Low Glycolysis Is Neuroprotective during Anoxic Spreading Depolarization (SD) and Reoxygenation in Locusts

Abstract Migratory locusts enter a reversible hypometabolic coma to survive environmental anoxia, wherein the cessation of CNS activity is driven by spreading depolarization (SD). While glycolysis is recognized as a crucial anaerobic energy source contributing to animal anoxia tolerance, its influence on the anoxic SD trajectory and recovery outcomes remains poorly understood. We investigated the effects of varying glycolytic capacity on adult female locust anoxic SD parameters, using glucose or the glycolytic inhibitors 2-deoxy-d-glucose (2DG) or monosodium iodoacetate (MIA). Surprisingly, 2DG treatment shared similarities with glucose yet had opposite effects compared with MIA. Specifically, although SD onset was not affected, both glucose and 2DG expedited the recovery of CNS electrical activity during reoxygenation, whereas MIA delayed it. Additionally, glucose and MIA, but not 2DG, increased tissue damage and neural cell death following anoxia-reoxygenation. Notably, glucose-induced injuries were associated with heightened CO2 output during the early phase of reoxygenation. Conversely, 2DG resulted in a bimodal response, initially dampening CO2 output and gradually increasing it throughout the recovery period. Given the discrepancies between effects of 2DG and MIA, the current results require cautious interpretations. Nonetheless, our findings present evidence that glycolysis is not a critical metabolic component in either anoxic SD onset or recovery and that heightened glycolysis during reoxygenation may exacerbate CNS injuries. Furthermore, we suggest that locust anoxic recovery is not solely dependent on energy availability, and the regulation of metabolic flux during early reoxygenation may constitute a strategy to mitigate damage.


Introduction
CNSs with restricted extracellular space and protected by diffusion barriers can reversibly shut-down because of a process known as spreading depolarization (SD; Robertson et al., 2020).In migratory locusts (Locusta migratoria), environmental anoxia triggers SD through the failure of aerobic metabolism and ion homeostasis, which suppresses neural activity at the onset of a hypometabolic coma (Robertson et al., 2020;Robertson and Van Dusen, 2021).SD is characterized by the abrupt collapse of ion gradients across neural cell membranes that propagates as self-regenerating waves.Such a process involves large surges of extracellular potassium ([K 1 ] o ), intracellular sodium and calcium ([Na 1 ] I , [Ca 21 ] i ), respectively, across the neural cell membrane, as well as the ensuing cell swelling from water influx (Robertson et al., 2020;Lemale et al., 2022).The loss of potential energy stored as transmembrane ion gradients presents a substantial energetic challenge for SD recovery, because of the need to restore changes in the intracellular and extracellular milieu.Thus, while often fully recoverable and benign, SD frequently predisposes neural tissues to injury and cell death.At least in higher animals, including humans, SD is the characteristic change in the state of neurons that virtually always occurs in the dying process, i.e., at the transition from life to death (Carlson et al., 2018;Dreier et al., 2018Dreier et al., , 2019)).In the case of anoxic SD, although reperfusion is necessary for survival and recovery, the process of reactivating oxidative metabolism could further compound tissue injuries (Gomez et al., 2023).
Glucose availability is a critical factor modifying tissue susceptibility for SD and the resultant damage (Hoffmann et al., 2013;Sarrafzadeh et al., 2013;Lourenço et al., 2017;Rogers et al., 2017).In locusts, glycogen is depleted in the CNS following anoxia, suggesting heightened glucose utilization during anoxic SD (Hochachka, 1993).Anaerobic glycolysis is often erroneously described as the source of lowered cellular pH during SD via "lactic acid" production (Rossi et al., 2007;Uzdensky, 2019;Bo et al., 2020;Tóth et al., 2020;Grech et al., 2021;Lima et al., 2022); however, only lactate is formed which does not contribute to acidification (Robergs et al., 2018).Nonetheless, few studies have examined the impact of glycolysis and glycolytically-derived energy on SD in metabolically challenged tissues, and the conclusions have been difficult to reconcile.Some studies, for instance, show that hyperglycemia and anaerobic glycolysis enhance tissue SD resistance and reduce injury under cerebral ischemia (Hartings et al., 2008;DiNuzzo et al., 2012;Hoffmann et al., 2013;Hertz et al., 2015;Wotton et al., 2020), while others associate excess glucose supply and neuronal glycolysis with worsened secondary damage post-SD (De Courten-Myers et al., 1994;Herrero-Mendez et al., 2009;Li et al., 2019;Barros et al., 2021).In mammals, the relation between ischemic injuries and serum glucose presumably presents a U-shaped curve (Dreier and Reiffurth, 2015), and a recent matched cohort study puts the optimal serum glucose for aneurysmal subarachnoid hemorrhage (aSAH) recovery at higher than physiological level (Eagles et al., 2022).Ultimately, it remains unclear whether and how increased CNS glycolytic flux is beneficial for anoxic SD recovery or neural cell survival.
The insect CNS provides an excellent model for examining the fundamental impact of neuroenergetics on SD events.Such an approach is advantageous by circumventing the additional complexity arising from mammalian neurovascular responses (Spong et al., 2017).Both the Drosophila brain and the locust thoracic ganglia generate SD reliably, and share similar triggers and electrophysiological characteristics with mammalian models (Spong et al., 2016;Shuttleworth et al., 2020).Moreover, many aspects of CNS homeostasis and energy metabolism, including K 1 spatial buffering and neuron-glia metabolite shuttling, are conserved (Spong and Robertson, 2013;Weiler et al., 2017;Rittschof and Schirmeier, 2018).Lastly, the universal reliance on glycolysis by both insect and mammalian glia underscores the central role of glycolytic flux in modulating CNS homeostasis across species (Bélanger et al., 2011;Volkenhoff et al., 2015).
In this study, we investigated the impact of CNS glycolysis on anoxic SD in the locust metathoracic ganglion (MTG).To do so, we manipulated the animal's glycolytic capacity pharmacologically, by supplying glucose as substrate to promote glycolysis, and monosodium iodoacetate (MIA) or 2-deoxy-Dglucose (2DG) to inhibit it.We took multiple approaches to monitor the physiological effects of our manipulations on N 2induced anoxic SD, including electrophysiology to monitor SD trajectories and CNS ion disturbances by measuring the transperineurial potential (TPP) across the ganglion sheath, respirometry to assess the effects on aerobic metabolism, and quantifying CNS injury through neural cell death and oxidative damage.

