11 and Hoxd-13 (Izpisúa-Belmonte et al., 1991), chicken Fgf-4 (Niswander et al., 1994) or chicken Shh and Bmp-7 (Francis-West et al., 1995) transcripts were prepared as described by the authors cited.

Immunohistochemistry

Cloned cell lines that had been stably transfected with hBMP-2/pCRNCM or Control/pCRNCM were grown to 70% confluence and fixed with 2% (w/v) paraformaldehyde at room temperature for 15 minutes. Immunohistochemistry was performed using the Vectastain kit (Vector Labs), following the manufacturer’s instructions. The primary antibody, a monoclonal antibody against human BMP-2 (Genetics Institute, Cambridge MA), was used at a dilution of 1/300 in PBS.

Expression of Shh, Bmp-2, Fgf-4 and Hoxd-13 genes in stage 20 chick limb buds

To compare the endogenous Bmp-2 expression domain with those of other posteriorly-restricted genes, we performed whole mount in situ hybridization to stage 20 chick wing buds.

Bmp-2 expression was found in posterior mesenchyme and throughout the apical ridge (Fig. 1A; Francis et al., 1994). The Shh domain lay within the mesenchymal Bmp-2 domain (Fig. 1B; Laufer et al., 1994), which in turn lay within the Hoxd-13 domain (Fig. 1C; Francis et al., 1994; Laufer et al., 1994). Fgf-4 was expressed in posterior ridge (Fig. 1D; Niswander et al., 1994), in a region overlying the Hoxd-13 domain and the distal part of the mesenchymal Bmp-2 domain, but lying distal and anterior to the Shh domain (Fig. 1A-D). The positions of these domains are summarised in Fig. 1E.

Fig. 1.

In situ hybridization to Bmp-2 (A), Shh (B), Hoxd-13 (C) and Fgf-4 (D) transcripts in whole mounts of normal stage 21 chick wing buds. Sketches of the relative positions of expression domains at stages 21 (E) and 26 (F) are shown. *The relative positions of the anterior boundaries of the Fgf-4 and Hoxd-13 domains have not been mapped precisely and may not be absolutely colinear. Data for F are from Fig. 4D (Fgf-4), Fig. 5A (Hoxd-13), Fig. 5C (Shh) and Francis et al. (1994) (Bmp-2).

Fig. 1.

In situ hybridization to Bmp-2 (A), Shh (B), Hoxd-13 (C) and Fgf-4 (D) transcripts in whole mounts of normal stage 21 chick wing buds. Sketches of the relative positions of expression domains at stages 21 (E) and 26 (F) are shown. *The relative positions of the anterior boundaries of the Fgf-4 and Hoxd-13 domains have not been mapped precisely and may not be absolutely colinear. Data for F are from Fig. 4D (Fgf-4), Fig. 5A (Hoxd-13), Fig. 5C (Shh) and Francis et al. (1994) (Bmp-2).

Fig. 2.

(A) Map of the genome of replication defective recombinant retrovirus hBMP-2/pCRNCM. The hBMP-2 coding sequences are in the sense orientation. Negative control virus Control/pCRNCM contained hBMP-2 coding sequences in reverse orientation. Viral transcripts are shown below the map. CMV, cytomegalovirus immediate early gene promoter. (B) Immunocytochemical staining for human BMP-2 in Q2bn cells stably transfected with Control/pCRNCM DNA. (C) Immunocytochemical staining for human BMP-2 in Q2bn cells stably transfected with hBmp-2/pCRNCM DNA. Cells were stained with a mouse monoclonal antibody against human BMP-2, followed by peroxidase-conjugated horse anti-mouse IgG.

Fig. 2.

(A) Map of the genome of replication defective recombinant retrovirus hBMP-2/pCRNCM. The hBMP-2 coding sequences are in the sense orientation. Negative control virus Control/pCRNCM contained hBMP-2 coding sequences in reverse orientation. Viral transcripts are shown below the map. CMV, cytomegalovirus immediate early gene promoter. (B) Immunocytochemical staining for human BMP-2 in Q2bn cells stably transfected with Control/pCRNCM DNA. (C) Immunocytochemical staining for human BMP-2 in Q2bn cells stably transfected with hBmp-2/pCRNCM DNA. Cells were stained with a mouse monoclonal antibody against human BMP-2, followed by peroxidase-conjugated horse anti-mouse IgG.

Fig. 3.

Localisation of replication defective hBmp-2/pCRNCM virus in whole mount preparations of limbs grafted anteriorly at stage 19/20 with hBmp-2/pCRNCM transfected Q2bn cells. Viral transcripts were detected by in situ hybridization with a probe specific for human Bmp-2 transcripts, 24 hours (A) and 40 hours (B) after grafting. Arrows show viral transcripts in grafted cells. Arrowheads show viral transcripts in host mesenchyme adjacent to the graft.

