Overgrowth of a Mouse Model of the Simpson–Golabi–Behmel Syndrome Is Independent of IGF Signaling

Developmental Biology 243, 185–206 (2002) doi:10.1006/dbio.2001.0554, available online at http://www.idealibrary.com on

Overgrowth of a Mouse Model of the Simpson– Golabi–Behmel Syndrome Is Independent of IGF Signaling Eric Chiao,* Peter Fisher,† Laura Crisponi,‡ Manila Deiana,‡ Ioannis Dragatsis,* David Schlessinger,§ Giuseppe Pilia,‡ and Argiris Efstratiadis* ,1 *Department of Genetics and Development, and †Department of Pathology, Columbia University, New York, New York 10032; ‡Istituto di Ricerca sulle Talassemie ed Anemie Mediterranee CNR, Cagliari, Italy; and §Laboratory of Genetics, National Institute on Aging, National Institutes of Health, Baltimore, Maryland 21224

The type 1 Simpson–Golabi–Behmel overgrowth syndrome (SGBS1) is caused by loss-of-function mutations of the X-linked GPC3 gene encoding glypican-3, a cell-surface heparan sulfate proteoglycan that apparently plays a negative role in growth control by an unknown mechanism. Mice carrying a Gpc3 gene knockout exhibited several phenotypic features that resemble clinical hallmarks of SGBS1, including somatic overgrowth, renal dysplasia, accessory spleens, polydactyly, and placentomegaly. In Gpc3/⌬H19 double mutants (lacking GPC3 and also carrying a deletion around the H19 gene region that causes bialellic expression of the closely linked Igf2 gene by imprint relaxation), the Gpc3-null phenotype was exacerbated, while additional SGBS1 features (omphalocele and skeletal defects) were manifested. However, results from a detailed comparative analysis of growth patterns in double mutants lacking GPC3 and also IGF2, IGF1, or the type 1 IGF receptor (IGF1R) provided conclusive genetic evidence inconsistent with the hypothesis that GPC3 acts as a growth suppressor by sequestering or downregulating an IGF ligand. Nevertheless, our data are compatible with a model positing that there is downstream convergence of the independent signaling pathways in which either IGFs or (indirectly) GPC3 participate. © 2002 Elsevier Science (USA) Key Words: growth; proteoglycans; glypican; insulin-like growth factor; Simpson–Golabi–Behmel syndrome.

INTRODUCTION Mammalian growth is regulated by the actions of gene products participating in signaling pathways that predominantly promote or inhibit cell proliferation through systemic or local effects (reviewed by Efstratiadis, 1998). Derangement of such genetic controls can lead to conditions of growth retardation or overgrowth. Human overgrowth syndromes comprise an ill-defined group of disorders characterized by prenatal and postnatal higher-than-normal weight and length usually associated with various malformations (reviewed by Cohen, 1999). In a 1

To whom correspondence should be addressed at Russ Berrie Medical Science Pavilion, Columbia University, 1150 St. Nicholas Avenue, New York, NY 10032. Fax: (212) 851-5236. E-mail: [email protected]. 0012-1606/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.

minority of these conditions, however, a genetic cause has been determined. This is the case for the Simpson–Golabi– Behmel syndrome type 1 (SGBS1; MIM 312870; reviewed by Neri et al., 1998; Lin et al., 1999; Li et al., 2001) caused by mutations of the X-linked gene GPC3 (Pilia et al., 1996), which resides on Xq26 and encodes glypican-3, one of six known mammalian cell-surface heparan sulfate proteoglycans attached to the plasma membrane lipid via a glycosylphosphatidylinositol (GPI) anchor (see Selleck, 1999, 2000, 2001; Bernfield et al., 1999; Perrimon and Bernfield, 2000, 2001; De Cat and David, 2001). Today, a total of ⬃120 SGBS1 patients have been described, and GPC3 deletions/ mutations have been identified in 40% (26/65) of the examined cases (see Lin et al., 1999; Li et al., 2001). In addition, a second SGBS locus associated with a more severe form of the disorder (SGBS2; MIM 300209; see Terespolsky et al., 1995) has been mapped in a single family

185

186

Chiao et al.

(four maternally related male cousins) within an ⬃6-Mb region of Xp22 (Brzustowicz et al., 1999). In male patients, in addition to excessive growth, the rather broad spectrum of SGBS1 manifestations includes coarse facies, macroglossia, organomegaly, renal dysplasia, cardiac defects, and skeletal malformations. The mechanistic relationship between the pathogenesis of developmental abnormalities in SGBS1 and various GPC3 loss-of-function mutations (large and small deletions and also nonsense, frameshift, missense, and splice-site mutations; Pilia et al., 1996; Hughes-Benzie et al., 1996; Lindsay et al., 1997; Xuan et al., 1999; Veugelers et al., 2000) continues to remain unknown. It is thought, however, that glypicans can function as coreceptors affecting particular signaling pathways involved in proliferative and morphogenetic processes by binding and modulating the actions of growth factors and morphogens (Perrimon and Bernfield, 2000). In this context, it can be surmised that the manifestation of overgrowth in SGBS1 would result from elimination of the normal GPC3 function that inhibits growth and/or promotes apoptosis. SGBS exhibits overlapping phenotypic features with the Beckwith–Wiedemann syndrome (BWS; MIM 130650), a genetically heterogeneous and complex overgrowth disorder characterized by macrosomia, macroglossia, abdominal wall defects, and other manifestations (reviewed by Elliot and Maher, 1994; Weng et al., 1995; Weksberg and Squire, 1996; Li et al., 1998; Cohen, 1999; Maher and Reik, 2000). In a subset of BWS cases, excess of growth-promoting IGF2 during embryonic development appears to be causally involved in pathogenesis (see Morison and Reeve, 1998; Li et al., 1998; Maher and Reik, 2000). Despite the fact that BWS and SGBS are distinct clinical entities of different etiology, an involvement of IGF signaling in SGBS was considered as possible because IGF2 overexpression in mouse mutants caused overgrowth and other aspects of the BWS phenotype (Eggenschwiler et al., 1997; Sun et al., 1997), but features resembling hallmarks of SGBS1 were also manifested (Eggenschwiler et al., 1997). An attractive model that was previously advanced posited that GPC3 potentially binds and somehow sequesters or downregulates IGF2 (Hughes-Benzie et al., 1996; Pilia et al., 1996). The validity of this hypothesis remained questionable, however, because preliminary biochemical data suggesting a human GPC3–IGF2 interaction (Pilia et al., 1996; Xu et al., 1998) were not reproduced with OCI-5 (the rat GPC3 homolog; Song et al., 1997). On the other hand, some GPC3–IGF2 relationship was implied by the observation that cell line-specific promotion of apoptosis by OCI-5 was rescued by adding IGF2 to the culture medium (Gonzalez et al., 1998). Still, the possibility remained that this evidence might have simply reflected ordinary antiapoptotic signaling through the type 1 IGF receptor (IGF1R; see Baserga et al., 1997). Genetic experiments with mouse mutants have established that, among the classic growth factors, the signaling system consisting of the IGFs (IGF1 and IGF2) and their

cognate receptor (IGF1R) is the major determinant of mammalian growth (see Efstratiadis, 1998; Lupu et al., 2001). In rodents, IGF2 is essential for normal embryonic growth (Igf2 expression ceases after weaning; DeChiara et al., 1990, 1991), whereas IGF1 has a continuous growth-promoting function throughout development (Liu et al., 1993; Baker et al., 1993). Because of reduced growth rate, Igf1 nullizygotes exhibit ⬃60% of normal weight at birth (⬃60%N) and, if they survive depending on genetic background, they become ⬃30%N in size at the age of 2 months. Because the Igf2 gene is subject to parental imprinting and only the paternal allele is expressed in most tissues (DeChiara et al., 1991), the heterozygotes with a paternally derived mutated allele (Igf2 ⫹/p⫺) and the Igf2 nullizygotes (Igf2 ⫺/⫺) are phenotypically indistinguishable; they are fully viable dwarfs with ⬃60%N birth weight (this relative size is maintained throughout life). Igf1r nullizygotes are severely growth retarded (⬃45%N birth weight) and die invariably immediately after birth. To investigate genetically whether there is a relationship between GPC3 and IGF2 in growth control, we sought to generate double-mutant mice carrying knockouts of both the Gpc3 and Igf2 genes and compare their phenotype with the phenotypes of the respective single mutants. For the purpose of these matings, we generated mutants lacking GPC3 and used available Igf2 mutant mice. While this work was in progress, targeted inactivations of Gpc3 were also described independently by two other groups (CanoGauci et al., 1999; Paine-Saunders et al., 2000), but genetic analyses in connection to the IGFs were not included in these reports. Moreover, despite manifestation of overgrowth in all three cases, some phenotypic differences were noted between our data and the published results (see Results and Discussion). In the genetic investigation of GPC3 function in relation to IGF signaling, our prediction was that, if overgrowth in the absence of GPC3 is due to excess of unleashed IGF ligand (which under normal conditions in wild-type mice is sequestered by GPC3), the phenotype of Gpc3/Igf2 double mutants would be identical to that of single mutants lacking only IGF2. This prediction was not fulfilled, as progeny with a body size intermediate between that of Igf2 mutants and wild-type mice were obtained. Similar observations were made with additional double mutants that we generated, lacking GPC3 and IGF1 or GPC3 and IGF1R, to ascertain whether any IGF signaling at all was related to GPC3 function. Thus, conclusive evidence was provided demonstrating that the GPC3 function is IGF-independent. This conclusion was further supported by using a reciprocal experimental design and generating double mutants lacking GPC3 and also overexpressing IGF2 because of a deletion around the H19 gene (⌬H19) that causes bialellic expression of the imprinted, closely linked Igf2 gene. Again, results inconsistent with a model positing a GPC3–IGF2 interaction were obtained, as the absence of GPC3 in Gpc3 ⫺//⌬H19 double mutants did not augment the overgrowth of single ⌬H19 mutants that results from excess of IGF2.

© 2002 Elsevier Science (USA). All rights reserved.

187

Genetic Analysis of Glypican-3 Null Mice

MATERIALS AND METHODS Gpc3 Gene Targeting and Breeding of Mutants A targeting vector (Fig. 1a) was constructed in several steps from subcloned fragments of a BAC clone carrying exon 1 and flanking sequences of the murine Gpc3 gene that was isolated from a 129/SvJ genomic library and characterized by restriction analysis and partial DNA sequencing. The final construct (cloned into pBluescript II SK⫹; Stratagene) consisted of a 2.25-kb HindIII–Bgl II 5⬘ genomic fragment, a neomycin resistance gene (neo) cassette under the control of the Pgk gene promoter (for positive selection with G418; 1.6 kb), a 3.9-kb EagI–HindIII 3⬘ genomic fragment, and a ␤-actin promoter-driven diphtheria toxin A gene cassette (for negative selection against random integration; see Yagi et al., 1990). Thus, a 1.0-kb BglII–EagI genomic fragment containing exon 1 flanked by 0.6 kb of 5⬘ and 0.1 kb of 3⬘ sequences was replaced with the neo cassette. Linearized targeting vector DNA was introduced by electroporation into ES cells [W9.5; 129/Sv (JR2448) genetic background; see Ge´ rard et al., 1999] and, after drug selection, DNA of resistant clones was analyzed by Southern blotting using a probe corresponding to genomic sequence downstream from the 3⬘ region of homology in the vector (0.7-kb HindIII fragment; Fig. 1a). Of 69 G418resistant clones, 12 were targeted correctly (single insertion) and 2 of them were injected into C57BL/6J host blastocysts, to generate chimeras transmitting the mutation (no phenotypic differences were observed between mice generated from the independent ES clones). Male chimeras were mated with wild-type C57BL/6J or 129/ Sv/Ev females, and F 1 heterozygous female progeny were crossed with males carrying Igf1, Igf2, or Igf1r targeted mutations (for details, see Results and Discussion). To initiate the generation of progeny carrying the Gpc3 mutation in combination with a deletion mutation of the H19 gene (⌬H19; see Results and Discussion), transmitting chimeras were first mated with ⌬H19 ⫺/⫺ females. The strains of mutant mice (other than those lacking GPC3) that were used in this study and procedures for their genotyping have been described previously (see DeChiara et al., 1990; Liu et al., 1993; Leighton et al., 1995). Histological and anatomical analyses of embryos and tissues, including skeletal preparations, were performed as described (Eggenschwiler et al., 1997).

DNA and RNA Analyses

performed as described (Eggenschwiler et al., 1997). The primary goat polyclonal antibody (Research Diagnostics) used at a concentration of 1 ␮g/ml was raised against amino acids 25– 44 of the human IGF2 precursor. The secondary antibody was biotinylated sheep anti-goat IgG (Boehringer) used at a 1:2500 dilution. The immunoreactive bands were visualized with ECL detection reagents (Amersham). Quantitation of luminescent signals was performed on a Kodak Digital Science 1D 440CF Imaging Station. To determine mutant/wild-type ratios, IGF2 arbitrary values were first normalized, using as internal standards either ␣-fetoprotein (serum) or HMG-I(Y) (embryonic extracts) that was detected with rabbit polyclonal antibodies (from ICN Biochemicals and D. Thanos, respectively).

Biochemical Analyses Blood glucose levels were determined by using a digital glucometer. Disaccharidase activities in small intestine homogenates were measured as described (Koldovsky et al., 1969), and specific activities were expressed in international units (␮moles substrate hydrolyzed per minute per g protein).

Growth Analysis Cumulative postnatal growth curves are best fitted with a logistic equation W ⫽ A/1 ⫹ exp[⫺b(t ⫺ c)], where W and A are the weight at time t and the asymptotic weight, respectively, while b and c are constants. Body weights were determined on fixed postnatal days for all mice and averages were calculated for each genotype. Considering that postnatal growth is triphasic, these average weight data were fitted simultaneously using a computer program (SigmaPlot) by summation of three logistic functions with different sets of A, b, and c values (see Koops, 1986; Koops et al., 1987). From each fit, curves of predicted absolute growth rates (dW/dt) were derived by subtracting the value of cumulative weight at each day (W i) from that estimated for the previous day (W i⫺1). Absolute growth rates were then transformed to specific growth rates [i.e., growth rates per unit size; (dW/dt) ⫻ (1/W)] using weight values for normalization: (W i ⫺ W i⫺1)/[1/2⫻(W i ⫹ W i⫺1)]. Organ weights were determined as described (Eggenschwiler et al., 1997).

RESULTS AND DISCUSSION

For genotyping by Southern analysis, DNA was prepared from yolk sacs of embryos or the tail tip of 10-day-old mice. Mice were genotyped for the Gpc3 mutation by digesting the DNA with EcoRI and hybridizing the blots with the 0.7-kb HndIII probe (Fig. 1a), as described above for targeted ES cells. For Northern analysis of total embryonic RNA, blots were hybridized with a 763-bp Gpc3 cDNA probe (Pellegrini et al., 1998) representing most of the sequence of exons 3–5 (positions 6281390; GenBank Accession No. AF185614). Nonisotopic in situ hybridization analysis on frozen sections was performed as described (Wilkinson, 1992), using digoxigenin-labeled sense and antisense RNA probes corresponding to a region of Gpc3 cDNA (positions 1020 –1918).

