Whereas neurons of the lower layers (VI–V) of the cerebral cortex are first born from dividing precursors at the ventricular zone, upper layer neurons (II–IV) subsequently arise from divisions of intermediate neuronal precursors at the subventricular zone (SVZ). Little is known about mechanisms that control the proliferation of SVZ neuronal precursors. We herein report that the restricted expression of the homeodomain transcription factor Cux-2 in the SVZ regulates the proliferation of intermediate neuronal precursors and the number of upper layer neurons. In Cux-2-deficient mice (Cux-2−/−), there is excessive number of upper layer neurons and selective expansion of SVZ neuronal precursors. Double-labeling experiments demonstrate that Cux-2−/− upper layer precursors reenter the cell cycle in a higher frequency than wild-type precursors. Overexpression studies indicate that Cux-2 controls cell cycle exit in a cell-autonomous manner. Analysis of Cux-1−/−; Cux-2−/− double mutant revealed that Cux-2 controls SVZ proliferation independently of Cux-1, demonstrating that this is a unique function of Cux-2, not redundant with Cux-1 activities. Our results point to Cux-2 as a key element in the control of the proliferation rates of the SVZ precursors and the number of upper cortical neurons, without altering the number of deep cortical layers.
The vertebrate cortex is organized into distinct layers, with the neurons within each layer sharing similar functions, morphology, and birthdates. An important advance in the understanding of brain development is that we now know that the sequential birth of the distinct cortical layers is achieved via the generation of an intermediate neuronal precursor for the upper layers. Whereas neurons of the lower layers arise first from asymmetric divisions of the radial glial cells in the ventricular zone (VZ) (Malatesta et al. 2000; Noctor et al. 2001), neurons of the upper cortical layers are born from intermediate neuronal precursors dividing symmetrically in the subventricular zone (SVZ) (Haubensak et al. 2004; Miyata et al. 2004; Nieto et al. 2004; Noctor et al. 2004; Zimmer et al. 2004; Englund et al. 2005; Wu et al. 2005; Cappello et al. 2006). This 2-step mechanism of development has been proposed to confer an evolutionary advantage by enabling both the expansion of cortical layers and the appearance of new neuronal types (Englund et al. 2005; Gotz and Huttner 2005; Hill and Walsh 2005; Kriegstein et al. 2006).
Indeed, the evolution of the cerebral cortex extended the cortical surface area and increased the number of neuronal layers and neuronal circuits. Furthermore, the number of distinguishable upper layers increases from mice to humans (reviewed in Marin Padilla 2001) so that the circuit diagram differs between primates and other mammals (Hill and Walsh 2005).
Remarkably, this increment in the number of superficial layers of the cortex that occurs during evolution correlates with an expansion of the SVZ during embryonic development. This might be indeed attributable to an increase in the number of intermediate neuronal precursors. Furthermore, the selective expansion of the upper layers in the outer part of the cortical gyri and of their precursors at the SVZ has recently been proposed to contribute to the formation of the gyri typical of the cerebral cortex of carnivores and primates (Kriegstein et al. 2006).
Any account of the evolution and development of the cerebral cortex therefore requires an understanding of the molecular basis regulating the proliferation of SVZ precursors and differentiation of the upper layer neurons. Several molecules have been reported to control precursor cell proliferation (Takahashi et al. 1995; Chenn and Walsh 2002; Nowakowski et al. 2002), and the sequential activation of transcription factors governs the acquisition of specific laminar fates (Englund et al. 2005; Guillemot et al. 2006). However, little is known about factors involved in the selective proliferation and differentiation of SVZ precursors and upper layer neurons (Calegari et al. 2005; Guillemot et al. 2006; Glickstein et al. 2007). Genes selectively expressed in the neuronal precursor population of the SVZ during embryonic development, such as SVET, Tbr-2, Nex/Math-2, and Cux-2 (Tarabykin et al. 2001; Nieto et al. 2004; Zimmer et al. 2004; Englund et al. 2005; Wu et al. 2005), may participate in these processes. Remarkably, 2 of these genes, Cux-2 and SVET, are also expressed in the upper layer neurons.
