Id4 regulates neural progenitor proliferation and differentiation in vivo

The increased cortical surface area of the human brain is associated with its ability to perform higher order cognitive functions, and hence understanding the molecular mechanisms that regulate cortical progenitor proliferation and differentiation is of great importance. The mammalian telencephalon develops from the rostral-most region of the neural plate. The dorsal part of the telencephalon gives rise to the hippocampus, neocortex,lateral cortex, and parts of the amygdala, along with other smaller structures. These different structures arise from distinguishable progenitor regions in the ventricular zone (VZ) called the medial pallium (MP), dorsal pallium (DP), lateral pallium (LP) and ventral pallium (VP), which are identifiable by combinatorial patterns of homeobox gene expression(Puelles et al., 2000; Yun et al., 2001). From within each zone, there is a sequential generation of neurons and glia in the telencephalon, such that neurogenesis precedes gliogenesis. Most neurons are generated from the proliferative zone between E11.5 and E17.5, whereas glial cells are generated during early postnatal days (P0-P20)(Sauvageot and Stiles,2002).

During early telencephalic development, there is an exponential expansion of neuroepithelial progenitors in the VZ as the result of rapid, symmetric cell divisions. The nuclei of these neuroepithelial cells go through characteristic movements within the VZ, known as the interkinetic nuclear movements. M-phase nuclei are found at the ventricular surface, while S-phase nuclei are found away from the ventricle in the distal region of the VZ(Takahashi et al., 1993; Takahashi et al., 1995a). These early VZ cells (E11.5-E12.5) divide continuously – their cell cycle time is estimated to be about 8 hours(Takahashi et al., 1993; Takahashi et al., 1992)– leading to rapid lateral expansion of the neuroepithelium. From the VZ emerge the cells that give rise to the secondary proliferative population(SPP) in the subventricular zone (SVZ)(Takahashi et al., 1995b) and differentiating neurons of the telencephalon.

Many genes known to affect proliferation and differentiation of the developing mammalian nervous system are transcription factors. In particular,members of the basic helix-loop-helix (bHLH) family have received much attention because of the potency with which they are able to instruct cellular differentiation (Guillemot,1999). Consistent with this role in cell fate determination and differentiation, many bHLH genes are expressed in a tissue-specific manner(class B bHLH proteins). A sub-class of HLH genes, which lack the basic DNA-binding domain, is known as Inhibitors of DNA binding or ID genes. The proteins encoded by these genes act as dominant-negative regulators of bHLH proteins by forming inactive heterodimeric complexes(Benezra et al., 1990; Norton, 2000; Yokota, 2001).

In mammals, there are four known ID gene family members. Id1, Id2,Id3 and Id4 are expressed in progenitors of the central nervous system (CNS) in overlapping but distinctive patterns(Andres-Barquin et al., 2000; Jen et al., 1997). Consistent with their expression patterns, the activities of ID1 and ID3 seem functionally redundant in the nervous system. Deletion of either Id1or Id3 individually does not lead to an observable phenotype, but when they are both inactivated, telencephalic neurogenesis and angiogenesis are greatly perturbed (Lyden et al.,1999). Mice lacking Id2 are viable, although they have an abnormal immune system and a lactation defect(Mori et al., 2000; Yokota et al., 1999). Id1,Id2 and Id3 have been shown to interact with cell cycle regulatory molecules (Zebedee and Hara,2001). For example, they negatively regulate the expression of cyclin D1, p16^Ink4a and p21^{CIP1/WAF1/SDI1}(Alani et al., 2001; Lasorella et al., 1996; Prabhu et al., 1997). In addition, both genetic and biochemical evidence indicates that ID2 inhibits RB(Retinoblastoma) protein function to enhance the G1 to S transition in proliferating cells (Iavarone et al.,1994; Toma et al.,2000).

The role of the most recently identified member of the ID family, Id4, in nervous system development has not been explored. In vitro studies suggest that ectopic expression of Id4 prevents oligodendrocytic differentiation (Kondo and Raff, 2000). However, its role in regulating neural stem cell proliferation and differentiation is not known. Here, we identify a role of Id4 during cortical development in vivo and reveal a novel function for Id4 in regulating cortical neurogenesis.

