Development 126, 221-228 (1999) 221 Printed in Great Britain © The Company of Biologists Limited 1998 DEV2331

Conserved regulation of mesenchymal gene expression by Fgf-8 in face and

Abigail S. Tucker1, Abdul Al Khamis1, Christine A. Ferguson1, Ingolf Bach2, Michael G. Rosenfeld2 and Paul T. Sharpe1,* 1Department of Craniofacial Development, GKT Dental Institute, Kings College, Guy’s Hospital, London SE1 9RT, UK 2Howard Hughes Medical Institute, Department and School of Medicine, University of California at San Diego, La Jolla, California, USA *Author for correspondence (e-mail: [email protected])

Accepted 5 November; published on WWW 14 December 1998

SUMMARY

Clim-2 (NLI, Lbd1) is one of two related mouse proteins show that in the mandibular arch, as in the limb, Fgf-8 that interact with Lim-domain homeoproteins. In the functions in combination with CD44, a cell surface binding mouse, embryonic expression of Clim-2 is particularly protein, and that blocking CD44 binding results in pronounced in facial ectomesenchyme and limb bud inhibition of Fgf8-induced expression of Clim-2 and Lhx-6. in association with Lim genes, Lhx-6 and Regulation of gene expression by Fgf8 in association with Lmx-1 respectively. We show that in common with both CD44 is thus conserved between limb and mandibular arch these Lim genes, Clim-2 expression is regulated by signals development. from overlying epithelium. In both the developing face and the limb buds we identify Fgf-8 as the likely candidate Key words: LIM homeodomain, Clim-2, NLI, Ldb-1, Fgf-8, CD44, signalling molecule that regulates Clim-2 expression. We Branchial arch, Limb, Mouse

INTRODUCTION thought to be similar to, or the same as Lhx-8 (L3)), are largely restricted to anterior mesenchymal cells of the mandibular and Early embryonic development of the facial processes and limb maxillary arches (Grigoriou et al., 1998; Wanaka et al., 1997; buds has been considered to share some common basic Kitanaka et al., 1998). Removal of the oral epithelium results mechanisms despite the mesenchymal cells having different in a downregulation of Lhx-6/-7 expression, which can be embryonic origins. The mesenchymal cells that contribute to rescued by the addition of Fgf-8 protein on beads (Grigoriou et oral and facial hard tissues are derived from cranial neural crest al., 1998). Since Fgf-8 is normally expressed in oral epithelium cells whereas limb mesenchyme cells are derived from axial it is an ideal candidate for the endogenous inducer of Lhx-6/-7. . The outgrowth of facial processes has been Limb bud epithelium, a source of Fgf-8, is capable of inducing compared with limb bud outgrowth, and the tooth bud enamel Lhx-6/-7 expression in branchial arch mesenchyme cells but not knot has been identified as a signalling centre with similarities in limb bud mesenchyme cells, indicating that the neural crest to both the limb ZPA and AER signalling centres (Koyama, et origin of the mesenchymal cells determines their ability to al., 1996; Jernvall et al., 1994; Vaahtokari et al., 1996). In activate Lhx-6/-7 expression in response to Fgf-8 (Tucker et al., addition, several homeobox-containing genes have been 1999). The expression of Lmx-1 is restricted to the dorsal implicated in both branchial arch and limb development, such mesenchymal cells of the limb bud. Removal of the dorsal limb as members of the Msx, Dlx and Lim-homeobox families (Chen bud epithelium results in downregulation of Lmx-1 expression, et al., 1996; 1997; Bulfone et al., 1993; Ferrari et al., 1995; Qiu which can be rescued by ectopic expression of Wnt-7a, a gene et al., 1995; Grigoriou et al., 1998; Riddle et al., 1995). Msx-1 normally expressed in dorsal epithelium (Riddle et al., 1995). expression, for example, has been shown to be maintained in Overexpression of Lmx-1 results in limbs with a double dorsal face to limb mesenchyme grafts (Brown et al., 1993). phenotype, while loss of Lmx-1 expression, as seen in Wnt-7a The Lim domain is a specialised double-zinc finger motif mutant embryos, results in limbs with a double ventral found in a variety of proteins. Lim domains are thought to phenotype (Vogel et al., 1995; Cygan et al., 1997). Restricted function as protein interaction modules, and have important expression of both Lhx-6/-7 in facial mesenchyme and Lmx-1 functions in cell lineage determination and pattern formation in limb bud mesenchyme are thus under the control of specific (Dawid et al., 1998). Regulation of the Lim-homeobox genes but different signals from the overlying epithelium. in the face and limb has been shown to be under the control of In the mouse, the Clim family of proteins have been the epithelium. The expression of the Lim-domain gene Lhx-6, identified as interacting (co-factors) with Lim-domain proteins and its closely associated family member Lhx-7 (which is (Bach et al., 1997). Clim-2 (cofactor of Lim homeodomain 222 A. S. Tucker and others proteins 2) was identified by its ability to bind to the Lim were detected was considered as E0.5. Cultures were carried out at domain of Lhx-3 (Bach et al., 1997). It is also known in the E10.5. Mandibular arch and forelimb buds were dissected in D-MEM. mouse as Ldb-1 and NLI (Agulnick et al., 1996; Jurata et al., The epithelium was removed using Dispase in calcium- and 1996). Homologues are found in Xenopus (Xldb-1), Zebrafish magnesium-free PBS at 2 units per ml, for 10 minutes at 37°C. After (Zfldb-1-4) and Drosophila (Chip) (Agulnick et al., 1996; incubation the tissues were washed in D-MEM with 10% fetal calf Toyama et al., 1998; Morcillo et al., 1997), all of which have serum (FCS), and the epithelium was dissected using fine tungsten needles. The mesenchyme was placed on membrane filters supported been characterised by their interactions with Lim proteins. α by