Protein Analysis Relative amounts of IGF2 in e18.5 embryonic extracts or serum and in e16.5 amniotic fluid were determined by Western blotting

Targeted Disruption of the Mouse Gpc3 Gene To generate mutants lacking GPC3 functions, we replaced, by targeting, the Gpc3 exon1 (carrying the ATG initiation codon) and flanking sequences with a neo gene cassette in ES cells (see Fig. 1a, and Materials and Methods) and produced chimeras transmitting the mutation to their progeny (Fig. 1b). The targeted allele was designated Gpc3 ⌬ex1, but it will be referred to here simply as Gpc3 ⫺. Northern analysis of total RNA isolated from genotyped e12.5 embryos using a Gpc3 cDNA probe showed that Gpc3 mRNA was undetectable in mutants lacking GPC3 (Fig. 1c). Thus, targeting of the Gpc3 locus resulted in generation of a null mutation, as expected from the deletion of at least some portion of the presumptive promoter region together with exon 1.

© 2002 Elsevier Science (USA). All rights reserved.

188

Chiao et al.

Genetic Crosses To test genetically whether there is a relationship between GPC3 function and IGF signaling, we performed three types of crosses and generated double mutants lacking GPC3 and IGF2 (“G/II cross”), GPC3 and IGF1 (“G/I cross”), or GPC3 and IGF1R (“G/R cross”). In a fourth cross, using ⌬H19 mutants (Leighton et al., 1995; Eggenschwiler et al., 1997), we generated double mutants lacking GPC3 but overexpressing IGF2 (“G/H cross”). In mice, the paternally expressed, imprinted Igf2 gene is closely linked to the reciprocally imprinted H19 locus encoding a nontranslatable RNA of unknown function. In progeny with a maternally inherited deletion mutation (⌬H19), which eliminates H19 and upstream sequence containing an insulator element (see Leighton et al., 1995; Wolffe, 2000), the normally silent maternal Igf2 allele present in cis becomes transcriptionally active. As a consequence, Igf2 is expressed biallelically and this results in a twofold increase of IGF2 tissue levels and overgrowth (birth weight ⬃130%N). To initiate all crosses, male chimeras transmitting the Gpc3 ⫺ mutation in 129/Sv genetic background were mated with wild-type C57BL/6J females, to obtain heterozygous Gpc3 ⫹/⫺ females (129/Sv ⫻ C57BL/6J hybrids; 129⫻B6). The latter were crossed: (1) with Igf2 ⫹/p⫺ males (G/II cross); (2) with Igf1 ⫹/⫺ males, to generate and then intercross (G/I cross) double heterozygous females (Gpc3 ⫹/⫺/Igf1 ⫹/⫺) and males hemizygous for the Gpc3 mutation and heterozygous for the Igf1 mutation (Gpc3 ⫺//Igf1 ⫹/⫺); and (3) with Igf1r ⫹/⫺ males, to generate and then intercross (G/R cross) Gpc3 ⫹/⫺/ Igf1r ⫹/⫺ female and Gpc3 ⫺//Igf1r ⫹/⫺ male progeny. To combine the Gpc3 ⫺ and ⌬H19 mutations, male transmitting chimeras were first mated with ⌬H19 ⫺/⫺ females, to generate Gpc3 ⫹/⫺/⌬H19 ⫹/⫺ females, which were then crossed with Gpc3 ⫺ males (G/H cross). The genotypes of progeny derived from the described breeding schemes are shown in Table 1. This Table also includes limited results of matings in pure 129 background (crosses of chimeras with 129 females). In the G/II cross, the129⫻B6 Gpc3 ⫹/⫺ females were mated with males carrying the Igf2 mutation in inbred 129 background. Thus, the progeny were enriched in 129 strain component (up to 3/4). On the other hand, the progeny of the G/R cross were derived from parents enriched in B6 component (up to 3/4), which were generated by matings of the Gpc3 ⫹/⫺ 129⫻B6 female hybrids with Igf1r ⫹/⫺ male partners from a purebred B6 colony. The background of the progeny derived from the G/H cross was also mixed 129⫻B6, but the relative contributions of the components are unknown (records of the breeding history of the ⌬H19 colony were not kept). Finally, the animals used for the G/I cross and their offspring carried components of four strain backgrounds (129, B6, MF1, and DBA) in variable mixtures. The particular background of Igf1 ⫹/⫺ heterozygotes (MF1/ DBA) was chosen empirically during previous experiments (see Lupu et al., 2001), to increase the frequency of surviving nullizygotes in case of intercrossing, as there is a

strain-dependent, high incidence of neonatal lethality among Igf1 ⫺/⫺ mice (Liu et al., 1993). Despite strain differences, pairwise comparisons of weight distributions between progeny of the same age and genotype from the four crosses did not reveal any statistically significant differences (Student’s t test; P ⬍ 0.05).

Phenotypic Features of Mutants In addition to overgrowth (discussed in detail in a separate section), which is readily evident in Gpc3 ⫺/ and Gpc3 ⫺/⫺ neonates compared with their normal siblings (Fig. 2a), the lack of GPC3 results in reduced survival and manifestation of other abnormalities described below. Viability. Three Gpc3 knockouts were generated independently by using ES cells from 129 substrains known to have differences in genetic backgrounds (see Simpson et al., 1997; Threadgill et al., 1997): 129/J (Cano-Gauci et al., 1999), 129/SvJ (Gpc3 tm108Sau allele in RW4 ES cells; PaineSaunders et al., 2000), and 129/Sv (JR2448; this work). An insertional mutation derived separately from a gene trap protocol was in 129/Ola background (Gpc3 TgN(␤geo)ex136Ska allele in CGR8 ES cells; Paine-Saunders et al., 2000). In all cases, transmitting chimeras were mated with B6 females (generation of F 1 progeny). Phenotypic differences between the t108 and Ex136 mutations were not observed and there was no incidence of embryonic lethality in any of the mutations. Postnatally, however, variable, strain-dependent lethality was observed in all cases. Table 1 shows the numbers of recovered stillborn progeny from the four crosses that we described above (G/II, G/I, G/R, and G/H) and also the numbers of additional mice that died before weaning (after weaning, some animals were sacrificed for analysis, but other deaths were not observed). These results were used to assess the viability of single mutants lacking GPC3 and double Gpc3 ⫺//⌬H19 m⫺/⫹ or Gpc3 ⫺/⫺/⌬H19 m⫺/⫹ mutants. Because mice lacking IGF1R die invariably immediately after birth, while neonates lacking IGF1 exhibit variable, strain-dependent lethality, the data concerning double mutants nullizygous for either one of these mutations and also lacking GPC3 were not taken into consideration. Maximum-likelihood statistical analysis of the data first indicated that there are no significant differences between the four sets of data in regard to the viability of Gpc3 ⫺/ or Gpc3 ⫺/⫺ mutants and wild-type controls. Thus, in evaluating death frequencies, all data were considered collectively. As Table 1 shows, random deaths of newborn progeny were recorded not only for male or female mutants lacking GPC3, but also for wild-type animals and Gpc3 ⫹/⫺ heterozygous females. Our analysis showed that the frequency of stillborn Gpc3 ⫺ and Gpc3 ⫺/⫺ mutants was not significantly different from that of other random deaths. In contrast, excluding stillborn animals, the frequency of mutants lacking GPC3 that died before weaning (8/76; 10.5%) was statistically significantly higher than that expected for random events (P ⬍ 0.025 or P ⬍ 0.034, depending on

© 2002 Elsevier Science (USA). All rights reserved.

189

Genetic Analysis of Glypican-3 Null Mice

TABLE 1 Breeding Data Cross

Sex

Genotype

No. of mice recovered

Observed frequency

Expected frequency

G/II

M or F M or F M M F F M M M M or F M or F M or F F F F M M M M or F M or F M or F F F F M M M or F M or F F F M or F M F

W II ⫹/p⫺ G ⫺/ ⫺/ ⫹/p⫺ G II G ⫹/⫺ ⫹/⫺ ⫹/p⫺ G II W I ⫺/⫺ I ⫹/⫺ ⫺/ G or G ⫺/⫺ G ⫺/I ⫺/⫺ or G ⫺/⫺I ⫺/⫺ G ⫺/I ⫹/⫺ or G ⫺/⫺I ⫹/⫺ G ⫹/⫺ G ⫹/⫺I ⫺/⫺ G ⫹/⫺I ⫹/⫺ W R ⫺/⫺ R ⫹/⫺ G ⫺/ or G ⫺/⫺ G ⫺/R ⫺/⫺ or G ⫺/⫺R ⫺/⫺ G ⫺/R ⫹/⫺ or G ⫺/⫺R ⫹/⫺ G ⫹/⫺ G ⫹/⫺R ⫺/⫺ G ⫹/⫺R ⫹/⫺ W H m⫺/⫹ G ⫺/ or G ⫺/⫺ ⫺/ m⫺/⫹ G H or G ⫺/⫺H m⫺/⫹ G ⫹/⫺ G ⫹/⫺H m⫺/⫹ W G ⫺/ G ⫹/⫺

22 27 15 14 17 11 12 9 30 20 8 33 14 3 27 7 2 16 14 7 24 11 6 17 23 23 35 27 20 25 41 3 17

0.208 0.255 0.141 0.132 0.160 0.104 0.0769 0.0577 0.192 0.128 0.0513 0.212 0.0897 0.0192 0.173 0.067 0.019 0.154 0.135 0.067 0.231 0.106 0.058 0.163 0.150 0.150 0.229 0.176 0.131 0.163 0.672 0.049 0.279

0.250 0.250 0.125 0.125 0.125 0.125 0.0625 0.0625 0.125 0.125 0.125 0.250 0.0625 0.0625 0.125 0.0625 0.0625 0.125 0.125 0.125 0.250 0.0625 0.0625 0.125 0.125 0.125 0.250 0.250 0.125 0.125 0.500 0.250 0.250

G/I

G/R

G/H

129

whether only wild-type or both wild-type and Gpc3 ⫹/⫺ heterozygous mice are taken into consideration). Nevertheless, the Gpc3 ⫺/ or Gpc3 ⫺/⫺ hybrids with variably segregating strain components were able to survive in their vast majority (⬃90%), and this gave us the opportunity to analyze postnatal growth in detail (see below), a study not performed previously with other Gpc3 mutants (CanoGauci et al., 1999; Paine-Saunders et al., 2000). In contrast to surviving hybrids, matings in pure 129 background, albeit limited, indicated that, in the absence of GPC3, modifiers in this genetic background affect significantly postnatal viability, since we were able to recover only three mutant mice (⬃20% of the expected number; Table 1). We presume that the deficit can be attributed to maternal cannibalism of dead mutant neonates, because we found that Gpc3 ⫺/ and Gpc3 ⫺/⫺ embryos were alive and present at the expected frequency at e18.5 (data not shown). We note that, even in pure 129 strain background, lethality

Stillborn

Addional deaths before weaning

2 1 3

1 1 6

1 4

3

2 4 4 3 1 1 1 2 1 7 2

4 1 1

2 6

6 2 1 4 19 1 3

5 5 1

1

in the absence of GPC3 is not 100%, as two of the three recovered Gpc ⫺/ neonates reached adulthood and appeared to be healthy. However, because of the viability problem, attempts to obtain double mutants lacking both GPC3 and IGF2 in a pure 129 strain background were unsuccessful. It is interesting that apparently different adverse modifiers affect the perinatal lethality associated with lack of GPC3, which was increased not only in a pure 129 background (our results), but also in a background significantly enriched in B6 component (Cano-Gauci et al., 1999), in contrast with the improved survival in hybrid states. However, some of the perinatal deaths previously reported (Cano-Gauci et al., 1999) were clearly due to a high incidence of respiratory infections attributed to presumptive abnormalities of the bronchi and the respiratory epithelium. We think that these deaths were unrelated to genetic causes, as we found no evidence of abnormal lung develop-

© 2002 Elsevier Science (USA). All rights reserved.

190

Chiao et al.

FIG. 1. Targeting of the mouse Gpc3 locus and molecular analyses of mutants. (a) The restriction map in the region of exon 1 (black rectangle) of the Gpc3 gene (top), the targeting construct (middle) used for gene disruption by homologous recombination (denoted by large X symbols), and the structure of the derived targeted allele (bottom) are shown. The transcriptional orientation of the neo cassette (replacing the deleted exon 1 and flanking sequences) is the same as that of the Gpc3 gene. An open rectangle and a wavy line represent a diphtheria toxin (DT) gene cassette and plasmid vector sequence, respectively. Restriction sites are: HindIII (H), EcoRI (R), BglII (B), EagI (E), and KpnI (K). The position of the probe (a 0.7-kb HindIII fragment located in the Gpc3 locus downstream from the 3⬘ region of homology present in the vector) that was used for Southern analyses of ES cell clones and mice is indicated, and the sizes (in kb) of the endogenous and targeted EcoRI genomic DNA fragments recognized by this probe are shown. (b) Example of Southern analysis of EcoRI-digested tail DNA from wild-type (lanes 1, 2, and 6) and Gpc3 ⫹/⫺ heterozygous females (lanes 3–5 and 7). The blot was hybridized with the probe shown in (a). (c) Northern analysis of total RNA (30 ␮g per lane) isolated from wild-type primary embryonic fibroblasts (lane 5; positive control) or e12.5 embryos of the following genotypes: wild-type (lane 1), double Gpc3 ⫺/ /Igf2 ⫹/p⫺ mutant (lane 2), single Igf2 ⫹/p⫺ mutant (lane 3), and single Gpc3 ⫺/ mutant (lane 4). The blot was hybridized with a Gpc3 cDNA probe (see Materials and Methods), which detected a single ⬃2.3-kb Gpc3 mRNA species in the samples from wild-type and Igf2 mutant embryos (lanes 1 and 3, respectively), whereas a hybridization signal was undetectable in the samples from embryos carrying the Gpc3 mutation (lanes 2 and 4), even after overexposure of the membrane (middle panel). A picture of the gel (bottom panel) displaying the 18S and 28S ribosomal RNA species stained with ethidium bromide indicates that approximately equal amounts of RNA were loaded per lane. In the top panel, the positions of these RNAs (size markers) are indicated with arrows. (d) Western analysis of serum from e18.5 embryos (5 ␮l per lane) used to assay relative amounts of IGF2 (bottom) using ␣-fetoprotein (AFP) as a standard (top). The specificity of the assay is indicated by the absence of immunoreactive IGF2 in sera from Igf2 null mutants (lanes 3 and 6), in contrast with wild-type (lane 4) and Gpc3 mutant (lanes 1, 2, and 5) embryos. Results are also presented for ⌬H19 and Gpc3/⌬H19 mutants (lanes 7 and 8, respectively). Differences in the amounts of immunoreactive IGF2 were not detected between the positive samples. In regard to the ⌬H19 specimen, the result is in agreement with previous data (the level of IGF2 is normal in the serum of ⌬H19 mutants, in contrast to a twofold increase detected in embryonic tissues; Eggenschwiler et al., 1997).

ment or respiratory infections in the analogous 129⫻B6 hybrids that we have studied. Whether maintenance of our mice in a pathogen-free barrier facility could explain this discrepancy remains unknown. It is notable that the incidence of perinatal death is quite high for SGBS patients (in a total of 92 informative cases, deaths of 17 neonates, including 1 stillborn, were described;

18.5%; see Lin et al., 1999). It is also notable that absence of GPC3 in combination with overexpression of IGF2 (Gpc3 ⫺/ or Gpc3 ⫺/⫺ mutants also carrying a maternally derived ⌬H19 mutation) results in a highly significant incidence of stillborn progeny (19/27; ⬃70%; P ⬍ 0.00025; Table 1), whereas neither one of the single mutations causes neonatal death at a frequency higher than random.

