Small-molecule activation of
neuronal cell fate
Jay W Schneider1, Zhengliang Gao2, Shijie Li3, Midhat Farooqi3,
Tie-Shan Tang4, Ilya Bezprozvanny4, Doug E Frantz5&
We probed an epigenetic regulatory path from small molecule
to neuronal gene activation. Isoxazole small molecules
triggered robust neuronal differentiation in adult neural stem
cells, rapidly signaling to the neuronal genome via Ca2+influx.
Ca2+-activated CaMK phosphorylated and mediated nuclear
export of the MEF2 regulator HDAC5, thereby de-repressing
neuronal genes. These results provide new tools to explore the
epigenetic signaling circuitry specifying neuronal cell fate and
new leads for neuro-regenerative drugs.
The transduction of fate signals to the genome of undifferentiated stem
cells remains one of the most fundamental problems in stem cell and
developmental biology. Despite their immaturity, neural stem cells
(NSCs) are receptive to neurotransmission signals regulating neuro-
sphere growth in vitro1; they are also receptive to those regulating
neuroD expression and differentiation in vitro and in vivo2,3. However,
the mechanism that transduces neurotransmitter-evoked Ca2+signals
to the neuronal gene program in stem cells is unknown. The brain-
enriched epigenetic regulatory module comprised of class IIa histone
deacetylases (HDACs) and myocyte enhancer factor-2 (MEF2) tran-
scription factors has an important role in neuronal maturation and
survival4, but a role for this mechanism in NSCs has been elusive. We
previously conducted a screen for cardiogenic small molecules that
induce expression of Nkx2.5, a homeodomain transcription factor
gene, in P19 embryonal carcinoma cells5. Unexpectedly, a subset of
these Nkx2.5-activating small molecules induced a neuronal (and not a
cardiogenic) phenotype. Indeed, Nkx2.5 has reported roles in neural
development and differentiation6,7, which provides biological rationale,
in retrospect, for the discovery of neurogenic molecules from the
original screen. Thus, we counterscreened our validated hits for
neurogenic small molecules and identified several isoxazole com-
pounds that selectively convert NSCs into neurons.
Initial screening efforts identified five hits (1, 2, 3, 4 and 5) that all
contain a 3,5-disubstituted isoxazole as a defining structural motif.
Each was capable of activating transfected neuroD and gluR2 neuronal
reporter genes in NSCs in a dose-dependent manner, with activity
peaking in the low micromolar range (Supplementary Fig. 1 online).
In an effort to generate a more potent and soluble lead, we synthesized
several small, targeted libraries of 3,5-disubstituted isoxazoles
( B75 analogs total) using the chemistry outlined in Supplementary
Methods online. Through these efforts, we identified 6, 7, 8 and 9
(Supplementary Fig. 1), which displayed not only the highest activity
but also adequate aqueous solubility for our downstream mechanistic
studies. Thus, we used 9 as our primary chemical probe to delineate the
mechanistic pathway for excitation-neurogenesis signaling to the stem
cell genome. Studies were done in hippocampal NSCs from adult rat
HCN cells—a well-characterized continuous line of multipotent stem
and progenitor cells that rarely exhibit spontaneous lineage differentia-
tion in the presence of fibroblast growth factor-2 (FGF-2) (ref. 8).
Compound 9 not only induced robust neuronal differentiation, but
also dominantly blocked competing astrocyte differentiation inducible
by leukemia inhibitory factor (LIF) and bone morphogenetic protein-2
(BMP-2) in HCN cells (Fig. 1a,b and Supplementary Fig. 2 online).
Within a few hours of exposure, 9 activated the endogenous neuronal
gene program in HCN cells (Fig. 1c), which is consistent with the
results of neuronal reporter genes (Fig. 1a). Taken together, these data
indicate that 9 sends a rapid biochemical signal to the nucleus of HCN
cells, thereby initiating the neuronal differentiation program while
blocking non-neuronal fates in the presence of strong gliogenic signals.
Compound 9 mediated an instructive fate signal in uncommitted
HCN cells and increased proliferation of committed progenitor cells
(neuroblasts) while causing only a modest amount of cell death
(Supplementary Fig. 3 and Supplementary Discussion online). In
addition to its effects on HCN cells, 9 induced neuronal differentiation
in adult mouse whole brain (MWB), subventricular zone (SVZ)
progenitors and P19 embryonal carcinoma cells, thereby establishing
a wide range of neurogenic activity for this class of molecules
(Supplementary Fig. 2).
