In developing mammalian telencephalon, the loss of adherens junctions and cell cycle exit represent crucial steps in the differentiation of neuroepithelial cells into neurons, but the relationship between these cellular events remains obscure. Atypical protein kinase C (aPKC) is known to contribute to junction formation in epithelial cells and to cell fate determination for Drosophila neuroblasts. To elucidate the functions of aPKCλ, one out of two aPKC members, in mouse neocortical neurogenesis, a Nestin-Cre mediated conditional gene targeting system was employed. In conditional aPKCλ knockout mice, neuroepithelial cells of the neocortical region lost aPKCλ protein at embryonic day 15 and demonstrated a loss of adherens junctions, retraction of apical processes and impaired interkinetic nuclear migration that resulted in disordered neuroepithelial tissue architecture. These results are evidence that aPKCλ is indispensable for the maintenance of adherens junctions and may function in the regulation of adherens junction integrity upon differentiation of neuroepithelial cells into neurons. In spite of the loss of adherens junctions in the neuroepithelium of conditional aPKCλ knockout mice, neurons were produced at a normal rate. Therefore, we concluded that, at least in the later stages of neurogenesis, regulation of cell cycle exit is independent of adherens junctions.
In the central nervous system of mammals, neuroepithelial cells serve as neural progenitors, differentiating into either neurons or glial cells(Ever and Gaiano, 2005; Gotz and Huttner, 2005; Haubensak et al., 2004; Miyata et al., 2001; Noctor et al., 2001). In this paper, we use the term `neuroepithelial cells' to indicate both actual neuroepithelial cells and radial glial cells. In mouse neocortex, neurogenesis precedes gliogenesis and commences around embryonic day 11 (E11), peaks around E15 and finishes around birth (Jacobsen,1991). Newly generated neurons migrate towards the pial side,forming the six-layered neocortex (Gupta et al., 2002; Marin and Rubenstein, 2003). Neuroepithelial cells form a well-packed cell layer called the ventricular zone and line the lateral ventricles. Although the ventricular zone looks like multiple cell layers, most cells face both the ventricular/apical and pial/basal surfaces of this cell layer with processes extending from the cell body. Basal side cellular processes are elongated towards the pial surface of the telencephalic vesicles, serving as a radial scaffold for the migration of newly generated neurons. However, apical side cellular processes are elongated towards the ventricular surface to form adherens junctions (Aaku-Saraste et al.,1996; Astrom and Webster,1991).
Once neurogenesis starts, neuroepithelial cells undergo asymmetric cell division producing daughter cells with a different cell fate; one keeps the characters of neuroepithelial cells, while the other differentiates into a post-mitotic neuron or an intermediate progenitor, which produces two neurons after the next cell division in the subventricular zone(Haubensak et al., 2004; Miyata et al., 2004; Noctor et al., 2004). Importantly, post-mitotic neurons and intermediate progenitors lose adherens junctions and move out from the ventricular zone. Thus, the loss of either adherens junctions or the apical domain defined by those junctions could represent a cue for the differentiation of neuroepithelial cells(Kosodo et al., 2004; Wodarz and Huttner, 2003). In support of this notion, asymmetric inheritance of the components of adherens junctions is occasionally observed with cells dividing at the apical surface of the ventricular zone (Chenn et al.,1998; Manabe et al.,2002). However, neither the importance of adherens junctions to differentiation nor the molecular mechanisms for differentiation-dependent regulation of the integrity of adherens junctions in neuroepithelial cells has been elucidated.
A subgroup of protein kinase C (PKC), atypical PKC (aPKC) regulates cell polarity in several organisms (Macara,2004; Ohno, 2001). aPKC forms a complex with cell polarity proteins PAR3 and PAR6, localizes predominantly at tight junctions in mammalian epithelial cells, and its kinase activity is required for the establishment of apicobasal cell polarity and the formation of tight junctions (Hirose et al., 2002; Suzuki et al.,2001; Yamanaka et al.,2001). Drosophila genetic studies have shown that aPKC,along with PAR3 and PAR6 are responsible for junction formation in epithelial cells and for cell fate determination in neuroblasts(Kuchinke et al., 1998; Petronczki and Knoblich, 2001; Rolls et al., 2003; Wodarz et al., 2000; Wodarz et al., 1999). In mammalian neural tissues, aPKC localizes with PAR3 and PAR6 at the adherens junctions of embryonic neuroepithelial cells(Manabe et al., 2002). However, the role of aPKC during mammalian neurogenesis remains unknown.
To explore the role of aPKC in the development of mouse neocortex, we inactivated the gene for aPKCλ, one of two aPKC isotypes, by employing a Nestin promoter and an intronic enhancer-driven Cre-mediated conditional gene targeting system. The phenotype of the mutant mice indicates that aPKCλ is indispensable for neuroepithelial cells to form adherens junctions and maintain cell polarity in the neuroepithelium. However,aPKCλ is not required for cell cycle exit and the subsequent radial migration in developing neocortex.
