The pleiotropic mouse phenotype extra-toes spotting is caused by translation initiation factor Eif3c mutations and is associated with disrupted sonic hedgehog signaling

Abstract

Polydactyly is a common malformation and can be an isolated anomaly or part of a pleiotropic syndrome. The elucidation of the mutated genes that cause polydactyly provides insight into limb development pathways. The extra-toes spotting (Xs) mouse phenotype manifests anterior polydactyly, predominantly in the forelimbs, with ventral hypopigmenation. The mapping of Xs^J to chromosome 7 was confirmed, and the interval was narrowed to 322 kb using intersubspecific crosses. Two mutations were identified in eukaryotic translation initiation factor 3 subunit C (Eif3c). An Eif3c c.907C>T mutation (p.Arg303X) was identified in Xs^J, and a c.1702_1758del mutation (p.Leu568_Leu586del) was identified in extra-toes spotting-like (Xsl), an allele of Xs^J. The effect of the Xs^J mutation on the SHH/GLI3 pathway was analyzed by in situ hybridization analysis, and we show that Xs mouse embryos have ectopic Shh and Ptch1 expression in the anterior limb. In addition, anterior limb buds show aberrant Gli3 processing, consistent with perturbed SHH/GLI3 signaling. Based on the occurrence of Eif3c mutations in 2 Xs lines and haploinsufficiency of the Xs^J allele, we conclude that the Xs phenotype is caused by a mutation in Eif3c, a component of the translation initiation complex, and that the phenotype is associated with aberrant SHH/GLI3 signaling.—Gildea, D. E., Luetkemeier, E. S., Bao, X., Loftus, S. K., Mackem, S., Yang, Y., Pavan, W. J., Biesecker, L. G. The pleiotropic mouse phenotype extra-toes spotting is caused by translation initiation factor Eif3c mutations and is associated with disrupted sonic hedgehog signaling.

Investigation into the genetic causes of congenital malformations in animal models provides insight into the mechanisms of development and is useful to improve understanding of human birth defects. The extra-toes spotting (Xs) mouse manifests anterior polydactyly, predominantly of the forelimbs, and/or ventral coat hypopigmentation. Xs is a semidominant phenotype that originated in an ethylnitrosourea-mutagenized mouse line (1). The locus for Xs was mapped to mouse chromosome 7, flanked by the chinchilla (c^ch) and pink-eyed dilution (p) loci, which proved that it was not allelic to Gli3^Xt, a phenotype with which it shares a number of phenotypic attributes. Further genetic mapping of another allele of Xs, Xs^J (Jackson allele), identified a 3.5-cM region on mouse chromosome 7 (2).

The extra-toes (Gli3^Xt) mutant mouse is an animal model for the Greig cephalopolysyndactyly syndrome (GCPS; refs. 3–6). Gli3^Xt is a semidominant phenotype that arose from a spontaneous mutation and comprises hindlimb anterior polydactyly, forelimb anterior and posterior polydactyly, hemimelia or shortened limb length, and ventral hypopigmentation (7). GLI3 is bifunctional, operating either as a repressor or activator of transcription, regulated by sonic hedgehog (SHH). The GLI3 protein is a transcriptional effector of the SHH signaling pathway (8–10). SHH signaling determines whether cytoplasmic GLI3 protein is translocated to the nucleus intact to activate gene transcription or is instead proteolytically processed into a truncated transcriptional repressor form.

Considering the phenotypic overlap of Xs and Gli3^Xt, we hypothesized that the Xs phenotype was caused by a mutation in a gene that affects SHH/GLI3 signaling. To test this hypothesis, we further delineated the phenotypic overlap of Xs and Gli3^Xt, positionally cloned the gene mutated in Xs, analyzed the expression of key genes in the SHH/GLI3 pathway, and tested the effect of the Xs mutation on GLI3 processing.

MATERIALS AND METHODS

Mouse husbandry to generate B6C3/J-Xs^J mice

Xs^J/+ mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA; stock #001750). Animal work was done in the National Human Genome Research Institute's animal facility according to an approved animal care and use protocol.

Mouse husbandry for linkage mapping

Xs^J/+ mice were maintained on a C57BL/6 × C3H/HeJ (B6C3/J) background. B6C3/J-Xs^J/+ mice were crossed to CAST/EiJ wild-type mice to increase heterozygosity for linkage mapping. For genotyping, DNA was isolated from tail biopsies using standard methods. Since Xs^J arose on C3H/HeJ, animals were typed for the C3H/HeJ alleles in the Xs region as described by Ko (2). D7Mit330 and D7Mit105 flank the locus defined by Ko and have distinct alleles on C57BL6/J, C3H/HeJ, and CAST/EiJ chromosomes. B6C3/J-Xs^J/+ × CAST/EiJ (G1) heterozygotes were crossed to B6C3F1/J wild-type mice. G2 mice were examined for ventral hypopigmentation and/or digit malformation. Because Xs^J was incompletely penetrant, an affected-only analysis was performed (Supplemental Fig. S1).