Cux-2 encodes a vertebrate homolog of the Drosophila transcription factor Cut (Quaggin et al. 1996). In the peripheral nervous system, Cut determines the neuronal fate of external sensory organ precursor cells (Bodmer et al. 1987; Blochlinger et al. 1990) and the dendrite morphology of their neuronal progeny (Grueber et al. 2003). Upper layer neurons express Cux-2 and a second Cut homolog, Cux-1, which does not show the same highly SVZ restricted expression during development (Nieto et al. 2004). The function of mammalian Cux genes in the nervous system is unknown. The overlapping expression patterns of Cux-1 and Cux-2 suggest redundant functions for these genes in neural cells and accordingly Cux-1-deficient mice (Cux-1−/−) show no specific phenotype related to the development of the nervous system (Luong et al. 2002). However, the restricted expression pattern of Cux-2 in the SVZ suggests specific functions for this transcription factor, possibly in the generation of neurons of the upper cortical layers. To address these issues, we generated a Cux-2-deficient mouse strain by targeted gene deletion and investigated the development of the cerebral cortex in these animals.
Our analysis of Cux-2−/− mice demonstrates that Cux-2 negatively regulates the proliferation of precursors of the upper layer neurons in the SVZ, which reenter the cell cycle at a higher frequency in the Cux-2−/− mice. Thus, our results point to Cux-2 as a gene, the expression of which limits the proliferation of SVZ intermediate precursor. In this context, mechanisms modulating Cux-2 expression and functions appear to be candidate for playing important roles in the expansion of the cerebral cortex that occurs during evolution.
Animals, Construction of a Mouse Cell Line Conditionally Targeting Cux-2/Cutl2, and Generation of Cux-2−/− Mice
All animal procedures were approved by the Centro Nacional de Biotecnología Animal Care and Use Committee, in compliance with National and European Legislation. Cux-2 gene encodes 4 DNA-binding domains: 3 Cut repeats and 1 homeodomain. A targeting construct was designed to conditionally eliminate exons 22 and 23, which encode the third Cut repeat and part of the homeodomain (exons 22 and 23) near the C terminus of the Cux-2 protein, and to create a premature stop codon 3′ after the deletion. The targeting vector was constructed using a 1.9-kb DNA fragment that contains regions located 1.4 kb 5′ and 0.5 kb 3′ of exon 21. The middle arm, containing exons 22 and 23, is a 1.7-kb polymerase chain reaction (PCR) product. The long arm is a 7.5-kb NdeI genomic fragment from a mouse lambda library. A total of 10 μg of targeting vector was linearized with NotI and then transfected into J1-129 embryonic stem cells by electroporation. After selection in G418, surviving colonies were expanded and analyzed by Southern blot to identify clones that had undergone homologous recombination. Correctly, targeted ES cell lines were microinjected into C57BL/6J blastocysts. Resulting chimeric mice were crossed with C57BL6 mice to obtain Cux-2+/loxP mice. Mice carrying the conditional allele (Cux2loxP) were mated with mice expressing CRE recombinase under the human beta-actin promoter (Tg (ACTB-CRE)2Mrt deleter mice; Jackson Laboratories, Bar Harbor, ME) on a Swiss Webster background, to obtain mice giving germ line transmission of the floxed null allele. Heterozygous (het) mice for this allele (hereafter Cux-2+/−) were crossed with wild-type (WT) mice to segregate from the beta-actin transgene. Cux-2+/− animals were mated to obtain Cux-2 homozygous mutant mice (Cux-2−/−). Previous studies with antibodies that recognize different regions of the Cux-2 protein have found 2 isoforms of Cux-2 protein in a human cell line. Both isoforms are recognized by an antibody against the N-terminal part of the protein (anti-Cux-2 antibody 356) encoded by the undeleted part of the targeted Cux-2 gene. Anti-Cux-2 antibody 356 shows immunoreactivity with mouse Cux-2 (Gingras et al. 2005). Western blot analysis of the expression of Cux-2 protein in Cux-2−/− mice using antibody 356 reveals 2 bands in adult cerebral cortex homogenates (Supplementary Fig. 1). No bands were detected on brain homogenates from adult Cux-2−/− animals (Supplementary Fig. 1). Immunohistochemistry (IHC) studies using antibody 356 shows expression of Cux-2 protein in the WT brains but not in Cux-2−/− animals. These experiments demonstrate the complete absence of Cux-2 protein and lack of Cux-2 truncated mutated forms of the protein in the Cux-2−/− animals. Cux-1−/− mice have been described previously (Luong et al. 2002) and were obtained from A. J. van Wijnen (University of Massachusetts Medical School, MA, USA). Animals were maintained on a C57BL6: Swiss Webster background. Morning of the day of the appearance of the vaginal plug was defined as embryonic day (E) 0.5.
