Essential role of protein kinase Bγ (PKBγ/Akt3) in postnatal brain development but not in glucose homeostasis

Protein kinase B (PKB/Akt) is a second messenger-regulated kinase that has been implicated in many crucial cellular processes, such as glucose metabolism, transcription, cell proliferation, apoptosis, migration and growth(Brazil and Hemmings, 2001; Chan et al., 1999; Datta et al., 1999; Kandel and Hay, 1999; Scheid and Woodgett, 2001). Deregulation of PKB activity contributes to cell transformation and diabetes(Brazil and Hemmings, 2001; Nicholson and Anderson, 2002). In mammalian cells, three closely related isoforms of the PKB family have been identified and termed PKBα/Akt1, PKBβ/Akt2 and PKBγ/Akt3(Brodbeck et al., 1999; Cheng et al., 1992; Jones et al., 1991a; Jones et al., 1991b; Masure et al., 1999; Nakatani et al., 1999). These proteins are products of three separate genes located on distinct chromosomes and are widely expressed with a few isoform-specific features(Altomare et al., 1998; Yang et al., 2003). All three members of the PKB family contain a highly conserved N-terminal pleckstrin homology (PH) domain, a central catalytic domain and a short C-terminal regulatory domain (Hanada et al.,2004).

PKB is activated by numerous stimuli, including growth factors, hormones and cytokines (Brazil and Hemmings,2001; Chan et al.,1999; Datta et al.,1999). Activation of PKB occurs in response to signalling via phosphoinositide 3 kinase (PI3K) and requires the membrane-bound second messenger phosphatidylinositol-3,4,5-triphosphate[PtdIns(3,4,5)P₃ or PIP3](Burgering and Coffer, 1995; Cross et al., 1995; Franke et al., 1995). The current model for PKB regulation proposes that PtdIns(3,4,5)P₃, generated following PI3K activation,interacts with the PH domain of PKB, recruiting the inactive kinase from the cytoplasm to the plasma membrane and promoting a conformational change that allows phosphorylation on two regulatory sites by upstream kinases at the plasma membrane. One of these critical phosphorylation sites resides in the activation loop of the kinase domain (Thr308 in PKBα) and the other is located in the C-terminal regulatory domain (Ser-473 in PKBα), termed the hydrophobic motif (Alessi et al.,1996; Brodbeck et al.,1999; Meier et al.,1997). The upstream kinase that phosphorylates Thr308 in the activation loop of the kinase domain of PKBα in a PIP3-dependent-manner has been identified and termed 3-phosphoinositide-dependent kinase 1 (PDK1)(Alessi et al., 1997; Stokoe et al., 1997). Thr308 phosphorylation is necessary and sufficient for PKB activation; however,maximal activation requires additional phosphorylation at Ser473(Alessi et al., 1996; Yang et al., 2002a; Yang et al., 2002b). Several different protein kinases and mechanisms have been proposed for the phosphorylation of the hydrophobic motif(Brazil and Hemmings, 2001; Yang et al., 2002b).

Mice with targeted disruption of Pkbα and/or Pkbβ have been obtained recently with Pkbα mutant mice displaying an increased neonatal lethality and a reduction in body weight of ∼30% (Chen et al.,2001; Cho et al.,2001b; Yang et al.,2003). Moreover, loss of Pkbα leads to placental hypotrophy with impaired vascularisation(Yang et al., 2003). By contrast, Pkbβ-deficient mice are born with the expected Mendelian ratio and exhibit a diabetes-like syndrome with elevated fasting plasma glucose, hepatic glucose output, peripheral insulin resistance and an compensatory increase of islet mass (Cho et al., 2001a). Compared with Pkbα mutant mice, Pkbβ-deficient mice are only mildly growth retarded(Cho et al., 2001a; Garofalo et al., 2003). However, mice lacking both isoforms die after birth, probably owing to respiratory failure (Peng et al.,2003). Pkbαβ double mutant newborns display a severe reduction in body weight (≈50%), prominent atrophy of the skin and skeletal muscle, as well as impaired adipogenesis and delayed ossification.

Here, we report the generation and characterisation of mice with targeted disruption of the Pkbγ gene. Compared with Pkbα^-/- and Pkbβ^-/- mice, Pkbγ mutant mice display a distinct phenotype without increased perinatal mortality,growth retardation or altered glucose metabolism. However, loss of PKBγprofoundly affects postnatal brain growth. Brains from adult Pkbγ mutant mice show a dramatic reduction in size and weight. Taken together, our results reveal a novel and important physiological role for PKBγ in postnatal brain development.

