The control of growth, patterning, and differentiation of the mammalian forebrain has a large genetic component, and many human disease loci associated with cortical malformations have been identified. To further understand the genes involved in controlling neural development, we have performed a forward genetic screen in the mouse (Mus musculus) using ENU mutagenesis. We report the results from our ENU screen in which we biased our ascertainment toward mutations affecting neurodevelopment. Our screen had three components: a careful morphological and histological examination of forebrain structure, the inclusion of a retinoic acid response element-lacZ reporter transgene to highlight patterning of the brain, and the use of a genetically sensitizing locus, Lis1/Pafah1b1, to predispose animals to neurodevelopmental defects. We recovered and mapped eight monogenic mutations, seven of which affect neurodevelopment. We have evidence for a causal gene in four of the eight mutations. We describe in detail two of these: a mutation in the planar cell polarity gene scribbled homolog (Drosophila) (Scrib) and a mutation in caspase-3 (Casp3). We find that refining ENU mutagenesis in these ways is an efficient experimental approach and that investigation of the developing mammalian nervous system using forward genetic experiments is highly productive.
THE early stages of mammalian forebrain development involve early tissue patterning events followed by an exquisitely coordinated series of cell division and differentiation events. The mammalian telencephalon develops from the most rostral portion of the developing neural tube at very early embryonic stages and ultimately gives rise to the neocortex, hippocampus, olfactory bulbs, and basal ganglia. A precise regional regulation of proliferation (and cell death) is crucial for ultimately creating the proper brain architecture. After terminal division and acquisition of neuronal identity, neural cells must migrate via long, and often circuitous, routes to their final destination and begin the processes of synapse and circuit formation. All of the above events require a wide variety of cell biological processes to occur with high fidelity. While a clearer understanding of the genetic regulation of forebrain development is slowly emerging, much remains to be discovered. Further understanding the formation of the brain will likely help to decipher the causes of many brain diseases (Herronet al. 2002), including developmental disorders, degenerative diseases of adult life, and tumorigenesis.
One particularly powerful approach to understanding human gene function is the study of the mouse. Murine anatomy and physiology are very similar to that of humans, and there are significant genetic tools available in the mouse. While genetics in the mouse has generally been extremely informative for developmental biology, it is perhaps surprising that relatively few mouse mutants are informative for the many complex processes of brain patterning. This may be because many of the genes required for brain patterning have roles in early embryonic development, leading to death or significant phenotypes that preclude later analysis of forebrain patterning [e.g., Fgf8 (Meyerset al. 1998), Shh (Chianget al. 1996), and Bmpr1a (Mishinaet al. 1995)]. Consequently, crucial neurodevelopmental genes are not easily studied at organogenic stages.
In an attempt to learn more about the molecular regulation of mammalian neurodevelopment, we have performed a screen of mice treated with ENU for recessive, monogenic mutations. The treatment of mice with ENU is an highly effective means for generating heritable changes in the genome with low morbidity and/or mortality (Stottmann and Beier 2010). Lines of mutagenized mice can be efficiently screened for mutations causing organogenic phenotypes (Herronet al. 2002). Previous experiments have shown that mutagenesis can uncover mutations affecting the late embryonic brain (Herronet al. 2002; Zarbaliset al. 2004), but the tissue is complex and not easily examined using histology or molecular analysis. We sought to determine if ENU mutagenesis can be used in a more tissue-specific manner and have taken three complementary approaches to enrich our screen for mutations specifically affecting forebrain development. We have performed morphological and histological analysis of the forebrain, incorporated a reporter allele to highlight the development of distinct structures in the developing forebrain, and genetically sensitized the mutagenized population to increase the incidence of neurodevelopmental defects.