Mouse phenotyping

Mice were examined manually for digit malformations and hypopigmentation. Limb lengths were measured on animals generated from the B6C3/J-Xs^J/+ × B6C3F1/J cross. Limb lengths were measured from the proximal end of the humerus (forelimb)/femur (hindlimb) to the distalmost digital tip of all limbs, using a ruler.

Embryo dissections

B6C3/J-Xs^J/+ × B6C3F1/J wild-type timed matings were set up with standard methods. Yolk sacs were used to isolate genomic DNA. Embryos for in situ analysis were fixed in 4% formaldehyde (Ted Pella, Redding, CA, USA) overnight at 4°C with constant rocking. After tissue was fixed, embryos were washed 3 times in cold 1× PBS for 5 min. Embryos were stored in 100% methanol at −20°C before in situ hybridization. Embryos for cDNA purification were dissected and stored at −20°C in RNAlater (Ambion, Foster City, CA, USA) before RNA isolation.

Blastocyst flushes

Xs^J/+ × Xs^J/+ timed matings were established by standard methods, and blastocysts were collected at embryonic day (E) 3.5. Blastocysts were identified under a stereomicroscope, washed in 1X PBS, and placed in 15 μl of 1× PCR buffer (Applied Biosystems, Foster City, CA, USA). The blastocysts were heated to 98°C for 15 min, and the recovered DNA was used for genotyping.

DNA isolation

DNA isolation was performed using standard lysis and phenol-chloroform extraction.

Hydrolysis probe assay for Xs^J genotyping

Fluorescent hydrolysis probe genotyping (TaqMan; Applied Biosystems) was used to screen for Eif3c^Xs-J/+ mice on the B6C3Fe/J background. A fluorophore-quenching probe was designed to screen directly for the Eif3c^Xs-J mutation, detecting both the wild-type allele (VIC-CCACCTCgGACCCG) and Eif3c^Xs-J allele (FAM-CCACCTCaGACCCG). The following PCR primers flanked the fluorophore-quenching probe: forward, GATGAGGAAGAGGAGGACAATGAG; and reverse, CCATGCTCACCTTAACAAGTGGTA. Reactions consisted of 12.5 μl 2× TaqMan Universal PCR master mix (Applied Biosystems), 0.625 μl 40× assay mix containing probes and primers, and 10 ng genomic DNA. Reaction volumes were brought to 25 μl with dH₂O. PCR and fluorescence detection were carried out on a real-time PCR machine (ABI 7500; Applied Biosystems). Cycling conditions were 10 min at 95°C, then 40 cycles of 92°C for 15 s and 60°C for 1 min. The relative fluorescence of probes was used to determine genotype.

Short tandem repeat polymorphism genotyping

Fluorescently labeled primers (forward primers were labeled) were designed to amplify known MIT markers in the Xs region. Twenty-two additional primer pairs were designed to amplify novel repeats. The Tandem Repeat Finder computer program (http://tandem.bu.edu/trf/trf.html) was used to screen the mouse genome for repeats ≥11 U in the Xs^J region. Primer3 (http://frodo.wi.mit.edu/) was used to design primers. PCR amplicons were separated on a DNA analyzer (ABI 3100; Applied Biosystems). PCR reactions included the following: 1.5 μl of 10× PCR buffer (Applied Biosystems), 1.5 μl of 2.5 mM dNTPs, 1.5 μl of 25mM MgCl₂, 1 μl of 10 mM each forward and reverse primer, 1.2 μl of DNA (50 ng/μl), 0.12 μl AmpliTaq (Applied Biosystems), and 8.12 μl of dH₂O. Cycling conditions were as follows: 95°C for 12 min; 10 cycles of 94°C for 15 s, 55°C for 15 s, 72°C for 30 s; 20 cycles of 89°C for 15 s, 55°C for 15 s, 72°C for 30 s; and 72°C for 10 min. Amplicon (2 μl) was added to 9 μl of Hi-Di formamide and 0.5 μl of ROX-labeled 400 HD size standard (Applied Biosystems) and denatured at 95°C for 2 min. Sizing of alleles was done using GeneMapper software (Applied Biosystems).