Animals
Adult L. migratoria aged four to five weeks past adult moult were reared in a crowded colony located in the Animal Care facility at the Bioscience Complex at Queen's University, Kingston, Ontario, Canada.The colony is maintained at around 30°C under light, and room temperature when dark, with a 12/12 h light/dark cycle.The animals were fed daily with a diet of wheatgrass, bran, milk protein, and yeast.All animals selected for experiments were aged on average five weeks past the final moult and were randomly assigned to treatment groups.Our preliminary electrophysiology experiments revealed greater influence of glucose in females compared with males.Thus, all subsequent data were collected using only females to maximize treatment effects and minimize variabilities because of sex.Animals were transported in well-ventilated plastic containers to the labs.

Pharmacology
The final concentrations of the drugs used in all experiments were: 10 mM (180 mg/dl) for glucose, 5 mM (104 mg/dl) for MIA, and 50 mM (820 mg/dl) for 2-deoxy-D-glucose (2DG).The concentration of glucose was chosen based on a prior study, where 10 mM maximally stimulated O 2 uptake of isolated ganglia (Clement and Strang, 1978).The reference range for hemolymph glucose concentrations is between 10 and 20 mM from start to 1 h after feeding (180-360 mg/dl; Zanotto et al., 1996).Compared with hemolymph, the semiintact preparation allows for greater fluid volume (;2 ml), which provides as a larger glucose reserve for the CNS.Initially, equimolar concentrations of 2DG and MIA were used with glucose.However, the concentration of 2DG was increased to 50 mM as no noticeable effects were observed (preliminary observations), and the concentration of MIA was reduced to 5 mM because of spontaneous and irreversible negative shift of TPPs.
For bath application during electrophysiology, aliquots of 100 mM glucose or 25 mM MIA or 100 mM 2DG (MIA; Sigma-Aldrich) in standard locust saline (147 mM NaCl, 10 mM KCl, 4 mM CaCl 2 , 3 mM NaOH, and 10 mM HEPES buffer; pH 7.2) were diluted to 10, 5, or 50 mM, respectively, before each experiment.For whole-animal injections, the volume of hemolymph in an animal is estimated to be 200 ml based on a prior study (Ayali and Pener, 1992).Accordingly, 30 ml of 80 mM glucose or 40 mM MIA or 400 mM 2DG in standard locust saline were delivered to achieve the appropriate final concentration in circulation.

