doi:10.1016/j.cub.2006.04.012


Dispatch
R361
Dispatches

Eye Development: Random Precision in Color Vision
In insect and vertebrate eyes, different types of color-detecting
photoreceptors are randomly distributed throughout the retina. A recent
study has provided important new insights into the developmental
mechanisms that generate the retinal mosaic required for color vision.
Thomas Hummel
and Christian Klämbt

It is not quantity which counts
[with colors], but choice and
organization

(Henri Matisse)

To collect information about the
outside world, eyes have evolved in
different animal species as highly
specialized sensory structures
which, although anatomically quite
diverse, show common
organizational features. The visual
system not only allows the
perception of shape and motion,
but also can often discriminate
colors. Light of different
wavelength is detected by a set of
Opsin type receptor molecules with
distinct spectral properties. They
are expressed in specific
photoreceptor types following the
‘one-receptor-one-neuron’ rule
characteristic for most sensory
neurons [1,2]. Interestingly, within
the human and the fly retina, the
different color detectors are
distributed in a random fashion, but
so far the developmental
mechanisms underlying stochastic
pattern formation have been
elusive. Using the Drosophila eye
as a model, the Desplan lab [3] has
now been able to link the
stochastic pattern of color-
sensitive photoreceptors to the
expression of a single transcription
factor.

The Drosophila compound eye
has been an extremely useful
model system for understanding
the logic of pattern formation [4,5].
Here 750–800 single eye units
called ommatidia are assembled
into an almost crystalline-like
arrangement. This ordered pattern
continues in the organization of the
eight photoreceptor cells (R cells)
inside each ommatidium. On the
basis of their different spectral
sensitivities, the eight R cells can
be grouped into two functional
classes. Six monochromatic outer
photoreceptor cells (R1–R6), the
functional homologues of
vertebrate rods, express only one
type of rhodopsin (Rh1) and are
involved in motion detection [6]. In
contrast, two inner photoreceptor
cells, R7 and R8, express one out of
several rhodopsins with distinct
spectral sensitivities (Rh3–Rh6),
allowing them to function in color
discrimination, much as cone cells
do in vertebrate eyes [7].

In the fly retina, two invariable
R7/R8 combinations define the
corresponding two color-sensitive
ommatidia classes (Figure 1A). In
‘pale’ ommatidia, sensitive to
shorter wavelengths, an R7 cell
expressing Rh3 always matches
with an R8 cell expressing Rh5; and
in ‘yellow’ ommatidia, specialized
in the perception of longer
wavelengths, an Rh4-positive R7
cell is housed together with an
Rh6-positive R8 cell. The
distribution of pale and yellow
ommatidial subtypes does not
follow the morphologically
homogeneous retinal pattern, but
seems to be random, resulting in
a functional mosaic inside the
retina [8]. Although the distribution
is stochastic, however, the ratio of
photoreceptor cell types is well
defined, with 30% pale and 70%
yellow ommatidia (Figure 1A).
How is a random but biased
mosaic pattern generated in an
otherwise regularly organized
sensory field?

The process of R cell
differentiation in Drosophila
extends over a period of five days
and can be divided into an early
specification period, in which the
different R cell subtypes first
acquire their neuronal ground
states, followed by their terminal
differentiation into functional
photoreceptors with class-specific
rhodopsin expression [9].
Important insights into the
mechanisms underlying
photoreceptor maturation came
from the analysis of mutations that
disrupt the strict pairing of Rh3/
Rh5 and Rh4/Rh6 in the pale and
yellow ommatidia, respectively
[10]. When all R7 cells are missing,
as in the sevenless mutant, the
number of R8 cells is unaffected,
but they express exclusively Rh6,
indicating that the decision
between a yellow or pale fate is
initiated when the R7 cell in each
ommatidium makes a choice of
whether to express Rh3 or Rh4
(Figure 1B,C). An Rh3 decision is
then communicated to the adjacent
R8 cell, leading to the induction of
Rh5 expression. Although the
signal from R7 to R8 is still obscure,
we now have a satisfying
understanding of what triggers the
initial choice of R7 between the Rh3
or Rh4 status.

