Af9/Mllt3 interferes with Tbr1 expression through epigenetic modification of histone H3K79 during development of the cerebral cortex

Abstract

Mutations of leukemia-associated AF9/MLLT3 are implicated in neurodevelopmental diseases, such as epilepsy and ataxia, but little is known about how AF9 influences brain development and function. Analyses of mouse mutants revealed that during cortical development, AF9 is involved in the maintenance of TBR2-positive progenitors (intermediate precursor cells, IPCs) in the subventricular zone and prevents premature cell cycle exit of IPCs. Furthermore, in postmitotic neurons of the developing cortical plate, AF9 is implicated in the formation of the six-layered cerebral cortex by suppressing a TBR1-positive cell fate mainly in upper layer neurons. We show that the molecular mechanism of TBR1 suppression is based on the interaction of AF9 with DOT1L, a protein that mediates transcriptional control through methylation of histone H3 lysine 79 (H3K79). AF9 associates with the transcriptional start site of Tbr1, mediates H3K79 dimethylation of the Tbr1 gene, and interferes with the presence of RNA polymerase II at the Tbr1 transcriptional start site. AF9 expression favors cytoplasmic localization of TBR1 and its association with mitochondria. Increased expression of TBR1 in Af9 mutants is associated with increased levels of TBR1-regulated expression of NMDAR subunit Nr1. Thus, this study identified AF9 as a developmental active epigenetic modifier during the generation of cortical projection neurons.

AF9/MLLT3 is one of multiple fusion partners of the histone methyltransferase MLL1 in acute leukemia (1, 2). However, mutations of the AF9/MLLT3 gene alone are not associated with leukemia but are implicated in anterior homeotic transformations during mouse development (3), and in neurodevelopmental diseases, such as mental retardation, epilepsy, and ataxia in human patients (4, 5). Little is known about AF9 function in the CNS, although its expression pattern implies a role during development of the forebrain and cerebellum (6). In the mouse forebrain, Af9 is transcribed in the subventricular zone (SVZ), a neurogenic compartment that harbors progenitors for upper layer neurons (7, 8). Af9 is also expressed in neurons dispersed over all cortical layers and diverges in this respect from other SVZ markers, such as Svet1 and Cux2 (6). On the molecular level, AF9 mediates transcriptional activation and was classified as a proto-oncogene (9). The AF9 protein interacts with many different factors and has been implicated in different cellular processes (9–11). In the extensive network of interacting proteins, AF9 associates with DOT1L (12, 13), the main enzyme responsible for histone H3 methylation at lysine 79 (14–16). Dot1 is implicated in UV damage repair (17), and affects gene expression in yeast, flies, and mammals (18, 19), whereas histone H3 lysine 79 (H3K79) methylation correlates with gene activation (20) and suppression (21). Following this line, AF9-DOT1L complexes mediate transcriptional activation through increased levels of dimethylated H3K79 (13), but are also capable of transcriptional repression through hypermethylation of H3K79 at the ENaCα promoter (12). Although increased levels of H3K79 methylation might be related to aging of the brain (22), the function of H3K79 methylation in the development or function of the CNS has not been studied in detail. However, other positions of histone H3 methylation mediate gene activities associated with developmental and cognitive functions in the CNS (23–28). The growing amount of data describing the implications of epigenetic modifications in neurodevelopmental disorders underlines the necessity to investigate and understand the role of histone modifications during normal brain development. We report on Af9 function in the developing forebrain and show that Af9 expression prevents premature depletion of progenitors during corticogenesis. AF9 suppresses Tbr1 expression in neurons and alters H3K79 dimethylation at the Tbr1 transcriptional start site concomitant with decreased levels of RNA polymerase II (RNAPolII). This study provides unique evidence that regulation of H3K79 methylation during cortical development and differentiation of neuronal subtypes may contribute to a specific layer identity. Furthermore, this epigenetic modification indirectly influences the NMDA receptor layout of projection neurons.

