Molecular and circuit analysis of stable contrast processing in the visual systemDoctoral thesis
Date of Examination:2022-07-14
Date of issue:2022-08-01
Advisor:Prof. Dr. Marion Silies
Referee:Prof. Dr. Silvio Rizzoli
Referee:Dr. Jan Clemens
Referee:Prof. Dr. Jochen Staiger
Referee:Prof. Dr. Tobias Moser
Referee:Prof. Dr. Alexander Ecker
Files in this item
This file will be freely accessible after 2023-07-13.
EnglishVisual perception gives us a reliable estimate of the world throughout the day, thanks to the computations implemented in visual circuits. In a given space or time, the relative change in luminance, contrast, forms the basis for many downstream computations like the detection of edges, shape, orientation and motion. Stable contrast estimation is challenged by the immensely changing lightning conditions that can slowly change throughout the day or much more rapidly when viewing natural scenes. To ensure reliable behavioral responses to the same (contrast) stimulus, visual systems face the monumental task of keeping contrast representations stable. Gain mechanisms operate across the nervous system to match neural responses to relevant stimulus distribution. In visual systems, photoreceptors have numerous mechanisms that implement sensory gain to keep contrast representations stable across the slow changes of luminance occurring throughout the day. However, major changes in luminance also occur at fast timescales due to the luminance distributions within natural scenes which we view or through which we navigate. The extend of photoreceptor gain is often insufficient in these conditions but vision can still operate luminance-invariantly as shown in human perception, neurons in the vertebrate cortex, lateral geniculate nucleus and the retina. This highlights the necessity of post-receptor luminance gain mechanisms and evidence from a few studies suggests that this happens in the retina. However, a detailed understanding of the underlying circuitry, including the contribution of different cell types, their specialization and the molecular mechanisms underlying the post-receptor gain, is not understood in any visual system. A recent study from Drosophila melanogaster showed that fly behavior to stimuli containing contrast decrements (OFF stimuli) is also luminance-invariant in rapidly changing conditions (Ketkar et al. 2020). Two second order neurons located in the lamina and post-synaptic to photoreceptors, L2 and L3, contribute to the OFF behavior by encoding contrast and luminance respectively. L3 neurons underlie the luminance-invariant behavior by providing a rapid luminance gain to the OFF pathway (Ketkar et al. 2020). This opened up the possibility of investigating post-receptor gain using the advanced fly genetic arsenal. In the first part of this thesis, we asked if the ON pathway also utilizes a post-receptor gain and which second-order neurons support this. The fly ON pathway was thought to receive major inputs only from a single second-order neuron type, L1 neurons, instead of the contrast and luminance sensitive parallel pathways formed by L2 and L3 within the OFF pathway. Using calcium imaging and behavioral paradigms, we revealed that all lamina neurons encode distinct features of the visual scenery and distribute these to both ON and OFF pathways showing that ON and OFF pathways only arise downstream of the lamina neurons. ON behavior also requires a luminance gain and this is provided by both L1 as well as L3 inputs. For the rapid luminance gain, L1 and L3 provide distinct contributions since their luminance encoding differ. Thus, a combination of temporal filtering leading to distinct contrast-luminance specializations in the second-order visual neurons is vital for downstream stable contrast processing. The bipolar cells, the second-order visual neurons of the vertebrate retina, also form parallel temporal channels suggesting that these specializations can serve distinct behavioral roles like in the fly visual system. In the second study, I investigated the molecular factors shaping the distinct feature encoding properties of the lamina cell-types. Biophysical properties of neurons that are largely shaped by ion channel expression underlie a significant fraction of their signal processing properties. Using RNA-seq data and endogenous protein tagging, I have identified a specific subtype of voltage gated potassium channels, the Ka channels Shaker and Shal, as being highly expressed in the contrast sensitive L2 neurons. Using pharmacological manipulations, I have revealed the circuit wide role of Ka channels in enhancing L2 contrast responses. Additionally using cell-type specific RNAi, I have shown that Ka channels are involved in sharpening the L2 contrast responses. Similarities in the other fly species suggest the conserved role of Ka channels in shaping distinct cell-types and conferring their computational role within visual circuits for contrast processing. In the third study I investigated visual circuits of the OFF pathway that implement the rapid luminance gain. I characterized the contrast encoding properties of all major celltypes in the OFF circuitry and identified the dendrites of two distinct third-order medulla neurons as the location where the rapid luminance gain arises. Using the specific features provided by the lamina neurons, the medulla neurons Tm1 and Tm9 implement a rapid luminance gain which scale their representations of contrast in distinct ways. Whereas Tm1 neurons reach luminance-invariant estimations of contrast, Tm9 neurons boost contrast signals in low luminances. Spatial pooling underlies the luminance gain in both neurons and both neurons receive wide glutamatergic signals yet use distinct molecular mechanisms to the implement luminance gain. Genetic analysis shows that Tm9 relies on inhibition mediated by GluClα channels whereas the molecular basis for luminanceinvariance in Tm1 remains yet to be identified. This study reveals the details of how visual systems implement a gain mechanisms vital for stable vision in natural scenes. A post-receptor gain acting specifically in dim light in vertebrates is also implemented in the third-order neurons suggesting the presence of convergent strategies for stable vision. Testing if distinct luminance gain channels exist in vertebrates and whether similar spatial and molecular properties underlie the luminance gain can reveal if convergent strategies are implemented for natural scene processing across species. My initial studies added to the numerous examples of convergent and divergent processing happening within neural circuits. One outcome of such complex wiring of neural circuits is that specific synaptic connections made between distinct synaptic partners, can contribute to information processing differently. To get a detailed causal understanding of specific cell-type to cell-type connections, manipulations restricted to specific synapses are necessary. However current manipulations maximally achieve specificity that affects all presynapses of one cell-type and thus limit the level of detail at which can understand neural circuit function. In my final study, we are developing a genetic tool, Synapse Targeted Activity Block (STAB), to manipulate synaptic connections between distinct partners. STAB is based on two components each expressed in one of the partners and only activates when these components coincide, which happens in the synaptic connections. One component is a viral protease (TEV protease - TEVp) which cleaves a conditionally integrated cleavage site (TEV cleavage site - TEVcs) in an endogenous synaptic protein which constitutes the second component. We generated TEVp fusion proteins that localize to synaptic regions and function extracellularly both in vitro and in vivo. Concentrating on the transsynaptic cleavage of postsynaptic receptors that are known to be essential for visual circuit function, in vitro experiments suggested that TEVp cleavage leads to degradation of post-synaptic receptor GluClα containing a TEVcs but the levels of degradation was not enough to recapitulate LOF phenotypes in vivo. We generated an optimized TEVp version which showed higher levels of GluClα degradation. STAB with the enhanced TEVp needs to be validated in vivo before applying it for investigating specific synaptic connections. STAB is the first example of synapse specific manipulations that opens up a door of highly specific investigations of neural circuits and get detailed causal insights on neural circuits, computations and behavior. Overall, my work thus combined the molecular and circuit analysis of stable contrast computation in dynamically changing environments. The development of the genetic STAB tool will allow circuit analysis at higher resolution than possible to date, and can be applied to studying visual computations, as well as any other type of circuit level analysis.
Keywords: visual system; drosophila; genetic tools; luminance invariance