Mutation screening by genomic DNA sequencing

Primers were designed using the ExonPrimer (http://ihg.gsf.de/ihg/ExonPrimer.html) to amplify exons; primers were designed in intronic sequence ≥50 nt beyond intron boundaries. A total of 280 primers were designed (Supplemental Table S2). PCR reactions included the following: 2.5 μl of 10× PCR buffer (Applied Biosystems), 2.0 μl of 2.5 mM dNTPs, 2.0 μl of 25 mM MgCl₂, 1 μl of 10 mM each forward and reverse primer, 5.0 μl of DNA (10 ng/μl), 0.25 μl AmpliTaq (Applied Biosystems), and 11.25 μl of dH₂O. Cycling conditions were as follows: 95°C for 8 min; 35 cycles of 95°C for 30 s, 56°C for 30 s, 72°C for 1 min; and 72°C for 10 min. PCR products were purified using alkaline lysis (QIAQuick; Qiagen, Valencia, CA, USA). Sequencing reactions were primed with PCR primers. Bidirentional sequencing was performed using v3.1 BigDye terminator cycle sequencing kit (Applied Biosystems) as per the manufacturer's protocol. DNA fragments were separated on an ABI 3100 Genetic Analyzer (Applied Biosystems) and analyzed with Sequencher v4.7 (GeneCodes, Ann Arbor, MI, USA).

Sequencing cDNA

RT-PCR was done using the Qiagen OneStep RT-PCR kit according to the manufacturer's protocol (Qiagen). RNA (50 ng) and Q-solution were used in all reactions. The primers used for the RT-PCR were as follows: forward, TAAAAACAACGCCAAGGCTC; and reverse, CAGCATGAGTGATCTCAGTTCC. Cycling conditions were as follows: 50°C for 30 min; 95°C for 15 min; 35 cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 2 min; and 72°C for 10 min. PCR products were gel purified using QIAQuick PCR purification kit (Qiagen) according to the manufacturer's protocol. Sequencing reactions were primed by the same primers used for RT-PCR. Forward and reverse sequences were obtained using the sequencing protocol used for mutation screening.

Quantitative RT-PCR

RNA was obtained from E11.5 Eif3c^Xs-J/+ embryos using RNeasy Mini kit (Qiagen) according to manufacturer's protocol. Embryos were disrupted in buffer RLT (Qiagen) using a hand-held motor pestle and repeated pipetting. Tissue was homogenized on a homogenizer column (QIAshredder; Qiagen). Total RNA from 3 Eif3c^Xs-J/+ E11.5 embryos and 3 wild-type E11.5 embryos were purified. Reverse transcription was performed using SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol, using 250 ng of random primers, 100 ng of total RNA, and 4 μl of dNTP mix (2.5 mM each dNTP). qPCR was performed using TaqMan probes (Applied Biosystems) for eukaryotic translation initiation factor 3 subunit C (Eif3c) and Gapdh as an endogenous control according to the manufacturer's protocol, utilizing 10 ng of cDNA. Three technical replicates for each cDNA sample were used. PCR and fluorescence detection was done on an ABI 7500 real-time PCR unit. Cycling conditions were 2 min at 50°C, 10 min at 95°C, and then 40 cycles of 95°C for 15 s and 60°C for 1 min. Relative expression of Eif3c^Xs-J/+ and wild-type samples was done using the ddCT method (11).

Whole-mount in situ hybridization

Digoxigenin-labeled antisense RNA probes for Shh, Ptch1, Gli1, Fgf8, Hoxd13, Gli3, and Eif3c were used for in situ analysis. Probes were made by reverse transcription of linearized plasmids in a reaction containing digoxigenin-labeled dUTP (reagents from Roche Diagnostics, Indianapolis, IN, USA). Whole-mount in situ hybridization was performed as described previously (12), with the following modifications: embryos for Fgf8 expression were treated with proteinase K for 2 min. For all other probes, embryos were treated for the following durations: E10.5 for 11 min, E11.5 for 13 min, E12.5 for 17 min, and E13.5 for 20 min.

Western blotting for GLI3 activator and repressor

GLI3 analysis by Western blot was performed as described previously (13). Individual band intensities on Western blots were measured using the extended measurement and histogram analysis tools in Photoshop CS5 (Adobe Systems, San Jose, CA, USA). GLI3 83-kDa [GLI3 repressor (GLI3R)] and 190-kDa [GLI3 activator (GLI3A)] bands were normalized to the relative vinculin band intensities on the same blot, and GLI3R/A ratios were calculated from the normalized values. GLI3R/A ratios were determined for 3 independent Xs-mutant E11.5 anterior and posterior forelimb bud lysates compared with wild-type siblings.