© 2002 Elsevier Science (USA). All rights reserved.

191

Genetic Analysis of Glypican-3 Null Mice

FIG. 2. Body size, pigmentation defects, and skeletal abnormalities of mutant mice. (a) Example of body-size phenotypes of wild-type (W) neonates and mutant littermates lacking GPC3 (G), IGF2 (II), or both (G/II). (b) p15 wild-type (W) and Gpc3 ⫺/ (G) siblings. The mutant, which is smaller than wild-type at this age (see text), lacks pigmentation in the feet and the distal third of the tail. (c) White belly spotting in a Gpc3 ⫺/ mutant. (d, e) Postaxial polydactyly (arrows) in e18.5 embryos affecting only a forelimb (d) or both a forelimb and a hindlimb (e). (f) Agnathia in a stillborn Gpc3 ⫺/ mutant [absence of the mandible and presence of only a small skin bud (arrow) in its place]. (g, h) Ribs and sternum of e18.5 wild-type (W) and Gpc3 ⫺/ /⌬H19 double mutant (G/H) embryos. The skeletons were stained with Alizarin red and Alcian blue. In the mutant, articulation between the sternum and the ventral tips of the costal cartilages has occurred asymmetrically and in a staggered fashion on opposite sides, resulting in ossification of sternal segments (h; arrow), in contrast to wild type (g; arrow). The sternum of the mutant is broader than normal and has a bifurcating xiphoid process (arrowhead). (i, j) The top two ribs of the double mutant shown in (h) are fused in their middle portions, but attached separately to the spine (i; arrow) and to the sternum (j; arrow). (k) The body of each lumbar vertebra in the skeletal preparation of a wild-type embryo has a single ossification center (arrow). (l) Twin ossification centers are present in the bodies of the L 2–L 6 lumbar vertebrae of the same Gpc3 ⫺/ /⌬H19 double mutant embryo as in (h). The ossification center in L 1 is normal, but this vertebra possesses an extra (14th) pair of rudimentary ribs (arrowhead). (m, n) In Igf1r ⫺/⫺ single mutant embryos (R; e18.5), the tibia is positioned at a right angle to the calcaneous (lines), as in normal mice. (o, p) Some double mutants lacking both GPC3 and IGF1R (G/R) exhibit a deformity resembling a club foot (the calcaneous is not positioned at an angle with respect to the tibia; lines).

© 2002 Elsevier Science (USA). All rights reserved.

192

Chiao et al.

FIG. 3. (a– d) Intestinal abnormality in Gpc3 ⫺ mutants. An example of severe gaseous abdominal distension in a mutant neonate is shown in (a). The typical appearance of gas bubbles in the intestine after dissection of a p5 pup is displayed in (b) and, in blowup, in (c). (d) Gaseous inflation of the cecum in a p15 mutant. (e–i) Spleen phenotypes. The spleen and accessory spleen (arrow) of a Gpc3 ⫺ mutant is shown in (e). Sections of wild-type (f, h) and Gpc3 ⫺ mutant (g, i) spleens from p15 (f, g) and p170 (h, i) mice are compared at low and high magnification, respectively. The mutant spleen is small and, at an advanced age (p170), exhibits a relative enlargement of follicles in an anastomosing pattern (arrow). (j–m) Renal phenotypes. Sections of kidneys from a wild-type neonate (j, k) and a stillborn Gpc3 ⫺ mutant (l, m) are compared in low (j, l) and high magnification (k, m). In the mutant, cystic dysplasia is evident in both the renal cortex and the medulla. The inset in (m) displays a hydronephrotic mutant kidney (cf. k, inset). (n–r) Placental phenotypes. The sections of e18.5 placentas are from: wild-type embryo (n), Gpc3 ⫺ mutant (o), ⌬H19 mutant (p), and Gpc3 ⫺/ /⌬H19 double mutant (q). The macroscopic appearance of the placentas in (n– q) prior to sectioning is shown (in the same order, from top to bottom) in (r). The bar corresponds to 1 cm. (s) In situ hybridization analysis of Gpc3 expression in e17.5 wild-type placenta. Sections were hybridized with an antisense (top, and blowup in the middle) or a sense (control) probe (for details, see text). (t) e16.5 ⌬H19 mutant embryo. (u) A Gpc3 ⫺/ /⌬H19 double mutant littermate of the embryo in (t) exhibiting ompalocele.

193

Genetic Analysis of Glypican-3 Null Mice

Pigmentation defects. The majority of mice lacking GPC3 (50/62; 80%) lacked pigmentation in their feet and in the distal third of their tails (Fig. 2b; the latter feature was detectable, despite tail clipping for genotyping). Interestingly, the same abnormalities were observed in Gpc3 ⫹/⫺ females, but at a lower frequency (15/58; 26%). In addition, ⬃10% of the Gpc3 ⫺/ or Gpc3 ⫺/⫺ and ⬃2% of the Gpc3 ⫹/⫺ mice (6/62 and 1/58, respectively) exhibited white belly spotting (Fig. 2c). The patch of white fur, which was not particularly extensive, was confined to the ventral side around the umbilicus and had sharp borders (intermingling with pigmented hairs was not observed). These pigmentation defects, which were absent from wild-type siblings, suggest that during embryogenesis GPC3 may play some role in neural crest derivatives, making contributions to pigmented tissue (melanocytes) in the trunk. Hypopigmentation has been attributed to failure of neural crest cells to colonize those areas of the skin that are the last to receive migrating melanoblasts (Rawles, 1947). The white-spotting phenotype could be the consequence of defects in the migration, differentiation, or survival of melanocytes. Gpc3 is highly expressed in the paraxial mesoderm surrounding the neural tube, but not in the neural tube itself (Pellegrini et al., 1998). Therefore, one possibility is that GPC3 interacts with extracellular factors that provide directional or survival signals for the neural crest cells as they migrate through mesenchyme. Skeletal defects. A low incidence of polydactyly and rare occurrence of agnathia were observed in mutants lacking GPC3, but other obvious abnormalities were not detected in skeletal preparations (n ⫽ 7) of e18.5 embryos. Notably, cleft palate, a feature observed quite frequently in SGBS, was never detected in our mutant mice either by additional analysis of histological sections (n ⫽ 12) or by examination of many mutant animals during autopsy. The observed polydactyly was always postaxial (Figs. 2d and 2e) and was manifested with incomplete penetrance (6/79 mutants were affected; 7.6%). It affected mostly the forelimbs, but cases with extra digits appearing in both forelimbs and hindlimbs (Fig. 2e) or exclusively in hindlimbs were also observed. Polydactyly was described in 18/73 (⬃25%) SGBS cases (for references, see MIM 312870 and Garganta and Bodurtha, 1992) and was also mostly postaxial (12/18). Apparently, polydactyly was not observed in the Gpc3 ⫺/ mutants described previously (Cano-Gauci et al., 1999; Paine-Saunders et al., 2000). However, Gpc3 ⫺/ hemizygotes that were also heterozygous for the geneencoding bone morphogenetic protein 4 (Bmp4 ⫹/⫺) exhibited postaxial polydactyly with 100% penetrance and variable expressivity (Paine-Saunders et al., 2000), whereas only 12% of the Bmp4 ⫹/⫺ mutants were found to be polydactylous (manifesting preaxial polydactyly of the right hindlimb; Dunn et al., 1997). The combination of BMP4 haploinsufficiency and lack of GPC3 not only increased the severity and changed the pattern of polydactyly from preaxial to postaxial but also resulted in manifestation of sternal and rib malformations not observed with either

mutation alone (Paine-Saunders et al., 2000). Although these results did not reveal mechanistic details, some kind of interaction between two signaling pathways was implied. We think that this is an example of a phenomenon known as “emergenesis” or “digenic inheritance” (emergence of novel phenotypic features in compound mutants carrying independently segregating mutant alleles; see Kajiwara et al., 1994; Helwig et al., 1995; De Arcangelis et al., 1999). An additional skeletal deformity manifested in 2/79 (2.5%) of our mutants was agnathia, i.e., complete absence of the mandible (only a small bud of skin was present instead of the lower jaw; Fig. 2f). This trait also appeared in 10% of other Gpc3 ⫺/ mutants after extensive backcrossing into the B6 background (Cano-Gauci et al., 1999). The observed agnathia was all-or-none, as we did not encounter a variable degree of mandibular hypoplasia (micrognathia). In this regard, the Gpc3 ⫺ mutants differ from Smad2 ⫹/mh1 heterozygous mutants, 10% of which exhibited variable degrees of hypoplasia up to complete absence of the mandible (Nomura and Li, 1998). It is notable that conditional inactivation of Fgf8 in the mesenchyme of the first branchial arch that is derived from the neural crest results in severe craniofacial defects, including agnathia (Trumpp et al., 1999). Double mutants carrying a combination of the Gpc3 and ⌬H19 mutations exhibited skeletal defects, predominantly of the sternum, which were not observed in either of the single mutants (Fig. 2h). The normal, fully developed sternum consists of six ossified segments: the manubrium, four sternebrae, and the xiphoid process (xiphisternum) remaining cartilagenous in its lower part. This segmentation results from the attachment of ribs (of 13 ribs, 7 attach to the sternum). The sternum develops from e12 onward from a pair of parallel condensations (bands) derived from migrating lateral plate mesodermal cells (Chen, 1952; Kaufman, 1992; Kaufman and Bard, 1999). The bands extend caudally, move ventromedially, and eventually fuse with each other in the midline in a craniocaudal sequence. They differentiate into cartilage by e16.0. Subsequent differentiation of cartilage into bone commences from ossification centers in the sternal segments that become well-ossified by the time of birth and are separated by cartilage where the ribs make contact. The most ventral parts of the ribs, which do not ossify but differentiate into costal cartilages, inhibit the hypertrophy of cartilagenous cells of the sternum near the sites of their attachment (in appearance, there is an impression that costal cartilage crosses the sternum; Fig. 2g, arrow). If the attachment of ribs to the sternum occurs in a slightly asymmetrical fashion, the sternebrae become correspondingly oblique. This variation (not true abnormality), which is described as “crankshaft-sternum” (Theiler, 1989), was consistently observed (4/4) in skeletal preparations of single Gpc3 ⫺/ mutants (not shown). In the Gpc3 and ⌬H19 double mutants, however, the ordered pace of the developmental events described above was apparently disrupted more severely. As in other analogous cases, articulation

© 2002 Elsevier Science (USA). All rights reserved.

194

Chiao et al.

between the sternum and the ventral tips of costal cartilages occurred asymmetrically in a staggered fashion, thus allowing undue ossification and fusion of sternal segments. This has been observed as a consequence of experimental amniotic sac puncture performed at e12.5– e13.5 (Chang et al., 1996), and in mutants lacking myogenin (Hasty et al., 1993; Vivian et al., 1999), ETL1 (a SWI2/SNF2 family member; Schoor et al., 1999), ZIC1 (a zinc finger transcription factor; Aruga et al., 1999), ␦EF1 (a zinc finger and homeodomain transcription factor; Takagi et al., 1998) and, interestingly, heparan sulfate 2-sulfotransferase (an enzyme catalyzing the transfer of sulfate to the iduronic acid component of heparan sulfate; Bullock et al., 1998). The same phenotype (staggered attachment of ribs and ossification of the sternum throughout its length; Fig. 2h, arrow) was observed in 3/6 skeletal preparations of Gpc3 and ⌬H19 double mutants that we examined. In the same mutants, the cartilage of the xiphisternum was bifurcated (Fig. 2h, arrowhead), whereas in the remaining three specimens of double mutants, a “punched-out” hole was observed in the cartilagenous part, a feature that is not considered as an abnormality (see Kaufman, 1992). In the latter skeletons, the abnormal, excessive ossification did not involve the entire sternum, but was confined to the region between one or two pairs of attached costal cartilages. In one of these cases, the sternum was also split vertically in the region extending between ribs 4 and 7, suggesting local failure of fusion between the sternal rudiments. In addition to sternal defects, one severely affected Gpc3 and ⌬H19 double mutant animal exhibited unilaterally a fusion of the first two ribs in their middle portion (they were separately attached to the spine and the sternum; Figs. 2i and 2j). The same mutant also possessed an extra (14th) pair of rudimentary ribs (Fig. 2l, arrowhead) associated with the first pair of lumbar vertebrae (L 1), whereas the body of the remaining vertebrae in the lumbar region (L 2–L 6) exhibited two (bilateral) ossification centers (Fig. 2l, arrow), rather than a single one in the middle (Fig. 2k) as normally observed (no defects were detected in the ossification centers of the vertebral arches). In addition to the emergenesis affecting their skeleton, the Gpc3 and ⌬H19 double mutants exhibited a high incidence of omphalocele, a trait never observed in animals carrying either of the single mutations of the compound genotype (omphalocele was manifested in 13/16 double mutant embryos examined at e16.5; cf. Figs. 3u and 3t). Although umbilical hernias appear quite frequently in SGBS patients, the occurrence of omphalocele is rare, in contrast to BWS (reviewed by Eggenschwiler et al., 1997). Unexpectedly, still another phenomenon of emergenesis was encountered in double mutants lacking both GPC3 and IGF1R. When examined at e18.5, 9/14 of these double mutant embryos exhibited a hind limb defect resembling club foot (Figs. 2o and 2p). Specifically, the hind paws were abnormally bent medially with respect to the axis of the tibia. Normally, the bones of the tibia and fibula are approximately positioned at a right angle to the calcaneus

(Figs. 2m and 2n). Examination of skeletal preparations from two of the Gpc3 and Igf1r double mutants indicated that the calcaneus was not positioned at an angle with respect to the tibia (cf. Figs. 2p and 2n). Occurrence of clubfoot (but also clubhand) has been observed in embryos after experimental amniotic sac puncture performed at e12.5– e13.5 (see Kaufman and Chang, 2000). Analogous observations were made with the mouse mutant cl (clubfoot; Robins, 1959). We also note that ⬃30% of Smad3 nullizygotes exhibit medially torqued wrist joints and, less frequently, ankle joints (Datto et al., 1999). Clubfeet have been described in three SGBS patients (Hughes-Benzie et al., 1996; Kim et al., 1999). Intestinal abnormality. A striking, quite frequently observed phenotypic feature of mutant neonates lacking GPC3 was the presence of intestinal gas resulting in abdominal distension that was occasionally severe (Figs. 3a– 3d). Upon dissection, the gas was usually localized in the ileum, but was also found throughout the entire length of the small and large intestine (sometimes including an inflated cecum; Fig. 2d). However, presence of gas in the stomach was not observed. In this regard, the condition of the Gpc3 mutants apparently differs from the insufflation of the entire gastrointestinal tract due to accumulation of air that was described in mouse neonates with feeding and/or breathing difficulties carrying, for example, targeted mutations of the genes Hoxa3 (Chisaka and Capecchi, 1991), Hoxa2 (Gendron-Maguire et al., 1993), goosecoid (Rivera-Pe´ rez et al., 1995; Yamada et al., 1995), etc., or gain-of-function Msx2 transgenes (Winograd et al., 1997). In the latter case, in which the gaseous abdominal distension was associated with a variable degree of micrognathia and cleft secondary palate, it was thought that glossoptosis (posterior malpositioning of the tongue within the cleft in conjunction with the micrognathia) led to airway obstruction and insufflation of the gastrointestinal tract. An analogous respiratory malfunction, which results in swallowing of air during suckling that subsequently reaches the stomach and the intestines in large amounts, was described in mutants of the unidentified locus bradypneic (Green and Jahn, 1977). In the case of Gpc3 mutants, the gaseous abdominal distension was consistently correlated with runting. In the most severe cases, the mutants were significantly growthretarded and moribund, but animals affected less severely were able to survive. In such cases, the gas had dissipated by p18. More detailed analysis at two ages (e18.5 and p15) indicated that the accumulation of gas was not associated with other intestinal abnormalities. Thus, the length and weight of the intestine were normal in the Gpc3 mutants and examination of cross-sections at 1-cm intervals failed to reveal histological differences from the controls (data not shown). Moreover, comparative immunohistochemical analysis of intestinal sections using several neuronal markers confirmed the presence of myenteric plexi in both mutants and controls (not shown). Thus, the presumptive neural crest defect resulting in pigmentation abnormalities

© 2002 Elsevier Science (USA). All rights reserved.