Results

Loss of Af9 Results in Reduced Numbers of Cortical Progenitors.

Because mutations of human AF9 are associated with neurodevelopmental diseases, we analyzed the Af9 mouse mutant (Af9^−/−) with regard to developmental defects of the CNS. Because Af9^−/− mice die shortly after birth, we concentrated on embryonic and early postnatal stages. As shown in Fig. 1A, the brains of P0 Af9^−/− mice did not show any gross abnormalities, despite an improper bundling of callosal fibers that was accompanied by a displacement of ventricular cells and slightly enlarged ventricles (Fig. S1 A and B). Various mouse mutants of factors involved in cortical development also display defects in the corpus callosum. We therefore judged the bundled callosal axons as evidence that a detailed analysis of Af9^−/− mice might also potentially reveal variations in the cortical plate.

Fig. 1.

Loss of Af9 results in reduced proliferation of cortical progenitors. (A) Nissl-stained cerebral cortex at P0 of WT and Af9^−/− littermates, showing no gross abnormalities despite a displacement of ventricular cells accompanied by improper bundling of fibers (arrowhead). (Scale bar, 200 μm.) (B) During main stages of neurogenesis, fewer progenitors incorporate BrdU after a pulse at either E11.5, E13.5, or E15.5 and immunohistochemical (IHC) analyses at P0 in Af9^−/−, reaching statistical significance at E15.5; t test, n = 3 for Af9^−/−, as well as controls for E11.5, n = 2 for Af9^−/− and controls for E13.5 and E15.5 BrdU injections, ±SEM; *, P < 0.05. (C) Af9^−/− progenitors incorporate less BrdU at E17.5 after a 30-min pulse. For statistical analyses, data of control animals from different litters were set to 100% and Af9^−/− data are expressed as percentage of control levels; t test, n = 4 for control and Af9^−/−, ±SEM; **, P < 0.001. (D) BrdU immunostaining at E17.5 after a 30-min pulse of control and mutant cortices. (Scale bar, 100 μm.)

Af9^−/− embryos were smaller than control mice (Fig. S1C). Because Af9 is expressed in progenitors located in the SVZ of the developing neocortex, we analyzed whether cell proliferation might be affected within the neocortex. BrdU labeling at different stages of development revealed fewer cells in S-phase at all main stages of neurogenesis in Af9^−/− cortices (Fig. 1B). This loss of progenitors is statistically significant only during later stages of neurogenesis, namely at E15.5 (Fig. 1B) and E17.5 (Fig. 1 C and D). Analyses of cell proliferation in the thymus revealed an increased number of cells in M-phase in Af9^−/− compared with WT littermates, indicating that decreased proliferation in the cerebral cortex was not a consequence of general growth retardation (Fig. S1 D and E). To analyze whether Af9 mutation affected not only cell proliferation but also migration into upper positions during cortical development, we injected BrdU at different time points (E11.5, E13.5, and E15.5) and analyzed the distribution of BrdU-positive cells at P0. As shown in Fig. S2, Af9^−/− mice displayed fewer BrdU-positive cells than control littermates, but despite their decreased numbers, these cells were mostly found in the correct positions. Mutants labeled at E15.5 showed a substantial loss of cycling cells in apical regions, indicating that the pool of progenitors might be depleted during development. Thus, Af9 mutation seemed to interfere with the proliferation of progenitors, especially during late stages of neurogenesis, but not with the capacity to migrate into the developmentally determined cortical layer position.

Because Af9 mutation affected proliferating progenitor cells, we further analyzed this cell population at E17.5 with PAX6 and TBR2 antibodies. PAX6-positive radial glia seemed reduced in some individuals (Fig. 2A). However, quantification of PAX6-positive cells did not reveal significant differences (Fig. 2B). In contrast, Af9^−/− brains had significantly fewer TBR2-positive intermediate precursor cells (IPCs) than WT littermates at E17.5 (Fig. 2 A and C). This loss was most pronounced at the pallial/subpallial boundary. Analyses of PAX6 and TBR2 expression at E15.5 did not reveal significant changes between Af9^−/− and WT (Fig. S3 A and B), indicating that Af9 mostly influenced progenitors during late stages of neurogenesis.

