TI - Discussion . AB - We report here a requirement of HDAC activity for the generation of neurons from the GE of the embryonic mouse telencephalon and a modulatory role of HDACs in cortical neurogenesis . TSA , a potent and specific inhibitor of class I and II HDACs , was seen in the GE to greatly reduce neurogenesis in vivo and almost completely block neurogenesis in vitro . In contrast , in the cortex neurogenesis was actually enhanced upon HDAC inhibition . We have identified the BMP2/4 signaling pathway as the target of HDAC activity , both in the GE and the cortex . During brain development HDACs inhibit BMP2/4 signaling , as this signaling pathway is upregulated , beginning at the transcriptional level , upon inhibition of HDACs by TSA . We observed that TSA inhibition of HDAC activity resulted in an increase of histone H3 and H4 acetylation levels already within one hour of drug application , suggesting that corresponding changes in gene expression could take place very quickly , as has been reported before [62] . In principle , the suppression of BMP signaling could be accomplished at the transcriptional level in a number of ways , including the ligands BMP2/4 , their receptors and genes in the signaling pathway downstream of the BMP ligands , such as the Smad family members . Indeed , within 12 hours of TSA application to cortical or GE-derived precursor cultures , we saw an upregulation in the expression of Bmp2 , and a down-regulation in the inhibitory cytoplasmic factor Smad7 . Both of these changes could contribute to an upregulation in the BMP2/4 signaling pathway . To prove this point , extracellular inhibition of BMP2/4 activity by noggin or Alk3-ECD restored both neurogenesis and astrogliogenesis to normal levels in TSA-treated cultures derived from both cortex and GE . Although noggin can inhibit other BMP family members , the binding and inhibition of BMP2 and -4 by Alk3-ECD , the high affinity receptor for these growth factors , identifies this BMP subfamily as responsible for the reported effects . Together , the two complementary transcriptional effects in Bmp2 and Smad7 may act in a synergistic fashion to tip the balance from neurogenesis to astrogliogenesis in the GE , and to further reinforce neurogenesis in the cortex . We propose that an important role of HDACs in GE-derived neural progenitor cells is to suppress the responsiveness of the progenitors to astrocyte fate-promoting signals from BMP2 and BMP4 . Two lines of evidence suggest that these two factors are produced by the progenitors themselves . In vitro , the cultures are homogenous at the level of nestin staining in the minimal time period in which TSA exerts its effects , i.e.in the 24 hour period before bFGF withdrawal ( Fig 4C , D , E ) . The observed levels of Bmp2 and Bmp4 gene expression are therefore probably reflective of production by the nestin-positive progenitors . In vivo , direct identification of Bmp2 - and Bmp4-expressing cells by in situ hybridization , followed by the labeling of radial glia cells with an antibody recognizing BLBP , clearly identifies radial glia as a significant minority of the cells in the ventricular zone of both GE and cortex that respond to HDAC inhibition by upregulation of Bmp2 ( Fig 11A , A' ) . During mammalian brain development , neural progenitors at first generate primarily neurons and toward the end of gestation switch to the production of astrocytes and oligodendrocytes [1] . How do BMP2/4 affect this switch? In our study , a 24-hour treatment with TSA just before bFGF withdrawal is sufficient to inhibit neurogenesis in GE-derived precursor cultures (Fig 6A) , and up to 4 cell divisions could occur between the cell type that is affected by TSA in this 24-hour period and the generation of neurons or astrocytes 3-4 days later , based upon a calculated cell -cycle length of 21 hours ( Maya Shaked , unpublished data ; [75] ) . The affected cell type could in principle be a bipotential progenitor , but it is more likely to represent a progenitor such as a radial glial cell [76] , that gives rise to separate neuroblasts and glioblasts , with BMP2/4 favoring the formation of glioblasts . Indeed , the cell type shown to respond to an upregulation of Bmp2 transcription in the developing cortex and GE proved to be almost exclusively RC2-positive neural precursor cells , as identified by nuclear staining for PHOSphorylated Smad1/5/8 ( Fig 10 ) . Interestingly , we saw an increase in vivo in the cortex in the RC2-positive population , with a corresponding slight increase in neurogenesis . This would be consistent with a neurogenic role of these precursors at this developmental time point . Indeed , it has been shown that ectopic application of BMP2 or BMP4 to acute telencephalic slices [21] or dissociated cultures [20] from this time period ( 145 dpc in the mouse , E16 in the rat ) promotes neurogenesis . Interestingly , after HDAC inhibition in the GE we saw an opposite effect upon neurogenesis and no such increase in the radial glial population . In contrast to the cortex , radial glia in the GE are poised to start generating astrocytic precursors at 14.5-15.5 d.p.c. , the timeframe of our investigation [2] . The advanced developmental state of the radial glia in the GE could explain the relative ease with which HDAC inhibition , leading to an upregulation of BMP2/4 signaling , results in a switch to an astrogliogenic program . In contrast , neocortical radial glia are in the middle of their neurogenic program in the time frame we investigated . In this context , the strong expression of the neurogenic factor Ngn1 has been shown to actively inhibit astrogliogenesis in