INSIGHTS | PERSPECTIVES
1566 26 SEPTEMBER 2014 • VOL 345 ISSUE 6204
The world’s growing population and industrialization result in an ever- increasing demand for energy. This poses society with a questionof criti- cal importance: Are we willing to accept the consequences of climate
change linked to the combustion of fossil
fuels in order to meet our energy demand?
An answer of “no” requires us to quickly
solve the challenging problem of developing alternative sources of energy that can
replace fossil fuels. On page 1593 of this issue, Luo et al. (1) report an important step
toward achieving that goal.
Sunlight is by far the most abundant
energy resource; harvesting just a small
fraction of 1% of available solar energy
would be enough to indefinitely satisfy
the world’s entire energy demand (2). One
major drawback of sunlight, however, is
the daily, seasonal, and regional variability of its intensity. Thus, if solar energy is
ever to be used at the scale of fossil fuels,
a method to store the energy is necessary.
Nature provides a great example of solar
energy storage by using sunlight to oxidize water to release O2 and reduce CO2 to
produce hydrocarbon (plant) materials.
The overall (sunlight to stored energy) efficiency of photosynthesis is estimated to
be only ~1%, which prevents it from being
a practical fuel source on a global scale (3).
Therefore, there has been an ongoing effort
to develop artificial photosynthesis systems
to convert sunlight and water directly into
high-energy-density chemical fuels such as
hydrogen gas, a process termed solar water
splitting (4) (see the figure).
Any viable solar water-splitting system
must satisfy four primary criteria: It needs
to be efficient, cost effective, and scalable—
thus composed of Earth-abundant materials—and stable enough to run for years
By Thomas Hamann
show promise for solar
CHEMISTRY sonic hedgehog (SHH) and bone morphogenetic protein (BMP), progenitor cells in
the neural tube acquire defined identities
characterized by specific transcription factors that condition the type of neurons they
can produce (2). For example, progenitors
located ventrally are exposed to higher concentrations of SHH (produced by the underlying floor plate and notochord tissues) and
generate neurons, including motoneurons,
that control motor output. Progenitors located dorsally are exposed to high concentrations of BMP (produced by the overlying
roof plate tissue) and produce interneurons
involved in processing sensory information.
Once morphogens establish the dorsoventral mosaic of precursor territories, the
neural tube undergoes massive growth accompanied by an increase in total cell number. In many developmental systems where
cell fate is controlled by morphogen gradients (such as the fruit fly imaginal discs),
the size of precursor territories increases
proportionally when the tissue grows (3, 4).
The situation is, however, strikingly different for the neural tube. Kicheva et al. show
that as the tube grows, the size of its individual precursor territories changes but not
in scale with overall neural tube growth.
The dimensions of each progenitor pool appear to be independently controlled and follow distinct growth patterns. Remarkably,
the different pools keep similar proportions
between embryos of different species such
as mouse and chicken.
What controls these evolutionarily conserved growth kinetics of neural progenitor pools? The rates of cell proliferation or
cell death could vary among the different
progenitor territories. Alternatively, newly
specified progenitors could be recruited
in response to morphogen exposure, thus
increasing the pool size. The rate of differentiation that controls cell exit from
the progenitor territory could also vary.
Kicheva et al. performed a thorough quan-
titative analysis of these three different pro-
cesses in chicken and mouse embryos. They
observed that progenitors in the different
territories divide at the same rate and that
cell death is negligible in the neural tube
during early stages of development. This
rules out regulation of the proliferation or
cell death rates as mechanisms that control
progenitor pool size.
Kicheva et al. also analyzed how the progenitor identity re-specification rate varies in time and demonstrate that while it
changes early on, this rate rapidly falls to
zero, indicating that the identity of the progenitor populations is stabilized early. Remarkably, the commitment of populations
to their definitive fate appears to coincide
with the peak of BMP and SHH activity in
the neural tube. The key parameter that appears to vary across progenitor territories is
the differentiation rate.
On the basis of these observations,
Kicheva et al. propose a two-phase model
for neural tube patterning (see the figure).
In the first phase, naïve progenitor territories are specified in response to exposure to
morphogens such as SHH and BMP. Subsequently, progenitors acquire their definitive
identity, which is associated with a characteristic differentiation rate that is specific
for each neuronal population. In the second
phase, growth of the progenitor pools become largely independent of the morphogen concentration and follows a specific
differentiation program imposed by the
profile of transcription factors expressed by
the progenitors. Growth kinetics of the various progenitor territories is then controlled
by the difference between the proliferation
and the differentiation rates. The model
overturns the idea that proportions are determined by scaling morphogen gradients.
Thus, the differentiation rate of motoneurons peaks early, leading to their precocious
differentiation in the ventral horn. By contrast, the differentiation rate of interneurons peaks late, allowing the formation of
a much larger compartment of progenitors
that will form the dorsal interneurons.
A major challenge will be to determine
what controls the differentiation rate in
different progenitor populations. The signaling pathway controlled by the protein
Notch, which regulates cell differentiation
in the nervous system (5), is an interesting candidate for this role. The findings of
Kicheva et al. may be relevant to growth dynamics in other tissues as well, and more
broadly, may help to explain size variability
within and between anatomical domains
across species. ■
1. A. Kicheva et al ., Science 345, 1254927 (2014); 10.1126/
2. W.A.Alaynick, T.M.Jessell, S.L.Pfaff, Cell 146,178(2011).
3. D. Ben-Zvi, B. Z. Shilo, A. Fainsod, N. Barkai,Nature 453,
4. O. Wartlick et al ., Science331, 1154 (2011).
5. A.Louvi, S.Artavanis-Tsakonas, Nat. Rev. Neurosci. 7,93
“The model overturns the
idea that proportions are
determined by scaling