32
efficiency of Purkinje cell differentiation, it requires a large amount of growth factors,
and it does not appear to induce the characteristic morphology, such as dendritic
extension and large soma. This protocol has not been widely used by other groups
thus far. Improved differentiation protocol of mESC-derived Purkinje cells utilized
coculture system with feeder cells to promote generation, survival, and maturation
(Tao et al. 2010 ). The efficiency was a twofold to ninefold increase in numbers of
Purkinje cells. They showed somewhat morphological and electrophysiological fea-
tures of Purkinje cells, but reproduced results using hPSCs have not been shown.
2.2.3 Self-Induction of Cerebellar Neurons from hPSCs
Several labs had attempted cerebellar induction from PSCs (Erceg et al. 2010 ;
Salero and Hatten 2007 ; Su et al. 2006 ; Tao et al. 2010 ), but the efficiency of dif-
ferentiation has been low. Therefore, we took a different indirect strategy that
focuses on mimicking the innate program of initial cerebellar development associ-
ated with isthmic organizer (Muguruma et al. 2010 ). In SFEBq culture of mESCs,
robust differentiation (>80% efficacy) of En2+ progenitors, representing the caudal
midbrain to rostral hindbrain regions, is observed when the cell aggregate is tran-
siently treated with moderate caudalizing factors (Fgf2 and insulin). A majority of
the induced progenitors express rostral hindbrain markers. Further treatment with a
hedgehog inhibitor, which promotes dorsalization indirectly by inhibiting the ven-
tralizing effect of endogenous Shh, causes expression of Ptf1a and Kirrel2 (a cell
surface marker upregulated by Ptf1a, a transcription factor essential for Purkinje
cell generation) (Mizuhara et al. 2010 ), suggesting efficient generation of cerebellar
neuroepithelium. The induced Kirrel2+ progenitors subsequently differentiate into
Purkinje cells at high efficiency when cocultured with mouse RL-derived granule
cells. It is not exactly known why the combination of Fgf2 and insulin is so effective
for induction of cerebellar neurons in SFEBq culture. An intriguing finding in this
context is that insulin, a commonly used additive in various culture media, has mod-
erate but substantial caudalizing effects on ESC-derived neural progenitors. Indeed,
the addition of insulin suppresses the expression of Six3 and Rax (rostral forebrain
markers) and increases the expression Fgf8 and En2 (expressed in MHB) (Muguruma
et al. 2010 ; Wataya et al. 2008 ), although the exact in vivo role of insulin (or IGF)
signaling in mammalian neural patterning remains to be elucidated.
These two factors do not seem to play instructive roles in rostral hindbrain induc-
tion; rather, these moderate signals may be permissive for differentiation of broad
MHB regions in the mESC aggregate. Importantly, the combination of insulin and
Fgf2 can reproducibly support the self-formation, albeit small in area, of around
isthmic organizer tissue (En2+/Otx2+) in culture, which secondarily promotes the
specification of cerebellar anlage in the aggregate by emanating the organizer
K. Muguruma