Generation of mice with a conditional Foxp2 null allele

Disruptions of the human FOXP2 gene cause problems with articulation of complex speech sounds, accompanied by impairment in many aspects of language ability. The FOXP2/Foxp2 transcription factor is highly similar in humans and mice, and shows a complex conserved expression pattern, with high levels in neuronal subpopulations of the cortex, striatum, thalamus, and cerebellum. In the present study we generated mice in which loxP sites flank exons 12–14 of Foxp2; these exons encode the DNA-binding motif, a key functional domain. We demonstrate that early global Cre-mediated recombination yields a null allele, as shown by loss of the loxP-flanked exons at the RNA level and an absence of Foxp2 protein. Homozygous null mice display severe motor impairment, cerebellar abnormalities and early postnatal lethality, consistent with other Foxp2 mutants. When crossed to transgenic lines expressing Cre protein in a spatially and/or temporally controlled manner, these conditional mice will provide new insights into the contributions of Foxp2 to distinct neural circuits, and allow dissection of roles during development and in the mature brain. genesis 45:440–446, 2007. Published 2007 Wiley-Liss, Inc.

There are now multiple examples of disruption of the human FOXP2 gene, with affected individuals displaying speech and language deficits of varying severity (Lai et al., 2001;MacDermot et al., 2005;Shriberg et al., 2006;Zeesman et al., 2006). The first example to be discovered, and hence the most well studied, is that of the KE family. A heterozygous missense mutation is inherited by affected members (Lai et al., 2001), who have an impaired ability to learn and produce the sequences of coordinated mouth movements necessary for speech; a condition commonly referred to as developmental verbal dyspraxia . These problems are accompanied by expressive and receptive deficits in oral and written language (Watkins et al., 2002). Functional neuroimaging studies have shown abnormal patterns of brain activation, including underactivation of Broca's area, during language-based tasks (Liegeois et al., 2003). To date, FOXP2 is the only gene clearly linked to this aspect of neurological function, providing a unique opportunity to study the molecular mechanisms involved (Marcus and Fisher, 2003;Vernes et al., 2006).
The FOXP2 protein belongs to a group of transcription factors characterized by the presence of a forkheadbox (FOX) DNA-binding domain. FOX proteins regulate a diverse variety of processes from early embryogenesis through to adulthood, and have been implicated in disorders of human or mouse development (Carlsson and Mahlapuu, 2002). Several domains have been identified in FOXP2 in addition to the characteristic DNA-binding motif, including polyglutamine tracts, a zinc finger, a leucine zipper, and an acidic C-terminal region (Wang et al., 2003). The FOXP2 amino acid sequence is highly similar across a number of distantly-related vertebrate species; the human and mouse proteins are distinguished by only three amino acid substitutions and a single-residue difference in polyglutamine-tract length (Enard et al., 2002). Moreover, orthologues of FOXP2 show conserved expression in equivalent brain structures in humans, rodents, birds, reptiles, and fish, with notable similarities in sublocalisation (Fisher and Marcus, 2006;Vargha-Khadem et al., 2005). Key expression sites lie within the cortex (pallium in nonmammals), striatum, thalamus, and cerebellum. In the mammalian cortex the gene is mainly expressed in the deepest layers, in the rat striatum it is enriched in striosomes, while hindbrain expression is confined to inferior olives, Purkinje cells, and deep cerebellar nuclei in all species studied thus far (Bonkowsky and Chien, 2005;Ferland et al., 2003;Haesler et al., 2004;Lai et al., 2003;Takahashi et al., 2003;Teramitsu et al., 2004).
These data point to widely conserved functions of FOXP2 orthologues in distributed vertebrate circuits involved in sensory processing, sensorimotor integration, and control of skilled coordinated movements. Analyses of primate sequence variation suggest that the precise roles of FOXP2 may have undergone modifications during human history, perhaps in relation to speech (Enard et al., 2002). Nevertheless, animal models will yield crucial insights into the contributions of the gene to the development and function of relevant neural circuits, and how they may go awry in cases of disorder. Gene disruption in the mouse is a highly amenable tool for addressing this question. Given the complexities of Foxp2 expression during embryogenesis, postnatal development and adulthood, in both neural (Ferland et al., 2003;Lai et al., 2003) and non-neural tissues (Shu et al., 2001), a strategy that allows spatiotemporal control of gene disruption is particularly valuable. Thus, in the present study, we have generated a mouse line in which critical exons of Foxp2 are flanked by loxP sites (floxed), and demonstrated successful Cre-mediated disruption of the gene.