Fig. 3.

Localisation of replication defective hBmp-2/pCRNCM virus in whole mount preparations of limbs grafted anteriorly at stage 19/20 with hBmp-2/pCRNCM transfected Q2bn cells. Viral transcripts were detected by in situ hybridization with a probe specific for human Bmp-2 transcripts, 24 hours (A) and 40 hours (B) after grafting. Arrows show viral transcripts in grafted cells. Arrowheads show viral transcripts in host mesenchyme adjacent to the graft.

At stage 26, Shh (see Fig. 5C) and Bmp-2 (Francis et al., 1994) transcripts remained localised to posterior mesenchyme of the bud, with the Shh domain lying wholly within the Bmp-2 domain. Hoxd-13 transcripts were no longer restricted to posterior mesenchyme, but were found in mesenchyme right across the distal tip of the bud (see Fig. 5A). Fgf-4 expression was found in the apical ridge, with levels being high in posterior ridge and fading anteriorly (see Fig. 4D). The positions of these domains are summarised in Fig. 1F.

Fig. 4.

Distribution of Fgf-4 transcripts in whole mount preparations. (A) 24 hours after anterior grafts of hBmp-2/pCRNCM-transfected Q2bn cells at stage 19/20. (B) 24 hours after anterior grafts of Control/pCRNCM-transfected Q2bn cells at stage 19/20. (C) 40 hours after anterior grafts of hBmp-2/pCRNCM-transfected Q2bn cells at stage 19/20. (D) Normal contralateral wing of C. Grafts were placed in right wing buds, and views were from the ventral aspect to show the ridge more clearly.

Fig. 4.

Distribution of Fgf-4 transcripts in whole mount preparations. (A) 24 hours after anterior grafts of hBmp-2/pCRNCM-transfected Q2bn cells at stage 19/20. (B) 24 hours after anterior grafts of Control/pCRNCM-transfected Q2bn cells at stage 19/20. (C) 40 hours after anterior grafts of hBmp-2/pCRNCM-transfected Q2bn cells at stage 19/20. (D) Normal contralateral wing of C. Grafts were placed in right wing buds, and views were from the ventral aspect to show the ridge more clearly.

Fig. 5.

Distribution of Hoxd-11, Hoxd-13 and Shh transcripts in whole mount preparations after anterior grafts of hBmp-2/pCRNCM transfected Q2bn cells to right wing buds at stage 19/20. Dorsal views. (A) Ectopic activation of Hoxd-13 expression (arrowed), 40 hours after grafting. (B) Ectopic expression of Hoxd-11 (arrowed), 48 hours after grafting. (C) No change in Shh expression, 48 hours after grafting.

Fig. 5.

Distribution of Hoxd-11, Hoxd-13 and Shh transcripts in whole mount preparations after anterior grafts of hBmp-2/pCRNCM transfected Q2bn cells to right wing buds at stage 19/20. Dorsal views. (A) Ectopic activation of Hoxd-13 expression (arrowed), 40 hours after grafting. (B) Ectopic expression of Hoxd-11 (arrowed), 48 hours after grafting. (C) No change in Shh expression, 48 hours after grafting.

Ectopic expression of human BMP-2 in chick wing buds

The human BMP-2 (hBMP-2) coding region was inserted into the replication defective retroviral vector pCRNCM, yielding plasmids hBMP-2/pCRNCM and Control/pCRNCM (Fig. 2A), which were stably transfected into Q2bn cells. Immunocyto-chemistry with a monoclonal antibody against hBMP-2 showed that hBMP-2/pCRNCM-transfected cells synthesized hBMP-2 (Fig. 2C), while Control/pCRNCM-transfected cells did not (Fig. 2B). In addition, supernatant from hBMP-2/pCRNCM-transfected cells stimulated a marked increase in cartilage formation when added to micromass cultures of stage 21/22 chick wing bud mesenchyme cells, compared to supernatant from Control/pCRNCM-transfected Q2bn cells (data not shown). This shows that hBMP-2/pCRNCM-transfected cells secrete hBMP-2 protein that is biologically active on chicken cells (Roark and Greer, 1994).

Aggregates of hBMP-2/pCRNCM-transfected cells were grafted to the anterior margin of stage 19/20 White Leghorn wing buds. The distribution of virus was monitored by in situ hybridisation to whole mount preparations, using a species-specific probe for hBmp-2 transcripts that hybridizes to viral genomic RNA transcribed from the viral LTR and to hBmp-2 mRNA transcribed from the CMV promoter (Fig. 2A). Transcripts were restricted to the anterior margin of the bud (Fig. 3), in both grafted cells and immediately adjacent host mesenchyme, at both 24 hours (n=6, Fig. 3A) and 48 hours (n=5, Fig. 3B) after grafting. This indicates that hBMP-2/pCRNCM acts as expected for a replication-defective retrovirus and is transcribed locally at the anterior of the bud, although the levels of hBMP-2 produced by different grafts may vary.

