Ablation of BRaf Impairs Neuronal Differentiation in the Postnatal Hippocampus and Cerebellum

This study focuses on the role of the kinase BRaf in postnatal brain development. Mice expressing truncated, non-functional BRaf in neural stem cell-derived brain tissue demonstrate alterations in the cerebellum, with decreased sizes and fuzzy borders of the glomeruli in the granule cell layer. In addition we observed reduced numbers and misplaced ectopic Purkinje cells that showed an altered structure of their dendritic arborizations in the hippocampus, while the overall cornus ammonis architecture appeared to be unchanged. In male mice lacking BRaf in the hippocampus the size of the granule cell layer was normal at postnatal day 12 (P12) but diminished at P21, as compared to control littermates. This defect was caused by a reduced ability of dentate gyrus progenitor cells to differentiate into NeuN positive granule cell neurons. In vitro cell culture of P0/P1 hippocampal cells revealed that BRaf deficient cells were impaired in their ability to form microtubule-associated protein 2 positive neurons. Together with the alterations in behaviour, such as autoaggression and loss of balance fitness, these observations indicate that in the absence of BRaf all neuronal cellular structures develop, but neuronal circuits in the cerebellum and hippocampus are partially disturbed besides impaired neuronal generation in both structures.


Introduction
Binding of growth factors to their cognate receptors leads to the activation of the RAS-RAF-MEK-ERK mitogen activated protein kinase (MAPK) cascade and is involved in regulation of many aspects of cellular growth and differentiation [1,2]. The pathway consists of the small guanine nucleotide binding protein RAS and the protein kinases RAF, MEK, and ERK [1]. The activation of members of the RAF serine/threonine protein kinase family is initiated by RAS-GTP association with the RAS binding domain of RAF located at the N-terminus [3]. Most of the RAF functions appear to be mediated by phosphorylation/activation of the mitogen-activated protein kinase (MAPK)-extracellular signalregulated kinases 1 and 2 (MEK1 and MEK2). ERK1 and ERK2 (extracellular-signal regulated protein kinase) phosphorylate multiple downstream substrates [1]. The duration and intensity of their activity is thought to control the response to growth factor signals [4]. Two distinct types of mutations have been identified in human diseases in various genes encoding components of this cascade. Miscoding oncogenic somatic mutations that cause tumorigenesis often confer enhanced and growth-factor independent, constitutive activity of the mutant protein. Examples are frequent somatic mutations of KRAS codons 12 and 13 [5] in various types of cancers in endodermal organs (pancreas, colon, lung, etc.), and the prevalent mutation in the kinase domain of BRAF, BRAFV600E [6,7] in cancer cells of endodermal (thyroid, colon) and ectodermal (melanoma) tissues. Germ line mutations in RAS or BRAF introduce distinct amino acid changes from those found in somatic cancer cells, and they can cause a spectrum of developmental defects such as cardiofacio-cutaneous (CFC) syndrome and Noonan or Costello syndrome [8], but do not appear to be overtly oncogenic. Features of CFC include congenital heart defects, a characteristic facial appearance, gastrointestinal dysmotility, moderate-to-severe intellectual disability, and short stature [9,10]. BRAF mutations have been documented in ,75% of affected individuals while ,25% have a mutation in MEK1 or MEK2 [10]. The BRAF mutations found in CFC can confer either weakly elevated kinase activity (mutations of Q257R, S467A, L485F, K499E) or impaired kinase activity versus wild-type BRAF (E501G and G596V) [11,12].
The three functional mammalian RAF proteins (ARAF, BRAF and CRAF, the latter is also termed c-Raf-1) display redundant as well as specific functions. RAF enzymes form homo-and heterodimers [13][14][15]. It has been demonstrated that oncogenic mutant BRAF with impaired kinase activity are still able to transactivate CRAF because the mutation induced an active conformation of the enzyme under basal growth condition [14]. One important difference among RAF kinases is found in the regulation of phosphorylation sites. The serine 445 of BRAF is constitutively phosphorylated while the homologous site in CRAF must be phosphorylated for maximal activation [16]. This difference is thought to be the basis of the observation that a single point mutation at codon 600 results in constitutive BRAF activation, whereas the homologous mutation in CRAF does not [17]. LTP in the dentate gyrus of adult rats induces BRaf but not CRaf expression [18]. Lung tumour formation in mice by oncogenic KRAS requires CRaf, but not BRaf [19,20]. Distinct biological features of Raf kinases are further suggested by the lack of compensation between Raf proteins in mice with genetic ablation of a single Raf gene and the different phenotypes of Rafdeficient mice ( [21]; and see below).
