Knockdown of a Cyclic Nucleotide-Gated Ion Channel Impairs Locomotor Activity and Recovery From Hypoxia in Adult Drosophila melanogaster

Cyclic guanosine monophosphate (cGMP) modulates the speed of recovery from anoxia in adult Drosophila and mediates hypoxia-related behaviors in larvae. Cyclic nucleotide-gated channels (CNG) and cGMP-activated protein kinase (PKG) are two cGMP downstream targets. PKG is involved in behavioral tolerance to hypoxia and anoxia in adults, however little is known about a role for CNG channels. We used a CNGL (CNG-like) mutant with reduced CNGL transcripts to investigate the contribution of CNGL to the hypoxia response. CNGL mutants had reduced locomotor activity under normoxia. A shorter distance travelled in a standard locomotor assay was due to a slower walking speed and more frequent stops. In control flies, hypoxia immediately reduced path length per minute. Flies took 30–40 min in normoxia for >90% recovery of path length per minute from 15 min hypoxia. CNGL mutants had impaired recovery from hypoxia; 40 min for ∼10% recovery of walking speed. The effects of CNGL mutation on locomotor activity and recovery from hypoxia were recapitulated by pan-neuronal CNGL knockdown. Genetic manipulation to increase cGMP in the CNGL mutants increased locomotor activity under normoxia and eliminated the impairment of recovery from hypoxia. We conclude that CNGL channels and cGMP signaling are involved in the control of locomotor activity and the hypoxic response of adult Drosophila.


INTRODUCTION
Insects have evolved remarkable adaptive mechanisms to tolerate hypoxia or anoxia. Apart from a few specialized, hypoxia-tolerant species, anoxia in vertebrates leads to irreversible tissue damage or death within minutes. However, the median lethal time (LT50) of midge, Chironomus plumosis, larvae exposed to anoxia can be as long as~200 days (Nagell and Landahl, 1978). Also, the larvae of tiger beetles, Cicindela togata, survive anoxia at 25°C with an LT50 of more than three days (Hoback et al., 1998;Hoback et al., 2000) and locusts, Locusta migratoria, are able to survive exposure to 100% nitrogen for up to six hours at room temperature (Wu et al., 2002). Similarly, adults of fruit flies, Drosophila melanogaster, can survive several hours in anoxia (Haddad, 2006;Callier et al., 2015;Campbell et al., 2019b). At extremely low oxygen tension (0.1 kPa), flies reduce metabolism around 10-fold relative to normoxia (Van Voorhies, 2009). The exceptional survival and physiological responses to hypoxia or anoxia in Drosophila provide a good opportunity to explore its underlying molecular and genetic basis (Campbell et al., 2019a).
Rapid behavioral and electrophysiological responses to hypoxia involve the regulation of ion channels by reduced O 2 levels (López-Barneo, 1996). The first identified ion channel responsible for O 2 sensing was a voltage-dependent K + channel (López-Barneo et al., 1988) found in Type I chemoreceptor cells in the mammalian carotid body. Hypoxia (PO 2 dropping from 150 to 10 mm Hg) reduces the K + current through the channel by 25-50%. Another O 2 -sensing channel is the O 2 -sensitive ATP-inhibitable K + channel, which is found in neocortical and substantia nigra neurons in the rat CNS . Hypoxia activates this channel and causes it to open and close with increased frequency. Moreover, in rat hippocampal CA1 neurons, the voltagedependent, fast-inactivating Na + inward current and neuronal excitability are depressed with decreased O 2 levels (Cummins et al., 1991). In Caenorhabditis elegans, the hypoxia responses require cyclic guanosine monophosphate (cGMP)-gated cation channels (CNGs) including tax-2 and tax-4, and the atypical soluble guanylyl cyclases (sGCs) such as GCY-35 (Gray et al., 2004;Chang et al., 2006;Zimmer et al., 2009).