Effects of anterior BMP-2 expression on gene expression in wing buds

Anterior expression of BMP-2 induced expression of Fgf-4 in anterior apical ectodermal ridge and of Hoxd-11 and Hoxd-13 in anterior wing bud mesenchyme. Twenty-four hours after grafting hBMP-2/pCRNCM-transfected Q2bn cells to the anterior margin of the bud, strong ectopic expression of Fgf-4 was visible in the anterior ridge (Fig. 4A; Table 1), whereas in normal wing buds (Fig. 4A), and in buds grafted with Control/pCRNCM-transfected cells (Fig. 4B; n=5), the domain of Fgf-4 expression faded anteriorly. In contrast, the grafts had no effect on Hoxd-13 expression at this time (Table 1).

Table 1.

Activation of ectopic gene expression following grafting of hBMP-2/pCRNCM transfected Q2bn cells to the anterior margin of the wing bud

Activation of ectopic gene expression following grafting of hBMP-2/pCRNCM transfected Q2bn cells to the anterior margin of the wing bud
Activation of ectopic gene expression following grafting of hBMP-2/pCRNCM transfected Q2bn cells to the anterior margin of the wing bud

After 40 or 48 hours, ectopic expression of Fgf-4 was still clearly visible in the anterior ridge of wing buds grafted with hBMP-2/pCRNCM-transfected cells, but was barely detectable in posterior ridge (Fig. 4C; Table 1). In normal wing buds, Fgf-4 expression was restricted to posterior ridge (Fig. 4D; Table 1). At this time, ectopic expression of Hoxd-13 (Fig. 5A; Table 1) and Hoxd-11 (Fig. 5B; Table 1) was also induced by grafts of hBMP-2/pCRNCM-transfected cells, whereas grafts of Control/pCRMCM-transfected cells had no effect on Hoxd-13 or Hoxd-11 expression (n=3). Ectopic Hoxd-11 and Hoxd-13 expression was found in discrete domains in anterior mesenchyme immediately adjacent and distal to the graft. In some cases, where grafts were placed closer to the tip of the bud than as shown in Fig. 5A, the ectopic Hoxd-13 domain merged with the anterior margin of the endogenous Hoxd-13 domain (n=2; data not shown).

Since BMP-2 induced ectopic Fgf-4 expression after 24 hours and ectopic Hoxd-13 expression after 40 hours, it is possible that FGF-4 mediates induction of Hoxd-13 by BMP-2. However, when we grafted beads soaked in FGF-4 to the anterior margin of the bud, we found no evidence of Hoxd-13 activation (n=4), suggesting that BMP-2 acts together with FGF-4 to induce Hoxd-13 in anterior distal mesenchyme.

Previous work has shown that SHH induces Hoxd-13 in anterior mesenchyme and Fgf-4 in anterior ridge. To determine whether ectopic expression of BMP-2 induces Shh expression, we grafted hBMP-2/pCRNCM-transfected Q2bn cells to the anterior margin of the bud and analyzed Shh expression. Shh transcripts were not detectable in grafted cells or in adjacent wing bud mesenchyme, at either 24 or 48 hours after grafting (Fig. 5C; Table 1), and neither were chicken Bmp-2 or Bmp-7 transcripts (data not shown).

Effects of anterior BMP-2 expression on digit pattern

Grafting of hBMP-2/pCRNCM-transfected cells to the anterior margin of the wing bud led to a change in bud shape, which was visible 40 and 48 hours after grafting and involved anterior extension of the bud (compare Fig. 4C and D; Fig. 5). We therefore examined skeletons of wings that developed 7 days after grafting. In 28 out of 45 cases (62%), extra digital elements were seen (Fig. 6; Table 2). In 14 of these cases there was an additional digit 2 (Fig. 6A), and 13 cases there was bifurcation of digit 3 and loss of digit 2 (Fig. 6B). In one case there was bifurcation of digit 3 without loss of digit 2 (Fig. 6C). Bifurcation of digit 3 was more often observed when grafts were positioned closer to the tip of the bud, rather than closer to the body wall.

Table 2.

Phenotypes of the wings grafted at stage 19/20 with Q2bn transfected cells with hBMP-2/pCRNCM or Control/pCRNCM

Phenotypes of the wings grafted at stage 19/20 with Q2bn transfected cells with hBMP-2/pCRNCM or Control/pCRNCM
Phenotypes of the wings grafted at stage 19/20 with Q2bn transfected cells with hBMP-2/pCRNCM or Control/pCRNCM
Fig. 6.

