Identifying the Mechanisms That Govern Synaptic Targeting in the Visual System

reln ctb

Retinal projections to the mouse LGN labeled by cholera toxin B. Axons from the ipsilateral eye are shown in green and those from the contralateral eye are shown in red. Note the mistargeted retinal axons (arrowheads) in reeler mutants (reln rl/rl), which lack functional reelin protein. (Images adapted from Su et al. 2013; see also Su et al. 2011, Fox and Guido 2011)

Our efforts to uncover mechanisms that drive synaptic targeting focus on the developing visual system. Specifically, we are interested in how synapses are formed between retinal ganglion cells (RGCs), the output neurons of the retina, and target neurons within the brain. Spurred on by Roger Sperry’s postulation of the chemoaffinity hypothesis (which postulated that regionalized chemical cues direct topographic mapping of RGC axons within the brain), many groups have searched for and identified chemical/molecular cues necessary for the correct exiting of retinal ganglion cell axons from the retina, the correct crossing of RGC axons at the optic chiasm, the repulsion of RGC axons from non-retinal targets, and the topographic mapping of RGC axons within the lateral geniculate nucleus (LGN) and superior colliculus (SC). Despite these monumental advances, it still remains unclear how different classes of retinal ganglion cells (RGCs)– of which there are more than 30 – target functionally distinct nuclei within the brain. One brain region where class-specific targeting of RGC axons is most evident is the LGN – a thalamic relay nucleus that contains three structurally and functionally distinct subnuclei: the ventral and dorsal LGN (vLGN and dLGN, respectively) and the intergeniculate leaflet that separates them. Since different classes of RGCs target either vLGN and IGL or dLGN, we hypothesized that regionalized guidance cues must exist in the LGN to direct axonal targeting. We have identified many candidate molecules that may act as class-specific targeting cues for retinogeniculate targeting and are now testing their necessity and sufficiency in retinogeniculate circuit formation.

These studies have further motivated us to examine how all axons target visual thalamus, not just retinal axons. For example, in the LGN non-retinal inputs originate from the cortex, superior colliculus, thalamic reticular nucleus and brainstem and together outnumber retinal inputs onto thalamic relay neurons 9 to 1. Despite the vast number of these non-retinal inputs in the mammalian LGN, we know nothing regarding the molecular mechanisms governing their formation. In a set of recent studies with the Guido lab (U.Louisville) we discovered that retinal inputs and aggrecan both play an instructive role in the unique timing of cortical inputs in LGN. We continue to actively investigating the molecular mechanisms that regulate the spatial and temporal targeting of these axons to the LGN in mice.

cg timing

Top panels: Schematic depiction of the timing of retinal and cortical innervation to the dLGN. Retinal axons are shown in red; cortical axons are shown in green. Synapses are illustrated by red or green dots. Bottom panels: Development of corticogeniculate projections in golli-tau-gfp transgenic mice. Projections from layer VI cortical neurons are labeled with tau-GFP in these mice. For all panels dLGN are encircled by white dots. D – dorsal; V – ventral; M- medial; L – lateral. Adapted from Brooks et al. 2013.

Identifying Molecules that Direct Synaptic Differentiation in the Mammalian Brain

Once synaptic partners have correctly targeted each other, both sides of the synapses must transform to form a functioning synapse (a process called synaptic differentiation). We are interested in identifying and characterizing trans-synaptic cues that direct synaptic differentiation in the mammalian central nervous system.

Interestingly, retinal synapses in the dLGN appear morphologically distinct from retinal synapses in all other regions of the brain. In fact, it appears that single retinal axons that branch to innervate multiple brain regions will form morphologically and functionally unique synapses in dLGN compared with other retino-recipient regions. For example, in dLGN at least 2 flavors of retinogeniculate synapses exist – those with a single retinal input (termed simple retinogeniculate synapses) and those which contain retinal inputs that originate from many (as many as 14!) different retinal axons (termed complex retinogeniculate synapses). We are now exploring molecules that are necessary for distinct steps in terminal development in dLGN. We found that FGF22, a molecule we previously identified as being critical for nerve terminal assembly at the neuromuscular junction, is important for the initial formation of retinal terminals in mouse dLGN (Singh et al. 2012). Now we are investigating what transforms these immature synapses into simple and complex retinogeniculate synapses. Several intriguing candidates have been identified and are under current investigation. We are particularly interested in what drives the assembly of complex retinogeniculate synapses as such studies may provide tools to begin to investigate the functional significance of this newly appreciated high level of retinal convergence onto thalamic relay cells. Image # 1

Left: Serial Block Face Scanning Electron Microscopy 3D reconstructions of a single retinal axon (red) that synapse onto both dendrite and soma of a relay cell (yellow) in adult mouse dLGN, forming a complex RG synapse.

Right: Serial Block Face Scanning Electron Microscopy 3D reconstructions of a several retinal axon synapsing onto shared region of a relay cell dendrite (yellow) in adult mouse dLGN.

In a second set of synaptogenic studies, we continue to define novel roles for a unique family of extracellular matrix proteins in synaptic development in the brain. This family of extracellular matrix molecules, termed unconventional (or non-fibrillar) collagens, has been found to direct synaptic differentiation and maturation at the neuromuscular junction (NMJ) — a large peripheral synapse between motoneurons and muscle fibers. Specifically, controlled proteolysis of several collagen molecules at the NMJ generates soluble peptides that exhibit unique bioactivities compared to the full-length molecule from which they are derived. These proteolytically released fragments of collagen molecules are termed ‘matricryptins’ and at the NMJ collagen-derived matricryptins have been shown to direct pre- and postsynaptic assembly and maturation. Based upon bio-activities of these matricryptin-releasing collagens at the NMJ, we are now asking whether similar collagens (or their matricryptins) are necessary and sufficient to induce the formation of central synapses. Why is this important? Besides advancing our basic knowledge of brain development, these families of ECM molecules are highly mutated in humans and many of these mutations cause unexplained neurological deficits (including schizophrenia, autism spectrum disorders, and epilepsy).

Characterizing How Infectious Agents Alter Neural Circuits in the Mammalian Brain

Our labs newest direction involves understanding how infectious agents alter the maintenance of neural circuits in the mammalian brain. We are particularly interested in Toxoplasma gondii, an obligate, intracellular parasite that can resides in the brain and skeletal muscle of most warm-blooded animals. Approximately 25% of the US population is infected with this parasite and once infected you remain infected for life. Active infections with Toxoplasma gondii   in infants, HIV/AIDS patients, or those with weakened immune systems can lead to toxoplasmosis. However, a number of more recent studies have revealed chronic infections of this parasite can lead to altered behaviors and have been associated with various neurological diseases. In fact, infection with Toxoplasma gondii appears to be a higher risk factor for developing schizophrenia than any single gene mutation identified to date. With this in mind, we are actively investigating how infection with Toxoplasma gondii  alters neural circuits that have been previously linked to schizophrenia.

Image # 2 for website

Left: Serial Block Face Scanning Electron Microscopy of a Toxoplasma gondii cyst (green) residing inside a cortical neuron (yellow).

Right: Immunostaining of excitatory and inhibitory synapses in control and toxoplasma-infected mice.  Infected mice display spontaneous seizures, as shown in the EEG and EMG traces.