An ∼11-kb HindIII fragment was subcloned containing exons 4 and 5 of Pkbγ. A NcoI site was generated in exon 4 for insertion of a ∼5-kb IRES-lacZ-Neo cassette. The targeting vector was linearised with SalI and electroporated into 129/Ola ES cells. An external probe was used for ES cell screening following XbaI digestion. An internal probe and a lacZ-Neo probe were used to characterize ES clones positive for homologous recombination (data not shown). Correctly targeted ES cells were used to generate chimeras. Male chimeras were mated with wild-type C57BL/6 females to obtain Pkbγ ^+/- mice. Pkbγ ^+/- mice have been backcrossed twice with pure C57BL/6 mice and the progeny of Pkbγ ^+/- intercrosses have ∼75% C57BL/6 genetic background. Progeny were genotyped by multiplex PCR with the following three primers: (1) P35, GGTTCTGTGGGAGGTAGTTCTC; (2) Neo-2,GCAATCCATCTTGTTCAATGGCCG; and (3) E4γ, CCATCGGTCGGCTACGGCTTGG.

The levels of PKB isoforms in wild-type and mutant mice were determined by quantitative Q-RT-PCR. The experiment was performed as described previously(Yang et al., 2003). Briefly,total RNA was purified using Trizol Reagent (Invitrogen). For the Q-PCR reaction, 50 ng total RNA was mixed with 5′ and 3′ primers, Taqman probe, MuLV reverse transcriptase, RNase inhibitor and the components of the TaqMan PCR reagent kit (Eurogentec) in a total volume of 25 μl following the TaqMan PCR reagent kit protocol.

For Western blot analysis, protein lysates were processed as previously described (Yang et al., 2003). PKB isoform-specific antibodies were obtained by immunizing rabbits with isoform-specific peptides as previously described(Yang et al., 2003). Antibodies against phospho-PKB (Ser473), phospho-GSK3α/β (Ser21/9),phospho-TSC2 (Thr1462) and phospho-p70S6K were purchased from Cell Signalling Technologies. Antibodies against p27 and ERK were purchased from Santa Cruz Biotechnology. The antibody against phospho-ERK (Thr202/Tyr204) was purchased from Promega and the Pan-Actin antibody was obtained from NeoMarkers. Western blots were scanned using a GS-800 BioRad densitometer with a resolution of 63.5 μm × 63.5 μm and bands were quantified using Proteomweaver 3.0.0.6 (Definiens).

For histological analysis, animals were perfused with phosphate-buffered saline and 4% paraformaldehyde in phosphate-buffered saline. Organs were dissected and kept in the same fixation solution overnight at 4°C. Samples were embedded in paraffin following dehydration in ethanol. Tissues were cut into 6 μm sections and stored for staining. For Haematoxylin-Eosin and Cresyl-Violet (Sigma) staining, sections were freed of paraffin and stained.

To determine the number of cells in a whole brain, the DNA content was determined as described by Labarca and Paigen(Labarca and Paigen, 1980). This method is based on the enhancement of fluorescence after binding of bisbenzimid (Riedel-de Haen) to DNA. A linear standard curve (1-10 μg/ml)was prepared to calculate the DNA concentration. The relative cell size in posterior cortex was measured on plane-matched, parasagittal brain sections stained with DAPI (Biotium). Image-Pro® Plus (Media Cybernetics) was used to count the cell number and to calculate the mean area occupied by one cell. The relative cell size is expressed as percent of wild type.

MRI studies of 4-month-old female Pkbγ wild-type(n=5) and mutant mice (n=5) were performed at 2.35 T using a MRBR 4.7/400 mm magnet (Magnex Scientific, Abingdon, England) equipped with BGA20 gradients (100 mT m^-1) and a DBX system (Bruker Biospin,Ettlingen, Germany). For in vivo examinations, animals were anaesthetized(1.0-1.5% halothane in 70:30 N₂O:O₂) and treated as previously described (Natt et al.,2002). Briefly, radiofrequency (rf) excitation and signal reception were accomplished with use of a Helmholtz coil (∅ 100 mm) and a surface coil (∅ 20 ×12 mm), respectively. Three-dimensional T₁-weighted (rf-spoiled 3D FLASH, repetition time TR=17 mseconds,echo time TE=7.6 mseconds, flip angle 25°, measuring time 84 minutes) and T₂-weighted MRI data sets (3D fast spin-echo, TR/TE=3000/98 mseconds, measuring time 58 minutes) were acquired with an isotropic resolution of 117 μm. Volumetric assessments were obtained by analysing T₁-weighted images using software provided by the manufacturer(Paravision, Bruker Biospin, Ettlingen, Germany). After manually outlining the whole brain and the ventricular spaces in individual sections, respective areas were calculated (in mm²), summed and multiplied by the section thickness.