Compound 9 has neurotransmitter-like effects in undifferentiated
HCN cells—it rapidly activates Ca2+influx through both voltage-gated
Ca2+channels and N-methyl-D-aspartic acid (NMDA) receptors. We
first searched for neurotransmitter-like activity based on previous
reports of NSC response to neurotransmitter growth signals1–3. Given
that both glutamate and GABAAreceptors induce hippocampal neu-
rogenesis through Ca2+signaling2,3, we first investigated whether 9
triggers increased [Ca2+]i. Indeed, 9 did increase [Ca2+]iin HCN cells,
with a slow but steadily rising current of appropriate magnitude for
undifferentiated stem cells with primitive Ca2+handling mechanisms
(Fig. 1d and Supplementary Fig. 4 online). We successfully inhibited
the 9-mediated [Ca2+]irise in NSCs by blocking all major sources of
Ca2+influx in HCN cells with a cocktail of inhibitors (AP5 (11), CNQX
(12) and nifedipine (13), targeting NMDA- and AMPA-type glutamate
receptors and L-type Ca2+channels, respectively) (Fig. 1d and Supple-
mentary Fig. 4). Next, we identified the most probable source of Ca2+
influx by demonstrating that MK801 (14), a specific NMDA receptor
antagonist, can block the 9-mediated Ca2+signal (Fig. 1d and
Supplementary Fig. 4). To confirm the physiological importance of
9-triggered Ca2+signaling, we demonstrated that the inhibitor cocktail
Received 17 March; accepted 16 May; published online 15 June 2008; doi:10.1038/nchembio.95
1Department of Internal Medicine,2Department of Molecular Biology and Cecil H. and Ida Green Center for Reproductive Biology Sciences,3Department of Molecular
Genetics,4Department of Physiology and5Department of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas
75390, USA. Correspondence should be addressed to J.H. (firstname.lastname@example.org).
408VOLUME 4NUMBER 7JULY 2008
NATURE CHEMICAL BIOLOGY
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
and MK801 block 9-mediated neuroD reporter gene induction
(Fig. 1e). Nifedipine alone also attenuated the neuroD reporter, and
L-type Ca2+channel agonists Bay K 8644 (15) and FPL 64176 (16)
increased neuroD activity in a dose-dependent manner, which indicates
that Ca2+influx via L-type Ca2+channels is sufficient to promote
neuroD expression leading to neuronal differentiation (Fig. 1e and
Supplementary Fig. 4). To further confirm the contribution of Ca2+/
NMDA signaling to 9-mediated excitation-neurogenesis, we applied the
inhibitor cocktail and MK801 to HCN cells, which inhibited basal and
enhanced neuroD mRNA levels elicited by 9 within 3 h (Fig. 1f). FPL
64176 alone was able to stimulate neuroD expression and neuronal
differentiation, although to a lesser extent compared with 9 (Supple-
mentary Fig. 4). Importantly, NMDA receptor and Ca2+channel
blockers MK801 and nifedipine strongly attenuated the 9-induced
increase in neuroD expression, but failed to completely block pheno-
typic differentiation. Breakthrough of the neuronal phenotype could be
due to kinetic or timing effects of 9, MK801 and nifedipine actions at
their respective targets, or it could indicate that 9 acts as well through a
second, Ca2+-signaling–independent pathway to the genome.
So far, there has been no evidence that Ca2+-regulated MEF2
transcription factors play a role in the early steps of embryonic or
adult NSC differentiation. In fact, despite overall high-level expression
in the central nervous system, MEF2 mRNA and protein expression is
notably absent from the proliferative zones of the brain under basal
conditions9. Yet in our gene expression studies, 9 strongly induced the
endogenous NMDA receptor subunit-1 (NR1) gene (Fig. 1c), one of
few validated MEF2 target genes in mature neurons10. To provide
evidence that MEF2 is involved in 9-mediated induction of NR1, we
confirmed that 9 signaling activates a MEF2 response element
(MREx3) reporter gene in a dose-dependent manner in HCN cells
(Fig. 2a). The rapid and strong activation of MREx3 reporter gene by
9 suggests that MEF2 transcription factor activity is available but
negatively regulated in these undifferentiated NSCs. The failure to
demonstrate MEF2 transcriptional activity on reporter genes in
undifferentiated HCN cells, despite the presence of MEF2 proteins
and DNA binding activity (Supplementary Fig. 5 and Supplementary
Discussion online), suggests that MEF2 is actively repressed in
undifferentiated NSCs and that 9 can relieve this repression.