In Drosophila, oxygen sensing and behavioral responses to hypoxia were undescribed until a class of sGCs was identified, including Gyc88E, Gyc89Da and Gyc89Db (Morton, 2004) which contribute to the hypoxia response in Drosophila larvae (Vermehren-Schmaedick et al., 2010). Drosophila has a family of 4 different CNG channel genes, one of which, CNGA, regulates an escape response to hypoxia in Drosophila larvae (Vermehren-Schmaedick et al., 2010). However, similar responses have not been attributed to the other CNG channel family members, such as CNG-like (CNGL), CG3536 and CG17922. Flies with down-regulation of cGMP-specific phosphodiesterase (PDE) recover locomotor ability rapidly from anoxia, suggesting the involvement of cGMP in the modulation of anoxia recovery speed in adult flies (Xiao and Robertson, 2017). In Drosophila larvae, reduced levels of cGMP in O 2 -sensitive neurons result in longer times to respond to hypoxia, indicating that cGMP also regulates the escape response to hypoxia in Drosophila larvae (Vermehren-Schmaedick et al., 2010). cGMP has several downstream targets, such as CNG channels and cGMP-dependent protein kinase G (PKG). Flies with lower PKG activity show an increased time to the onset of anoxic coma and are more behaviorally resistant to anoxia and hypoxia (Dawson-Scully et al., 2010;Spong et al., 2016). However, whether CNG channels regulate the anoxia or hypoxia responses of adult flies is unknown. Under anoxia, neural activity is shutdown (Robertson et al., 2020), whereas under hypoxia, the CNS remains functional, indicating that it is feasible to investigate the role of CNG channels in maintaining neural function and regulating locomotor activity under hypoxia. To date, one of the CNG channel family members, the CNGL channel, has received little attention.
CNGL is detected in fly optic lobe, central brain and thoracic ganglia, shows similarity to the mammalian CNG channel a and ß subunits, and is predicted to form heteromeric channels with similar sensitivity to cAMP and cGMP (Miyazu et al., 2000;Vermehren-Schmaedick et al., 2010). The CNGL channel does not contribute to the hypoxia escape response in Drosophila larvae (Vermehren-Schmaedick et al., 2010). However, CNGL knockdown reduced visual orientation memory (Kuntz et al., 2017) and a CNGL homolog in Hawaiian crickets is a candidate gene underlying interspecific variation in centrally-generated song patterns (Xu and Shaw, 2019) indicating a role in motor patterning in the nervous system.
Considering that cGMP and PKG are involved in responses to hypoxia and that CNG channels contribute to the escape response to hypoxia in Drosophila larvae, yet CNGL channels do not, we investigated the role of CNGL channels in regulating the hypoxia response of adult flies. We hypothesized that mutation of CNGL would alter responses to hypoxia and tested the hypothesis using a locomotor assay and comparing CNGL mutant flies in a w1118 genetic background and w1118 control flies under normoxia, under hypoxia, and during recovery. In addition, we examined the effects of hypoxia on locomotor activity in pan-neural or panglial CNGL knockdown flies. Finally, we used fly lines with overexpression of Gyc88E (soluble guanylate cyclase) or mutation of a phosophodiesterase, Pde1c, to examine the possible interaction between CNGL and cGMP in response to hypoxia.
Male progeny (CNGL MB01092 /y) were obtained by crossing female CNGL MB01092 flies with male w1118 flies. This was to reduce as much as possible any differences in the genetic background of CNGL MB01092 and w1118 flies. Male progeny (CNGL MB01092 /y; Pde1c KG05572 /+) were obtained by crossing female CNGL MB01092 flies with male Pde1c KG05572 flies. These Pde1c KG05572 flies were confirmed in previous investigations in our laboratory to have reduced Pde1c transcript levels (Xiao and Robertson, 2017). Further synchronization of the genetic background was difficult due to the lack of an independent eye-color marker in the CNGL MB01092 .
Flies were raised on standard medium (0.01% molasses, 8.2% cornmeal, 3.4% killed yeast, 0.94% agar, 0.18% benzoic acid, 0.66% propionic acid) at room temperature 21-23°C, 60-70% humidity. A 12h/12 h light/dark cycle was provided by three light bulbs (Philips 13 W compact fluorescent energy saver) with lights on at 7 am and off at 7 pm. Male flies were collected within two days after eclosion and raised for at least three additional days before experiments. 100% N 2 gas was used to knock down flies during collection but flies were not exposed to 100% N 2 for 3 days prior to testing. All experiments took place between 10 am and 4 pm on flies younger than 9 days old.