Fig. 2.

Mutation of Af9 leads to a loss of TBR2-expressing cortical progenitors at E17.5. (A) PAX6- and TBR2-positive IPCs are reduced in E17.5 Af9^−/− mice. (Scale bar, 100 μm.) (B) Quantification of PAX6-positive cells in WT and Af9^−/− animals, n = 2 for both, ±SEM. (C) Quantification of TBR2-positive cells in WT and Af9^−/− animals. t test, n = 10 for Af9^−/−, n = 8 for WT, ±SEM; *, P < 0.05; **, P < 0.001. Given are the numbers of positive cells per square millimeter in the respective bin.

Because we could not detect increased apoptosis in Af9^−/− cortices, we investigated the progenitors that stayed in the cell cycle. We labeled cycling cells with BrdU at E16.5 for 24 h and analyzed for BrdU and KI67-positive cells (Fig. 3 A and B). Loss of Af9 resulted in a significantly smaller cycling fraction (Fig. 3C). Thus, Af9 expression is partially necessary to maintain the pool of progenitors during cortical development.

Fig. 3.

Af9 mutation leads to a premature exit from the cell cycle. (A and B) BrdU (green) and KI67 (red) double labeling of E17.5 Af9^−/− and Af9^+/− cerebral cortex after a 24 h BrdU pulse. (C) Quantification of BrdU and BrdU/KI67-positive cells, one-tailed t test, n = 2 for ^−/− animals, n = 3 for controls, ±SEM; *, P < 0.05. (Scale bar, 50 μm.)

Loss of Af9 Increases TBR1-Positive Neurons.

It has been shown that differentiation of cortical progenitors from radial glia occurs over IPCs to postmitotic projection neurons. This developmental process is characterized by the expression of the transcription factors PAX6, TBR2, and TBR1 in a sequential series that corresponds to the progressing development of progenitors toward a neuronal subtype (29). Because Af9 mutation led to a loss of TBR2-positive progenitors, we hypothesized that the consequence would either be a reduced number of differentiated neurons (e.g., TBR1-positive neurons) or in case of premature exit of the cell cycle, an increase in a subset of cortical neurons. Analyses of TBR1-positive cells revealed a significant increase in the number of TBR1-expressing cells in the mutant neocortex (Fig. 4 A and B). These TBR1-positive cells were not only found in the lower layers (subplate and layer VI) with strong TBR1 expression, but were increased in upper layers II/III with usually lower TBR1 expression levels (30, 31). This increased TBR1 expression after Af9 knockout was corroborated by Western blot analysis of proteins from control and mutant cortices (Fig. 4C). To investigate whether other layers of the neocortex were affected by the loss of Af9, we performed Western blot (Fig. 4C) and quantified immunohistochemical (IHC) analyses (Fig. S3C). Af9 expression overlaps with Svet1 and Cux2 in the SVZ of the neocortex, two proteins involved in the generation of upper layer neurons (7, 32). We therefore reasoned that Af9 mutation might influence upper layer neuron formation similar to Cux2, where mutants display an increase in the number of upper layer neurons. However, we found no evidence for a general change in the quantity of upper layer neurons, as indicated by SATB2 Western blot analyses of mutant and control brain extracts (Fig. 4C). AF9 regulates expression of CTGF in nonneuronal cells, and CTGF is expressed in layer VII of the neocortex. However CTGF expression was not affected by the loss of AF9 in mutant brains (Fig. 4C). CTIP2 is expressed mainly in a subset of neurons located in layer V and weakly in layer VI, but its expression seemed unaffected by Af9 mutation (Fig. 4C). In summary, these data indicated that Af9 mutation affected TBR2-positive progenitors, which might subsequently differentiate into TBR1-positive neurons. Although postmitotic TBR1-positive neurons were significantly increased in Af9^−/− brains, we observed no indication that other layers were affected by the loss of Af9. We therefore used BrdU labeling at different developmental stages to follow this differentiation of progenitors into TBR1-expressing neurons in mutant and control cortices. As shown in Fig. S4 A to D, Af9 mutants displayed only a slightly higher differentiation of progenitors into TBR1-positive neurons at E11.5, E13.5, and E15.5. Furthermore, real-time RT-PCR assays of Tbr1 expression revealed that Tbr1 was up-regulated more than 2-fold in P0-derived mutant versus control brains (Fig. 5B). Samples generated from E16.5 embryos did not show changes in Tbr1 expression levels (Fig. S5A). Taken together, these observations indicated that AF9-mediated regulation of Tbr1 expression might mainly take place in postmitotic neurons during the differentiation from progenitors to mature neurons.