RESULTS

Manifestations associated with the Xs phenotype

Phenotyping was performed with Xs^J on 2 backgrounds: C57BL/6J × C3H/HeJ (B6C3/J) and B6C3/J × CAST/EiJ. Digit malformations and hemimelia were observed in Xs^J/+ mice (Fig. 1). Digit malformations ranged from an enlarged digit 1 to anterior polydactyly with up to 6 digits and were nonrandomly distributed among the 4 limbs (Table 1; χ² P=3.5×10⁻¹⁸). Most of this difference was attributable to a difference between forelimbs and hindlimbs, as the anomalies were 5 times more frequent in the former (χ² P=4.1×10⁻¹¹). Overall, left digit malformations were more common than right (χ² P=0.0027). However, this effect was reversed in the hindlimbs.

Figure 1.

Xs phenotype. A) Wild-type mouse (left) and Eif3c^Xs-J/+ mouse (right). Note the ventral hypopigmentation of the Eif3c^Xs-J/+ mouse. B) Left forelimb in an Eif3c^Xs-J/+ mouse. Arrow indicates an enlarged, triphalangeal pollex. C) Normal left forelimb of a wild-type mouse with a small, biphalangeal pollex.

Table 1.

Frequencies of the two major Xs phenotypic manifestations on two genetic backgrounds

Variable	B6C3/J		B6C3/J × CAST/EiJ
	Wild-type	Eif3cXs-J/+	Wild-type	Eif3cXs-J/+
Mice (n)	62	47	217	199
Males	55		229
Females	54		187
Total	109		416
Major phenotype
Hypopigmentation
Ventral trunk	0	31	3	65
Dorsal cranium	0	2	0	6
Digit malformation
Forelimb right	0	13	0	9
Forelimb left	0	37	0	24
Hindlimb right	0	12	0	1
Hindlimb left	0	4	0	0

Data represent mice generated from the breeding of Eif3c^Xs-J/+ on two different mouse genetic backgrounds: B6C3Fe/J and B6C3Fe/J × CAST/EiJ. Numbers of occurrences of hypopigmentation and digit malformation are shown. Distribution of digit malformation among the 4 limbs was nonrandom, with forelimbs more commonly affected than hindlimbs and left limbs more commonly affected than right (χ² P=3.5×10⁻¹⁸).

Forelimb hemimelia was observed in Xs^J/+ animals on the B6C3/J background (Fig. 2). To adjust for the overall size of animals, we normalized forelimb length to hindlimb length. From these normalized data, there was a significant difference in the forelimb lengths between Xs^J/+ and wild-type mice (P<0.0001, Mann-Whitney 2-tailed). We conclude that hemimelia of the forelimbs is a manifestation of Xs^J/+.

Figure 2.

Hemimelia in Eif3c^Xs-J/+ mice. A) X-ray of Eif3c^Xs-J/+ and wild-type mice. Arrows indicate short forelimbs. Also note the triphalangeal, enlarged pollex on the left and right forelimbs of the Eif3c^Xs-J/+ mouse. B) Limb lengths for Eif3c^Xs-J/+ and wild-type mice. Measurements were performed on 60 wild-type and 47 Eif3c^Xs-J/+ mice. Each data point represents a limb measurement. Bars at right of data points designate means ± sd for each category. Note that the difference in forelimb lengths is modest, although the difference in sd in Eif3c^Xs-J/+ vs. wild-type animals is substantial. This is due to a subset of animals with markedly shortened forelimbs. C) These data were further analyzed by normalizing forelimb length to hindlimb length. See main text for details.

Coat hypopigmentation occurred either on the ventral midline of the trunk or the dorsal cranium (Fig. 1). Hypopigmentation occurred ∼17 times more frequently on the ventral trunk surface compared with the dorsal cranium (Table 1). Ventral hypopigmentation was observed in 96 Xs^J/+ and 3 wild-type mice.

Mode of inheritance of Xs is semidominant with incomplete penetrance

Of the 525 genotyped mice, 279 were wild-type and 246 were Xs^J/+, not different from the expected 1:1 ratio (χ² P=0.149). Of the 47 Xs^J/+ mice on a B6C3/J background, 42 had digit malformations or coat hypopigmentation (our definition of affected for penetrance calculations), giving a penetrance of 89% (42/47). Of the 199 Xs^J/+ mice on a B6C3/J × CAST/EiJ background, penetrance was 39% (77/199). We conclude that there was no evidence for segregation distortion or embryonic lethality for Xs^J/+ animals. These data also suggest that CAST/EiJ harbors modifier alleles for the Xs phenotype similar to that observed for crosses of Xs^J/+ mice to the MOLF/Ei or FVB/N backgrounds (2).