Locomotor Assay
The locomotor assay has been described (Xiao and Robertson, 2015). Illumination for video-capture (frame rate of 15 frames/ s) was provided with a white light box (Logan Portaview slide/ transparency viewer with 15-W color-corrected fluorescent bulbs) and the reflection from white cardboard screens surrounding the set up. Individual flies were gently aspirated into circular arenas (1.27 cm diameter; 0.3 cm depth) that prevented flying but allowed their walking to be video recorded for later analysis. A custom-written fly tracking software using Open Computer Vision 2.0 (OpenCV2.0) was used for computing fly positions (the center of mass) and calculating path length per minute. Gas (air or the hypoxia mixture) continuously flowed through the apparatus at 2 L/min. After 5 min of adaptation in the arena and 10 min of locomotor activity recorded under normoxia, hypoxia was induced by switching from air to a mixture of 2% O 2 and 98% N 2 for 15 min. These parameters for the hypoxia were chosen based on preliminary results indicating a good balance between the severity of the effect and the ability to recover in a short enough time to prevent the flies being stressed (dehydration and starvation) in the locomotor chambers. The ratio of O 2 and N 2 chosen here was based on previous studies, which define such a composition as "hypoxia" (Allweis et al., 1984;Donohoe et al., 2000). After switching back to air, the flies recovered under normoxia for 40 min. Experiments were conducted in the light and at least three hours away from light-dark transit, in order to avoid morning and evening activity peaks (Grima et al., 2004).
Every video recording was inspected to ensure that the flies were walking apparently normally in the arenas throughout the experiment and not changing position due to uncoordinated movements. This pertains particularly to repo-Gal4/UAS-CNGL-RNAi flies in which locomotor activity was severely impaired and they were excluded from the analysis of recovery from hypoxia. For this genotype, the large transient changes in path length represented movements associated with seizures and were evident only for the flies with CNGL knockdown in glia. Flies do not walk continuously in one direction in the arena but pause for variable periods and/or change direction (Xiao et al., 2018). To estimate the frequency of pausing during the first 10 min in normoxia, we counted the number of 1 s intervals during which the path length was zero (e.g. a fly motionless for 10 min would have 600 s of zero path length, represented as 600 stops/10 min). To estimate maximum walking speed, we chose the farthest distance travelled in 1 s during the first 10 min, assuming that during normoxia each fly would have at least one 1 s episode of continuous walking. Every minute the total path length, including stationary periods, was averaged and plotted under three different conditions including normoxia (10 min), hypoxia (15 min), and recovery after hypoxia (40 min). In addition, the path length per minute under normoxia was averaged over 10 min (1-10 min) (L normoxia ), under hypoxia during the last 10 min (16-25 min) (L hypoxia) and under normoxia during the first 10 min after 5 min adaptation (31-40 min) and the last 10 min of recovery (56-65 min) (L recovery ). L normoxia was compared within and between fly genotypes. To determine the effect of O 2 level changes on fly locomotion, L normoxia , L hypoxia and L recovery were compared within the same fly line. Percent recovery was calculated only for the fly lines with significant effects of O 2 level changes on fly locomotor activity (p < 0.05 when performing Kruskal-Wallis test). For each fly, the percent recovery after hypoxia was compared within and between fly genotypes using the values for L normoxia and L recovery (i.e. the final 10 min of recovery). At the end of the experiment, flies were removed from the locomotor assay arenas and discarded without further observation.

Statistics
Three replicates of each experiment with groups of flies were conducted. The sample sizes for each test are stated in the Results section. It should be noted that, due to the inconsistent yields of progeny obtained after fly crosses, the sample sizes for the experiments varied. Statistical analysis was performed using Prism version 5.0 (GraphPad Software, San Diego, CA to examine the data distribution. Because some of the data had non-Gaussian distributions, nonparametric Mann-Whitney or Kruskal-Wallis tests with post-hoc comparisons were performed to examine the difference of medians between groups. Kruskal-Wallis test (without post-hoc comparisons) was performed to determine the effect of O 2 level changes within the same fly line. p < 0.05 was considered as indicating statistical significance. Path lengths over time for different genotypes are displayed as mean ± standard error. Average path lengths during different stages of the experiment are displayed as medians with whiskers outlining the interquartile range, overlaid with individual data points.