Effects on digit pattern of grafting hBmp-2/pCRNCM-infected Q2bn cells to the anterior margin of stage 19/20 wing buds. Whole mount preparations were stained with alcian green on embryonic day 10. Three grafted wings (A,B,C) and one normal contralateral wing from the same embryo as C (D) are shown. In A there is a duplication of digit 2. In B and C the distal phalanges of digit 3 are forked. Digit 2 is missing from B, but is present in C. 2, 3, 4 = digits 2, 3 and 4; b3 = bifurcated digit 3.

Fig. 6.

Effects on digit pattern of grafting hBmp-2/pCRNCM-infected Q2bn cells to the anterior margin of stage 19/20 wing buds. Whole mount preparations were stained with alcian green on embryonic day 10. Three grafted wings (A,B,C) and one normal contralateral wing from the same embryo as C (D) are shown. In A there is a duplication of digit 2. In B and C the distal phalanges of digit 3 are forked. Digit 2 is missing from B, but is present in C. 2, 3, 4 = digits 2, 3 and 4; b3 = bifurcated digit 3.

To see whether increasing the number of cells in the graft would increase the extent of digit duplication, we grafted 400 μm aggregates which contained approximately seven times as many cells as the 100 μm aggregates. In five out of ten cases, the larger grafts induced duplication of digit 2 similar to that seen with smaller grafts (Table 2). Similar grafts of Control/pCRNCM-transfected cells had no effect on digit development (n=3). We also grafted pellets of cells obtained by centrifugation of sub-confluent cells (Riddle et al., 1993), in case leaving cells to form aggregates led to loss of viability and activity. Again, we saw only formation of an extra digit 2, in 2 out of 6 cases.

When grafts were made to the apex or posterior margin of the wing bud, the wings were frequently truncated (Table 2). Grafts of Control/pCRNCM-transfected cells to the anterior margin, apex or posterior margin of the wing bud had no effect on digit pattern (Table 2).

The effects of hBMP-2/pCRNCM-transfected cells on gene expression and digit pattern could be mediated solely by hBMP-2 secreted by the grafted cells, or could require infection of wing bud mesenchyme cells adjacent to the graft. To distinguish between these possibilities, we grafted hBMP-2/pCRNCM-transfected cells to the anterior margin of C line chick embryo wing buds, which cannot be infected by subgroup A hBMP-2/pCRNCM virus, and monitored digit pattern as a convenient indicator of the activity of grafted cells. Duplication of digit 2 was seen in 4 out of 9 cases, demonstrating that hBMP-2 produced by the grafted cells is sufficient to induce an additional digit 2.

BMP-2 and signalling in the limb bud

Signalling by the polarizing region requires cooperation with signalling by the apical ectodermal ridge. Previous work has shown that both retinoic acid and SHH can act as polarizing signals and cooperate with FGF-4, a ridge signal, to induce Bmp-2 and HoxD gene expression in anterior mesenchyme. SHH-expressing cells can induce Fgf-4 expression in anterior ridge (Laufer et al., 1994). Our current results show that BMP-2-expressing cells also induce Fgf-4 expression in anterior ridge, suggesting that BMP-2 may mediate the induction of Fgf-4 expression by SHH (Fig. 7). Consistent with this, the endogenous expression domain of Fgf-4 is more closely related spatially to the mesenchymal Bmp-2 domain, than to that of Shh (Fig. 1). Fgf-4 expression in posterior ridge was weaker in manipulated wing buds than in normal contralateral buds, but the reason for this is unclear. A similar expression pattern is seen in Rim4 and Strong’s Luxoid mice, which have partial mirror image digit duplications and a domain of Fgf-4 expression that extends more anteriorly than normal, but fades posteriorly (Masuya et al., 1995; Chan et al., 1995).

Fig. 7.

Proposed model of molecular interactions in the polarizing region signalling pathway. FGF-4 and WNT-7a maintain SHH expression in posterior mesenchyme. SHH activates early posterior HoxD expression, and induces expression of BMP-2. BMP-2 maintains FGF-4 expression in the ridge and, in concert with FGF-4, activates late HoxD gene expression in distal mesenchyme.

Fig. 7.

Proposed model of molecular interactions in the polarizing region signalling pathway. FGF-4 and WNT-7a maintain SHH expression in posterior mesenchyme. SHH activates early posterior HoxD expression, and induces expression of BMP-2. BMP-2 maintains FGF-4 expression in the ridge and, in concert with FGF-4, activates late HoxD gene expression in distal mesenchyme.

Bmp-2 transcripts are normally present throughout the ridge, as well as in posterior mesenchyme (Francis et al., 1994). If BMP-2 activates Fgf-4, why is Fgf-4 expression normally restricted to posterior ridge? One possibility is that BMP-2 forms different species of dimers in ridge and mesenchyme, and that only the mesenchymal species regulate Fgf-4. Another possibility is that the concentration of BMP-2 produced by ridge alone is insufficient to activate Fgf-4, and that this activation requires additional BMP-2 produced by posterior mesenchyme. Niswander and Martin (1993) showed that BMP-2 inhibits FGF-4-induced outgrowth of mouse limb buds from which apical ridge has been removed. This may relate to our finding that grafting of hBMP-2/pCRNCM-transfected cells to the distal tip of the bud, immediately beneath the ridge, resulted in limb truncations.