195

Genetic Analysis of Glypican-3 Null Mice

in the Gpc3 mutants was not accompanied by overt anomalies of the enteric nervous system, in contrast to other mouse mutations causing coat-color spotting and megacolon (see Gershon, 1999). The cause of abdominal bloating when GPC3 is lacking remains unknown. The condition is not associated with reduced gastrointestinal motility, while the absence of gastric involvement suggests that occurrence of aerophagia (swallowing of air) is rather unlikely. Analysis of postnatal growth (see below) has shown that all mutant mice, including those without an abdominal distension, grow with a reduced rate before weaning, although they are fed (suckling is successful and milk can be seen in the stomach of mutants in amounts comparable to those present in control siblings). Because the growth retardation of mutants is transient, it is likely to be due to malnutrition resulting, for example, from insufficiency of digestive enzymes. During the suckling period, intestinal lactase displays high activity making the neonates capable of digesting lactose in the milk. This activity then declines, while maltase, sucrase, and other enzymatic activities appear or rise at weaning (see, for example, Aubin et al., 1999). Preliminary measurements at p7 of intestinal disaccharidase activities in randomly selected Gpc3 mutants (n ⫽ 3) and wild-type mice (n ⫽ 4) indicated that the lactase activity was approximately twofold lower in the mutants (9.2 ⫾ 2.0 units vs. 17.5 ⫾ 3.0 in the controls). In contrast, the still low levels of maltase activity were similar in mutants and controls (14.57 ⫾ 1.2 and 12.45 ⫾ 1.7, respectively), whereas the sucrase activity was practically undetectable in the wildtype mice and barely detectable (1.2 ⫾ 0.3 units) in the mutants. It remains to be seen whether the reduction of lactase in the mutants could be correlated with fermentation of undigested lactose by the intestinal flora and production of excess gas (hydrogen, carbon dioxide, and volatile short-chain fatty acids). We note that measurements of blood glucose levels were consistent with the notion that the preweaning growth retardation observed in Gpc3 mutants could be caused by malnutrition (although alternate interpretations for the observed correlation between hypoglycemia and low weight, which is described below, cannot be excluded). At p1, i.e., at a time when the Gpc3 mutant neonates were still larger than their normal siblings (2.32 ⫾ 0.1 g, n ⫽ 9 vs. 1.77 ⫾ 0.04 g, n ⫽ 6), a difference in glucose concentration was not detected (71.0 ⫾ 2.8 and 76.8 ⫾ 4.4 mg/dL for mutants and controls, respectively). At p15, however, when two nonoverlapping body weight groups of Gpc3 mutants were examined (exhibiting either low or close to normal weights), the situation was different. Whereas the blood glucose concentration in the mutants that were not severely affected (average weight 8.95 ⫾ 0.9 g; glucose 130.6 ⫾ 9.0 mg/dL; n ⫽ 7) did not differ from that of controls (10.95 ⫾ 0.7 g; glucose 122.3 ⫾ 6.7 mg/dL; n ⫽ 10), the severely growth-retarded mutants (4.5 ⫾ 0.3 g; n ⫽ 7) were markedly hypoglycemic (glucose 51.7 ⫾ 4.2 mg/dL).

Microsplenia and accessory spleens. Gross examination during autopsy indicated that the spleens of Gpc3 mutants were smaller than normal (cf. Figs. 3f and 3g). Further analysis indicated that the mutant spleen exhibited a negative allometric relationship to body weight (the average spleen to body weight ratio did not exceed ⬃60% of the normal value; data not shown). This was surprising because the spleen phenotype, which was not described previously for Gpc3 mutants (Cano-Gauci et al., 1999; Paine-Saunders et al., 2000), is opposite to the frequently described “splenomegaly” detected in SGBS patients by clinical examination. We note, however, that in one case with an available autopsy record (Opitz et al., 1988), the spleen was 100 g, which, according to our calculations (for formulas, see Kock et al., 1997), was normal for the age (25 months) and body weight (estimated to be ⬃20 kg) of the patient. Despite a significant reduction in spleen size correlating with a histologically detectable decrease in follicular dimensions and level of hemopoietic activity, abnormalities in the architectural arrangement of the white and red pulp compartments were not observed in stillborn Gpc3 mutants (n ⫽ 2; compared with 4 wild-type controls) or in mutants sacrificed at p15 (n ⫽ 4; compared with 8 controls). Moreover, preliminary flow cytometric analysis indicated that the correct populations and proportions of B and T cells were present in the mutants (data not shown). However, histopathological examination of mutant spleens at ⬃6 months of age (n ⫽ 4; compared with 5 controls) demonstrated an apparent enlargement of follicles present in a complex anastomosing pattern (Fig. 3i), in contrast to the discrete and evenly sized normal counterparts in the control specimens (Fig. 3h). We note that, in several Gpc3 mutants, we observed the presence of accessory spleens associated anatomically with the tail of the pancreas (Fig. 3e, arrow). Although such ectopic splenic tissue separate from the main body of the spleen was not detected in any of the wild-type animals that we have examined, it is not straightforward to assign this feature as part of the Gpc3-null phenotype, as accessory spleens can be found in normal mice as a variation. In fact, sometimes they are present at a very high, strain-dependent frequency (for example, ⬃30% of C57BL/6 mice possess accessory spleens; Hummel et al., 1975). Similarly, the occurrence of one or more such supernumerary spleens, which should not be confused with polysplenia (presence of multiple, small spleens instead of a single regular one), has been occasionally described in SGBS patients (Opitz, 1984; LeMerrer et al., 1989; Chen et al., 1993), but this usually asymptomatic feature is found in at least 10% of unrelated autopsies (see Freeman et al., 1993). Renal dysplasia. Previously, the manifestation of renal cystic abnormalities in the absence of GPC3 was either described briefly (Paine-Saunders et al., 2000) or studied in great detail in Gpc3 mutants of the N4 backcross onward (Cano-Gauci, 1999; Grisaru et al., 2001). Therefore, our account of the kidney phenotype need not be extensive and

© 2002 Elsevier Science (USA). All rights reserved.

196

Chiao et al.

it will be mainly focused on differences from the previous reports that described the observed dysplasia as confined to the renal medulla. In contrast, we have also detected cystic changes in the renal cortex (cf. Figs. 3l and 3m with Figs. 3j and 3k). The reason for this discrepancy is unknown and, in our view, cannot be attributed to strain differences (we examined F 2 hybrids rather than progeny of backcrosses), as it concerns only the topography and not the nature of these lesions. Overall, in agreement with previous observations (CanoGauci et al., 1999), we found extensive variability in the degree of severity of this dysplasia, which appeared to correlate with mortality. Thus, we did not encounter significant histopathological findings in the particular embryonic specimens (e16.5– e18.5) that we examined (n ⫽ 3), whereas in stillborn mutants (n ⫽ 7), we observed dilatation of renal tubules throughout the kidneys with a variable degree of severity. Scattered small tubular cysts corresponding predominantly to proximal tubules (with less involvement of distal tubules) were found in both the renal cortex and medulla and were most prominent in the proximity of the junction between these regions (Fig. 3m). At p15, we observed consistently that the renal cortex was thinner in the mutants (n ⫽ 4) than in the controls (n ⫽ 2). This aberration was accompanied by an apparent expansion of the medulla in the form of irregular clusters of distal tubules protruding unevenly beyond the medulla into the renal cortex to variable levels (not shown). In long-lived survivors, however, the kidney abnormalities were not very prominent. Thus, in ⬃6-month-old mice, only a slight distension of the medullary collecting ducts was evident in the mutants (n ⫽ 3), in comparison with the controls (n ⫽ 3), whereas tubular cysts were absent. In a single mutant, the collecting duct abnormality was associated with mild dilatation of the renal pelvis, consistent with early hydronephrosis (not shown). Although we did not keep a detailed record, we note that, in contrast to wild-type mice, hydronephrotic kidneys were observed upon dissection of several adult mutants that were sacrificed because they were found in distress with distended abdomens. Usually, only one of the kidneys was enlarged and filled with clear fluid (Fig. 3m, inset; cf. Fig. 3k, inset). The exact site of the presumptive obstruction for the development of hydronephrosis remains unknown. In addition to somatic overgrowth, the renal phenotype of Gpc3 mutants is the closest feature also listed among the hallmarks of SGBS. Overall, nephromegaly has been described in 19/73 cases (for references, see MIM 312870 and Garganta and Bodurtha, 1992). Hydronephrosis was noted in 3/19 cases, while the kidneys were characterized as “dysplastic” and/or “cystic” in 10/19 cases. However, autopsy details have been reported only twice. “Renal dysplasia” is a descriptive term implying abnormal kidney morphogenesis that is used without histopathological discrimination for a diverse group of renal disorders (see Matsell, 1998). Most dysplasias, however, are associated with the presence of greatly dilated tubules (cysts). The

cystic changes described in one of the SGBS autopsies (4-month-old patient) involved both tubules and glomeruli (Golabi and Rosen, 1984), while in the second available autopsy (death occurred at the age of 3 days), the presence of “bilateral peripheral cortical cystic renal dysplasia” was recorded (Chen et al., 1993). Accordingly, the cortical and medullary cystic abnormalities that we observed in our Gpc3 mutants are seemingly at least analogous to the limited observations from SGBS patients. Placentomegaly. Although most of the SGBS records do not include descriptions of the placenta, significant placentomegaly was reported in two cases (Hughes-Benzie et al., 1994; Yamashita et al., 1995). The weights of these placentas were 900 and 1110 g, i.e., ⬃30 and 60% higher than the normal average of 680 g (see Hindmarsh et al., 2000). Analogous placentomegaly was observed in Gpc3 mutant mice. Thus, the e18.5 placentas of embryos lacking GPC3 (0.1139 ⫾ 0.0027 g; n ⫽ 16) were 38.6% heavier than those of the controls (0.0822 ⫾ 0.0026 g; n ⫽ 13), whereas a dramatic increase in weight was observed in the placentas of Gpc3/⌬H19 double mutants (0.1595 ⫾ 0.0047 g; n ⫽ 24), which were 94% heavier than normal. Considering that the placental weight of single ⌬H19 mutants (0.1282 ⫾ 0.0035 g; n ⫽ 5) was 56% higher than normal, it is interesting that the combined effects of excessive IGF2 (due to the ⌬H19 mutation) and absence of GPC3 in the double mutants were exactly additive (56 ⫹ 38.6 ⫽ 94.6%). These differences in placental weights (wild-type ⬍ Gpc3 ⫺/ or Gpc3 ⫺/⫺ ⬍ ⌬H19 ⬍ double mutant) could be attributed predominantly to corresponding differences in placental disc diameters (see Fig. 3r) with smaller, but easily detectable differences in thickness, especially in the case of the double mutant (see Figs. 3n–3p). However, the overall architectural and cytological appearances of the decidua, spongiotrophoblast, and labyrinthine layers were preserved in the mutants, except that the incremental gains in thickness were due to expansion of the labyrinth, whereas the width of the spongiotrophoblast did not differ from wild type (Figs. 3n–3p). Similar observations, based both on weight measurements and histological analyses, were made for e16.5 placentas of all four genotypes described above (data not shown). The placentas lacking GPC3 were also ⬃22% heavier than wild type at e14.5 (the earliest time point that we have examined; 0.1059 ⫾ 0.0026 g, n ⫽ 30 vs. 0.0866 ⫾ 0.0042 g, n ⫽ 15). Considering that a known effect of IGF2 excess is hyperplasia (see, for example, Ludwig et al., 1996; Eggenschwiler et al., 1997), and assuming that lack of GPC3 has (indirectly) similar consequences, it is likely that the increased placental size observed in the single and double mutant specimens, in which overt histopathological findings were absent, is due to hyperproliferation, potentially through paracrine actions. It is notable, in this regard, that in situ hybridization analysis of Gpc3 expression in normal e17.5 placentas demonstrated the presence of intense signal in all cells of the yolk sac and an additional less intense signal in an arborizing pattern restricted to mesenchymal cords of

© 2002 Elsevier Science (USA). All rights reserved.

197

Genetic Analysis of Glypican-3 Null Mice

the labyrinthine trabeculae, whereas adjacent trophoblastic and endothelial cells were negative for hybridization (Fig. 3s). On the other hand, Gpc3 expression was undetectable in the spongiotrophoblast, i.e., in the placental zone that does not expand in the mutants (Fig. 3s). Expression of Igf2 in the rodent placenta analogous to that of Gpc3 has been detected in the yolk sac and in the trabeculae of the labyrinth (Zhou and Bondy, 1992; Redline et al., 1993; Correia-da-Silva et al., 1999). In contrast to Gpc3, however, Igf2 is expressed in the entire mouse placenta, i.e., in both the spongiotrophoblast and labyrinthine zones (Redline et al., 1993), and also in the structurally homologous regions of the human placenta, the extravillous trophoblast, and the floating chorionic villi, respectively. More specifically, human IGF2 is expressed in the mesodermal core and the surrounding villus cytotrophoblasts, but not in the villous syncytiotrophoblast sheath (reviewed by Han and Carter, 2000). In contrast, both the syncytiotrophoblast and cytotrophoblast layers of the villi express GPC3 at high and low levels, respectively, whereas expression is undetectable in the mesenchymal elements (Khan et al., 2001). These apparent differences between the human and mouse patterns of placental Gpc3 expression require further investigation.