MATERIALS AND METHODS
Mice harboring a floxed aPKCλ gene in which exon 5 was flanked by loxP sequences were generated by homologous recombination(Hashimoto et al., 2005; Koike et al., 2005; Matsumoto et al., 2003). Nestin-Cre mice (T.S., unpublished) harboring tandem arrays of the Cre gene driven by a rat nestin promoter (Zimmerman et al., 1994), the internal ribosome entry site (IRES) sequence,the lacZ gene and a nestin neuron-specific enhancer(Zimmerman et al., 1994).
Detection of β-galactosidase activity
Any β-galactosidase expressed from Rosa26R gene that lost the suppressor neo-cassette following cre-mediated recombination was detected in frozen sections fixed using 4% PFA for 10 minutes by staining with X-gal according to the standard protocol. The Nestin-Cre transgene bears the lacZ gene; however, the expression of β-galactosidase from this lacZ gene was negligible compared with that from the Rosa26Rgene (data not shown).
Western blot analysis
Telencephalic vesicles from E13.5 or E15.5 embryos were cut out in ice-cold PBS, the meninges removed and vesicles homogenized in 1 ml of SDS-PAGE sample buffer. Samples were appropriately diluted to provide equal protein amounts and used for SDS-PAGE. Western blot analysis was performed according to standard protocols using the following antibodies: anti-aPKCι antibody 1/1000 (clone 23, BD); anti-aPKCζ antibody 1/1000 (rabbit, Santa Cruz);and anti-β-actin 1/2000 (AC-15, Sigma). For secondary antibodies,horseradish peroxidase was conjugated with anti-rabbit or - mouse IgG 1/2000(goat, Amersham). Enzyme activity was detected using an ECL system (Amersham)and luminescence was quantified using a FUJI Las 3000 luminescence image analyzer (Fuji Photo Film, Tokyo). All images were arranged and labeled using Photoshop 7.0 (Adobe Systems).
Embryos were fixed with 4% paraformaldehyde (PFA)/phosphate-buffered saline(PBS) for paraffin wax-embedded sectioning (5-6 μm). When required, BrdU was injected intraperitoneally into pregnant mice at the time points indicated in the text. Paraffin sections were hydrolyzed and heated at 120°C for 20 minutes in 10 mM sodium citrate (pH 6.0). For staining with MAP2, CSPG neurogenin 2 antibody, embryos were frozen directly in OCT compound and sectioned at 8 μm. Frozen sections were then fixed with a methanol/acetone 1:1 mix at -20°C for 10 minutes and air-dried. Immunostaining was performed according to standard protocols using 10% normal goat serum in TBST[10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05% Tween 20] as a blocking reagent,and primary and secondary antibodies were diluted in 0.1% BSA/1.5% normal goat serum/TBST, as follows. Primary antibodies: affinity purified anti-aPKCλ rabbit antibody 1/1000 (λ3)(Suzuki et al., 2001);anti-N-cadherin 1/1000 (clone 32, BD); anti-β-catenin 1/1000 (clone 14,BD); anti ZO-1 1/1000 (clone ZO1-1A12, Zymed); anti-PAR-6β 1/500 (BC32AP)(Yamanaka et al., 2003);anti-ASIP/PAR-3 1/1000 (C2-3) (Hirose et al., 2002); anti-γ-tubulin 1/500 (rabbit, Sigma); anti-BrdU 1/1000 (clone 2B1, MBL); anti-phospho-histone H3 1/1000 (rabbit, Upstate);anti-βIII-tubulin 1/2000 (clone TuJ1, Babco); anti-Ki67 1/200 (rabbit,Novocastra); anti-CSPG 1/1000 (clone CS-56, Sigma); anti-MAP2 (clone HM-1;Sigma) and anti-neurogenin 2 1/50 (clone 5C6)(Lo et al., 2002). Secondary antibodies: Cy3-conjugated anti-rabbit or -mouse IgG 1/2000 (goat, Amersham);Alexa488-conjugated anti-rabbit or -mouse IgG (goat, Molecular Probes); and TOPRO-3 1/100 (Molecular probes). DAPI (2.5 μg/ml, Sigma) was included in the final wash buffer (TBST) for nuclear staining. For immunohistochemistry,biotinylated secondary anti-mouse IgG antibodies 1/1000 (goat, Vector) were detected using a Vectastain Elite ABC kit (Vector). Images were captured using a BX50 fluorescent microscope (Olympus) equipped with a CCD camera(Photometrics) or an LSM510 (Zeiss). All images were arranged and labeled using Photoshop 7.0 (Adobe Systems). Some paraffin sections were stained using Carazzi's hematoxylin (Muto Pure Chemicals) and Eosin B (Sigma) or with 0.5%Cresyl Violet.