Previous work (2) showed that Xs^J/Xs^J manifests 100% lethality at or before midgestation. To confirm these results and to refine the timing of the lethality, E3.5 blastocysts were collected from Xs^J/+ × Xs^J/+ timed matings. Eighteen blastocysts were recovered: 3 wild type; 15 Xs^J/+; 0 Xs^J/Xs^J. The probability of obtaining no Xs^J/Xs^J genotypes by chance among 18 blastocysts is 0.00564. We conclude that Xs^J/Xs^J is lethal before E3.5. Considering that Xs^J/Xs^J is embryonic lethal and Xs/+ manifests digit malformation, coat hypopigmentation, and hemimelia, we confirmed that the mode of inheritance of the Xs^J is semidominant.

Mutations in Eif3c cause the Xs phenotype

The Xs locus was mapped to a 3.5-cM region of mouse chromosome 7 defined by D7Mit255 and D7Mit103 (2). This interval contained 111 RefSeq genes [University of California–Santa Cruz (UCSC) database, July 2007; http://genome.ucsc.edu/]. Meiotic mapping was performed to further narrow this interval. Although the B6C3/J background is optimal for expressing an external phenotype, C57BL/6 and C3H/HeJ share extended haplotypes in this region. To increase heterozygosity, Xs^J/+ animals were generated by a B6C3/J-Xs^J/+ × CAST/EiJ cross. With these G1 offspring, (B6C3/J-Xs^J/+×CAST/EiJ) × B6C3F1/J matings were established, and G2 mice were typed with microsatellite markers (Supplemental Fig. 1B). Hypopigmentation was not used to determine affection status in this cross, and we used an affected-only approach for the polydactyly. Eight MIT markers were used, with an additional 16 markers identified in this study. Of the 416 typed G2 mice, 34 recombinations occurred between D7Mit330 and D7Mit109, which would be ∼8.2 cM. The genetic distance between D7Mit330 and D7Mit109 is 8.5 cM [Mouse Genome Database (MGD); http://www.informatics.jax.org/], suggesting that gross chromosomal rearrangements were unlikely. Xs mapped to an interval defined by D7Nih19 and D7Nih22, which was a physical distance of 322 kb. Fifteen known RefSeq genes were in this interval (Fig. 3).

Figure 3.

Map of the informative markers used in this study. The initial Xs region was flanked by D7Mit255 and D7Mit103. The Xs region was narrowed and defined by markers D7Nih19 and D7Nih22. According to Build 37 of the mouse genome, the physical distance between D7Nih19 and D7Nih22 is 322 kb, which includes 15 RefSeq genes that were sequenced during mutation screening.

Sequencing of the exons and exon/intron boundaries was performed on Xs^J/+ and C3H/HeJ wild-type DNA to screen for mutations in genes in the refined Xs interval. Six heterozygous sequence alterations were detected in Xs^J/+ animals that were absent in wild-type animals. Four alterations were identified in Rabep2, three of which occurred in introns (c.393+15A>C, c.504+18_504+19insGAGG, and c.1200+50T>C; GenBank NM_030566.2) and 1 occurred in the 3′ UTR (c.*357G>A). One alteration was identified in Coro1a, which occurred in intronic sequence (c.-1–9A>G; GenBank NM_009898.2). One alteration (c.907C>T; GenBank NM_146200.1) was identified in exon 9 of the Eif3c, a gene with a total of 21 exons (Fig. 4). This alternation creates a premature stop codon in a nonterminal exon, predicting p.Arg303X. Eif3c was also sequenced in Xsl, a putative allele of Xs^J (14). A heterozygous c.1702_1758del variant in exon 15 of Eif3c was found in Xsl/+. This variant predicts the 19-aa deletion p.Leu568_Leu586del from EIF3C, with the full-length protein having 911 aa (GenBank NP_666312.1). Given that 2 independent mutants harbor mutations in the same gene, we concluded that mutations in Eif3c cause Xs, and we suggest that these alleles should be Eif3c^Xs−J and Eif3c^Xsl. We use these designations hereafter.

Figure 4.