Based on the neurological phenotypes reported for CFC, the generation, differentiation and functional maintenance of neurons is likely affected by mutated BRAF. However, little is known about the effects of these mutant proteins on cell cycle regulation of neural stem/precursor cells and neuronal differentiation. Mice with constitutive elimination of BRaf are embryonic lethal at midgestation [22]. In addition embryonic sensory and motor neurons of BRaf-deficient mice are not able to survive in the presence of neurotrophic survival factors. This effect is BRafspecific and cannot be rescued by CRaf [23]. Global expression of constitutively active BRAFV600E in mouse tissues causes early embryonic lethality [24]. Conditional elimination of BRaf using Nestin-Cre demonstrated reduced expression of glial cell linederived neurotrophic factor receptor RET in sensory dorsal root ganglion neurons at postnatal age and did not reveal an essential role of BRaf to relay the survival signal in neurons [25]. Furthermore, the essential role of BRaf for oligodendrocyte maturation and myelination during postnatal central nervous system development was demonstrated [26]. In addition, BRaf is involved in synaptic plasticity, because Cre-mediated conditional elimination of BRaf in neurons expressing calcium/calmodulindependent protein kinase II alpha showed impaired hippocampal learning and memory [27].
Many neurons of the adult brain are generated prenatally but in the cerebellum [28], olfactory bulb [29] and hippocampus [30,31] neurons are also born in postnatal life. Neurogenesis persists even in adult mammals, including humans, in the subventricular zone (SVZ) and the subgranular zone (SGZ) of the dentate gyrus in the hippocampus [28,[32][33][34]. In the hippocampus, the majority of new cells migrate from the SGZ into the granule cell layer, differentiate into dentate granule neurons, and integrate into the existing neural circuitry [34]. Thus, neurogenesis is a cellular form of plasticity that accompanies the classical synaptic forms of plasticity. The question whether altered BRaf kinase levels or activity affect postnatal neuronal functions in the dentate gyrus and cerebellum has not been studied so far. Here, we analyse the effects of BRaf ablation in the postnatal mouse dentate gyrus and cerebellum. We observe that postnatal hippocampal dentate gyrus precursor cells show increased rates of proliferation besides impaired dendritic differentiation in both structures.

Ablation of BRaf Impairs the Postnatal Growth of the Dentate Gyrus
To facilitate an analysis of the role of BRaf kinase in brain development, postnatal neural stem cell proliferation and neuronal differentiation, we created a conditional allele by flanking exon 3 of the BRaf gene with two loxP sites. We targeted exon 3 because it encodes part of the Ras binding domain, which is essential for the activation of the kinase and predicted that an exon 3 deleted BRaf allele would be a null allele (Figure 1 A; Figure S1A). Raf kinases shuttle between closed inactive and open active conformation. The N-terminal regulatory domain maintains an inactive conformation of the kinase such that the C-terminal catalytic domain is inhibited [3,35,36] ( Figure 1G). The RAS-binding domains (RBD), encoded by exons 3-5 and encompassing 81 amino acid residues [37] is located within the N-terminal regulatory domain ( Figure 1G). Autoinhibition mediated by the regulatory domain is relieved by binding of Raf`s RBD to activated Ras. Since residues of the bstrand B2 of the RBD which are instrumental for binding to Ras/ Rap are encoded in exon 3, BRaf with an exon 3 deletion would be predicted to be deficient in binding to GTP-bound Ras and shuttling Raf into the active conformation [38,39]. All other domains in BRaf are predicted to remain intact after in-frame deletion of exon 3 ( Figure 1G). We first investigated the effect of deletion of exon 3 in mice obtained from two independent targeted floxed BRaf ES cell clones (nfl152 and nfl156, Figure S1B, C) that were both successfully transmitted into the germ line. To obtain a line with exon 3 deletion, we crossed heterozygous female BRaf wt/nfl156 mice with male EIIaCre deleter mice [40]. PCR analysis of tail DNA revealed that compound offspring harbouring both the EIIaCre transgene and the floxed BRaf allele contained cells with partial or complete deletions of the loxP flanked regions in a mosaic fashion (not shown), in line with the known function of the EIIaCre strain as a general deleter. Back-crossing of these mosaic mice to wild-type mice showed that individual offspring harboured either a complete deletion of the loxP flanked locus (BRaf wt/del ) or a deletion removing only the neomycin resistance gene (BRaf wt/fl ; Figure 1A). The latter allele excluded any potential detrimental effect of the neo cassette in BRaf wt/nfl mice on BRaf expression. Intercrosses of BRaf wt/nfl or BRaf wt/fl mice yielded homozygous BRaf nfl/nfl and BRaf fl/fl mice at a frequency expected for a functional allele (nfl152:129 animals born, 31 wt, 69 wt/nfl, 29 nfl,nfl; nfl156:72 animals born, 18 wt, 37 wt/nfl, 17 nfl,nfl; fl156:69 animals born, 18 wt, 31 wt/fl, 20 fl,fl;). Homozygous BRaf nfl/nfl and BRaf fl/fl mice had a normal life span, were fertile and showed no macroscopic pathological phenotype in the brain. Heterozygous BRaf wt/del mice were healthy, fertile and displayed no body weight difference compared to BRaf wt/wt mice ( Figure  S1F), arguing against a dominant activity of the exon 3-deleted BRaf del allele. In contrast, intercrosses of heterozygous BRaf wt/del mice yielded no live born del/del offspring ( Figure S1D) indicating that the deletion of exon 3 from BRaf had generated a nonfunctional BRaf expressing mouse ( Figure 1G). At E10.5 already, the number of del/del embryos was lower than expected, but some of the BRaf del/del embryos were still alive ( Figure S1D), as observed previously upon constitutive ablation of BRaf [22]. In two litters isolated at E14.5, only one dead del/del embryo could be recovered ( Figure S1D, E, G). We noted that del/del embryos as well as their placentas were smaller and retarded in their development at E10.5, compared to wt/del or wt/wt littermates ( Figure S1E, and not shown). Western blot analysis from E10.5 embryo extracts with antibodies against either the N-or Cterminal domains of BRaf showed that the wild type mouse BRaf locus specifies the expression of two slightly different-sized isoforms of approximately 92 and 89 kDa in the embryo ( Figure 1B). In the BRaf wt/del embryo a new band was also identified, at an approximate molecular weight of 82 kDa in the immunoblot with the antibody against the C-terminus ( Figure 1B). This ,82 kDa band persisted in the del/del extracts when probed with the antibody against the C-terminus ( Figure 1B). The ,92 kDa and ,89 kDa bands were absent in del/del embryo extracts with either of the two antibodies ( Figure 1B). These protein isoforms could be due to different phosphorylation states of BRaf or to translation of alternatively spliced exons. In order to discriminate between these possibilities, we performed RT PCR on total mRNA from embryos. Using primers located in exons 1 and 4 for cDNA amplification, we observed two PCR fragments of around 420 and 320 bp in wild type embryos ( Figure S2). DNA sequencing of gel-purified PCR products revealed ( Figure S2) that the smaller 320 bp band corresponded to the known cDNA sequence of BRaf (Genebank accession NM_139294) whereas the larger 420 bp fragment had a sequence present within exon 3 that was identical to the genomic sequence (Genbank accession NC_000072). The fragment present in the wt/del embryos was a fusion of exon 2 to exon 4 ( Figure S2, Figure 1G). Since the deletion of exon 3 would not change the open reading frame, the translation of a novel BRaf protein from the BRaf del allele is plausible.
In order to evaluate a specific role of BRaf in neural development, we crossed compound Nestin-Cre BRaf wt/fl and Nestin-Cre BRaf wt/nfl (ctrl) mice with BRaf fl/fl and BRaf nfl/nfl mice. Nestin-Cre BRaf fl/fl and Nestin-Cre BRaf nfl/nfl (cKO) mice were born alive at the same frequency as their ctrl littermates and all animals remained alive up to postnatal day 16 ( Figure 2A). We observed no difference among the three lines of mice in terms of survival or body weights. At postnatal day 10, the body weight of cKO animals was 4.360.11 g, 4.060.15 g and 4.460.3 g for nfl152, nfl156 and fl animals, respectively (n = 10). Body weight in all three lines was dramatically decreased compared to control littermates (5.960.14 g, 6.160.34 g and 5.860.27 g, n = 14, p#0.003). Therefore we have used both BRaf fl and BRaf nfl animals in our experiments. CKO mice were smaller at P5 compared to ctrl litter mates and were unable to gain weight after P10 ( Figure S3A). CKO mice began to die at around P17 and none survived beyond P28 (Figure 2A). In order to analyse the level of BRaf elimination in hippocampi of P18/P19 mice, we performed western blot analysis of microdissected hippocampal tissue with the antibody against the N-terminal of BRaf using conditions of high signal sensitivity (see Materials and Methods) and observed that there was only a minor residual level of BRaf in cKO detectable ( Figure 1C) indicating efficient gene ablation. The loss of BRaf in cKO mice could also be observed in other microdissected brain regions at P21 ( Figure S3D) and in hippocampus at P6, P12 and P22 ( Figure S3E). The low residual level of BRaf in cKO ( Figure 1C) is likely due to a low number of BRaf-positive cells where Cre was unable to delete the BRaf gene as observed by immunohistochemistry ( Figure 1D, Figure S4). The ,82 kDa band was observed in ctrl and cKO mice with the antibody against the C-terminal of BRaf (not shown). In order to demonstrate that the ,82 kDa protein is a non-functional kinase, we analysed two downstream targets of BRaf by immunblotting. We observed in brain lysates that two downstream targets of BRaf signalling, the phosphorylation levels of the kinases ERK1,2, as well as the levels of the early growth response 1 transcription factor Egr1 were significantly reduced in the hippocampus of cKO mice compared to ctrl mice ( Figure 1C). This provides additional biochemical evidence that the smaller ,82 kDa BRaf protein in our cKO mice is a non-functional kinase. In summary, we show that both wild-type isoforms of BRaf (,92 kDa and ,89 kDa) are eliminated by our Cre/loxP strategy in del/del embryos. The smaller ,82 kDa protein expressed from the del allele represents a non-functional protein that could not act as a BRaf kinase ( Figure 1G). We conclude that our targeting strategy has created a conditional BRaf mouse with loxP-flanked exon 3. The deletion of exon 3 generates a non-functional allele of BRaf that cannot support embryonic development and survival beyond midgestation. This phenotype is similar to that noted before in mice with constitutive ablation of BRaf [22] and two previously reported conditional alleles of BRaf where either exon 3 or exon 12 [25,27] had been deleted.