Samples were fixed in 1% glutaraldehyde/4% PFA in 0.1 M sodium cacodylate buffer overnight at 4°C, post-fixed with 2% osmium in 0.1 M sodium cacodylate buffer for 2 hours at 4°C, and processed for Epon embedding. Sections were examined at 75 kV using an H-7500 transmission electron microscope (Hitachi).
Slice culture analysis
Coronal slices of E15.5 embryonic telencephalon (200-300 μm) were prepared and cultured in collagen gel, as previously described(Miyata et al., 2004). To examine cell shapes, dissected telencephalon was immersed for 10 seconds in DiI suspension (∼1 mg/ml culture medium), DiI was deposited on the pial surface prior to making the slices and the dye was left to diffuse for more than 2 hours. Time-lapse photographs were taken by hand.
Nestin-Cre-mediated conditional disruption of the aPKCλ gene
Homozygous aPKCλ allele lacking exon 9(PKCλΔ9/Δ9)mice show lethality at an early embryonic stage(Soloff et al., 2004). To assess the function of aPKCλ in neocortical neurogenesis, we generated Nestin-Cre-mediated conditional aPKCλ knockout mice (aPKCλ cKO). In the conditional aPKCλ allele(PKCλfloxed), exon 5 of the aPKCλ gene is flanked by loxP sequences. As exon 5 is 86 bp long, deletion results in a frame-shift mutation upstream of the kinase domain-coding sequence (Hashimoto et al.,2005; Koike et al.,2005; Matsumoto et al.,2003). This conditional allele can thus be converted into a strong loss-of-function allele(PKCλΔ5) by Cre recombinase-mediated recombination. With the Nestin-Cre allele, Cre recombinase express under the control of the rat Nestin promoter and intronic enhancer (T.S., unpublished). To generate aPKCλ cKOs carrying the aPKCλ floxed allele in homo and Nestin-Cre allele in hetero (PKCλfloxed/floxed; Nestin-Cre), we bred mice carrying the aPKCλ floxed allele in homo(PKCλfloxed/floxed) and mice carrying the aPKCλ floxed allele in hetero and Nestin-Cre allele in hetero (PKCλfloxed/+; Nestin-Cre). PKCλfloxed/+; Nestin-Cre mice found as littermates of the aPKCλ cKO mice are viable, fertile and show no obvious abnormalities, similar to the heterozygous aPKCλ mutant(PKCλΔ5/+) mice (data not shown) that were used as controls for most of the experiments presented in this manuscript. Genotypes and numbers of newborn pups obtained from this breeding program were 21 for aPKCλ cKO(PKCλfloxed/floxed; Nestin-Cre),13 for PKCλfloxed/floxed, 25 for PKCλ floxed/+; Nestin-Cre and 19 for PKCλ floxed/+, most fitting the expected ratio of 1:1:1:1. Although for some unknown reason the number of PKCλ floxed/floxed pups was somewhat lower than the expected value, comparable numbers of aPKCλ cKO (PKCλ floxed/floxed; Nestin-Cre) and control(PKCλfloxed/+; Nestin-Cre) pups were obtained; indicating that the most aPKCλ cKO mice can survive during embryonic and early postnatal days. These aPKCλ cKO mice were healthy and indistinguishable from littermates up to 5 days after birth, although after this point growth retardation started to be evident. Thereafter, all aPKCλ cKO mice died within 1 month of birth and all displayed hydrocephalus (21/21, 100%).
To test the tissue specificity of the Nestin-Cre-mediated recombination,Nestin-Cre transgenic mice were crossed with Rosa26R reporter mice in which a floxed neo cassette interrupts the ubiquitous expression of the lacZ gene (Soriano,1999). In Rosa26R mouse embryos carrying the Nestin-Cre transgene,Cre-mediated recombination monitored by the expression of β-galactosidase activity was specifically detected in the central nervous system of E15.5 embryos, including in the telencephalon, diencephalon, mesencephalon,metencephalon, myeloncephalon and spinal cord. In the telencephalon,recombination was detected in the ganglionic eminence and rostral neocortial region, but no recombination was evident at this stage in the caudal neocortical region, including the hippocampus(Fig. 1A).
We then tested if aPKCλ protein was abolished in embryonic telencephalon of aPKCλ cKO mice. Western blot analysis using a specific antibody against aPKCλ demonstrated a substantial reduction of aPKCλ protein from telencephalic vesicles at E15.5, while the protein was still present at E13.5 (Fig. 1B, upper). Remnant aPKCλ protein would have originated from the caudal region of the telencephalon, where Cre-mediated recombination did not proceed even at E15.5 (Fig. 1A). As the activity of Cre recombinase monitored by lacZexpression from the modified Rosa locus had been detected in a large part of E13.5 brain (data not shown), the subtle decrease in the amount of aPKCλ protein at E13.5 might indicate the relatively low susceptibility of the aPKCλ locus to Cre recombinase or the long half-life of aPKCλ protein.