CNGL Transcriptional Down-Regulation in CNGL MB01092 Flies
RT-PCR was used to examine transcriptional changes in CNGL mutants. According to the National Center for Biotechnology Information (NCBI) nucleotide database, there are five mRNA variants including CNGL-RD, RE, RJ, RI and RF for CNGL ( Figure 1A). Among them, CNGL MB01092 showed a reduction of CNGL-RD, RE, RJ and RI transcripts compared with w1118 ( Figure 1B). The other variant, CNGL-RF, was likely unchanged. Thus, CNGL MB01092 flies had reduced levels of CNGL transcripts.

Minimal Locomotor Activity Under Normoxia in Flies With CNGL Knockdown in Glia
RNAi knockdown of CNGL using the pan-glial driver repo-Gal4 (Sepp et al., 2001) showed no apparent morphological abnormality or developmental delay. However, it led to greatly reduced locomotor activity compared with controls under normoxia ( Figures 4A-D, Supplementary Video S5). ;;repo-Gal4/UAS-  Figure 4E). An acute increase in path length per minute representing seizures could be observed in ;;repo-Gal4/UAS-CNGL-RNAi flies at the beginning of hypoxia treatment (Supplementary Video S5; Figure 4C).  (Xiao and Robertson, 2017). However, it is unclear whether cGMP modulates locomotor activity under normoxia and whether it regulates the hypoxia response in adult flies. Moreover, it would be possible for increased activation of a reduced number of channels in the mutant to mitigate the locomotor impairment. Therefore, an atypical soluble guanylyl cyclase Gyc88E, which catalyzes cGMP production, was expressed in CNGL mutant flies to examine whether cGMP upregulation could compensate for the effects of CNGL downregulation. Under normoxia, the flies CNGL MB01092 /y; UAS-Gyc88E/+; had improved locomotor activity compared with CNGL MB01092 /y flies ( Figures 5A-D). Although the average path length per minute of flies CNGL MB01092 /y; UAS-Gyc88E/+; (n = 16, median 43.2 cm/min, IQR 39.5-49.5 cm/min) was comparable to flies ;UAS-Gyc88E/+; (n = 25, median 44.5 cm/min, IQR 36.9-48.9 cm/min) (p > 0.05, Kruskal-Wallis test with Dunn's multiple comparison), the average speed of flies CNGL MB01092 /y; UAS-Gyc88E/+; was significantly higher compared with CNGL MB01092 /y flies (n = 14, median 6.2 cm/min, IQR 3.2-10.2 cm/min) (p < 0.001, Kruskal-Wallis test with Dunn's multiple comparison) ( Figure 5E, left panel). The alterations in O 2 level affected locomotor activity in all three fly lines (p < 0.001 for all three comparisons within same genotype, Kruskal-Wallis test without post-hoc test) ( Figure 5E). The CNGL MB01092 /y flies also showed an acute increase in path length per minute within 2 min at the onset of hypoxia ( Figures 5A,D).
Therefore, the overexpression of Gyc88E in CNGL mutant flies suppressed the reduced recovery in response to hypoxia.

Pde1c Mutation Eliminated the Reduced Recovery in Response to Hypoxia in CNGL Mutant
Overexpression of Gyc88E, which leads to the upregulation of cGMP, suppresses the w-RNAi induced delay of locomotor recovery (Xiao and Robertson, 2017) and eliminates the reduced recovery in response to hypoxia in CNGL mutant. Genes for cGMP-specific PDEs, which regulate intracellular levels of cGMP by hydrolyzing cGMP, have been identified in Drosophila (Morton and Hudson, 2002;Day et al., 2005). Therefore, PDE mutation provides another approach to investigate the impact of cGMP upregulation (Xiao and Robertson, 2017). One of the PDE mutants, Pde1c KG05572 , could compensate for the w-RNAi-induced delay of locomotor recovery from anoxia. Here we examined whether the Pde1c mutation could also eliminate the reduced locomotor activity under normoxia and the reduced recovery in response to hypoxia in CNGL mutant flies.
