Figure 3.4: Comparison of the Fly and Mouse Motion Vision Pathway.
a Fly photoreceptors depolarize in response to light, but the lamina monopolar cells invert the
signal. The medulla Mi and Tm cells are sensitive to either ON or OFF signals and differ in their temporal properties. Medulla cells with slower or faster temporal properties connect to different parts of the T4 or T5 dendrites, enabling their direction selectivity. T4 and T5 cells provide direction-selective, excitatory input to lobula plate tangential cells. In addition, T4 and T5 cells connect to lobula intrinsic cells, which provide null direction inhibition to lobula plate tangential cells. aIn contrast to invertebrates, vertebrate photoreceptors hyperpolarize in response to light.
The bipolar cells are subdivided in ON and OFF sensitive cells, similar to the medulla Mi and Tm cells of flies. The bipolar cells also differ in their temporal tuning properties, different subtypes connect to the more distal or proximal part of the starburst amacrine cell dendrites. Similar to T4 and T5 cells, the direction selectivity of starburst amacrine cells arises from both preferred direction enhancement and null direction suppression. They provide both inhibition and excitation to direction-selective retinal ganglion cells, with localized inhibitory synapses but more distributed excitatory synapses.
Comparing the fly optic lobe to the mouse retina, the photoreceptor input to the system differs substantially: while invertebrate photoreceptors depolarize in response to light, vertebrate photoreceptors hyperpolarize. However, the signal of the fly photoreceptors becomes inverted at the level of lamina monopolar cells via histamine gated chloride channels. Thus, on the level of
lamina monopolar cell output and the vertebrate photoreceptor output, the signals in the fly optic lobe and the vertebrate retina are OFF sensitive. Similar to the fly visual system, ON and OFF sensitive cells exist in the mouse retina (Stell et al., 1977); the bipolar cells as well as the starburst amacrine cells are either ON or OFF sensitive. The medulla Mi and Tm cells in the fly brain are comparable to the bipolar interneurons in the retina (Fig. 3.4). Both exist in ON and OFF sensitive subtypes and they are not direction-selective. The subtypes of bipolar cells but also Tm and Mi cells differ in their temporal response properties. Therefore, they enable preferred direction enhancement or null direction suppression in their postsynaptic neurons (Baden et al., 2013; Arenz et al., 2017).
The first direction-selective cells are the T4/T5 neurons in the fly optic lobe and the ON/OFF starburst amacrine cells in the mouse retina. The direction selectivity of T4 cells most likely originates from different mechanisms: release from inhibition together with fast excitation can enhance preferred direction response. In addition, slow inhibition at the dendritic base can inhibit responses to motion in the null direction. Starburst amacrine cells also combine null direction inhibition and preferred direction enhancement. The four different subtypes of T4 cells detect motion in the four cardinal directions and their dendrites are oriented towards the null direction of the cell. In the mouse retina, one amacrine cell with a symmetrical dendrite detects motion in all directions: the dendrite is sensitive to motion in the direction from proximal to distal (Euler et al., 2002).
Comparable to T4 dendrites, the starburst amacrine cells receive inhibition at their null direction onset side. However, starburst amacrine cells connect to each other with dendrites tuned to the opposite direction inhibiting the distal dendritic area (Lee and Zhou, 2006), while T4 cells receive non-directional inhibition. T4 cells of the same subtype connect to each other and likely enhance the response to motion in the preferred direction.
Starburst amacrine cells provide both excitation as well as a null direction inhibition to the direction-selective retinal ganglion dendrites. In contrast, T4/T5 cells are only cholinergic and excitatory, nevertheless they connect to bistratified, glutamatergic lobula plate intrinsic cells. These cells are postsynaptic to T4/T5 in one lobula plate layer and presynaptic to lobula plate tangential cells in the oppositely tuned layer (Mauss et al., 2015). Therefore, T4/T5 indirectly provide null direction inhibition to the lobula plate tangential cells, comparable to starburst amacrine cells.
3.2
Object Orientation
Information about local luminance changes could be used for different tasks: to find attractive objects and fly or walk towards them, to orient according to the position of a visual landmark or to detect the position of a potential predator and avoid it. While all these behaviors require motion vision, the detection of the position of objects is equally important.
Closed-loop fixation behavior of flies as well as their turning reaction to local motion can be explained by an asymmetric front-to-back than back-to-front motion response (Reichardt, 1973; Götz, 1975). However, such an asymmetric motion response fails to explain the strongly position dependent turning response of houseflies towards pure flicker stimuli (Pick, 1974). Several studies suggested that fly orientation towards local objects can be mediated without motion vision. Laser ablation of lobula plate tangential cells in houseflies reduced their optomotor response, but the flies could still orient towards single objects (Geiger and Nässel, 1981). Similar experiments were
performed with the motion blind Drosophila mutant omb and the double mutant rol sol. These
flies have difficulties following whole field motion, but react to landmark position (Bausenwein et al., 1986; Wolf and Heisenberg, 1986). This indicated, that a motion independent position system exists in the fly visual system, which is involved in object orientation behavior. Following this idea,
Reichardt and Poggio (1976) described the turning response ofDrosophilato a vertical bar moving
around a fixed fly as the sum of a symmetrical motion response and a position dependent response. We used a specific driver line to block the synaptic output of T4 and T5 cells by overexpression
of shibirets. These flies were not able to perform an optomotor response, which confirmed that
they are motion blind. Nevertheless, the T4/T5 blocked flies could fixate a vertical bar, albeit weaker than control flies. Presenting a rotating bar in an open-loop experiment demonstrated, that motion blind flies do respond to the position of the bar, but not to the direction of the bar motion. The motion blind flies reacted the same way to a flicker stimulus (a vertical bar appearing and disappearing) presented at different positions as control flies. To compare the flicker response with the response to local motion, we presented three stimuli with the same luminance change: a lateral bar moving locally in front-to-back direction, a lateral bar moving in back-to-front direction and a lateral appearing bar. While control flies reacted strongest to the front-to-back motion and weakest to the back-to-front motion, T4/T5 blocked flies did not distinguish between the three conditions. Therefore, flies are able to observe the position of luminance changes without functional T4/T5 cells, but not the direction of motion. We concluded, that the fixation behavior of flies is dependent on a motion-independent position system in combination with a symmetrical motion response. The position system induces strong reactions to local luminance changes at a lateral position.
Another study testing the dependence of flight fixation behavior on T4/T5 cells confirmed our results when using a low gain feedback in the closed-loop fixation, that means flight turns induce slow movements of the bar. However, the T4/T5 blocked flies were not able to fixate a bar under high feedback gain closed-loop conditions or with additional background motion (Fenk et
al., 2014). The bar fixation they found under low gain feedback conditions is very weak compared to the fixation we measured with walking flies. The lower fixation could arise from differences between walking and flight behavior. An alternative explanation would be an effect of the GAL4 driver line on the flight behavior. We found the T4/T5 driver line used by Fenk et al. (2014) to drive expression in motor neurons important for flight (data not published). Therefore, we used a more specific split-GAL line to block T4/T5 cells when measuring flight behavior (see the third manuscript included in this thesis). The impairments in flight could reduce the fixation ability of the flies and explain the weak flight fixation under all conditions.
In contrast to our results describing symmetric motion responses, Fenk et al. (2014) suggested the fixation behavior to rely on asymmetric motion responses of the horizontal system lobula plate tangential cells. A clarifying experiment could be to test fixation behavior while blocking the output of horizontal system lobula plate tangential cells. For example, expression of the light gated anion channel GtACR was shown to hyperpolarize these cells during illumination (Busch et al., 2018). However, depolarization as well as hyperpolarization of the horizontal system lobula plate tangential cells induced flies to walk slower (Busch et al., 2018), therefore blocking these cells directly might interfere with a fixation behavior.
A cellular implementation of a position system could be T4/T5-independent input to the lobula plate or a pathway from the medulla to the lobula. Lobula plate tangential cells do respond to local flicker without a T4/T5 input, but partly dependent on the medulla neurons Mi1 and Tm3 (Bahl et al., 2015). So far, no T4/T5 independent input to lobula plate tangential cells is described. Since many medulla cells project to the lobula and connect to lobula columnar cells, the cellular implementation of object recognition could be lobula columnar cells. For example, LC11 cells react specifically to small objects (Keleş and Frye, 2017; Wu et al., 2016). Such cells might receive information about local luminance changes directly from the medulla and additional motion information from the lobula plate. Therefore, lobula columnar cells might enable flies to react to local flicker without functional T4/T5 cells.