E16.5 murine hippocampal neurons were isolated from timed matings. Cells were kept in culture for 7 days on polylysine coverslips coated with B27/Neurobasal (GIBCO Life Technologies). At day 7, cultures were treated with glutamate (15 mM/24 hours), staurosporine (50 nM/12 hours) or were left untreated. For the detection of apoptotic cells, five fixed cultures per genotype were stained using a TUNEL-assay (Roche) according to the manufacturer's instructions and counterstained with DAPI (Biotium). At least 200 cells per culture were counted and the percentage of apoptotic cells was calculated.

Mice were housed according to the Swiss Animal Protection Laws in groups with 12 hours dark/light cycles and with free access to food and water. All procedures were conducted with the approval of the appropriate authorities.

Random-fed and fasting blood glucose levels were determined in 5- to 6-month-old mice. Blood samples were collected from tail veins and glucose levels were determined using Glucometer Elite (Bayer). Mice (aged 5-6 months)were fasted overnight before the start of the glucose and insulin tolerance tests. Glucose (2 g/kg) was given orally to conscious mice. Insulin (1 U/kg)was administered by intraperitoneal injection to conscious mice. Blood samples were collected at indicated times from tail veins and glucose levels were determined using Glucometer Elite (Bayer). Blood glucose levels were expressed as percentage of initial value. Blood insulin levels were measured with a Mouse ultra-sensitive insulin ELISA (Immunodiagnostic Systems).

Microarray analysis was performed using murine MOE430A GeneChips™(Affymetrix). Total RNA (10 μg) was reverse transcribed using the SuperScript Choice system for cDNA synthesis (Life Technologies) and biotin-labelled cRNA generated using the Enzo BioArray High Yield RNA transcript labelling kit (Enzo Diagnostics) following the manufacturer's protocol. cRNA fragmentation and hybridisation were performed as recommended by Affymetrix. Expression data were calculated using the RMA algorithm from BioConductor (Irizarry et al.,2003). A gene was considered to be significantly altered in its expression if it had an Affymetrix change P-value of less than 0.003 for either increase or decrease in at least two-thirds of replicate comparisons and it had a minimum expression value of 100 in at least one condition. A fold-change threshold of 1.5 was then applied and the resulting genes were subjected to a one-way ANOVA with a P-value cut-off of 0.05. A Benjamini and Hochberg multiple testing correction and a Tukey post-hoc test were applied.

To compare body weight, brain weight and volume, brain/body weight ratio,DNA content, cell number and percentage of apoptotic cells between Pkbγ^+/+ and Pkbγ ^-/-, an unpaired Students t-test was performed. P values under 0.05 were considered as significant and values below 0.01 as highly significant.

Pkbγ ^-/- mice were generated by targeted disruption of exon 4 as shown in (Fig. 1A). The ablation of Pkbγ was confirmed by PCR,Q-RT-PCR and Western blot analysis using brain tissue samples from Pkbγ wild-type (Pkbγ^+/+),heterozygous (Pkbγ^+/-) and mutant(Pkbγ^-/-) mice(Fig. 1B). Quantitative RT-PCR displayed complete ablation of the PKBγ mRNA in brain samples of Pkbγ mutant mice (n=3, data not shown). To confirm the absence of PKBγ at the protein level, Western blot analysis was performed with an isoform-specific anti-PKBγ antibody. In contrast to Pkbγ^+/+ brain tissue samples, PKBγ was not detectable in samples derived from mutant mice(Fig. 1C). In addition, Western blot analysis with antibodies specific for PKBα and PKBβ,respectively, showed no compensatory upregulation of PKBα and/or PKBβ, respectively, in the brain of Pkbγ^-/-mice (Fig. 1D).

It has been reported that the PKBα and β isoform are widely expressed in all organs, but with some isoform-specific features(Altomare et al., 1998; Yang et al., 2003). Less is known about the tissue distribution of the PKBγ isoform. Previous reports suggest that PKBγ has a more restricted distribution with high levels in the adult brain and foetal heart and low levels in liver and skeletal muscle (Brodbeck et al.,1999; Masure et al.,1999; Yang et al.,2003). We assessed the distribution of PKBγ mRNA in 15 major tissues of adult mice by quantitative RT-PCR and normalised to the level of PKBγ in the brain (Fig. 2A). PKBγ mRNA was found at the highest level in brain and testis, and at lower levels in lung, mammary gland, fat and spleen.

