Drosophila larval light avoidance is negatively regulated by temperature through two pairs of central brain neurons

Animal’s innate avoidance behavior is crucial for its survival. It subjects to modulation by environmental conditions in addition to the commanding sensorimotor transformation pathway. Although much has been known about the commanding neural basis, relatively less is known about how innate avoidance behavior is shaped by external conditions. Here in this paper, we report that Drosophila larvae showed stronger light avoidance at lower temperatures than at higher temperatures. Such negative regulation of light avoidance by temperature was abolished by blocking two pairs of central brain neurons, ACLPR60F09 neurons, that were responsive to both light and temperature change, including cooling and warming. ACLPR60F09 neurons could be excited by pdf-LaN neurons in the visual pathway. On the downstream side, they could inhibit the CLPNR82B09 neurons that command light induced reorientation behavior. Compared with at warm temperature, ACLPR60F09 neurons’ response to light was decreased at cool temperature so that the inhibition on CLPNR82B09 neurons was relieved and the light induced avoidance was enhanced. Our result proposed a neural mechanism underlying cross-modal modulation of animal innate avoidance behavior.


Introduction
The neural mechanism underlying animal preference behavior has been intensively studied as it reflects the most common form of sensorimotor transformation (Glimcher, 2003;Poeppl et al., 2016;Song and Lee, 2018;Salamone et al., 2018;Lowenstein and Velazquez-Ulloa, 2018;Marachlian et al., 2018). For animal innate preference behaviors, in addition to the commanding internal neural mechanism, they can also be modified by external environmental conditions. In Drosophila for example, preference for food containing different concentration of sugar was biased by hardness of food (Jeong et al., 2016). Adult females choose site of egg laying according to the interplay between sweet taste and mechanical feeling of hardness (Wu et al., 2019). Flies' humid preference was reported to be affected by temperature sensation while rapid heat avoidance was negatively related to environmental humidity on the other hand (Frank et al., 2017). Also, temperature preference in adult fly was positively influenced by environmental light conditions (Head et al.,2015). Such environmental modulation of innate preference behavior may facilitate the animals' response to those environmental conditions under which certain external stimuli usually co-exist.
Comparatively, environmental modulation of Drosophila light preference has received less investigation. Drosophila avoids light and prefers darkness at larval stage (Keene et al., 2011;Keene and Sprecher 2012). In Drosophila larva, the anteriorly located larval photoreceptors, Bolwig's organs, which sense low-or intermediate-intensity light and are required for light avoidance response (Mazzoni et al., 2005;Keene et al. 2011;Keene and Sprecher, 2012;Humberg and Sprecher, 2017;Humberg et al., 2018). Bolwig's organs project their axons into the larval optic neuropil (LON) and form synapses with downstream neurons such as pdf-expressing lateral neurons (pdf-LaN) (Larderet et al., 2017), the fifth lateral neuron (5 th -LN) (Keene et al. 2011) and the PVL09 neurons (Humberg et al., 2018), with the latter two reported to be required for larval light avoidance.
Downstream to these neurons includes a neural pathway consisting of PTTH neurons (Gong et al., 2010), EH neurons and Tdc2 motor neurons that commands light induced deceleration response , and another pathway including LRIN R13B07 neurons and CLPN R82B09 neurons that mediates light induced reorientation .
As light is closely related to temperature such as in daily or seasonal cycle, it is assumable that Drosophila larval light avoidance is under the impact of temperature. Indeed, we could observe stronger light avoidance at lower temperatures than at higher temperatures. In Drosophila larva cold-sensing neurons are anteriorly located with cell bodies in the dorsal organ ganglion (DOG) and the terminal organ ganglion (TOG) (Liu et al. 2003;Klein et al., 2015;Li and Gong, 2015;Ni et al., 2016;Li and Gong, 2017). They send sensory dendrites to terminal organs and dorsal organs in the very anterior tip and axonal projections to central brain. DOG neurons sense cold stimuli through mediation of ionotropic receptors IR21a and IR25a (Ni et al., 2016). On the other hand, neurons expressing TRPA1 and painless were found to sense innocuous warmth and nociceptive heat in Drosophila larva (Barbagallo and Garrity, 2015;Rosenzweig et al., 2005;Tracey et al., 2003).
Interestingly, the light receptor protein Rh5 and Rh6 were also reported to regulate larval temperature preference behavior (Sokabe et al., 2016). There is relatively little study on how cold and warmth signal is processed in downstream neurons (Li and Gong, 2015).
In this study, we report that Drosophila larval light avoidance was negatively regulated by temperature. Such thermal regulation of light avoidance was abolished by blocking two pairs of ACLP R60F09 neurons that could sense both light stimulation and temperature changes including cooling and warming. At lower temperature, the light induced response in ACLP R60F09 neurons was repressed so that their inhibition on downstream CLPN R82B09 neurons that are known to mediate light induced reorientation response was relieved. Our results reveal a neural mechanism underlying modulation of light evoked animal avoidance behavior by environmental temperature.

