Targeting the Fgf8 locus. a, The wild-type Fgf8 locus is depicted on the top line, exons are shown as green boxes and are identified numerically. Fgf8^C targeting vectors were constructed by inserting loxP sites in untranslated regions flanking exon 5; Cre-mediated recombination deletes all sequences between the two loxP sites and generates the inactive allele, Fgf8^CR. The 5′ loxP site (red arrowhead) was inserted into a SmaI site (S) located 147 bp 5′ of exon 5. A cassette containing frt-flanked pol II/neo^r, loxP followed by a splice acceptor and coding sequences for human alkaline phosphatase (AP, yellow arrow) was inserted into a SpeI (Sp) site, 230 bp downstream of the translation stop. The frt-flanked pol II/neo^r gene was subsequently deleted from the Fgf8^C allele by exposure to Flp recombinase (leaving a residual Frt site, purple line). Cre-mediated recombination of this AP-containing allele not only inactivates Fgf8, but also causes AP to be expressed under control of the Fgf8 promoter as a reporter of the recombination event in cells in which Fgf8 is transcribed. The null allele, Fgf8^N, was generated by Cre-mediated recombination of a hypomorphic conditional allele (not shown) in the germ line. b, Genotyping was performed by PCR analysis on DNA prepared from yolk sacs or tails using the primers shown in (a) (black arrowheads). P1 and P2 distinguish the wild-type (194 bp) and Fgf8^C (246 bp) alleles. The product amplified with P1 and P3 from the Fgf8^N allele is 149 bp, whereas that amplified using P1 and P4 from the Fgf8^CR allele is 355 bp.

Fgf8 expression and alkaline phosphatase activity in embryos bearing mutant alleles of Fgf8. In this and all subsequent figures, Fgf8 genotypes are listed at the top of the column. +, wild-type allele; N, null allele; C, conditional allele; CR, Cre-recombined conditional allele. All in situ hybridizations were performed with stage-matched littermates whenever possible and control and mutant embryos were hybridized in the same vials. a, Fgf8 expression in E10.5 embryos. Top, whole embryos; bottom, higher magnification of forelimb buds. Fgf8 transcripts were detected in hindlimb (blue arrow) but not forelimb AER (red arrow) of Fgf8^CR/N conditional mutants. Fgf8^C/N embryos (no RARCre transgene) have detectable Fgf8 mRNA in both limb buds and normal bud morphology. Note small embryo size and abnormal head shape indicating abnormal hindbrain development (green asterisks) of Fgf8^C/N and Fgf8^CR/N animals derived from the hypomorphic conditional allele. b, Fgf8 expression in E11.5 embryos. Note stepwise decrease in the amount of Fgf8 transcripts with additional mutant alleles. The conditional mutant Fgf8^CR/N forelimb bud is small, abnormally shaped and lacks Fgf8 transcripts (red arrow). c, Alkaline phosphatase activity in E10.5 embryos. There is no staining detectable in the absence of Cre recombinase (Fgf8^C/+). Expression of RARCre (Fgf8^CR/+ or Fgf8^CR/N) resulting in recombination places AP coding sequences under control of the Fgf8 promoter. AP enzymatic activity occurs only in those cells that express Fgf8 and in which Cre-mediated recombination has occurred: the signal is strong and uniform in the forelimb AER (red arrow), whereas only a subset of hindlimb AER cells are stained (blue arrow)

Morphology of Fgf8 conditional mutants. a, Gross morphology of newborn and postnatal day 1 animals and forelimbs. Conditional mutants (Fgf8^CR/N) display severe forelimb deformity resulting from ablation of Fgf8 function during forelimb development. b, Skeleton preparations of limbs of newborns. The radius and first digit were absent from 100% (27/27) of conditional mutants (red arrowheads). Most also lacked the humerus (blue arrowhead; 19/27, 70%) and a second digit (14/27, 52%). At low penetrance (2/18, 10%), animals generated with the hypomorphic conditional allele lacked the first hindlimb digit, indicating that formation of this structure is sensitive to decreased levels of Fgf8; in animals bearing the nonhypomorphic conditional allele, this digit was always present. r, radius; h, humerus; s, scapula; c, clavicle; u, ulna. c, Photomicrographs of E10.0 and E11.5 forelimb buds from conditional mutants and control littermates. Arrows point from anterior to posterior. Fixed, dehydrated embryos were briefly stained with eosin and haematoxylin to improve contrast. The deficiency in outgrowth and anterior mesenchyme is already apparent by E10.0, consistent with onset of Cre activity in the forelimb before E9.5.

Fgf8 regulates Fgf4 expression in the AER. Expression of Fgf4 in E10.5 (a) and E11.5 (b) embryos was detected by in situ hybridization. E11.5 conditional mutants (n=5) displayed increased intensity of signal and disorganization of the AER (red arrows) compared with controls (n=20). c, Proposed regulatory mechanism between Fgf4 and Fgf8 mediated by a repressor effect of Fgf8. This model is consistent with observations that a decrease (Fgf8^C/N) or absence (Fgf8^CR/N) of Fgf8 leads to increased expression of Fgf4 in the AER.

Abnormal morphology and outgrowth of Fgf8 conditional mutant forelimb buds results from increased apoptosis and decreased Fgf10 expression. a, TUNEL assay on E10.5 embryos (n=5 mutants) reveals increased programmed cell death in the AER, anterior (blue arrow) and proximal (yellow arrow) limb bud mesenchyme, and somites of conditional mutants. Fgf10 expression is decreased in the anterior mesenchyme of conditional mutant forelimbs (red arrows) at E10.5 (b) and E11.5 (c) (n=4 each). Expression in the posterior forelimb mesenchyme is also decreased at E11.5 (blue arrows); there is a central region with relatively intact expression. Hindlimb expression seemed to be normal.

Shh expression is not maintained in the ZPA of Fgf8 conditional mutants, but asymmetry of Bmp2 expression is preserved. a, At E10.5 Shh expression appears intact. b, By E11.5 a decreased amount of Shh mRNA is detected in the forelimbs of conditional mutant embryos (red arrows; n=6). c, Bmp2 expression appears normal or slightly increased in conditional mutant forelimbs at E10.5. d, By E11.5 there is an increased Bmp2 signal in the AER (n=3).