TI - Bidirectional control of neurogenesis by HDACs in vivo . AB - In order to determine whether the HDAC -mediated modulation of neurogenesis and astrogliogenesis observed in vitro is of physiological relevance , TSA was employed to inhibit HDACs during embryogenesis in vivo . Timed-pregnant C57BL/6J female mice were injected every 12 hours with vehicle ( 8% ethanol in 1x PBS ) or with 12.5 ug TSA in vehicle , an amount of TSA that had been previously shown to be nontoxic to murine embryos [69] . Injections began at 13.5 days post coitum ( dpc ) , a time point when neurogenesis has already begun in GE and cortex , and embryos were examined two days later ( Fig 6A ) . In order to quantify neurogenesis , we performed FACS analysis upon dissociated cortical and GE-derived cells from TSA-injected mice of the tauGFP line , in which the cDNA encoding EGFP has been inserted at the tau locus [60] . In this line , EGFP is expressed at high levels in all neurons shortly after their birth in a time frame similar to induction of beta tubulin III , and both markers can therefore be used for the quantification of neurogenesis [70] . As judged by the fraction of cells that were GFP-positive , neurogenesis in GE was reduced by 45% ( Fig 6B , D ) , whereas neurogenesis increased in the cortex modestly by 10% ( Fig 6C , E ) . The validity of FACS-sorting of the cortical and GE-derived cells was subsequently confirmed . Cells were collected with the same gating conditions that were used for the determination of the neuronal fraction of the populations , and the GFP-positive populations were examined by staining for TuJ1 , a standard marker for newborn neurons that recognizes beta tubulin III [71] . In the cortex and GE , over 90% and 80% , respectively , of the cells judged by FACS to be GFP-positive were found to be TuJ1 positive ( Fig 6F ) . TuJ1 staining of acutely-dissociated cortex and GE from embryos exposed to TSA confirmed the FACS results ( data not shown ) . We next examined the influence upon telencephalic development of HDAC inhibition , using immunohistofluorescence upon embryos after one or two days of injections starting at 13.5 d.p.c . Inspection of the entire brain revealed no apparent anatomical abnormalities . As expected , acetylation of histone H3 increased strongly in cells throughout the cortex and GE upon TSA treatment (Fig 7A) , using an antibody that specifically recognizes histone H3 . Using TuJ1 as a marker for newborn neurons , we could clearly observe a decrease in signal in GE , reflecting the decrease in neurogenesis seen with FACS analysis (Fig 7B) . Many of the newborn neurons in the medial GE start migrating tangentially into the cortex , using a route through the cortical intermediate zone [8] . Using an antibody against GABA to label these migrating inhibitory interneurons , we were clearly able to see a reduction in staining of both the GE and also the cortical intermediate and marginal zones in TSA-treated embryos , reflective of a decrease in neurogenesis in the GE (Fig 7C) . To quantitate this decrease , we counted the number of GABA-positive neurons in the ventrolateral cortex ( at the boundary with the mantle zone of the lateral GE ) , where individual GABA-positive neurons are easy to identify . We observed a 50% reduction in the numbers of GABA-positive neurons in TSA-treated cultures , corresponding to the overall decrease in neurogenesis seen with FACS analysis and TuJ1 staining (Fig 7D) . This decrease in GE-derived neurons was not caused by an increase in apoptosis of precursors or newborn neurons , as we observed very few apoptotic figures in either vehicle - or TSA-injected GE or the cortex at 15.75 d.p.c , as judged by immunolabeling of activated caspase 3 (Fig 8D) . As would be expected for the modest increase in cortical neurogenesis revealed by FACS analysis ( Fig 6C , E ) , no obvious change in TuJ1 staining could be observed in the cortex ( Fig 7B ) . Interestingly , the number of mitotic cells increased both in the ventricular ( VZ ) and subventricular zones ( SVZ ) of GE , but not in the VZ / SVZ of the cortex ( Fig 8A ) , as judged by the use of an antibody recognizing histone H3 , which becomes PHOSphorylated specifically during mitosis [72] . In order to quantify histone H3-positive cells , the VZ was delimited with a 30 um line normal to the ventricle , and all positive cells within this area were counted and displayed as the number of cells within this area along the length of the ventricular surface ( per 100 um , Fig 8A ) . SVZ cells were counted as those cells lying between 30 and 200 um basal to the ventricular surface , and displayed as the number of cells within this area ( per 104 um2 , Fig 8A ) . To identify the cells whose proliferation was increasing as a result of TSA treatment , markers for radial glial cells were examined , as radial glia are known to be precursors for both neurons as well as astrocytes in the developing cortex and GE [2] , [3] . An RC2 antibody , which detects neural precursors and radial glia during embryogenesis [73] , revealed a distinct upregulation upon TSA treatment in both GE and the cortex ( Fig 8B ) . Similarly , we observed an increase in the staining intensity of the transcription factor Pax6 in the cortex of TSA-treated animals , most prominently in the ventromedial cortex ( Fig 8C ) , and an increase in the staining intensity of BLBP , another marker for radial glia , in both GE and cortex after TSA treatment (Fig 8E) . In order to test if these changes reflected an increase in the numbers of radial glia in the TSA-treated embryos , or just an increase in the expression level in cells already expressing these markers , we plated out