Drosophila larval light avoidance was negatively regulated by temperature
We investigated the effect of temperature on larval light avoidance by testing 3 rd instar larval light preference using a simple light/dark choice assay (Gong et al., 2010;Zhao et al., 2019) with white light intensity of 550 lux (about 23.3 μW/mm 2 ) for 10 minutes at temperatures ranging from 15 ℃ to 27 ℃ using a wildtype strain of WT-CS. In this range, larval light avoidance was promoted as temperature decreased (Fig. 1a). Furthermore, we asked if the behavioral output was stable over longer time of testing. We tested 3 rd instar larval light avoidance of another commonly used control strain w 1118 at 18 o C and 27 o C for 5-, 10-, 15-and 20 minutes respectively. In all cases, larvae always showed a higher preference for darkness at a cool temperature of 18 ℃ than at a warm temperature of 27 ℃ (Fig. 1b). These results together showed that larval light avoidance was negatively regulated by temperature.

R60F09-GAL4 labeled neurons were putatively required for regulation of light avoidance by temperature
In order to explore the neural basis of lower temperature enhancement of larval light avoidance, we screened for candidate neurons by using Gal4 strains to drive the expression of tetanus toxin (TNTG) to block the activity of neurons and comparing larval light avoidance performance to 170 lux white light at 18 ℃ and 27 ℃. The difference between 18 ℃ and 27 ℃ seen in control lines disappeared when TNTG was expressed by R60F09-GAL4 (Fig. 2a). In larval central nervous system, R60F09-GAL4 marks three pairs of neurons in the brain hemispheres and three pairs of neurons in the suboesophageal zone (SEZ) (Fig. 2b). We used flp-out technology  to reveal single neuronal morphology in the brain and found that the three pairs of neurons in the brain were divided into two categories: in the first category, each of the two pairs of neurons has dendrites and cell body ipsilaterally located in anterior brain hemisphere and an axon projecting contralaterally to the other hemisphere (Fig. 2c, e); in the second category, each of the third pair of neurons has cell body and dendritic arborizations as well as axonal arborizations all in the same brain hemisphere (Fig. 2d). For convenience, we named neurons in the first category ACLP R60F09 neurons (anterior contralateral projecting neurons). Thus, R60F09-GAL4 labeled neurons were potentially required for the regulation of larval light avoidance by temperature.