the neocortex at this time [77] , and indeed , Ngn1 was seen to be upregulated upon HDAC inhibition in precursor cultures derived from cortical progenitors (Fig 3J) . However , Ngn1 was also seen to be upregulated by HDAC inhibition in GE-derived precursor cultures (Fig 3I) , but it clearly did not show a neurogenic effect upon the cultures , perhaps explained by the fact that its basal level of transcription is much lower in the GE than in neocortex . Finally , HDACs have catalytic activity upon proteins other than histones , including alpha tubulin [78] , p53 and CBP/p300 [36] , so there is also the possibility that HDACs regulate the acetylation of histone proteins important for BMP signaling . Clearly , other factors must be at work to explain the difference in response to HDAC inhibition in the cortex and the GE , and these differences will best be investigated by performing an unbiased gene expression analysis using microarray gene chips . It is unclear why GE precursors do not precociously generate GFAP+ astrocytes in vivo after HDAC inhibition , as they certainly do so in precursor cultures derived from the GE ( Fig 5C , D ) . Premature astrogliogenesis in vivo has been seen in several mouse mutants in neurogenic genes [9] , [79] and for the DNA-methylating enzyme Dnmt1[74] . In the Dnmt1 mutant , the astrogliogenic transcription factors Stat1 and Stat3 are upregulated at the transcriptional level , a change correlating with a demethylation of the Stat1 gene promoter . Interestingly , we also observed an upregulation of both Stat1 and Stat3 in both cortical and GE cultures upon TSA treatment ( Fig 3G , H ) , without seeing any increase in astrogliogenesis in cortical cultures , which could be explained by the anti-astrogliogenic effect of Ngn1 in cortical precursors [77] . Similarly , it could be that pharmacological modulation of histone acetylation levels is not sufficiently strong to overcome inhibitory signals to astrogliogenesis present in the embryonic GE , and that these inhibitory signals are either absent or reduced in in vitro cultures . Indeed , astrogliogenesis was clearly promoted in dissociated cultures prepared from the GE of embryos that had been treated with TSA in utero and then treated with either BMP2 , LIF , or both factors together ( Fig 9C ) . However , the nature of these signals remains unclear . Which HDAC is responsible for the control of neurogenesis in the cortex and GE? Of the five different HDAC inhibitors used in this study , only VPA , which specifically inhibits class I HDACs at the concentrations we have used [58] , did not show any affect upon neurogenesis , although it did inhibit the production and maturation of oligodendrocytes , as reported previously [43] . This suggests that class II HDACs may be responsible for controlling neurogenesis in the telencephalon . Of the class II HDACs that we examined , we detected protein expression of HDAC4 , -5 , -6 , and -7 ( Fig 4 ) . The expression of HDAC9 has also been reported at the mRNA level in differentiating neurosphere cultures [80] . As it is possible that these gene products may perform similar functions in the neural progenitors , identification of the individual HDACs involved in neurogenesis may be difficult . In this respect , comparison with myogenesis may prove useful . In the mouse , individual knock-outs of Hdac4[48] , Hdac5[50] and Hdac7[52] have not been reported to show a disruption in the skeletal myogenesis program , although all three proteins are clearly important for the in vitro differentiation of myoblasts to myotubes [36] . In addition , HDAC5 and HDAC9 have been reported to play functionally redundant roles in heart development in vivo[50] . Possible redundancy of HDAC function in neurogenesis from neural stem cells is currently under investigation through sequential and combinatorial knockdown experiments using RNA interference . It will also be important , however , to undertake a closer examination of the various HDAC knockout strains for defects in nervous system development . Our results stand in agreement to a previous report [81] showing that several HDAC inhibitors , including TSA , VPA and sodium butyrate , dramatically increase neurogenesis from neurospheres derived from adult rat hippocampus , which develops as an invagination of the dorsal telencephalon at the same time as the cortex . We did not see such a large increase in neurogenesis in embryonic cortex in vivo , but we did observe a doubling of neurogenesis in vitro . In addition to the difference in species and developmental stage examined in these two studies , these results may reflect the different roles that BMPs play in neurogenesis in various regions of the brain . In the developing cortex , BMP2/4 promote the formation of neurons [21] , but in contrast they have been shown to inhibit neurogenesis in neural stem cells derived from the subventricular zone of the embryonic GE [14] . Our data are consistent with an inhibitory role of BMP2/4 upon neurogenesis in the ventral telencephalon . It is of interest to note that ablation of hdac1 in the embryonic zebrafish also leads to a reduction in neurogenesis , both in the retina [46] and in motor neurons of the spinal cord [47] . In the former case , Hdac1 was found to suppress both Wnt and Notch signaling . We investigated both of these pathways and found that neither of them seem to play a role in the HDAC -mediated promotion of neurogenesis , atleast in GE-derived neural progenitor cells . Whether this reflects a vertebrate class difference between the two model organisms or differences in HDAC function between various regions of the nervous system is unclear .