metal grids according to the Trowel technique as modified by Clim-2 is required for the synergistic activation of the GSU Saxén (Trowel, 1959; Saxén, 1966). For the recombinations, gene by other Lim homeodomain genes, but is unable to epithelium and mesenchyme were aligned in the correct orientation activate transcription itself (Bach et al., 1997). Thus Clim-2 is on top of transparent Nucleopore membrane filters (0.1 µm pore believed to be an essential co-activator of Lim protein function. diameter; Costar). For the bead experiments, limb buds and From work in Xenopus, it has been shown that Xldb-1 binds mandibular arches were placed on top of Millipore filters (0.1 µm pore the Lim domains through its carboxyl-terminal group and diameter) and beads added. For both these experiments it was forms homodimers via its amino-terminal region. Both regions important that the mandibular arch mesenchyme was facing oral side are necessary for the synergism of Ldb1 with Xlim-1 in the up. activation of downstream genes (Breen et al., 1998). For the Fgf and BSA control protein heparin, acrylic beads (Sigma) were used. These were washed several times in PBS, then incubated Misexpression of Clim-2 alone when injected into Xenopus overnight at 4°C in 1 µg/µl Fgf-8b, or Fgf-9 (R & D systems). For embryos had no obvious effect on development, whereas co- the BMP2 and 4 proteins, Affi-Gel-blue beads (Bio Rad) were used injection with Lhx-1 resulted in partial axis duplication, (Genetics Institute). These were washed and dried out before being indicating again that Clim-2 acts as a co-factor for LIM- placed in the protein for 1 hour at 37°C. The BMPs were used at a domain proteins (Agulnick et al., 1996). Similarly in chick concentration of 100 µg/ml, a concentration which has been shown to limb development, misexpression of Clim-2 using retroviruses induce Msx-1 and repress Pax-9 expression in a similar assay (Tucker has no effect on development (Bach et al., unpublished). et al., 1998; Neubüser et al., 1997). Protein-soaked beads were stored In the mouse, Clim-2 is detected from E8.5, at the time when at 4°C for up to 3 weeks. the Lim homeodomain genes are first starting to be expressed. The explants were cultured for 4-24 hours in D-MEM with 10% At E9.0-E9.5, high expression is seen in the brain, neural tube, FCS. A standard incubator was used at 37°C with an atmosphere of 5% CO2 in air and 100% humidity. All solutions used contained the forming branchial arches and limbs, overlapping with the penicillin and streptomycin at 20 IU/ml. After the period of culture, expression patterns of several Lim domain genes including, cultures were washed in ice-cold methanol for 1 minute then fixed in Lhx-2 (LH-2), Lhx-3 (P-Lim), Lhx-6, Lhx-7 (Lhx-8) and Lmx1 fresh 4% paraformaldehyde for 1 hour at RT. Cultures were then (Bach et al., 1997, Grigoriou et al., 1998; Wanaka et al., 1997; prepared for whole-mount digoxigenin in situ hybridisation. Kitanka et al., 1998; Riddle et al., 1995). We show that in common with Lhx-6 and Lmx-1, the epithelium is required for CD44 blocking experiments Clim-2 expression in underlying mesenchyme. However Clim- Digested oligosaccharide fragment preparations of hyaluronic acid 2 differs from the Lim-domain genes in that its expression is (HAOS) were prepared by treating hyaluronan with testicular regulated by the same signal, Fgf-8, in both the limb buds and hyaluronidase (40U/mg hyaluronan) for 4 hours in 0.2 M sodium mandibular arches. acetate and 0.15 M sodium chloride pH 5.0. The reaction was stopped by boiling for 5 minutes, which caused precipitation of the enzyme, In the limb, Fgf-8 has recently been shown to act in then spun at 3,000 rpm for 5 minutes (Rahmanian et al., 1997; Skelton conjunction with CD44, a cell surface hyaluronan binding et al., 1998). The resulting crude supernatant containing the protein (Sherman et al., 1998). CD44 is expressed in a number oligosaccharide fragments was diluted to a stock concentration of 200 of tissues, such as the and heart, which require µg/ml. Affigel blue beads were soaked in the stock supernatant for 1 hyaluronan, but is also expressed in hyaluronan-independent hour at 37°C. Beads were then added to whole mandibular arch tissues, such as the limb and oral epithelium (Wheatley et al., cultures and cultures minus epithelium which had had Fgf8 beads 1993). In these epithelia a specific splice variant of CD44, added (see above for methods). Whole mandibles were cultured in which contains the v6 exon, is highly expressed (Yu and Toole, serum-free medium, with 100 µg/ml HAOS, and 50 µg/ml apo- 1997; Sherman et al., 1998). Blocking of this specific variant transferrin (Sigma). Control mandibles were cultured in serum-free in the limb results in inhibition of Fgf-8 function (Sherman et medium with the equivalent concentration of sodium acetate and sodium chloride. The effect of the HAOS on Clim-2, Lhx-6 and Ptc al., 1998). From the expression pattern of v6CD44 it would be expression was then followed by whole-mount digoxigenin in situ predicted that the association between Fgf-8 and CD44 hybridisation. described in the limb is also occurring in the mandibular arch. We show that CD44 is closely associated with Fgf-8 in oral In situ hybridisation epithelium, and that if CD44 binding is blocked, Fgf-8 function Whole-mount digoxigenin in situ hybridisation was carried out as is inhibited, resulting in loss of both Clim-2 and Lhx-6 in described by Pownall et al. (1996). DIG in situs were vibratomed in mandibular arch mesenchyme. Thus in the face, as in the limb, 5% Agarose at 50-100 µm and mounted in 80% glycerol. v6CD44 and Fgf-8 act in concert to allow signalling from the Radioactive 35S in situ hybridisation procedures were carried out epithelium to the mesenchyme. as described by Wilkinson (1992), with the following modifications. Embryonic heads were sectioned at 8 µm and floated onto TESPA (3- aminopropyltriethoxysilane) coated slides. The slides were pre-treated with 5 µg/ml Proteinase K and 0.25% (v/v) acetic anhydride to reduce MATERIALS AND METHODS background. Hybridisation was carried out overnight in a humidified chamber at 55°C. The slides were then washed twice at high Cultures stringency (20 minutes at 65°C in 2× SSC, 50% formamide, 10 mM CD-1 strain of mice were used; noon of the day on which the plugs DTT) and treated with 40 µg/ml RNAse A for 30 minutes at 37°C to Conserved regulation in limb and face 223 remove any non-specifically bound probe. The high stringency washes (at 65°C in 2× SSC, 50% formamide, 10 mM DTT) were repeated, followed by a further wash at 65°C in 0.1× SSC, 10 mM DTT. The sections were then washed in 0.1× SSC at room temperature, and dehydrated through 300 mM ammonium acetate in 70% ethanol, 95% ethanol and absolute ethanol. The slides were air dried and dipped in Ilford K.5 photographic emulsion. Autoradiography was performed by exposing the sections in a light tight box at 4°C for 10-14 days. Slides were developed using Kodak D19, fixed in Kodak UNIFIX and counter stained with malachite green. The antisense probes used were generated from mouse cDNA clones. Clim-2 was linearised with BamHI and transcribed with T3. Lhx-6 was linearised with NotI and transcribed with T3. Fgf-8 was linearised with HindIII and transcribed with T3. Bmp-4 and Bmp-2 were linearised with EcoRI and transcribed with SP6. CD44 was linearised with NaeI and transcribed with T3. Ptc was linearised with BamHI and transcribed with T3. Fgf-9 antisense probe was generated from a rat cDNA clone, linearised with EcoRI and transcribed with SP6.

RESULTS

Expression of Clim-2 in the mesenchyme of the mandibular arch and limb buds is dependent on a signal from the epithelium Clim-2 is expressed widely in the developing embryo. In sections through the branchial arches at E9.5 Clim-2 expression was apparent in the branchial arch ectomesenchyme, in a domain that overlaps with Lhx-6 expression, but excluded from the epithelium (Fig. 1A). The expression of Lhx-6 has been shown to be dependent upon the epithelium at this stage (Grigoriou et al., 1998). To test whether Clim-2 expression is also dependent on signals from the epithelium, the epithelium was removed from E10.5 mandibular arch explants which were then cultured for 4-24 hours. After 4 hours the expression of Clim-2 in the mandibular arch mesenchyme had been Fig. 1. Expression of Clim-2 is dependent on signals from the epithelium. (A) Vibratome frontal section through the head of E9.5 completely lost (Fig. 1B). Replacement of the epithelium back mouse embryo showing Clim-2 expression in the mesenchyme of the onto the mesenchyme after 4 hours followed by culture for a first branchial arch but excluded from the overlying epithelium. further 4 hours, resulted in expression of Clim-2 being restored, (B) E10.5 mandibular arch culture minus epithelium cultured for 4 indicating that epithelium is capable of inducing and hours, showing no expression of Clim-2. (C) E10.5 mandibular arch maintaining expression (Fig. 1C). Similar experiments were culture minus epithelium cultured for 4 hours, then recombined with repeated on limb buds at E10.5, when Clim-2 was expressed mandibular arch epithelium and cultured for a further 4 hours. Note throughout the limb bud mesenchyme (Fig. 1D), overlapping rescue of Clim-2 expression. (D) Intact forelimb bud cultured from with the expression of Lmx-1 which is expressed only on the E10.5 for 4 hours. (E) E10.5 forelimb bud minus epithelium cultured dorsal side. When the epithelium was removed from the limb, for 4 hours. Note loss of Clim-2 expression. (F) E10.5 mandibular expression of Clim-2 was lost (Fig. 1E). Thus expression of arch mesenchyme minus epithelium recombined with limb bud epithelium and cultured for 4 hours, showing rescue of Clim-2 Clim-2 is dependent on signals from the epithelium, both in the expression. (G) Vibratome section through a mandibular arch mandibular arches and limb buds. recombined with limb epithelium as in F. Red dashes outline To test whether a similar signal from the epithelium was epithelium. n=15. responsible for maintenance of expression in both the mandibular arch and the limb, limb bud epithelium was recombined with mandibular arch mesenchyme. In this arch and limb bud epithelium at E10.5 (Fig. 2A-C). Fgf-8 recombination expression was again found to be restored in the (fibroblast growth factor-8) is a likely candidate since it is mesenchyme after 4 hours (Fig. 1F,G). Thus, a similar signal, expressed in the oral epithelium of the first branchial arch and found in both limb bud and mandibular arch epithelium, is limb before limb bud out growth is evident responsible for induction/maintenance of Clim-2 expression in (Heikinheimo et al., 1994; Crossley and Martin, 1995). Fgf-8 is underlying mesenchymal cells. thus expressed at the onset of Clim-2 expression. Fgf-8 continues to be expressed in the oral epithelium until the bud Fgf-8 and Fgf-9 induce/maintain expression of Clim- stage of tooth development, and is highly expressed in the AER 2 in both branchial arch and limb bud mesenchyme () of the limb (Fig. 2B). Fgf-9, a closely The expression pattern of Clim-2 was compared to the related family member, is also expressed in oral epithelium and expression patterns of genes that are expressed in both branchial weakly in limb ectoderm (Kettunen et al., 1998; Martin, 1998). 224 A. S. Tucker and others

Fig. 2. Comparison of the expression patterns of Clim-2 with potential inducers as determined by DIG in situ hybridisation at E10.5. (A) Expression of Clim-2. (B) Expression of Fgf-8. (C) Expression of Bmp-4.