The class IIa HDAC-MEF2 epigenetic/transcriptional regulatory
network is the next logical target mechanism for the neurogenic effects
of 9 in HCN cells. MEF2-dependent gene expression at the level of
HDAC nuclear transport has been demonstrated in multiple cell types
including heart, skeletal muscle and mature cells of the brain, including
hippocampal neurons11. A nodal point in HDAC-MEF2 network
regulation is activation of HDAC kinases. Nuclear MEF2 transcription
factors are liberated from HDAC repression through 14-3-3–chaper-
oned nuclear export of phosphorylated HDAC (ref. 12). We first
observed that 9 indeed activates an HDAC kinase in HCN cells,
which leads to hyperphosphorylation of HDAC5 at Ser259 and
Ser498, two 14-3-3 docking sites (Supplementary Fig. 6 online). To
correlate 9-induced HDAC5 hyperphosphorylation with export of this
transcriptional repressor protein from the NSC nucleus, we fractio-
nated control and 9-treated HCN cells into nuclear and cytoplasmic
components for blotting experiments (Fig. 2b). Compound 9 had no
effect on the subcellular distribution of MEF2A, MEF2C or cAMP
response element–binding protein (CREB), a second class of neuronal
Ca2+-regulated nuclear transcription factors, in differentiated HCN
cells (Fig. 2b). Indeed, cytoplasmic accumulation of phosphorylated
HDAC5, with a concomitant decrease in the nucleus, is the only
substantial 9-mediated change we observed in these extracts (Fig. 2b).
Compound 9 did not significantly affect phosphorylated HDAC4 levels
(Fig. 2b), which is consistent with the previous finding that HDAC5
and HDAC4 are differentially regulated by neural inputs in hippocam-
pal neurons13. We confirmed 9-mediated nuclear export of HDAC5 in
NSCs using a green fluorescent protein (GFP)-HDAC5 fusion protein
delivered by adenovirus (Fig. 2c and Supplementary Fig. 6). GFP-
HDAC5 (S-A), a signal-resistant mutant of GFP-HDAC5 bearing
mutated 14-3-3 chaperone protein docking sites (Ser259 and
Ser498), was predominantly nuclear in NSCs, even after 9 treat-
ment (Fig. 2c). We also confirmed cytoplasmic enrichment of
3 h1 d 3 d
01 2 35
40 17 10
Tuj1-positive cells (%)
LIF + BMP
LIF + BMP + 9
RLU (104) NeuroD-luc
0 min2 min 4 min8 min 12 min 16 min
Figure 1 Isoxazoles trigger neuronal differentiation
through a neurotransmitter-evoked Ca2+signal.
(a) Compound 9 induces a dose-dependent
increase in both neuroD-luc and gluR2-luc reporters,
exceeding the gold standard, retinoic acid and
forskolin (RA/FSK), in HCN cells. RLU, relative
luciferase units. (b) Compound 9 promotes robust
neuronal differentiation of HCN cells and dominantly
blocks gliogenesis in 4 d cultures. Scale bar, 25 mm.
Quantification of Tuj1+cells over 4 d is shown.
(c) Reverse-transcriptase PCR gene expression
profiles and protein blotting analysis of neuronal-
specific genes and p27KIP1levels in HCN cells.
Glyceraldehyde 3-phosphate dehydrogenase (GAPDH)
is used as a normalization control. (d) Representative
images showing Fura-2 (340/380) ratios in HCN cells. Scale bar, 5 mm. (e) 50 mM inhibitor cocktail, MK801 or nifedipine (Nif) in 9-treated HCNs
attenuated neuroD-luc activity after 24 h. (f) Reverse-transcriptase PCR analysis of neuroD expression of 9-treated HCN cells in combination with 50 mM
cocktail, MK801 or Nif at 3 and 24 h. Values in a and e represent the average of 12 replicates ± s.d. Values in b represent the average of 2 replicates ± s.d.
of one representative experiment from three independent experiments.