Fig. 4.

Loss of Af9 increases TBR1-positive neurons. (A) TBR1 IHC of Af9^−/− and Af9^+/− cortices at P0, showing an increase of TBR1-positive cells in mutants. Indicated is the subdivision of 10 bins for quantification (Upper), and the corresponding regions in an overview (Lower). (B) Quantification of TBR1-positive cells in the corresponding bins indicated in A. Given are the numbers of TBR1-positive cells per square millimeter per bin from WT and Af9^−/− animals. n = 6 for −/− animals, n = 3 for +/+ animals, one-tailed t test, ±SEM; *, P < 0.05. (C) Western blot analysis of proteins with different cortical layer localization showing increased TBR1 expression in P0 Af9^−/−. CTGF, CTIP2, and SATB2 did not show differences compared to GAPDH or β-TUBULIN.

Fig. 5.

AF9 interferes with nuclear TBR1 function. (A) AF9 overexpression results in a significant loss of nuclear TBR1 protein compared to residual cytoplasmic TBR1 protein. Quantification of TBR1 localization in AAV-AF9 or AAV-GFP transduced primary cortical neurons. ++, strong expression; +, moderate expression. Given is the percentage of the number of TBR1-positive cells with nuclear or cytoplasmic localization among strong and moderate AF9 or GFP expressing cells, ±SEM, n = 3; **, P < 0.001; ***, P < 0.0001. (B) Quantification of Tbr1 and Nr1 mRNA expression levels in P0 cerebral cortices of Af9^−/− and WT littermates showing a mean of 2.114- (Tbr1) and 2.167- (Nr1) fold more transcripts in −/− samples as determined by real-time RT-PCR, ±SEM, Af9^+/+ n = 3, Af9^−/− n = 3; *, P < 0.05, one-sample t test with a theoretical mean of 1 (representing no change of expression between Af9^−/− and Af9^+/+ samples). (C) Staining of TBR1 and mitochondrial marker Prohibitin in a healthy neuron (Upper) and neuron with fragmented nucleus (Lower). (Scale bars, 10 μm.)

Overexpression of AF9 Suppresses TBR1 Expression.