Southern Blot and PCR
Genomic DNA was obtained from ES cell clones or tail biopsies and digested with KpnI. Southern blot was performed using ExpressHyb hybridization solution (BD Biosciences, Mountain View, CA) according to the manufacturer's protocol. A 1-kb cDNA probe corresponding to exon 24 was used to screen for positive ES cell clones and to genotype mice carrying the loxP and null alleles. Mice were screened by Southern blot and the PCR. Primers for the detection of the null floxed allele were 5′-AAGGGCGGTGATTACAGAGA-3′ and 5′-GCCTGCTGTGGTAGACAGGT-3′; primers for the WT allele were 5′-TCAGCACATGGTGTCTGGAT-3′and 5′-CCTACTTTCTGCCTGCTTG-3′. PCR was carried out over 35 cycles of 94 °C for 1 min, 60 °C for 1 min, and 72 °C for 1 min in 2.5 mM Mg2Cl and 5 mM Betaine.
Antibodies, IHC, and Histology
Mice were perfused transcardially with 0.1 M phosphate-buffered saline (PBS; pH 7.4) followed by cold 4% paraformaldehyde in PBS. The perfused brains were removed and postfixed in 4% paraformaldehyde at 4 °C overnight. Fixed brains were cryoprotected in 30% sucrose in PBS and sectioned on a cryostat to produce either 10–20 μM cryosections on Superfrost Plus microscope slides (Fisher Scientific, Pittsburgh, PA) or 50–100 μM floating cryosections. Sections were blocked for 1 h at room temperature (r.t.) with 5% horse serum in PBS containing 0.5% Triton-X 100 (blocking solution) and then incubated for 1 h at r.t. or overnight at 4 °C with primary antibodies diluted in blocking solution. Fluorescent-tagged secondary antibodies (in PBS, 5% horse serum) were applied for 1 h at r.t., and sections were counterstained with Hoechst 33342 (Molecular Probes, Eugene, OR) and mounted in Aqua-polymount mounting medium (Poly-Labo, Strasbourg, France). Peroxidase/diaminobenzidine IHC staining was performed as described (Cubelos et al. 2005). Briefly, after incubation with primary antibody, sections were incubated with biotinylated donkey anti-rabbit IgG (Sigma, St Louis, MO) for 1 h. Sections were then washed 3 times in PBS, incubated with streptavidin-biotinylated horseradish peroxidase complex, washed, and incubated with 0.1 mg/ml H2O2 and 0.5 mg/mL diaminobenzidine in PBS. Sections were mounted in glycerol–gelatin.
The following primary antibodies were used at the dilutions indicated: rabbit polyclonal anti-Cux-1 (clone M222) (1:10) and anti-Brn-1 (1:50) (Santa Cruz Biotechnologies Inc., St Cruz, CA), rabbit polyclonal antiphosphohistone H3 (pH3) (1:500) (Upstate, Spartanburg, SC), rabbit anticleaved caspase-3 (1:500) (Cell Signaling Tech Inc., Danuers, MA), rabbit anti Ki67 (Novocastra, Newcastle on Tyne, UK), rabbit polyclonal anti-Cux-2 (antibody 356, a gift from Dr Alex Nepveu of McGill University Health Centre, Canada), rat anti-bromodeoxiuridine (BrdU) (1:50) (BD Biosciences), and rabbit anti-Tbr-2 antibody (1:500) (Chemicon Inc., Hampshire, UK). Goat anti-rabbit and goat anti-rabbit and anti-mouse secondary antibodies, conjugated to Alexa 488 and 594, respectively (Molecular Probes), were applied at 1:500. Before staining for Cux-1, Cux-2, Ki67, BrdU, clorodeoxiuridine (CldU), iododeoxiuridine (IdU), or pH3, sections were boiled for 30 min in antigen retrieval solution (Vector Laboratories, Burlingame, CA). This was followed in the case of staining for BrdU, CldU, IdU, and pH3 by 30-min treatment with 2 N HCl.
