Zebrafish care and breeding were performed under standard conditions (Westerfield, 2000) (http://zfin.org). Zebrafish embryos were incubated at 28.5°C in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄) containing 0.2 mM phenylthiourea to avoid melanin pigmentation. Embryos were staged according to Kimmel et al., 1995. All work on zebrafish embryos and adults were in accordance with German laws for animal care under permission from the Regierungspräsidium Freiburg.

We used the zebrafish AB/TL wildtype strain in all experiments. For the analysis of Islet1 positive cranial motor neurons, we used the transgenic Tg(islet1:GFP) ^rw0 reporter line (Higashijima et al., 2000). For experiments, heterozygous Tg(islet1:GFP) ^rw0 fish were outcrossed to AB/TL fish.

Supplementary Figure 1 lists all small molecule compounds used in the screens. Selected epigenetic small molecule compounds were obtained from the Collaborative Research Centre 992 Medical Epigenetics (CRC992 MEDEP; Laboratory of Manfred Jung) at University of Freiburg, and from the Structural Genomics Consortium (SGC). Additionally, the following chemical compounds were obtained commercially: Bromosporine (Sigma #SML0992), Ex-527 (Selleckchem #S1541), MM-102 (Selleckchem #S7265), Mocetinostat (Selleckchem #S1122), OICR9429 (Tocris # 5267), PFI-3 (Selleckchem #S7315), UNC0631 (Selleckchem, S7610). All small molecule compounds were dissolved in DMSO (99.5%, PanReac AppliChem #A3672) at 10 mM concentration. For experiments, the 10 mM stock was diluted with E3 medium to obtain the desired working stock concentration of 60 µM containing 1% DMSO, which was used to prepare the dilution series. The γ-secretase inhibitor DAPT (increased DA neurogenesis; (Mahler et al., 2010); Sigma, #D5942) and the prodrug neurotoxin MPTP (neurotoxic to DA neurons; (Sallinen et al., 2009); Sigma #M0896) were used as positive controls for drug delivery into embryos and dopaminergic neurons.

Small molecule compounds were screened at a concentration range of 30, 10, 3, and 1 µM in E3 medium containing 1% DMSO. These compounds were tested alongside 1% DMSO as vehicle control and 100 µM DAPT and 1 mM MPTP as positive controls for pharmacological effects on DA neurogenesis and neuron survival respectively. Zebrafish embryos at desired developmental stages from age-matched separate spawnings were pooled and arrayed into 24-well plates with 12 embryos per well. Five hundred microliters of E3 medium containing 0.2 mM phenylthiourea (PTU) and 1% DMSO was added to each well. Working solutions of small molecule compounds were prepared in a separate 24-well plate. A serial dilution was prepared at 60, 20, 6, and 2 µM in E3 medium/1× PTU/1% DMSO. Five hundred microliters of each working solution was added to the respective well containing 500 µl E3 medium and the embryos. The final volume of 1,000 µl in each well corresponded to compound concentrations of 30, 10, 3, and 1 µM. Screening assay controls for drug delivery were used at active concentrations resulting in impaired DA or NA neurogenesis or survival as described before (Sallinen et al., 2009; Mahler et al., 2010). Working solutions of DAPT and MPTP were prepared at 200 µM and 2 mM respectively. Addition of 500 µl of each working solution to the embryos in 500 µl E3 medium/1× PTU/1% DMSO yielded a final concentration of 100 µM DAPT and 1 mM MPTP. Embryos were incubated at 28.5°C until 72 h post fertilization (hpf) and the small molecule solution was then washed out by five E3 medium washes. Embryos were incubated for an additional 24 h in E3 medium/1× PTU at 28.5°C. Embryo morphology was assessed and documented at 96 hpf using a stereomicroscope. For subsequent immunohistochemistry, embryos from each treatment were transferred into separate 2.0 ml reaction tubes. Zebrafish embryos at 96 hpf were fixed in 4% PFA in PBST (1xPBS, 0.1% Tween-20), dehydrated in increasing concentrations of methanol/PBST and stored in absolute methanol at -20°C.