Genetic studies identified the
spineless gene, which encodes
a bHLH-PAS transcription factor,
as the critical determinant of both
the initial yellow R7 fate choice and
the stochastic distribution of the
two types of color-sensitive
ommatidia [3]. In spineless
mutants, R cells develop normally
through the first phase of cell type
specification, leading to a regularly
patterned and normal sized
compound eye. Visualizing
rhodopsin expression, however,
revealed the presence of only the
pale ommatidia subtype with a
wild-type Rh3/Rh5 pairing.
Subsequent clonal analysis
demonstrated a strictly cell-
autonomous function of spineless
in the yellow R7 cell for the
induction of Rh4 expression.

The notion that spineless
expression triggers Rh4
expression received further
support from gain-of-function
experiments: spineless expression
in all developing photoreceptor



Current Biology Vol 16 No 10
R362
Yellow 70%Pale 30%

R6R1

R7

R8

rh4

rh6

rh3

rh5

spineless

20–40%

spineless

Pupae

R6R1

R7

R8

R7

R8

60–80%

R7

R8

R7

R8

R7

R8

A

B

C

Adult

Current Biology

Figure 1. Organization and development of Drosophila color-sensitive photoreceptors.

(A) In the adult retina, each ommatidium contains 6 outer (R1–R6) and 2 inner (R7 and
R8) photoreceptor cells. Specific combinations of rhodopsin expression define pale
(Rh3/Rh5) and yellow (Rh4/Rh6) ommatidia, which are randomly distributed throughout
the retina in a 3:7 ratio. (B,C) Different stages of R7/R8 differentiation in late (B) and mid
(C) pupal development. Different levels of spineless expression lead to a cell-autono-
mous Rh3/Rh4 fate choice in R7, followed by an induced or default R8 differentiation.
neurons resulted in an efficient
ectopic Rh4 activation. In R8, the
endogenous, cell type specific
rhodopsin expression was
maintained leading to a violation of
the ‘one-receptor-one-neuron’
rule; in R7 cells, however, spineless
induced a perfect switch from Rh3
to Rh4 terminal differentiation
status. Furthermore, expression of
spineless even in late
differentiating pale R7 cells, using
a rhodopsin promoter construct,
was able to trigger the switch from
Rh3 to Rh4 expression. The
induced Rh5 expression in the
neighboring R8 cell cannot be
reverted, however, resulting in
untypical ommatidia with Rh4/Rh5
inner receptor cells pairing. From
these results, Wernet et al. [3]
concluded that a single
transcription factor, Spineless, is
both necessary and sufficient to
initiate the assembly of the yellow
type ommatidium, first specifying
Rh4 expression in R7, which then
prevents the induction of the R8
cell to express the Rh5 protein
(Figure 1C).

Spineless also has a direct role in
establishing the stochastic
distribution of the color-sensitive
ommatidia in the retina [3].
Expression of spineless RNA
occurs in only a randomly
distributed 60–80% of R7 cells,
which presumably will later
differentiate into yellow type R7
photoreceptors (Figure 1C).
Furthermore, the level of
expression varies greatly among
the Spineless-positive R7 cells,
suggesting a direct link between
a Spineless threshold and
Rhodopsin expression choice.
Temporally enhanced spineless
expression leads to a clear
increase in the number of yellow at
the expense of pale ommatidia. In
contrast, this ratio decreases
following experimental reduction of
spineless expression, either by
inactivating one gene copy or by
adding transgenes carrying the
spineless eye enhancer, which
compete for endogenous
activating factors. These data thus
support a two-step model
combining cell-autonomous
mechanisms and cell–cell
communication to establish cell
fate and distribution of the two
color-sensitive ommatidia types.