Staining for CldU and IdU was performed as described (Aten et al. 1994). Briefly, slides were incubated overnight at 4 °C with a mouse monoclonal anti-BrdU antibody (1:50; BD Biosciences; cross-reactivity with IdU), then washed in Tris buffer, pH 8.0, containing 0.5 M NaCl and 1% Tween 20 for 20 min at r.t. to eliminate background and possible cross-reactivity with CldU, and incubated overnight at 4 °C with rat monoclonal anti-BrdU antibodies (1:250; Abcam, Cambridge, UK; cross-reactivity with CldU). Secondary anti-mouse conjugated to Alexa 495–conjugated and anti-rat Alexa 598–conjugated antibodies (1:500; both from Molecular Probes) were applied for 2 h at r.t. The presence of single positive neurons for IdU and CldU in the stained cortical sections demonstrated the specificity of the antibodies and the staining.
BrdU, CldU, and IdU Injections and Cell Counting
All BrdU, CldU, and IdU quantification analysis show results obtained from the primary sensory cortex (primary somatosensory cortex barrel field [S1BF]. Interaural 2.34-2.22, Bregma 1.46-1.58 according to the mouse atlas of Paxinos and Franklin 1997). Similar results were observed in the primary motor cortex and other regions of the sensory cortex (M1 and primary somatosensory cortex forelimb region [S1FL], according to the mouse atlas of Paxinos and Franklin). BrdU, IdU, and CldU were administered intraperitoneally at 5 mg/mL in PBS. A single CldU injection was given 10 h after IdU injection at E14.5. Anatomically matched sections were selected from each mouse at each stage after BrdU injection (n = 3 Cux-2+/? control mice and n = 3 Cux-2−/− mice), and BrdU was detected by IHC. The total cortical thickness was subdivided into 10 bins of equal area, and the number of cells in which at least half of the nucleus was BrdU positive was counted in each bin. Identical results were obtained from independent analyses by 2 investigators.
Cortical Layer Thickness and Neuronal Density
For neuronal cell density and cortical thickness, serial sections were cut from brains of control and Cux-2−/− animal (n = 3). Then sections were Nissl stained, matched, and photographed. Measurements and cell counts were performed on the primary sensory cortex (S1BF, Interaural 2.34-2.22, Bregma 1.46-1.58 according to the mouse atlas of Paxinos and Franklin 1997). Similar results of increased layer thickness were observed in S1FL and M1 (Paxinos and Franklin 1997). Analysis was performed blind by 2 investigators in independent analyses. Cell density was calculated using the optical fractionator method (Sterio 1984). Optical dissector was given by a preexisting microscope grid (x, y, z = 61, 61, 10 μm). For cortical thickness, 4–5 measurements were taken of each brain region and mean values and standard deviations obtained. Statistical significance was calculated using the 2-tailed Student's t-test.
Confocal Microscopy and Imaging
Confocal microscopy was performed with a Radiance 2100 (Bio-Rad, Hercules, CA) Laser Scanning System on a Zeiss Axiovert 200 microscope. For fluorescence excitation, an argon ion laser (488 nm), a Krypton–Neon laser (543 nm), and a red diode (637 nm) were employed. The filter combinations used for detection of Alexa 488 and Alexa 594 were a 560 DCLPXR beam splitter and HQ 515/30 emission filter and a 650 DCLPXR beam splitter with HQ 590/70, respectively. Sequential images were taken with LaserSharp v5.0 software (Bio-Rad) and analyzed using LaserPix v.4 image software (Bio-Rad).
Neurosphere Culture, Nucleofection, and Constructs
Dorsal telencephalons of E14.5 WT embryos were dissected and cells dissociated as described (Nieto et al. 2004). Dissociated single cells were cultured in DMEM-F12 supplemented with N2, fibroblast growth factor (10 ng/mL), and epidermal growth factor (10 ng/mL) (Gibco, Invitrogen, Carlsbad, CA). Neurosphere cultures were passaged every 3 days by mechanical dissociation. Dissociated cells were cotransfected using the nucleofector technic (Amaxa Biosystems, Gaithersburg, MD) with 2.5 μg each of DNA constructs containing Cux-2 and green fluorescence protein (GFP) cDNAs under the cytomegalovirus enhancer, chicken β-actin promoter, and rabbit β-globin poly(A) signal (CAG) (cytomegalovirus [CMV] and beta-actin) promoter or the empty vector and GFP construct. Neurosphere formation was monitored after 10 days in culture.