Among the three fly genotypes under normoxia, flies CNGL MB01092 /y; Pde1c KG05572 /+; displayed better locomotor activity compared with CNGL mutant flies (Figures 6A-D). The average path length per minute in CNGL MB01092 /y; Pde1c KG05572 /+; was 31.2 cm/min (n = 15, IQR 29.2-34.0 cm/ min). Even though it was a smaller path length per minute than flies; Pde1c KG05572 /+; (n = 12, median 48.2 cm/min, IQR 42.8-51.7 cm/min), it was more than for CNGL mutant flies CNGL MB01092 /y flies (n = 8, median 7.3 cm/min, IQR 2.4-18.5 cm/min) (p < 0.001 for both comparisons, Kruskal-Wallis test with Dunn's multiple comparison) ( Figure 6E, left panel). These data indicate that Pde1c mutation suppressed the reduced locomotor activity under normoxia in CNGL mutant. In addition, O 2 level changes affected fly locomotor activities in all three fly genotypes (p < 0.001 for all three comparisons within same genotypes, Kruskal-Wallis test without post-hoc test) ( Figure 6E). Similarly, the path length per minute in CNGL MB01092 /y flies reached a peak at the beginning of the hypoxia treatment and immediately dropped~10 cm/min, which was then followed by a gradual increase (Figures 6A,D).

DISCUSSION
CNGL MB01092 /y mutant w1118 flies had low levels of CNGL transcripts and reduced locomotor activity in normoxia compared with control w1118 flies. Reduced locomotor activity in the mutant was a result of more time spent motionless in between episodes of walking and a lower walking speed during episodes. In control flies, hypoxia immediately reduced locomotor activity measured by path length per minute, however there was complete recovery upon the return of air (~40 min to recover after 15 min of hypoxia). However, CNGL MB01092 /y flies had an immediate and transient increase in path length per minute at the beginning of hypoxia, Frontiers in Physiology | www.frontiersin.org April 2022 | Volume 13 | Article 852919 followed by a gradual increase during the hypoxia period and poor recovery from hypoxia. CNGL knockdown in neurons replicated the properties of the CNGL MB01092 /y mutant. However, CNGL knockdown in glial cells impaired locomotor activity to such an extent that it was not possible to characterize effects of hypoxia and return to air. Finally, we showed that crossing the mutant with UAS-Gyc88E or the Pde1c KG05572 mutant, both of which would upregulate cGMP, compensated for CNGL downregulation and enabled locomotor recovery from hypoxia in the mutant line. We show that CNGL transcript level is lower in the w1118 CNGL MB01092 /y mutants than in w1118 controls, however we do not know the effect of this on channel abundance, kinetics or ion currents. Our conclusions therefore relate to the clear and undeniable effects of this genetic mutation on locomotor activity and responses to hypoxia rather than to a defined manipulation of channel function. We propose that CNGL channel activity is reduced in the mutant but this requires confirmation in future experiments. In addition, although the genetic backgrounds of both control and mutant lines were w1118, it remains a possibility that other genetic factors may have influenced the results. However, the effects of the mutation, which reduced levels of CNGL transcript, were recapitulated by RNAi targeting CNGL in neurons and were compensated by two independent manipulations (Gyc88E and Pde1c) that would increase cGMP levels. Given previous published data on CNG channels, responses to hypoxia in larvae and the role of a white/cGMP/ PKG signaling pathway in responses of adult flies to anoxia (Dawson-Scully et al., 2010;Vermehren-Schmaedick et al., 2010;Spong et al., 2016;Xiao and Robertson, 2017), we propose that our results were due to changes in CNGL ion channel content and activity. We found that CNGL MB01092 /y had reduced locomotor activity under normoxia, suggesting that CNGL is necessary for normal motor activity in adult flies. This notion is supported by the structural similarity of CNGL with cyclic nucleotide-gated HCN channels (Xu and Shaw, 2019), which are intimately involved in the central generation of motor patterns in many taxa (Picton et al., 2018). Indeed, it is intriguing that many of the effects of the CNGL mutation on fly walking are like the effects of blocking HCN channels on swimming of larval Xenopus, including shorter episodes of slower locomotion. In larval Xenopus, HCN currents compensate for the hyperpolarizing activity of the Na + /K + -ATPase linking the locomotor effects to ion homeostasis in the central circuitry (Picton et al., 2018). Alternatively, the increased duration of quiescence with fewer