As previously reported, aPKCλ protein was highly concentrated in the apical ridge of the ventricular zone, predominantly in the adherens junctions of neuroepithelial cells (Manabe et al.,2002). When localization of aPKCλ was examined under immunofluorescent microscopy, aPKCλ protein was undetectable in the anterior neocortical region of aPKCλ cKO mouse embryos at E15.5, while apical localization of this protein was obvious in control embryos(Fig. 5A,B).
In mammals, aPKC comprises two closely related isotypes, aPKCλ and aPKCζ (Suzuki et al.,2003). Although aPKCλ comprises a single protein with a molecular mass of ∼70 kDa, aPKCζ has two variants, at 70 kDa and 55 kDa. The 55 kDa variant is called PKMζ and lacks an N-terminal regulatory domain including PB1, a pseudosubstrate and a cysteine-rich domain(Hernandez et al., 2003; Hirai et al., 2003). Contrasting with aPKCζ and aPKCλ, which are found in a variety of tissues, PKMζ is expressed predominantly in the brain(Akimoto et al., 1994;Hernandes et al., 2003; Hirai et al.,2003). To test protein levels of aPKCζ variants in the aPKCλ cKO telencephalon, Western blot analysis was employed using antibody recognizing all aPKCs (aPKCλ, aPKCζ and PKMζ). A 70 kDa protein detected in controls was greatly reduced in aPKCλ cKO embryos at E15.5 (Fig. 1B,middle), indicating that the expression level of aPKCζ in the telencephalon is rather low in aPKCλ cKO embryos, and probably also in control embryos. PKMζ was clearly detected in E15.5 control telencephalon, and protein levels remained unchanged in aPKCλ cKO telencephalon. These results indicate that in embryonic telencephalon,aPKCλ and PKMζ are mainly expressed, and that the loss of aPKCλ does not affect protein levels of aPKCζ variants.
Loss of aPKCλ causes disruption of neuroepithelial tissue architecture in the telencephalon
To examine the effects of aPKCλ disruption on the gross morphology of the neocortical region, histological analysis was initially performed under light microscopy. At E13.5, when the expression of aPKCλ protein was not severely impaired, no morphological abnormalities were observed in the neocortical regions of aPKCλ cKO embryos (data not shown). Abnormalities became evident at E15.5, when aPKCλ protein in this region decreased to undetectable levels. In control embryos, the ventricular zone comprised uniformly packed neuroepithelial cells and had a smooth ventricular surface (Fig. 2A). By contrast, the ventricular surface in aPKCλ cKO embryos was rough and neuroepithelial cells were loosely packed. The ventricular zone was thus barely distinguishable from the subventricular zone(Fig. 2E). Nevertheless, the subplate and cortical plate were comparable with these layers in control embryos, and little abnormality was apparent in the ventricular zone at the caudal neocortical region where apical localization of aPKCλ protein was maintained (Fig. 5E, data not shown). Morphological abnormalities became more evident at E16.5. The ventricular and subventricular zones were severely disorganized, and the ventricular, subventricular and intermediate zones were difficult to distinguish (Fig. 2C,G). Gross death of neuroepithelial cells was considered unlikely as the primary cause of this morphological abnormality, as no significant increase in the number of cells undergoing apoptosis was observed (see Fig. S1 in the supplementary material) and the total number of proliferating cells as assessed by the number of cells in S-phase was not significantly changed in the neocortical region of aPKCλ cKO embryos (Fig. 3C). Instead, a dispersion of neuroepithelial cells into the outer layers, and the subventricular and intermediate zones might have caused these abnormalities. This possibility is supported by the unusual distribution of TuJ1-positive differentiated neurons. In control embryos, TuJ1-positive cells were for the most part excluded from the ventricular and subventricular zones(Menezes and Luskin, 1994). However, in aPKCλ cKO embryos, TuJ1 staining was found in all layers of the neocortical region (Fig. 2H). To identify the somata of TuJ1-positive differentiated neurons, we stained sections with an antibody for the proliferation marker Ki67 and TuJ1. In such staining, Ki67-negative nuclei surrounded by TuJ1 staining represented somata of differentiated neurons found in the most apical region in aPKCλ cKO embryos (Fig. 2H′). In contrast to the severe disorganization of ventricular-side layers of the neocortical region, pial-side layers comprising the subplate and cortical plate were unaffected, and staining with chondroitin sulfate proteoglycans and MAP2, as markers for the subplate and cortical plate, respectively, yielded a pattern indistinguishable from that in control embryos (Fig. 2C,G; see Fig. S2 in the supplementary material). The abnormality was more prominent in the ganglionic eminence, where the loss of aPKCλ took place earlier than in the neocortical region. Typically, fusion of the right and left medial ganglionic eminences was observed at E13.5, and at E16.5 irregular protrusions formed at the ventricular surface and the anterior horn of the lateral ventricle was filled with neuroepithelial cells and neurons(Fig. 2F, see Fig. S3 in the supplementary material). These results suggest that aPKCλ is indispensable for neuroepithelial cells to form pseudostratified epithelium in the ventricular zone.