Bmp-4 ( morphogenetic protein-4), another candidate, is expressed slightly later at E9.0 in oral epithelium and in early limb ectoderm (Åberg et al., 1997; Jones et al., 1991). Bmp-4 is also expressed early on in limb bud mesenchyme, becoming restricted later to the anterior and posterior margins (Jones et al., 1991; Francis et al., 1994) (Fig. 2C). Bmp-2, a closely related family member to Bmp-4, is co-expressed with Bmp-4 in oral epithelium and limb ectoderm (Åberg et al., 1998; Lyons et al., 1990) (data not shown). In the oral epithelium, Bmp-2 expression is not seen until E10.5, at least a day after Bmp-4 expression is first observed (Åberg et al., 1998). In the limb mesenchyme, Bmp-2 is expressed at the posterior margin in a region corresponding to the ZPA (zone of polarising activity), slightly after the onset of Bmp-4 expression (Jones et al., 1991). To test the ability of these signalling molecules to maintain expression of Clim-2, mandibular arches were cultured without epithelium but with beads soaked in recombinant proteins. Mandibular arch cultures without epithelium but with control BSA beads showed no expression of Clim-2 (Fig. 3A). A similar lack of expression was seen with both BMP2 and BMP4 beads, which were able to induce the expression of Msx- 1 in control assays (Fig. 3B, only BMP4 beads shown). Fgf-8 and Fgf-9 beads however, were able to maintain strong expression around the bead (Fig. 3C,D). Fgf-8 and Fgf-9 beads were also able to induce expression of the Lim homeodomain gene Lhx-6 (Grigoriou et al., 1998; data not shown). Similar experiments were carried out on the limb bud mesenchyme. No expression was seen with BSA, BMP2 or BMP4 beads (Fig. 3E), while strong induction occurred after addition of Fgf-8 or Fgf-9 beads (Fig. 3F, only Fgf-8 bead shown). It would thus appear that Fgf-8 and/or Fgf-9 are good candidates for the endogenous inducer of Clim-2 expression in the first branchial arch and limb buds. In the mandibular arch the expression of Pax-9 has been shown to be restricted through induction by Fgf-8 and Fig. 3. Induction/maintenance of Clim-2 expression in mandibular inhibition by Bmp-4 (Neübuser et al., 1997). A similar arch and limb cultures at E10.5. (A-D) Mandibular arch mesenchyme antagonism between Fgf-8 and Bmp-4 has also been noted in minus epithelium plus beads soaked in various recombinant proteins. the growth of the chick limb bud (Buckland et al., 1998). The (A) BSA control bead; no endogenous Clim-2 is present. (B) BMP4 ability of BMP4 and BMP2 to inhibit Clim-2 induction by Fgf- bead; no rescue of Clim-2 expression is apparent. (C) Fgf-8 bead; (D) 8 was tested by placing two beads together in mandibular arch Fgf-9 bead; both cultures showing strong induction of Clim-2. and limb bud mesenchyme. In both cases the induction of (E,F) Forelimb mesenchyme minus epithelium plus beads. (E) BSA control beads result in no endogenous Clim-2 expression. Clim-2 expression by an Fgf-8 bead was not affected by the (F) Induction of Clim-2 by Fgf-8 beads. (G,H) Mandibular arch presence of beads loaded with either of the other proteins (Fig. mesenchyme minus epithelium plus two adjacent beads: (G) Fgf-8 3G,H). This is perhaps not surprising since, unlike Pax-9, (white) and BMP4 (blue) beads; (H) Fgf-8 (white) and BMP2 (blue) Clim-2 has a very wide expression pattern, incorporating most beads. Note BMP2 and BMP4 do not inhibit the induction of Clim-2 of the branchial arch and limb bud mesenchyme. by Fgf-8 in the mandibular arch. n=10-15. Conserved regulation in limb and face 225

CD44 and Fgf-8 are co-localised in limb and oral variants. The expression of the two genes was found to epithelium coincide in both the oral epithelium and, as previously A specific splice variant of CD44 containing the v6 exon has indicated, the limb AER (Fig. 4A,B). In the oral epithelium the previously been shown to be expressed in the AER (apical co-expression of CD44 and Fgf-8 was especially pronounced ectodermal ridge) of developing limb buds, where it is in the more distal (i.e. presumptive incisor) regions of the associated with Fgf-8 expression (Yu et al., 1996; Yu and maxillary and mandibular processes of the 1st branchial arch Toole, 1997; Sherman et al., 1998). The same variant of CD44 (Fig. 4C,D). The fact that the expression patterns of CD44 and is also expressed in oral epithelium (Yu and Toole, 1997; Fgf-8 do not exactly overlap could act to further restrict the Sherman et al., 1998). Inhibition of CD44 function, by area of efficient Fgf-8 signalling. treatment of the AER with blocking antibodies specific to v6, interferes with the binding of Fgf-8 to its receptors in the Blocking of CD44 binding prevents Fgf8-induction of mesenchyme, leading to a failure in bud outgrowth (Sherman Clim-2 and Lhx-6 expression et al., 1998). Thus v6CD44 has an essential role in Fgf-8 The co-localisation of CD44 and Fgf-8 in oral epithelium mediated epithelial-mesenchymal interactions (reviewed by Ponta et al., 1998). Since Fgf-8 expression in oral epithelium is involved in regulation of mesenchymal genes, including Clim-2 and Lhx-6, we compared the expression of CD44 with Fgf-8 in oral epithelium. Radioactive in situ hybridisation was performed on consecutive frontal sections using Fgf-8 and CD44 riboprobes. The CD44 cDNA template used was full length and therefore the riboprobe hybridised to all splice