Dissection
An illustration of a semi-intact preparation is shown in Figure 1Ai.The appendages were snipped near the body and a portion of the dorsal pronotum was removed.An incision was made along the dorsal midline, starting near the seventh abdominal segment and ending at the head capsule.The animal was then pinned to the cork floor of a sealable acrylic chamber (5 Â 5.2 Â 2 cm), exposing the thoracic cavity.The air sacs, fat bodies and eggs were subsequently removed, and the gut was clipped at the posterior end and pulled over the head capsule.Standard locust saline was added into the thoracic cavity immediately afterward to prevent dehydration.The ventral diaphragm was then removed, exposing the metathoracic ganglion and the associated nerve roots.An aquarium air pump ventilated the chamber during dissection and subsequent recording sessions with a flow rate of 100 ml/min.

Recording setup
The animal was grounded using a chlorided silver wire placed at the caudal end of the abdominal cavity in contact with bathing saline.The electrode placements are shown in Figure 1Aii.TPP was measured with an extracellular DC electrode made from 1-mm filamented glass capillary (World Precision Instruments) and pulled to a tip resistance of ;5 MV.The electrode and holder were filled with 500 mM and 3 M KCl, respectively, and connected to a DC amplifier (Model 1600, A-M Systems).The DC electrode was calibrated in the bathing saline before penetration of MTG at the start of the experiment and after the removal from the MTG at the end of the experiment, to adjust for baseline shift.Nerve activity was measured by an AC suction electrode on the dorsal median nerve of the MTG.The AC electrode was made from 1-mm nonfilamented glass capillary (World Precision Instruments) and pulled to ;5-MV tip resistance; the tip was then broken off with forceps to align with the diameter of the dorsal median nerve.The AC electrode was connected to a differential AC amplifier (model 1700, A-M Systems).The analog signals from both amplifiers were digitized (DigiData 1322A, Molecular Devices) and recorded with Axoscope 10.7 (Molecular Devices).

Recordings of anoxic SD
Each day of the experiment included an equal number of control and treatment animals to account for daily variations in the colony.Following dissection, the saline in the thoracic cavity was changed to either the treatment or the control saline (washed three times), with a 20-min pretreatment period before anoxia.During pretreatment, the DC and AC electrodes were calibrated and positioned in place, and the chamber was sealed with cellophane tape.N 2 anoxia was introduced immediately following the pretreatment period; 10 min after the abrupt negative DC shift marking SD onset, the N 2 was switched off and the air pump circulated the air back to reoxygenate the animal.A constant flowrate of 100 ml/min was maintained for air and N 2 .
The time to motor pattern failure (F patt ), the time to excitability failure (F exc ) and the time to SD (F TPP ) were the times taken from the start of anoxia to the disappearance of distinct rhythmical nerve activity, nerve excitation, and the abrupt negative DC shift of TPP, respectively.The time to motor pattern recovery (R patt ), the time to excitability recovery (R exc ) and the time to SD recovery (R TPP ) were the times taken for the return of rhythmical nerve activity, nerve excitation, and the positive peak value of TPP after the air returns, respectively.SD amplitude was taken as the difference between the baseline TPP preanoxia and the steady TPP value immediately following the abrupt negative shift; post-SD positivity represents the difference between the maximum recovered TPP value and the preanoxia baseline.TPP recovery slope was measured as the linear slope of TPP increase at the initial phase of recovery.The recovery time constant (t ) was obtained by fitting the DC recovery trace with the standard exponential model in Clampfit 10.7 (Molecular Devices).

Molecular and biochemical assays Tissue preparation
Intact animals were restrained with scotch tape and injected with either saline or drugs using a 50-ml Hamilton syringe (Hamilton Company).They were then placed in the same chamber used in respirometry experiments for 30 min following injection for pretreatment, and subjected to 30-min N 2 anoxia except for control animals, which did not receive anoxia.The animals subsequently recovered for 2 h in normoxia to allow injuries to develop, before each MTG was extracted through a ventral incision and snap frozen in liquid N 2 for later use.All tissue samples were homogenized in lysis buffers supplied by the corresponding assay kits, using a TissueLyzer II bead mill (one 5-mm stainless steel bead per tube, 30 s À1 frequency for 1.5 min; QIAGEN Inc.).The homogenate protein concentrations were determined using a QuantiPro BCA protein assay kit (Sigma-Aldrich; catalog #QPBCA).All colorimetric and fluorometric assays were performed using the SpectraMax Paradigm microplate reader (Molecular Devices).