Hoxd-11, Hoxd-12 and Hoxd-13 are activated in two distinct phases during normal chick limb bud outgrowth (Morgan and Tabin, 1994; Laufer et al., 1994). First, Hoxd-11, Hoxd-12 and Hoxd-13 are activated in response to SHH, in a nested set of domains in posterior mesenchyme. Then, the three genes become expressed in approximately the same region of distal mesenchyme, whilst posterior expression fades. It has been shown previously that retinoic acid, polarizing region grafts and SHH-expressing cells activate Bmp-2 and HoxD expression anteriorly within 24 hours (Izpisúa-Belmonte et al., 1991; Riddle et al., 1993; Francis et al., 1994; Laufer et al., 1994), with the HoxD genes being activated in nested domains corresponding to the early phase of posterior HoxD expression (Izpisúa-Belmonte et al., 1991). In contrast, we have shown here that BMP-2 requires 40 hours to activate Hoxd-13 anteriorly. This suggests that BMP-2 plays no part in regulating the early phase of HoxD expression but, rather, may be involved in regulating the later phase (Fig. 7). Most interestingly, Luo et al. (1995) have recently shown that limb buds of Bmp-7 null mice have a normal Shh domain and a posterior Hoxd-13 domain, but that Hoxd-13 expression was much reduced in the distal domain. We have previously shown that Bmp-7 is co-expressed with Bmp-2 in normal limb buds and is activated anteriorly by retinoic acid (Francis-West et al., 1995). Taken together, these data suggest that BMP-2 and BMP-7 may work together to control distal HoxD gene expression in late buds (Fig. 7).

Since grafts of BMP-2-expressing cells activated Fgf-4 before Hoxd-13, it is possible that FGF-4 itself is responsible for late activation of Hoxd-13. However, FGF-4 soaked beads did not activate Hoxd-13, suggesting that BMP-2 and FGF-4 may work together to induce HoxD expression (Fig. 7). A similar result is obtained when FGF-4 is applied in the absence of the apical ridge (Laufer et al., 1994; Niswander et al., 1994). This close relationship between Bmp-2, Fgf-4 and HoxD expression is seen in early talpid3 (ta3) mutant limb buds, where posterior HoxD genes and Bmp-2 are expressed uniformly throughout distal limb bud mesenchyme, and Fgf-4 is expressed throughout the apical ridge, instead of being posteriorly restricted as in normal limb buds (Francis-West et al. 1995). Recombination experiments have shown that the defect in ta3 limb patterning lies with mesenchyme and not with ridge (Ede, 1980). One possibility is that anarchic expression of BMP-2 in ta3 mesenchyme induces FGF-4 expression throughout the ridge, and that BMP-2 and FGF-4 then induce Hoxd-13 expression throughout distal mesenchyme. All of this occurs independently of SHH, which remains restricted to posterior mesenchyme in ta3 buds, as in normal buds (Francis-West et al., 1995). Some similar features are also found in limb buds that develop from reaggregated anterior leg bud mesenchyme (Hardy et al., 1995), where Bmp-2 and Hoxd-13 expression are found together, and the apical ridge is maintained in the absence of Shh expression.

While the simplest explanation of these data is that hBMP-2 produced by grafted cells is responsible for inducing Fgf-4 and HoxD expression in host buds, it is possible that another factor produced specifically by hBMP-2/pCRNCM-transfected Q2bn cells, and perhaps induced by hBMP-2, is responsible. Our results show that this putative factor is not likely to be BMP-7 or SHH.

Induction of digit duplication and bifurcation by BMP-2

Local mis-expression of BMP-2 at the anterior margin of the chick wing bud induced an additional digit 2 or bifurcation of digit 3, but never resulted in mirror image digit duplications. Previous work with BMP-2-soaked beads also failed to give digit duplications (Francis et al., 1994). Therefore BMP-2 alone does not appear to mediate the activity of SHH, which induces mirror image digit duplications (Riddle et al., 1993), even though BMP-2 expression is induced by SHH-expressing cells (Laufer et al., 1994). It seems unlikely that any other BMP homodimers or heterodimers would give mirror image duplications since grafts of anterior mesenchyme from ta3 mutant limb buds induce only an additional anterior digit, even though they express Bmp-2, Bmp-4 and Bmp-7 (Francis-West et al., 1995). This contrasts with development of the Drosophila wing imaginal disc, where patterning by Hedgehog, which is the Drosophila homologue of Shh, appears to be mediated entirely through activation of Dpp, which is the homologue of BMP-2 and BMP-4 (Heberlein et al., 1993; Basler and Struhl, 1994; Felsenfeld and Kennison, 1995).