Because AF9 is not only expressed in progenitors but also in the cortical plate, we investigated whether the modulation of TBR1 expression would depend on the AF9 protein in postmitotic cortical neurons. We used an adeno-associated virus (AAV) that overexpressed a C-terminally AU-1 epitope-tagged AF9 under the control of the neuron-specific Synapsin promoter to visualize the expression of AF9 in cortical cultures generated from E17.5 WT mice. Cells that stained positive for AU-1 showed a characteristic focal nuclear localization of the protein (Fig. S6). Western blot of proteins isolated from transduced cultures after 12 days in vitro (DIV) showed a specific expression of AF9 as indicated by AU-1 immunoreactivity. Nontransduced and GFP-transduced control cells did not express the approximately 64 kDa recombinant protein (Fig. 6A). Overexpression of AF9 resulted in a loss of the TBR1 protein in a titer-dependent manner, though TBR1 expression was unaffected when a control virus expressing GFP was applied (Fig. 6A). To further prove that AF9 suppressed TBR1 expression, we analyzed AF9 overexpression in vitro and in vivo. In vitro, neurons strongly overexpressing AF9 did not express TBR1, and vice versa: strong expression of TBR1 correlated with low or undetectable AF9 expression. Expression of GFP did not correlate with strong or weak TBR1 expression but rather showed uniform distribution. We quantified the TBR1 expression level against the expression levels of AF9 and GFP, respectively, and showed an inverse correlation of AF9 and TBR1 expression (Fig. S6 B and C). On the other hand, TBR1 expression seemed independent of the expression of GFP, indicating that suppression of TBR1 by AF9 was not a result of nonspecific effects caused by the viral transduction procedure. We next injected AAV-AF9 in the frontal brain of newborn mice and analyzed at P23 and P34 for AF9 and TBR1 expression. As shown in Fig. 6 B and C, strong AF9 overexpression in vivo also correlated with weak TBR1 expression, and vice versa. Quantification of these in vivo results corroborated our in vitro findings of TBR1 suppression by AF9 (Fig. 6B), showing a significant increase of TBR1-negative (Tbr1–) cells among cells with strong AF9 (Au-1++) expression compared with cells with moderate AF9 (Au-1+) expression. Furthermore, we counted significantly increased numbers of moderate TBR1- (Tbr1+) expressing cells among moderate AF9- (Au-1+) expressing cells compared with strong AF9- (Au-1++) expressing cells.

Fig. 6.

Overexpression of AF9 suppresses TBR1 expression in vitro and in vivo. (A) Western blot analysis of transduced primary cortical neurons. Transduction of AAV-AF9 or AAV-GFP (3 × 10⁷ or 6 × 10⁷ transforming units per 250,000 cells) results in the overexpression of the recombinant AF9, as indicated by the AU-1 band, absent in AAV-GFP-transduced cells or untransduced controls. TBR1 expression is decreased with increased AF9, although unaffected through overexpression of GFP. (B) Quantification of TBR1 compared to AF9 expression in vivo in adult cortices (P23 or P34) transduced with AAV-AF9 at P3. Given is the percentage of the number of TBR1-positive cells of a certain expression level among strong and moderate AF9-expressing cells. ++, strong expression; +, moderate expression; -, weak expression; –, no expression. n = 4, student's t test; *, P < 0.05; *¹, P < 0.05 (one-tailed); **, P < 0.001, ±SEM. (C) IHC analysis of adult cortices transduced with AAV-AF9/AAV-GFP at P3. (Upper) AU-1 and TBR1 costaining at lower and higher magnification. (Lower) GFP and Tbr1 expression. In vivo, strong AF9 expression (arrowheads) correlated with weak TBR1 expression, and vice versa: strong TBR1 expression (arrows) correlated with weak AF9 expression. GFP expression correlated with strong (*), moderate (#), or lacking (arrowhead) TBR1 expression. TBR1 expression was also observed in GFP-negative cells (arrows). (Scale bars: first row, 50 μm; second row, 20 μm; third row, 20 μm.)

AF9 Directs H3K79 Dimethylation in the Neocortex.