It is unfeasible to target the entire Foxp2 locus, since it spans several hundred kilobases. Instead, we chose to insert loxP sites around exons 12-14, which encode the DNA-binding motif (Fig. 1). In addition to removing an essential functional domain, loss of these exons is predicted to yield premature termination of protein translation early in exon 15, because of a frameshift. Of note, Shu et al. previously targeted a similar region of Foxp2 in FIG. 1. Schematic representation of the Foxp2 conditional targeting strategy. Recombination of the targeting vector with the Foxp2 genomic locus results in the introduction of a FLPe-Neo cassette, flanked by FRT sites (ovals), 5 0 of exon 12, and leaves two loxP sites (arrows) surrounding the cassette and exons 12-14. The thymidine kinase gene at the 3 0 end of the targeting vector enables selection against clones containing randomly integrated vector. FLPe-mediated recombination enables removal of the selection cassette, and generates the floxDneo allele. The D12-14 allele can be generated from either the floxneo or floxDneo allele by Cre-mediated recombination. PCR genotyping primers and relevant restriction enzyme sites and probes used for Southern analysis are indicated.
standard knockout experiments, and reported that replacement of exons 12 and 13 by a neomycin resistance cassette produced a null allele (Shu et al., 2005). For the present study, Bruce-4 ES cells (Kontgen et al., 1993), of C57BL/6 origin, were transfected with linearized conditional targeting vector. Clones surviving selection were screened by Southern blotting for appropriate integration of the 5 0 and 3 0 loxP sites, and the neomycin cassette (Fig. 2a). Correctly targeted clone 1 cells were injected into C57BL/6 albino blastocysts to obtain a male chimera. Breeding to C57BL/6 albino females yielded germline transmission, and Foxp2 floxneo/þ heterozygotes were identified by coat color and PCR genotyping (Fig. 2b).
We removed the neomycin resistance gene, since it is well established that the presence of this selection cassette can influence expression of the floxed gene and/or neighboring loci (Lewandoski, 2001). The targeting construct incorporates a FLPe gene, driven by the testes-specific ACE promoter, which theoretically drives ''self-excision'' of the FRT-flanked selection cassette in the testes. However, we found no evidence that the cassette had been removed in chimeric offspring, probably because of insufficient levels of FLPe protein expression. We therefore mated Foxp2 floxneo/þ heterozygotes to ACTB-FLPe hemizygotes [ Fig. 1(ii)], which express FLPe ubiquitously under control of human b-actin regulatory sequences (Rodriguez et al., 2000) and offspring were genotyped by PCR (Fig 2b). Resulting Foxp2 floxDneo/þ heterozygotes were then bred to C57BL/6 wildtype animals to remove the ACTB-FLPe transgene from the floxed strain. Mice carrying the floxDneo allele will be used for future crosses to transgenic strains expressing Cre recombinase in a region-and/or temporal-specific manner [ Fig. 1 To verify functionality of the loxP sites in vivo, and to determine if Cre-mediated excision of exons 12-14 produces a null allele, we crossed Foxp2 floxneo/þ males to Foxp2 D12-14/þ ;Sox2-Cre þ/À females [ Fig. 1(iv)]. Sox2-Cre mice provide an efficient means for deleting loxP-flanked sequences, particularly when the Cre transgene is carried on the maternal line. In this case excision occurs throughout the early embryo, in all offspring, irrespective of whether they receive the transgene (Hayashi et al., 2003;Vincent and Robertson, 2003). As expected, these crosses produced general deletion of exons 12-14 in offspring and successfully yielded Foxp2 D12-14/þ and Foxp2 D12-14/D12-14 pups, demonstrated by PCR-based genotyping (Fig. 2b). Quantitative real-time RT-PCR was used to analyse Foxp2 expression in the striatal precursor region of E16.5 embryos from such crosses. Three sets of Foxp2 primers were employed from different regions of the transcript. Foxp2 D12-14/D12-14 homozygotes lacked