Alkaline phosphatase based standard colorigenic whole mount in situ hybridization (WISH) was performed based on established protocols (Holzschuh et al., 2003). The following digoxigenin-labeled antisense riboprobes were generated: sox2 (Chapouton et al., 2006) and th (Holzschuh et al., 2003). Antisense probes were in vitro transcribed respectively with T7 or T3 RNA polymerase along with DIG labeling mix (Roche). Fixed embryos were rehydrated in decreasing concentrations of methanol/PBST and transferred to PBST. Embryos at 96 hpf were treated with proteinase K (10 µg/ml in PBST) for 1 h at room temperature (RT). Proteinase K permeabilization was stopped by postfixing the embryos in 4% PFA/PBST for 20 min at RT. After three PBST washes, embryos were pre-hybridized in Hyb-Mix (50% formamide, 5x SSC, 50 µg/ml heparin, 0.1% Tween-20, 5 mg/ml torula RNA) for 2 h at 65°C. Hyb mix was replaced by 300 µl Hyb mix containing 50 ng digoxygenin labeled antisense probe. Hybridization was performed overnight at 65°C. High-stringency washes in 50% formamide/5× SSCT (0.30 M NaCl, 0.030 M trisodium citrate, pH 7.0), 2× SSCT buffer and 0.2× SSCT buffer were performed at 65°C. Embryos were washed twice in PBST and then transferred to WISH blocking solution (2% NGS, 2 mg/ml BSA in PBST) for 2 h at RT. The embryos were then incubated in anti-Digoxygenin alkaline phosphatase conjugated Fab fragment (Roche) at a 1:5,000 dilution in PBST overnight at 4°C. Excess antibody was washed out by five washes in PBST at RT. Prior to staining, embryos were transferred for 15 min to 100 mM Tris-HCl (pH 9.5) followed by incubation in 1x staining buffer (100 mM NaCl, 50 mM MgCl₂, 100 mM Tris HCl pH 9.5, 1 mM levamisole, 0.1% Tween-20) for 15 min at RT. BCIP/NBT was used as a chromogenic substrate for alkaline phosphatase (160 ng/ml BCIP and NBT in staining buffer). Staining was performed in 24-well plates for 1 ½ h at RT in the dark. Staining reaction was stopped by three washes with WISH stop solution (PBST/EDTA, pH 5.5). Stained embryos were cleared and stored in 100% glycerol.

Peroxidase based colorigenic immunohistochemistry was used to detect Islet1:GFP transgene expression in Tg(islet1:GFP) embryos and to detect pH3 immunoreactive cells in WT embryos. Following fixation but prior to permeabilization, embryos were incubated in 1% H₂O₂/PBST solution to inactivate endogenous peroxidase activity. Embryos at 96 hpf were treated with proteinase K (10 µg/ml in PBST) for 1 h at RT and the reaction was stopped by postfixing the embryos in 4% PFA/PBST for 20 min. After several washes in PBSTD (1× PBS, 0.1% Tween-20, 1% DMSO) embryos were incubated in blocking reagent (2 mg/ml BSA, 5% NGS, 1% Boehringer Blocking Reagent in PBSTD) for 2 h at RT. Primary chicken anti-GFP antibody (Invitrogen) or rabbit anti-pH3 antibody (Stern et al., 2005) (Sigma/Millipore; 06-570) were used at a 5 µg/ml dilution in blocking reagent. Embryos were incubated in the presence of primary antibody overnight at 4°C. Ten washes were performed for 15 min each in PBSTD. Secondary biotinylated goat anti chicken or biotinylated goat anti rabbit sera (Vector Laboratories, Burlingame USA) were used at a 1:1,000 dilution in blocking reagent and embryos were incubated in secondary antibody overnight at 4°C. Antigen detection was performed using the Vectastain Elite ABC Kit (Vector Laboratories, Burlingame USA) according to the manufacturer's recommendation. Diaminobenzidine (DAB) was used as a chromogenic substrate for the peroxidase. Embryos were pre-equilibrated for 15 minutes in DAB solution (0.2 mg/ml DAB in PBSTD). Staining was performed in DAB solution containing 0.3% H₂O₂ for 5 min in the dark. Stained embryos were washed several times in PBST and transferred to 100% glycerol for clearing and storing.