Electropherograms of Eif3c sequencing in Eif3c^Xs-J/+ (A) and Eif3c^Xsl/+ (B) mice. A) Forward sequence from Eif3c^Xsl/+ DNA. Arrows indicate the c.907C>T mutation detected in Eif3c^Xs-J/+ mice, which predicts p.Arg303X. B) Forward sequence from C3H/HeJ wild-type DNA. A c.1702_1758del was detected in Eif3c^Xsl/+ mice, which predicts p.Leu568_Leu586del. Bottom: forward sequence from the Eif3c^Xsl allele, which was generated from a gel-purified PCR product.

Reduced Eif3c mRNA levels in Eif3c^Xs-J/+ mice

To test whether the Eif3c^Xs-J c.907C>T mutation would be subjected to nonsense-mediated decay (NMD), cDNAs from 2 E11.5 Eif3c^Xs-J/+ embryos were sequenced, and the mutant allele was not detected (Fig. 5A). Quantitative RT-PCR was used to measure Eif3c mRNA levels. mRNA levels in E11.5 Eif3c^Xs-J/+ embryos were reduced ∼50% relative to wild-type levels (Fig. 5B). Taken together, these data are consistent with NMD of the Eif3c^Xs-J allele and suggest that significant up-regulation of the wild-type allele does not occur, that there is haploinsufficiency of Eif3c in Eif3c^Xs-J/+ mice, and that Eif3c^Xs-J is a null allele.

Figure 5.

Expression of Eif3c in Eif3c^Xs-J/+ mice. A) Electropherograms of cDNA and genomic DNA of Eif3c^Xs-J/+ animals. Arrows indicate the location of the c.907C>T mutation that was identified in Eif3c^Xs-J/+ mice. No mutant transcript was detected when sequencing cDNA from Eif3c^Xs-J/+ mice. B) Relative expression of Eif3c in Eif3c^Xs-J/+ and wild-type (+/+) mice. RNA was isolated from 3 wild-type E11.5 and 3 Eif3c^Xs-J/+ E11.5 embryos. For each embryo, 3 qPCR reactions were performed to obtain gene expression measurements for Eif3c, using Gapdh as an endogenous control. Error bars represent the sd of the 3 technical replicates. Arrow indicates sample against which relative expression was calculated. Relative expression of the Eif3c^Xs-J/+ samples is approximately half that of the wild-type samples. Expression levels were calculated using the ddCT method (11).

Ectopic expression of SHH pathway genes in Eif3c^Xs-J/+ embryos

Due to the phenotypic overlap of Xs and Gli3^Xt and the role of GLI3 in SHH signaling, we hypothesized that SHH/GLI3 signaling is disrupted in Eif3c^Xs-J/+ mice. In situ hybridization was used to analyze SHH/GLI3 pathway genes in Eif3c^Xs-J/+ limb buds (Fig. 6). Only normal, posterior expression of Shh (15, 16) was observed in Eif3c^Xs-J/+ limbs at E10.5 (n=6) and E11.5 (n=5). However, at E12.5, ectopic anterior expression of Shh was seen (2 of 4 embryos). This occurred when normal posterior Shh expression ceased. The expression pattern of Gli3 mRNA was also screened by in situ hybridization. No alterations in the expression pattern of Gli3 were observed in Eif3c^Xs-J/+ limbs at E10.5 (n=5), E11.5 (n=5), and E12.5 (n=3).

Figure 6.

In situ hybridization for mRNA expression of genes in the SHH/GLI3 pathway. Arrows indicate ectopic expression. In +/+ limb buds, no ectopic expression was seen when probing for Shh (n=15), Ptch1 (n=10), Gli3 (n=12), Gli1 (n=11), Hoxd13 (n=13), and Fgf8 (n=17). For Fgf8, the horizontal dashed line is drawn to parallel the proximal/distal axis of the limb. Then, a perpendicular to that line is drawn at the point where the perpendicular would contact the posterior Fgf8 expression boundary. The expression is assessed as anteriorly extended if the anterior expression signal extends proximal to the perpendicular. ΨE11.5, matched developmentally to stage E11.5. This embryo was dissected at E12.5; however, it exhibits developmental delay and is developmentally staged as an E11.5 embryo. Limb orientation in photographs: anterior = up; posterior = down; distal = left; proximal = right.