The cell autonomous function of
Spineless implies the existence of
R7-specific or combinatorial acting
upstream factors acting on a 1.6
kilobase enhancer fragment of the
spineless regulatory region. Neither
the activating factors nor the final
mechanisms that set the Spineless
threshold for inducing Rh4
expression are known. It is
possible that, once the fate choice
is made, further stabilization is
needed, as it has been reported for
the Rh5/6 decision in the R8 cell
[11]. At the same time, Spineless,
which is expressed only in a short
pulse during eye development,
likely activates downstream
targets that control the expression
of Rh4 sensory receptor, but
concomitantly leads to the
exclusion of Rh3 receptor
co-expression. In vivo, spineless
and rhodopsin expression are
temporally separated by one day,
but even when spineless
expression is forced during the
time of rhodopsin expression, it
can reprogram terminal fate [3].
Additionally, Spineless or its
downstream targets antagonize
the generation of an inductive
signal, yet to be identified,
emanating from the R7 cell.
Interestingly, like Spineless, other
members of the bHLH/PAS
transcription factor family often
control the initial choice in various
cell fate decisions [12]. As bHLH/
PAS transcription factors
commonly act as heterodimers, the
identification of Spineless
interaction partners will be very
informative regarding the
mechanism through which
Spineless leads to a selective
activation of distinct differentiation
programs.

Finally, sensory receptor
specificity in the peripheral
nervous system has to be matched
with precise synaptic connections



Dispatch
R363
in the central brain to allow color
discrimination. In Drosophila, the
integration of sensory information
coming from randomly distributed
receptor neurons into the visual
system’s topographic organization
has not been analyzed. Although
most of the cellular components
and their projection patterns in the
fly visual system have been
described [13], our understanding
of how the brain ‘sees’ the
colorful world is still in its
beginnings.

References
1. Mombaerts, P. (2004). Odorant receptor

gene choice in olfactory sensory neurons:
the one receptor-one neuron hypothesis
revisited. Curr. Opin. Neurobiol. 14, 31–36.

2. Mazzoni, E.O., Desplan, C., and Celik, A.
(2004). ‘One receptor’ rules in sensory
neurons. Dev. Neurosci. 26, 388–395.

3. Wernet, M.F., Mazzoni, E.O., Celik, A.,
Duncan, D.M., Duncan, I., and Desplan, C.
(2006). Stochastic spineless expression
Bacterial Cell Bio
Magnetosomes

Sensing of magnetic fields by living
best understood in magnetotactic ba
new insight into the biogenesis of ba
these organelles to a newly recognize
which organizes magnetosomes into
aligning cells with the geomagnetic

Craig Stephens

Several centuries ago, humans
learned to construct navigational
compasses that could sense the
earth’s magnetic field [1]. By that
time, the living world was millions
of years ahead of us in
geomagnetic sensing technology.
While the significance and
mechanisms of magnetosensing in
animals, such as migratory birds,
fish and insects, that execute
remarkable global navigational
feats have been debated for years,
the biological compass mechanism
we know the most about at the
cellular level is found in
magnetotactic bacteria. These
aquatic microbes are thought to
use their internal magnets for the
relatively mundane task of pointing
themselves downward, toward
their preferred homes in oxygen-
depleted sediments [2]. We will
creates the retinal mosaic for colour
vision. Nature 440, 174–180.

4. Voas, M.G., and Rebay, I. (2004). Signal
integration during development: insights
from the Drosophila eye. Dev. Dyn. 229,
162–175.

5. Reifegerste, R., and Moses, K. (1999).
Genetics of epithelial polarity and pattern
in the Drosophila retina. Bioessays 21,
275–285.

6. Heisenberg, M., and Buchner, E. (1977).
The role of retinula cell types in visual
behavior of Drosophiila melanogaster.
J. Comp. Physiol. 117, 127–162.