Detection of apoptotic cell death by TUNEL staining was performed using the Chemicon ApopTag In Situ Apoptosis Detection Kit (Millipore Bioscience Research). The protocol was modified from (Abdelilah et al., 1996). Embryos at 96 hpf were permeabilized in proteinase K (10 µg/ml in PBST) for 90 min at RT and postfixed in 4% PFA for 20 min at RT. Subsequently, embryos were incubated in Equilibration Buffer (provided in ApopTag In Situ Apoptosis Kit) for 1h at RT. The TdT enzyme was diluted in Reaction Buffer (provided in the kit) (30%, v/v). Embryos were incubated in this TdT enzyme/reaction buffer solution overnight at 37°C. The DNA end-labeling reaction was stopped by transferring the embryos to working strength Stop buffer (provided in the kit) for 10 min at RT. After several PBST washes, embryos were incubated in blocking solution (2% NGS, 2 mg/ml BSA in PBST) for 2 h at RT. Subsequently, embryos were incubated in anti-Digoxygenin alkaline phosphatase conjugated Fab fragment (Roche) at a 1:3,000 dilution in PBST overnight at 4°C to detect digoxygenin labeled DNA fragments. BCIP/NBT (160 ng/ml in PBST) was used as a chromogenic substrate for alkaline phosphatase. Staining reaction was stopped by several washes in Stop solution (PBST/EDTA, pH 5.5) and embryos were transferred to 100% glycerol for clearing and storing.

Transmitted light images were recorded using an Axioskop compound microscope (Carl Zeiss) with DIC optics and an AxioCam MRc digital camera (Carl Zeiss). For quantification of cells at single cell resolution, images of stained embryos were acquired using an Axio Examiner D.1 (Carl Zeiss) with DIC optics and an LD LCI Plan Apochromat 20× (NA = 1.0) objective. Z-stacks were recorded in 2 µm steps using the ZEN 2010 software (Carl Zeiss). We used the ZEN 2010 Blue software to count cell numbers in z-stacks. The image stack was converted to greyscale to enhance contrast. We used the ZEN 2010 Blue Event Marker tool to manually label each cell in the x, y, and z plane. We validated that each cell was counted only once by re-tracking each cell through the z-stack. For images representing distributions of neurons in z-stacks, minimum intensity projections were generated in ImageJ or ZEN 2010 Blue.

In the primary screen, all compounds except OF-1 and AGK-1 were tested at 1, 3, 10, 30 µM with 7 to 16 embryos in each treatment (no independent replicates; Supplementary Table 1 report data for compounds with phenotype, Supplementary Table 3 reports compounds without phenotype). All compounds including OF-1 and AGK-1, but excluding PFI-2 and UNC1215, were also tested at 100 µM with at least 10 embryos in each treatment (two independent replicates; Supplementary Table 2 ). For all 10 compounds that generated changes in th expression, one additional experiment was performed, such that data were obtained at least in duplicate ( Supplementary Table 1 ). In the secondary screen, all th WISH assays with cell counts were performed at least in triplicates (three to six independent experiments; Supplementary Table 4 ). The isl1 assays with cell counts were performed in triplicate ( Supplementary Table 5 ). The quantitative evaluation of sox2 expression was performed in triplicate ( Supplementary Table 6 ). The apoptosis assays with cell counts were performed in triplicates ( Supplementary Table 7 ), except for Vorinostat, Mocetinostat, and JQ1, which were only performed in duplicate. Retina diameters of inhibitor and DMSO treated embryos were determined in three to six replicates ( Supplementary Table 8 ). Retina diameters of 72 and 120 hpf control embryos were determined in a single experiment on eight embryos each.

Statistical analyses were performed using GraphPad Prism 6 software (GraphPad Software). Embryos from the WISH experiments were scored according to staining intensity and distribution. Comparison of phenotypes of DMSO treated control embryos and small molecule compound treated embryos was analyzed by two-way Fisher's Exact Test. Cell numbers and retina diameter measurements from independent experimental replicates were analyzed using two-way-ANOVA with ad hoc Bonferroni correction for multiple comparisons. For retina diameter comparison to 72 and 120 hpf larvae, statistical analysis used Mann-Whitney test. Graphs show mean cell numbers with standard deviation. For cell counts or numbers of embryos affected by a given phenotype, the averages for control embryos were set to 100 (phenotypes) or 1.0 (cell numbers), and all measurements of treated embryos were normalized with respect to the control value. Supplementary Tables 4–8 provide the detailed statistical analysis for each experiment.