Caspase-3 activity assay
The level of neural caspase three activation was assessed using a Caspase-3 Assay kit, Fluorometric (Sigma-Aldrich; catalog #CASP3F), following manufacturer's instructions.Modifications were made to recommended kit reagent volumes to ensure each sample assayed an individual MTG.In brief, each MTG was homogenized in 150ml lysis buffer, loaded as duplicates on a 96-well microplate (50 ml per well), and then reacted with equal volume of 2Â substrate working solutions.The concentration of fluorescent end-product 7-amino-4-methylcoumarin (AMC) was measured with excitation and emission wavelength of 360 and 460 nm, respectively.Results are presented as mmol AMC cleavage per hour per mg protein in the sample.Outliers in data were removed using Grubb's test online (https://www.graphpad.com/quickcalcs/grubbs1/),figure with outliers is presented in Extended Data Figure 7-1.

TBARS assay
The extent of lipid peroxidation damage was estimated with QuantiChrom TBARS Assay kit (BioAssay Systems; catalog #DTBA-100), following manufacturer's instructions.Briefly, two MTGs were pooled into each sample, homogenized in 200 ml of cold PBS.The homogenate was then incubated with equal volumes of thiobarbituric (TBA) reagent for 1 h at 100°C.The reaction mixtures were cooled and loaded into a 96-well microplate in duplicates, then measured fluorometrically using excitation and emission wavelength of 530 and 550 nm, respectively.Results are presented as mmol MDA per mg protein.

Immunoblotting
The protein carbonyl content in ganglionic tissue was measured with Protein Carbonyl Assay kit (Western blotting; Abcam; catalog #ab178020), following manufacturer's instructions.In short, two MTGs were pooled in each sample and homogenized in 100 ml tris-buffered saline (pH 7.5) containing cOmplete ULTRA protease inhibitor cocktail (Sigma-Aldrich).The homogenates were combined with equal volumes of 2Â extraction buffer and spun at 18,000 Â g for 20 min at 4°C, the supernatants were split equally and reacted with either dinitrophenylhydrazine (DNPH) or control solutions before gel electrophoresis.
To optimize protein loading amounts, a standard curve was generated and analyzed using the Empiria 2.0 software linear range detection function (LI-COR Biosciences).20 mg of control or derivatized protein were resolved on a 7.5% SDS-PAGE gel and transferred onto a nitrocellulose membrane (0.45 mM pore size).Total protein signal was measured with the Revert 700 total protein stain kit (LI-COR), before acquiring the signal for anti-dinitrophenyl (DNP; supplied in ab178020, 1:5000).An Odyssey XF imaging system was used to image the membrane; the whole-lane signals were quantified and then normalized against total protein, using the Empiria 2.0 software (LI-COR Biosciences).Data were compiled from five blots (all biological replicates), and normalized signal intensities are plotted as fold change compared with control.The unprocessed blot images are provided in Extended Data Figure 7-2.

Respirometry
A simple flow-through respirometry system was constructed (Fig. 1B).A manually operated three-way valve controlled the feeding of either compressed N 2 or normoxic air supplied by an air pump (Qbit Systems Inc.), which passed through a CO 2 scrubber containing soda lime before entering the animal chamber.Water vapor was removed with a Drierite filter before the gas entered an infrared CO 2 analyzer (QS-151, Qbit Systems Inc.); the analog signals from the analyzer were digitized (DigiData 1322A, Molecular Devices) and acquired in Axoscope 10.7 (Molecular Devices).The flowrate of oxygenated air and N 2 were maintained at a constant 100 ml/min as indicated by the flow monitor (Q-G268, Qbit Systems Inc.).
Intact animals were used for respirometry and were injected as described above, and immediately placed in the chamber.The animals were allowed 30 min to recover from manipulation and for their CO 2 output to stabilize under normoxic air.Afterwards, N 2 was switched on to induce a 35min anoxia, followed by 2 h of recovery in normoxic air.Following recovery, the MTG was immediately collected and stored to supplement downstream molecular and biochemical assays.Data analysis was performed in Clampfit 10.7 (Molecular Devices), where the CO 2 emission trace was integrated to calculate gas output for a given time interval.Raw data in PPM were converted to emission rate expressed in ml CO 2 per minute using the following formula:  and equality of group variances were determined by Shapiro-Wilk and Brown-Forsythe tests, respectively.For electrophysiology results, the means of normally distributed data were compared by two-samples independent t test; otherwise, Mann-Whitney ranked sum test was used to compare medians.For all other data, one-way ANOVA test was used.For representative traces in the figures, the sample closest to the mean were selected and processed using Clampfit 10.7 (Molecular Devices), where abrupt and short-lived (,1 ms) electrical artifacts were replaced with local mean values.
To compare recovery kinetics between different pretreatment regimens, we calculated the initial linear phase TPP recovery rate and the recovery time constant t (Fig. 3). 10 mM glucose pretreatment had a significantly higher initial rate of TPP recovery (two-sample Student's t test, vs control; Control: n ¼ 5, Glucose: n ¼ 6, p ¼ 0.042; Fig.
To determine whether the altered postanoxic TPP level from 2DG and MIA group originates from different magnitudes of negative DC shift, we measured the SD amplitudes (Fig. 5C).Surprisingly, 10 mM glucose significantly reduced SD amplitudes (two-sample Student's t test, vs control; Control: n ¼ 5, Glucose: n ¼ 6, p ¼ 0.027; Fig. 5Ai