A NcoI-HindIII fragment, including the EGFP cDNA from the pEGFP-N1 plasmid vector (Clontech) ligated to a 1.1 kbp HindIII-NcoI Id4-5′ flanking genomic fragment and a 4 kb NotI-PstI fragment(Id4-3′-flanking genomic fragment), was inserted into the neo-tk cassette (Qiu et al.,1995) and introduced into JM-1 ES cells by electroporation. Of the 24 clones examined, 4 clones were found to contain a 6.5 kb band corresponding to the mutant allele. PCR primers specific for the wild-type and targeted alleles that were used included:

• Id4-wild-type sense, 5′-CTGCAGTGCGATATGAACG-3′;

• Id4-wild-type antisense, 5′-GCGGCTGTCTCAGCAAAG-3′;

• Id4-mutant sense, 5′-GTCGATTTCTGGAGCTCGGA-3′); and

• Id4-mutant antisense, 5′-AACTTGTGGCCGTTTACGTC-3′.

The PCR cycling conditions were 95°C for 30 seconds, 60°C for 40 seconds and 72°C for 1 minute for 30 cycles. The amplified PCR products were analyzed on gels to separate the wild-type (166 bp) and targeted (220 bp)allele fragments.

Coronal and saggital sections of Id4^+/– and Id4^-/- tissues at matching levels of the CNS were stained with Cresyl Violet and compared by measuring the VZ surface area using the ImagePro program. Three separate measurements of the same sections were averaged and compared. In each age group, three separate littermate brain pairs, sectioned in coronal and saggital planes at similar angles and matched to comparable levels along the rostrocaudal and mediolateral axis were analyzed.

In situ hybridization on frozen sections was carried out using ³⁵S-labeled antisense-riboprobes(Yun et al., 2001). Each probe was examined on three to five different mutant-control littermate pairs from at least two independent crosses at each age shown. Photographs were taken using dark-field optics on an Olympus SZH10 microscope. The cDNAs used in this study included: Wnt3a (A. McMahon, Harvard University, USA); Emx1 (J. Rubenstein, UCSF, USA); Dbx1 (T. Jessell, Colombia University, USA); Ngn2 and NeuroD (F. Guillemot, NIMR, UK); Tbr2 (A. Bulfone, TIGEM, Italy); and Id2 and Id4(M. Israel, Dartmouth Medical School, USA).

Labeling of S-phase cells by BrdU incorporation was performed by injecting BrdU (1 mg/10 g body weight) into pregnant females and waiting for 45-60 minutes before sacrificing. A standard immunofluorescence protocol was used. Primary antibodies used in this study included: phosphorylated Histone 3 (PH3)(rabbit IgG; Upstate Biotech), BrdU (rat; Harlen), MAP2 (mouse IgG; Sigma,catalog number M1406), PCNA (mouse IgG; Novocastra), TBR1 (gift of M. Sheng,Harvard Medical School, USA). Confocal images of PCNA staining were collected as 2 μm optical slices.

E12.5 dorsal telencephalon from Id4^-/- and control littermates was dissected in cold PBS. Cells were dissociated in 0.05%trypsin, rinsed, and resuspended in 100 μl of PBS. Cells were fixed in 70%ethanol, rinsed, resuspended in 1 ml of solution containing 100 μg/ml RNaseA/0.2% Triton X-100/50 μg/ml Propidium Iodide, and incubated at 37°C for 30 minutes. Stained cells were scanned in a Becton Dickinson (San Jose, CA) FACScan flow cytometer, and the cell cycle profile was quantitated using ModFit software.

To understand the in vivo function of Id4, we used homologous recombination to generate a mutant mouse in which the Id4 locus was disrupted (Fig. 1A). Exons 1 and 2 of Id4 were replaced by the coding regions of GFP(green fluorescent protein) and the neomycin-resistance gene. Following transfection into embryonic stem cells, recombination events were confirmed by PCR and Southern blot analysis (Fig. 1A). Two independent lines of recombinant animals carrying an altered Id4 locus were identified. In both lines, animals lacking functional Id4 survived postnatally in clean environments, but their brain (Fig. 1B) size was consistently smaller than that of their littermates (not shown).