To identify chromatin regulatory mechanisms that function during embryonic DA neurogenesis in vivo, we performed a chemical genetics screen in zebrafish embryos ( Figure 1A ). The screen was organized in two phases: in the primary screen, we tested 32 compounds to establish lethal concentrations and identify compounds with effects on DA neurogenesis based on changes in th expression. In the secondary screen, we compared effects on DA neurons with other neuronal population and tried to distinguish effects on NSCs, neuronal differentiation and survival.

We reasoned that applying active compounds during specific stages of DA neuron development might affect chromatin regulation in NSCs, DA progenitor populations, or mature neurons. Based on previous reported cell permeability and in vivo activity, we selected 32 small molecule compounds that target different groups of chromatin regulators ( Supplementary Figure 1 ). This set includes 6 histone deacetylase (HDAC) inhibitors, two sirtuin inhibitors, 4 histone acetyltransferase (HAT) inhibitors, 8 bromodomain inhibitors, 7 histone lysine methyltransferase (HKMT) inhibitors (including 2 MLL1-WDR5 interaction inhibitors), 4 histone lysine demethylase (HKDM) inhibitors, and 1 histone methyl reader protein inhibitor ( Supplementary Figure 1 ). The active working concentrations and lethal concentrations for these compounds were mostly unknown for zebrafish. We tested each compound at 1, 3, 10, and 30 µM concentrations to determine if any compound showed lethal effects in the postulated range of active concentration (see black interrupted lines in Figure 2 ; lethality data for all compounds are reported in Supplementary Tables 1–3 ). Previous small molecule screens in zebrafish embryos reported high hit rates and low off-target toxicity within this dose range (Peterson and Fishman, 2011). We then evaluated the dose–response curves for each of these compounds with regard to DA or NA neurogenesis. Small molecule inhibitors that caused no overall morphological alterations were subsequently also tested at 100 µM concentration, which however, in many cases turned out to be lethal ( Supplementary Table 2 ).

Zebrafish DA and NA neurons develop in temporally distinct phases of neurogenesis (Mahler et al., 2010). We exposed zebrafish embryos to small molecule compounds during a broad interval of DA and NA neuron development, while aiming not to interfere with early pattern formation in embryogenesis. Therefore, in our primary screen, we added the compounds at 24 hpf and incubated embryos until 72 hpf. After removal of the compounds, embryos developed for another 24 h in standard E3 medium until fixation at 96 hpf for further analyses. All incubation media also contained PTU to suppress pigmentation (Elsalini and Rohr, 2003) and facilitate imaging of the brains. Treated embryos were assayed for DA and NA neurogenesis defects at 96 hpf by WISH using th expression as a marker for DA and NA neurons (Holzschuh et al., 2003). The th expressing DA clusters in the telencephalon (Tc), the pretectum (PrT), and the diencephalon (DC1–DC6) were assayed individually. We determined olfactory bulb and subpallial DA groups, which undergo neurogenesis in similar time windows (Mahler et al., 2010), together as telencephalic DA groups. The locus coeruleus (LC) NA cluster was analyzed separately, while NA clusters within the medulla oblongata (MO) and area prostrema (AP) were analyzed together and labeled MO/AP. The DA and NA phenotypes were classified into three categories based on the level and pattern of the th transcript detection: (1) no differences detected compared to the DMSO treated controls, (2) increase or (3) decrease in cell numbers and/or stain intensity. While we refined the analysis by cell counts in the secondary screen, in the primary screen we did not distinguish increase in cell number or th WISH stain intensity. Dose-effect curves were calculated and display the percentage of treated embryos showing altered th staining intensity for each compound concentration ( Figure 2 ). Figure 3 shows only compounds that affected DA or NA development in treated embryos. For examples of compounds with no effects on th expression see Supplementary Figure 2 . We considered those small molecule compounds that caused a change in th mRNA expression intensity (compared to DMSO controls) in more than one third of tested embryos as a compound potentially affecting DA and NA neurogenesis, and selected them for further characterization in the secondary screen ( Figures 1A, B ).