To further analyse the BRaf gene deletion in the brain, we performed immunostaining of P21 brain sections. Almost all cells in the cellular hippocampal regions as well as their dendritic processes and cerebellar Purkinje cells failed to stain positive for BRaf demonstrating an efficient non-functional BRaf expressing mouse. Few BRaf-positive granule cells in the hippocampus and a low amount of BRaf-positive Purkinje cells in the cerebellum presumably represent cells that escaped the deletion by Cre recombinase ( Figure 1D, Figure S4). To test the efficiency of Cre recombinase we performed Cre and BRaf immuno-histological staining on tissue slices. We did not expect a high amount of Cre in Purkinje cells of the cerebellum at P21, because these cells are Nestin-negative at this time point. We could not find any BRaf positive cell expressing Cre at the same time (data not shown). We therefore conclude that BRaf positive cells might result from Creescaping cells that have already been described earlier [41,42].
The gross appearance of the brain and of hippocampal sections of cKO mice at around P20 did not reveal any obvious abnormality ( Figure 1D, E; Figure S3F), but the weights of the brains were significantly reduced at P10 and P20 ( Figure S3B). In the hippocampus, migration defects can lead to cell dispersal [43]. As BRaf has been implicated to play a role in neuronal migration in the embryonic cortex [44], we inspected hippocampal sections of cKO mice at P21 but did not observe cell dispersal in the dentate gyrus ( Figure 1D, E). In order to exclude sex-related differences, quantification of the granule cell layer volume and all other subsequent experiments were performed exclusively with male littermates. A quantitative analysis of the volume of the hippocampal granule cell layer revealed that cKO mice had a reduced volume of this neuronal cell layer at P21 but not at P12 ( Figure 1F). This finding indicates that BRaf has a specific role in the postnatal generation or differentiation of dentate granule neurons that are born and differentiating after P12.

Pathophenotypes in the Cerebellum of BRaf Ablated Mice
Whereas the gross appearance of the brain of cKO mice was normal ( Figure S3F), their behaviour showed signs of abnormality. This included autoaggression, as evident from biting of their toes, 13 out of 15 mice showed this phenotype ( Figure 2B). CKO mice could walk normally ( Figure S3G). However, they showed deficiencies in their ability to walk and balance on a rod ( Figure 2C). In line with this observation, several cytoarchitectonic alterations were observed in sagittal sections of the hypoplastic cerebellum of P21 cKO mice in which BRaf expression was efficiently eliminated as shown by Western blotting ( Figure S3D) and immunostaining ( Figure S4). The abnormalities affected all cerebellar lobes at P21 ( Figure 3A, A'). Preliminary birthdating experiments suggest that altered proliferation of cerebellar progenitor cells in the second and third postnatal week may be the cause for the hypoplasticity in cKO mice. Lobuli LI/II, LIII, LVII, LIX and LX showed reduced lengths with an impairment of 30% in LI/II, 21% in LIII, 18% in LV and 23% in LX together with significant alterations in LV (n = 3, p-value 0.049) ( Figure 3D). In the internal granule cell layer of lobuli LI/II, LVII and LX the boundaries of single glomeruli were less well demarcated and tended to become indistinct with fuzzy borders ( Figure 3B, B'). Glomeruli were reduced in their size indicating that the synapses of the granule cells with mossy fiber and Golgi cells had not Notably, in cKO mice the primary dendrite of the Purkinje cells appeared elongated and the arborization in the molecular layer was reduced and irregular ( Figure 3C, C'). Quantitative analysis revealed a more than 26 lengthening of primary dendrite length in cKO (LI, LIX and LX) compared to ctrl (not shown). These findings could additionally be observed in vivo using MAP2 staining of P21 mouse hippocampal slices ( Figure S6).

Ablation of BRaf in Neural Precursor Cells Expands the Pool of Proliferating Hippocampal Progenitor Cells in the Third Postnatal Week
To examine whether increased apoptosis or reduced proliferation of stem/progenitor cells could explain the reduced hippocampal granule cell volume in the absence of BRaf ( Figure 1F), we first stained hippocampal sections of P24 mice for the presence of activated caspase 3. On average, one or two apoptotic cells were usually visible in a section of the dentate gyrus, and the apoptotic cells were mostly located close to the subgranular region ( Figure 4A). Quantification revealed no significant increase of apoptotic cells detectable in the dentate gyrus of cKO mice, as compared to control littermates ( Figure 4A). To determine whether reduced proliferation of stem/progenitor cells in the granular cell layer could be the cause of the reduced cell volume, we applied a single pulse of 5-bromo-2-deoxyuridine (BrdU) to male P20 mice and sacrificed the animals 2 h later. BrdU can be incorporated into DNA of dividing cells only during the S-phase of the cell cycle and is a useful tool for monitoring cell proliferation and birth dating [45]. Unexpectedly, BrdU immunostaining and quantification of BrdU-positive cells in the dentate gyrus of cKO mice showed an approximately 50% higher number of BrdU-positive cells as compared to ctrl mice ( Figure 4B, E). As an independent measure of proliferation of neural stem cells, we stained hippocampal sections for Ki-67 protein. Since Ki-67 is expressed in all phases of the cell cycle except the resting phase [46], the fraction of Ki-67 positive cells represents the number of dividing progenitor cells. The dentate gyrus of cKO mice harboured an increased number of Ki-67 positive cells ( Figure 4C, E). To address the question whether the elimination of BRaf in cKO mice altered the fraction of neural progenitors in the subgranular zone of the dentate gyrus that exited from the cell cycle, we applied a two-hour pulse of BrdU before tissue fixation and performed a double staining for BrdU and Ki-67. The fraction of BrdU positive cells that were negative for Ki-67 was increased in cKO mice as compared to ctrl ( Figure 4D, E; BrdUpositive, Ki-67-negative cells marked with an arrowhead). We conclude that the elimination of BRaf mediated an increased and aberrant cell cycle exit involving loss of Ki-67 expression during the cell cycle.