Generation of Cux-2−/− Mice
To investigate the functions of Cux-2 in the development of the nervous system, we generated mice carrying a null allele of Cux-2 and analyzed the resulting brain phenotype. The mutation was designed to remove exons 22–23 that encode the third Cut repeat and the homeodomain near the C-terminal end of the protein. Het Cux-2+/− mice were obtained by breeding mice carrying a Cux-2 conditional allele (Cux-2+/loxP) with CRE deleter mice (see Experimental Procedure and Fig. 1a–c). Homozygous Cux-2 mutant mice (Cux-2−/−) were obtained by crossing het Cux-2+/− animals. Cux-2−/− mice were born at the expected Mendelian ratios and showed overall normal development and growth (not shown). To confirm that the mutation of the Cux-2 C terminus effectively removed expression of Cux-2 protein in mutant mice, postnatal brain sections of WT and Cux-2−/− mice were stained with a Cux-2 antiserum against the N-terminal part of the protein that recognizes the 2 described isoforms of Cux-2 protein but does not recognize Cux-1 (Gingras et al. 2005). This demonstrated the complete absence of Cux-2 immunoreactivity in all brain regions of Cux-2−/− animals, including the upper layer neurons and neurons of the ventrolateral and piriform cortex (Fig. 1d, left panels). Western blot experiments confirmed the complete absence of Cux-2 protein in Cux-2−/− cortical homogenates (Supplementary Fig. 1). In contrast, staining with a Cux-1-specific antiserum showed that the Cux-1 protein expression was indistinguishable between Cux-2−/−, Cux-2+/−, and WT littermates (Fig. 1d and not shown). Thus, loss of Cux-2 protein expression does not affect the expression of Cux-1. Also unaffected were the expression patterns of the layer-specific transcription factors, Brn-1 (II-V), Foxp-1 (III–V), Id-2 (II, III, V, and VI), and Tbr-1 (VI) (Bulfone et al. 1995; Rubenstein et al. 1999; Hevner et al. 2001; McEvilly et al. 2002; Sugitani et al. 2002; Ferland et al. 2003) (Fig. 1e). These results thus show that in mice completely lacking Cux-2 protein expression, upper and lower layer neurons each acquire their correct laminar position and correctly expressed the early layer-specific markers that we analyzed.
Cell Density and Total Neuronal Numbers Are Increased in the Upper Layers of the Cux-2−/− Cerebral Cortex
Visual analysis indicated that adult Cux-2−/− brains were moderately but consistently bigger in volume than WT brains (Fig. 2a). Examination of histological sections showed that both cell density and the thickness of upper cortical layers (II–IV) of Cux-2−/− mice was greater than in WT animals (Fig. 2b,c). The increased cell density was significant in layers II–III, (20% increment) but was most pronounced in layer IV (30% increment). In contrast, we observed no differences in neuronal density in layers V and VI (Fig. 2b). The increased cell density in the Cux-2−/− upper layers was accompanied by a moderate increase in the thickness of upper layers (15% increment) (Fig. 2c). Again, this increment was not observed in lower layers and resulted in a modest increase in the overall cortical thickness (Fig. 2c). The combination of increased cellular density and increased upper layer cortical thickness indicates an important increase in the total number of neurons in layers II–III and IV of the Cux-2−/− cortex.
BrdU birthdating experiments performed at E13.5, E14.5, and E16.5 (Fig. 3) confirmed the selective increase in the number of upper layer neurons. In accordance with the correct expression of layer-specific markers (Fig. 1e), the BrdU experiments suggested normal neuronal migration: neurons born at E14.5 and E16.5 populated the upper part of the postnatal cortex (P21) and not the deep cortex in control and Cux-2−/− animals (Fig. 3a–c). Likewise, neurons born at E13.5 populated the deep cortical layers in both groups of mice (Fig. 3d). However, the number of BrdU-positive neurons that populated the superficial bins of Cux-2−/− animals and that were born at E14.5 and E16.5 was increased 40% and 30%, respectively, in Cux-2−/− mice compared with control mice (WT or Cux-2+/−) (Fig. 3b,c; bins 2–4 at E16.5 and bins 2–5 at E14.5). In contrasts, no differences were found between the populations born at E13.5 (Fig. 3c). Interestingly, the increased number of labeled upper layer neurons was produced by an increase in the number of dimly fluorescent BrdU–labeled cells (Fig. 3a–c, right panels), suggesting that the pulse label had been diluted by additional rounds of cell division. Hence, upper layer precursors appear to undergo additional rounds of division when Cux-2 is removed. Also supporting an increase in proliferation, the rates of apoptosis in the brains of WT and Cux-2−/− mice at P0 and P16 postnatal stages were the same (Fig. 4a,b). Thus, the increment in the number of upper layer neurons in the Cux-2−/− mice cannot be attributed to defective neuronal apoptosis. Decreased postnatal apoptosis in Cux-2−/− mice would also predict less significant differences in the number of BrdU-labeled upper layer neurons at early postnatal stages than in the late postnatal days. But we found increased numbers of E16.5 BrdU–labeled neuronal population in the Cux-2−/− cortex already at the first postnatal days (P0–P5), and the increment was equivalent to that found at P21 (Fig. 4c and see also Fig. 6). All together, we conclude that the increase number of upper layer neurons in the Cux-2−/− mutant mice is due to increased proliferation of neuronal precursors during development and not to decreased apoptosis.