the FOX domain-encoding exons -primer pair 13/14 yielded no product in these embryos, and a half-dosage in Foxp2 D12-14/þ heterozygotes (Fig. 3a). RT-PCR assays for exons 6-7 and 16 showed reduced expression levels in Foxp2 D12-14/D12-14 embryos as compared to wildtype littermates, with intermediate levels in  pups to right themselves after being placed on their backs (error bars represent SE of the mean); note that WT and heterozygote pups are able to right themselves with virtually no delay. (f) Cerebellar morphology at P22 in WT (i, ii), heterozygotes (iii, iv), and homozygotes (v, vi). Homozygotes show reduced cerebellar size and foliation, as shown by whole brains (i, iii, v) and Nisslstained sagittal sections through the vermis (ii, iv, vi). All whole brain photographs were taken at the same magnification. ther a significant proportion of transcripts from the conditional allele are unstable/degraded e.g. by nonsense mediated mRNA decay, or genomic regulatory sequences between exons 12-14 influence expression of Foxp2. Of note, mutant mice that carry an early stop codon in exon 7 of Foxp2 show comparable reductions in Foxp2 mRNA expression levels (Groszer et al., in preparation). No changes in relative RNA expression were observed for Foxp1, Foxp4, or other striatally expressed genes, Gad2 (Katarova et al., 2000) and Dlx2 (Porteus et al., 1991) at this developmental time (Fig. 3a).
Extracts from the striatal precursor region of E16.5 embroys were analysed by Western blotting using an Nterminal Foxp2 antibody, to determine the impact of Cre-mediated deletion at the protein level. A single band corresponding to full-length Foxp2 was observed in wildtype extracts and in Foxp2 D12-14/þ extracts at a reduced intensity (Fig. 3b). No bands could be detected in Foxp2 D12-14/D12-14 extracts; there was no evidence of any truncated protein resulting from loss of exons 12-14 and termination of translation in exon 15 (Fig. 3b).
Homozygous Foxp2 D12-14/D12-14 pups have a reduced body weight when compared with littermates, and typically die around postnatal day 21 (Fig. 3c). In addition, they display developmental delays (Fig. 3d) and severe motor dysfunction, including an impaired righting reflex (Fig. 3e). Although most of the brain appears grossly normal, there is a substantial reduction in the size and foliation of the cerebellum (Fig. 3f). By contrast, heterozygous Foxp2 D12-14/þ pups display no overt abnormalities-they gain weight at the same rate as their wildtype littermates, do not show significant righting deficits, and have normal cerebellar size and foliation (Fig. 3c-f). Phenotypes of these homozygous and heterozygous animals are highly consistent with those observed in mutant mice carrying an early stop codon in exon 7 of Foxp2 (Groszer et al., in preparation). The gross phenotype of Foxp2 D12-14/D12-14 homozygotes also recapitulates that obtained by Shu et al. for their standard targeted knockout of exons 12-13 (Shu et al., 2005). However, heterozygotes in the Shu et al. study displayed modest developmental delay and reduced rate of weight gain (Shu et al., 2005), unlike our Foxp2 D12-14/þ heterozygotes, which show no such deficits.
In conclusion, we have generated a conditional Foxp2 allele and shown that homozygous general deletion of exons 12-14 results in absence of Foxp2 protein, accompanied by developmental delays, severe motor dysfunction, and neural abnormalities. Crossing our floxed line with transgenic mice which express Cre-recombinase in a spatially and/or temporally controlled manner should circumvent problems associated with early postnatal lethality and allow investigations of adult animals. Such studies will facilitate investigations of Foxp2 function in the various tissues where it is expressed, including brain, lung, intestine, and cardiovascular system. Most importantly, our conditional allele represents a powerful tool for dissecting the differential contributions of Foxp2 to development and function of distinct neural networks in the mammalian central nervous system.