Here we report the results of this mutagenesis experiment where we mapped eight monogenic mutations, seven of which have significant neurodevelopmental defects. All of these are mapped, and we report the causal locus for four. We further describe two of our mutations. One is a new allele of scribbled homolog (Drosophila) (Scrib), which carries a mutation in a potentially informative region of the protein. We also report a new allele of caspase-3 (Casp3), a regulator of apoptosis crucial for neural and craniofacial development. Finally, we also discuss the success of each attempt to bias the screen toward neurodevelopmental phenotypes.
Materials and Methods
Mutagenesis and breeding
ENU mutagenesis followed standard protocols (Stottmann and Beier 2010). In brief, 6- to 8-week-old A/J mice (Jackson Labs) were injected i.p. with a fractionated dose of ENU once weekly for 3 weeks with 90, 95, or 100 mg/kg ENU (Sigma). A/J mice were used due to their ability to tolerate larger doses of ENU with acceptable morbidity and mortality, thus increasing their mutation load and increasing productivity of the screen (Weberet al. 2000). ENU was prepared using standard methods and dissolved in ethanol. After allowing for a period of infertility following the mutagenesis, a standard three-generation breeding scheme was followed (Bodeet al. 1988; Herronet al. 2002; also see Results and Figure 1). The A/J males were crossed with FVB/J females (Jackson Labs), retinoic acid response element (RARE)-lacZ–positive females, or Lis1 heterozygous females (largely from a 129 background). The resulting G2 females were backcrossed to generate G3 embryos or postnatal day (P) 14 pups. For timed pregnancies, successful matings were identified by the presence of a copulation plug, and noon of the day of detection was set as embryonic day (E) 0.5. All animals were housed in accordance with the Harvard Medical School Center for Animal Resources and Comparative Medicine.
RARE-lacZ mice were genotyped with standard lacZ primers (F-TTTAACGCCGTGCGCTGTTCG, R-GATCCAGCGATACAGCGCGTC). Lis1 heterozygotes were identified using primers for the neomycin resistance cassette (F-TCCTGCCGAGAAAGTATCCATCAT, R-CCAGCCGGCCACAGTCGT) as previously described (Hirotsuneet al. 1998). Looptail mice carry a mutation in the vang-like 2 (Vangl2) gene, and heterozygous mice were identified by their abnormally looped tail. Direct sequencing of the mutation (in the eighth exon of vangl2) was done to genotype embryos (F-AGAGGATGAAGGGTGGGTG, R-GTGTCAGGGCCAGAGAACC).
Whole-genome single nucleotide polymorphism scanning
To map our mutations, we genotyped a custom whole-genome panel of 768 single nucleotide polymorphism (SNPs) with the Illumina GoldenGate technology at the Broad Institute Center for Genotyping and Analysis. A total of 256 of the 768 SNPs on this panel were polymorphic between A/J and FVB/J. The initial genomic interval carrying the mutation was defined by flanking SNPs, which identified a homozygous A/J haplotype shared by all affected animals. Fine mapping was done by analyzing further meioses with microsatellite markers, restriction fragment length polymorphisms [identified via the MGI database http://www.informatics.jax.org/strains_SNPs.shtml or the algorithm SNP2RFLP (Becksteadet al. 2008)], or direct sequencing of SNPs. Sequencing of candidate genes was done with standard methods, either exon-based sequencing or using random-primed cDNA from mutant RNA for transcript analysis. Primer design was either with Primer3 (http://frodo.wi.mit.edu/primer3/) or via the Exon Primer function in the University of California at San Francisco (UCSC) Genome browser (http://genome.ucsc.edu/). All DNA sequencing other than SNP mapping was done at the Dana-Farber/Harvard Cancer Center DNA Resource Core. In two of these mutant lines, crn2 and hith2, our initial mapping was with a cohort of affected animals with multiple phenotypes and resulted in two candidate regions. Analysis of more embryos and standard microsatellite mapping demonstrated that not all phenotypes were allelic and allowed us to focus on the chromosomal location reported in Table 2 for mapping crn2 and hith2.