episodes of walking may have been due to a disruption of neuromodulation resulting from the downregulation of CNGL. In larval zebrafish, fictive swimming with an episodic pattern is evoked by the application of N-methyl-Daspartate (NMDA), which is an activator of the larval zebrafish swimming central pattern generator (McDearmid and Drapeau, 2006;Wiggin et al., 2012). In C. elegans episodic locomotion is also observed with spontaneous alternation between active swimming and inactivity, and this transition is promoted by acetylcholine (ACh) signaling (Ghosh and Emmons, 2008). So far, no direct evidence has been found for the involvement of the CNGL channel in the release of ACh, however, CNG channels can modulate glutamatergic synaptic transmission (Zagotta and Siegelbaum, 1996;Bradley et al., 1997;Savchenko et al., 1997).
The CNGL channel is critical for normal locomotor activity in adult flies. CNGL MB01092 /y flies showed reduced locomotor activity and slower recovery after hypoxia, effects which may both be consequences of the same deficiency. This was comparable with CNGL knockdown in neurons, however, a full comparison with CNGL knockdown in glia was not possible because of the greatly impaired locomotor activity in these flies. Indeed, RNAi-mediated knockdown of CNGL in either neurons or glia had more extreme effects than the mutation, suggesting that the mutant retained more CNGL function. The CNGL gene is highly expressed in the nervous tissue, particularly in neuropil regions of the brain (Miyazu et al., 2000), indicating a role in neuronal circuit operation as noted above. However, it is unclear how much CNGL is in glia, which is also critical for neural function and behavior by regulating ion homeostasis and synaptic transmission. It is likely that the CNGL channels in glial cells have different functions from the CNGL channels in neurons, leading to different sensitivities to hypoxia or varied effects of O 2 level changes on locomotor activities. Indeed, the flies with CNGL knockdown in glia were more sensitive to hypoxia to the extent that they convulsed and entered anoxic coma with O 2 levels that were permissive for locomotor activity in all the other genotypes tested. This is an interesting observation that deserves further research in the context of the role of glia in controlling ion homeostasis and regulating anoxic depolarization in the brain (Armstrong et al., 2011). Nevertheless, compared with Gal4 control flies, the reduced recovery of flies with CNGL knockdown in neurons suggests that CNGL channels expressed in neurons regulate hypoxic recovery. To determine the contributions of specific neuronal or glial cell types to locomotor activity or the response to hypoxia, specific driver lines for neurons or glia of the fly CNS could be used to cross with UAS-CNGL-RNAi flies.
The behavioral response to hypoxia in Drosophila larvae depends on the nitric oxide (NO)/cyclic GMP (cGMP) signaling pathway (Wingrove and O'farrell, 1999). In swimming Xenopus laevis larvae the application of S-nitroso-N-acetylpenicillamine (SNAP) to increase NO levels, results in a reduced episode duration (the time for which an animal swims in response to a brief sensory stimulus) and an increased cycle period (interval between the onset of a burst in one cycle and the onset of the burst in the next cycle) (McLean and Sillar, 2000). On the other hand, treatment with nitric oxide synthase (NOS) inhibitors, such as L-NAME and L-NNA, to reduce NO levels increases the episode duration and decreases the cycle period. The effect of SNAP on the swimming motor pattern is similar to that of sodium azide (NaN 3 ) when it is applied to Xenopus laevis larvae (Robertson et al., 2010), which is widely used to induce chemical hypoxia in neurons (Varming et al., 1996;Jørgensen et al., 1999;Selvatici et al., 2009) and is reported to generate NO (Smith et al., 1991;Smith and Wilcox, 1994). In adult flies, it is still unclear why an acute increase of locomotor activity was generated at the beginning of hypoxia, especially in the flies with CNGL down-regulation or pan-neuronal knockdown. However, it demonstrates that the locomotor system can operate faster. Reduced locomotor activity followed by the acute increase in the fly lines noted above as well as the decreased activity in all other tested fly lines during hypoxia might be due to increased levels of NO, reducing episode duration and thus decreasing the path length per minute. An intriguing possibility is that CNGL potentiates motor patterning and hypoxic activation of other CNG channels can compensate for its knockdown.