signalling. The phosphorylation levels of the kinases ERK1,2, as well as the levels of the early growth response 1 transcription factor Egr1 were significantly reduced in the hippocampus of cKO mice compared to ctrl mice whereas the expression of Erk1,2 was unaltered. The residual level of BRaf in cKO may occur from ''escaper'' cells. Gapdh served as loading control. (D) Analysis of BRaf expression by immunohistochemistry. Upper panels are representative sagittal sections of P21 hippocampus immunostained for BRaf with an antibody against the BRaf N-terminus. Lower panels are images taken from boxed regions in upper panels; note presence of BRaf stain in cell body of singular granule neurons (arrows) and their dendrite extending into the molecular layer that might have ''escaped'' Cre recombinase-mediated BRaf deletion in Nestin-Cre/BRaf fl,fl mice. Scale bars; upper row, 200 mm; lower row, 25 mm. (E) Representative sagittal sections of P12 and P21 hippocampus stained with Nissl. Scale bars; 800 mm. (F) Volume of hippocampal granule cell layer (gcl) in 12 and 21 day old mice. Data are mean 6s.e.m.; P12, n = 3; P21, n = 7. (G) Exon organization and location of regulatory regions in BRaf isoforms. Boxes indicate exons with their sizes in nucleotides aligned to the regulatory, catalytic and RAS-binding domains (RBD) of BRaf protein. The vertical arrows above exon 3 indicate the positions of the 59 end and 39 end, respectively of an intron that has been spliced out in the small cDNA harbouring exon 3* in embryonic RNA (see Figure S2). This in-frame splicing retains the reading frame and is predicted to encode the 89 kDa BRaf isoform. The scheme is deduced from cDNA sequencing of wild-type and exon 2-4 spliced BRaf del samples (see text). The molecular masses of BRaf proteins present on the gel (Fig. 1B)

Ablation of BRaf in Neural Precursor Cells Impairs Neuronal Differentiation
We were puzzled by the observation that in three week old cKO mice the volume of the dentate granule cell layer was decreased ( Figure 1E), yet we observed an increased number of proliferating cells in the dentate gyrus, but no change in apoptosis (Figure 4). One potential explanation could be that the cells divide but do not differentiate into neurons which occupy a larger volume than progenitor cells with their dendritic and axonal processes. To address this point, we used birthdating experiments [45]. A cohort of mice were treated with two pulses of BrdU at days P10 and P11 and then sacrificed at P22. Using a 12 days chase period after two consecutive BrdU applications at P10 and P11, would enable us to analyse whether BrdU-labelled progenitor cells retained their  radial glial GFAP molecular marker, whether they had differentiated into new NeuN-positive neurons or whether they were lost during this time interval. Notably, the number of BrdU positive cells at P12 (at the beginning of the chase period after two BrdU applications at P10 and P11) was not different in cKO mice as compared to controls ( Figure 5A). This finding is in contrast to the data obtained at P20 with a two-hour BrdU pulse. The number of cells remaining BrdU positive cells at P22 after the long chase period of 12 days was reduced compared to the amount of BrdUpositive cells at the beginning of the chase period in both control and BRaf-deleted dentate gyrus, indicating that either the BrdUlabel was diluted by cell proliferation, or by cell loss ( Figure 5B). The fraction of BrdU/NeuN positive neurons was significantly reduced in cKO mice compared to controls, indicating that neuronal differentiation is impaired in the absence of BRaf ( Figure 5C). In contrast, the fraction of BrdU/GFAP positive radial glia-like stem cells was approximately twofold increased in cKO mice as compared to controls ( Figure 5D). The fraction of horizontal BrdU/GFAP-positive cells corresponding to differentiated astrocytes was unchanged in the absence of BRaf, indicating that no switch from neural to glial fate had occurred ( Figure S5). Taken together, these results demonstrate an essential role of BRaf in the differentiation of precursor cells into neurons. In the absence of BRaf, the number of GFAP-positive precursor cells that were proliferating 12 days ago increased at P22 in cKO dentate gyrus.