To investigate whether PKBγ ablation leads to compensatory increase in PKBα and/or PKBβ, total RNA isolated from brain, testis, lung,mammary gland, fat and spleen of three Pkbγ wild-type and three mutant mice was subjected to quantitative RT-PCR. The levels of PKBα andβ were normalised to the level of PKBα in wild-type brain and set as 100%. Overall, no marked upregulation of PKBα and/or PKBβ was observed, including the brain (Fig. 2B,C). These results are consistent with Western blot analysis of protein extracts of brains from Pkbγ wild-type and mutant mice(Fig. 1D). To investigate the distribution and levels of individual PKB isoforms within the brain, protein lysates were prepared from ten anatomically and functionally different regions. In general, all three isoforms were expressed in all examined regions but with certain isoform-specific features(Fig. 2D). PKBα is expressed in all regions at similar levels, whereas PKBβ is expressed at moderate levels in cortex, cerebellum, hippocampus and olfactory bulb, and at lower levels in the hypothalamus, midbrain, brain stem and spinal cord. PKBγ is expressed in all examined regions but at higher levels in the cortex and in the cerebellum.

The phosphorylation/activation state of PKB in protein extracts from brains of 3-month-old Pkbγ^+/+ and Pkbγ^-/- mice was analysed using an antibody specific for the phosphorylation site (anti-phospho-Ser-473) in the hydrophobic motif of the regulatory domain of all three isoforms(Fig. 3A). As expected, levels of activated PKB (quantification was done using actin as loading control) in Pkbγ^-/- mice was significantly lower than in Pkbγ^+/+ littermate controls (100% versus 38%), but was not completely abolished (Fig. 3B). The phosphorylation levels of glycogen synthase kinase 3(GSK3), tuberous sclerosis complex 2 (TSC2), p70S6K, ERK and p27 in Pkbγ^-/- mice were not significantly changed when compared with wild-type littermate controls (levels of phosphorylated protein were normalised with the level of unphosphorylated protein). We repeated this experiment using brain samples of 14 days old mice and obtained comparable results with significantly reduced levels of total phosphorylated PKB and unchanged levels of phosphorylated substrates (data not shown).

Several recent studies emphasize the importance of insulin-like growth factor 1 (IGF1) signalling in the survival and proliferation of oligodendrocytes and myelination in the central nervous system(D'Ercole et al., 2002). Mice lacking IGF1 also have reduced brain weight and reduced size of the corpus callosum and anterior commissure (Ye et al., 2002). Therefore, we performed a western blot analysis with cell lineage-specific markers using antibodies against myelin basic protein(oligodendrocytes), glial fibrillary acidic protein (astrocytes) and M-neurofilaments (neurons). Significantly, no obvious differences in the expression levels of these marker proteins were observed when comparing samples of Pkbγ^+/+ and Pkbγ^-/- mice (n=5 per genotype, data not shown).

There was no evidence for increased mortality of Pkbγ^-/- mice after birth in an analysis of more than 400 pups (aged 3-4 weeks) [Pkbγ^+/+: Pkbγ^+/-: Pkbγ ^-/-=104 (25.6%): 206 (50.7%): 96 (23.6%)]. As PKBγis highly expressed in the testis, fertility of mutant mice was tested using male Pkbγ^-/- × female Pkbγ^+/+ and female Pkbγ^-/-× male Pkbγ^+/+ matings. Both male and female mutant mice matings gave normal pregnancies and births, indicating that fertility is not impaired in either male or female Pkbγ mutant mice (data not shown).

As PKB has been implicated in the regulation of cell and organ growth, body weight was measured in male Pkbγ mutant mice and wild-type littermate controls (n=5-8 per group) at different time points(Chen et al., 2001; Cho et al., 2001b; Peng et al., 2003; Yang et al., 2003). Body weight did not differ significantly between male Pkbγ mutant mice and wild-type controls at any time point(Fig. 4A). A similar result was obtained with female Pkbγ^-/- mice and Pkbγ^+/+ controls (n=5-8 per group, data not shown), indicating that PKBγ does not play a significant role in the overall growth of mice.