Neuronal Proliferation Is Increased in the SVZ of Cux-2 Mutant Mice
The restricted expression of Cux-2 mRNA (Nieto et al. 2004) and protein in the SVZ (Supplementary Fig. 2) and the increased upper layer neurons in Cux-2−/− mice suggest that Cux-2 has specific functions in the proliferation of intermediate precursors for upper layer neurons. To identify possible differences in cell cycle parameters, we combined immunostaining for Ki67 antigen, a protein expressed in all dividing cells, and a pulse of BrdU to label cells in the S phase. After a 2-h pulse of BrdU in E15.5 embryos, precursor cells in the SVZ were identified as those set apart from the BrdU-positive precursors aligned along the VZ (Fig. 5a). Remarkably, the number of Ki67- and BrdU-positive cells was dramatically increased in the SVZ of Cux-2−/− cortex (Fig. 5a, left panel graph). No significant differences were observed in the proportion of cycling cells that were in the S phase (i.e., the ratio of BrdU+/Ki67+) between the control and Cux-2−/− mutant mice. Because this ratio provides an estimate of cell cycle length, this therefore indicates that Cux-2−/− and control SVZ precursors cycled at similar rates (Fig. 5a, right panel graph).
We next studied the number and location of cells undergoing mitosis with an antibody against the phosphorylated form of histone 3 (pH3). pH3 staining of E15.5 cortical sections revealed 2 mitotic zones: apical mitosis occurring in radial glial cells at the ventricular walls and basal mitosis, clearly separated from the ventricle, occurring in cycling cells of the SVZ (Fig. 5b). Quantification analysis revealed that the number of cells in basal mitosis was greatly increased in the SVZ of E15.5 Cux-2−/− mice compared with their WT counterparts (Fig. 5b, graph). In contrast, no differences were observed in numbers of cells undergoing apical mitosis (Fig. 5b, graph). This indicates that proliferation of VZ precursors, which do not express Cux-2, is not affected in Cux-2−/− mice and demonstrates a selective increase in the proliferation of SVZ neuronal precursors but not of radial glial cells. In agreement with this, the number of Tbr-2 positive cells, which mark intermediate neuronal progenitors in the proliferative regions (Englund et al. 2005), was dramatically increased in the cortex of Cux-2−/− mice (Fig. 5c). Similarly, analysis of the number of Tbr-2 and pH3 in the SVZ of E14.5 revealed increments of proliferating SVZ cells in E14.5 Cux-2−/− mice compared with WT animals. In contrast, no differences were found on cortical basal mitosis at E12.5, when lower layer neurons are generated (Supplementary Fig. 3).
Cux-2 Regulates the Number of Upper Cortical Neurons Independently of Cux-1
VZ and SVZ cells express low levels of Cux-1 protein, which also shows persistent expression in upper layer neurons (Nieto et al. 2004). To investigate the possibility that Cux-1 and Cux-2 might show redundant or interacting roles, we examined Cux-1−/−; Cux-2−/− double-mutant mice. This double gene deletion is embryonically lethal for reasons that are still undetermined. The lethality of Cux-1−/−; Cux-2−/− double mutation is due to defects manifest at early embryonic stages (between E9 and E15) (Cubelos B and Nieto M., unpublished data). However, although this phenotype is highly penetrant, a small proportion of mice develop to birth. Nissl staining of P0 coronal sections showed no gross abnormalities in the double and single Cux-1 and Cux-2 mutants and their cortex appeared to be normally laminated (Fig. 6a). We analyzed the migration and the number of E16.5 BrdU-labeled neurons at birth (P0) in Cux-1−/−; Cux-2−/− double mutants, compared with single Cux-2-/- mutants, and control littermates (Fig. 6b,c). The migration patterns of BrdU-labeled cells to the upper layer of the cortex were indistinguishable in all genotypes. However, the number of E16.5-birthdated neurons was significantly increased in the cortex of Cux-1−/−; Cux-2−/− double-mutant mice compared with controls, but this increase was no greater than that observed in the Cux-2 single-mutant mice (Fig. 6c). These experiments rule out the possibility that an unrepressed function of Cux-1 is responsible for the increased proliferation of upper layer neurons in Cux-2 mutant mice. They moreover provide evidence that the control of upper layer neuron number is a unique function of Cux-2 and is not redundant with Cux-1 activities.