Induction of extra digital elements by hBMP-2 could result from its activation of FGF-4 and/or HoxD expression. This seems more likely than the possibility that hBMP-2 produced by grafted cells has a late chondrogenic effect, since grafted cells remain in the proximal part of the bud during outgrowth, far from the site of digit formation. FGF-4 expression in anterior ridge might stimulate sufficient growth in underlying mesenchyme to result in formation of these extra digital elements. In previous work, ectopic expression of FGF-2 in anterior wing bud induced the appearance of an extra digit 2 (Riley et al., 1993). Alternatively, induction of Hoxd-11, Hoxd-13 and, presumably, Hoxd-12 expression in anterior mesenchyme by BMP-2 might also provide sufficient extra local growth to accommodate an extra digit 2 and/or bifurcation of digit 3 (Dollé et al., 1993; Davis and Capecchi, 1994; Morgan and Tabin, 1994). These changes in digit pattern are thus not due to changes in the positional identity of wing bud cells, but rather reflect the role of BMP-2 in linking growth and patterning.

We are grateful to Elizabeth Wang of the Genetics Institute (Cambridge, USA), for the human Bmp-2 cDNA clone and the monoclonal antibody against human BMP-2; to Stephen Hughes for plasmid Cla12; to Rolf Zeller for plasmid pCRNCM and Q2bn cells; to Juan-Carlos Izpisúa-Belmonte and Denis Duboule for Hoxd-11 and Hoxd-13 cDNA clones; to Juan-Carlos Izpisúa-Belmonte for the Shh cDNA clone; and to Lee Niswander for the Fgf-4 cDNA clone. We also thank Anne Crawley for technical help.