Histology and immunohistochemistry
Histological analysis used standard methods (Nagy 2003). Embryos were fixed in either Bouin's fixative or 4% paraformaldehyde followed by paraffin embedding and sectioning at 14 μm. Sections were stained with hematoxylin and eosin or Cresyl violet. Caspase-3 immunohistochemistry was performed on paraffin sections following the manufacturer's instructions using reagents described below. After deparaffinization of the slides, antigen retrieval was performed with Citra Buffer (Biogenex). Blocking against background from avidin–biotin immunohistochemistry was done with commercial reagents (DAKO, Vectorlab). Slides were incubated with a primary antibody against full-length caspase-3 (Cell Signaling) overnight at 4° at 1:200. Visualization was with a biotin-labeled secondary antibody (Vectastain Universal ABC kit) and counterstained with hematoxylin QS (Biogenex).
Analysis of crn2 cochlear phenotype
E18 whole-mount cochleae were fixed in 4% paraformaldehyde, dissected to expose the sensory epithelium, and stained with Alexa Fluor-conjugated phalloidin (1:200; Invitrogen). Images shown are confocal projections of the cochlear sensory epithelium. Absolute rotation of hair-cell stereocilia bundles was assessed at positions of 5, 25, and 50% of the total cochlear length from the basal end of the duct, as previously described (Montcouquiolet al. 2003). Data are presented as the average deviation from 0°; error bars represent SEM. Significant differences were determined by two-tailed t-test.
We performed an independent ENU mutagenesis experiment in a similar manner to our previous experiments done with the goal of obtaining novel mouse mutations that model human congenital defects (Herronet al. 2002). We used a standard three-generation breeding scheme designed for the recovery of recessive traits (McDonald and Bode 1988; Kasarskiset al. 1998; Herronet al. 2002; Figure 1). An initial population of mice (G0) was treated with ENU to induce random single nucleotide mutations throughout the animals, including the germline. These were used to create male offspring (G1), each of which will carry some proportion of the mutations in the paternal gametes. The G1 male animals are again outcrossed to create the G2 generation, each member of which has a 50% chance of carrying any specific mutation. The G2 female progeny are then mated back to the G1 males in an attempt to homozygose an ENU mutation in the G3 generation. These embryos are analyzed for gross morphological abnormalities at E18.5, one day before birth, which allows for significant organogenesis and the development of phenotypes that would be lethal upon birth (Herronet al. 2002). We have found this strategy to be useful for the discovery of phenotypes that model defects found in human populations (Ackermanet al. 2005). As in our previous studies (Herronet al. 2002), we have performed the mutagenesis breeding as an outcross. We mutagenized males from an A/J background and crossed to FVB/J females to create the G1 generation. The G2 generation was again generated with an FVB/J outcross. This strategy ultimately creates genetically heterogeneous G3 embryos, which are informative for mapping purposes.
We used three distinct strategies to perform our phenotypic analysis in efforts to more specifically query neurodevelopment. First, embryos were carefully examined for gross phenotypes with an initial examination to assess overall body patterning, including features such as head size and appearance, limb patterning, digit number, and tail morphology (Figure 1A). This was followed by removal of the brain from the skull for a more detailed examination, noting such features as presence/absence of the anterior-most olfactory bulbs, size and location of the telencephalic vesicles, and midbrain/hindbrain structures. Each brain was then embedded in paraffin and sectioned in the coronal plane to more carefully examine the cortical architecture.