Cux-2 Controls Reentry of Precursor Cells into the Cell Cycle in a Cell-Autonomous Manner
The higher number of weakly BrdU-labeled cells in the cortex of Cux-2−/− mice suggested that neuronal precursors in these animals underwent more cell divisions than WT cells immediately before the generation of upper layer neurons. To investigate this further, we set out to unequivocally identify whether cortical neurons had been generated from precursors that had divided once, or more than once, by sequential injection with 2 different halogenated nucleotide analog. The nucleotide analog IdU was given in a first injection, and CldU was injected 10 h later. The interval between injections allows many cells to exit the S phase after the first nucleotide injection but is shorter than the whole cell cycle, ensuring that cells reentering the cell cycle are labeled with the second nucleotide. Thus, cells that exited the cell cycle after the first injection are labeled with IdU only, whereas cells generated after subsequent subdivisions are double labeled with IdU and CldU (see diagram Fig. 7c). Cells that were not in the S phase at the first injection, or that diluted any incorporated IdU sufficiently during successive divisions, appear labeled only with CldU. We analyzed the upper layer neurons labeled at E14.5 in postnatal (P21) brain sections of control and Cux-2−/− animals (Fig. 7), and we found that the number of double-positive cells was much greater in the Cux-2−/− cortex (Fig. 7a). The ratio between the number of CldU-positive cells (second injection; red cells) and the total number of IdU-postive cells (first injection; green cells) shows that the proportion of precursor cells that reentered the cell cycle after the first injection was 2.5-fold higher in the Cux-2−/− mice than in control animals (Fig. 7b). Thus, our double-labeling analysis suggests that Cux-2 controls exit from the cell cycle (Fig. 7c,d).
To investigate whether the phenotype of Cux-2-deficient precursors reflects a cell- or noncell-autonomous function, we overexpressed Cux-2 in dividing embryonic neural stem cells grown as neurospheres in vitro. After cotransfection with Cux-2, or the empty vector (control), together with a GFP reporter, dissociated single GFP-positive cells were sorted by flow cytometry fluorescence activated cell sorting (FACS) (Fig. 7e). Sorted cells were then seeded at clonal density, and the number of neurospheres was quantified after 7 days in culture. As shown in Figure 7f,g, overexpression of Cux-2 significantly reduced the number of neurospheres generated from the sorted cells, compared with control cells. Given that Cux-2-expressing cells and control cells were equally able to survive and differentiate into glia and neurons when plated in differentiating conditions (not shown), the reduced neurosphere formation by Cux-2 overexpressing progenitors was not due to increased death rate or to changes in cell fate. These results therefore strongly suggest that Cux-2 is able to direct cells out of the cell cycle in a cell-autonomous manner.
In the present study, we demonstrate that expression of Cux-2 selectively controls the proliferation of intermediate neuronal precursors in the SVZ and the number of upper layer neurons. In the context of evolution, the actions of Cux-2 during development might provide mechanisms controlling the appearance and expansion of upper layer neurons.
Our data indicate that the restricted expression of Cux-2 in the SVZ selectively controls the proliferation rates of neuronal intermediate precursors during mouse development. The role of Cux-2 in cell cycle regulation was previously suggested by its reported expression in tissues undergoing extended proliferation during embryonic development such as the limb bud progress zones or the urogenital system (Iulianella et al. 2003) but has not been demonstrated before. Hence, we herein provide the first evidence demonstrating a role for Cux-2 in cell proliferation.