acutely-dissociated cells from control and TSA-treated embryos and fixed them after 2 hours , followed by a subsequent staining with the corresponding markers . Here , we clearly saw an increase in the number of cells staining positive for both RC2 and Pax6 in the cortex ( Fig 8C ) , with no change in GE ( data not shown ) . As had been seen in vitro , a reduction in neurogenesis in GE may be explained by a precocious astrogliogenesis , which had been previously seen in a genetic ablation of DNA methyltransferase 1 ( Dnmt1 ) [74] . To examine this , an antibody recognizing S100beta , a marker for newborn astrocytes , was used to examine TSA-treated embryos . An increase in staining could be observed specifically in GE (Fig 9A) , which was accompanied by an increase in the number of S100beta-positive cells detected in acutely dissociated cultures (Fig 9B) . The high background in S100beta ( 7-9% ) staining came from endothelial cells of blood vessels , seen as tubular structures in Fig. 9A , whereas the increase in staining in GE of TSA-treated embryos is derived from small punctuate cells with one or two processes ( Fig 9A , arrows ) . Whether this change reflects the birth of newborn astrocytes was difficult to ascertain , as we could not confirm their identity in vivo with a stain for GFAP , a marker for more mature astrocytes ( data not shown ) . In order to see if the in situ environment is prohibitive for the production of astrocytes at this time point , acutely-dissociated cultures were made from both GE and cortex of vehicle - and TSA-injected embryos . The freshly dissociated cultures were treated for 2 days with either BMP2 ( 10 ng/ml ) , LIF ( 25 ng/ml ) , or both factors , both of which have been previously shown to promote astrogliogenesis in vitro from embryonic brain precursors [12] , [14] . The cultures were then examined five days later at 7 DIV , using an anti-GFAP antibody to detect newborn astrocytes . In both acutely-dissociated GE-derived and cortical cells , both BMP2 and LIF could promote the production of GFAP-positive astrocytes , and the combination of the two factors showed an even stronger effect , as reported previously [12] (Fig 9C) . However , in cultures derived from GE of TSA-treated embryos , astrogliogenesis was substantially promoted not only in the presence of LIF and BMP2 but also in their absence ( Fig 9C ) . In distinct contrast , BMP2 and LIF-promoted astrogliogenesis was significantly inhibited in cortical cultures derived from TSA-treated embryos ( Fig 9C ) . Therefore , HDAC inhibition showed the same effect upon astrogliogenesis from acute cultures derived from TSA-injected embryos as it did upon in vitro-differentiated neurosphere cultures prepared from these two brain regions . In order to see if the effects of HDAC inhibition upon neurogenesis and astrogliogenesis were occurring through a BMP2/4 -dependent pathway , as demonstrated in vitro , the nuclear localization of PHOSphorylated Smad1/5/8 was examined in vehicle - and TSA-treated embryos . In both the cortex ( Fig 10A , arrows ) and GE ( Fig 10B , arrows ) of vehicle-treated embryos , faint staining could be seen in cells of the ventricular zone of the cortex and GE and the subventricular zone of GE . In contrast , nuclear localization of PHOSphorylated Smad1/5/8 , as revealed by DAPI colocalization ( data not shown ) , was increased in these regions in TSA-treated embryos ( Fig 10A , B , left panels ) . Quantitation of the cells with nuclear localization of PHOSphorylated Smad1/5/8 revealed a large increase in these three regions ( Fig 10C , D , E ) , using the same counting method as for histone H3 (Fig 8A) . Most of the cells responding to TSA-treatment were also seen to be co-labelled for the neural precursor marker RC2 ( Fig 10A , B , middle and right panels ) , indicating that neural precursors / radial glia are the cell population responding to HDAC inhibition by nuclear localization of the downstream effector of BMP2/4 signaling , PHOSphorylated Smad1/5/8 . To see which cells were producing BMP2 and BMP4 in the brain , in situ hybridization was coupled with immunohistofluorescence for the radial glial marker BLBP ( Fig 11 ) . As before , 13.5 d.p.c.timed-pregnant mice were injected every 12 hours with vehicle or with 12.5 ug TSA in vehicle , and embryos were examined 24 hours later . In vehicle-treated 14.5 d.p.c.embryos , low levels of Bmp2 expression could be seen throughout the cortex and GE ( Fig 11A , top panel ) . At the ventricular zone of the cortex and the GE , some BLBP-positive radial glia were seen to express Bmp2 ( Fig 11A , thin arrows , middle and bottom panels ) , but many Bmp2-expressing cells were not labeled with the BLBP-antibody , especially in the subventricular zones of GE and cortex ( Fig 11A , thick arrows , middle and bottom panels ) . Upon 24-hour treatment with TSA , a strong up-regulation of Bmp2 could be observed in both GE and cortex ( Fig 11A' , top panel ) . Although some of the cells strongly-expressing Bmp2 co-labelled for BLBP ( Fig 11A' , thin arrows , middle and bottom panels ) , most of them did not ( Fig 11A' , thick arrows , middle and bottom panels ) . In contrast , the expression pattern of Bmp4 did not appear to change significantly upon TSA treatment ( Cf Fig 11B with Fig 11B' , labeled as above ) . Together , these data are in correlation with that seen in in vitro cultures , in which HDAC inhibition upregulated the mRNA expression of Bmp2 but not of Bmp4 . It also supports the idea that not only are radial glia expressing BMP2/4 , but they are also responding to these factors , as seen with the nuclear localization of PHOSphorylated Smad1/5/8 after HDAC inhibition ( Fig 10 ) .