We used an in vitro culture system in order to investigate whether neuronal differentiation involving the growth of neurites in hippocampal neurons is impaired when BRaf is eliminated. Postnatal day P0/P1 hippocampi were dissected from the brain, and dissociated cell cultures were maintained in serum-free medium. After 6 days of culture, we fixed the cells and stained the samples for expression of BRaf and Map2, a marker of dendritic differentiation; nuclei were visualized by DAPI ( Figure 6A

Discussion
This study focuses on the role of the kinase BRaf in postnatal brain development. Using a conditional loxP recombination-site flanked allele of BRaf we obtained a mouse expressing nonfunctional BRaf that enabled the efficient deletion of functional BRaf in postnatal cerebellum, hippocampus and other brain areas of cKO compound mice. Mice expressing non-functional BRaf in neural stem cell-derived tissue survived up to four weeks. They showed alterations in the cerebellum, with reduced numbers and misplaced ectopic Purkinje cells, decreased sizes of the glomeruli in the granule cell layer and altered structure of the dendritic arborizations of the Purkinje neurons. The finding that virtually all Purkinje neurons of cKO compound animals failed to stain for BRaf expression argues against the view that a small number of ''escaper'' cells in which Cre recombinase did not delete exon 3 can maintain the Purkinje cell architecture. In the hippocampus, the overall cornus ammonis (CA) architecture appeared to be unchanged, but the growth of the granule cell layer in the dentate gyrus was diminished. Notably, reduced granule cell volume was obvious at P21 but normal at P12 in cKO male compound mice, compared to control littermates. In vitro cultivation of P0/P1 hippocampal cells revealed that BRaf deficient neurons were impaired in their differentiation into neurons lacking Map2 expressing dendrites. We propose, in analogy to the reduced arborization of the cerebellar Purkinje cells that a reduced hippocampal granule cell volume of cKO compound animals is caused also by reduced growth of their dendrites. This interpretation is supported by the absence of overt apoptosis in corresponding brain areas and by cell proliferation studies. BrdU birth-dating analysis of dentate gyrus progenitor cells at P10/P11 showed that their differentiation into NeuN-positive granule cell neurons over a 12-day period was severely reduced in mice lacking BRaf. Together with alterations in mouse behaviour, these observations indicate that all neuronal cellular structures develop in the absence of BRaf, but neuronal generation and maturation of dendrites are impaired in some neuronal subtypes and therefore synaptic circuits are unlikely to be fully functional. Impairment of sensory neuron differentiation has been observed before upon conditional elimination of BRaf [25]. Furthermore, conditional elimination of BRaf in neural precursors cells caused defective oligodendrocyte differentiation accompanied by dysmyelination [47].
The targeting strategy employed here is similar to the protocol chosen by Zhong and colleagues [25] in which exon 3 was also targeted for deletion by Nestin promoter driven Cre recombinase. In both strains of mice death occurs in the fourth postnatal week. Immunoblotting with a C-terminal BRaf antibody revealed the existence of a truncated, non-functional ,82 kDa BRaf protein that most likely arises from a splice isoform from the BRaf del allele due to splicing from exon 2 to exon 4 (see below). This splicing is plausible since there is no change of reading frame in the first four exons of BRaf. Nevertheless, homozygous BRaf del/del embryos expressing the ,82 kDa BRaf protein isoform were unable to stay alive beyond embryonic day E11. This is in accordance with observed phenotypes of mice with constitutive elimination of BRaf on embryonic development [22]. Notably, we demonstrate that two downstream targets of BRaf signalling, the phosphorylation levels of the kinases ERK1,2, as well as the levels of the early growth response 1 transcription factor Egr1 are significantly reduced in the hippocampus of cKO mice compared to control mice. This constitutes biochemical evidence that the smaller ,82 kDa BRaf protein we detect in our cKO mice is a nonfunctional kinase.
The most prominent function of neural progenitors in the subgranular zone (SGZ) of the dentate gyrus in the hippocampus of adult mammals, including humans, is the generation of new granule cell neurons [32][33][34]48]. It is a striking example of structural anatomical reorganization of a neuronal circuit leading to functional changes at the level of synapse formation. Notably, neurogenesis in the SGZ of the dentate gyrus is not a simple replacement event to compensate for the loss of cells, but instead leads to the addition of new glutamatergic neurons in the granule cell layer. The functional importance of adult neurogenesis has been demonstrated in mice where impaired spatial memory function was noted after elimination of the stem cells [49]. A small fraction (about 7%) of Sox2-positive neural progenitors in the subgranular zone has the potential to differentiate into astrocytes [50]. Using BrdU birth-dating in conjunction with GFAP immonostaining, we have observed an increased fraction of radial glia precursor cells. The fraction of horizontal BrdU/GFAPpositive cells corresponding to differentiated astrocytes was unchanged in the absence of BRaf indicating that the cell fate of  the progenitor cell progeny did not change from neuron to glia, as has been observed for example upon transient Notch activation [51]. The use of Cre transgenic mice that are specific for neural stem cells and can be activated in adult animals by the hormone tamoxifen, such as CreGli1-CreER T2 transgenic mice [52] may provide a means for a more detailed study of the role of BRaf in adult hippocampal neurogenesis. The increased number of BrdU positive cells in the subgranular zone of P20 cKO male compound mice could reflect a reactive response. Our immunostaining data indicate that BRaf is expressed in the SGZ band of the dentate gyrus, albeit at lower levels than in the differentiated granule neurons. It has been observed before that the mRNA encoding BRaf is expressed in the dentate gyrus and upregulated after longterm potentiation [18]. The increased fraction of BrdU positive cells that have lost Ki-67 expression is intriguing. The Ki-67 antigen is a nuclear and nucleolar protein expressed during all active phases of the cell cycle (G1, S, G2, and mitosis) and thus is a marker for proliferating precursor cells, but is absent from differentiated neurons or quiescent (G0) stem cells [46].