Interestingly, our studies demonstrate that Cux-2 limits the number of times SVZ neuronal precursors go through the cell cycle. Our neuronal birthdating and double-labeling experiments demonstrate that Cux-2−/− upper layer precursors reenter the cell cycle at a higher frequency than their WT counterparts (Fig. 7d). Concurrently, we observed an expansion of intermediate neuronal precursors in the SVZ of the developing Cux-2−/− mice. This selective expansion is most definitively demonstrated by an increase in Tbr-2-expressing cells but no changes in precursors of the VZ. This selective effect of Cux-2 in SVZ cells supports the notion that the functions of Cux-2 are cell autonomous. Cell-autonomous mechanisms are also strongly supported by our experiments showing that overexpression of Cux-2 decreases proliferation in neural progenitor cells in vitro.
The increased numbers in the number of SVZ precursors in Cux-2−/− mice are quite significant (about 40% increment). This excessive proliferation temporally correlates with increased production of late-born neurons that migrate to the upper layers in normal patterns (30–40% in BrdU experiments). This causes parallel increments in cell density of layers II–IV (20–30% increment), as well as a moderate increase in upper layer thickness (15% increment).
Somewhat surprisingly, Cux-2 does not appear to be functionally redundant with Cux-1 in the SVZ precursor. The nonredundancy of Cux-2 in intermediate neuronal precursors is suggested by its restricted expression in the SVZ, which is not shown by Cux-1, because it is expressed in VZ precursors as well. Our analysis of Cux-1−/−; Cux-2−/− double-mutant mice confirms this notion and demonstrates that the proliferation of upper layer neuronal precursor does not depend on Cux-1 activity. The 2 Cux genes thus do not cooperate to control these aspects of neuronal development.
Our findings are of particular interest in the search for a mechanism to explain how the appearance and selective expansion of upper cortical layers during development evolved (Marin Padilla 2001; Hill and Walsh 2005; Kriegstein et al. 2006; Martinez-Cerdeno et al. 2006). A larger cortex implies that precursor cells undergo more rounds of division. In addition, the extended period of neurogenesis adds the upper layer to the cortex. Indeed, upper cortical layer neurons are highly represented in the primate cerebral cortex, especially in humans (reviewed in Hill and Walsh 2005). During development and evolution, inhibition of Cux-2-mediated cell cycle exit in the SVZ could be a target mechanism to expand the number of upper layer neurons without affecting the number of deep layer neurons. Thus, the functions of Cux-2 fulfill both of the developmental and evolutionary necessities of expanding the number of layers. In this regard, evolution might have required Cux-2 functions in processes such as the addition of upper layer neurons to the 3 layers of the reptilian cortex, the later selective expansion of upper layers in higher mammals (Reiner 1991; Marin Padilla 2001; Kriegstein et al. 2006), or the focal expansion of the upper layers and the SVZ that has been proposed to occur during the formation of the gyrencephalic cortex (Kriegstein et al. 2006).
Ministerio de Educación y Ciencia (SAF2005–0094); Mutua Madrileña Automovilistica (2004); RYC-2003-006143 to M.N.; JdC-05-162-74c to B.C.; the Ministerio de Educación y Ciencia (BES-2006-13901) to A.S.-S.; National Institute of Neurological Disorders and Stroke (2RO1 NS032457) to C.A.W.; the Spanish Ministry of Health and Consumer Affairs and the Pro-CNIC Foundation to Centro Nacional de Investigaciones Cardiovasculares.
We thank B. Alarcón, S. Bartlett, F. Guillemot, M. Mellado, J. M. Rodríguez-Frade, P. Bovolenta, M. Guzmán, and H. M. van Santen for critical reading of the manuscript and for their experimental advice. We thank A. Nepveu (McGill University, Canada) for the anti-Cux-2 antibody, M. Sheng (Picower Institute, MIT, Cambridge, MA) for the Tbr-1 antiserum, and A. J. van Wijnen (University of Massachusetts Medical School, Worcester, MA) for the Cux-1 mutant mice. We are grateful to S. Montalban, S. Gutierrez-Erlandsson, and D. Esteban for technical assistance. During the generation of the Cux-2loxP conditional mice, M.N. was a Lefler fellow in the laboratory of C.A.W. We are indebted to C.A.W. for allowing us to continue the project at the Centro Nacional de Biotecnología (Madrid, Spain). C.A.W. is an Investigator at the Howard Hughes Medical Institute. Conflict of Interest: None declared.