The expression of Ptch1 (17–21), Gli1 (8, 21, 22), Fgf8 (23, 24), and Hoxd13 (25, 26) has defined expression patterns in the developing limb and is regulated by SHH/GLI3 signaling. Aberrant expression of these genes would indicate that downstream processes regulated by SHH/GLI3 pathway are disrupted. In situ hybridization analysis of Ptch1, which is a transcriptional target of hedgehog signaling and encodes a receptor of hedgehog, showed ectopic anterior expression in Eif3c^Xs-J/+ limbs at E10.5 (1 of 2 embryos), E11.5 (2 of 3 embryos), and E12.5 (3 of 4 embryos), along with normal posterior expression. Ectopic anterior Gli1 expression was also seen in Eif3c^Xs-J/+ limb buds at E11.5 (4 of 4 embryos), but expression was normal at E10.5 (n=4) and E12.5 (n=3). Fgf8 is expressed in the AER, and its expression is maintained by SHH (15, 27, 28), and aberrant Fgf8 expression could indicate perturbed SHH/GLI3 signaling. At E10.5 (2 of 3 embryos) and E11.5 (3 of 4 embryos), Fgf8 expression was extended anteriorly in Eif3c^Xs-J/+ limbs. At E12.5, Fgf8 expression was not detected. SHH/GLI3 signaling also regulates Hoxd13 expression in the developing limb (16, 29). Alteration of Hoxd13 expression was observed in 1 Eif3c^Xs-J/+ limb (n=2), where expression was extended anteriorly. At E13.5, prolonged Hoxd13 expression was seen in 1 Eif3c^Xs-J/+ limb (n=1) compared with the corresponding wild-type control. Ectopic expression of Shh, Ptch1, Gli1, Fgf8, and Hoxd13 was observed in Eif3c^Xs-J/+ limbs, indicating that SHH/GLI3 signaling is disrupted in Eif3c^Xs-J mice.

Assessment of GLI3 processing by Western blot analysis

Limb buds from Eif3c^Xs-J/+ heterozygous and sibling wild-type embryos were divided into anterior and posterior halves, and the relative amounts of full-length GLI3A compared with processed GLI3R were evaluated by Western blot analysis using a GLI3 antibody that recognizes both forms. Compared with wild type, the Eif3c^Xs-J/+ heterozygotes showed reduced levels of GLI3R in anterior limb buds. Furthermore, the GLI3A/GLI3R ratio in the mutants were comparable between the anterior and posterior halves, whereas in the wild type, GLI3R levels are significantly higher in the anterior than the posterior limb bud (representative example shown in Fig. 7; n=4). These results indicate reduced processing of GLI3 in the anterior limb bud, which would lead to ectopic pathway activation, consistent with the changes in SHH target gene expression and skeletal phenotypes observed in Eif3c^Xs-J/+ embryos.

Figure 7.

Western blot analysis of GLI3 processing. Individual E11.5 limb buds from Eif3c^Xs-J/+ heterozygous and sibling wild-type embryos were divided into anterior (A) and posterior (P) halves, solubilized in RIPA buffer, and probed with anti-GLI3 antibody on Western blots as described previously (13). Blots were reprobed with an anti-vinculin antibody as a loading control. Note that the ratio of full-length GLI3A to processed GLI3R is dramatically increased in the Eif3c^Xs-J/+ mutant compared with wild-type anterior limb bud, whereas the ratio in posterior limb buds is similar between the mutant and wild-type samples. Example shown is right hindlimb.

DISCUSSION

Using a strategy of outbreeding Xs^J heterozygotes to castaneus with test crossing of recombinants to B6C3F1/J animals, we narrowed the candidate region for this locus and sequenced the candidate genes. The only potentially pathological variant in Xs^J was a c.907C>T change in Eif3c, which predicts p.Arg303X. We sequenced Eif3c in Xsl and found that this allele contained the c.1702_1758del sequence variant, which predicts p.Leu568_Leu586del. Neither of these variants was present in wild-type C3H/HeJ DNA or in C57BL/6. We also showed a 50% reduction in total Eif3c mRNA levels in Eif3c^Xs-J/+ animals. The evidence supporting causation of Eif3c mutations for the Xs phenotype includes the following: the presence of two novel sequence variants in mutant animals not found in the background strain DNA, one variant that is a nonsense mutation, the absence of other apparently pathological variants in the 15 genes in the linkage region, and the reduction of Eif3c mutant mRNA in affected animals. We conclude that these two mutations in Eif3c (now designated as Eif3c^Xs-J and Eif3c^Xsl) cause the Xs phenotype and that the genetic mechanism, at least for Eif3c^Xs-J, is haploinsufficiency, most likely as a result of nonsense mediated decay of the Eif3c^Xs-J transcript.