To investigate a potential role of PKBγ in the regulation of glucose metabolism, blood glucose levels were measured in adult (5-6 months) Pkbγ mutant mice under random-fed and fasting condition, and compared with age- and gender-matched wild-type controls. Interestingly, both random-fed and fasting blood glucose levels, were not significantly different between wild-type and mutant mice (Fig. 4B). Additionally, blood insulin levels in random-fed condition did not differ significantly between Pkbγ^+/+ and Pkbγ ^-/- mice (1.32±0.08 μg/l versus 1.30±0.13 μg/l; n=6). To further investigate glucose metabolism, overnight fasted mice were challenged with insulin (insulin tolerance test) or glucose (glucose tolerance test). To test the insulin responsiveness, insulin (1 U/kg) was applied by intraperitoneal injection und blood glucose levels were measured at indicated time points using blood from tail veins. No obvious differences of blood glucose levels in the insulin tolerance test were found between the groups with mutant and control mice(Fig. 4C). Additionally, mice were challenged with orally applied glucose (2 g/kg) and blood glucose levels were measured at indicated time. Compared with wild-type mice,Pkbγ ^-/- mice displayed a very similar response to the glucose load (Fig. 4D). Taken together, these results suggest that PKBγ does not play a significant role in the maintenance of glucose homeostasis.

Next, adult Pkbγ^+/+ and Pkbγ^-/- mice were dissected and all major organs were investigated macroscopically. Compared with Pkbγ^+/+ littermate controls, the overall size of brains from adult Pkbγ^-/- mice was strikingly reduced. A representative example is shown in Fig. 5B-D. Furthermore, the weights of freshly dissected brains of Pkbγ wild-type and Pkbγ mutant mice were measured at different ages(Fig. 5A). Compared with age-and gender-matched wild-type littermate controls, brains from adult Pkbγ^-/- mice (3-12 months old) exhibited a highly significant reduction in weight of about 25% (range 22%-29%), affecting both males and females (data for females not shown, n=5-8). Interestingly,at birth, brain weight did not differ significantly between Pkbγ^+/+ and Pkbγ^-/- mice. The reduction in brain size and weight was first observed at the age of 1 month, but was less pronounced compared with adult mice (≈18%). In contrast to Pkbα, Pkbβ and IgfI-null mutant mice(Beck et al., 1995; Cheng et al., 1998; Garofalo et al., 2003; Powell-Braxton et al., 1993),the brain/body weight ratio of Pkbγ mutant mice was also significantly reduced to a similar extent.

MRI is an excellent non-invasive technique for studying brain anatomy in transgenic and mutant mice (Kooy et al.,2001; Natt et al.,2002). For further characterization of Pkbγ mutant brain anatomy, five adult female mutant and five age and gender matched wild-type littermate controls (with the same genetic background) from Pkbγ^+/- matings were examined using high-resolution 3D MRI as previously described (Natt et al., 2002). Fig. 5E-J shows representative T₂-weighted images of Pkbγ^+/+ and Pkbγ^-/- brains in a sagittal, horizontal, and coronal section orientation. Complementary in vivo volumetry based on T₁-weighted 3D MRI confirmed the much smaller brain size of all five Pkbγ mutant mice. Compared with wild-type littermates, whole brain volume in Pkbγ^-/-mice was significantly reduced by about 24% (513±14 mm³versus 391±16 mm³, P<0.01; n=5). Although several brain regions were affected, including olfactory bulb, cortex and hippocampal formation, no alteration in the structural organization of the brain was observed, confirming the macro-pathological findings. Moreover,because the volume of the ventricular spaces in Pkbγ^-/- mice was reduced (7.85±2.2 mm³ versus 5.55±2.8 mm³) in proportion to that of the whole brain and MRI revealed no enlargement of the subarachnoid space, the occurrence of an internal or external hydrocephalus is excluded as a cause of reduced brain size. Interestingly, in contrast to any wild-type littermate controls, all five Pkbγ^-/- mice presented with a marked thinning of the corpus callosum (downward arrowheads in Fig. 5E,F). White matter fibres connections are depicted as hypointense structures in T₂-weighted images. Although unequivocally identifiable in Pkbγ^+/+ mice(Fig. 5E,G,I), the corpus callosum is partly indistinguishable from surrounding grey matter in Pkbγ^-/- mice(Fig. 5F,H,J). Anterior and posterior commissures as well as the hippocampal fimbria are less prominently affected.

Histological examination of Pkbγ^-/- brain was performed to investigate changes at the microscopic level. Various brain regions, including cerebellum, hippocampus and corpus callosum, were examined following Haematoxylin/Eosin staining (Fig. 6A-F). With the exception of the reduced size of all regions, no abnormalities in the overall structure of the different brain regions were observed. In line with the in vivo MRI findings, the thickness of the corpus callosum in Pkbγ mutant mice was markedly reduced (downward arrowheads in Fig. 6G,J). Additionally, myelin staining with luxol Fast Blue revealed not only a reduction in thickness of the corpus callosum but also less intense staining of the structure (Fig. 6H,K).