In our Western blot experiments of total embryonic, as well as postnatal brain extracts we observed two BRaf bands of ,92 and ,89 kDa, respectively. Two BRaf bands have been noted before [53]. We further observed that these two bands are absent in tissue extracts from embryos with homozygous exon 3 deletion (lanes del/del; Figure 1B). These genetic data do not support the idea that the known splicing in exons 8b/9b [54] is the basis for the isoforms observed in the immune blots but rather argue in favour of a splicing within exon 3 of the BRaf gene. Sequencing of RT-PCR products confirmed the existence of an intron in exon 3 that was spliced from the mRNA such that in the encoded smaller BRaf protein 33 amino acid residues would be missing. This intron/spliced-out segment (GT, AAT,GGA,-,CTT,TCA,G) starts with GT at the 59-splice site and ends with an AG at the 39 splice site and conforms to the canonical splicing pathway [55]. The encoded protein domain is situated N-terminal in close proximity to the RAS binding domain of RAF located in conserved region CR1 which is instrumental in recruitment of RAF to the cell membrane and the phosphorylation in the activation segment (residues threonine 599 and serine 602 in CR3 of human BRAF). Whether this alternative BRaf isoform can be activated in a distinct manner remains to be determined. Dimer formation in wild-type BRaf and transactivation of CRaf by wildtype BRaf are dependent on the interaction with activated Ras via the RBD domain [13][14][15]. CRaf/BRaf heterodimerization levels have been reported to be significantly reduced in BRaf knockdown cells [56]. Therefore, a BRaf protein lacking exon 3 would be expected to be unable to dimerize.

Animals
Mice were housed under barrier condition in air-filtered, temperature-controlled units with a 12 h light/dark cycle with free access to food and water. Animals were monitored for signs of distress by daily inspection. All mice were of similar genetic background ($90% C57Bl/6, #10% 129Sv) and maintained as heterozygotes. Nestin-Cre mice have been described elsewhere [57] and were obtained from M. Sendtner. Compound mice were obtained by intercrossing heterozygotes. Genotype analysis of tail DNA was done by PCR at the age of three weeks. All progeny were confirmed by standard PCR-based genotyping of DNA isolated from tail biopsies or yolk sacs. Embryo staging was based on the presence of a plug and considered 0.5 days post coitus. The following primer sequences were employed: BRaf (see Figure 1A for their position): The primer pair for detection of the wild-type allele (wt) was brafgs12 (5-TGT AGC CTC GGC TGT GGA ACT C ) and brafgas8 (59-GAG ACC AAA CCA AGG ACC TCT G) yielding a fragment of 281 bp; the primer pair for the nfl allele was brafgs12 and neo5a (59-ACG GAG CCG GTT GGC GCC TA) yielding a fragment of 326 bp; alternatively, the primer pair 5xhoI (59-CCT GAA AGC TGC TAG TAG AAG AC) and baxh (59-ACA TTG TTG ATG CTG CTC GAT CC) was used yielding a fragment of 317 bp. The primer pair for the del allele was 5xhoI and brafgas8 yielding a fragment of 421 bp. Nestin-Cre, 59-CCG TTT GCC GGT CGT GGG and 59-CGT ATA TCC TGG CAG CGA TC, yielding a single PCR product of ,395 bp. All animal studies were performed in accordance with German legislation and were approved by the Bavarian State authorities for animal experimentation. All animal experiments were approved by the Committee on the Ethics of Animal Experiments of the Government of Lower Franconia (Permit Number: 55.2-2531.01-83/09). All surgery was performed under KetanestH/RompunH anesthesia, and all efforts were made to minimize suffering.

Generation of ''Floxed'' BRaf Mice
The targeting construct was generated by inserting a phosphoglycerate kinase promoter-neomycin resistance cassette, flanked by two loxP sites, into the SpeI restriction site ,500 bp 39 of exon 3 in a genomic fragment of BRaf. A third loxP site was inserted into the XhoI site between exons 2 and 3, ,300 bp 59 of exon 3 ( Figure  S1A). The targeted allele was generated via homologous recombination by introducing the linearized targeting construct into mouse embryonic stem cells using standard methods. Among ,300 ES colonies, 12 were identified as targeted clones using a PCR assay (not shown). A 59 ,500 bp and a 39 ,600 bp genomic DNA fragment (shown in Fig. S1A) were used as probes for Southern blot confirmation of ES clones harbouring the targeted allele ( Figures S1B and S1C). Four targeted clones were injected into C57BL/6J blastocysts and then placed into pseudopregnant C57BL/6J females. Two clones gave germ line transmission. Chimeric mice carrying the floxed braf allele were crossed to C57BL/6J mice. The BRaf wt/nfl progeny of that cross were then interbred to obtain BRaf nfl/nfl mice from both ES clones. In addition, BRaf wt/nfl progeny from one ES clone was crossed to mice expressing Cre recombinase under the control of the adenovirus EIIa promoter [40] to remove the neo cassette yielding Bra wt/fl mice (lacking the neo cassette in the BRaf allele) and BRaf wt/del (lacking exon-3 in the braf allele, Figure 1A). The PCR for the deletion of the neo cassette employed the primer pair spba (59-TGG CAC TTA AAT ATA AGT ACT AGA TC) and brafgas8 yielding a fragment of 161 bp. Homozygous BRaf nfl/nfl and BRaf fl/ fl were born at Mendelian ratios, were viable and fertile and showed no obvious phenotype. The presence of the neo cassette in BRaf fl/fl mice did not alter BRaf expression at the mRNA level, as assessed by semi-quantitative RT PCR compared to wild type controls (not shown). For conditional removal of the targeted BRaf gene in neural progenitors, BRaf fl/fl mice were crossbred with Nestin-Cre mice [57]. Nestin-Cre mice were obtained from M. Sendtner.

Birthdating and Histological Analysis
For cell proliferation studies we employed male littermates that were injected intraperitoneally with bromodeoxyuridine (BrdU) (50 mg/g of body weight). After appropriate labelling periods, mice were anaesthetized with KetanestH/RompunH and perfused transcardially with saline, followed by 4% paraformaldehyde (PFA) in PBS, pH7.4. Brains were dissected and postfixed in 4% PFA/PBS overnight at 4uC and subsequently embedded in paraffin and serially sectioned as described [44]. The left part of the brain was serially cut into a spate of ten 10 mm sagittal sections always using the first slide for Nissl staining. Quantified data were normalized to the Nissl stained DG-volume data which were measured from the beginning of the DG formation until its fusion, summarized and multiplied with slide thickness. Volume data were mentioned in mm 3 . Cerebellar lobuli length was measured from the central core axis outward including molecular layer size in comparable slices of Nissl or HE stained sections. Glomeruli/ granule cell distribution in the cerebellum of Nissl stained sections was analysed using Keyence 9000 brightness quantification software in three comparable selected areas in each mice to normalize data to a defined area size (mm 2 ). Glomeruli parts were shown as % of analysed area. Primary dendrite length of cerebellar Purkinje cells was analysed using Keyence 9000 measuring software. Dendrite length was quantified from the Purkinje cell body until the first arborization. Analysed data are shown in mm.

RNA Isolation and RT-PCR Analysis
Total RNA was extracted from embryo tissue using Trizol followed by purification with spin columns from the Pure Link RNA Mini Kit (Ambion). cDNA was synthesized with recombinant M-MuLV reverse transcriptase using 2 mg of total RNA using random hexamer primers (Fermentas). Aliquots of the reaction mixture were used for the subsequent PCR amplification. The primer sequences for BRaf amplification were 5-GAC CCG GCC ATT CCT GAA G (sense, located in exon 1) and 5-CTG TCT CGA ACT GTA ACA CCA CAT (antisense, located in exon 4). PCR conditions were as follows: a reaction volume of 30 ml for 5 minutes at 95uC for initial denaturation, followed by 35 cycles of 30 seconds at 95uC, 30 seconds at 60uC, 30 seconds at 72uC, and a final extension at 72uC for 10 minutes. PCR products were visualized on 2.5% agarose gels stained with ethidium bromide. Reactions without reverse transcriptase were used as a negative control. The RT-PCR for ß-actin with primers (sense, 59-GTA TGG AAT CCT GTG GCATC; antisense, 59 AAG CAC TTG CGG TGC ACG A) was used to check the quality of the RNA extraction and reverse transcription.

Hippocampal Cell Culture and Staining
Hippocampi were dissected from postnatal day P0/P1 pups and treated with trypsin (15 min, 37uC, 1%, Worthington). After addition of trypsin inhibitor, 25.000 cells were cultured on 10 mm poly-L-lysine coated coverslips in 100 ml neurobasal medium containing B27 and glutamax. After 6 days culturing in vitro, cells were fixed and stained for immunofluorescence.

Mouse Behavioural Analyses
Walking and balancing on a pencil was used to measure motor coordination; the time period spent on the rod was determined; cut-off time of this assay was one minute. Control mice could balance without problem for even longer time periods.

Statistical Analysis
Data are presented as mean 6 s.e.m. Paired t-test (two-tailed) was used to compare two groups (P,0.05 was considered significant). Data values in each group were assessed for normal distribution using Statsistica 8 (Statsoft) quantile plot test.