NATURE CHEMICAL BIOLOGY
VOLUME 4NUMBER 7 JULY 2008409
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
phosphorylated HDAC5 in response to 9 by immunoprecipitation and Download full-text
protein blotting of nuclear and cytoplasmic extracts from 9-treated and
control NSCs using adenovirus encoding a FLAG-tagged HDAC5
(Supplementary Fig. 6). Finally, we confirmed that the HDAC-
MEF2 axis regulates the NR1 MEF2 target gene in HCN cells by
demonstrating that the signal-resistant mutant HDAC5 (S-A) (Fig. 2d)
and a dominant-negative MEF2C-engrailed repressor fusion protein
(Supplementary Fig. 5) completely block the ability of 9 to activate the
NR1 reporter gene. In addition to a MEF2 site, the NR1 promoter also
contains tandem binding sites to the ubiquitously expressed transcrip-
tion factor Sp1. Indeed, activation of NR1 triggered by Isx was
dependent, in part, on MEF2 binding, as well as binding of Sp1,
although MEF2 and Sp1 did not appear to have synergistic effects in
HCN cells10(Fig. 2e). However, neither HDAC5 (S-A) nor MEF2C-
engrailed proteins completely blocked 9-induced neuronal differentia-
tion in HCN cells, which implicates the involvement of other pathways.
Regulatory phosphorylation of class II HDAC-MEF2 activity has
been mainly attributed to CaM kinases (CaMK) and protein kinase D
(PKD), which is phosphorylated and activated by protein kinase C
(PKC)14,15. Therefore, we tested inhibitors of CaMK and PKC for their
ability to block 9-induced HDAC5 phosphorylation and neuronal
differentiation (Supplementary Fig. 7 and Supplementary Discus-
sion online). KN93 (17), a specific inhibitor of CaMK, and not the
analog KN92 (19), which blocks potassium channels but not CaMK,
effectively suppressed 9-mediated differentiation (Fig. 2f,g). These
data suggest that CaMK is the main HDAC kinase activated by 9
mediating the neurogenic signal in HCN cells.
Our experiments with 9 provide compelling evidence that a
neurotransmitter-like excitation signal in NSCs can regulate the
neuronal genome through an epigenetic transcriptional network
involving HDAC5 and MEF2, thus providing the first evidence that
the HDAC-MEF2 circuitry participates in early neural cell fate events,
at least in vitro. Isoxazoles trigger a Ca2+signal, involving voltage-
gated Ca2+channels and NMDA receptors, that activates CaMKII, the
major HDAC kinase in NSCs. Compound 9–induced phosphorylation
of HDAC5 leads to export of this chromatin-modifying enzyme and
repressor protein from the NSC nucleus, thereby de-repressing MEF2
and other transcription factors to directly activate MEF2 target genes
such as NR1 and indirectly activate neuroD and other neuronal genes,
which together promote early phenotypic differentiation. Though our
data demonstrate that NSC Ca2+channels and NMDA receptors play a
major role in mediating the neurogenic isoxazole response, we cannot
exclude alternative, more direct, pathways for electrochemical
epigenetic signaling to the neuronal genome. Future studies will
explore whether de-repression of MEF2 by nuclear export of
HDAC5 in NSCs underlies the brain’s neurogenic response to patho-
logical stress that involves excitation (for example, drugs, ischemia
or seizures), and whether 9-mediated nuclear export of HDAC5
modulates NRSF/REST, an important HDAC5 binding partner and
co-regulator of NR1 and many other neuronal genes.
Note: Supplementary information and chemical compound information is available on
the Nature Chemical Biology website.
We thank L. Zhang, B. Hannack, T. Hsu and K. Ure for technical assistance.
We thank T. McKinsey (Gilead Colorado, Inc.) for the phosphorylated HDAC5
antibody, G. Bai (University of Maryland) for the NR1-luciferase construct,
S. Goetsch for technical assistance and artwork, and A. Barbosa (University of
Texas Southwestern), J. Backs (University of Texas Southwestern) and E. Olson
(University of Texas Southwestern) for sharing reagents and helpful critique of the
manuscript. We also thank F. Gage, S. McKnight, O. Cleaver and A. Chang for
helpful comments on the manuscript. Work was funded in part by grants from
the Reynolds Foundation (J.W.S.), the US National Institutes of Health, National
Institute of Neurological Disorders and Stroke (R01NS38082; I.B.), the Ellison
Medical Foundation (AG-NS-0371-06; J.H.) and the Welch Foundation (I-1660;
J.H.). J.H. is the recipient of a fellowship from the Esther A. and Joseph
Klingenstein Fund and of the University of Texas Southwestern President’s
Research Council Young Researcher Award.