Among 32 compounds tested, the primary screening identified 10 compounds to affect th expression in the larval brain ( Figures 2 , 3 , 4 ; Supplementary Table 1 ). Namoline caused changes only at one concentration, and UNC1215 had effects that did not appear to be dose dependent. Therefore, we did not further analyze these compounds. Treatments with the other 20 compounds resulted in normal th expression within the brain of more than two thirds of treated embryos ( Supplementary Figure 2 ; Supplementary Table 3 ). We found that some compounds interfering with histone acetylation and acetylation reader proteins containing Bromodomains decrease th expression in treated embryos. The HDAC class 1 inhibitors Entinostat and Mocetinostat reduced th expression within telencephalic, pretectal and diencephalic DC1 DA groups in zebrafish larvae ( Figures 2A, B , 3B, C , and 4 ). We also observed a less pronounced decrease within the DC4/5 and 6 DA groups and MO/AP NA groups ( Figures 2A, B ). The observed alterations in th expression in Tc, PrT and DC were dose-dependent, with higher concentrations affecting higher percentages of treated embryos ( Figures 2A, B ). The concentration-dependent effects became prominent at concentrations of 10 µM and higher. More than 80% of treated embryos showed altered th expression in Tc, PrT and DC groups in treatments with 10 or 30 µM of both compounds ( Figures 2A, B ). The pan-HDAC inhibitor Vorinostat (SAHA) affected th expression in Tc and PrT DA groups ( Figures 2C , 3D , and 4 ). The effects on th expression in Tc and PrT were dose-dependent with more than 60% of treated embryos showing reduced th expression at 30 µM.

We observed a reduction in th expression upon treatment with the pan-Bromodomain inhibitor Bromosporine and the Bet-Bromodomain inhibitors I-Bet151 and JQ1 ( Figures 2D–F , 3E–G , and 4 ). Treatments with Bromosporine, in a dose-dependent manner, caused reduced th expression within the Tc and the PrT DA groups and the MO/AP NA groups (80% at 30 µM; Figures 2D and 3E ). Furthermore, we detected a minor reduction in th staining intensity in DC4/5/6 groups in 20% of treated embryos at 30 µM ( Figure 2D ). Treatments with JQ1 reduced th expression in Tc, PrT, and DC4/5/6 DC groups and the MO/AP NA groups at 10 µM (80% of treated embryos; Figures 2E , 3F , and 4 ). I-Bet151 also caused a reduction in th expression within the Tc, PrT and DC4/5/6 DA groups ( Figures 2F and 3G ).

Embryos treated with either pan-HAT inhibitor PU139 or PU141 showed slightly increased th stain intensity in DA groups DC4/5/6 at 30 µM ( Figures 2G, H , 3H, J , and 4 ). Treatments with each compound at 30µM also caused morphological abnormalities as evident in a smaller head size ( Figures 3H, I ).

We assayed two compounds that specifically disrupt the interaction between histone methyltransferase MLL1 and its binding partner WDR5. For MM102 we found reduced th expression in Tc, but surprisingly at 10 and 30 µM th expression appeared increased in DC4/5/6 DA groups ( Figures 2I, J , 3J, K , and 4 ). For PrT and DC1 DA groups, the effect was variable; with few embryos having increased or decreased th expression (see also Supplementary Table 2 ). We observed similar effects with the MLL1-WDR5 interaction inhibitor OICR9429 ( Figures 2J , 3K , and 4 ). At compound concentrations between 3 and 30 µM we found 30% of treated embryos to show increased th staining intensities in DC4/5/6 ( Figures 2J and 3K ).

Distinct DA and NA groups develop in distinct temporal waves of neurogenesis during development (Mahler et al., 2010). To investigate whether specific DA and NA neuron progenitor pools may be sensitive to epigenetic inhibitors in shorter time windows of neurogenesis, we applied HDAC, HAT, bromodomain and MLL1 interaction inhibitors for 24 or 36 h time windows (12 to 48 hpf, 24 to 48 hpf, or 48 to 72 hpf). These treatments resulted in similar decreases in th expression in Tc and PrT DA neurons for each time window ( Supplementary Figure 3 ). However, early born DC2 DA neurons were differently affected in specific time windows by Entinostat and Mocetinostat: the effect of early treatment starting at 12 hpf on DC2 DA neurons was stronger when compared to treatments starting at 24 hpf or later, suggesting that these HDAC inhibitors might act on DC2 DA progenitor cells. We conclude that different phases of neurogenesis are differentially affected by these HDAC inhibitors.