Fig. 5. Inhibition of Fgf-8 signalling by blocking of CD44 binding. E10.5 mandibular arch explants cultured for 24 hours. DIG in situ hybridisation for Clim-2 (A,B,C,E,F), Lhx-6 (D,G,H) and Ptc (I,J). (A,B) Mandibular arch mesenchyme minus epithelium plus adjacently placed beads. (A) Fgf-8 (white) and HAOS (blue) beads. Note induction of Clim-2 by Fgf-8 is inhibited in the cells immediately surrounding the HAOS bead. (B) Fgf-8 (white) and BSA control (blue) beads. The BSA beads have no effect on Clim-2 expression. (C,D) Whole mandibular arch with HOAS beads, showing inhibition of endogenous Clim-2 expression in the cells near Fig. 4. Co-expression of CD44 and Fgf-8 at E10.5 in facial and limb to the beads (C) and inhibition of endogenous Lhx-6 expression in epithelium. Adjacent frontal sections of E10.5 embryos were the cells near to the bead (D). (E-J) Whole mandibular arches hybridised with a CD44 riboprobe (A,C) and with an Fgf-8 riboprobe cultured in HAOS and control medium. (E,G,I) Control cultures. (B,D). (A,B) Sections showing head and body. (C,D) Close up of the (F,H,J) HAOS cultures. (F) Loss of Clim-2 expression after culture in oral epithelium. AER, apical ectodermal ridge; HT, heart; mand, HAOS. (H) Loss of Lhx-6 expression after culture in HAOS. (J) No mandibular process of the 1st branchial arch; max, maxillary process; effect on Ptc expression. n=10-20. White dashes in C,D indicate OC, oral cavity. endogenous expression domain. 226 A. S. Tucker and others suggested that the association between these proteins, mandibular arch can be rescued by recombination with previously shown to occur in early limb outgrowth (Sherman epithelium from either the limb bud or the mandibular arch et al., 1998), may also be found in mandibular arch (Fig. 1). This implies that the signal regulating Clim-2 development. To test this, we used the ability of digested expression is present in the epithelium of both developing oligosaccharide fragments of hyaluronic acid (HAOS) to limbs and mandible. compete for CD44 binding. HAOS-soaked beads were From comparisons of the expression patterns of Clim-2 with implanted into E10.5 mandibular arch mesenchyme adjacent to genes expressed in the epithelium, together with bead Fgf-8-soaked beads and cultured for 24 hours. In mandibular experiments, both Fgf-8 and Fgf-9 are likely candidates for the arch mesenchyme without epithelium the expression of Clim- endogenous inducers of Clim-2 expression (Fig. 2). Fgf-8 is 2 was downregulated but could be induced by the presence of expressed in early oral epithelium and limb bud ectoderm at an Fgf bead (see Fig. 3). When the HAOS beads were placed the same time as Clim-2 is first seen in the mesenchyme near to the Fgf beads, a zone of inhibition of Clim-2 expression (Heikinheimo et al., 1994; Crossley and Martin, 1995). Fgf-9 was seen immediately around the HAOS beads (Fig. 5A). No is expressed slightly later, and at much lower levels (data not effect was seen with BSA control beads (Fig. 5B). Thus, shown, Kettunen et al., 1998). Beads loaded with Fgf-8 or Fgf- inhibition of CD44 binding, and hence its interaction with Fgf- 9 protein induced/maintained high levels of Clim-2 in both 8, resulted in loss of Clim-2 expression. The ability of the limb bud and mandibular arch mesenchyme, while other HAOS beads to affect endogenous Clim-2 was also tested. signalling molecules, such as Bmps, had no effect (Fig. 3). HAOS beads were placed in whole mandibular arch explants Soluble dominant-negative forms of Fgf receptors expressed at E10.5 close to the putative Fgf-8 domain and cultured for during embryogenesis have been shown to produce 24 hours. In this case, endogenous Clim-2 was shown to be craniofacial and limb abnormalities, indicating the important inhibited in the area immediately surrounding the beads (Fig. role that Fgfs play in both facial and limb development (Celli 5C). A similar inhibition was also seen for Lhx-6, which can et al., 1998). Mice homozygous for null mutations in Fgf-8 die also be induced by Fgf-8 (Fig. 5D), no effect being seen for early in development prior to the formation of limbs or BSA control beads (data not shown). The degree of inhibition branchial arches (Meyers et al., 1998). However, when Fgf-8 seen was dependent on the position of the bead. Beads is specifically inactivated in the AER at E10.5 by the use of implanted deep in the mesenchyme or at a distance from the the Cre/Lox method, severe limb abnormalities are induced endogenous Fgf-8 expression domain had no effect on the (reviewed in Martin, 1988). Mice homozygous for null expression of the mesenchymal genes. This indicates that mutations in Fgf-9 have normal limbs, which may indicate that either the blocking oligosaccharide fragments were unable to Fgf-9 is redundant along with other Fgfs in limb development diffuse far from the beads, or that the Fgf-8 protein is localised (reviewed by Martin, 1998). close to the source (the oral epithelium), and exerts its long Regulation of Clim-2 expression appears to be conserved in range effects on the mesenchyme by a mechanism that does the developing face and limb. This is in contrast with Lim- not involve protein diffusion. To further test the ability of the domain proteins with which Clim-2 interacts in these regions. oligosaccharides to block Fgf-8 signalling, HAOS were added Lhx-6 is expressed in first branchial arch