Id4 is widely expressed in the developing mouse embryo nervous system (Andres-Barquin et al.,2000; Jen et al.,1997). Because we found that animals lacking Id4 had smaller brains, we examined telencephalic development in the Id4mutant animal to better understand the physiological role of Id4. Although we did not observe a phenotype in the Id4 heterozygote, the reduced brain size shown in Fig. 1B is apparent in homozygous animals from E11.5 onwards. When littermate brains were sectioned and compared at E12.5(Fig. 1C,C′), the ventricular surface of the mutant telencephalon was found to be approximately 30% smaller than that of the wild type along the mediolateral axis (compare the distance between the arrows in Fig. 1C,C′,F). In particular, the dorsal and medial pallial regions, primordia for the neocortex and the hippocampus, respectively,appeared much smaller than the more ventral regions in both the cortical(lateral and ventral pallium) and subcortical (ganglionic eminences) tissues(Fig. 1C,C′,E,F). Saggital sections at E15.5 and E18.5 showed smaller lateral ventricles and reduced ventricular surface along the rostrocaudal axis as well(Fig. 1D,D′,G). Concurrently, the mutant cortical plates are consistently thicker than those of their littermates (Fig. 1D,D′,H). From this analysis, we conclude that Id4function is required for the development of a normal size brain, and especially for the lateral expansion of the cortical neuroepithelium.

To determine whether the smaller cortex in Id4^-/-animals reflected the loss of a specific region or an overall reduction in all cortical regions, we examined the expression of molecular markers that distinguish different progenitor zones. Genes expressed at the dorsal and ventral extremes of the pallial (cortical) progenitor zone appeared normal. For example, Wnt3a expression in the cortical hem(Lee et al., 2000) was unchanged in mutant animals (Fig. 2A,A′). In addition, no consistent differences were found in the expression patterns of genes marking the ventral and lateral pallium progenitors (Dbx1, Emx1, Tbr2 and Wnt7b; Fig. 2B-D′, and not shown), or the mantle (not shown) (Puelles et al., 2000; Yun et al.,2001) in Id4^-/- animals. To quantitate the difference in the surface area of the cortical proliferative zone, we measured the cortical VZ width on matched coronal sections from littermate control and Id4^-/- animals. We found progressive shortening of the VZ surface area (Fig. 2G) and consistent reduction in the hippocampal primordium (approximately 50%, Fig. 2H) in Id4^-/- animals. These findings, in the context of the histological analysis shown in Fig. 1, indicate that the hypoplasia of the Id4^-/-telencephalon most severely affects dorsomedial structures.

As Id4 is highly expressed in all cortical progenitor zones(Fig. 2F), we were surprised to find these regional differences. We hypothesized that ID4 function is unique in the dorsomedial cortex and that other ID genes compensate for Id4loss in other regions of the telencephalon. Consistent with this possibility,we observed elevated levels of Id1, Id2 and Id3 mRNA in the Id4^-/- telencephalon by RT-PCR (data not shown). Because Id2 is expressed in a lateral-to-medial gradient in the VZ of the cortex (levels are undetectable between the arrowheads in Fig. 2E), we used in situ hybridization to determine whether an altered pattern of Id2expression could be identified. We found that Id2 expression was increased beyond its normal dorsal boundary in the cortex extending all the way to the midline in Id4′ animals(Fig. 2E′). This observation raises the possibility that Id4 normally represses Id2 expression in the dorsomedial region, or that the hypoplastic region of the Id4^-/- mouse cortex corresponds to areas in which Id2 is normally not expressed.

To determine whether the molecular abnormalities associated with the loss of Id4 also contributed to cortical hypoplasia by enhancing progenitor cell death, we examined apoptosis in the developing telencephalon of Id4^-/- animals. We did not detect differences in the number of apoptotic cells in the Id4^-/- and control telencephalon at E11.5, E12.5 or E15.5 (data not shown). Furthermore, the insertion of DNA encoding GFP into the Id4 locus allowed the identification of cells in which Id4 transcription would occur. GFP-expressing cells were clearly present in Id4^-/-mutants at E12.5 (Fig. 2F′), indicating that the loss of Id4 function does not result in extensive cell death.