Embryonic and Postnatal Growth of Mutants During development, increase of the mammalian organism in size (growth) is brought about predominantly by proliferative events increasing cell number rather than cell size. In fact, measurements of cell sizes or counting the number of cell nuclei present within a length of 100 ␮m in sections of heart, kidney, and liver of p15 Gpc3, ⌬H19, and Gpc3/⌬H19 mutants did not reveal statistically significant differences from wild type (data not shown). The rate and duration of the proliferation that determines the growth process are reflected in a progressive weight increase until a final body size is attained when a state of balance between proliferation and apoptosis is reached. Thus, growth curves (plots of weight vs. developmental age) and their derivatives (plots of growth rates) can provide a relatively coarse, but informative index in comparing growth patterns between mutant and wild-type mice. Statistical analysis of weight data of wild-type and mutant embryos lacking GPC3 at various embryonic days indicated that the overgrowth of mutants was undetectable at e11.5 (0.0414 ⫾ 0.0017 g, n ⫽ 4 for mutants vs. 0.0366 ⫾ 0.0009, n ⫽ 4 for wild-type controls). However, a statistically significant difference became evident from e12.5 onward (the weights for mutant and control embryos at e12.5 were, respectively, 0.1416 ⫾ 0.0066 g, n ⫽ 9; 0.1113 ⫾ 0.0045 g, n ⫽ 11). At e18.5, the mutant mice were ⬃30% heavier than normal (1.77 ⫾ 0.025 g, n ⫽ 44 vs. 1.35 g, n ⫽ 27), in agreement with previous reports (Cano-Gauci et al., 1999; Paine-Saunders et al., 2000). Interestingly, when the Gpc3 and ⌬H19 mutations were combined, there was no additivity in the excess in embry-

onic weights, in contrast to the results that we described above for the placenta (see Concluding Remarks, for a discussion of this difference). Thus, the weight of e18.5 mutants lacking GPC3 (1.77 g, as reported above) was the same as that of ⌬H19 mutants (1.70 ⫾ 0.06 g, n ⫽ 5) and double mutants (1.70 ⫾ 0.03 g, n ⫽ 25). To study postnatal growth, we compared growth curves (representing cumulative weight at a given time) derived by regression analysis of a robust set of primary weight data (Figs. 4a and 4b). From this analysis, a difference between sexes in wild-type controls was detected only shortly after weaning (p25), as expected (due to the emerging actions of distinct gonadal steroids during the period of pubertal growth spurt). Exactly the same gender-specific growth patterns were observed in mutants lacking GPC3 (Gpc3 ⫺/ or Gpc3 ⫺/⫺). The growth-spurt period is reflected clearly in a graph of absolute growth rate (absolute weight gain per unit time), which is biphasic (Fig. 4c). Although the cumulative weight of a mouse continues to increase during the approximate time interval between p10 and p15 (Fig. 4a), it does so with a lower rate than previously (Fig. 4c). However, after this initial decline, the growth rate increases again starting at ⬃p20 (growth spurt), but only transiently (it begins declining again at a time close to p30). Unexpectedly, we observed that the p10 –p15 decline in growth rate was greater in Gpc3 ⫺/ mutants than in wildtype males, i.e., the mutants exhibited transient growth retardation relative to the controls (Figs. 4c– 4e; the same was observed in a comparison between wild-type and Gpc3 ⫺/⫺ females; data not shown). This phase, however, was followed by a higher-than-normal rate in the mutants during the period of the pubertal growth spurt (Fig. 4c). These two growth phases could also be observed clearly when the already attained weight at each time point was used as a normalizing factor to convert absolute growth rates to specific growth rates (Fig. 4d). At around p40, the specific growth rates were practically indistinguishable between mutants and controls and tended toward zero, as the growth of animals of both genotypes was approaching a steady state. Because of the described relative growth retardation, the cumulative weight of Gpc3 ⫺/ mice, which was ⬃120%N at birth, began decreasing continuously, and between p14 and p19 (Fig. 4e; see also Fig. 2b), the mutants were no longer larger than the controls (in fact, they were slightly smaller on average, and some of them at the low end of the weight distribution exhibited significant runting). This decline in relative weight was followed by a period of catch-up growth (i.e., growth at an accelerated rate after an interval of growth failure). Thus, by the end of the period of observation (p45), the mutants had acquired a relative weight that was the same as that at birth (⬃120%N). This relative weight was maintained throughout adulthood. For example, measurements at p170 showed that the weight of Gpc3 ⫺/ mutants was ⬃25% higher than that of normal mice (41.1 ⫾ 1.63 g vs. 32.8 ⫾ 0.95 g; n ⫽ 5 for each group).

© 2002 Elsevier Science (USA). All rights reserved.

198

Chiao et al.

FIG. 4. Growth analysis. (a, b) Average growth curves (derived as described in Materials and Methods) of male and female mice, respectively. The genotypes were: wild-type mice (W), Gpc3 ⫺/ or Gpc3 ⫺/⫺ mutants (G), Gpc3 ⫹/⫺ heterozygous females (G⫾), Igf2-null (Igf2 ⫹/p⫺) mice (II), Gpc3 ⫺/ /Igf2 ⫹/p⫺ male double mutants (G/II), Gpc3 ⫹/⫺/Igf2 ⫹/p⫺ female double mutants (G⫾/II), and ⌬H19 (H) male mutants. The number of animals of each genotype used to calculate average weights is shown in parentheses. The flags represent standard errors. From the curves in (a), absolute growth rates (dW/dt) and specific growth rates [(dW/dt) ⫻ (1/W)] were calculated and are shown in (c) and (d), respectively. The weight values of the curves in (a) were also used to calculate relative (% of normal) weights that are shown in (e). In (f) and (g), available weight and height values, respectively, of SGBS patients are superimposed on standard growth charts for males (values for the 50th, 75th, 90th, 95th, and 97th percentiles are represented, from bottom to top, by the curves and, in the insets, by dots). Weight or height values for patients older than 20 years are lumped and positioned arbitrarily at a single time point between 250 and 300 months of age. In the insets, each dash represents the weight or height at birth of an individual SGBS patient (in some cases, dashes are superimposed). The data are from: Simpson et al., 1975; Golabi and Rosen, 1984; Tsukahara et al., 1984; Behmel et al., 1984, 1988; Opitz, 1984; Opitz et al., 1988; Neri et al., 1988; LeMerrer et al., 1989; Ko¨ nig et al., 1991; Hughes-Benzie et al., 1992a, 1994, 1996; Garganta and Bodurtha, 1992; Gurrieri et al., 1992; Chen et al., 1993; Yamashita et al., 1995; Lapunzina et al., 1998; Kim et al., 1999; Savarirayan and Bankier, 1999; Okamoto et al., 1999; Wabitsch et al., 2001.

Interestingly, heterozygous Gpc3 ⫹/⫺ females possessing a normal amount of GPC3 but in half of their cells (mosaics because of random X-chromosome inactivation in the soma), exhibited half the excess of body weight over wildtype at p45 (⬃110%N vs. ⬃120%N in Gpc3 ⫺/⫺ females; see Table 2). It is notable that female SGBS carriers exhibit some degree of overgrowth and variably mild SGBS phenotypic features (Hughes-Benzie et al., 1992a). However, complete manifestations of SGBS phenotypes have been described in two females with X;autosome translocations disrupting the GPC3 gene (Punnett, 1994; Pilia et al., 1996;

Hughes-Benzie et al., 1996). Apparently, as in other cases with analogous translocations, skewing in X-chromosome inactivation affected preferentially the structurally normal X (see, for example, Krepischi-Santos et al., 2001, and other references therein). The cause(s) of transiently occurring relative growth retardation and the mechanism of catch-up growth remain unknown. One possibility is that the growth rate is briefly reduced in the mutants because of malnutrition, since the time window for this observation coincides with the intestinal abnormalities that we have described. An analogous

© 2002 Elsevier Science (USA). All rights reserved.

199

Genetic Analysis of Glypican-3 Null Mice

TABLE 2 Additivity of GPC3 and IGF Actions on Body Weight p0

p45 (A) Males

[II(0.85) ⫹ G(1.87) ⫺ W(1.54)]/GII(1.15) ⫽ 1.03 [I ⫹/⫺(1.41) ⫹ G(1.87) ⫺ W(1.54)]/GI ⫹/⫺(1.81) ⫽ 0.96 [R ⫹/⫺(1.49) ⫹ G(1.87) ⫺ W(1.54)]/GR ⫹/⫺(1.85) ⫽ 0.98

[II(17.66) ⫹ G(32.20) ⫺ W(27.12)]/GII(21.2) ⫽ 1.07 [I ⫹/⫺(25.00) ⫹ G(32.20) ⫺ W(27.12)]/GI ⫹/⫺(30.10) ⫽ 1.00 [R ⫹/⫺(24.41) ⫹ G(32.20) ⫺ W(27.12)]/GR ⫹/⫺(28.11) ⫽ 1.05 (B) Females

⫹/⫺

⫹/⫺

[II(0.85) ⫹ G (1.70) ⫺ W(1.54)]/G II(1.03) ⫽ 0.98 [G ⫺/⫺(1.87) ⫺ W(1.54)]/G ⫹/⫺(1.70) ⫺ W(1.54)] ⫽ 2.06

[II(14.59) ⫹ G ⫹/⫺(24.58) ⫺ W(22.41)]/G ⫹/⫺II(17.00) ⫽ 0.99 [G ⫺/⫺(26.68) ⫺ W(22.41)]/G ⫹/⫺(24.58) ⫺ W(22.41)] ⫽ 1.97

Note. With the exception of the last line, the ratios shown, which do not deviate significantly from unity, indicate that the GPC3 and IGF actions on body weight are additive (if there is additivity, the weight of each double mutant should be equal to the weight of the single mutant exhibiting growth deficiency increased by the weight excess of the single mutant exhibiting overgrowth; for details, see text). The calculations shown in the last line indicate that the weight excess of Gpc3 female homozygous mutants is twice that exhibited by heterozygotes. The weights at birth and at the age of 45 days used in the calculations are averages from the following numbers of mice of each genotype shown in parentheses: W (wild-type; 46), G (Gpc3 ⫺/; 38), II (Igf2 ⫹/p⫺; 23), GII (Gpc3 ⫺//Igf2 ⫹/p⫺; 15 for p0; 5 for p45), I ⫹/⫺ (Igf1 ⫹/⫺; 30), GI ⫹/⫺(Gpc3 ⫺//Igf1 ⫹/⫺; 14), R ⫹/⫺(Igf1r ⫹/⫺; 14), GR ⫹/⫺(Gpc3 ⫺//Igf1r ⫹/⫺; 6), G ⫹/⫺(Gpc3 ⫹/⫺; 55), G ⫺/⫺(Gpc3 ⫺/⫺; 24), G ⫹/⫺II (Gpc3 ⫺/⫺/Igf2 ⫹/p⫺; 12).

situation has been observed, for example, in mice nullizygous for the Mf3 gene (encoding a winged helix/forkhead transcription factor), which exhibit early postnatal growth retardation because of potential feeding defects but, if they survive, grow with a normal rate after weaning (Labosky et al., 1997). However, full restoration of body weight to normal does not occur. In contrast, mice nullizygous for the homeobox-containing gene Otx1 (apparently involved in growth hormone gene expression, but only transiently) become growth retarded during the first month of postnatal life, but they recover completely by 4 months of age (Acampora et al., 1998). Catch-up growth (for a review, see Boersma and Wit, 1997) seems to be affected by some kind of systemic compensatory action in response to a mechanism apparently monitoring whole-body size and sensing that a presumptive set-point appropriate for a particular age has not been reached. In this regard, the example of the Gpc3 ⫺/ mutants is currently unprecedented, as it presumably involves set-points in a growth trajectory that are somehow adjusted to the level of overgrowth already attained during embryogenesis. This is quite remarkable because, at the time when the unknown cause of transient growth retardation is eliminated, the size of the mutants has become by default “normal,” and the expectation would have been that they would continue to growth with a rate equal to that of controls. Instead, for a time, they grow with a rate that is higher than normal. Interestingly, however, this process does not overshoot and there is no overgrowth beyond the attainment of a relative body size matching the one at birth. We surmise, therefore, that if the transient growth retardation had not occurred, the Gpc3 ⫺/ mutants would have grown with a normal rate and would have maintained a 120%N relative size throughout their postnatal development (i.e., they would have exhibited a

practically invariable pattern of relative body weight similar to that of ⌬H19 mutants; see Fig. 4e; Leighton et al., 1995; Eggenschwiler et al., 1997). We propose then that the particular GPC3 function that is related to overall somatic growth operates only during embryonic development, i.e., the overgrowth of Gpc3 ⫺/ or Gpc3 ⫺/⫺ mutants is exclusively an embryonic phenomenon. This view seems to conflict with the widely accepted generalization that a predominant phenotypic feature of SGBS is not only prenatal but also postnatal overgrowth. We think, however, that, similarly to the mouse Gpc3 mutants, the overgrowth of SGBS patients also occurs only during embryogenesis and that the expression “postnatal overgrowth” is used loosely to simply describe “big body size,” instead of referring to higher-than-normal growth rate. Horizontal weight and height data for SGBS patients are scarce and limited (for references, see legend to Fig. 4). However, if they are used collectively, including values recorded at single time points, and are superimposed on smoothed percentile curves (ranging between the 50th and 97th percentiles) of postnatal weight and height growth standards for males (http://www.cdc.gov/growthcharts), it becomes immediately obvious that the patient growth data follow normal trajectories (Figs. 4f and 4g). In fact, if the allometric scaling equation y ⫽ a ⫻ x b is used in its logarithmic form logy ⫽ loga ⫹ blogx, where y is body length (BL), x is body weight (BW), b is the slope of the regression line, and a is the y-intercept, it can be seen from the results of regression analysis (not shown) that the equation BL ⫽ 30 ⫻ BW 0.4 (coefficient of correlation r ⫽ 0.99) fits almost exactly the growth standards and also the data for SGBS patients (36 pairs of values from 25 patients). Moreover, as Figs. 4f and 4g show, only very few of the SGBS weight or height values exceed significantly the

© 2002 Elsevier Science (USA). All rights reserved.

200

Chiao et al.