We next investigated whether AF9 represses Tbr1 transcription by direct association with the Tbr1 promoter. Cortical neurons were transduced with AAV-AF9 and AAV-GFP and after 12 DIV subjected to ChIP with nonspecific IgG or anti-AU-1 antibodies. Untreated cells served as an additional control. Real-time PCR with primers surrounding position −310, the transcriptional start site (TSS, +1), +145, and +15,000 of the mouse Tbr1 gene revealed that the AU-1 antibody specifically immunoprecipitated the AF9 protein that associated with the Tbr1 gene with an increased enrichment near the TSS (Fig. 7A). To analyze whether overexpression of AF9 affected transcriptional initiation and RNAPolII recruitment, we performed ChIP analysis of RNAPolII on different parts of the Tbr1 gene after overexpression of AF9 or in the respective controls. AF9 overexpression led to a statistically significant reduced association of RNAPolII with the Tbr1 TSS (Fig. 7B). These data suggested that AF9 suppressed Tbr1 expression in mature cortical neurons by interfering with transcriptional initiation and RNAPolII recruitment.

Fig. 7.

AF9 associates with the Tbr1 transcriptional start site and reduces RNAPolII binding through hypermethylation of H3K79. (A) ChIP of AAV-AF9 and AAV-GFP transduced or untransduced primary cortical cells. Antibodies used were IgG and anti-AU-1. Precipitated DNA was assessed by real-time PCR at the transcriptional start site (TSS), and +15.000. Given is the amount of precipitated DNA relative to the respective input DNA ±SEM of triplicates. Dotted line indicates the mean amplification of IgG control samples. (B) ChIP of primary cortical cells transduced with AAV-AF9 and AAV-GFP or untransduced. Antibodies used were IgG and RNAPolII, real-time PCR and presentation of data as in A. t-test; *, P < 0.05; **, P < 0.001. (C) ChIP of untreated cultured primary cortical neurons (12 DIV) with IgG and anti-H3K79me2. DNA was amplified from the positions −310, TSS, and +145, data presentation as in A. (D) ChIP of WT cerebral cortex tissue with IgG and anti-H3K79me2. Presentation of data as in C. (E) ChIP of Af9^+/+ and Af9^−/− P0 cerebral cortices with anti-H3K79me2 antibody, one-tailed t test; *, P < 0.05; **, P < 0.001.

Because AF9 interacts with the methyltransferase DOT1L, which mediates di- and trimethylation of K79 of Histone H3 (H3K79me2, me3), we next investigated whether the observed decrease of Tbr1 expression through AF9 activity was connected to a change in H3K79me2. We performed ChIP with chromatin isolated from WT neocortex as well as cultured cortical neurons and examined H3K79me2 at different regions in and flanking the Tbr1 gene. As shown in Fig. 7 C and D, we were able to precipitate several fragments of the Tbr1 gene from both samples using the H3K79me2 antibody. To analyze for differences in the H3K79me2 pattern in Af9^−/− cortices, we performed ChIP from mutant and control forebrains. Fig. 7E shows that Af9 mutation led to a reduction of H3K79me2 at all sites on the Tbr1 gene that we investigated. These data showed that AF9 modulated the H3K79me2 pattern at the Tbr1 locus, probably suggesting that AF9-directed H3K79me2 near the transcriptional start site suppressed Tbr1 transcription.

AF9 Interferes with Nuclear TBR1 Function.

Our analyses of AF9 overexpression in cortical neurons not only revealed a reduced expression of TBR1, but indicated that residual TBR1 protein was found in the cytoplasm of the neurons. Nuclear protein was nearly undetectable in AAV-AF9-transduced neurons, while significantly more neurons retained nuclear TBR1 in AAV-GFP-transduced cells (Fig. 5A). Cytoplasmic TBR1 has been reported in adult rodent brain and was shown to be localized to synaptosomes (33). These authors proposed that synaptic TBR1 regulates expression of NMDA receptors that is dependent on the nuclear localization of TBR1 (34). To further analyze AF9 function in this context we investigated whether loss of AF9 would not only result in increased TBR1 expression but also in increased expression of NMDAR. As shown in Fig. 5B, consistent with known TBR1 function, loss of AF9 resulted in a more than 2-fold increase in the expression of Tbr1 and Nr1 in real-time RT-PCR assays. To further investigate the cytoplasmic localization of TBR1, we performed immunocytochemical stainings against a subset of synaptic markers including Bassoon, Gephyrin, NR1, PSD95, Synapsin, and Synaptophysin, but observed no colocalization with TBR1 (Fig. S5C). In contrast, we found partial colocalization of TBR1 with the mitochondrial marker Prohibitin (Fig. 5C). Although healthy cells showed partial overlap of TBR1 with Prohibitin (Fig. 5C, Upper), the colocalization was more apparent in neurons with fragmented nuclei, which were presumably undergoing apoptosis (Fig. 5C, Lower).