mesenchymal cells to the medium of whole mandibular cultures. After 24 hours and expression is induced by Fgf-8 produced from oral the expression of Clim-2 and Lhx-6 had been dramatically epithelium. However, despite the presence of Fgf-8 in the AER, reduced (Fig. 5E-H). To confirm that the effect of the HAOS no expression of Lhx-6 is observed in limb bud mesenchyme. was specific to Fgf8-inducible genes, the expression of Ptc was This is because only neural crest-derived mesenchymal cells assayed after HAOS treatment. Ptc expression is regulated by appear able to respond to the Fgf-8 signal and express Lhx-6. Shh in mandibular arch mesenchyme, and Fgf-8 has been In the limb, Lmx-1 is induced not by Fgf-8 but by Wnt-7a from shown to have no effect on its expression (Hardcastle et al., the epithelium. Thus two different signalling pathways are 1998 & unpublished observations). Expression of Ptc was involved in the regulation of the Lim proteins, with which found to be unaffected by addition of HAOS (Fig. 5I,J). Clim-2 is thought to interact. The ability of HAOS to inhibit Clim-2 and Lhx-6 expression In the limb, Fgf-8 needs a specific splice variant of CD44, indicates that CD44 and Fgf-8 do indeed have an interactive the hyaluronan receptor, in order for it to function (Sherman et role together in both mandibular arch and limb development, al., 1998). CD44 is expressed in an number of tissues during and that Fgf-8 is the likely endogenous inducer of Clim-2 in development, which can be divided into those that have been both the limb and the face. shown to require hyaluronan, such as the heart, as can be seen in Fig. 4A, and those that act independently of hyaluronan, such as limb and facial epithelia (Wheatley et al., 1993). In DISCUSSION these two epithelial tissues the highest expression of the splice variant v6 are found (Yu and Toole, 1997; Sherman et al., The mouse Clim-2 (Ldb-1, NLI) encodes a protein which 1998). In the epithelium of the limb, v6CD44 has been shown interacts with the Lim domain of a variety of Lim homeobox to be required for presentation of Fgf-8 to its receptor, rather proteins during development. We have shown that the than as a hyaluronan receptor (Sherman et al., 1998), and we expression of Clim-2 in facial ectomesenchyme and limb bud show here that a similar mechanism is acting in the mandibular mesenchyme is regulated by the overlying epithelium at E10.5. arch epithelium. CD44 and Fgf-8 are co-expressed in oral Removal of the epithelium in both areas results in rapid epithelium at E10.5 (Fig. 4). Using blocking oligosaccharide downregulation of Clim-2 expression, along with fragments (HOAS) to CD44 at E10.5, Fgf-8 signalling can be downregulation of the Lim homeodomain genes, with which inhibited, resulting in a loss of both Clim-2 and Lhx-6 Clim-2 is thought to associate. The expression in the expression (Fig. 5). This is a specific response of Fgf-8- Conserved regulation in limb and face 227 induced genes since expression of Ptc, which is part of the G. (1997). A family of LIM domain-associated cofactors confer pathway, is unaffected. This is consistent with transcriptional synergism between LIM and Otx homeodomain proteins. the fact that whereas oral epithelium will induce tooth Genes Dev. 11, 1370-1380. Breen, J. J., Agulnick, A. D., Westphal, S. and Dawid, I. B. (1998). development in non-oral branchial arch mesenchyme, non-oral Interactions between LIM domains and the LIM domain-binding protein branchial arch epithelium cannot induce tooth development Ldb1. J. Biol. Chem. 273, 4712-4717. even with the addition of exogenous Fgf-8 (Tucker et al., Brown, J. M., Wedden, S. E., Millburn, G. H., Robson, L. G., Hill, R. E., 1999). Therefore Fgf-8 alone is not sufficient to induce tooth Davidson, D. R. and Tickle, C. (1993). Experimental-analysis of the control of expression of the homeobox- gene Msx-1 in the developing limb development. The loss of expression of Clim-2 after addition and face. Development 119, 41-48. of the oligosaccharide fragments indicates that in the face, as Buckland, R. A., Collinson, J. M., Graham, E., Davidson, D. R. and Hill, in the limb, CD44 is necessary for Fgf-8 signalling, and R. E. (1998). Antagonistic effects of FGF4 on BMP induction of apoptosis provides strong evidence that Fgf-8 is the endogenous inducer and chondrogenesis in the chick limb bud. Mech. Dev. 71, 143-150. of Clim-2 in both of these tissues. Bulfone, A., Kim, H. J., Puelles, L., Porteus, M. H., Grippo, J. F. and Rubenstein, J. L. R. (1993). The mouse Dlx-2 (Tes-1) gene is expressed in Clim-2 and Fgf-8 are also co-expressed in the developing spatially restricted domains of the forebrain, face and limbs in midgestation brain. Clim-2 is found at high levels in the mesencephalon, mouse embryos. Mech. Dev. 40, 129-140. metacephalon and telencephalon (Bach et al., 1997). Fgf-8 is Celli, G., LaRochelle, W. J., Mackem, S., Sharp, R. and Merlino, G. (1998). expressed strongly at the junction between the midbrain and Soluble dominant-negative receptor uncovers essential roles for fibroblast hindbrain, and has been shown to have a role in organisation growth factors in multi-organ induction and patterning. EMBO J. 17, 1642- 1655. of the midbrain (Crossley and Martin, 1995; Crossley et al., Chen, Y. P., Bei, M., Woo, I., Satokata, I. and Maas, R. (1996). Msx1 1996). Fgf-8 is also expressed at the