(normal) 97th percentile. However, such graph points should have been the majority, if the postnatal growth rate were higher than normal. This can be surmised from the fact that 67% (18/27) of the known SGBS birth weights and 75% (9/12) of the known height values recorded at birth are above the 97th percentile (Figs. 4f and 4g, insets). Nevertheless, it cannot be suggested from the available evidence that there is transient growth retardation in patients as in mutant mice (growth retardation associated with feeding problems or illness has been described in very few SGBS cases that are difficult to evaluate; see Opitz, 1984; Ko¨ nig et al., 1991). The body size differences between Gpc3 ⫺/ and normal mice were reflected in the weights of various organs examined at embryonic and postnatal ages. Preliminary comparative allometric regression analyses of organ and body weights (log–log plots; data not shown) indicated that the growth of the body and organs were mostly, but not always, proportional in the Gpc3 ⫺/ mutants. Thus, within the limitations of these results, it can be concluded that the heart, lungs, liver, kidney, stomach, and intestines followed closely the changes in mutant body weight, whereas the spleen-to-body allometric relationship was negative (smaller spleen for the same weight, compared to wild type). In contrast, the tongue and possibly the brain exhibited positive allometric relationships in the mutants. Importantly, the detailed analysis of somatic growth (Figs. 4a, 4b, and 4e) allowed us to derive conclusive evidence indicating that apparently there is no direct relationship between IGF signaling and the GPC3 role in somatic growth. The hypothesis that the overgrowth of Gpc3 ⫺/ mutants is brought about by excess of IGF2 resulting from the absence of GPC3 that normally sequesters or downregulates this ligand makes the strong prediction that, in terms of somatic growth, double mutants lacking both IGF2 and GPC3 would be phenotypically indistinguishable from single Igf2 ⫹/p⫺ mutants (proportionate dwarfs). Our data show clearly that this is not the case, as the double mutants are larger than the dwarfs lacking IGF2 (Fig. 4a). Moreover, combination of Gpc3 heterozygosity with lack of IGF2 in females (Gpc3 ⫹/⫺/Igf2 ⫹/p⫺ double mutants) again resulted in mice larger than single Igf2 ⫹/p⫺ female mutants (Fig. 4b). Without further analysis, these results cannot exclude altogether an IGF2 involvement, but can be interpreted minimally as indicating that, if GPC3 interacts with IGF2 at all, this ligand is not the only effector involved in the growth-related function of GPC3. However, when the weights of mutants and controls were compared quantitatively at birth (p0) and at p45 (i.e., at times when relative growth retardation could not affect the calculations), it was possible to show that the actions of IGF2 and GPC3 are just additive and, therefore, most likely unrelated. If W, G, II, and GII are, respectively, the average weights of wild-type controls and Gpc3 ⫺/, Igf2 ⫹/p⫺, and double Gpc3 ⫺//Igf2 ⫹/p⫺ mutants at p0 or p45, the growth excess in the absence of GPC3, as reflected in whole-body weight, is

TABLE 3 Relative Concentrations of IGF2

Serum (e18.5) Embryonic extract (e18.5) Amniotic fluid (e16.5)

Wild type

Gpc3 ⫺/

⌬a

100 ⫾ 14.0 (n ⫽ 5) 100 ⫾ 11.2 (n ⫽ 5)

93.2 ⫾ 16.3 (n ⫽ 5) 83.6 ⫾ 8.2 (n ⫽ 5)

– –

100 ⫾ 9.4 (n ⫽ 3)

94.8 ⫾ 19.9 (n ⫽ 3)

–

a

There is no statistically significant difference (⌬) between the samples from wild-type controls and mutants lacking GPC3.

a ⫽ G ⫺ W, whereas the growth deficit in the absence of IGF2 is b ⫽ W ⫺ II (Table 2). If the actions of GPC3 and IGF2 are independent, the weight GII of the double mutants should be equal to the weight of the mutants lacking IGF2, but increased by the excess a [II ⫹ (G ⫺ W)] or, from the reciprocal point of view, it should be equal to the weight of the mutants lacking GPC3, but decreased by the deficit b [G ⫺(W ⫺ II)]. Therefore, the ratio (G ⫹ II ⫺ W)/GII should be ⬃1.0. If, on the other hand, the weight excess a of Gpc3 ⫺/ mutants is due to the absence of two GPC3 actions (a ⫽ a 1 ⫹ a 2), one of them independent (a 1) and the other related to IGF2 action (a 2), the ratio (G ⫹ II ⫺ W)/GII should be ⬎1 because the a 2 component of the excess would be missing from the double mutants (same numerator, but smaller denominator). As Table 2 shows, when the excess weight of the Gpc3 ⫺/ mutants is added to the weight of Igf2 mutants, the sum matches almost exactly the weight of double mutants, indicating that the actions of GPC3 and IGF2 on somatic growth must be unrelated. This independence is readily evident in the graph of Fig. 4e, where the values are expressed as percentages of normal weight (%N). At p0 and p45, the Gpc3 ⫺/ and Igf2 ⫹/p⫺ mutants are, respectively, ⬃20% heavier and ⬃40% lighter than normal (⬃120%N and ⬃60%N), whereas the double mutants are ⬃20% lighter (⬃80%N). Importantly, additional calculations demonstrated that the described relationships also hold for Gpc3 ⫹/⫺ and Gpc3 ⫹/⫺/Igf2 ⫹/p⫺ female mutants (Table 2). Further supporting evidence was provided by analyzing detailed growth curves (data not shown) of Igf1 ⫹/⫺ and Igf1r ⫹/⫺ heterozygotes, alone or in combination with the Gpc3 ⫺/ mutation and performing calculations, as described above. Clearly, the known haploinsufficiency of Igf1 ⫹/⫺ and Igf1r ⫹/⫺ heterozygotes, as reflected in body weight (90 – 92%N), was not maintained in double mutants also lacking GPC3, but rather the effects of the mutations were additive (see Table 2). Consistent with the conclusion that there is no physical interaction between GPC3 and IGF2, comparative measurements of IGF2 concentrations in the serum and tissues of Gpc3 null and wild-type embryos by Western analysis (Fig. 1d, and Table 3) demonstrated that there is no statistically significant difference between the two groups, i.e., the levels of IGF2 are normal, rather than elevated in the

© 2002 Elsevier Science (USA). All rights reserved.

201

Genetic Analysis of Glypican-3 Null Mice

mutants, in agreement with analogous previous measurements (Cano-Gauci et al., 1999). Additional assays were performed to assess IGF2 levels in amniotic fluid that reportedly contains in humans soluble GPC3 capable of binding IGF2 with high affinity (Xu et al., 1998). Again, a difference between mutants and controls was not detected (Table 3). Interestingly, as already noted for embryos, the absence of GPC3 in double Gpc3 ⫺//⌬H19 mutants did not augment the somatic overgrowth resulting from excess of IGF2. Thus, the birth weight of offspring with a Gpc3 ⫺//⌬H19 genotype was 2.09 ⫾ 0.05 g on average (n ⫽ 21), whereas in single ⌬H19 and Gpc3 ⫺/ mutants, it was 2.02 ⫾ 0.09 g (n ⫽ 19) and 1.90 ⫾ 0.04 g (n ⫽ 38), respectively. At p45, the weights of single male ⌬H19 and Gpc3 ⫺/ mutants were the same (32.7 ⫾ 0.89 and 32.2 ⫾ 0.55 g, respectively). Male double mutants did not survive to this age to be analyzed, but the average weight of three female Gpc3 ⫺/⫺/⌬H19 survivors was 24.87 ⫾ 0.09 g, i.e., slightly lower than that of Gpc3 ⫺/⫺ females (26.68 ⫾ 0.54 g, n ⫽ 24; ⌬H19 females were not available for comparison, as they were not obtained from the G/H cross).

CONCLUDING REMARKS The whole-body overgrowth of Gpc3 ⫺/ mutants that is manifested with concomitant organ-specific differences in allometric relationships to body size, including both positive and negative effects, in conjunction with the detected tissue-specific (regional) histopathological changes reflecting morphogenetic lesions, suggest that GPC3 interacts with various effectors. Unraveling the identity and roles of these effectors will probably turn out to be a slow and difficult process because, in future experiments, the nature of the abnormalities in growth and morphogenesis that the absence of GPC3 provokes would continue to require the use of the entire mammalian organism as an assay system providing a proper context. Still, at least some guiding clues may be gained from in vitro assays, experiments with cultured cells, or paradigms from other model organisms, such as the interactions of a Drosophila glypican (the dally gene product) with components of the Wg/Wnt and other signaling pathways (Tsuda et al., 1999; Lin and Perrimon, 1999, 2000). However, despite a likely participation of some FGF or TGF␤ family members, which potentially interact with GPC3, in regional patterning events (see De Cat and David, 2001), there is no known and immediately recognizable candidate that can be proposed as being involved in general (whole-body) overgrowth. A few comments focusing exclusively on this aspect of the Gpc3 ⫺/ phenotype are presented below. Because the increased growth of Gpc3 ⫺/ mutants corresponds to hyperplasia and not to hypertrophy, a key mechanistic question is whether the overgrowth observed in the absence of GPC3 is brought about by an increase in proliferation or a decrease in apoptosis. The limited information

that is currently available (Gonzalez et al., 1998) suggests that GPC3 apparently promotes apoptosis by reducing antiapoptotic signaling, which should be increased in its absence. However, the effect appears to be cell-specific (at least by the criterion of the cell lines used in the experiments) and, therefore, inconsistent with a general somatic role. Moreover, against expectation from these results, a significant increase in apoptosis was observed in the kidneys of Gpc3 ⫺/ mutants (Grisaru et al., 2001). Considering the situation further, we think that it is quite unlikely that antiapoptotic function could be involved in Gpc3 ⫺/ somatic overgrowth. Apoptotic functions are effected by a complex multicomponent network that presumably includes participants with overlapping roles. Most probably, therefore, opportunistic compensation of function would be provided upon gene targeting of any single member. In fact, although knockout of the antiapoptotic Bcl2 gene resulted in severe postnatal growth retardation (Veis et al., 1993), generalized overgrowth was not observed by knocking-out genes encoding proapoptotic effectors. For example, elimination of either caspase 3 (Kuida et al., 1996; Woo et al., 1998) or caspase 9 (Kuida et al., 1998; Hakem et al., 1998) resulted only in neuronal hyperplasia. Although some of the regional phenotypic characteristics of Gpc3 ⫺/ mutants may be related to apoptosis, we found no evidence that such effects could be predominant, at least in mice. A characteristic example of the developmental role of apoptotic pathways is the sculpting of limbs during embryogenesis. However, interdigital webbing occasionally observed in SGBS patients (Neri et al., 1988; Hughes-Benzie et al., 1992a) was absent from Gpc3 ⫺/ mutants against expectation, although it would have been detectable, had it occurred. Persistence of interdigital tissue that normally undergoes apoptosis and disappears by e15.5 is, for example, a feature of hammertoe mouse mutants (see Ahuja et al., 1997) and of surviving Bax ⫺/⫺Bak ⫺/⫺ double nullizygotes (the corresponding genes encode proapoptotic proteins; Lindsten et al., 2000). The mechanism by which GPC3 participates in overall somatic growth regulation remains elusive. We favor, however, the view that, to play this particular role, the membrane-bound GPC3 interacts with a systemically acting and, therefore, diffusible factor, rather than with components of the extracellular matrix. For simplicity, we will assume that this presumptive factor (G3IF, for “glypican 3-interacting factor”) is unique. For an involvement in proliferation and, therefore, organismal growth, the signaling that G3IF elicits after binding to its cognate receptor should affect the cell cycle. In principle, however, the mode of action can be either positive or negative. GPC3 could be, for example, a coreceptor augmenting the binding of growth-inhibitory G3IF to its receptor. Inefficient binding in Gpc3 ⫺/ mutants would result in reduction of negative signaling, overactivity of growth promoters that are no longer counterbalanced, and overgrowth. An alternative hypothesis, which we prefer because it is more parsimonious, is that G3IF could be a growth-promoting factor that is

© 2002 Elsevier Science (USA). All rights reserved.

202

Chiao et al.

sequestered and becomes ineffective upon binding to GPC3. In this scenario, which is identical to the one originally proposed for IGF2, overgrowth would result from excess of G3IF in Gpc3 ⫺/ mutants. Although, as we have shown here, IGF2 and G3IF are different factors serving independent signaling pathways, we propose a model postulating that there is convergence of their signal transduction relays to common downstream effectors. This view can account for the similarity in Igf2 and Gpc3 expression patterns (DeChiara et al., 1991; Pellegrini et al., 1998) and for the growth saturation phenomenon that we described in Gpc3 ⫺//⌬H19 double mutant embryos, which is accompanied by additive (nonconverging) effects on placental growth of excess IGF2 and postulated excess of G3IF. Moreover, consistent with the model are additional phenotypic manifestations in the same mutants, for example, omphalocele and vertebral anomalies. We have previously described such features in double mutants carrying the ⌬H19 mutation and also lacking the type 2 IGF receptor (IGF2R) because of knockout of the normally active maternal allele of the imprinted cognate gene. These mutants (Igf2r m⫺/⫹/⌬H19) express IGF2 at very high levels because of elimination of IGF2R-mediated IGF2 turnover in combination with biallelic Igf2 transcription (Eggenschwiler et al., 1997). In contrast, the single ⌬H19 mutants do not manifest omphalocele or skeletal anomalies (Leighton et al., 1995; Eggenschwiler et al., 1997). Therefore, the partial phenotypic overlap between the Igf2r m⫺/⫹/ ⌬H19 and Gpc3 ⫺//⌬H19 double mutants, which share the ⌬H19 mutation, is highly suggestive of the existence of common downstream effectors for IGF2 and G3IF, since the presumptive excess of G3IF signaling in Gpc3 ⫺//⌬H19 double mutants or the extra amount of IGF2 (provided by the Igf2r m⫺/⫹ mutation) in Igf2r m⫺/⫹/⌬H19 double mutants appear to have similar consequences. Interestingly, however, G3IF does not simulate IGF2 action in all tissues. For example, the single Igf2r m⫺/⫹ and double Igf2r m⫺/⫹/⌬H19 mutants exhibit cardiac abnormalities (heart enlargement with persistence of early embryonic sinusoids resulting in heart failure) that are absent from Gpc3 ⫺//⌬H19 double mutants. Although SGBS and BWS (see Introduction) are clinically distinct conditions, they exhibit overlapping phenotypic features, including overgrowth, to such a degree that misdiagnosed cases are not uncommon (Hughes-Benzie et al., 1992b; Verloes et al., 1995). Keeping in mind that at least some of the BWS cases can be attributed to IGF2 excess (see Li et al., 1998; Maher and Reik, 2000) and that the Igf2r m⫺/⫹/⌬H19 double mutants manifest features of both BWS and SGBS (Eggenschwiler et al., 1997), we think that our model of independent but converging IGF2 and G3IF signaling can account for the common and distinct phenotypic elements of these two syndromes. Overstimulation by either factor would result in somatic overgrowth associated with common or different, tissue-specific dysmorphogenetic consequences. It may turn out that the principle of convergence could be extended to explain clinical overlaps

between the BWS or SGBS and Perlman syndrome (see Verloes et al., 1995; Coppin et al., 1997; Li et al., 2000) or additional overgrowth conditions.

ACKNOWLEDGMENTS We thank Joe Terwilliger for invaluable help with the statistical analyses, Shirley Tilghman for the ⌬H19 mutant mice, Dimitris Thanos for reagents, and Mian Su, Qiong Li, Lejuan Chatman, and Tommy Kolar for expert technical assistance. This work was supported by NIH Grants HD34526, MH50733 (project 2), and CA75553 (project 3) (to A.E.) and by Telethon Grants E357 and E867 (to G.P.) and Assessorato Igiene e Sanita`, Legge Regionale n.11 del 30.04.1990.