To confirm this possibility, we examined whether the typical characteristics of pseudostratified epithelium were maintained in aPKCλcKO embryos. One such characteristic is interkinetic nuclear migration, a cell cycle-dependent translocation of the nucleus along the radial axis of the ventricular zone. Upon cell cycle progression, nuclei of S-phase cells located deep (pial side) in the ventricular zone move towards the ventricular surface,where cells undergo mitosis (Jacobsen,1991). To examine the position of S- and G2-phase cell nuclei in the neocortical region, labeling was performed by intraperitoneal injection of BrdU into pregnant mice 2 hours prior to embryo dissection(Fig. 3A,B). No significant differences were noted between control and aPKCλ cKO neocortical regions with regard to the ratio of BrdU-positive nuclei to total nuclei in the defined area (Fig. 3C). In addition, most BrdU-positive nuclei were restricted to within the ventricular zone and subventricular zone in both cases(Fig. 3D). However, when positions of BrdU-positive nuclei within the ventricular zone and subventricular zone were compared, differences were evident. In control embryos, BrdU-positive nuclei were concentrated in the middle layer of the ventricular zone (layer 2 in Fig. 3A,D), but were dispersed throughout the ventricular zone and subventricular zone in aPKCλ cKO embryos, showing a slight tendency to concentrate on the ventricle side (layer 3, Fig. 3B,D). Mitotic cell nuclei were then labeled by immunostaining using the phospho-histone H3 antibody,with the majority being located at the ventricular surface in control embryos,while a relatively minor population was found in the non-surface area of the ventricular zone and subventricular zone and the intermediate zone(Miyata et al., 2004; Noctor et al., 2004). By contrast, only a minor portion of the mitotic cells were found on the ventricular surface in aPKCλ cKO embryos, with the majority found in non-surface areas of the ventricular zone and subventricular zone(Fig. 3E). These results indicate that proper interkinetic nuclear migration of neuroepithelial cells is impaired by the loss of aPKCλ; thereby, each cell is at a random position in the cell cycle.
Another common feature of cells in pseudostratified epithelium is the presence of basal and apical cellular processes extending to the pial and ventricular surfaces, respectively (Astrom and Webster, 1991; Miyata et al., 2001). Although the length of each process changes depending on the interkinetic nuclear migration, centrosomes are always anchored at the tip of the apical process, except in mitotic round-up cells, so centrosomes are apically localized in pseudostratified epithelium(Astrom and Webster, 1991). In aPKCλ cKO embryos, centrosomes were dispersed throughout the ventricular zone and subventricular zone with minor apical localization,indicating a considerable degree of cell-shape changes(Fig. 4A). When neuroepithelial cells in control embryos were labeled applying DiI from the pial surface, most cells displayed cellular processes extending to both the pial and ventricular surfaces (left panels in Fig. 4B,C). In aPKCλ cKO embryos, most apical cellular processes were shortened and detached from the ventricular surface (right panels in Fig. 4B,C).
Time lapse observations in aPKCλ cKO slice cultures revealed that apical cellular processes gradually retracted after detaching from the ventricular surface (Fig. 4C). Shortened apical cellular processes represent the normal characteristic of some fractions, such as neuroepithelial and intermediate progenitors that will divide at the subventricular zone (Miyata et al., 2004; Noctor et al.,2004). However, in aPKCλ cKO embryos, the majority of neuroepithelial cells (55 out of 60 cells observed) displayed apical processes detached from the ventricular surface. The retraction of apical processes may also occur upon the production of intermediate progenitors(Haubensak et al., 2004; Miyata et al., 2004; Noctor et al., 2004). Therefore, the vast increase in the production of intermediate progenitor could also have the same result. The increase in the number of intermediate progenitor may results in the increase in the number of neurogenin 2-positive cells (Miyata et al., 2004). However, it did not change significantly (see Fig. 7). Therefore, this apical process phenotype cannot be explained by the over-production of intermediate progenitor. Taken together, these results indicate that aPKCλ protein is essential for the maintenance of apical cellular processes.