To determine whether the dorsomedial cortical progenitor zones are reduced as a result of precocious differentiation or compromised proliferation of neural stem cells, we first examined early neurogenesis in the mutant telencephalon. At E11.5, cortical neurogenesis is just beginning in the dorsal and medial pallium, whereas in the more ventral regions of the cortex, the ventral and lateral pallium, neurons are produced from E10.5. Expression of pan-neuronal markers at E11.5 (MAP2 and βIII-tubulin, Fig. 3A,A′, and not shown) was reproducibly increased throughout the postmitotic zone of the Id4^-/- cortex, suggesting precocious neurogenesis. Therefore, we examined the expression of other genes that either regulate neurogenesis in cortical progenitors or are markers of cortical neurons. At E11.5, neurogenin 2 (Ngn2; Neurog2 – Mouse Genome Informatics), a proneural bHLH gene (Fode et al., 2000; Gradwohl et al.,1996), was expressed in a similar pattern in both wild type and Id4^-/- animals. However, the expression of markers identifying early differentiating neurons, NeuroD(Fig. 3C,C′) and Tbr2 (arrow in Fig. 2D,D′), was consistently increased in the mantle zone of the dorsomedial cortex in Id4^-/- animals. Furthermore,scattered Tbr2-expressing cells were found ectopically in the VZ(dorsal to its normal expression domain in the VP and LP(Bulfone et al., 1999; Yun et al., 2001)) of the dorsal and medial pallium (arrowhead in Fig. 2D,D′). Each of these findings is indicative of precocious differentiation of early cortical progenitors.

Early-born (E11.5-E13.5) neurons of the cortex form deep layer neurons(layers 5 and 6) (Takahashi et al.,1999). At E15.5 and E18.5, TBR1 expression in postmitotic cortical neurons (Hevner et al., 2001)identified an increased number of deep layer neurons in Id4^-/- animals (Fig. 3D-E′). In addition, TBR1-expressing neurons were found in more dispersed positions in Id4^-/- animals at E15.5(Fig. 3D′). Consistently, Id2 expression at E18.5(Rubenstein et al., 1999)revealed disorganized layering/differentiation within the cortical layers(Fig. 3F,F′). These observations provide additional evidence that early cortical progenitors are precociously differentiating in the absence of the functional Id4gene.