Gene Targeting and Generation of Mutant Mice
Exon numbering for murine Foxp2 is concordant with that found for orthologous exons in human FOXP2 (Mac-Dermot et al., 2005), C57BL/6 DNA containing exons 11-16 of Foxp2 was cloned into the vector pEASY-FLIRT (Casola, 2004). The resulting construct contained a loxP-FRT-FLPe-Neo R -FRT cassette between exons 11 and 12, with a second loxP site inserted between exons 14 and 15. Bruce-4 ES cells were cultured on a mitotically inactive primary MEF feeder layer as described previously (Torres and Kuhn, 1997). 1.2 3 10 7 ES cells were transfected by electroporation with 25 lg of ClaI-linearized targeting construct. Colonies surviving G418 (163 lg/ mL active ingredient) and gancyclovir (2 lM) selection were screened for targeted recombination by Southern blot analysis. Morula-stage embryos were harvested 2.5 days post coitum from superovulated C57BL/6 albino females (C57BL/6J-Tyr c-2J , Jackson Laboratories) and cultured overnight in M16 media. The following day, healthy blastocysts were injected with ES cells and transferred to pseudo-pregnant CD-1 females. Resulting chimeras were bred to C57BL/6 albino mice, to enable germline transmission to be determined by coat color. All regulated procedures were carried out under UK Home Office Project Licence 30/2016.

Western Blotting
Tissue from the striatal precursor region of E16.5 embryos was dissected into RIPA lysis buffer and disrupted by sonication. Lysates were centrifuged at 10,000g for 10 min, and the protein concentrations of the supernatants were determined by Bradford assay. Proteins (30 lg per lane) were separated on 12% SDSpolyacrylamide gels and transferred to PVDF membranes. Membranes were blocked in 5% milk, before overnight incubation with the primary antibody at 48C. Goat anti-Foxp2 (N-16) polyclonal antibody (Santa Cruz Biotechnology), and the loading control, mouse antiactin monoclonal antibody (Sigma), were both used at a 1:2,000 dilution. Rabbit anti-goat (Dako) and goat anti-mouse (BioRad) HRP-conjugated secondary antibodies were applied at a 1:5,000 dilution for 1 h at room temperature. Proteins were visualized by chemiluminescence detection using LumiGLO (Cell Signaling Technology).

Quantitative Real Time RT-PCR
E16.5 striatal tissue, from three brains of each genotype, was dissected into RNAlater (Ambion), before being snap-frozen and stored at À808C. When required, samples were thawed and transferred to buffer RLT (QIAGEN) with b-mercaptoethanol, and then disrupted using a piston homogenizer. RNA was extracted using an RNeasy kit (QIAGEN), and included an on-column DNase digestion step. cDNA was synthesized from 2 lg RNA using Superscript III reverse transcriptase (Invitrogen), according to the manufacturer's protocol. PCR amplification was carried out using 25 lL reaction volumes with 12.5 lL of SYBR Green Supermix (BIO-RAD), 0.5 lL of each primer (10 lM) and 1 lL of cDNA. Thermal cycling was performed on the iCycler iQ system (BIO-RAD) with amplification for 50 cycles at 958C for 15 s, 608C for 30 s, and 728C for 30 s. Melting curve analysis was performed to exclude amplification of nonspecific products. Relative changes in expression were calculated using the 2 ÀDDC T method (Livak and Schmittgen, 2001), using GAPDH as the internal control and the average of the wild type samples for each primer pair as the calibrator.

Histological Analysis
Mice were deeply anaesthetized and transcardially perfused with 4% paraformaldehyde (PFA) in phosphate buffer. Brains were removed and postfixed for a further 24 h, before being embedded in paraffin. Serial sections were cut sagittally at 5 lM and stained with cresyl violet. Whole brains were snap-frozen in liquid nitrogen before being postfixed for 3 h in 4% PFA. Brains were photographed with a drop of bromophenol blue for enhanced contrast.