Disruption of aPKCλ results in the disappearance of neuroepithelial adherens junctions
The apical tips of neuroepithelial cells are interconnected at the ventricular surface by adherens junctions, in which aPKCλ, N-cadherin,β-catenin and ZO-1 are localized(Chenn and Walsh, 2002; Manabe et al., 2002). Loss of aPKCλ protein presumably causes abnormalities of the adherens junctions, together with the abnormalities in the neuroepithelium as described above. To test this possibility, formation of adherens junctions in the neuroepithelium of control and aPKCλ cKO embryos was assessed. We first examined the localization of β-catenin, which was highly concentrated in dot-like structures located on the ventricular surface of control embryos(Fig. 5A,C). In aPKCλcKO embryos, such localization of β-catenin was rarely seen on the ventricular surface, even though weak staining around cell bodies was maintained (Fig. 5B,D). Localization of other components of adherens junction, such as N-cadherin and the aPKC-binding polarity proteins PAR3 and PAR6β, was also severely impaired in aPKCλ cKO embryos (see Fig. S4 in the supplementary material). These results clearly indicate the loss of adherens junctions in the neuroepithelium of aPKCλ cKO embryos. We then used electron microscopy to confirm the disappearance of adherens junctions. Typical adherens junctions were not identified at either the apical surface or the deeper area of the ventricular zone and subventricular zone in the rostral neocortical region of aPKCλ cKO embryos(Fig. 5F,H), but were always found as continuous electron dense lines at the apical ridge of neuroepithelium in the caudal neocortical region(Fig. 5E,G), where aPKCλprotein was still present at E15.5. Fragmented adherens junctions were occasionally observed on the ventricular surface in aPKCλ cKO embryos(Fig. 5F,H, arrow), possibly corresponding to the remnant dot-like structures stained with β-catenin(Fig. 5D, arrows). Interestingly, these structures were not stained with aPKCλ antibody(Fig. 5B). Loss of aPKCλprotein may thus precede the disappearance of adherens junctions;nevertheless, no continuous intact adherens junctions were found without the association of aPKCλ, indicating that the integrity of adherens junctions depends almost totally on the presence of aPKCλ. Loss of adherens junctions is most probably a cell-autonomous effect of aPKCλloss, as adherens junctions were unaffected in a caudal region of neocortex at E15.5, where aPKCλ protein was still present, even though this region was adjacent to the more rostral affected region.
Disruption of aPKCλ does not affect neurogenesis and radial migration
Previous studies have shown that aPKC is involved in the cell fate determination of neuroblasts in the Drosophila central nervous system(Wodarz et al., 2000, Rolls et al., 2003). In mammals, loss of adherens junctions presumably causes the loss of polarity cue in neuroepithelial cells, and, depending on asymmetric cell division, affects neurogenesis (Wodarz and Huttner et al., 2003). In the neocortex of aPKCλ cKO mice at postnatal day (P) 3, lateral ventricles were swollen as in the hydrocephalus, but no serious abnormalities in the laminated structure of the neocortex were found (Fig. 6B). The apparently normal structure of the neocortex at P3 suggests that neurogenesis and the radial migration that follows are unaffected by the loss of aPKCλ. To further test this possibility,neurons produced at around E15.5 or E17.5 were labeled with BrdU, and the position of these neurons in the neocortex was examined at P3. In both control and aPKCλ cKO neocortex, neurons labeled at E17.5 were located in the outermost layer of the neocortex adjacent to the marginal zone (layers II or III of mature neocortex) at P3, and neurons labeled at E15.5 were located slightly closer (Fig. 6C,D). These results indicate that radial migration was unaffected in aPKCλcKO mice; moreover, no obvious differences were identified in the numbers of labeled cells, possibly reflecting the rate of neurogenesis at E15.5 or E17.5. However, directly comparing numbers of labeled neurons was difficult because of the enlarged shape of the aPKCλ cKO mouse telencephalon. To clarify this point, the rate of neurogenesis was monitored by measuring the cell cycle exit rate (Chenn and Walsh,2002), as follows. BrdU was loaded at E15.5 and brains were fixed at E16.5 for immunostaining using antibodies against BrdU and Ki67, a protein marker for proliferating cells. Ki67-negative cells in the BrdU-positive cell cohort thus correspond to cells that have exited the cell cycle and differentiated in the period between E15.5 and E16.5(Fig. 7A,B). The ratio of Ki67-negative/BrdU-positive cells to all BrdU-positive cells was about 50% in the neocortical region of both control and aPKCλ cKO embryos(Fig. 7C). We then compared the expression of neurogenin 2, an early differentiation marker, which is essential for the commitment to neuronal linage(Kageyama et al., 2005; Ross et al., 2003). Again, no significant changes in the ratio of neurogenin 2-positive nuclei were observed among the total number of nuclei in the ventricular zone and subventricular zone of E15.5 aPKCλ cKO and control embryos(Fig. 7D-F). Notably, the number of GFAP-positive glial cells in the neocortex at E15.5, E17.5 and P3 remained unchanged with the loss of aPKCλ (data not shown). Moreover,neuroepithelial cells from the neocortical region of aPKCλ cKO embryos at E15.5 or E16.5 proliferated in suspension culture forming neurospheres, and these cells retained the potential to differentiate into neurons and glia(data not shown). Taken together, these results indicate that aPKCλ is not essential for the neurogenesis of neocortical development, at least in the later stages after E15.5.