To determine whether defects in progenitor proliferation also contributed to the Id4^-/- mutant phenotype, we examined markers of cell cycle phases and found that G1-S transition of the Id4^-/-neuroepithelial cells is compromised at E12.5. We compared progenitor cells in wild-type and Id4^-/- cortex by analyzing the expression of markers that identify all cells actively proliferating (PCNA, proliferating cell nuclear antigen), or cells in M-phase (PH3: Phosphorylated Histone 3), or cells in S-phase, by a short pulse of BrdU (Bromodeoxyuridine)-labeling. At E12.5, the density of proliferating cells in the VZ is indistinguishable in wild-type and Id4^-/- animals (PCNA⁺ cells in Fig. 4A,A′,D′), but the VZ is approximately 15% thicker in the Id4^-/- mutant(see Fig. 1C,C′, Fig. 4A,A′). To determine whether this increased thickness could be attributed to cells at a particular phase of the cell cycle, we first examined cells in M phase. We found that the actual number of cells in M-phase was indistinguishable in the wild-type and Id4^-/- animals at this age(Fig. 4C,C′, and not shown). BrdU labeling at E12.5 identified cells that were in S phase(Fig. 4B,B′). Although the thickness of the S-phase zone varied within the cortical VZ (thinner dorsomedially) in the mutant animal, the most striking difference between the control and Id4^-/- animals we found was in the thickness of the zone corresponding to the G-phases of the cell cycle. The G-phase zone is sandwiched between the BrdU-positive S-phase cells(Fig. 4B,B′) and the PH3-positive M-phase cells (Fig. 4C,C′). The thickening of this zone (distance between the arrowheads in Fig. 4B,B′)clearly accounts for the overall increased thickness of the VZ in mutant animals. The presence of cycling cells in this region, as identified by PCNA expression (Fig. 4A,A′,D′), suggested that progenitor cells in the Id4^-/- animal might make a prolonged transition between G1-S and/or G2-M phases of the cell cycle. We, therefore, evaluated the cell cycle distribution of cortical cells isolated from E12.5 dorsal telencephalon by FACS analysis. In multiple experiments (n=7) examining cells from E12.5 or E13.5 littermate embryos, we invariably observed a reduction in the percentage of cells in S-phase and a corresponding increase in the percentage of cells in G0/G1, as indicated in representative experiments from E12.5 shown in Table 1. In wild-type animals at E12.5, virtually all VZ cells are dividing and no G0 cells are found among them (Takahashi et al.,1995a). Our finding that the thicker mutant VZ is composed of PCNA⁺ proliferating cells (Fig. 4A′,D′) indicates that the increased percentage of cells in G0/G1 detected by FACS analysis(Table 1) represents a prolonged G1 in the mutant cells at this age. Although the increased number of prematurely differentiating cells observed at this age may also contribute to the higher percentage of G0/G1 cells in the ID4 mutants, the fact that most cells in the G-phase zone are PCNA positive(Fig. 4A′,D′)indicates that a contribution of prematurely differentiating cells to the thickening of the VZ is unlikely. Together, these two independent lines of investigation indicate that Id4 is required for normal G1-S transition in early neural stem cells. Thus, the reduced brain size in the Id4^-/- cortex results from both premature differentiation and compromised cell cycle transition of early neural stem cells.

To determine whether the proliferation defect in Id4^-/-animals at E12.5 was also detectable at later stages of development, we examined markers of cortical progenitors and cellular proliferation at E15.5. Surprisingly, we found a greater number of proliferating cells outside of the VZ in the Id4^-/- animals. In addition to progenitors in the SVZ, many cells in the mantle (postmitotic) region ectopically expressed markers of VZ and SVZ, such as Ngn2, Ngn1, NeuroD, Emx1 and cadherin 8 (Fig. 4E-F′, and not shown). We interpret this expression pattern to reflect the presence of proliferating cells in the postmitotic region. Therefore, we sought evidence for ectopic proliferation at E15.5 by evaluating cortical tissue following a 45-minute BrdU pulse to identify S-phase cells and PH3⁺immunostaining to identify M phase cells. As indicated in Fig. 4G-H′, we observed many BrdU-positive, as well as PH3-positive, cells in the mantle zone. This observation indicates that, in contrast to the compromised proliferation observed at E12.5, cells in the postmitotic region are aberrantly proliferating in the Id4 mutant at E15.5.

Unlike other mouse lines in which inactivation of a single ID gene family member does not show an observable CNS phenotype(Lyden et al., 1999; Yokota, 2001), the Id4^-/- mouse has an obviously smaller brain. For example, Id1 and Id3 single mutants do not show an obvious CNS phenotype, although a phenotype in the telencephalon does emerge when Id1 and Id3 are simultaneously inactivated(Lyden et al., 1999). Here, we provide multiple lines of evidence indicating that the Id4 mutant phenotype arises from the requirement, in vivo, of ID4 function in cortical stem cells for their proliferation and differentiation. Importantly, we show different effects of Id4 loss of function on cortical cell proliferation at different times during development.