Ahrens
,
P. B.
,
Solursh
,
M.
and
Reiter
,
R. S.
(
1977
).
Stage related capacity for limb chondrogenesis in cell culture
.
Dev. Biol
.
60
,
69
82
.
Basler
,
K.
and
Struhl
,
G.
(
1994
).
Compartment boundaries and the control of Drosophila limb pattern by hedgehog protein
.
Nature
368
,
208
213
.
Chan
,
D. C.
,
Laufer
,
E.
,
Tabin
,
C.
and
Leder
,
P.
(
1995
).
Polydactylous limbs in Strong’s Luxoid mice result from ectopic polarizing activity
.
Development
121
,
1971
1978
.
Crossley
,
P. H.
and
Martin
,
G. R.
(
1995
).
The mouse Fgf-8 gene encodes a family of polypeptides and is expressed in regions that direct outgrowth and patterning in the developing embryo
.
Development
121
,
439
451
.
Davis
,
A. P.
and
Cappechi
,
M. R.
(
1994
).
Axial homeosis and appendicular skeleton defects in mice with a targeted disruption of hoxd-11
.
Development
120
,
2187
2198
.
de la
Pompa
, J. L. and
Zeller
,
R.
(
1993
).
Ectopic expression of genes during chicken limb pattern formation using replication defective retroviral vectors
.
Mech. Dev
.
43
,
187
198
.
Dollé
,
P.
,
Dierich
,
A.
,
LeMeur
,
M.
,
Schimmang
,
T.
,
Schuhbaur
,
B.
,
Chambon
,
P.
and
Duboule
,
D.
(
1993
).
Disruption of the Hoxd-13 gene induces localized heterochrony leading to mice with neotenic limbs
.
Cell
75
,
431
441
.
Ede
,
D. A.
(
1980
).
Role of the ectoderm in limb development of normal and mutant mouse (Disorganisation, Pupoid Foetus) and fowl (talpid3) embryos
.
In Teratology of the Limbs
, pp.
53
66
.
New York and Berlin
:
Walter de Gruyter and Co
.
Felsenfeld
,
A. L.
and
Kennison
,
J. A.
(
1995
).
Positional signaling by hedgehog in Drosophila imaginal disc development
.
Development
121
,
1
10
.
Francis
,
P. H.
,
Richardson
,
M. K.
,
Brickell
,
P. M.
and
Tickle
,
C.
(
1994
).
Bone morphogenetic proteins and a signalling pathway that controls patterning in the developing chick limb
.
Development
120
,
209
218
.
Francis-West
,
P. H.
,
Robertson
,
K. E.
,
Ede
,
D. A.
,
Rodriguez
,
C.
,
Izpisúa-Belmonte
,
J.-C.
,
Houston
,
B.
,
Burt
,
D. W.
,
Gribbin
,
C.
,
Brickell
,
P. M.
and
Tickle
,
C.
(
1995
).
Expression of genes encoding Bone Morphogenetic Proteins and Sonic Hedgehog in talpid (ta3) limb buds: their relationships in the signalling cascade involved in limb patterning
.
Dev. Dynam
.
203
,
187
197
.
Hamburger
,
V.
and
Hamilton
,
H.
(
1951
).
A series of normal stages in the development of the chick embryo
.
J. Morphol
.
88
,
49
92
.
Hardy
,
A.
,
Richardson
,
M. K.
,
Francis-West
,
P. H.
,
Rodriguez
,
C.
,
Izpisúa-Belmonte
,
J.-C.
,
Duprez
,
D.
and
Wolpert
,
L.
(
1995
).
Gene expression, polarising activity and skeletal patterning in reaggregated hind limb mesenchyme
Development
121
4329
4337
.
Heberlein
,
U.
,
Wolff
,
T.
and
Rubin
,
G. M.
(
1993
).
The TGFβ homolog dpp and the segment polarity gene hedgehog are required for propagation of a morphogenetic wave in the Drosophila retina
.
Cell
75
,
913
926
.
Hughes
,
S. H.
,
Greenhouse
,
J. J.
,
Petropoulos
,
C. J.
and
Sutrave
,
P.
(
1987
).
Adaptor plasmids simplify the insertion of foreign DNA into helper-independent retroviral vectors
.
J. Virol
.
61
,
3004
3012
.
Izpisúa-Belmonte
,
J.-C.
,
Tickle
,
C.
,
Dollé
,
P.
,
Wolpert
,
L.
and
Duboule
,
D.
(
1991
).
Expression of the homeobox Hox-4 genes and the specification of position in chick wing development
.
Nature
350
,
585
589
.
Izpisúa-Belmonte
,
J.-C.
,
Brown
,
J. M.
,
Crawley
,
A.
,
Duboule
,
D.
and
Tickle
,
C.
(
1992a
).
Hox-4 gene expression in mouse/chicken heterospecific grafts of signalling regions to limb buds reveals similarities in patterning mechanisms
.
Development
115
,
553
560
.
Izpisúa-Belmonte
,
J.-C.
,
Brown
,
J. M.
,
Duboule
,
D.
and
Tickle
,
C.
(
1992b
).
Expression of Hox-4 genes in the chick wing links pattern formation to the epithelial-mesenchymal interactions that mediate growth
.
EMBO J
.
11
,
1451
1457
.
Laufer
,
E.
,
Nelson
,
C. E.
,
Johnson
,
R. L.
,
Morgan
,
B. A.
and
Tabin
,
C.
(
1994
).
Sonic hedgehog and Fgf-4 act through a signaling cascade and feedback loop to integrate growth and patterning of the developing limb bud
.
Cell
79
,
993
1003
.
López-Martínez
,
A.
,
Chang
,
D. T.
,
Chiang
,
C.
,
Porter
,
J. A.
,
Ros
,
M. A.
,
Simandl
,
B. K.
,
Beachy
,
P. A.
and
Fallon
,
J. F.
(
1995
).
Limb-patterning activity and restricted posterior localization of the amino-terminal product of Sonic hedgehog cleavage
.
Curr. Biol
.
5
,
791
796
.
Luo
,
G.
,
Hofmann
,
C.
,
Bronckers
,
A. L. J. J.
,
Sohoki
,
M.
,
Bradley
,
A.
and
Karsenty
,
G.