We reported here in that conditional disruption of the aPKCλ gene in mouse neocortex causes the loss of adherens junctions in neuroepithelial cells. Adherens junctions were unaffected until E14.5 in the neocortical region of aPKCλ cKO embryos, as aPKCλprotein was still present until this time, supporting the proposition that the loss of adherens junctions is the cell-autonomous effect caused by the disappearance of aPKCλ. Because a zebrafish mutant heart and soul (has), in which aPKCλ is disrupted at the C-terminal region, has been reported to cause the disappearance of apical ZO-1 or F-actin staining in the retina, the role of aPKC for adherens junctions seems to be conserved (Horne-Badovinac et al.,2001; Peterson et al.,2001). Loss of adherens junctions is accompanied by gross abnormalities in the neuroepithelial tissue architecture, including the invasion of differentiated neurons, retraction of apical processes and impaired interkinetic nuclear migration. Similar abnormalities in neuroepithelium have been observed by blocking either the function or expression of the adherens junction components N-cadherin and β-catenin,respectively, so the loss of adherens junctions is likely to represent the cause of these abnormalities(Ganzler-Odenthal and Redies,1998; Machon et al.,2003; Masai et al.,2003). Importantly, cell cycle exit and the radial migration that follows seem not to be affected by the conditional disruption of the aPKCλ gene in mouse neocortex after E15.5, providing a clear contrast to the previously reported mutants.
Adherens junctions is not required for mammalian neurogenesis
The loss of adherens junctions is a natural cellular event observed when neuroepithelial cells differentiate into neurons or a certain cohort of progenitor, intermediate progenitor. Recent reports have described that adherens junction-free progenitors dividing at the subventricular zone or intermediate zone undergo symmetric cell divisions producing two neurons in the most cases (Miyata et al.,2004; Noctor et al.,2004). Adherens junctions play a crucial role in the maintenance of proliferative status in a variety of stem cells, including Drosophila germinal cells, mouse hematopoietic stem cells and epidermal cells by maintaining the stem cell niche(Fuchs et al., 2004; Song et al., 2002; Yamashita et al., 2003). For example, expression levels of adherens junction components are important for generating hair follicles and the inactivation of α-catenin impairs hair follicle morphogenesis and adherens junctions(Jamora et al., 2003; Vasioukhin et al., 2001). This has been predicted to remain true for neuroepithelial cells, the neural stem cells. However, if adherens junctions play a crucial role in the proliferative status in neuroepithelial cells, loss of adherens junctions may cause temporal overproduction of neurons, resulting in a considerable decrease in neuroepithelial cell numbers. In fact, loss of adherens junctions has also been reported in conditional β-catenin knockout and Lgl1knockout mice (Klezovitch et al.,2004; Machon et al.,2003). In such cases, decreases or increases in cell proliferation rate have been reported in addition to the disorganization of the neuroepithelium in the neocortical region or ganglionic eminence. However, the present results indicate that adherens junctions are not required for either maintenance of proliferative status or cell cycle exit in the mouse neocortex,at least at the late stage of neurogenesis. This is supported by three observations. First, the total number of BrdU-labeled S-phase cells was not altered in neocortical regions where adherens junctions were lost. Second, the cell cycle exit rate reflecting the rate of neural cell differentiation and the neurogenin 2-positive cell rate were also unaffected. Third, neurons produced 2 days after the loss of adherens junctions (E17.5) were found in normal positions in the laminated neocortex after birth. These different effects of the loss of adherens junctions on cell proliferation may reflect the function of β-catenin and Lgl1, rather than the maintenance of the adherens junction structure, as β-catenin acts as a component of the transcription factor required for induction of Wnt signaling-dependent cell proliferation (Chenn and Walsh,2002), and Lgl is characterized as a tumor suppressor gene(Bilder et al., 2000). Adherens junctions were unaffected until E14.5 in the neocortical region of aPKCλ cKO embryos, as aPKCλ protein was still present until this time. Thus, adherens junctions might play essential roles in the maintenance of proliferative status or cell cycle exit of neuroepithelial cells at stages earlier than E14.5.
aPKCλ regulates integrity of adherens junctions
Although the presence of adherens junctions does not affect neurogenesis as discussed above, the integrity of adherens junction may be regulated by a mechanism that accords to cell fate determination. One possibility is that the orientation of the cleavage plate regulates adherens junction integrity. When the cleavage plate is oriented parallel to the ventricular surface, a daughter cell located distal to the surface may inherit few or no adherens junctions(Chenn et al., 1998; Manabe et al., 2002). However,recent studies have revealed that this type of cell division is rather rare in developing mouse neocortex (Haydar et al.,2003) and that only one daughter cell might inherit adherens junctions even when the cleavage plate is oriented perpendicular to the ventricular surface (Kosodo et al.,2004). In this case, differentiation of neuroepithelial cells into intermediate progenitors or neurons may trigger the destruction of adherens junctions. Our finding that the inactivation of aPKCλ causes a loss of adherens junctions suggests the involvement of this protein kinase in such a differentiation-dependent regulation of adherens junctions.