Neural epithelium of the developing cortex expands laterally through the rapid division of early stem cells. Cortical neurogenesis commences at around E11.5 and the onset of an observable Id4^-/- phenotype coincides with this major milestone in cortical development. At E12.5, the mutant cortical VZ lateral expansion is greatly compromised, and a progressively reduced cortical VZ area is found from this time forward. For example, at E15.5, the cortical VZ surface area is less than 50% of the control littermate brains (∼60% mediolaterally×∼70% rostrocaudally). At E12.5, the Id4 mutant VZ is thicker and this thickening can be attributed to a prolonged G1 phase of the cell cycle (Fig. 4 and Table 1). This phenotype is consistent with in vitro studies indicating that ID genes enhance proliferation by facilitating the G1-S transition. For example, ID2 has been shown to bind to the RB (Retinoblastoma) protein thereby enhancing the G1-S transition (Iavarone et al.,1994), and other ID genes have been shown to regulate the expression of the G1 cyclin-dependent kinase inhibitors p16 and p21 at the transcriptional level (Alani et al.,2001; Lasorella et al.,1996; Prabhu et al.,1997). ID genes are in turn regulated during the cell cycle where their peak level of expression is observed early in G1, after serum stimulation (Christy et al.,1991; Hara et al.,1994). However, our analysis of these G1-S regulators in the Id4^-/- telencephalon did not reveal significant changes in their expression levels (data not shown), suggesting that Id4 does not regulate their expression during cortical development. We are currently examining the possibility that ID4 interacts with these molecules at the protein:protein level to affect their function.

Our findings provide evidence supporting an integrated proliferative regulatory mechanism that coordinates the timing, the number and the neuronal specificity of cortical neurons, as proposed by Caviness and his colleagues in their theoretical model (Caviness et al.,2003). Their model predicts that increasing the proportion of daughter cells exiting the cell cycle early will lead to reduced lateral expansion of the neuroepithelium, as a result of the reduced proportion of daughter cells available to expand the proliferating population. It also predicts that the early cortical plate will be transiently thicker (because of an increased output of postmitotic cells), but significantly thinner at the end of the neurogenic period because of the loss of stem cells early on. In addition, an increase in the proportion of daughter cells exiting the VZ is associated with a lengthening of the cell cycle time, particularly G1, during normal development. With the exception of a thicker, rather than thinner,cortical plate in the Id4 mutant at E18.5, the Id4 mutant phenotype is consistent with this model. Hence, our findings generally support the existence of an integrated regulatory mechanism, and identify Id4as a key regulator that controls the time domain, the output domain, and neuronal specification during cortical development.

We were intrigued by the region-specific requirement for ID4 function, and the relatively mild mutant phenotype at P0 despite the early loss of stem cells due to both premature differentiation and compromised cell cycle transition. As Id1, Id2 and Id3 are also expressed in the developing telencephalon, these ID genes may functionally compensate for Id4 in non-affected areas. For example, mutational studies of Id1 and Id3 have revealed that they are functionally redundant during telencephalic development. To examine this possibility, we monitored expression levels of other ID genes in Id4 mutant telencephalon. Indeed Id1, Id2 and Id3 expression levels were increased in the Id4 mutant, when compared with a control littermate (Fig. 2E′ and not shown). In particular, Id2 expression was altered both spatially(Fig. 2E,E′) and quantitatively (not shown). Hence, we examined possible functional redundancies between Id2 and Id4 by analyzing Id2/4double null embryos. We were unable to detect evidence of altered telencephalic development in Id2^-/- animals (data not shown). Interestingly, the brain size of the Id2/4 double null animal was only slightly smaller than that of the Id4 single mutant (data not shown), and no obvious defects were found in other areas of the telencephalon. This observation indicates that Id2 is not necessary to compensate for Id4 in other areas, and supports the idea that ID4 function is unique and necessary for the development of the telencephalic midline tissue. Id2/4 double mutant analysis also suggests the existence of other compensatory mechanisms, either at the molecular level by Id1 or Id3, or at the cellular level (see below).

In contrast to the reduced proliferation observed in the early neuroepithelium, the number of proliferating cells outside of the VZ in the Id4^-/- brain after E15.5 is dramatically increased(Fig. 4). These ectopically positioned cells proliferate, as indicated by both S-phase and M-phase markers, and differentiate, as indicated by NeuroD expression, in ectopic positions. These results indicate a potential compensation mechanism at the cellular level for the reduced proliferation of early VZ progenitors, and may account for the relatively mild Id4 mutant phenotype observed at birth.