While this style of gross embryonic phenotyping has been fruitful (Kasarskiset al. 1998; Herronet al. 2002) and has yielded interesting neurodevelopmental mutations, it is difficult to observe any fine detail in brain development using this approach. Histological or immunohistochemical analysis is potentially useful but may reduce the efficiency of screening. Another technique commonly used in mutagenesis experiments in flies and worms has been to incorporate a transgenic “reporter” allele, which highlights a structure of interest that would not otherwise be visible to the investigator. Thus, the sensitivity of the screen is greatly enhanced by employing either a simple histochemical reaction or visualization using fluorescence microscopy. We chose to use the RARE-lacZ transgenic mouse line (Rossantet al. 1991) to highlight several structures in the brain that we would otherwise not be able to readily score, including the optic nerves and hippocampus (Figure 1B). The RARE-lacZ transgene contains a retinoic acid response element immediately upstream of a β-galactosidase reporter cassette and is expressed in a gradient from low levels rostrally to high levels caudally in the forebrain (Luoet al. 2004). We chose the RARE-lacZ reporter allele for these expression patterns in discrete regions of the brain, not as part of a screen designed to obtain alleles with a role in retinoic acid signaling. The use of a lacZ-based reporter instead of GFP offers some flexibility in phenotyping the embryos, which can be useful when large numbers of embryos are dissected at once.
In a further attempt to enrich our screen for defects in neurodevelopment, we employed an approach commonly used in invertebrate mutagenesis experiments, but less commonly used in mammalian studies: the incorporation of a mouse carrying a mutation that genetically sensitizes, or predisposes, our mutagenized population toward defects in cortical development (Figure 1C). Mutations in Lis1 (Pafah1b1) cause lissencephaly (smooth brain) and neural migration defects in both mice and humans (Reineret al. 1993; Hirotsuneet al. 1998; Gambelloet al. 2003). Previous work has shown that Lis1 interacts genetically with several components of reelin signaling (Assadiet al. 2003), an essential pathway for cortical development. Mice heterozygous for mutations in Lis1 have a low incidence of hydrocephalus (7%) (Assadiet al. 2003). The incidence of hydrocephaly is dramatically increased when heterozygosity for Lis1 is combined with mutations in any one of several genes in the reelin pathway [e.g., 85% in rl/rl;lis1/+ animals (Assadiet al. 2003)]. We hypothesized that, by analyzing ENU mutations on a background with reduced Lis1 gene function, we may identify more components of the reelin pathway as well as other genes essential for normal brain development. That is, the heterozygous Lis1 mice represent a sensitized background, which will facilitate the identification of other loci that affect neuronal migration. The hydrocephalus phenotype is easily ascertained, making this approach logistically simple and amenable for the analysis of large numbers of progeny. We therefore allowed G3 animals from this arm of the screen to be born and we scored for hydrocephalic pups at P14. This approach was particularly attractive, given that interactions with a heterozygous Lis1 allele were observed for both heterozygous and homozygous mutations at the secondary locus, which could markedly enhance the yield of the screen (Assadiet al. 2003).
All three of these approaches (gross examination, inclusion of a reporter, or sensitizing allele) can be used with the same cohort of G1 animals. There is no difference between the breeding strategies until the mating of G1 males to create the G2 females (Figure 1). Furthermore, if mice carrying a modifying allele (reporter or sensitizing) locus can be bred to homozygosity, this precludes the need to genotype the G2 offspring females for that allele.
In this study, 76 G1 males were bred, 52 independent lines (pedigrees) were established, 4211 G3 progeny were screened from 535 litters, and 38 lines were comprehensively analyzed (Table 1). We consider an examination of three or more litters of at least eight embryos a comprehensive analysis, as this provides an ∼80% likelihood of identifying a fully penetrant monogenic mutation (Stottmann and Beier 2010). Fourteen abnormal phenotypes were found, of which 8 behaved as heritable monogenic traits (Table 2). These have all been mapped, and the mutant gene is identified for 4. Seven of these phenotypes affected development of the embryonic nervous system, and several of these are likely to be either novel genes or mutations in genes not previously known to be required for neurodevelopment.