In a secondary screen, we (1) quantified the effects of the compounds on DA and NA cell numbers, and tested whether (2) compound effects are selective for DA or NA populations or also affect other neuronal types, (3) effects on DA or NA neurons may be from loss of neural stem cells, (4) compounds may cause apoptosis, and (5) compounds may interfere with normal developmental progression and morphogenesis ( Figure 1B ). We screened in detail the 10 inhibitors active in our primary screen: We chose a 30 µM concentration, except for Mocetinostat (10 µM) and JQ1 (3 µM), based on our primary screen results. Compounds were applied from 24 until 72 hpf, and embryos analyzed at 96 hpf.

(1) To independently validate and quantify compound effects on DA or NA neuron number, we repeated the WISH analysis for th on treated embryos at 96 hpf. We recorded high-resolution z-stacks of the brains from control and treated embryos, and counted th expressing cells in specific DA and NA cell clusters. We also compared subcellular localization of the th WISH signal in pretectal DA and locus coeruleus NA neurons, which were imaged at highest resolution, and found th transcript to be localized predominantly in the cytoplasmic compartment for DMSO controls and compound treated embryos. (2) Given that chromatin regulation mechanisms may not be selective for DA or NA neurons, we analyzed whether the compounds also affect other types of early developing neurons. To visualize islet1 expressing neurons, which include primary and secondary cranial and spinal motor neurons, but also other neuron populations in the forebrain, we used transgenic Tg(isl1:GFP)^rw0/+ zebrafish (Higashijima et al., 2000) and stained 96 hpf embryos by anti-GFP immunohistochemistry. The DA or NA neuronal groups and the Isl1:GFP expressing neuronal groups are anatomically distinct, except for the prethalamic DC1 DA cluster, which requires isl1 for differentiation of DA neurons (Filippi et al., 2012). In our analysis of Isl1:GFP transgene expressing cranial motor neurons we did not include the Isl1 positive DC1 DA neurons. islet1 positive neurons were documented in image stacks and selected neuron groups were counted. (3) NSC states, including self-maintenance and segregation of progenitors, rely on chromatin regulation and may be affected by epigenetic inhibitors. Therefore, we analyzed sox2 expression, a stem cell marker expressed in all NCSs in the ventricular zone, by WISH of treated embryos fixed at 96 hpf, documented them, and visually evaluated effects of inhibitor exposure semi-quantitatively based on sox2 staining intensity. (4) We further asked whether a reduction in cell numbers may be due to increased apoptosis in inhibitor treated embryos. Previously, several small molecule inhibitors targeting HDACs and Bromodomain proteins have been reported to trigger apoptotic cell death in different cancer cell lines (Mertz et al., 2011). Therefore, we performed TUNEL assays on treated zebrafish embryos fixed at 96 hpf and quantified TUNEL positive cells in defined brain regions (nasal placode NP, telencephalon Tc, retina, and hindbrain HB; diencephalon and midbrain DC/MB were counted together because the boundary could not be distinguished in dorsal views) by cell counting. (5) Another possible reason for the reduction in cell numbers could be that the compounds slow down development of the larvae, resulting in a developmental delay as compared to the DMSO treated control larvae. To account for any such effects, we analyzed the zebrafish retina by morphometry. We used fixed 96 hpf larvae from the th WISH analysis. We measured the diameter of the right and left retinae at their maximum rostro-caudal extension in inhibitor treated larvae and compared them with the age-matched DMSO treated control larvae. To determine the potential extent of delay, we also measured retinae in DMSO treated control larvae at 72, 96, and 120 hpf.

We performed all five assays in at least three independent experimental replicates. For better comparison, cell counts were normalized to the cell counts obtained from DMSO treated control embryos. To assess whether differences between control and inhibitor treated embryos may be significant, we performed two-way ANOVA.