Discussion

In this study we have shown that AF9 contributes to the development of cortical neurons and thereby to the composition of the six-layered cerebral cortex. In this context, AF9 controls proliferation of progenitors and prevents a premature exit from the cell cycle, as indicated by fewer BrdU-incorporating cells as well as fewer IPCs in Af9 mutants. In differentiating IPCs and their neuronal derivatives, AF9 prevents the expression of TBR1 and therefore determines their upper layer identity, as indicated by increased TBR1-positive cells in mutants and less TBR1 in AF9 overexpressing neurons.

Af9 is expressed in cells of the SVZ and in neurons located in all layers of the developing cerebral cortex. Studies of other molecular markers of the SVZ, namely Svet1 and Cux2, suggested that Af9 might influence upper layer formation as postulated for Svet1 and shown for Cux2. A comparison of the mutant phenotypes of Af9 and Cux2 cerebral cortices shows that both genes exert opposing effects on the pool of progenitors. Although Af9^−/− brains have fewer proliferating cells, as indicated by markers such as BrdU, pHH3, and TBR2, Cux2 mutants incorporate more BrdU and have more pHH3- as well as TBR2-positive cells. As a consequence, Af9^−/− brains are smaller than controls and contain fewer BrdU-positive cells in upper layer positions of the cerebral cortex. However, Cux2^−/− brains are bigger, display more BrdU-positive neurons in upper layers, and have tightly packed upper layers (8). Thus, both AF9 and CUX2 differentially influence IPCs, where CUX2 promotes cell cycle exit at the time of upper layer formation and AF9 prevents premature exit from the cell cycle and further inhibits the acquisition of lower layer identity by suppressing TBR1 expression at the time of upper layer formation.

In addition to its role during proliferation, AF9 suppressed TBR1 expression in mature neurons. In vitro and in vivo overexpression showed that strong expression of AF9 correlated with weak or lacking TBR1 expression, thus indicating that sustained repression of TBR1 in mature neurons might be necessary for proper development of the neocortex. This suppression of TBR1 expression in mature neurons by AF9 might also explain why AF9 is expressed in neurons over the entire cortical plate, which differs from CUX2 (6).

To further investigate how AF9 might control TBR1 expression, we exploited the fact that Af9 interacts with DOT1L, which is responsible for H3K79me2 and me3. There is a strong correlation between H3K79me2/me3 and transcriptional activity (19). Our data support such a correlation with regard to the Tbr1 promoter that is subject to H3K79me2, the pattern of which was changed according to the absence of AF9. At the transcriptional start site of Tbr1, we observed decreased methylation in Af9 mutants. Decreased methylation through loss of Af9 corresponded with increased expression of TBR1. This finding of AF9-dependent transcriptional control via H3K79me2 supports other studies, reporting the AF9-dependent regulation of transcription through H3K79 methylation (12, 13). In the latter study, presence of AF9 is also coupled to an increase in H3K79me2 at the ENaCα promoter and its transcriptional repression.