Most mutant mice were identified by morphological or histological analysis. In the second arm of the screen in which we employed a lacz reporter, 1344 mice from 176 litters (16 lines) were screened, and 12 mice from multiple lines with abnormal lacZ expression profiles were identified. However, the RARE-lacZ reporter allele that we used in this screen was found to have significant variability in expression as we introduced different genetic backgrounds during the course of the breeding scheme in Figure 1B, and none of the abnormal phenotypes identified by the reporter proved consistent and heritable. This effect of genetic background on gene expression has been noted before in the mouse genetics literature (e.g., Hebert and Mcconnell 2000).
In the third arm of the screen, we analyzed 545 mice from 114 litters (eight lines). Using this approach, we identified two lines with obvious postnatal defects, including the hydrocephalus phenotype that we hoped to ascertain. However, neither of these phenotypes proved to be dependent on simultaneous heterozygosity for the Lis1 allele; that is, neither occurred as a consequence of a genetic interaction with the sensitizing allele.
Crn2 is a novel allele of Scrib
A mutation identified via gross embryonic examination was the craniorachischisis 2 (crn2) mutation. Most crn2 embryos (n = 32) had an open neural tube along the entire extent of the anterior–posterior axis and a kinked tail (Figure 2, B and D). We also noted exencephaly in a few embryos (both crn2/+ heterozygotes and homozygotes). Our initial genetic interval included Scrib, a known planar cell polarity (PCP) gene, which we considered as a candidate locus on the basis of the similarity of the crn2 phenotype to published alleles (Rachelet al. 2000; Murdochet al. 2001; Zarbaliset al. 2004). Sequencing of the crn2 transcripts revealed a missense mutation in the 19th exon of the Scrib transcript, converting a highly conserved glutamic acid residue at position aa 800 to glycine (Figure 2, E–G). The minimal interval from our initial mapping data also included the gene cadherin, EGF LAG seven-pass G-type receptor 1 (flamingo homolog) (Drosophila) (Celsr1). Celsr1 mouse mutants also have PCP defects, including craniorachischisis. However, sequence analysis of recombinant mice with the crn2 phenotype identified heterozygous SNPs in Celsr, which excluded it from the minimal interval, supporting our conclusion that this mutation in Scrib is almost certainly the causal gene in the crn2 mutant.
We performed two experiments to determine the extent of PCP perturbation and expressivity of this allele. First, we performed a cross with the looptail (Lp) allele of vang-like 2 (Vangl2), another known PCP gene. A strong interaction between the circletail (Crc) allele of Scrib and looptail (Lp) mice—in which a significant proportion of Crc/+;Lp/+ double heterozygous embryos have a craniorachischisis defect similar to mice homozygous for each mutation alone—has previously been demonstrated (Murdochet al. 2001). We performed the same cross with the crn2 mice and also saw a strong genetic interaction with the Lp allele. We recovered four double heterozygote (crn2/+; Lp/+) embryos. Two of these had the full craniorachischisis defect (Figure 2H) and two had the looped-tail phenotype, a similar finding to the analysis of Crc/+;Lp/+ mice (Murdochet al. 2001).
While further breeding the crn2 and Lp alleles on a highly mixed genetic background (A/J, FVB, 129, and B6), we noted a variable penetrance of the craniorachischisis phenotype. Some adult mice in this subpopulation are homozygous for the crn2 mutation, and one affected embryo appears heterozygous. As several of these mice are derived from a parent in which a recombination event occurred proximal to Scrib, there is a formal possibility that there is a mutation that causes a phenotype that is phenotypically identical to Scrib and interacts similarly with Lp. However, the minimal recombinant interval derived from these mice and those originally studied reveals that it contains only one gene, Efr3a, and direct sequencing did not reveal a coding change. Furthermore, the amino acid mutated in crn2 is conserved as a glutamic acid in all vertebrates examined, as well as in Drosophila. Multiple studies present evidence that penetrance of mouse neural-tube closure phenotypes can vary with genetic background in both mutations of planar cell polarity genes (Wanget al. 2006; Paudyalet al. 2010) and in other genes (Mattesonet al. 2008). Given these findings, we believe that the variable phenotype of the crn2 allele is a consequence of the mixed genetic background.