In the primary screen, the MLL1-WDR5 interaction inhibitors MM102 and OICR9429 emerged as potential hits. However, the qualitative primary analysis was not confirmed in the secondary screen: we observed that both OICR9429 and MM102 did not produce any significant changes in the number of th expressing cells in any of the DA or NA clusters ( Supplementary Figures 4A and 5A–C ). We also found no effects of MM102 on the number of isl1:GFP transgene expressing cells and sox2 expressing cells in the ventricular zone ( Supplementary Figures 4J, O and 5F, I ). OICR9429 induces a small increase in the isl1:GFP transgene expressing cells in the midbrain clusters MNIII/IV and hindbrain cluster MNVII but no specific effects were observed on the sox2 expressing cells ( Supplementary Figures 5D–I ). We did not observe any significant effects on apoptosis in OICR9429 or MM102 treated embryos ( Supplementary Figures 4T and 5J–L ). The morphometric analysis of retinae did not reveal delayed development in OICR9429 or MM102 treated larvae ( Supplementary Figure 6A ). With respect to th expression, MM102 and OICR9429 appear to have been false positive hits in the primary screen. However, at least for OICR9429, the isl1:GFP phenotype reveals some compound activity. A potential explanation for the differences in qualitative versus quantitative th expression analysis may be that the cell clusters may have appeared slightly more dispersed, while the cell numbers in each cluster were actually not affected.

In Entinostat and Mocetinostat treated embryos, th WISH cell counts confirmed the phenotypes observed in the primary screen, while Vorinostat treated embryos with respect to DA and NA cell numbers did not differ significantly from controls. For Entinostat and Mocetinostat we counted significantly less th expressing cells within Tc, PrT, retina and the diencephalic cluster DC1 (p < 0.001; Figures 5A–C ; Supplementary Figure 4B ). In addition, we also found a decrease in hindbrain NA medulla oblongata/area postrema neurons (MO/AP; p < 0.001; Figures 5A–C ). We could not detect any changes in cell numbers in DA neuron group DC2 and NA neurons in the locus coeruleus (LC). Both neuron populations become postmitotic before the onset of the small molecule compound exposure (Mahler et al., 2010), suggesting that Entinostat and Mocetinostat do not affect the maintenance of differentiated DA and NA neurons. These HDAC inhibitors also influence cell numbers of isl1 positive neurons ( Figures 5D–F ; Supplementary Figure 4G ). Entinostat treatments resulted in an increase in cell number only within the midbrain motor neuron clusters MNIII/IV ( Figures 5D–F ; p < 0.001). However, Mocetinostat treatments caused a reduction in isl1 positive neurons in Tc (p < 0.001), and hindbrain motor neuron cluster MNVp (p < 0.001) ( Figure 5F and Supplementary Figure 4G ). Vorinostat affects isl1 positive neurons in MNVa and MNVp ( Figure 5F ). Sox2 expression in NSCs within the ventricular zone (VZ) and the retina proliferation zone was unaffected in Entinostat and Vorinostat treated embryos ( Figures 5G–I ). However, Mocetinostat strongly reduced sox2 expression exclusively in the ventricular zone and the retina proliferative zone ( Figure 5I ; Supplementary Figures 4K, L ). We did not observe a reduction of sox2 transcript within the pharyngeal arches, indicating that sox2 expression in NSCs may be selectively affected. Treatments with Entinostat and Mocetinostat caused an overall increase in apoptosis in the zebrafish brain based on TUNEL assays. We counted significantly more apoptotic cells in the telencephalon, di-/mesencephalic area and hindbrain in Entinostat treated zebrafish larvae ( Figures 5J–L ; p < 0.001), and in the telencephalon and hindbrain of Mocetinostat treated larvae (p < 0.001, see also Supplementary Figure 4Q ). Vorinostat did not affect apoptosis ( Figure 5L ). The morphometric analysis of the retina diameter showed no significant changes in the retinae in Vorinostat and Mocetinostat treated larvae compared to the 96 hpf DMSO treated control larvae, indicating no developmental delay. In case of Entinostat treated larvae, the mean diameter of the retinae showed an increase of 10.65% (p < 0.001), which is higher than the change in mean diameter during normal developmental progression from 96 to 120 hpf ( Supplementary Figure 6 ), arguing for hyper-proliferation or morphological expansion of the retina.