rostral limit of the neural controls inductive signaling in mammalian tooth morphogenesis. plate, which folds to form the telencephalic commissural plate Development 122, 3035-3044. (Crossley and Martin, 1995). Thus the expression patterns of Chen, Y. P., Satokata, I., Heaney, S., Woo, I. and Maas, R. (1997). Control of limb bud pattern formation and apoptosis by mouse Msx genes. Dev. Biol. these two genes are again very closely associated. In the brain 186, B228-B228. Clim-2 is thought to interact with a variety of Lim domain Crossley, P. H. and Martin, G. R. (1995). The mouse Fgf8 gene encodes a genes, such as Lhx-2 (LH-2) and Islet -1 (Bach et al., 1997). family of polypeptides and is expressed in regions that direct outgrowth and Later in development, and in the adult, Clim-2 is highly patterning in the developing embryo. Development 121, 439-451. expressed in the pituitary gland (Bach et al., 1997). This Crossley, P. H., Martinez, S. and Martin, G. R. (1996). Midbrain development induced by Fgf8 in the chick-embryo. Nature 380, 66-68. expression coincides with that of Lhx-3 (P-Lim), with which Cygan, J. A., Johnson, R. L. and McMahon, A. P. (1997). Novel regulatory Clim-2 is known to bind (Bach et al., 1997). From knockout interactions revealed by studies of murine limb pattern in Wnt-7a and En-1 studies, Lhx-3 (P-Lim) has been shown to have an essential role mutants. Development 124, 5021-5032. in the specification of pituitary cell lineages (Sheng et al., Dawid, I. B., Breen, J. J. and Toyama, R. (1998). LIM domains: multiple 1996). Interestingly, Fgf-8 has recently been shown to have a roles as adapters and functional modifiers in protein interactions. Trends Genet. 14, 156-162. role in pituitary development, in conjunction with Bmp-2, Ferrari, D., Sumoy, L., Gannon, J., Sun, H. L., Brown, A. M. C., Upholt, Bmp-4 and Wnt-5a (Treier et al., 1998). From the data W. B. and Kosher, R. A. (1995). The expression pattern of the distal-less presented here on limb and mandibular arch development, it homeobox-containing gene Dlx-5 in the developing chick limb bud suggests would seem likely that Fgf-8 expression is responsible for its involvement in apical ectodermal ridge activity, pattern-formation, and differentiation. Mech. Dev. 52, 257-264. induction/maintenance of Clim-2 expression in specific regions Francis, P. H., Richardson, M. K., Brickell, P. M. and Tickle, C. (1994). of the brain. Moreover the CD44/ Fgf-8/ Clim-2/Lim Bone morphogenetic proteins and a signaling pathway that controls association may also be conserved in the brain. The v6 splice patterning in the developing chick limb. Development 120, 209-218. variant of CD44 is expressed in the developing forebrain Grigoriou, M., Tucker, A. S., Sharpe, P. T. and Pachnis, V. (1998). (Sherman et al, 1998) and our in situ hybridisation analysis at Expression of Lhx-6 and Lhx-7, a novel subfamily of Lim homeodomain genes, suggests a role in mammalian head development. Development 125, E10.5 shows that CD44 expression overlaps with the sites of 2063-2074. Fgf-8 expression in both the forebrain and midbrain (data not Hardcastle, Z., Rong, M., Chi-chung, H. and Sharpe, P. T. (1998). The Shh shown). This suggests that CD44 may also be necessary to signalling pathway in tooth development: defects in Gli2 and Gli3 mutants. present Fgf-8 to neighbouring cells in the developing brain. Development 125, 2803-2811. Heikinheimo, M., Lawshe, A., Shackleton, G. M., Wilson, D. B. and MacArthur, C. A. (1994). Fgf-8 expression in the post- mouse We thank Brigid Hogan for the Bmp-2 and Bmp-4 probes, Irma suggests roles in the developing face, limbs and central nervous system. Thesleff for the Fgf-9 probe, Ivor Mason for the Fgf-8 probe, Vassilus Mech. Dev. 48, 129-138. Pachnis for the Lhx-6 probe, Claire Isacke for the CD44 probe and Jernvall, J., Kettunen, P., Karavanova, I., Martin, L. B. and Thesleff, I. Mathew Scott for the Ptc probe. We also thank the Genetics Institute (1994). Evidence for the role of the Enamel Knot as a control center in for providing the BMP2 and BMP4 protein, and Mike Ballis for mammalian tooth cusp formation – Nondividing cells express growth- supplying the HAOS. This work was supported by grants from the stimulating Fgf-4 gene. Int. J. Dev. Biol. 38, 463-469. HFSPO and MRC. Jones, C. M., Lyons, K. M. and Hogan, B. L. M. (1991). Involvement of Bone-Morphogenetic Protein-4 (Bmp-4) and Vgr-1 in morphogenesis and neurogenesis in the mouse. Development 111, 531. Jurata, L. W., Kenny, D. A. and Gill, G. N. (1996). Nuclear Lim Interactor, REFERENCES a rhombotin and Lim homeodomain interacting protein, is expressed early in neuronal development. Proc. Nat. Acad. Sci. USA 93, 11693-11698. Åberg, T., Wozney, J. and Thesleff, I. (1997). Expression patterns of bone Kettunen, P. and Thesleff, I. (1998). Expression and function of FGFs-4, -8, morphogenetic proteins (Bmps) in the developing mouse tooth suggest roles and -9 suggest functional redundancy and repetitive use as epithelial signals in morphogenesis and cell differentiation. Dev. Dyn. 210,383-396. during tooth morphogenesis. Dev. Dyn. 211, 256-268. Agulnick, A. D., Taira, M., Breen, J. J., Tanaka, T., Dawid, I. B. and Kitanaka, J. I., Takemura, M., Matsumoto, K., Mori, T. and Wanaka, A. Westphal, H. (1996). Interactions of the Lim-domain-binding factor Ldb1 (1998). Structure and chromosomal localization of a murine Lim/homeobox with Lim homeodomain proteins. Nature 384, 270-272. gene, Lhx-8. Genomics 49, 307-309. Bach, I., Carriere, C., Ostendorff, H. P., Andersen, B. and Rosenfeld, M. Koyama, E., Yamaai, T., Iseki, S., Ohuchi, H., Nohno, T., Yoshioka, H., 228 A. S. Tucker and others