The molecular mechanisms underlying the regulation of adherens junctions by aPKCλ are largely unknown. In epithelial cells, aPKC regulates the integrity of tight junctions with its binding partners, PAR3 and PAR6(Macara, 2004; Ohno, 2001). PAR6 binds to Pals1, Lgl1 and Lgl2, and last two are the substrate for aPKCλ(Hurd et al., 2003; Plant et al., 2003; Yamanaka et al., 2003). PAR3 is also phosphorylated by aPKCλ, with the phosphorylation of these proteins being a crucial step for the establishment and probably also for the maintenance of tight junctions in epithelial cells(Hirose et al., 2002; Nagai-Tamai et al., 2002; Yamanaka et al., 2003). Although neuroepithelial cells lose tight junctions in their early embryonic stages, the remaining adherens junctions retain many of these aPKC-binding proteins (Aaku-Saraste et al.,1996; Astrom and Webster,1991). Some common molecular mechanisms may therefore act in the establishment of epithelial tight junctions and in the maintenance of neuroepithelial adherens junctions. This notion is in part supported by the observation that Lgl1, Pals1 and PAR3, as well as aPKC play crucial roles in the maintenance of adherens junctions in the neuroepithelium of the mouse telencephalon or zebrafish retina(Klezovitch et al., 2004; Wei et al., 2004; Wei and Malicki, 2002).
The role of aPKCλ on mammalian neurogenesis
In Drosophila neuroblasts, aPKC regulates cell fate by modifying cell polarity and proliferation (Rolls et al., 2003; Wodarz et al.,2000). As Drosophila neuroblasts lose adherens junctions upon delamination from the neuroectoderm, these functions of aPKC are independent of the maintenance of adherens junctions. Mouse aPKCs could also display such adherens junction-independent functions to regulate the differentiation in neuroepithelial cells. As aPKCλ cKO embryos exhibited no significant alteration of neurogenesis as described above, the argument could be made that another aPKC member protein, aPKCζ, may display such functions. However, a major form of aPKCζ expressed in neural tissue of both aPKCλ cKO and control embryos is PKMζ, which lacks the N-terminal PB1 domain essential for interaction with PAR6(Hirano et al., 2005; Suzuki et al., 2001; Wilson et al., 2003). The functions of PKMζ might thus be more restricted than intact aPKCζ or aPKCλ. In any case, a definitive answer for the issue of whether aPKCs regulate neural cell differentiation in mammals will be provided by analyzing the phenotype of aPKCλ/aPKCζ double-knockout mice.
Radial migration was not significantly affected in aPKCλ cKO embryos, and the radial glia scaffold comprising the basal cellular processes of neuroepithelial cells was normally aligned, even though the apical process was retracted (Fig. 4B,C). In contrast to the apical processes, in which integrity was totally dependent on adherens junctions, basal processes were probably maintained by interactions with the basal lamina at the pial surface and with differentiated neurons located in the subplate and cortical plate. Conversely, some reports have indicated the involvement of aPKCs in neural cell migration(Jossin et al., 2003; Solecki et al., 2004). Although the present data show little contribution of aPKCλ to radial migration, a signaling pathway to regulate cell migration could be driven by PKMζ and residual amounts of aPKCζ found in the aPKCλ cKO embryonic brain.
The loss of aPKCλ eventually causes brain malformation characterized by hydrocephalus and loss of the ependymal layer along with most of the striatum, which becomes prominent at P3 or later(Fig. 6B). The overall structure of the neocortex was relatively well maintained even in the absence of the ependymal layer, probably because neural fiber enriched layers such as cortical plate, subplate and intermediate zone defend neocortical cell layers against the gross disorganization observed in the striatum. In other words,brain malformation in aPKCλ cKO mice indicates that the maintenance of neuroepithelial tissue architecture by adherens junctions is essential for mechanical support of the developing brain, although the process is largely dispensable for neural cell differentiation and migration in the late embryonic stages. Moreover, appropriate regulation of aPKCλ activity could contribute to morphogenesis in brain ontogeny.
We thank David Anderson for the neurogenin 2 antibody; Noriko Ohsumi for the neurogenin 2 staining; Tomonori Hirose, Masa-aki Nakaya and Tomoyuki Yamanaka for their helpful comments; and Atsumi Kawaguchi and Yumi Bamba for animal care. F.I. was supported by Research Fellowships from the Japan Society for the Promotion of Science for Young Scientists. This work was also supported in part by grants from the Ministry of Education, Culture, Sports,Science and Technology of Japan (S.O.).