REFERENCES Acampora, D., Mazan, S., Tuorto, F., Avantaggiato, V., Tremblay, J. J., Lazzaro, D., di Carlo, A., Mariano, A., Macchia, P. E., Corte, G., Macchia, V., Drouin, J., Brulet, P., and Simeone, A. (1998). Transient dwarfism and hypogonadism in mice lacking Otx1 reveal prepubescent stage-specific control of pituitary levels of GH, FSH and LH. Development 125, 1229 –1239. Ahuja, H. S., James, W., and Zakeri, Z. (1997). Rescue of the limb deformity in hammertoe mutant mice by retinoic acid-induced cell death. Dev. Dyn. 208, 466 – 481. Aruga, J., Mizugishi, K., Koseki, H., Imai, K., Balling, R., Noda, T., and Mikoshiba, K. (1999). Zic1 regulates the patterning of vertebral arches in cooperation with Gli3. Mech. Dev. 89, 141–150. Aubin, J., Chailler, P., Menard, D., and Jeannotte, L. (1999). Loss of Hoxa5 gene function in mice perturbs intestinal maturation. Am. J. Physiol. 277, C965–C973. Baker, J., Liu, J. P., Robertson, E. J., and Efstratiadis, A. (1993). Role of insulin-like growth factors in embryonic and postnatal growth. Cell 75, 73– 82. Baserga, R., Resnicoff, M., D’Ambrosio, C., and Valentinis, B. (1997). The role of the IGF-I receptor in apoptosis. Vitam. Horm. 53, 65–98. Behmel, A., Plo¨ chl, E., and Rosenkranz, W. (1984). A new X-linked dysplasia gigantism syndrome: Identical with the Simpson dysplasia syndrome? Hum. Genet. 67, 409 – 413. Behmel, A., Plo¨ chl, E., and Rosenkranz, W. (1988). A new X-linked dysplasia gigantism syndrome: Follow up in the first family and report of a second Austrian family. Am. J. Med. Genet. 30, 275–285. Bernfield, M., Go¨ tte, M., Park, P. W., Reizes, O., Fitzgerald, M. L., Lincecum, J., and Zako, M. (1999). Functions of cell surface heparan sulfate proteoglycans. Annu. Rev. Biochem. 68, 729 – 777. Boersma, B., and Wit, J. M. (1997). Catch-up growth. Endocr. Rev. 18, 646 – 661. Brzustowicz, L. M., Farrell, S., Khan, M. B., and Weksberg, R. (1999). Mapping of a new SGBS locus to chromosome Xp22 in a family with a severe form of Simpson–Golabi–Behmel syndrome. Am. J. Hum. Genet. 65, 779 –783. Bullock, S. L., Fletcher, J. M., Beddington, R. S., and Wilson, V. A. (1998). Renal agenesis in mice homozygous for a gene trap mutation in the gene encoding heparan sulfate 2- sulfotransferase. Genes Dev. 12, 1894 –1906. Cano-Gauci, D. F., Song, H. H., Yang, H., McKerlie, C., Choo, B., Shi, W., Pullano, R., Piscione, T. D., Grisaru, S., Soon, S.,

© 2002 Elsevier Science (USA). All rights reserved.

203

Genetic Analysis of Glypican-3 Null Mice

Sedlackova, L., Tanswell, A. K., Mak, T. W., Yeger, H., Lockwood, G. A., Rosenblum, N. D., and Filmus, J. (1999). Glypican3-deficient mice exhibit developmental overgrowth and some of the abnormalities typical of Simpson–Golabi–Behmel syndrome. J. Cell Biol. 146, 255–264. Chang, H. H., Schwartz, Z., and Kaufman, M. H. (1996). Limb and other postcranial skeletal defects induced by amniotic sac puncture in the mouse. J. Anat. 189, 37– 49. Chen, J. M. (1952). Studies on the morphogenesis of the mouse sternum. I. Normal embryonic development. J. Anat. 86, 373– 389. Chen, E., Johnson, J. P., Cox, V. A., and Golabi, M. (1993). Simpson-Golabi-Behmel syndrome: Congenital diaphragmatic hernia and radiologic findings in two patients and follow-up of a previously reported case. Am. J. Med. Genet. 46, 574 –578. Chisaka, O., and Capecchi, M. R. (1991). Regionally restricted developmental defects resulting from targeted disruption of the mouse homeobox gene hox-1.5. Nature 350, 473– 479. Cohen, M. M. (1999). Overgrowth syndromes: An update. Adv. Pediatr. 46, 441– 491. Coppin, B., Moore, I., and Hatchwell, E. (1997). Extending the overlap of three congenital overgrowth syndromes. Clin. Genet. 51, 375–378. Correia-da-Silva, G., Bell, S. C., Pringle, J. H., and Teixeira, N. (1999). Expression of mRNA encoding insulin-like growth factors I and II by uterine tissues and placenta during pregnancy in the rat. Mol. Reprod. Dev. 53, 294 –305. Datto, M. B., Frederick, J. P., Pan, L., Borton, A. J., Zhuang, Y., and Wang, X. F. (1999). Targeted disruption of Smad3 reveals an essential role in transforming growth factor beta-mediated signal transduction. Mol. Cell. Biol. 19, 2495–2504. De Arcangelis, A., Mark, M., Kreidberg, J., Sorokin, L., and GeorgesLabouesse, E. (1999). Synergistic activities of alpha3 and alpha6 integrins are required during apical ectodermal ridge formation and organogenesis in the mouse. Development 126, 3957–3968. De Cat, B., and David, G. (2001). Developmental roles of the glypicans. Semin. Cell Dev. Biol. 12, 117–125. DeChiara, T. M., Efstratiadis, A., and Robertson, E. J. (1990). A growth-deficiency phenotype in heterozygous mice carrying an insulin-like growth factor II gene disrupted by targeting. Nature 345, 78 – 80. DeChiara, T. M., Robertson, E. J., and Efstratiadis, A. (1991). Parental imprinting of the mouse insulin-like growth factor II gene. Cell 64, 849 – 859. Dunn, N. R., Winnier, G. E., Hargett, L. K., Schrick, J. J., Fogo, A. B., and Hogan, B. L. (1997). Haploinsufficient phenotypes in Bmp4 heterozygous null mice and modification by mutations in Gli3 and Alx4. Dev. Biol. 188, 235–247. Efstratiadis, A. (1998). Genetics of mouse growth. Int. J. Dev. Biol. 42, 955–976. Eggenschwiler, J., Ludwig, T., Fisher, P., Leighton, P. A., Tilghman, S. M., and Efstratiadis, A. (1997). Mouse mutant embryos overexpressing IGF-II exhibit phenotypic features of the Beckwith– Wiedemann, and Simpson–Golabi–Behmel syndromes. Genes Dev. 11, 3128 –3142. Elliot, M., and Maher, E. R. (1994). Beckwith–Wiedemann syndrome. J. Med. Genet. 31, 560 –564. Freeman, J. L., Jafri, S. Z., Roberts, J. L., Mezwa, D. G., and Shirkhoda, A. (1993). CT of congenital and acquired abnormalities of the spleen. Radiographics 13, 597– 610.

Garganta, C. L., and Bodurtha, J. N. (1992). Report of another family with Simpson–Golabi–Behmel syndrome and a review of the literature. Am. J. Med. Genet. 44, 129 –135. Gendron-Maguire, M., Mallo, M., Zhang, M., and Gridley, T. (1993). Hoxa-2 mutant mice exhibit homeotic transformation of skeletal elements derived from cranial neural crest. Cell 75, 1317–1331. Ge´ rard, M., Hernandez, L., Wevrick, R., and Stewart, C. L. (1999). Disruption of the mouse necdin gene results in early post-natal lethality. Nat. Genet. 23, 199 –202. Gershon, M. D. (1999). Lessons from genetically engineered animal models. II. Disorders of enteric neuronal development: Insights from transgenic mice. Am. J. Physiol. 277, G262–G267. Golabi, M., and Rosen, L. (1984). A new X-linked mental retardation-overgrowth syndrome. Am. J. Med. Genet. 17, 345– 358. Gonzalez, A. D., Kaya, M., Shi, W., Song, H., Testa, J. R., Penn, L. Z., and Filmus, J. (1998). OCI-5/GPC3, a glypican encoded by a gene that is mutated in the Simpson–Golabi–Behmel overgrowth syndrome, induces apoptosis in a cell line-specific manner. J. Cell Biol. 141, 1407–1414. Green, M. C., and Jahn, C. L. (1977). Bradypneic, a new mutation in mice causing slow breathing, runting, and early death. J. Hered. 68, 150 –156. Grisaru, S., Cano-Gauci, D., Tee, J., Filmus, J., and Rosenblum, N. D. (2001). Glypican-3 modulates BMP- and FGF-mediated effects during renal branching morphogenesis. Dev. Biol. 231, 31– 46. Gurrieri, F., Cappa, M., and Neri, G. (1992). Further delineation of the Simpson–Golabi–Behmel (SGB) syndrome. Am. J. Med. Genet. 44, 136 –137. Hakem, R., Hakem, A., Duncan, G. S., Henderson, J. T., Woo, M., Soengas, M. S., Elia, A., de la Pompa, J. L., Kagi, D., Khoo, W., Potter, J., Yoshida, R., Kaufman, S. A., Lowe, S. W., Penninger, J. M., and Mak, T. W. (1998). Differential requirement for caspase 9 in apoptotic pathways in vivo. Cell 94, 339 –352. Han, V. K., and Carter, A. M. (2000). Spatial and temporal patterns of expression of messenger RNA for insulin-like growth factors and their binding proteins in the placenta of man and laboratory animals. Placenta 21, 289 –305. Hasty, P., Bradley, A., Morris, J. H., Edmondson, D. G., Venuti, J. M., Olson, E. N., and Klein, W. H. (1993). Muscle deficiency and neonatal death in mice with a targeted mutation in the myogenin gene. Nature 364, 501–506. Helwig, U., Imai, K., Schmahl, W., Thomas, B. E., Varnum, D. S., Nadeau, J. H., and Balling, R. (1995). Interaction between undulated and Patch leads to an extreme form of spina bifida in double-mutant mice. Nat. Genet. 11, 60 – 63. Hindmarsh, P. C., Geary, M. P., Rodeck, C. H., Jackson, M. R., and Kingdom, J. C. (2000). Effect of early maternal iron stores on placental weight and structure. Lancet 356, 719 –723. Hughes-Benzie, R. M., Hunter, A. G. W., Allanson, J. E., and Mackenzie, A. E. (1992a). Simpson–Golabi–Behmel syndrome associated with renal dysplasia and embryonal tumor: Localization of the gene to Xqcen-q21. Am. J. Med. Genet. 43, 428 – 435. Hughes-Benzie, R. M., Allanson, J., Hunter, A., and Cole, T. (1992b). The importance of differentiating Simpson–Golabi– Behmel and Beckwith–Wiedemann syndromes. J. Med. Genet. 29, 928. Hughes-Benzie, R. M., Tolmie, J. L., McNay, M., and Patrick, A. (1994). Simpson–Golabi–Behmel syndrome: Disproportionate fe-

© 2002 Elsevier Science (USA). All rights reserved.

204

Chiao et al.

tal overgrowth and elevated maternal serum alpha-fetoprotein. Prenatal Diagn. 14, 313–318. Hughes-Benzie, R. M., Pilia, G., Xuan, J. Y., Hunter, A. G. W., Chen, E., Golabi, M., Hurst, J. A., Kobori, J., Marymee, K., Pagon, R. A., Punnett, H. H., Schelley, S., Tolmie, J. L., Wohlferd, M. M., Grossman, T., Schlessinger, D., and MacKenzie, A. E. (1996). Simpson–Golabi–Behmel syndrome: Genotype/phenotype analysis of 18 affected males from 7 unrelated families. Am. J. Med. Genet. 66, 227–234. Hummel, K. P., Richardson, F. L., and Fekete, E. (1975). Anatomy. In “Biology of the Laboratory Mouse” (E. L. Green, Ed.), pp. 247–307. Dover Publications, New York. Kajiwara, K., Berson, E. L., and Dryja, T. P. (1994). Digenic retinitis pigmentosa due to mutations at the unlinked peripherin/RDS and ROM1 loci. Science 264, 1604 –1608. Kaufman, M. H. (1992). “The Atlas of Mouse Development.” Academic Press, New York. Kaufman, M. H., and Bard, J. B. L. (1999). “The Anatomical Basis of Mouse Development.” Academic Press, New York. Kaufman, M. H., and Chang, H. H. (2000). Studies of the mechanism of amniotic sac puncture-induced limb abnormalities in mice. Int. J. Dev. Biol. 44, 161–175. Khan, S., Blackburn, M., Mao, D. L., Huber, R., Schlessinger, D., and Fant, M. (2001). Glypican-3 (GPC3) expression in human placenta: Localization to the differentiated syncytiotrophoblast. Histol. Histopathol. 16, 71–78. Kim, S., Idowu, O., and Chen, E. (1999). Choledochal cyst in Simpson–Golabi–Behmel syndrome. Am. J. Med. Genet. 87, 267–270. Kock, K. F., Lammert, O., Thobo-Carlsen, B., and Garby, L. (1997). Organ weights in healthy and apparently healthy Danish infants. Am. J. Hum. Biol. 9, 35–38. Koldovsky, O., Asp, N. G., and Dahlqvist, A.(1969). A method for the separate assay of “neutral” and “acid” ␤-galactosidase in homogenates of rat small-intestinal mucosa. Anal. Biochem. 27, 409 – 418. Koops, W. J. (1986). Multiphasic growth curve analysis. Growth 50, 169 –177. Koops, W. J., Grossman, M., and Michalska, E. (1987). Multiphasic growth curve analysis in mice. Growth 51, 372–382. Ko¨ nig, R., Fuchs, S., Kern, C., and Langenbeck, U. (1991). Simpson– Golabi–Behmel syndrome with severe cardiac arrhythmias. Am. J. Med. Genet. 38, 244 –247. Krepischi-Santos, A. C., Carneiro, J. D., Svartman, M., Bendit, I., Odone-Filho, V., and Vianna-Morgante, A. M. (2001). Deletion of the factor IX gene as a result of translocation t(X;1) in a girl affected by haemophilia B. Br. J. Haematol. 113, 616 – 620. Kuida, K., Zheng, T. S., Na, S., Kuan, C., Yang, D., Karasuyama, H., Rakic, P., and Flavell, R. A. (1996). Decreased apoptosis in the brain and premature lethality in CPP32-deficient mice. Nature 384, 368 –372. Kuida, K., Haydar, T. F., Kuan, C. Y., Gu, Y., Taya, C., Karasuyama, H., Su, M. S., Rakic, P., and Flavell, R. A. (1998). Reduced apoptosis and cytochrome c-mediated caspase activation in mice lacking caspase 9. Cell 94, 325–337. Labosky, P. A., Winnier, G. E., Jetton, T. L., Hargett, L., Ryan, A. K., Rosenfeld, M. G., Parlow, A. F., and Hogan, B. L. (1997). The winged helix gene, Mf3, is required for normal development of the diencephalon and midbrain, postnatal growth and the milkejection reflex. Development 124, 1263–1274. Lapunzina, P., Badia, I., Galoppo, C., De Matteo, E., Silberman, P., Tello, A., Grichener, J., and Hughes-Benzie, R. (1998). A patient