Hayashi, Y., Leatherman, J. L., Golden, E. B., Noji, S. et al. (1996). variant of CD44 expressed in the apical ectodermal ridge presents to limb Polarizing activity, sonic hedgehog, and tooth development in embryonic mesenchyme and is required for limb outgrowth. Genes Dev. 12, 1058- and postnatal mouse. Dev. Dyn. 206, 59-72. 1071. Lyons, K. M., Pelton, R. W. and Hogan, B. L. M. (1990). Organogenesis Skelton, T. P., Zeng, C. X., Nocks, A. and Stamenkovic, I. (1998). and pattern-formation in the mouse – RNA distribution patterns suggest a Glycosylation provides both stimulatory and inhibitory effects on cell role for Bone Morphogenetic Protein-2a (Bmp-2a). Development 109, 833. surface and soluble CD44 binding to hyaluronan. J. Cell Biol. 140,431-446. Martin, G. R. (1998). The roles of FGFs in the early development of Toyama, R., Kobayashi, M., Tomita, T. and Dawid, I. B. (1998). Expression vertebrate limbs. Genes Dev. 12, 1571-1586. of LIM-domain binding protein (ldb) genes during zebrafish embryogenesis. Meyers, E. N., Lewandoski, M. and Martin, G. R. (1998). Generation of an Mech. Dev. 71, 197-200. Fgf-8 mutant allelic series using a single targeted mouse line carrying Cre Treier, M., Gleiberman, A. S., O’Connell, S. M., Szeto, D. P., McMahon J. and Flp recombinase recognition sites. Nature Genet. 18, 136-141. A., McMahon A. P. and Rosenfeld, M. G. (1998). Multistep signaling Morcillo, P., Rosen, C., Baylies, M. K. and Dorsett, D. (1997). Chip, a requirements for pituitary organogenesis in vivo. Genes Dev. 12, 1691-1704. widely expressed chromosomal protein required for segmentation and Trowel, O. A. (1959). The culture of mature organs in a synthetic medium. activity of a remote wing margin enhancer in Drosophila. Genes Dev. 11, Exp. Cell Res. 16, 118-147. 2729-2740. Tucker, A. S., Al Khamis, A. and Sharpe, P. T. (1998). Interactions between Neubüser, A., Peters, H., Balling, R. and Martin, G. R. (1997). Antagonistic Bmp-4 and Msx-1 act to restrict gene expression to odontogenic interactions between FGF and BMP signaling pathways: A mechanism for mesenchyme. Dev. Dyn. 212, 533-539. positioning the sites of tooth formation. Cell 90, 247-255. Tucker, A. S., Yamada, G., Grigoriou, M., Pachnis, V. and Sharpe P. T. Ponta, H., Wainwright, D. and Herrlich, P. (1998). The CD44 protein family. (1999). Fgf-8 determines rostral-caudal polarity in the first branchial arch. Int. J. Biochem. Cell Biol. 30, 299-305. Development 126, 11-51. Pownall, M. E., Tucker, A. S., Slack, J. M. W. and Isaacs, H. V. (1996). Vaahtokari, A., Åberg, T., Jernvall, J., Keranen, S. and Thesleff, I. (1996). eFGF, Xcad3 and Hox genes form a molecular pathway that establishes the The enamel knot as a signaling center in the developing mouse tooth. Mech. anteroposterior axis in Xenopus. Development 122, 3881-3892. Dev. 54, 39-43. Qiu, M. S., Bulfone, A., Martinez, S., Meneses, J. J., Shimamura, K., Vogel, A., Rodriguez, C., Warnken, W. and Izpisua Belmonte, J. C. (1995). Pedersen, R. A. and Rubenstein, J. L. R. (1995). Null mutation of Dlx-2 Dorsal cell fate specified by chick Lmx-1 during vertebrate limb results in abnormal morphogenesis of proximal first and 2nd branchial arch development. Nature 378, 716-720. derivatives and abnormal differentiation in the forebrain. Genes Dev. 9, Wanaka, A., Matsumoto, K., Kashihara, Y., Furuyama, T., Tanaka, T., 2523-2538. Mori, T., Tanno, Y., Yokoya, S., Kitanaka, J., Takemura, M. et al. (1997). Rahmanian, M., Pertoft, H., Kanda, S., Christofferson, R., Claesson LIM-homeodomain gene family in neural development. Dev. Neurosci. 19, Welsh, L. and Heldin, P. (1997). Hyaluronan oligosaccharides induce tube 97-100. formation of a brain endothelial cell line in vitro. Exp. Cell Res. 237, 223- Wheatley, S. C., Isacke, C. M. and Crossley, P. H. (1993). Restricted 230. expression of the hyaluronan receptor, CD44, during postimplantation Riddle, R. D., Ensini, M., Nelson, C., Tsuchida, T., Jessell, T. M. and Tabin, mouse embryogenesis suggests key roles in tissue formation and patterning. C. (1995). Induction of the Lim homeobox gene Lmx1 by Wnt7a establishes Development 119, 295-306. dorsoventral pattern in the vertebrate limb. Cell 83, 631-640. Wilkinson, D. G. (1992). Whole mount in situ hybridisation of vertebrate Saxén, L. (1966). The effect of Tetracyclin on osteogenesis in vitro. J. exp. embryos. In In situ Hybridisation: A practical Approach. (ed. D. G. Zool. 162, 269-294. Wilkinson), pp. 75-83. Oxford: IRL Press. Sheng, H. Z., Zhadanov, A. B., Mosinger Jr, B., Fujii, T., Bertuzzi, S., Yu, Q., Grammatikakis, N. and Toole, B. P. (1996). Expression of multiple Grinberg, A., Lee, E. J., Huang, S.-P., Mahon, K. A. and Westphal, H. CD44 isoforms in the apical ectodermal ridge of the embryonic mouse limb. (1996). Specification of pituitary cell lineages by the Lim homeobox gene Dev. Dyn. 207, 204-214. Lhx-3. Science 272, 1004-1007. Yu, Q. and Toole, B. P. (1997). Common pattern of CD44 isoforms expressed Sherman, L., Wainwright, D., Ponta, H. and Herrlich, P. (1998). A splice in morphogenetically active epithelia. Dev. Dyn. 208, 1-10.