Wiring the visual system,

from development to circuit function.


We seek to understand the developmental programs that lead to the self-assembly of neuronal circuits.

Put your brain in numbers: an adult human brain has ~80 billion neurons, and each can form multiple connections (synapses) with other neurons, adding up to more than 100 trillion synapses. It makes sense that brain function is supported by such patterns of synaptic connectivity. But how do particular anatomical features in individual neurons and circuits support information processing? And how are those circuits assembled during development?

Optic lobe of the adult fly. The first visual motion direction sensitive neurons are labeled in green.

We study the developmental programs underlying neural circuit development and wiring specificity in the fruit fly visual system. The fly and human eyes are remarkably similar in their neuronal diversity, cellular oganization and neuronal computations. Importantly, cellular and molecular strategies that are evolutionary conserved are thought to assemble both vertebrate and the fly visual systems. Altogether, this allows us to make use of the genetic toolkit available in the fly to gain insight into common design principles that act in the fly and in humans, which build and support the function of the nervous system.

We are using single-cell transcriptomics to identify which genes are active in each neuron during development. We then combine sophisticated genetic manipulations, live imaging, functional and behavioral approaches to study the mechanisms of neural circuits’ formation during development, and how function and behavior could be compromised when changes in neural wiring arise.

Our Research

We are currently focusing our studies on the fly motion detection circuitry. Here, as in humans, dedicated sets of neurons process directional motion information: some neurons are sensitive to a stimulus presented to the eye moving from left to right, others top to bottom, and so forth.

In both flies and humans, visual motion sensing neurons are crucial for the individual to perceive broad visual motion and thus navigate their environment.

A fundamental feature in the motion detection system is that all neurons share a general morphology. However, each neuron type is characterized by a combination of specific morphological attributes that dictate the neuron’s direction selectivity.

This well-known relationship between neuronal structure and function makes the motion detection system an invaluable model to unveil the developmental programs dictating the wiring specificity of each neuron in the network. In particular, we can use it to ask how changes in gene activity between highly similar neurons can lead to the specific morphologies that establish each neuron's function.

3D reconstruction of a motion sensing neuron dendrite highlighting its synaptic inputs

We manipulate the development of synaptic connectivity, and then use two-photon imaging and behavioral paradigms to ask how the integrated function of the circuit and the motion driven behaviors of the adult fly are affected.

If you want to know more...

In most visual systems, visual input from the retina is retinotopically mapped onto the brain, so that adjacent points in space are represented as adjacent regions on the retinotopic map optic lobe. In the fly compound eye, ~800-unit eyes collect information from 800 points in the visual space. For each point, visual information is passed on from photoreceptors onto a single columnar unit of neurons. The repeated array of columns builds a retinotopic map, maintained across the four optic neuropile ganglia: Lamina, Medulla, Lobula, and Lobula plate.

Within the fly optic lobe, visual motion information is processed in two parallel pathways: the ON and OFF pathways, which detect motion as brightness increments or brightness decrements, respectively. The first direction-selective neurons in each pathway are the T4 (ON) and T5 (OFF) neurons. This specialization results from their dendrites arborizing in different neuropiles. T4 dendrites innervate in the Medulla and T5 dendrites in the Lobula, where they receive synaptic input from different pre-synaptic neurons.

Schematic of an adult fly optic lobe highlighting the ON (T4) and OFF (T5) pathways of the motion detection circuit. Visual inputs propagate retinotopically via T4 and T5 neurons to the four layers of the lobula plate. Four subtypes of T4 and of T5 neurons, each directionally tuned to one of the four cardinal directions, converge into layers a and b of the lobula plate (horizontal motion sensing system), or layers c and d (vertical motion sensing system).

There are four T4 and four T5 cells per column, each tuned to one of the four cardinal directions of visual motion: front-to-back, back-to-front, upwards and downward. The orientation of dendritic arbors in each subtype correlates with its own directional tuning preference.It is thought that the specificity of directional tuning of each subtype results from this spatial organization.

All T4/T5 neurons retinotopically project their axons to the Lobula plate, where they segregate according to directional output, producing four layers, each tuned for motion in one of the cardinal directions:(layer a, front-to-back; layer b, back-to-front, layer c, upward; layer d, downward). In each Lobula Plate layer, the downstream postsynaptic partners of T4/T5 neurons integrate their local motion signals, producing direction-selective wide-field motion responses that match such selectivity. This allows the fly, like every other sighted animal, to use directional motion information in tasks such as visual course control, prey capture or predator avoidance.

Retinotopic T4 (green) and T5 (red) neurons.

Thus, direction selectivity involves the establishment of neuronal subtypes, morphologically defined by three parameters:

1) the projection of their dendrites to either the Medulla (T4) or Lobula (T5) neuropiles

2) the orientation of their dendritic arbors in one of the four cardinal directions allows them to connect to upstream neurons

3) the organization of axonal outputs in one of four differentiated Lobula Plate layers

By investigating the developmental programs that specify each T4/T5 subtype, we will be able to understand how a complex network of eight coincident retinotopic maps is established at the single cell level, and how the acquisition of neural identity generates a circuit of complex synaptic connections.

We have previously identified the mechanisms by which neuroprogenitors generate all T4/T5 neurons and concurrently establish their retinotopic organization (Pinto-Teixeira, Cell 2018). These findings, linking circuit development to computational function, inspire our current efforts. We aim to identify the developmental programs downstream of T4/T5 fate specification that differentiate and wire each subtype into complex synaptic circuits to support visual motion detection.

Drosophila head drawing based on original by Helfrich-Förster.

3D dendritic reconstruction done with NeuroNLP.