(
1995
).
BMP 7 is an inducer of nephrogenesis and is also required for eye development and skeletal patterning
.
Genes Dev
.
9
2808
2820
.
Mahmood
,
R.
,
Bresnick
,
J.
,
Hornbruch
,
A.
,
Mahomy
,
C.
,
Colquhoun
,
K.
,
Martin
,
P.
,
Lumsden
,
A.
,
Dickson
,
C.
and
Mason
,
I.
(
1995
).
A role for FGF-8 in the initiation and maintenance of vertebrate limb bud outgrowth
.
Curr. Biol
.
5
,
797
805
.
Mason
,
I. J.
(
1994
).
The ins and outs of fibroblast growth factors
.
Cell
78
,
547
552
.
Masuya
,
H.
,
Sagai
,
T.
,
Wakana
,
S.
,
Moriwaki
,
K.
and
Shiroishi.
T
. (
1995
).
A duplicated zone of polarizing activity in polydactylous mouse mutants
.
Genes Dev
.
9
,
1645
1653
.
Mima
,
T.
,
Ohuchi
,
H.
,
Noji
,
S.
and
Mikawa
,
T.
(
1995
).
FGF can induce outgrowth of somatic mesoderm both inside and outside of limb-forming regions
.
Dev. Biol
.
167
,
617
620
.
Morgan
,
B. A.
,
Izpisúa-Belmonte
,
J.-C.
,
Duboule
,
D.
and
Tabin
,
C. J.
(
1992
).
Targeted mis-expression of Hox-4.6 in the avian limb bud causes apparent homeotic transformations
.
Nature
358
,
236
239
.
Morgan
,
B. A.
and
Tabin
,
C.
(
1994
).
Hox genes and growth: early and late roles in limb bud morphogenesis
.
Development (supplement)
181
186
.
Niswander
,
L.
and
Martin
,
G. R.
(
1993
).
FGF-4 and BMP-2 have opposite effects on limb growth
.
Nature
361
,
68
71
.
Niswander
,
L.
,
Tickle
,
C.
,
Vogel
,
A.
,
Booth
,
L.
and
Martin
,
G. R.
(
1993
).
FGF-4 replaces the apical actodermal ridge and directs outgrowth and patterning of the limb
.
Cell
75
,
579
587
.
Niswander
,
L.
,
Jeffrey
,
S.
,
Martin
,
G. R.
and
Tickle
,
C.
(
1994
).
A positive feedback loop coordinates growth and patterning in the vertebrate limb
.
Nature
371
,
609
612
.
Noji
,
S.
,
Nohno
,
T.
,
Koyama
,
E.
,
Muto
,
K.
,
Ohyama
,
K.
,
Aoki
,
Y.
,
Tamura
,
K.
,
Ohsugi
,
K.
,
Ide
,
H.
,
Taniguchi
,
S.
and
Saito
,
T.
(
1991
).
Retinoic acid induces polarizing activity but is unlikely to be a morphogen in the chick limb bud
.
Nature
350
,
83
86
.
Parr
,
B. A.
and
McMahon
,
A. P.
(
1995
).
Dorsalizing signal Wnt-7a required for normal polarity of D-V and A-P axes of mouse limb
.
Nature
374
,
350
353
.
Riddle
,
R. D.
,
Johnson
,
R. L.
,
Laufer
,
E.
and
Tabin
,
C.
(
1993
).
Sonic hedgehog mediates the polarizing activity of the ZPA
.
Cell
75
,
1401
1416
.
Riley
,
B. B.
,
Savage
,
M. P.
,
Simandl
,
B. K.
,
Olwin
,
B. B.
and
Fallon
,
J. F.
(
1993
).
Retroviral expression of FGF-2 (bFGF) affects patterning in chick limb bud
.
Development
118
,
95
104
.
Roark
,
E. F.
and
Greer
,
K.
(
1994
).
Transforming growth factor-β and bone morphogenetic protein-2 act by distinct mechanisms to promote chick limb cartilage differentiation in vitro
.
Dev. Dynam
.
200
,
103
116
.
Saunders
,
J. W.
and
Gasseling
,
M. T.
(
1968
).
Ectodermal-mesenchymal interactions in the origin of limb symmetry
.
In Epithelial-Mesenchymal Interactions
(ed.
R.
Fleischmajer
and
R. E.
Billingham
), pp.
78
97
.
Baltimore
:
Williams and Wilkins
.
Stoker
,
A. W.
and
Bissel
,
M. J.
(
1988
).
Development of avian sarcoma and leukosis virus-based vector-packaging cell lines
.
J. Virol
.
62
,
1008
1015
.
Tabata
,
T.
and
Kornberg
,
T. B.
(
1994
).
Hedgehog is a signaling protein with a key role in patterning Drosophila imaginal discs
.
Cell
76
,
89
102
.
Tickle
,
C.
,
Summerbell
,
D.
and
Wolpert
,
L.
(
1975
).
Positional signalling and specification of digits in chick limb morphogenesis
.
Nature
254
,
199
202
.
Tickle
,
C.
,
Alberts
,
B.
,
Wolpert
,
L.
and
Lee
,
J.
(
1982
).
Local application of retinoic acid to the limb bud mimics the action of the polarizing region
.
Nature
296
,
564
566
.
Tickle
,
C.
,
Lee
,
J.
and
Eichele
,
G.
(
1985
).
A quantitative analysis of the effect of all-trans retinoic acid on the pattern of chick wing development
.
Dev. Biol
.
109
,
82
95
.
Tickle
,
C.
(
1995
).
Vertebrate limb development
.
Curr. Biol
.
5
,
478
484
.
Wanek
,
N.
,
Gardiner
,
D. M.
,
Muneoka
,
K.
and
Bryant
,
S. V.
(
1991
).
Conversion by retinoic acid of anterior cells into ZPA cells in the chick wing bud
.
Nature
350
,
81
83
.
Wozney
,
J. M.
,
Rosen
,
V.
,
Celeste
,
A. J.
,
Mitsock
,
L. M.
,
Whitters
,
M. J.
,
Kriz
,
R. W.
,
Hewick
,
R. M.
and
Wang
,
E. A.
(
1988
).
Novel regulators of bone formation: molecular clones and activities
.
Science
242
,
1528
1534
.
Yang
,
Y.
and
Niswander
,
L.
(
1995
).
Interaction between the signaling molecules WNT7a and SHH during vertebrate limb development: dorsal signals regulate anteroposterior patterning
.
Cell
80
,
939
947
.