We also examined the cochlea of the crn2 mutant as PCP is necessary for proper elongation of the developing cochlea and the stereotypical alignment of the stereocilia on the hair cells (Montcouquiolet al. 2003). We found that the cochlea had a normal morphology but was shorter in crn2 mutants (Figure 3, A and B). Examination of the stereocilia on wild-type cochlear hair cells revealed them to be largely perpendicular to the long axis of the cochlea. In mutants, however, the stereocilia are less uniformly oriented (Figure 3, C–I). These results are consistent with the crn2 mutation acting as a null or highly hypomorphic allele.
This is the fourth reported allele of Scrib (Figure 2I). The classic circletail mutation is a premature stop codon resulting in a protein of 971 amino acids and translation of only the first two of four PDZ domains (Murdochet al. 2003). Two previous ENU screens have identified missense mutations in Scrib: one in a leucine-rich region 5′ of Scrib (Zarbaliset al. 2004) and one creating a premature stop eliminating 10% of the amino acid sequence (Wansleebenet al. 2010). The crn2 mutation reported here is in the last amino acid of the first PDZ domain. The fact that all of these mutations cause such similar phenotypes suggests that multiple regions of the SCRIB protein are required for proper SCRIB function.
Hith2is likely a novel allele ofcaspase-3
Hole in the head 2 (hith2) mutants were initially identified in the third arm of our screen focused on postnatal phenotypes. Hith2 mutants have significant encephaloceles, some of which are visible through the skin of the postnatal pups (Figure 4, A–C), a phenotype very reminiscent of the hith mutation in forkhead box c1 (Foxc1) (Zarbaliset al. 2007). We also noted an incompletely penetrant cleft palate phenotype in embryonic dissections. Most homozygous hith2 mutants do not survive past birth, and a substantial proportion of hith2 heterozygous have an abnormal gait consistent with a vestibular defect.
We initially mapped the hith2 mutation to chromosome 8 and identified Casp3 as a candidate gene on the basis of the similarity to published phenotypes (Kuidaet al. 1996; Takahashiet al. 2001; Leonardet al. 2002). Sequencing of the Casp3 locus in the hith2 embryos revealed a mutation in the intron between exons 2 and 3, two nucleotides upstream of the beginning of the third exon (Figure 4D). Subsequent cDNA analysis by RT-PCR showed the mutant transcript to have two isoforms. Sequencing the RT-PCR products revealed the larger of these products to be an imprecise splicing event four nucleotides into the third exon and the smaller species to be a complete skipping of the third exon (Figure 4E). Both alternate transcripts are predicted to result in missense proteins with premature stop codons. Casp3 is an established effector of apoptotic cell death (Nijhawanet al. 2000), and full-length CASPASE-3 is normally cleaved into an active form during apoptotis. Consistent with the alternate transcripts that we identified, immunohistochemical analysis demonstrated that little or no CASPASE-3 is produced in hith2 mutants (Figure 3G). Multiple alleles of Casp3 have been created with similar phenotypes, four of which produce no protein (Kuidaet al. 1996; Wooet al. 1998; Keramariset al. 2000; Morishitaet al. 2001), and a missense mutation at a catalytic residue (Parkeret al. 2010).
Additional mutant characterization
Homozygous rudolph (rud) mutants had skeletal defects, craniofacial defects, and neural defects. Some of the mutants had an accumulation of fluid in a bleb at the end of the snout that sometimes filled with blood (Figure 5A). Histological analysis revealed severe differentiation defects throughout the central nervous system (Figure 5B; data not shown). We have identified a causal mutation on chromosome 1 in the cholesterol biosynthetic enzyme hydroxysteroid (17-beta) dehydrogenase 7 (Hsd17b7) and further analysis of this mutation will be presented separately (R. W. Stottmann, H. Qiu, J. L. Moran, D. R. Beier, unpublished results).