with Simpson–Golabi–Behmel syndrome and hepatocellular carcinoma. J. Med. Genet. 35, 153–156. Leighton, P. A., Ingram, R. S., Eggenschwiler, J., Efstratiadis, A., and Tilghman, S. M. (1995). Disruption of imprinting caused by deletion of the H19 gene region in mice. Nature 375, 34 –39. LeMerrer, M., Roume, J., Narcy, F., Larroche, J. C., and Briard, M. L. (1989). Syndrome de Golabi Rosen: Un syndrome rare aux limites nosologiques impre´ cises, e´ chec du diagnostic pre´ natal. J. Genet. Hum. 37, 163–165. Li, C. C. (1987). A genetical model for emergenesis. Am. J. Hum. Genet. 41, 517–523. Li, M., Squire, J. A., and Weksberg, R. (1998). Molecular genetics of Wiedemann–Beckwith syndrome. Am. J. Med. Genet. 79, 253– 259. Li, M., Shuman, C., Fei, Y. L., Cutiongco, E., Bender, H. A., Stevens, C., Wilkins-Haug, L., Day-Salvatore, D., Yong, S. L., Geraghty, M. T., Squire, J., and Weksberg, R. (2001). GPC3 mutation analysis in a spectrum of patients with overgrowth expands the phenotype of Simpson–Golabi–Behmel syndrome. Am. J. Med. Genet. 102, 161–168. Lin, A. E., Neri, G., Hughes-Benzie, R., and Weksberg, R. (1999). Cardiac anomalies in the Simpson–Golabi–Behmel syndrome. Am. J. Med. Genet. 83, 378 –381. Lin, X., and Perrimon, N. (1999). Dally cooperates with Drosophila Frizzled 2 to transduce Wingless signalling. Nature 400, 281–284. Lin, X., and Perrimon, N. (2000). Role of heparan sulfate proteoglycans in cell– cell signaling in Drosophila. Matrix Biol. 19, 303– 307. Lindsay, S., Ireland, M., O’Brien, O., Clayton-Smith, J., Hurst, J. A., Mann, J., Cole, T., Sampson, J., Slaney, S., Schlessinger, D., Burn, J., and Pilia, G. (1997). Large scale deletions in the GPC3 gene may account for a minority of cases of Simpson–Golabi–Behmel syndrome. J. Med. Genet. 34, 480 – 483. Lindsten, T., Ross, A. J., King, A., Zong, W. X., Rathmell, J. C., Shiels, H. A., Ulrich, E., Waymire, K. G., Mahar, P., Frauwirth, K., Chen, Y., Wei, M., Eng, V. M., Adelman, D. M., Simon, M. C., Ma, A., Golden, J. A., Evan, G., Korsmeyer, S. J., MacGregor, G. R., and Thompson, C. B. (2000). The combined functions of proapoptotic Bcl-2 family members bak and bax are essential for normal development of multiple tissues. Mol. Cell 6, 1389 –1399. Liu, J. P., Baker, J., Perkins, A. S., Robertson, E. J., and Efstratiadis, A. (1993). Mice carrying null mutations of the genes encoding insulin-like growth factor I (Igf-1) and type 1 IGF receptor (Igf1r). Cell 75, 59 –72. Ludwig, T., Eggenschwiler, J., Fisher, P., D’Ercole, A. J., Davenport, M. L., and Efstratiadis, A. (1996). Mouse mutants lacking the type 2 IGF receptor (IGF2R) are rescued from perinatal lethality in Igf2 and Igf1r null backgrounds. Dev. Biol. 177, 517–535. Lupu, F., Terwilliger, J. D., Lee, K., Segre, G. V., and Efstratiadis, A. (2001). Roles of growth hormone and insulin-like growth factor 1 in mouse postnatal growth. Dev. Biol. 229, 141–162. Maher, E. R., and Reik, W. (2000). Beckwith–Wiedemann syndrome: imprinting in clusters revisited. J. Clin. Invest. 105, 247–252. Matsell, D. G. (1998). Renal dysplasia: New approaches to an old problem. Am. J. Kidney Dis. 32, 535–543. Morison, I. M., and Reeve, A. E. (1998). Insulin-like growth factor 2 and overgrowth: Molecular biology and clinical implications. Mol. Med. Today 4, 110 –115. Neri, G., Marini, R., Cappa, M., Borrell, P., and Opitz, J. M. (1988). Simpson–Golabi–Behmel syndrome: An X-linked encephalotrophoschisis syndrome. Am. J. Med. Genet. 30, 287–299.

© 2002 Elsevier Science (USA). All rights reserved.

205

Genetic Analysis of Glypican-3 Null Mice

Neri, G., Gurrieri, F., Zanni, G., and Lin, A. (1998). Clinical and molecular aspects of the Simpson–Golabi–Behmel syndrome. Am. J. Med. Genet. 79, 279 –283. Nomura, M., and Li, E. (1998). Smad2 role in mesoderm formation, left–right patterning and craniofacial development. Nature 393, 786 –790. Okamoto, N., Yagi, M., Imura, K., and Wada, Y. (1999). A clinical and molecular study of a patient with Simpson–Golabi–Behmel syndrome. J. Hum. Genet. 44, 327–329. Opitz, J. M. (1984). The Golabi–Rosen syndrome: Report of a second family. Am. J. Med. Genet. 17, 359 –366. Opitz, J. M., Herrmann, J., Gilbert, E. F., and Matalon, R. (1988). Simpson–Golabi–Behmel syndrome: Follow up of the Michigan family. Am. J. Med. Genet. 30, 301–308. Paine-Saunders, S., Viviano, B. L., Zupicich, J., Skarnes, W. C., and Saunders, S. (2000). glypican-3 controls cellular responses to Bmp4 in limb patterning and skeletal development. Dev. Biol. 225, 179 –187. Pellegrini, M., Pilia, G., Pantano, S., Lucchini, F., Uda, M., Fumi, M., Cao, A., Schlessinger, D., and Forabosco, A. (1998). Gpc3 expression correlates with the phenotype of the Simpson– Golabi–Behmel syndrome. Dev. Dyn. 213, 431– 439. Perrimon, N., and Bernfield, M. (2000). Specificities of heparan sulphate proteoglycans in developmental processes. Nature 404, 725–728. Perrimon, N., and Bernfield, M. (2001). Cellular functions of proteoglycans: An overview. Semin. Cell Dev. Biol. 12, 65– 67. Pilia, G., Hughes-Benzie, R. M., MacKenzie, A., Baybayan, P., Chen, E. Y., Huber, R., Neri, G., Cao, A., Farabosco, A., and Schlessinger, D. (1996). Mutations in GPC3, a glypican gene, cause the Simpson–Golabi–Behmel overgrowth syndrome. Nat. Genet. 12, 241–247. Punnett, H. H. (1994). Simpson–Golabi–Behmel syndrome (SGBS) in a female with an X-autosome translocation. Am. J. Med. Genet. 50, 391–393. Rawles, M. E. (1947). Origin of pigment cells from the neural crest in the mouse embryo. Physiol. Zool. 20, 248 –266. Redline, R. W., Chernicky, C. L., Tan, H. Q., Ilan, J., and Ilan, J. (1993). Differential expression of insulin-like growth factor-II in specific regions of the late (post day 9.5) murine placenta. Mol. Reprod. Dev. 36, 121–129. Robins, M. W. (1959). A mutation causing congenital clubfoot in the house mouse. J. Hered. 50, 188 –192. Rivera-Perez, J. A., Mallo, M., Gendron-Maguire, M., Gridley, T., and Behringer, R. R. (1995). Goosecoid is not an essential component of the mouse gastrula organizer but is required for craniofacial and rib development. Development 121, 3005–3012. Savarirayan, R., and Bankier, A. (1999). Simpson–Golabi–Behmel syndrome and attention deficit hyperactivity disorder in two brothers. J. Med. Genet. 36, 574 –576. Schoor, M., Schuster-Gossler, K., Roopenian, D., and Gossler, A. (1999). Skeletal dysplasias, growth retardation, reduced postnatal survival, and impaired fertility in mice lacking theSNF2/SWI2 family member ETL1. Mech. Dev. 85, 73– 83. Selleck, S. B. (1999). Overgrowth syndromes and the regulation of signaling complexes by proteoglycans. Am. J. Hum. Genet. 64, 372–377. Selleck, S. B. (2000). Proteoglycans and pattern formation: Sugar biochemistry meets developmental genetics. Trends Genet. 16, 206 –212. Selleck, S. B. (2001). Genetic dissection of proteoglycan function in Drosophila and C. elegans. Semin. Cell Dev. Biol. 12, 127–134.

Simpson, J. L., Landey, S., New, M., and German, J. (1975). A previously unrecognized X-linked syndrome of dysmorphia. Birth Defects 11, 18 –24. Simpson, E. M., Linder, C. C., Sargent, E. E., Davisson, M. T., Mobraaten, L. E., and Sharp, J. J. (1997). Genetic variation among 129 substrains and its importance for targeted mutagenesis in mice. Nat. Genet. 16, 19 –27. Song, H. H., Shi, W., and Filmus, J. (1997). OCI-5/rat glypican-3 binds to fibroblast growth factor-2 but not to insulin-like growth factor-2. J. Biol. Chem. 272, 7574 –7577. Sun, F. L., Dean, W. L., Kelsey, G., Allen, N. D., and Reik, W. (1997). Transactivation of Igf2 in a mouse model of Beckwith– Wiedemann syndrome. Nature 389, 809 – 815. Takagi, T., Moribe, H., Kondoh, H., and Higashi, Y. (1998). ␦EF1, a zinc finger and homeodomain transcription factor, is required for skeleton patterning in multiple lineages. Development 125, 21–31. Terespolsky, D., Farrell, S. A., Siegel-Bartelt, J., and Weksberg, R. (1995). Infantile lethal variant of Simpson–Golabi–Behmel syndrome associated with hydrops fetalis. Am. J. Med. Genet. 59, 329 –333. Theiler, K. (1989). “The House Mouse: Atlas of Embryonic Development.” Springer-Verlag, New York. Threadgill, D. W., Yee, D., Matin, A., Nadeau, J. H., and Magnuson, T. (1997). Genealogy of the 129 inbred strains: 129/SvJ is a contaminated inbred strain. Mamm. Genome 8, 390 –393. Trumpp, A., Depew, M. J., Rubenstein, J. L., Bishop, J. M., and Martin, G. R. (1999). Cre-mediated gene inactivation demonstrates that FGF8 is required for cell survival and patterning of the first branchial arch. Genes Dev. 13, 3136 –3148. Tsuda, M., Kamimura, K., Nakato, H., Archer, M., Staatz, W., Fox, B., Humphrey, M., Olson, S., Futch, T., Kaluza, V., Siegfried, E., Stam, L., and Selleck, S. B. (1999). The cell-surface proteoglycan Dally regulates Wingless signalling in Drosophila. Nature 400, 276 –280. Tsukahara, M., Tanaka, S., and Kajii, T. (1984). A Weaver-like syndrome in a Japanese boy. Clin. Genet. 25, 73–78. Veis, D. J., Sorenson, C. M., Shutter, J. R., and Korsmeyer, S. J. (1993). Bcl-2-deficient mice demonstrate fulminant lymphoid apoptosis, polycystic kidneys, and hypopigmented hair. Cell 75, 229 –240. Verloes, A., Massart, B., Dehalleux, I., Langhendries, J.-P., and Koulischer, L. (1995). Clinical overlap of Beckwith–Wiedemann, Perlman and Simpson–Golabi–Behmel syndromes: A diagnostic pitfall. Clin. Genet. 47, 257–262. Veugelers, M., Cat, B. D., Muyldermans, S. Y., Reekmans, G., Delande, N., Frints, S., Legius, E., Fryns, J. P., SchranderStumpel, C., Weidle, B., Magdalena, N., and David, G. (2000). Mutational analysis of the PC3/GPC4 glypican gene cluster on Xq26 in patients with Simpson–Golabi–Behmel syndrome: Identification of loss-of-function mutations in the GPC3 gene. Hum. Mol. Genet. 9, 1321–1328. Vivian, J. L., Gan, L., Olson, E. N., and Klein, W. H. (1999). A hypomorphic myogenin allele reveals distinct myogenin expression levels required for viability, skeletal muscle development, and sternum formation. Dev. Biol. 208, 44 –55. Wabitsch, M., Brenner, R. E., Melzner, I., Braun, M., Mo¨ ller, P., Heinze, E., Debatin, K. M., and Hauner, H. (2001). Characterization of a human preadipocyte cell strain with high capacity for adipose differentiation. Int. J. Obes. 25, 8 –15.

© 2002 Elsevier Science (USA). All rights reserved.

206

Chiao et al.

Weksberg, R., and Squire, J. A. (1996). Molecular biology of Beckwith–Wiedemann syndrome. Med. Pediatr. Oncol. 27, 462– 469. Weng, E. Y., Mortier, G. R., and Graham, G. M. (1995). Beckwith– Wiedemann syndrome: An update and review for the primary pediatrician. Clin. Pediatr. 34, 317–326. Wilkinson, D. G. (1992). Whole-mount in situ hybridization of vertebrate embryos. In “In Situ Hybridization: A Practical Approach” (D. G. Wilkinson, Ed.), pp. 75– 83. IRL Press, Oxford. Winograd, J., Reilly, M. P., Roe, R., Lutz, J., Laughner, E., Xu, X., Hu, L., Asakura, T., vander Kolk, C., Strandberg, J. D., and Semenza, G. L. (1997). Perinatal lethality and multiple craniofacial malformations in MSX2 transgenic mice. Hum. Mol. Genet. 6, 369 –379. Wolffe, A. P. (2000). Transcriptional control: Imprinting insulation. Curr. Biol. 10, R463–R465. Woo, M., Hakem, R., Soengas, M. S., Duncan, G. S., Shahinian, A., Kagi, D., Hakem, A., McCurrach, M., Khoo, W., Kaufman, S. A., Senaldi, G., Howard, T., Lowe, S. W., and Mak, T. W. (1998). Essential contribution of caspase 3/CPP32 to apoptosis and its associated nuclear changes. Genes Dev. 12, 806 – 819. Xu, Y., Papageorgiou, A., and Polychronakos, C. (1998). Developmental regulation of the soluble form of insulin-like growth factor-II/mannose 6-phosphatereceptor in human serum and amniotic fluid. J. Clin. Endocrinol. Metab. 83, 437– 442.

Xuan, J. Y., Hughes-Benzie, R. M., and MacKenzie, A. E. (1999). A small interstitial deletion in the GPC3 gene causes Simpson– Golabi–Behmel syndrome in a Dutch-Canadian family. J. Med. Genet. 36, 57–58. Yagi, T., Ikawa, Y., Yoshida, K., Shigetani, Y., Takeda, N., Mabuchi, I., Yamamoto, T., and Aizawa, S. (1990). Homologous recombination at c-fyn locus of mouse embryonic stem cells with use of diphtheria toxin A-fragment gene in negative selection. Proc. Natl. Acad. Sci. USA 87, 9918 –9922. Yamada, G., Mansouri, A., Torres, M., Stuart, E. T., Blum, M., Schultz, M., De Robertis, E. M., and Gruss, P. (1995). Targeted mutation of the murine goosecoid gene results in craniofacial defects and neonatal death. Development 121, 2917–2922. Yamashita, H., Yasuhi, I., Ishimaru, T., Matsumoto, T., and Yamabe, T. (1995). A case of nondiabetic macrosomia with Simpson– Golabi–Behmel syndrome: Antenatal sonographic findings. Fetal Diagn. Ther. 10, 134 –138. Zhou, J., and Bondy, C. (1992). Insulin-like growth factor-II and its binding proteins in placental development. Endocrinology 131, 1230 –1240.

© 2002 Elsevier Science (USA). All rights reserved.

Received for publication August 30, Revised December 3, Accepted December 3, Published online January 23,

2001 2001 2001 2002