2DG and glucose have opposing, temporally dependent effects on CO 2 emission during reoxygenation
Considering that 2DG inhibits glycolysis and thus, pyruvate production, we hypothesized that 2DG also influences aerobic metabolism.Whole animal flow-through respirometry revealed a striking observation that both control and glucose experienced a rapid rise in CO 2 emissions during reoxygenation, whereas 2DG led to a slower and more steady increase (Fig. 6A).MIA results were not included as the treatments completely abolished ventilation pattern and was deemed too toxic for the animals.Interestingly, only glucose significantly increased CO 2 output during the first 20 min of reoxygenation, the period that corresponds to the re-establishment of CNS ion-homeostasis and functioning (one-way ANOVA, Control: n ¼ 9, Glucose: n ¼ 8, 2DG: n ¼ 8, F ¼ 5.472, p ¼ 0.012; Holm-Sidak pairwise comparison, glucose vs control, p ¼ 0.033; glucose vs 2DG, p ¼ 0.016; 2DG vs control, p ¼ 0.56; Fig. 6Bi).On the other hand, both 2DG and glucose pretreatments significantly increased total CO 2 emission in the 2-h period following reoxygenation (one-way ANOVA, Control: n ¼ 9, Glucose: n ¼ 8, 2DG: n ¼ 8, F ¼ 4.033, p ¼ 0.032; Fig. 6Bii).

The pattern of CNS damage following anoxic SD recovery
To examine whether improved electrophysiological performance corresponds to reduced CNS damage following anoxic SD, we measured the protein carbonyls levels and tissue malondialdehyde (MDA) content after N 2 anoxia, to quantify the extent of protein oxidation and membrane lipid peroxidation, respectively (Fig. 7A,B).
Energy stress and cellular injuries could lead to neural cell apoptosis following anoxia-reoxygenation.We assessed the activated Caspase-3 (Cas-3) activities in the mesothoracic and metathoracic ganglia to measure CNS programmed cell death after N 2 anoxia.The ganglionic Cas-3 activity was significantly affected by pretreatments (one-way ANOVA, Control: n ¼ 14, Anoxic Control: n ¼ 16, Glucose: n ¼ 18, 2DG: n ¼ 17, MIA: n ¼ 16; F ¼ 5.945; p , 0.001; Fig. 7C); glucose increased Cas-3 activity as measured by substrate cleavage, compared with control and 2DG groups (Holm-Sidak multiple comparison; p ¼ 0.004 and p ¼ 0.016, respectively), MIA similarly exhibited a higher AMC cleavage rate than control and 2DG groups (p ¼ 0.005 and p ¼ 0.018, respectively).Pretreatment with 2DG had no effect on ganglionic Cas-3 activity compared with control and anoxic control (p ¼ 0.76 and p ¼ 0.66, respectively), 25 min N 2 anoxia also had limited effect on Cas-3 activity compared with control (p ¼ 0.36).

SD onset is independent of glycolytic capacity
Pharmacological manipulations of locust CNS glycolytic capacity did not affect parameters of anoxic SD onset in this study, consistent with the theory that insect SD constitutes a secondary adaptation to conserve CNS energy (Robertson et al., 2020).Indeed, delaying anoxic coma onset by reducing cGMP-dependent protein kinase (PKG) activity in Drosophila significantly worsens survival and recovery (Dawson-Scully et al., 2010).Additionally, SD induction in locusts is modulated by metabolic stress sensors like NO and AMPK signaling pathways (Armstrong et al., 2009;Rodgers-Garlick et al., 2011;Money et al., 2014), which may act to suppress anaerobic metabolism under anoxia.It must be noted that locusts would not encounter pure N 2 in their natural habitats, despite the experimental convenience of N 2 -anoxia.Nonetheless, the effects of N 2 -induced SD are similar to that of water submersion, an ecologically relevant hazard for many insects (Robertson and Van Dusen, 2021).