7. Chou, W.H., Hall, K.J., Wilson, D.B.,
Wideman, C.L., Townson, S.M., Chadwell,
L.V., and Britt, S.G. (1996). Identification
of a novel Drosophila opsin reveals
specific patterning of the R7 and R8
photoreceptor cells. Neuron 17,
1101–1115.

8. Franceschini, N., Kirschfeld, K., and
Minke, B. (1981). Fluorescence of
photoreceptor cells observed in vivo.
Science 213, 1264–1267.

9. Mollereau, B., Dominguez, M., Webel, R.,
Colley, N.J., Keung, B., de Celis, J.F., and
Desplan, C. (2001). Two-step process for
photoreceptor formation in Drosophila.
Nature 412, 911–913.

10. Chou, W.H., Huber, A., Bentrop, J.,
Schulz, S., Schwab, K., Chadwell, L.V.,
logy: Managing

organisms — magnetosensing — is
cteria. Recently work has provided
cterial magnetosomes, and links
d prokaryotic cytoskeletal filament
a sensory structure capable of

field.

discuss here recent insights into
how ‘magnetosomes’, the
membrane-enclosed magnetite
crystals central to bacterial
magnetosensing, are produced
and organized [3,4].
Magnetosome-like structures have
been observed in many animals,
and the work discussed here may
provide insight into the
development, function and
evolution of magnetosomes in
eukaryotes.

Experimental work on bacterial
magnetotaxis began over 30 years
ago, when Richard Blakemore
made the curious observation that
a population of motile bacteria
from salt marsh mud responded
dramatically to magnetic
manipulation [2]. Since
Blakemore’s initial discovery,
magnetotactic bacteria have been
found in freshwater and marine
sediments around the world [5].
Paulsen, R., and Britt, S.G. (1999).
Patterning of the R7 and R8
photoreceptor cells of Drosophila:
evidence for induced and default cell-fate
specification. Development 126, 607–616.

11. Mikeladze-Dvali, T., Wernet, M.F., Pistillo,
D., Mazzoni, E.O., Teleman, A.A., Chen,
Y.W., Cohen, S., and Desplan, C. (2005).
The growth regulators warts/lats and
melted interact in a bistable loop to
specify opposite fates in Drosophila R8
photoreceptors. Cell 122, 775–787.

12. Crews, S.T. (1998). Control of cell
lineage-specific development and
transcription by bHLH-PAS proteins.
Genes Dev. 12, 607–620.

13. Fischbach, K.F., and Dittrich, A.P. (1989).
The optic lobe of Drosophila
melanogaster. I. A Golgi analysis of
wild-type structures. Cell Tissue Res. 258,
441–475.

Institut für Neuro- und
Verhaltensbiologie, Universität Münster,
48149 Münster, Germany.
E-mail: hummel@uni-muenster.de;
klaembt@uni-muenster.de

DOI: 10.1016/j.cub.2006.04.012
Most magnetotactic bacteria seen
in the Northern hemisphere are
north-seeking, and most in the
Southern hemisphere are south-
seeking [6,7]. Why is this?
Blakemore hypothesized that,
because of the significant vertical
component of the geomagnetic
field at latitudes away from the
equator, alignment of a bacterial
cell with the geomagnetic field
would facilitate downward
migration by north-seeking
bacteria in the northern
hemisphere (and conversely in the
south) [2]. Since the magnetotactic
bacteria isolated so far prefer
anaerobic or microaerobic
conditions, if they find themselves
in an O2-rich environment, such as
the water column above the
sediment, following the
geomagnetic field
downward — and supplementing
magnetotaxis with O2 and/or redox
sensing and taxis — should help
them to find more anoxic
sediments [2,8].

The cell biology of magnetotaxis
is under active investigation [9,10].
Magnetosomes contain crystalline
particles of magnetite (Fe3O4) or
greigite (Fe3S4). The individual
crystals are generally 35–100 nm in
size, and constitute a permanent
single magnetic domain [11,12]. To
generate a sufficiently large

mailto:hummel@uni-muenster.de

	Eye Development: Random Precision in Color Vision
	References