Harvard researchers have discovered a previously unknown type of retinal
cell that plays an exclusive and unusual role in mice: detecting upward
motion. The cells reflect their function in the physical arrangement of
their dendrites, branch-like structures on neuronal cells that form a
communicative network with other dendrites and neurons in the brain.
The work, led by neuroscientists Joshua R. Sanes and Markus Meister, is described in a letter in the journal Nature.
“The structure of these cells resembles the photos you see in the
aftermath of a hurricane, where all the trees have fallen down in the
same direction,” says Meister, the Jeff C. Tarr Professor of Molecular
and Cellular Biology in Harvard’s Faculty of Arts and Sciences. “When
you look at these neurons in the microscope, they all point the same
way. There’s no other cell type in the retina that has that degree of
The cells, like other retinal neurons, are composed of a round cell
body surrounded by a tangle of dendrites. Most retinal neurons
distribute their dendrites evenly around the cell body, but the upward
motion-detecting cells arrange almost 90 percent of their dendrite
tangle exclusively on one side of the cell body.
“This lopsided arrangement literally directs the cell’s function,
orienting the dendrites downward like roots of great trees,” says
Sanes, professor of molecular and cellular biology and Paul J. Finnegan
Family Director of Harvard’s Center for Brain Science. “Because the
eye’s lens acts as a camera, reversing incoming light rays as they
strike the retinal tissue, an object moving up will result in a
downward-moving image at the back of the eye — the exact orientation
of the cells’ dendrites.”
The research builds on efforts by Meister to understand neural
processing in the retina, as well as work in Sanes’s laboratory to
identify and mark neurons in the retina using molecular tags. Recently,
they tracked down a family of molecules expressed exclusively by small
subsets of retinal cells in mice. One in particular, called JAM-B, was
present in cells that had a peculiar distribution and orientation.
According to Sanes, developmental neurologists have long tried to
identify different types of neural cells based on their function and
anatomy — how they appeared on the outside.
“But it’s a huge limitation because it’s essentially a qualitative
assessment,” he says. “We really need some way to reliably identify and
track these cells if we ever hope to study their development. So the
emergence of cell-specific molecular markers is a very big deal,
because it will do just that. Already we’ve seen that it helps us
identify new kinds of cells we didn’t know existed before. Once we have
a promising molecule, we can track down the cells that it corresponds
“The other important result,” continues Sanes, “is that we’re
actually mimicking how the brain itself identifies its cells. The brain
has to be able to reliably recognize and tell apart different kinds of
cells, and that’s going to happen on a molecular basis. In fact, it’s
possible that some of the molecules we’ve identified are, in fact, the
same molecules the brain uses to distinguish cell types.”
By identifying molecules that are solely expressed by specific types
of neurons, scientists hope to gain insights into how nerve cells form
synapses, or connections, with other nerve cells — in short, how the
brain controls its development on a molecular basis.
For the moment, however, researchers are busy puzzling over the results of the JAM-B mouse retinal cells.
“Why in the world would mice need to develop cells to detect upward motion?” Sanes wonders. “It’s a great mystery.”
Sanes and Meister’s co-authors on the Nature paper are In-Jung Kim,
Yifeng Zhang, and Masahito Yamagata, all of Harvard’s Department of
Molecular and Cellular Biology. In a separate Nature letter published
earlier this year, Yamagata and Sanes demonstrated a type of target
recognition not previously shown anywhere in the brain: They identified
four recognition molecules, each of which marks and specifies a circuit
in the retina, and showed the role of each for specific connectivity in
The current research was funded by the National Institutes of Health.