Given that the Eif3c^Xs-J phenotype is similar to Gli3^Xt-J, we hypothesized that polydactyly in Eif3c^Xs-J may be mediated by or associated with a perturbation of the SHH/GLI3 pathway. We found ectopic anterior expression of Shh in the limb buds of Eif3c^Xs-J/+ E12.5 embryos and Ptch1 in E10.5, E11.5, and E12.5 embryos and anterior extension of Hoxd13 and Fgf8 in the apical ectodermal ridge. Taken together, these results show that SHH/GLI3 signaling is perturbed in Eif3c^Xs-J/+ animals. To further explore these data, we went on to test whether GLI3 processing was perturbed in Eif3c^Xs-J/+ embryos. Indeed, we found an increased proportion of GLI3A relative to GLI3R in anterior limb buds, which is again consistent with up-regulation of SHH, as SHH inhibits the processing of GLI3 into the repressor form. The presence of ectopic Ptch1 expression and abnormal GLI3 processing in Eif3c^Xs-J/+ embryos before ectopic Shh expression suggests that the activation of this pathway occurs downstream of ligand binding (i.e., is ligand independent) and that the later anterior ectopic expression of Shh is an indication of SHH pathway activity.

The association of mutations in a translation initiation factor with perturbation of the SHH/GLI3 pathway is a surprising result, as there are no data to suggest a relationship of any Eif gene product to this pathway. The 12 known EIFs mediate translation by recruiting the machinery to initiate protein synthesis. EIF3 recruits the mRNA, Met-tRNA_i^Met, and the 40S ribosomal subunit, and it scans for the start codon. EIF3, EIF1A, and EIF5 facilitate the interaction of the ternary complex (EIF2, GTP, and Met-tRNA_i^Met) with the 40S ribosome, thus forming the 43S preinitiation complex. mRNAs are recruited to the 43S preinitiation complex, which allows for scanning and recognition of the start codon. On AUG recognition and pairing of Met-tRNA_i^Met, EIFs dissociate and the 60S ribosome is recruited to the 40S ribosome, Met-tRNA_i^Met, and mRNA, allowing for translation elongation (30). EIF3 has 13 known subunits, and 6 of these constitute a functional core (31). EIF3C is one of these core subunits. Interactions with other complexes in the translational machinery of Nip1, the Saccharomyces cerevisiae orthologue of EIF3C, have been investigated by expressing truncated forms that lack specific protein-binding domains. Deletion mutants of Nip1 showed that it is essential for stabilizing the interaction of the ternary complex (EIF2, GTP, and Met-tRNA_i^Met) to the 40S ribosome (32, 33). We speculate that a reduction of Nip1/EIF3C limits recruitment of the ternary complex to the 40S ribosome and that translational efficiency or even a complete lack of translation initiation results. We have shown that Eif3c^Xs-J/Xs-J animals manifest lethality before E3.5. In light of the function of the S. cerevisiae Nip1/EIF3C, we hypothesize that Eif3c^Xs-J/Xs-J homozygous cells lacking EIF3C are incapable of translation. Thus, early embryonic lethality of Eif3c^Xs-J/Xs-J is not surprising, considering that total loss of Eif3c function would preclude or severely inhibit protein translation, and it is known that transition to embryonic transcription and translation is essential for survival past the 2 cell stage (34).

In contrast to the general, lethal effect of homozygosity for the Eif3c^Xs-J mutation, Eif3c^Xs-J/+ heterozygotes have limited manifestations of digit and limb malformations and coat hypopigmentation. One explanation for this could be that proper limb patterning and melanocyte development require high levels of protein synthesis and that decreased protein translation has a greater effect on these developmental processes. Belly spot and tail (Bst) is a mouse phenotype that includes preaxial polydactyly and ventral midline hypopigmentation in heterozygotes and embryonic lethality in homozygotes. Bst results from a 4-nucleotide deletion in ribosomal-like protein L24 (Rpl24; refs. 35, 36). Rpl24^Bst/+ mouse embryonic fibroblasts (MEFs) have reduced protein synthesis compared with wild-type MEFs (36). Like Rpl24^Bst/+ animals, impairment of protein synthesis could occur in Eif3c^Xs-J/+ animals. However, it has recently been shown that up-regulation of p53 is a consequence of loss of function of Rpl24 and that the up-regulated p53 is the effector of the anomalies in Rpl24^Bst/+ mice (37). It is not known how p53 is upregulated by Rpl24 haploinsufficiency nor is it known how p53 mediates the anomalies in those mice. Further studies, beyond the scope of the present experiments, will be necessary to determine whether EIF3C mediates skeletal anomalies via reduced translation or other indirect effects, such as that observed in Rpl24^Bst/+ mice, or whether EIF3C interacts in as yet unappreciated ways with the SHH/GLI3 pathway.