However, our analyses of the Af9 mutation indicated a general increase in H3K79me2 in cortical and cerebellar protein extracts (Fig. S5B), and we therefore speculate that genetic loci other than Tbr1 will also be affected through AF9-mediated H3K79 methylation, possibly associated with a depression of methylation. AF9 mediates H3K79me2 by recruiting the methyltransferase DOT1L, but because AF9 has an extensive network of interacting proteins, AF9 might also recruit other factors that exert an inhibitory effect on the DOT1L activity, at least in neurons. Polycomb repression complexes (PRC), which inhibit initiation of transcriptional elongation, have been discussed as potential counter players to DOT1L, and PRC-repressive effects might be overcome by H3K79me2/3 (19). AF9 interacts with a member of the Polycomb group, MPC3 (11), which is also expressed in the CNS (35) and might therefore be a candidate for repressing the DOT1L-mediated methylation in the presence of AF9 in cortical neurons. PRC itself is an epigenetic factor that regulates transcription via histone modifications, such as ubiquitination of H2A and methylation of H3K27 (36, 37). The balancing of different posttranslational histone modifications may be an important feature for cell fate determination.

Our data clearly show that cell identity can be altered through epigenetic modifications during development (e.g., from a TBR1-negative into a TBR1-positive state). We further started to analyze whether these changes relate to distinct neuronal functions. TBR1 itself regulates the expression of NMDAR subunits in association with CASK (34). Accordingly, increased TBR1 expression through AF9-directed epigenetic modification resulted in increased transcription of Nr1, which might cause changed receptor composition of glutamatergic neurons. This result might potentially lead to different functions in CNS circuitry and may therefore be of significance for proper brain function. Furthermore, young neurons use glutamate as a chemoattractant (38, 39), which specifically influences migration into upper layer positions during cortical development. Although our initial analyses did not indicate a migratory defect in Af9^−/− mice, it is tempting to speculate that Af9 mediated down-regulation of Tbr1 and specific NMDAR gene expression might result in a variation of NMDAR layout. Such variation is observed during the switch from a migratory to a postmigratory neuronal phenotype. Further studies will be necessary to further test this hypothesis. Dysregulations of NMDAR subunits are also associated with different neurological and psychiatric disorders, such as epilepsy (40) and schizophrenia (41), and we hypothesize that this dysregulation might potentially account for some of the malfunctions observed in patients with AF9 mutation.

Epigenetic modifications, such as H3K4 or H3K27 methylation, have been shown to be associated with important neuronal promoters, such as GABAergic interneuronal Gad1, where altered patterns are associated with schizophrenia (26). Increased interneuronal differentiation has been shown after demethylation of H3K27 at the Dlx5 promoter (23), and in adult neural stem cells demethylation of H3K27 through MLL1 favors neuronal instead of glial differentiation through activation of Dlx2 transcription (24). To our knowledge our findings are unique in showing an implication of H3K79 methylation during developmental processes in general and specifically during neuronal differentiation and subtype specification of layer-specific neurons in the neocortex. Thus, our data contribute to the growing number of data that report on the importance of epigenetic modification for proper brain development and function.

Materials and Methods

Primary Cortical culture, AAV Vectors, and Viral Transduction.

Primary cortical cells were isolated and transduced as described in SI Materials and Methods. AF9 or EGFP expression in AAV vectors of the hybrid serotype 1/2 was driven by the neuron-specific human synapsin 1 promoter. AAV infections were performed at DIV3, and cells harvested at DIV12 for protein extraction and immunocytochemistry.

IHC, Western Blotting, and ChIP.

Antibodies used for IHC and Western blot experiments are listed in SI Materials and Methods. For ChIP, 2 μg of H3K79me2, AU-1, RNAPolII, and IgG antibodies were used essentially as described in SI Materials and Methods. For ChIP of cortical tissue, hemispheres were dissected in 1% formaldehyde in PBS, incubated under agitation for 15 min, and quenched with 1.25 M glycine for 5 min. After washes with PBS, tissue pellets were resuspended in IP buffer, homogenized with a Dounce homogenizer, and subsequently with a syringe. Tissue pellets were washed once with IP buffer and subsequently sonicated in IP buffer. After centrifugation the supernatant was precleared with Protein A-Sepharose and then subjected to immunoprecipitation (see Table S1 for primers used).