We also recovered the brain dimple (brdp) mutation in this portion of the screen. Brdp mutants were grossly indistinguishable from wild-type littermates upon dissection at E18.5. Microdissection of the brain revealed defects in brain development, including smaller olfactory bulbs and a significant thinning of the caudo-lateral telencephalic tissue (Figure 5D). Mapping of this mutation suggests that it is a novel locus on chromosome 13.
Several of the mutations that we recovered had craniofacial phenotypes along with neural phenotypes. Cleft palate and exencephaly (clpex) mutants have multiple phenotypes at E18.5, including cleft lip, cleft palate, exencephaly, and closed brain cavities with a flattened head. The exencephaly and clefting phenotypes are incompletely penetrant and occur both independently and in the same embryo (Figure 5, F–H). One mutant exhibited spina bifida aperta. Mapping of the mutation identifies a likely novel locus on chromosome 7.
Cleft face (clft3) mutants were initially identified by the failure of the craniofacial tissue to fuse at the midline (Figure 5I). Further dissections revealed mutants with an accumulation of apparently vascularized tissue on top of the skull (Figure 5J). Mutant embryos were often smaller, and we saw many dead and dying embryos at mid-embryonic stages (data not shown). Histological analysis revealed that the brain of clft3 mutants appears smaller than controls (Figure 5, K and L) and that the tissue on top of the cranium is encephaloceles where the neural tissue had grown through a patent surface ectoderm (data not shown). Clft3 is an apparently novel allele of a gene on chromosome 15.
We recovered one line (cleft palate 1) (clft1) with cleft palate but no CNS defects. We have identified this to be a hypomorphic allele of the known clefting gene, interferon regulatory factor 6 (Irf6) (Ingrahamet al. 2006; Richardsonet al. 2006; Stottmannet al. 2010).
The progressive hydrocephaly (prh) mutant was discovered in the third arm of the screen, although the phenotype is not dependent on heterozygosity at the Lis1 locus. Prh mutants are indistinguishable from littermates at birth but are visibly hydrocephalic by P14 (Figure 6, B and F), are less vigorous, and do not survive to weaning ages. The hydrocephaly is not present at birth (Figure 6D), is first visible histologically at P7, and gradually increases in severity through the rest of life (data not shown). Ventricular dilation is visible in the lateral and third ventricles.
We describe our recent efforts to expand the use of ENU mutagenesis in the mouse to uncover novel alleles in a particular tissue of interest, in this case, the forebrain. We used three strategies to increase the sensitivity of the screen for mutations with phenotypes relevant to neurodevelopment: (1) a morphological and histological analysis of the forebrain; (2) inclusion of a reporter allele, the RARE-lacZ transgene; and (3) the use of a sensitizing allele, Lis1.
We find that including various alleles into a traditional three-generation screen does not significantly increase the effort involved to generate and screen the mutant lines. Furthermore, the genetic heterogeneity introduced along with these modifying alleles can be accommodated and still allow rapid mapping of the mutations. However, as one might expect, this genetic heterogeneity can affect phenotype reproducibility; the RARE-lacZ reporter allele turned out to have variable expression in the backgrounds that we used, and no variants from that portion of the screen were proven heritable. The Lis1-sensitized portion of the screen also did not recover any novel interacting genes. However, in both of these cases, we were able to comprehensively analyze only a small number of lines with these modifications. In fact, the sensitized portion of our screen more realistically represents a pilot screen testing the approach of using a sensitizing locus with its own genetic background heterogeneity. Our experience shows that the addition of these components to the screen does not overly complicate the logistics and efficiency of the mutagenesis protocol and, with appropriate selection of alleles, can be an invaluable addition to the experimental design. In the experiment that we describe here, the most valuable approach was to dissect the brain and observe in whole mount. This was clearly fruitful, but is still a relatively blunt approach to observing neurodevelopment. We anticipate continuing to utilize reporter alleles targeting different aspects of neuronal development in future mutagenesis efforts. The continuing addition of transgenic mouse reagents to the neuroscience repertoire [e.g., the GENSAT project (Gonget al. 2003)] provides a growing resource for reporter screens, further enhancing the utility of this approach.