J.W.S., I.B., D.E.F. and J.H. designed experiments. J.W.S., Z.G., S.L., M.F., T.-S.T.,
D.E.F. and J.H. performed experiments. J.W.S. and J.H. wrote the paper.
Published online at http://www.nature.com/naturechemicalbiology/
Reprints and permissions information is available online at http://npg.nature.com/
1. Diamandis, P. et al. Nat. Chem. Biol. 3, 268–273 (2007).
2. Deisseroth, K. et al. Neuron 42, 535–552 (2004).
3. Tozuka, Y., Fukuda, S., Namba, T., Seki, T. & Hisatsune, T. Neuron 47, 803–815 (2005).
4. Lyons, G.E., Micales, B.K., Schwarz, J., Martin, J.F. & Olson, E.N. J. Neurosci. 15,
5. Sadek, H. et al. Proc. Natl. Acad. Sci. USA 105, 6063–6068 (2008).
6. Skerjanc, I.S. & Wilton, S. FEBS Lett. 472, 53–56 (2000).
7. Riazi, A.M., Lee, H., Hsu, C. & Van Arsdell, G. J. Biol. Chem. 280, 10716–10720
8. Gage, F.H. et al. Proc. Natl. Acad. Sci. USA 92, 11879–11883 (1995).
9. Mao, Z., Bonni, A., Xia, F., Nadal-Vicens, M. & Greenberg, M.E. Science 286,
10. Krainc, D. et al. J. Biol. Chem. 273, 26218–26224 (1998).
11. McKinsey, T.A., Zhang, C.L. & Olson, E.N. Trends Biochem. Sci. 27, 40–47 (2002).
12. Grozinger, C.M. & Schreiber, S.L. Proc. Natl. Acad. Sci. USA 97, 7835–7840 (2000).
13. Chawla, S., Vanhoutte, P., Arnold, F.J., Huang, C.L. & Bading, H. J. Neurochem. 85,
14. McKinsey, T.A., Zhang, C.L., Lu, J. & Olson, E.N. Nature 408, 106–111 (2000).
15. Vega, R.B. et al. Mol. Cell. Biol. 24, 8374–8385 (2004).
3 d1 d5 d
9 + KN93 9 + KN92
9 + KN93
9 + KN92
Figure 2 Isoxazole signaling leads to activation of CaMK, phosphorylation/
export of HDAC5 and MEF2-dependent gene expression in NSCs. (a) MEF2
reporter gene (3XMRE-luc) activation in 9-treated HCN cells.
(b) Accumulation of phosphorylated HDAC5 in the cytoplasm of 2-d 9-
treated HCNs (top). Compound 9 led to decreased phosphorylated HDAC5
in the nucleus over time, normalized to total HDAC5 and CREB levels
(bottom). (c) Shown are representative fields of live-cell GFP fluorescence in
vehicle or 20 mM 9-treated HCN cells expressing wild-type GFP-HDAC5
(AdGFP-HD5 (WT)) or S259A S498A mutant GFP-HDAC5 (AdGFP-HDAC5
(S-A)). Scale bar, 5 mm. (d) Expression of signal-resistant mutant HDAC5
(S-A) normalized to a control GFP plasmid. (e) Activation of NR1 is depen-
dent on MEF2 and Sp1 binding. (f) Blocking CaMK with 2.5 mM KN93, and
not KN92, resulted in an inhibition of 9-mediated neuronal differentiation
in HCN cells in 2 d cultures. Scale bar, 5 mm. (g) Quantification of Tuj1 and
Map2ab+cells is shown. Values in a, d and e represent the average of 12
replicates ± s.d. Values in g represent the average of two replicates ± s.d. of
one representative experiment from two independent experiments.
410VOLUME 4 NUMBER 7JULY 2008
NATURE CHEMICAL BIOLOGY
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology