Abstract Although the functional role of the optic tectum has yet to be well characterized in teleosts, it is thought to be responsible for initiating and coordinating visuomotor behaviors associated with the looming response. In this study, behavioral and electrophysiological responses to the looming stimulus were measured by differential suction electrodes at the optic tectum and retinal ganglion cells of adult male zebrafish (Brachydanio rerio). The looming stimulus was simulated by a light-emitting diode and by increasing irradiance durations. Responses to off-light events were found at both the retina and optic tectum, suggesting that light-induced startle responses begin as early in the visual synaptic pathway as the retina. However, several differences in signal shape and amplitude between recordings at these two regions suggest that higher-order visual processing of on- and off-responses occurs at the optic tectum. Latency times were measured between the presentation of the stimulus and the onset of the response, but data are currently inconclusive. Habituation was observed through decreases in startle response as a function of flash repetition. Future studies are necessary to correlate electrode position with signal strength to resolve neural organization at the optic tectum. Öther parameters such as flash duration, interstimulus interval, and age of the zebrafish can also be varied to obtain a more complete understanding of the startle response. Introduction In natural habitats, predator-prey interactions often provide the selection pressure for the evolution of behavioral escape responses, which range from immobility, changes in color, active flight, and the seeking of refuge (Roberts 1992). In this study, we focused on one specific escape response, the startle reaction, which occurs following a sudden, unexpected stimulus. The adult zebrafish (Brachydanio rerio) was used as our model organism because its startle reaction has been well characterized: a distinct tail flip, more pronounced than that in normal swimming. which both accelerates and turns the fish (Kimmel 1974). The net result is rapid movement at an angle to the original axis of swimming (Fig. 1). In general, these motions enable the fish to evade oncoming predators quickly and efficiently. In many teleosts, including the zebrafish, an abrupt decrease in light intensity triggers a particular startle reaction, now identified as the looming response (Diamond 1970). Shadows created from above can indicate an impending attack, especially by avian predators. Therefore, these fish tend to dive into deeper waters for safety when another creature "looms" above it and blocks the ambient sunlight. The pathway that mediates light received by the eye and ultimately generates muscle fiber contraction has been studied extensively. The looming response is understood to be transmitted via Mauthner’s neuron, or the M-cell, which initiates large muscular contractions to produce a full startle reaction following certain visual cues (Bartelmez 1915). Nevertheless, by what method certain structures function, such as the exact processes occurring within the optic tectum, is not yet well understood. The retina contains two distinct types of ganglion cells: on-center and off-center cells. On- center cells fire action potentials when their center is exposed to light and do not fire when the region around its center, the surround, is subjected to light. Off-center cells react in just the opposite manner. The axonal projections of retinal ganglion cells aggregate to form the optic nerve. In non-mammalian vertebrates, these axons synapse onto the optic tectum (Fig. 2). In fish, the optic tectum is the largest center for visual processing and appears to play a significant role in predator/prey discrimination by initiating either escape or hunting, respectively. The mammalian analog of the optic tectum is known as the superior colliculus. Another group of neurons connects the superior colliculus to the cerebral cortex, the primary visual center in mammalian vertebrates that is absent in lower vertebrates. Interestingly, the number of retinal fibers in the visual afferent pathway to the optic tectum decreases considerably from lower vertebrates to mammals (Cajal 1995). From the optic tectum, signals are relayed to the pair of Mauthner neurons that surround the brainstem (Zottoli 1987). Each Mauthner cell sends a crossed projection to a group of motoneurons and simultaneously inhibits the M-cell located on the opposite side of the fish. These motoneurons are responsible for the contraction of muscle fibers along the entire length of the zebrafish flank, which ultimately creates a visible startle response. In this study, we recorded both behavioral and electrophysiological responses of zebrafish to the looming stimulus. We stimulated on-center and off-center cells in the retina to create action potentials that traveled along the optic nerve to the optic tectum, which were then measured using electrophysiological techniques. Our objective was to resolve differences in the processing of on- and off-responses in the optic nerve and at the optic tectum. By observing discrepancies in shape, relative amplitude, and duration, we hoped to elucidate whether higher-order processing of visual input occurs in the optic tectum of adult zebrafish. Materials & Methods Subjects Adult zebrafish were bred in the lab or purchased from a commercial supplier and kept in aerated tanks with flowing water at 28°C. Only males approximately 3-4 cm in body length were used. Animal care and handling was conducted in accordance with guidelines established by Stanford University Behavioral Observations Five zebrafish were placed in a tank measuring 24 cm 12 cm 16 cm with aerated water at 28°C and allowed to acclimate for at least 1 hour. The tank was surrounded on three sides by black poster board and illuminated by ambient room light, as well as by a flashlight positioned approximately 1 m above the water’s surface. Shadows were created by swiftly interrupting the flashlight beam with a circular piece of cardboard 10 cm in diameter. The shadow was presented every 10 seconds until a noticeable decrease in animal response occurred. Behavioral responses were recorded by a high-speed digital video recorder (Sony DSC-T1). Retinal and Tectal Preparations Prior to surgery, the zebrafish was isolated from the supply tank and immobilized temporarily by hypothermic shock in an ice water bath. Skeletal muscle paralysis was induced by an intramuscular injection of 0.1-0.3 mL of curare into the mid-lateral tail line. Recordings from the retina and the tectum were performed in different fish. For retinal experiments, the eye was isolated from the fish by gently pulling and severing the optic nerve and surrounding muscle tissue (Fig. 3). The eye was submerged in aerated 75% Hanks in a glass¬ holding dish during experimentation. For tectal experiments, the skull was removed dorsally with forceps to expose the optic tectum (Fig. 4). Äfter surgery, the fish was positioned upright in a glass-holding dish, and the gills were perfused with aerated 75% Hanks solution. The solution level was maintained below the gills. Adequate oxygen delivery was monitored throughout the experiment by observing the blood flow in the gills and in the vessels on the optic tectum. The glass-holding dish was placed in a light-isolated Faraday cage for both retinal and tectal experiments. Electrode Placement Differential glass suction electrodes were used for recording from the retina and the tectum. The electrodes consisted of an Ag-AgCl wire encased in a glass pipette filled with 75% Hanks solution. The tectum electrode was positioned on the surface of either the right or the left tectal hemisphere. The retinal electrode was placed on the stump of the detached optic nerve. Suction and a mineral oil coating were used to establish a tight seal. A reference electrode was placed approximately 2 cm from the fish. Once the electrodes were positioned, 30 minutes were given for darkness adaptation. Light Stimuli and Recordings On- and off-light stimuli were produced by a white light-emitting diode (LED) positioned approximately 3 cm to the left of the subject. The intensity of the LED was constant during all periods of irradiance, and ambient light was minimal during dark periods. The flash duration and interstimulus interval were controlled remotely by a computer (National Instruments Lab VIEW). Experiments consisted of 5 repetitions of an irradiance interval followed by a darkness interval (interstimulus interval). Light stimulus duration started at 100 ms and was incremented to 2 minutes. Irradiance was preceded by a 100-ms delay and was followed by an interstimulus interval totaling the irradiance and delay time in duration. Approximations for these parameters were obtained and modified from Bullock (1990). Electrical signals were sent to an AC differential amplifier with low-pass filtering of 1 Hz and high-pass filtering of 1 kHz. The amplified signal (10,000—) was sent to an oscilloscope, a DAT recording deck, and a computer. Measurements were recorded as single sweeps and as averaged responses of five consecutive repetitions. Light frequencies were simultaneously recorded by a DAT. Results In each figure, the lower schematic depicts when the flashes occurred: the LED is on whenever the voltage is at a value significantly higher than the baseline. The upper diagram, on the other hand, is the recording from the electrode placed either in the optic nerve or at the optic tectum. Optic Nerve Recording from the retina was done from the distal stump of the optic nerve. A baseline flash duration of 100 ms and interstimulus interval of 1900 ms was established (Fig. 5). From the retina, only a single dip in response to each of the five light flashes was observed. Each of these potentials lasted longer than the light flash itself and was relatively sharp in its characteristic shape. The end of each signal was marked by hyperpolarization before the resting potential was reached. As the flash duration was increased gradually, two distinct dips began to appear: an on- response when the light appeared and an off-response when the light disappeared (Fig. 6). In order to see the two separate troughs, however, it was necessary to increase the flash duration past a certain threshold value. This ensured that the on- and off-responses did not overlap, which occurred in the initial trial. Although both of these responses may have been present physiologically at a flash duration of 100 ms, they were indistinguishable according to the voltage recordings. In addition, the on- and off-responses were similar in magnitude over five repetitions. As the flash duration and interstimulus interval were increased further, it was noticed that even with an increasing flash duration, the relative durations of both on- and off-reactions remain constant, within an approximate range of 200 and 300 ms (Fig. 7). Right Tectum Due to the anatomical connections that link the retina to the brain in lower vertebrates, when the LED was positioned more proximal to the left eye, the contralateral tectum received all of the neuronal signals output by the left retina (Fig. 8). As a result, it was expected that signals detected at the right tectum to correspond better to the stimulus being presented to the zebrafish when compared to those measured at the left tectum. When the same initial trial from the retinal experiment was repeated, two peaks became visible in the right tectum, one that represented the on-response and another that represented the off-response (Fig. 9). Moreover, the on- and off-responses appeared to be comparable in magnitude given these baseline parameters. Even though the recordings from the right tectum show slightly higher levels of electrical noise compared to those made in the optic nerve, the overall shapes present in a series of flashes remain consistent over repeated trails. As both the flash duration and interstimulus interval were lengthened, complex relationships between flash duration, on-response, and off-response magnitude were observed. For example, when the flash duration was increased from 100 ms to 5 seconds, the ratio between corresponding off- and on-responses was almost 50 percent greater in magnitude (Fig. 10). As flash duration was increased further, the off-response continued to grow relative to the on- response, but the increases were not as apparent, and the off-response reached its maximum amplitude at a flash duration of 10 seconds. The on-response also grew with longer flashes, but only very slightly. Once comparisons were made with regard to flash duration, patterns within a single trial of five repetitions were examined. The observation that the off-response decays markedly as a function of repetition number was made (Fig. 11). The greater the number of times our stimulus is presented to the zebrafish, the smaller the magnitude of the resulting off-response. As it turns out, within a set of five repetitions, there was an approximate 25 percent decrease in the magnitude of the off-response as repetition number increases. The on-response, on the other hand, was much more variable and showed both decreases and increases with repetition. Both the on- and off-responses were characterized by a short delay period during which only background electrical activity was measured, and a sharp increase in activity that immediately followed. The observed delay, or the time between the onset of a stimulus and the beginning of a signal, was referred to as the latency period. For both the on- and off-response, the latency period ranged between 20 and 100 ms. The signals themselves, however, had durations around 20 ms for the on-response and 130 ms for the off-response. Thus, although the on- and off-responses were generated at similar times after a light stimulus, they persisted to different extents and at varying magnitudes. In addition, changes in latency times were analyzed to determine a relationship as a function of flash duration. At a flash duration of 100 ms, the average latency period for the ÖFF response was 100 ms in the right tectum. When flash duration was extended to 5 s, however, the corresponding latency value was only around 20 ms. For the on-response, latency times remained relatively constant between 15 and 20 ms. Left Tectum Although the left optic tectum represented an ipsilateral configuration to the retina receiving most of the light input, recordings made from the left tectum displayed strong activity that corresponded to stimulus times. From our baseline trial with a flash duration of 100 ms, two peaks with highly variable amplitudes over five repetitions were the result (Fig. 12), one for the on-response and another for the off-response. The first peak was characterized by a high- amplitude sharp wave, whereas the second peak was slower and less pronounced. When flash duration was increased enough to distinguish between on- and off-responses, the off-response was consistently smaller in magnitude than its associated on-response (Fig. 13). As flash duration was increased further, off-responses increased disproportionately relative to their corresponding on-responses until a flash duration of 10s was reached. However, the ratio of off- responses to the first on-response of five successive repetitions showed no consistent trend over repeated trials. Although differences in the relative amplitudes of on- and off-responses increased with longer flash durations, they were not as significant as those observed in the right optic tectum. The components of both the on- and off-responses in the left tectum included a delay period that lasted between 15 and 30 ms. However, the on-response contained two distinct peaks, whereas the off-response involved only one such wave. Within the on-response, the first wave was very sharp in its shape and persisted from 40 to 70 ms. This wave was followed by a slower, shallower wave with a duration of around 130 ms. On the other hand, the off-response consisted of a single, long wave of activity that ranged between 200 and 300 ms in length. Latency periods were again compared over varying flash durations. At a flash duration of 100 ms and an interstimulus interval of 1900 ms, the average latency time for the ÖFF response was approximately 15 ms. When flash duration was increased to 5 s, the corresponding latency value was almost 200 ms. Similar to what occurred in the right tectum, the ON response latency times ranged very little, between 15 and 20 ms. Qualitative Comparisons In comparing recordings made at the optic nerve and two optic tecta, on- and off-responses were observed in all three locations. However, their relative magnitudes were found to be more similar in the retina. Signals observed in the tecta consisted of double peaks, whereas those measured in the retina consisted only of single peaks. The most distinct difference that was noticed was the reversal of the relative magnitudes of the on- and off- responses between the right and left tecta. In the right tectum, the off-response was considerably and consistently larger in magnitude than the on-response that preceded it. In contrast, in the left tectum, the on¬ response was larger than the off-response, although not quite as consistently in comparison with the right tectum. Moreover, the shapes of on- and off- responses were more defined in the right tectum than in the left tectum. Äfter a response was evoked, an extended series of several oscillations occurred at the left optic tectum before baseline potentials were established. These fluctuations were entirely absent in the right tectum. In addition, latency times for the OFF response decrease as a function of flash duration in the right tectum, but increase in the left tectum. The presence of such polar differences between the two tecta may indicate one method of determining the direction of a light stimulus. Discussion In this study, we investigated the behavioral and electrophysiological responses of adult male zebrafish to a looming stimulus. Electrical activity at both the optic nerve and tectum in response to sudden introduction and elimination of light showed that the on- and off-responses originate from the retina. However, several significant differences in the waveforms elicited at the post¬ synaptic layers of the tectum were observed, which might suggest the occurrence of higher-order processing of on- and off- responses in the optic tectum. Evidence of Higher-Order Tectal Processing Visible signals recorded from the optic nerve in response to on- and off-events at all tested durations of irradiation and interstimulus intervals show that neuron potentials elicited by sudden illumination or darkening around the field of vision begin as early in the visual synaptic pathway as the retina. This follows current consensus on the functional organization of retinal ganglion, on-, and off- bipolar cells in mammals (Masland 2001) and teleosts (Daw 1968, Yang et al. 1988). However, marked differences in waveforms and relative amplitudes were observed between the signals recorded from the retina and optic tectum, which seems to indicate higher tectal processing of the on- and off-stimulated signals from the ganglion cells. Signals may be processed" at the tectum for any number of quantitative and qualitative characteristics of the visual stimulus, such as light intensity, diffusivity, duration, and orientation, as well as integration of these various aspects. Additionally, differences between signals recorded from the 11 right and left hemispheres of the optic tectum may indicate independent processing at individual tectal lobes. The waves recorded from the ganglion nerve in response to the on- and off-light stimulus exhibited approximately equal amplitudes at all tested durations of irradiance. Differences in amplitude between the on- and off-stimulus were immediately apparent at the optic tectum. The on-response was visibly weaker than the off-response at the right tectum, whereas the reverse was observed at the left tectum. It is too early to postulate the functional significance of the amplitude differences observed in the waves recorded from the retina and optic tectum. Nevertheless, it is possible that such differences are the result of visual orientation and directionality processing, since the ganglion cells from the left eye synapse primarily onto the right optic tectum (the LED was positioned on the left side of the animal in this study). Any amplitude differences between the right and left tectum are thus likely to be influenced strongly by the position of the light source relative to the visual field of the zebrafish. While at least two depolarization events occurred at the optic tectum at all tested durations of irradiance (one in response to the on-stimulus and one in response to the off-stimulus), only one depolarization event was recorded from the optic nerve at the flash duration at 100 ms. However, durations of light longer than 100 ms produced distinct on- and off-responses at the optic nerve. These observations suggest the presence of at least three different functional cell populations in the ganglion, including on-center, off-center, and on-off response cells, in which on-off response cell potentials are evoked by a brief flash of light (Monasterio et al. 1975). Accordingly, it appears that the visual synaptic input to the optic tectum during a looming response derives primarily from on- and off-response cells in the ganglion. Comparisons of relative amplitudes of on- and off-responses as flash durations were increased indicated that the size of the off-response relative to the on-response at both the left and right optic tectum increased. In contrast, little to no increase was observed at the retina. Thus, the off-evoked potential at the optic tectum appeared more pronounced as the fish was exposed to longer periods of light. This suggests that the tectal processing of information regarding dark-induced visual input from the retina is influenced by the time of light exposure prior to the off-stimulus. Additional measurements of on- and off-responses to a short flash duration (100 ms) were recorded immediately after measurements of responses to a series of increasingly longer flash durations to confirm that the increases in the ratio of the off- and on¬ response amplitudes were not the result of temporal habituation during repeated testing. Off-Response Attenuation The magnitude of consecutive off-responses decreased relative to the first off-response within each experiment (5 repetitions per experiment), indicating the occurrence of habituation to the looming stimulus. Attenuation of responses to shadows was also observed in the behavioral experiments. While the occurrence of habituation in these studies suggests that zebrafish may become less responsive to repeated appearances of shadows of predators from overhead or may judge that such shadows do not represent imminent danger, this has yet to be studied in natural habitats. Limitations and Directions for Future Research This study represents a preliminary investigation of the electrophysiological responses of zebrafish to light-induced startle responses and is limited in several respects. Electrode placement was determined largely by the position yielding the most visible signal and did not 13 account for possible input biases resulting from the synaptic organization of the retina and optic tectum. Comparisons of signals from different locations on the tectum may help provide a more detailed understanding of the functional organization of the retina and optic tectum. Furthermore, a number of parameters have yet to be studied, including responses to flash durations of increased resolution, decreased interstimulus intervals, and from fish in different stages of development. Also, measurements of latency may reveal more retinal and tectal differences. While preliminary recordings seem to indicate a decrease in latency of the off- response at the right optic tectum and an increase at the left optic tectum with increasing durations of light, they are currently inconclusive and require a more accurate and standardized measurement method. Additionally, the looming stimulus was produced only as square-wave pulses of uniform light intensity in this study. It remains for future studies to employ variations in the simulation of the looming stimulus. Acknowledgements We are deeply indebted to Professor Stuart Thompson for his expertise and guidance, without which, this project would not have been possible. We also express our sincerest gratitude to Christian O’Reilly and Chris Patton for their encouragement and assistance. Additionally, we would like to thank all the members of the Thompson laboratory for their support and collaboration. 14 Literature Cited Bartelmez, G. W. 1915. Mauthner’s Cell and the Nucleus Motorius Tegmeni. J. Comp. Neurol. 25: 87-129. Bullock, T. H., M. Hofmann, F. Nahm, J. New, and J. Prechtl. 1990. Event-Related Potentials ir the Retina and Optic Tectum of Fish. J. Neurophysiol. 64: 903-914. Cajal, S. R. 1995. The Optic Lobe of Lower Vertebrates. pp. 161-185 in N. Swanson and L. W. Swanson, ed. Histology of the Nervous System of Man and Vertebrates, Vol. II., Oxford University Press, New York, NY. Cajal, S. R. 1995. The Optic Nerve, Chiasm, and Tract. pp. 303-313 in N. Swanson and L. W. Swanson, ed. Histology of the Nervous System of Man and Vertebrates, Vol. II., Oxford University Press, New York, NY. Daw, N.W. 1968. Colour-coded ganglion cells in the goldfish retina. J. Physiol. 197: 567-592. Diamond, J. 1970. The Mauthner Cell. pp. 265-346 in W. S. Hoar and D. J. Randall, ed. Fish Physiology, Vol V., Academic Press, New York, NY. Eaton, R. C., R. A. Bombardieri, and D. L. Meyer. 1977. The Mauthner-Initiated Startle Response in Teleost Fish. J. Exp. Biol. 66: 65-81. Kimmel, C. B., J. Patterson, and R. Kimmel. 1974. The Development and Behavioral Characteristics if the Startle Response in the Zebra Fish. Dev. Psychobiol. 7: 47-60. Masland, R. H. 2001. The fundamental plan of the retina. Nature Neurosci. 4: 877-886. Monasterio, F. M. and P. Gouras. 1975. Functional properties of ganglion cells of the rhesus monkey retina. J. Physiol. 251: 167-195. Roberts, B. L. 1992. Neural mechanisms underlying escape behaviour in fishes. Rev. Fish Biol. Fish. 2: 243-266. Yang, X. L., K. Tornqvist, and J. E. Dowling. 1988. Modulation of cone horizontal cell activity in the teleost fish retina. J. Neurosci. 8: 2259-2268. Zottoli, S. J., A. R. Hordes, and D. S. Faber. 1987. Localization of Optic Tectal Input to the Ventral Dendrite of the Goldfish Mauthner Cell. Brain Res. 401: 113-121. 15 t a CE kaaaa- Fe - 3 e tl S —- 21. — pre F. 58 e Si N s Sen e L . i 1tj orågete 1 ...- — 18 fne 1000 ms k — — Mimp M . 23 C.. R .. tpennntuenudenndpet 1000 ms 10 mfeenfntentlenfrannentngnnandenes — — — — L — L L 105 12 mnsns ltendmaldndn 1000 ms 13 dannnmennenenentsunemhe — — L — Fig. 1 Dorsal view of typical startle response in zebrafish. Silhouettes are drawn at 5 ms intervals beginning approximately 5 ms before onset of vibrational stimulus (reproduced from Eaton et al. 1977) Fig. 2 Frontal section through a teleost (Barbus fluviatilis) optic tectum (after P. Ramón) stained using the Golgi method. A, deep regions of layer 5; B, middle region and central plexus of layer 5; C, superficial region of layer 5 consisting primarily of fusiform cell dendritic branches; a, layer 5 ganglion cell; b, pyramidal cells with a central axon; c, shepherd’s crook cell with some of its dendritic branches extending into the central plexus; d, layer 5 transverse cell; e, small stellate cell (reproduced from Cajal 1995). Dorsal view of zebrafish eye and surrounding tissue after extraction from socket. The Fig. 3. thin, light tissue is the optic nerve, which contains the axonal projections of ganglion cells. Encircling the optic nerve are fat globules that supply nutrients to the region. Top-down view of zebrafish tectal region after surgical removal of protective head plate. Fig. 4. Clearly visible are the right and left lobes of the optic tectum, along with the cerebellum, which lies posterior to and between the tecta. Fig. 5 Diagram of an optic chiasm and the compensatory decussation of motor and somatosensory pathways in lower vertebrates. C, second-order visual centers; G, spinal ganglia and sensory roots; M, motor pathway; O, crossed optic nerves; R, spinal motor roots; S, crossed second-order somatosensory pathway (reproduced from Cajal 1995). Voltage data recorded from the retina at five repetitions given baseline parameters (flash Fig. 6. duration of 100 ms and an interstimulus interval of 1900 ms). Single dip responses occur following each flash. Each response lasts longer than the corresponding flash itself. Voltage data recorded from the retina given a flash duration and an interstimulus Fig. 7 interval of 10 s. Both an on- and an off-response are visible and are similar in peak amplitude. Voltage data recorded from the retina given a flash duration and an interstimulus Fig. 8 interval of 20 s. On- and off-responses remain constant in the duration, between 200 and 300 ms, as flash duration increases. Voltage data recorded from the right optic tectum at five repetitions given baseline Fig. 9 parameters. Both on- and off-responses result from repeated light stimuli. High levels of electrical noise are present in this run. Fig. 10 Voltage data recorded from the right optic tectum at five repetitions given a flash duration and an interstimulus interval of 5 s. The ratio of off-response magnitude to on- response magnitude increases as a function of repetition number. Fig. 11 Voltage data recorded from the right optic tectum at five repetitions given a flash duration and an interstimulus interval of 10 s. Off-response magnitude displays a noticeable decrease of approximately 25 percent over the complete series of light flashes. Fig. 12 Voltage data recorded from the left optic tectum at five repetitions given baseline parameters. Both on- and off-responses result from repeated light stimuli, and no clear trend governs their magnitudes. High levels of electrical noise are present in this run. Fig. 13. Voltage data recorded from the left optic tectum at five repetitions given a flash duration and an interstimulus interval of 5 s. The ratio of off-response magnitude to on-response magnitude decreases as a function of repetition number. 29