Next, we indirectly assessed the cell number by measuring the amount of DNA in whole brains derived from Pkbγ^+/+ and Pkbγ^-/- mice. The amount of DNA is considered as an indicator of cell number, whereas the amount of DNA per gram of tissue is an indicator of cell density which is reciprocal to cell volume(Zamenhof, 1976). The DNA contents of brains from newborns and 1-month-old mice were determined using the method described by Labarca and Paigen(Labarca and Paigen, 1980). Briefly, this method is based on the enhancement of fluorescence after binding of bisbenzimid to DNA. Compared with wild-type control samples, whole brain DNA content of Pkbγ^-/- newborns did not differ significantly, indicating that cell number was not changed. Similarly, the DNA content per gram of brain tissue was comparable between Pkbγwild-type and mutant mice, indicating a comparable cell size. In contrast to newborns, the DNA content in brains of 1-month-old Pkbγ mutant mice was slightly, but significantly, reduced compared with the Pkbγ^+/+ controls(Table 1). However, the DNA content per gram of tissue was, significantly increased in samples from Pkbγ^-/- compared with wild-type littermate controls,indicative of increased cell density (and reduced cell size).

Additionally, the relative cell size in the posterior cortex of adult mice(3 months) was determined based on histological slides stained with DAPI. Cells were analysed using Image-Pro Plus in a defined field of view (≈1000 per field), and the area occupied by one cell was subsequently calculated. In line with the results of the DNA content experiment, the relative cell size was significantly and consistently reduced in Pkbγ mutant mice(100±10% versus 80±7%; n=5 per genotype; P<0.05). Taken together, the results show that both cell size and cell number contribute to the reduction in brain size observed in mutant mice,but the relative contribution of cell size reduction plays a more important part.

It is established that the PI3K/PKB pathway plays a crucial role in cell survival in the central nervous system(Datta et al., 1999; Dudek et al., 1997; Kim et al., 2002). To investigate the potential role of PKBγ in apoptosis, primary cell cultures were established from Pkbγ^+/+ and Pkbγ^-/- hippocampal neurons. Immunocytochemistry at day 28 using antibodies against tau (for axons) and Map2C (for dendrites) did not reveal any obvious defects in the differentiation of Pkbγ^-/- hippocampal neurons(Fig. 7A-D). Consistent with the analysis of the expression pattern of PKB isoforms in various brain regions (Fig. 2D), we found that PKBα, PKBβ and PKBγ, respectively, were expressed in cultured wild-type primary hippocampal neurons. In accordance with the result of Fig. 1D, we did not find a compensatory upregulation of PKBα and PKBβ in Pkbγ-deficient cells (Fig. 7E). To test the potential role of PKBγ in the survival of hippocampal neurons, cell cultures were challenged with glutamate (15 mM/24 hours) or staurosporine (50 nM/12 hours) after 7 days in culture. Apoptotic cells were detected using the TUNEL assay. In untreated cells cultures, no significant difference in the percentage of apoptotic cells between Pkbγ^+/+ and Pkbγ^-/-hippocampal neurons was observed (Fig. 7F). After treatment with glutamate or staurosporine, the percentage of apoptotic cells was significantly increased (51% and 24%,respectively, P<0.01) in cultures from Pkbγ^-/- hippocampal neurons(Fig. 7F). Additionally, we analysed the number of apoptotic cells from adult brains (n=5 per genotype) using TUNEL staining on parasagittal sections. Similar to the results of untreated cultures, we did not find any difference between Pkbγ^+/+ and Pkbγ^-/- mice(data not shown).

In the CNS, a considerable amount of cellular and physiological events occur during postnatal development, such as neuronal outgrowth,differentiation, synaptogenesis and maturation. To investigate a potential role of PKBγ in these events, we analyzed gene expression patterns in brains of Pkbγ^+/+ and Pkbγ^-/- mice at different time points during postnatal brain development, i.e. day 1, day 7 and day 30. RNA was extracted from whole mice brains and microarray analysis was performed on three individual brains per genotype and time point using murine Affymetrix GeneChips™. Changes below 1.5-fold increase were considered insignificant, and differentially regulated genes were subjected to a one-way ANOVA analysis (Table 2). As the brain consists of various heterogeneous cell populations, any change of expression occurring in a specific cell type will be quenched in the pool of whole brain RNA. It is therefore likely that the real expression changes in specific tissues are higher than apparent in this experiment. Among the 17,000 expressed genes, we found 37 genes to be differentially expressed in Pkbγ^-/- brains versus Pkbγ^+/+ brains. Of these, 16 genes were upregulated and 21 genes downregulated. Among the genes with changed expression are several transcription factors and genes implicated in cell cycle and proliferation. Significantly, one of the most interesting findings is that at P30, several genes involved in synaptic transmission (pre- and postsynaptic),including ionotropic glutamate receptor (NMDAR1), potassium-chloride co-transporter 2 (KCC2), chapsyin 110 [channel-associated protein of synapses,110 kDa (DLG2)] and synaptotagmin 2 (SYT2), have significantly reduced expression levels in Pkbγ mutant brains. Additionally,Ca²⁺/calmodulin-dependent protein kinase II α (CaMKII), an enzyme that is essential in synaptic plasticity and memory formation, was found at reduced expression levels (Ninan and Arancio, 2004).