ACLP R60F09 neurons were responsive to both light and temperature changes
We proposed that thermal sensation might affect larval light avoidance by modulating the neurons in the underlying pathway. If so, the neurons crucial for the cross-modality sensory integration should be responsive to both light and temperature change. We then tested this hypothesis using calcium imaging. Upon light stimulation, the two ACLP R60F09 neurons in the same brain hemisphere showed obvious increase in calcium signal as indicated by GCAMP, with average peak response of about 10% increase in fluorescent intensity, while the third neuron in brain hemisphere as well as neurons in SEZ did not show obvious response ( Fig. 3a and 3b, Supplemental Figure 1). Treating the neurons with 20 μM tetrodotoxin (TTX) (Pirez et al., 2013;Luo et al., 2017), a voltage-gated Na + channels antagonist, could completely abolish the response, suggesting that the ACLP R60F09 neurons do not sense light by themselves but receive light signal from other neurons (Fig. 3a). Therefore, we focused only on the two pairs of ACLP R60F09 neurons in following experiments. We next tested the response of ACLP R60F09 neurons to temperature change. As shown in Fig 2a, blocking R60F09-GAL4 labeled neurons appeared to promote larval light avoidance at 27 ℃ but not at 18 ℃, suggesting that it could be the temperature rise that inhibited light avoidance. We first tested ACLP R60F09 neurons' response to temperature rise. As shown in Fig 3c and 3d, temperature rise monotonically from 18 ℃ to 27 ℃ within 30 seconds using a temperature controller led to a fast decrease in calcium signal by about 30% in peak response, suggesting that ACLP R60F09 neurons were repressed by temperature rise. Application of 20 μM tetrodotoxin (TTX) could largely remove the inhibitory response, suggesting that the thermal response originated from other neurons but not the ACLP R60F09 neurons themselves. We then wanted to know if temperature drop could produce an opposite response. Indeed, upon application of ice water, ACLP R60F09 neurons demonstrated obvious increase in calcium signal indicated by GCAMP (Supplemental Figure 2). As temperature first dropped and then rose rapidly after the application of ice water, it was not clear if the calcium response was caused by temperature drop or rise. We then used the temperature controller to control temperature dropping monotonically from 27 ℃ to 18 ℃ without recovery. Decreasing the temperature from 27 ℃ to 18 ℃ within 30 seconds caused a noticeable response in larval ACLP R60F09 neurons, with an average peak response of about 30% increase in fluorescent intensity ( Fig. 3e and 3f). ACLP R60F09 neurons' response to temperature drop was completely abolished in presence of TTX, indicating that they receive cold signal from other neurons (Fig. 3e). Thus, ACLP R60F09 neurons could respond to both acute temperature rise and drop. In addition to these acute temperature changes, long term exposure to constant cool temperatures such as incubation at 18 ℃ for 18 hours could produce strong signals as visualized by Ca-LexA technique, while 18 hours of incubation at a warm temperature of 27 ℃ did not induce detectable signal (Supplemental Figure 3). These data together showed that ACLP R60F09 neurons could respond to both light and thermal stimulation.
Taken together, integration of thermal signal and light signal in ACLP R60F09 neurons could be crucial for the negative regulation of larval light avoidance by temperature.

ACLP R60F09 neurons could be downstream to pdf-LaNs in visual pathway
As the ACLP R60F09 neurons are morphologically close to pdf-positive lateral neurons (pdf-LaNs) which are known secondary visual pathway components (Supplemental Figure 4a), we wondered if ACLP R60F09 neurons received light signal from pdf-LaN neurons. We tested possible interactions between ACLP R60F09 neurons and pdf-LaNs using the GFP reconstitution across synaptic partners (GRASP) technique. Robust GRASP signal was observed between putative dendrites of ACLP R60F09 neurons and axonal termini of pdf-LaNs ( Fig. 4a-d), suggesting that the ACLP R60F09 neurons might directly receive inputs from pdf-LaNs. To verify the existence of a functional connection, we combined optogenetics and calcium imaging to test whether directly exciting pdf-LaNs could activates ACLP R60F09 neurons. We expressed red light sensitive Chrimson in pdf-LaNs and GCAMP in ACLP R60F09 neurons in larvae fed on food supplied with trans-retinal. Upon red light stimulation of pdf-LaNs, ACLP R60F09 neurons showed obvious calcium response, with an average peak response of more than 20% increase in fluorescent intensity ( Fig. 4e and f). As pdf-LaNs can receive the light signal from larval photoreceptors (Larderet et al., 2017), it is thus possible that ACLP R60F09 neurons could receive light signal from pdf-LaNs.

ACLP R60F09 neurons could inhibit CLPN neurons.
As ACLP R60F09 neurons appeared to be morphologically close to CLPN R82B09 neurons that were found to mediate the light induced head cast response in larval light avoidance (Supplemental Figure 4a and 4b), we wondered if ACLP R60F09 neurons could directly target on CLPN R82B09 neurons to affect light avoidance. We used GRASP technique to test possible interaction between ACLP R60F09 neurons and CLPN R82B09 neurons which could be labeled by promoter GMR82B10.
ACLP R60F09 neurons and CLPN R82B09 neurons showed robust GRASP signal ( Fig. 5a and 5b) while no signal was seen in the controls ( Fig. 5c and 5d), suggesting that the ACLP R60F09 neurons might directly interact with CLPN R82B09 neurons. To further explore whether CLPN R82B09 neurons receive signal input from ACLP R60F09 neurons, we again combined optogenetics and calcium imaging to see if directly exciting ACLP R60F09 neurons could induce response in CLPN R82B09 neurons. When the ACLP R60F09 neurons expressing LexAop-Chrimson were excited by red-light, CLPN R82B09 neurons showed a maximal decrease of about 30% in calcium signal ( Fig. 5e and 5f).
This meant that CLPN R82B09 neurons received inhibitory signal input from ACLP R60F09 neurons.
Indeed, when we co-stained R60F09-Gal4 labeled neurons with the antibody against GABA, an inhibitory neurotransmitter, colocalization was seen in cell bodies of all the three R60F09-Gal4 labeled neurons in brain hemispheres ( Fig. 5g to 5r). This meant that ACLP R60F09 neurons were GABAergic. As CLPN R82B09 neurons had been identified to express a GABAA receptor RDL , it was thus highly possible that ACLP R60F09 neurons inhibited CLPN R82B09 neurons through GABA/RDL interaction. Therefore, ACLP R60F09 neurons seemed to play inhibitory roles in larval light avoidance behavior.

ACLP R60F09 s' response to light was reduced at lower temperature
As ACLP R60F09 neurons were activated by temperature drop and inhibited by temperature rise, how could the inhibitory ACLP R60F09 neurons mediate the negative regulation of light avoidance by temperature, that is, enhanced light avoidance at lower temperature and repressed light avoidance at higher temperature? To explain this apparent paradox, we need to find out how light and thermal information were integrated in ACLP R60F09 neurons. We then tested the response of ACLP R60F09 neurons to light at 18 o C and 27 o C. Calcium responses of larvae expressing GCAMP in ACLP R60F09 neurons to 1-second light pulses at interval of 20 seconds at 18 °C and subsequently 27 °C were recorded. To reduce the effect of spontaneous activity on neuronal response, we used the average amplitudes of peak responses to the first five light pulses to measure the response amplitude. ACLP R60F09 neurons' response to light stimulation decreased significantly at 18 °C compared to 27 °C ( Fig. 6a and 6b), by a relative amplitude of more than 30% on average (Fig. 6c). As light response in ACLP R60F09 neurons was weaker at lower temperature, the light induced inhibition of ACLP R60F09 neurons on CLPN R82B09 neurons was thus relieved to allow stronger aversive reorientation response. These results could well explain the paradox and were in consistent with our previous observation that blocking ACLP R60F09 neurons abolished the thermal regulation of larval light avoidance.

Discussion
In this study, we found that Drosophila larval light avoidance was negatively regulated by temperature. When temperature dropped, the response of ACLP R60F09 neurons to light was decreased.
As ACLP R60F09 neurons could inhibit the CLPN R82B09 neurons that command the light avoidance behavior, the reduction in ACLP R60F09 neurons' light response at lower temperatures facilitated the light avoidance behavior by relieving the inhibition on CLPN R82B09 neurons.
The enhanced light avoidance at lower temperature should be good for survival of Drosophila larva. It is assumable that when temperature drops before arrival of winter, Drosophila larvae are more likely to avoid light and hide in shelter-like places. This will help adult fruit flies to overwinter in shelters such as ground holes or crevices (Izquierdo, 1991) which can protect them from being found by predators. The reduced general activity and metabolism at low temperatures further add to the importance of staying in shelter for overwintering.
In our results, despite that ACLP R60F09 neurons' response to light was repressed at 18 °C, it was still quite significant. This meant that ACLP R60F09 neurons might still exert inhibition on CLPN R82B09 neurons. However, blocking ACLP R60F09 neurons did not change larval performance in light avoidance assay at 18 °C (Fig. 2). One possible explanation for such paradox could be that ACLP R60F09 neurons' activity upon light stimulation at 18 °C was not high enough to substantially suppress the function of CLPN R82B09 neurons.
For explanation of ACLP R60F09 neurons' function in the negative regulation of larval light avoidance by temperature, there are two remaining questions to be answered. The first question is from where do the ACLP R60F09 neurons received the cood and warmth signal, as ACLP R60F09 neurons do not sense temperature change by themselves. Cold sensing DOG and TOG neurons have been well characterized, but they all project directly into SOG area that is distant from ACLP R60F09 neurons.
There must be some interneurons that transfer the cold signal onto ACLP R60F09 neurons. Sensory neurons for warmth in larva is so far not known (Barbagallo and Garrity, 2015). Another possibility is that ACLP R60F09 neurons receive thermal signals from other uncharacterized thermal sensing neurons. Neurons that express temperature-sensitive molecules such as various channels of the TRP family (Flockerzi 2007;Venkatachalam and Montell, 2007;Rosenzweig et al., 2008;Zhong et al.,2012;Fowler and Montell, 2013;Turner et al., 2016) The second and probably more puzzling question is why ACLP R60F09 neurons' response to light was reduced as temperature dropped. Since the light stimulated ACLP R60F09 neurons were excited by temperature drop and inhibited by temperature rise, it is straightforward to assume that light and low temperature facilitate each other's excitatory effect on ACLP R60F09 neurons. Unexpectedly, the light response was repressed at lower temperature. This is probably because that lower temperature facilitates certain cellular event in ACLP R60F09 neurons that antagonizes the neurons' response to light, or that higher temperature inhibits cellular event that facilitates the neurons' response to light.
Another explanation is that the integration of light and thermal signal occurs in some upstream neurons. In this case, the reduced light response at lower temperature in ACLP R60F09 neurons is a result of the upstream event.
In summary, our work disclosed a new behavioral paradigm of integration of temperature and light sensing and characterized the related neural basis. Future unraveling of the underlying mechanism will add to our understandings of cross-modal sensory integration in animal central brain.

Fly strains.
Most flies were reared at 25 ℃ on standard culture medium under 12h:12h light/dark cycles. Flies Larval light avoidance assay.
The procedure of light spot assay was largely same as previously described . In short, half of the petri dish containing a 1.5% agar plate was covered to create a dark environment, white light above the petri dish illuminate the uncovered half. Light intensity of 550 lux corresponded to 23.3 μW/mm 2 , 170 lux corresponded to 9.1μW/mm 2 at maximal readings (S401C, Thorlabs, Inc). Twenty 3 rd instar larvae were placed on the agar plate to choose between light and dark for different time (5min or 10min or 15min or 20min) at different temperatures (15℃，18℃， 20℃，23℃，25℃ and 27℃). Light preference index (PI) was calculated as: PI = (number of larvae in the dark half-number of larvae in the light half)/(number of larvae in the dark half + number of larvae in the light half).

Calcium imaging.
Calcium imaging was done using 2 nd or 3 rd instar larvae as in previous report . If 3 rd instar larvae were used, they were dissected in AHL (Adult Hemolymph-Like) solution to remove the posterior part and keep the central brain exposed, and then transferred into a chamber formed by reinforcing rings on a cover glass and covered with another cover glass. If 2 nd instar larvae were used, the single larva was directly covered with a cover glass without reinforcing ring.
The cell bodies of neurons were imaged. Ca 2+ imaging was performed with Olympus FV-1000 twophoton microscope with 40X water objective lens at room temperature of 23 ℃. Infrared laser at 910 nm was used for illuminating the GCaMP6m. For calcium imaging response of ACLP R60F09 neurons to controlled cooling (from 27 ℃ to 18 ℃) and warming (from 18 ℃ to 27 ℃) or to light, the 2 nd instar larva or dissected 3 rd instar larva between cover glasses was placed on the aluminum plate surface of a custom-made temperature controller. Precise cooling and warming were controlled by the temperature controller. The light stimulus was 460nm blue light at intensity of about 28.076 μW/mm 2 . For calcium imaging response of ACLP R60F09 neurons to ice water, 200μl ice water was added on the glass slide.
For calcium imaging response of pharmacologically isolated ACLP R60F09 neurons to light and to thermal stimuli, larvae were dissected and soaked in AHL (adult hemolymph like) saline containing 20μM TTX for 4 minutes before recording the ACLP R60F09 neurons' calcium response. The soaking solution was AHL for control samples.
For Ca 2+ imaging ACLP R60F09 neurons' responses to optogenetic activation of pdf-LaNs and CLPN R82B09 neurons' responses to optogenetic activation of ACLP R60F09 neurons, eggs of proper genotypes were laid on food supplied with 0.2mM trans-retinal and raised at 25°C in constant darkness until 3 rd instar larvae. 590nm red light was used to activate pdf-LaNs or ACLP R60F09 neurons. Images were acquired at 1.109s per frame at resolution of 512 × 512 in pixels.
For quantitative analysis of Ca 2+ imaging data, we used ImageJ (imageJ.nih.gov/ij) to batch process images to determine fluorescence intensity of regions of interest. We subjectively selected a few sequential images right before thermal stimulation to calculate average fluorescence intensity (F) as the basal level. Changes in fluorescence intensity (ΔF) were calculated and ΔF/F was used to denote Ca 2+ responses.

Ca-LexA ( calcium-dependent nuclear import of LexA ) imaging.
Eggs of proper genotypes were laid on food and raised at 25°C under 12h:12 h light/dark cycles for two days. The larvae were then divided into two groups and placed in incubators with internal temperature of 18 ℃ and 27 ℃ respectively. After 18 hours, larvae were dissected, fixed, and imaged with confocal microscopy (see following).
Images were acquired using an Olympus FV1000 confocal laser scanning microscope with 20X-, 40X oil-or 60X oil-objective lens at resolution of 1024 × 1024 in pixels.

Single neuronal morphology labeling experiment.
Eggs of proper genotypes were laid and collected on food. The food containing the collected eggs was evenly dispersed in a plastic petri dish, then the petri dish was placed in a 37 ℃ water bath for 5 minutes, after which the food containing the eggs was transferred to a vial containing standard culture medium. Eggs were reared at 25 ℃ under 12h:12 h light/dark cycles for 3 days. 3 rd instar larvae were dissected, fixed, and imaged.

Statistics.
For all the tests, paired t-test, unpaired t-test or one-way ANOVA with Tukey's post hoc test were used. Error bars in scatter plot and shaded areas flanking curves represented SEM. (a) WT-CS larvae show stronger light avoidance at lower temperature in a light/dark choice assay for 10 minutes. *** P < 0.001, one-way ANOVA followed by post hoc Tukey's multiple comparison test; n = 16 for all temperatures except that n = 24 for 25 ℃.
White light of 550 lux (23.3 μW/mm 2 ) was used in both (a) and (b). Error bars, SEMs. (a) Blocking ACLP R60F09 neurons using TNTG abolished difference in light avoidance at cool and warm temperatures in a light/dark preference assay using white light at intensity of 170 lux. n.s. P > 0.05, *** P < 0.001, unpaired t-test; n = 26, 26, 29, 29, 38 and 38 from left to right. Note that performance index of larvae with ACLP R60F09 neurons blocked using TNTG is higher than in control Drosophila larvae at 27 ℃ (* P < 0.05, **P < 0.01, one-way ANOVA followed by post hoc Tukey's multiple comparison test; n = 26, 29 and 38 from left to right). Error bars, SEMs. (c, d) Morphology of single R60F09-GAL4 neurons in brain hemispheres, including ACLP R60F09 neuron (c) and the 3 rd neuron lacking contralateral projection (d).
(e) Morphology of two ACLP R60F09 neurons located in the same brain hemisphere. Arrowheads point to the cell bodies. Long thin arrows point to dendrites. Short thick arrows point to axonal projections.
Treating the neurons with 20 µM TTX for 4 minutes abolishes the response. n = 7 for TTX treated ACLP R60F09 neurons. n = 9 for control ACLP R60F09 neurons treated with AHL.
(g-r) ACLP R60F09 neurons are GABAergic. The three neurons in one brain hemisphere (g-i) are separately shown in (j-r). cell 1,2 and 3 are separately shown in (j-l), (m-o), and (p-r). anti-GABA signal is in magenta. GCAMP signal driven by R60F09-Gal4 is in green. Arrowhead in (g) indicates the cell bodies of the three neurons. Note that R82B10-Gal4 and R82B10-LexA were used to label CLPN R82B09 neurons. Scale bar is 50 µm in (g-i). (a) Ca 2+ imaging of ACLP R60F09 neurons' responses to blue light at 18 ℃ and 27 ℃. 460 nm light stimulation was applied for 1 second. n = 12.
(c) The percent of decrease in the average amplitude of ACLP R60F09 neurons' response to blue light at 18 ℃ compared with that at 27 ℃. ** P < 0.01, *** P < 0.001, t-test against zero level. n = 12.