Our attempt to refine the technique and query a specific organ was successful as seven of eight mutants that we recovered had some effect on forebrain development. We comprehensively screened only 38 families, giving a success rate in obtaining a neurodevelopmental mutant of 7/38 (18%). As five of these genes appear to be novel genes in forebrain developmental genetics, this continues to show that ENU mutagenesis is an efficient tool for gene discovery and compares favorably with other recent screening efforts. A previous approach in our own laboratory aimed at general morphological defects recovered 15 monogenic mutants in 54 lines [28% of lines comprehensively analyzed (Herronet al. 2002)]. A previous attempt to query neurodevelopment with a reporter of cortical tangential migration recovered 13 phenotypes from 305 lines (4% of all lines established) with 5 specific to the reporter allele used [1.6% (Zarbaliset al. 2004)]. Another reporter screen assaying axon guidance recovered 7 mutants from 57 lines (12%) with 6 affecting the reporter allele expression (Dwyer et al. 2011). A immunohistochemical approach to studying cranial nerve development yielded seven mutants from 40 pedigrees [18% (Maret al. 2005)].
We report a novel allele of Scrib, which contributes to a greater understanding of this large protein. The scribble protein has leucine-rich repeats at the 5′ end and multiple PDZ protein interaction domains at the 3′ end. The premature stop generated by the initial circletail mutation is predicted to produce only the first two PDZ domains and indicated that the PDZ domains are critical for the function of the Scribble protein (Murdochet al. 2003). Another ENU-induced mutation demonstrated that the leucine-rich regions are also critical (Zarbaliset al. 2004). A third allele, also from an ENU screen, retained ∼90% of the coding sequence, but mRNA levels were reported to be down 65% (Wansleebenet al. 2010). Our crn2 missense mutation is at the extreme 3′ end of the first PDZ domain and may prevent the full-length Scrib protein from folding correctly or interfere with other specific protein interactions. These alleles together support the notion that Scrib encodes a scaffolding protein in the PCP pathway with multiple important protein–protein interaction sites.
We have isolated several mutants of interest for neurodevelopment. All but one of the mutations that we report here were lethal at or before birth, suggesting that they each have pleiotropic effects, possibly leading to many other avenues of future investigation. Furthermore, our phenotype-driven analysis may have uncovered hypomorphic alleles for genes in which a true null allele would have led to early embryonic lethality. Thus, a reverse genetic, “knock-out allele” approach may not have implicated these genes in neurodevelopment. The imminent availability of a large set of conditionally targeted mutant ES lines should facilitate complementary studies of ENU alleles as they are generated (Collinset al. 2007).
We thank A. Tilt, H. Qiu, S. Hines, A. Bolton, and S. Nicholson for assistance in screening and mapping; M. Lun, Y. Yun, and M. Prysak for assistance with animal husbandry; and J. Min for assistance with caspase-3 immunohistochemistry. We thank U. Drager (University of Massachusetts Medical School) for the RARE-lacZ mouse line, Lisa Goodrich (Harvard Medical School) for the looptail mice, and A. Wynshaw-Boris (University of California at San Francisco) for the Lis1 mouse. Funding for this project is from the National Institutes of Health (grants R01HD0306404 and R01MH081187 to D.R.B. and grant F32HD053198 to R.W.S.).
↵1 Present address: Stanley Center for Psychiatric Genetics, Broad Institute of MIT and Harvard, Cambridge, MA 02142.
- Received January 14, 2011.
- Accepted April 13, 2011.
- Copyright © 2011 by the Genetics Society of America