Here, we report the phenotypic consequences of the ablation of the Pkbγ gene. Mice with targeted disruption of all single PKB isoform were generated and all demonstrated a distinct phenotype. Mice deficient in Pkbα are smaller with a 30% reduction in body size and partial neonatal lethality, which might be caused by placental insufficiency (Chen et al.,2001; Cho et al.,2001b; Yang et al.,2003). By contrast, Pkbβ knockout mice exhibited a diabetes-like syndrome with hyperglycemia and impaired insulin action in fat,liver and skeletal muscle and, depending on the genetic background, mild growth retardation (Cho et al., 2001; Garofalo et al., 2003). Recently, Peng and colleagues reported the generation of Pkbαβ double mutant mice with severe intrauterine growth retardation that die shortly after birth with multiple defects in skin, bone and fat tissue (Peng et al.,2003). However, our results demonstrate that Pkbγ-deficient mice display a distinct phenotype without increased mortality, growth retardation and altered glucose homeostasis. Inactivation of the Pkbγ gene leads to a considerable reduction in total phosphorylated/activated PKB in the mutant brain without any compensatory increase of the α and β isoforms (assessed at the mRNA and protein levels), suggesting a failure to fully compensate for the loss of PKBγ. This result is consistent with findings in Pkbα and Pkbβ mutant mice, where no compensatory upregulation of 2 remaining isoforms was found (Cho et al.,2001b; Garofalo et al.,2003; Yang et al.,2003). It has been speculated that the distinct phenotypes of Pkbα and Pkbβ mutant mice are due to specific and distinct functions of the different PKB isoforms. By contrast, Peng and colleagues proposed that the individual phenotypes are due to loss of the dominant isoform in a specific tissue, which leads to significant reduction of total activated PKB below a crucial level(Peng et al., 2003). For example, the observed diabetes mellitus-like syndrome of Pkbβ^-/- mice could be due to ablation of the dominant isoform in the classical insulin-responsive tissues such as fat, liver and muscle. Moreover, the loss of a second isoform would have an additive effect in reducing the level of total PKB and, therefore, would intensify the phenotype. This possibility is supported by the observation of Pkbαβ double knockout mice(Peng et al., 2003) and our unpublished data on the Pkbαγ double mutant mice (Z.-Z.Y. and B.A.H., unpublished). Ablation of a single copy of Pkbγ in Pkbα-deficient mice (Pkbα^-/-Pkbγ^+/- led to a higher perinatal mortality compared with Pkbα single mutant mice and the ablation of both Pkbγ alleles in Pkbα^-/- mice led to more pronounced dwarfism and intra-uterine death of all Pkbα^-/-Pkbγ^-/- double mutant animals. However, it cannot be confirmed yet whether the observed phenotypes are due to a combination of reduced level of activated PKB and the loss of isoform-specific functions.

The IGF1/PI3K/PKB pathway plays a crucial role in mammalian brain development and function (D'Ercole et al.,2002; Rodgers and Theibert,2002). Besides the severe growth retardation, adult mice with IGF1 deficiency exhibit a significant (38%) brain weight reduction(Beck et al., 1995). Similar to the Pkbγ mutant mice, all brain parts of the Igf1^-/- mice were affected but the general anatomical organisation was normal. Furthermore, ablation of IGF1 resulted in a cell type-dependent loss of neurons, as well as a reduced total number of oligodendrocytes and hypomyelination (Beck et al., 1995; Ye et al.,2002). In addition, targeted deletion of IRS2 in mice also produced a pronounced brain growth deficiency, but in contrast to Pkbγ mutants, the reduction was already apparent during embryonic (E15.5) development (Schubert et al., 2003). By contrast, an increased brain mass was observed in mice overexpressing IGF1 (Mathews et al.,1988; Ye et al.,1995). Moreover, mice with brain-specific deletion of PTEN, a negative regulator of the PI3K/PKB pathway, exhibited an enlarged brain with seizures and ataxia resembling Lhermitte-Duclos disease(Backman et al., 2001; Kwon et al., 2001). Less is known about the consequences of Pkbα and Pkbβinactivation for mouse brain development. Compared with Pkbγ^-/- mice, adult Pkbα and Pkbβ mutant mice showed only a slight decrease in brain weight(Garofalo et al., 2003; Yang et al., 2004). In both Pkbα and Pkbβ mutant mice, no changes in the gross brain morphology were reported.

However, inactivation of the Pkbγ gene resulted in a significant reduction of brain weight and size. Interestingly, Pkbγ deficiency did not affect the general anatomical organization of the brain. In vivo 3D MRI and histological analysis excluded the absence of a specific brain region as the main cause of the weight reduction, consistent with the result of the broad expression profile among brain regions. More specifically, the proportionally reduced ventricular system rules out major disturbances in production, circulation and absorption of cerebrospinal fluid as a cause of reduced cell size/number in Pkbγ^-/- mice. The biological relevance of the MRI signal alterations in white matter such as the corpus callosum requires further investigation. Nevertheless, it should be noted that the in vivo MRI results are consistent with a pronounced, but not complete, deficit in myelin deposition (Boretius, 2003),which is also strongly supported by the histological findings for myelin staining. In agreement with our results, mice deficient in IGF1, a potent activator of PKBγ, the myelin-rich white matter regions, including corpus callosum and anterior commissure, were overproportionally reduced by about 70% (Beck et al., 1995). By contrast, mice overexpressing IGF1 display an increased brain weight and the corpus callosum of the Igf1 transgenic mice was increased in excess of proportionality (Carson et al.,1993).

The PI3-K/PKB signalling pathway plays a crucial role in the determination of cell size (Scanga et al.,2000; Shioi et al.,2002; Tuttle et al.,2001; Verdu et al.,1999). Results from transgenic mice overexpressing PKB show larger cardiac myocytes and thymocytes, or hypertrophy and hyperplasia in the pancreas (Kovacic et al.,2003; Mangi et al.,2003). An increase in neuronal soma size was observed in mice with brain-specific deletion of PTEN (Backman et al., 2001; Kwon et al.,2001). By contrast, the size of skeletal muscle cells in Pkbαβ double mutant mice was dramatically reduced(Peng et al., 2003). Our results show that both cell number and cell size are affected, but that reduced cell size contributes more than the reduced cell number. Additionally,it has been shown that the mTOR signalling pathway is also involved in the determination of cell size (Montagne et al., 1999; Oldham et al.,2000; Zhang et al.,2000). PKB modulates mTOR activity by phosphorylating TSC2, with a subsequent disruption of the TSC1-TSC2 interaction(Inoki et al., 2002; Potter et al., 2003).

Recent publications have linked PI3-K/PKB with synaptic plasticity and memory (Dash et al., 2004; Kelly and Lynch, 2000; Lin et al., 2001; Robles et al., 2003; Sanna et al., 2002). Wang and colleagues demonstrated that the A-type γ-aminobutyric acid receptors(GABA_AR), which mediate fast inhibitory synaptic transmission, is phosphorylated by PKB (Wang et al.,2003). Phosphorylation of GABA_AR leads to an increased number of receptor on the cell membrane and an increased synaptic transmission. Additionally, Lin et al. established a role of the PI3-K/PKB pathway in fear conditioning in the amygdala(Lin et al., 2001). As we found several genes involved in neuronal circuit activity in our microarray experiment, future behavioural and electrophysiological studies of Pkbγ mutant brains will elucidate the specific role of PKBγ in synaptic transmission, learning and memory.

In summary, we have demonstrated that Pkbγ-deficient mice display a phenotype distinct from Pkbα and Pkbβmutant mice. Our results provide novel insights into the physiological function of PKBγ and suggest a crucial role in postnatal brain development of mammals. Identification of PKBγ specific substrates involved in postnatal brain development is now of critical importance.

The authors thank Michael Leitges (Max-Planck-Institut für Experimentelle Endokrinologie, Hannover, Germany) for providing the IRES/lacZ/Neo targeting cassette. From the FMI we acknowledge J. F. Spetz and P. Kopp for ES cell culture and generation of chimera, D. Hynx for animal care, H. Brinkhaus for help with neuronal cell culture, E. Oakeley and H. Angliker for microarray analysis, M. Sticker for helpful advice in histology and R. Portmann for advice on Western blot quantification. O.T. is supported by the Novartis Stiftung für Medizinisch-Biologische Forschung,Z.Z.Y. is funded in part by a grant from Bundesamt für Bildung und Wissenschaft (BBW #98.0176) and B.D. by the Swiss Cancer League (KFS 1167-09-2001 and KFS 01002-02-2000). The Friedrich Miescher Institute for Biomedical Research is funded by the Novartis Research Foundation.