The hypoxia escape response in Drosophila larvae requires functional CNGA channels but not CNGL channels (Vermehren-Schmaedick et al., 2010). This is consistent with the observation that CNGA channels promote the mobilization of calcium downstream of hypoxia-induced NO signaling but CNGL channels do not (Dijkers and O'farrell, 2009). cGMP modulates the speed of anoxic recovery in adult flies (Xiao and Robertson, 2017) and mediates the escape response to hypoxia in Drosophila larvae (Vermehren-Schmaedick et al., 2010). One of the cGMP downstream targets, PKG, is involved in behavioral tolerance to anoxia and hypoxia (Dawson-Scully et al., 2010;Spong et al., 2016). Our results show that the CNGL channel, one member of the CNG ion channel family, which represents downstream targets of cGMP, also regulated the hypoxia response in adult flies. However, whether the hypoxia response in adults also requires the other members of CNG channels such as CNGA, was not investigated in our research. In Drosophila larvae, hypoxia activates two heterodimeric atypical sGCs (Gyc-88E/Gyc-89Da and Gyc-88E/Gyc-89Db), which act as O 2 sensors (Morton, 2011). The sGCs then promote the production of cGMP, which activates CNG channels. Drosophila larvae withdraw from food in response to hypoxia (Vermehren-Schmaedick et al., 2010) or remain moving in anoxia for almost 40 min (Callier et al., 2015), whereas adults show complete paralysis within 30 s under anoxia (Callier et al., 2015). To examine the involvement of O 2 sensors in adult flies, the hypoxia responses in the atypical sGCs mutant flies should be compared with the controls.
We also found that the overexpression of Gyc88E or the mutation of Pde1c compensated for the reduced recovery from hypoxia in CNGL mutant flies. We did not measure cGMP levels when Gyc88E was overexpressed, however, the overexpression of Gyc88E eliminates the white (w) gene-RNAi-induced delay of locomotor recovery and the effect is similar to several PDE mutants, especially cGMP-specific PDE mutants (Xiao and Robertson, 2017), suggesting that the Gyc88E overexpression is likely to have increased cGMP levels. Genes for cGMP-specific PDEs, which also regulate intracellular levels of cGMP by hydrolyzing cGMP, have been identified in Drosophila (Morton and Hudson, 2002;Day et al., 2005). Therefore, PDE mutation is another approach to investigate the impact of cGMP upregulation (Xiao and Robertson, 2017). One of the PDE mutants, Pde1c KG05572 , compensated for the w-RNAi-induced delay of locomotor recovery from anoxia. It should be noted that ;UAS-Gyc88E/+; recovered poorly from hypoxia. The reason for this inconsistency between these two approaches might be that the Pde1c mutation was not a UAS-line and therefore, it lost tissue specificity and could not function as specifically as the UAS-Gyc88E fly line, in which Gyc88E was overexpressed only in CNGL-positive cells. The reason for this is unclear, however, compared with CNGL mutant flies CNGL MB01092 /y, the remarkable increased recovery of the flies with Gyc88E overexpression or Pde1c mutation in CNGL mutants suggests that cGMP upregulation eliminated the impaired recovery induced by the down-regulation of CNGL in response to hypoxia. cGMP upregulation may have potentiated the activity of remaining CNGL channels, however it is also possible that it compensated for the mutation by activating an alternative downstream pathway.
In summary, our results support the conclusion that the CNGL channel is required for proper locomotor activity and its recovery from hypoxia in Drosophila adults. With respect to central pattern generation, the role of cGMP and CNGL channels in episodic motor activity are future research areas that are likely to be particularly fruitful.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.