Ectopically positioned proliferating cells found in the Id4^-/- cortex are reminiscent of changes observed in animals lacking the Rb gene. In Rb^-/- animals,ectopically positioned proliferating cells are found in the cortex, and this phenotype has been interpreted as indicating that differentiating mantle cells have aberrantly re-entered the cell cycle in the absence of functional RB(Clarke et al., 1992; Lee et al., 1992; Lipinski et al., 2001; MacPherson et al., 2003). We therefore examined Rb levels in Id4^-/- animals,but we did not observe a significant difference in its expression, either at the RNA or protein level (not shown). ID2 has been shown to bind RB protein and to inhibit its function post-translationally(Iavarone et al., 1994). Hence, a possible explanation for the observed phenotype is that the increased level of Id2 expression in the Id4 mutant(Fig. 2E′) leads to inactivation of RB. However, direct examination of this hypothesis indicates that this phenotype is not the result of an epigenetic event (increased Id2 levels). The Id2/4 double mutant cortex also contained many ectopically proliferating cells in the mantle region (not shown). These observations suggest that if the Id4 phenotype were mediated through the RB pathway, ID4 most likely acts to enhance RB-mediated inhibition of proliferation, in contrast to ID2, which negatively regulates RB. Among the many possibilities, direct interactions can be envisioned in which ID4 competes with ID2 for RB binding in differentiating neurons. Or, ID4 may directly interact with other molecules in the cell cycle machinery, and depending on the interacting partner and cellular context, either enhance or inhibit cell cycle transition.

ID genes have been shown to inhibit differentiation in many different cell types in vitro and in vivo (Norton,2000). The main mechanism of this potent inhibitory effect is attributed to its ability to form dominant-negative dimers with bHLH proteins. In the developing cortical progenitor zone, at least three proneural bHLH genes, Ngn1, Ngn2 and Mash1, are important for progenitor proliferation, cell fate specification, and differentiation(Fode et al., 2000; Nieto et al., 2001; Parras et al., 2002; Ross et al., 2003). As ID proteins are thought to act as negative regulators of bHLH proteins by sequestering available E-proteins, we have postulated that proneural bHLH proteins are hyper-active in the Id4 mutant cortex. Consistent with this model, the Id4 mutant phenotype correlates well with both loss-of-function and gain-of-function studies of these proneural genes(Fode et al., 2000; Parras et al., 2002; Sun et al., 2001). For example, ectopic expression of Ngn1 clearly promotes neuronal differentiation (Sun et al.,2001), and the simultaneous loss of Ngn2 and Mash1 function results in reduced neurogenesis (and NeuroDexpression), particularly in the dorsomedial cortex(Fode et al., 2000). In Id4^-/- animals, where proneural bHLH proteins are presumably free from ID4 inhibition, neural stem cells prematurely differentiate and this is most obvious in the dorsomedial region(Fig. 2D,D′, Fig. 3A,A′,C,C′). It is unclear, at present, why the dorsomedial region of the telencephalon is particularly sensitive to both positive and negative HLH protein activity, but these in vivo studies reveal an important requirement for bHLH transcription factor function in this area, and emphasize the role of Id4 in regulating its development.

CNS development requires complex molecular interactions that control progenitor proliferation and differentiation, and Id4 can now be recognized to be important for these processes. Our finding that Id4inhibits neural progenitor differentiation is consistent with the assignment of this function to other ID family members. In addition, we provide in vivo evidence for a novel function of Id4 in preventing aberrant cell cycle entry in differentiating neurons. These observations underscore the importance of cell context for ID gene function, and in particular for Id4, which either enhances or blocks cell cycle transition in the same developing tissue at different times. This study reveals a crucial role for Id4 in cortical development, and identifies the molecular pathways over which its effects are mediated, while suggesting important new areas for future study.

We thank Sharon Liu, Jian Yang and Eric DuFour for their excellent help with animal care. We also thank John Hill for sharing the ID gene RT-PCR primers we used, and Nathan Watson for his assistance with western analysis. We thank Dr Bernard F. Cole for his advice on statistical analysis. This study was supported by NIH grant #CA84280 (M.A.I.), the Betz Foundation (M.A.I.),and the Jordan and Kyra Memorial Fund (M.A.I.).