The pan-Bromodomain inhibitor Bromosporine as well as the Bet-Bromodomain inhibitors JQ1 and I-Bet151 caused a severe reduction in th expression within DA clusters in TC and PrT in the primary screen. Embryos treated with Bromosporine or I-Bet151at 30 µM or JQ1 at 3 µM had reduced numbers of th expressing cells within TC and PrT (p < 0.001; Figures 6A–C ; Supplementary Figure 4C ). Furthermore, we detected less th expressing neurons in the retina after treatments with Bromosporine and JQ1 (p < 0.001; Figures 6A–C ). Cell numbers of th expressing NA neurons in the MO/AP cluster in the hindbrain were reduced after treatments with Bromosporine and JQ1 but were unaffected by I-Bet151 treatments. These effects were not restricted to DA or NA neurons, since in bromosporine treated embryos isl1 neurons were also reduced in the Tc and the hindbrain cluster MNV (p < 0.001; Figures 6D–F ). We observed a similar reduction in isl1 neurons in Tc and MNV in embryos treated with JQ1 and I-Bet151 (p < 0.001 each; Supplementary Figure 4H ). Bromosporine caused a strong decrease in sox2 expression in NSCs in the ventricular zone and retinal proliferation zone (p < 0.001; Figures 6G–I ), while pharyngeal sox2 expression was unaffected. JQ1 also causes a strong decrease in sox2 expression in NSC in the VZ and RPZ ( Figure 6I ). Loss of sox2 expression in NSCs suggests that reduced numbers of th and isl1:GFP expressing neurons may be caused by a depletion of sox2 expressing NSCs. I-Bet151 also resulted in reduction of sox2 expression in VZ ( Figure 6I ; Supplementary Figure 4M ). Treatments with Bromosporine and JQ1 also cause an overall increase in apoptosis. We counted significantly more apoptotic cells in telencephalon and retina of zebrafish embryos treated with bromosporine ( Figures 6J–L ; p < 0.001). JQ1 caused a significant increase in apoptosis in each of the anatomical regions analyzed except the nasal placode. Surprisingly, we did not observe any specific effect on apoptosis in the I-bet151 treated larvae. The morphometric analysis of the retinal diameters in Bromosporine, I-Bet151 and JQ1 treated larvae showed no significant changes with respect to the 96 hpf DMSO treated control larvae ( Supplementary Figure 6 ). In summary, the severe effects of Bromosporine, JQ1 and I-Bet151 on DA or NA as well as isl1 neuron development may be caused by loss of NSCs, increased apoptosis, but may also directly affect neuronal differentiation.

The primary screen revealed HAT inhibitors PU139 and PU141 to affect th expression within DA neuron groups in DC4/5/6. In contrast, cell counts revealed that PU139 or PU141 reduced the numbers of th expressing cells in the DA neuron clusters PrT, DC1 and in the retina (p < 0.005; Figures 7A–C ) but not in DC4/5/6. We found isl1 expressing MNVa and MNVII motor neurons numbers to be increased in PU141 treated embryos (p < 0.005; Figures 7D–F ). Furthermore, isl1:GFP transgene expression revealed in PU141 treated embryos a failure in proper neuronal migration of motor neuron cluster MNVII, which arise in rhombomere 4 and normally subsequently migrate into rhombomere 6. Combined with the small head size of the treated embryos this observation might indicate developmental delay caused by the inhibitor. Interestingly, sox2 expression levels appear elevated within ventricular and retinal proliferation zones after exposure to PU141 ( Figures 7G–I ), however the strength of the effect varied between treated embryos. PU141 caused a significant increase in apoptotic cells in the tel- and di/mesencephalon ( Figures 7J–L ; p < 0.001). In contrast, treatment with PU139 caused increased apoptosis in the hindbrain but no significant increase in other brain regions ( Figures 7J–L ; Supplementary Figure 4S ; p < 0.001). The morphometric analysis of the retinal also revealed a decrease in in PU139 and PU141 treated larvae by 6.27 and 8.38% respectively (p < 0.001) as compared to the 96 hpf control larvae. This decrease is comparable to the change during normal developmental progression from 72 to 96 hpf ( Supplementary Figure 6 ), suggesting that PU139 and PU141 cause severe developmental delay.