Discrepancies between 2DG and MIA
The contrasting effects of 2DG and MIA, which are both glycolytic inhibitors, on recovery kinetics and tissue damage patterns are challenging to interpret.Nonetheless, growing evidence suggests that MIA has effects beyond glycolysis inhibition.Iodoacetate derivatives, including MIA, nonspecifically modify a wide range of cysteine containing proteins, including the glycolytic enzyme glyceraldehyde 3phosphate dehydrogenase (GAPDH; Carneiro et al., 2011).The thiol-oxidizing properties of MIA also contribute to the depletion of cellular glutathione and antioxidant potentials (Schmidt and Dringen, 2009).Notably, the oxidative off-target effects of MIA might have played a significant role in preventing TPP recovery, as MIA inhibition of methioninesulfoxide reductase (MSR) functions could delay insect anoxic SD recovery and worsen tissue injury (Moskovitz et al., 2000;Franklin et al., 2013;Gu et al., 2016;Suthakaran et al., 2021).It is also worth mentioning the persistent post-SD negativity (which normally should be a transient positivity, see control, glucose, and 2DG groups; Fig. 5), that may suggest MIA challenges locust CNS in ways other than glycolytic inhibition.On the other hand, reports of 2DG's offtarget effects are sporadic and mainly in the context of cell proliferation (Mireuta et al., 2011;Pang et al., 2015).

Potential mechanism of glycolysis-related damage
Whereas glucose and 2DG have similar effects on recovery kinetics, the underlying mechanisms are likely different.It is possible that glucose hastens recovery by driving aerobic glycolysis and indirectly stimulating oxidative phosphorylation (OXPHOS), evident by the increased CO 2 output early on during reoxygenation.Such an overall increase in metabolic flux could rapidly replenish neural energy supply and bolster ion clearance capacity, yet it also precipitates latent tissue damage.
We speculate that heightened glycolysis likely contributes to neural injuries by disrupting mitochondrial redox balance and driving ROS production, thereby exacerbating tissue oxidative stress and infarction.During reoxygenation, mitochondria rapidly oxidize glycolytic end products, including lactate, pyruvate, and NADH, generated under both aerobic and anaerobic conditions (Hochachka, 1993).Therefore, high glycolytic flux stimulated by glucose may create a mismatch between substrate supply and mitochondrial demand, fostering conditions for reverse electron transfer (RET) at Complex I (Hoffman and Brookes, 2009).
For instance, Complex I reduces ubiquinone to ubiquinol with electrons from NADH.However, high NADH/NAD 1 and ubiquinol/ubiquinone ratio, resulting from oversupply of glycolytic end products, make the reverse reaction thermodynamically favorable (Onukwufor et al., 2019).Additionally, the TCA cycle converts accumulated lactate and pyruvate into succinate, which, on subsequent oxidation at Complex II, further increases the level of reduced ubiquinol contributing to RET (Bundgaard et al., 2019).Indeed, the greater CO 2 output in the glucose group is suggestive of the heightened TCA cycle flux during reoxygenation.
Consequently, the reversal of electron flow favors the formation of superoxide (O 2 -) and reactive oxygen species (ROS) that damages macromolecules and cellular components.These effects potentially give rise to the increased protein carbonyl content observed in glucose group, though it remains unclear why lipid peroxidation decreased simultaneously.Finally, ROS bursts increase the probability of mitochondrial permeability transition that precipitates programed neural cell death, reflected as greater Cas-3 activation in glucose group.

Potential mechanisms underlying 2DG's effects
Given that the glycolytic inhibitor 2DG also expedited the recovery of TPP and neuronal excitability similar to glucose, we speculate 2DG-animals may have switched to b -oxidation to fuel recovery (McMullen et al., 2023).Fatty acids metabolism is slower but generates greater energy compared with glycolysis, which may explain the initial low CO 2 output in 2DG group.Although insect CNS stores little lipids, glial cells may mobilize peripheral fat body reserves when glycolysis is impaired, and supply ketone bodies to neurons as oxidative fuel (Rittschof and Schirmeier, 2018).For instance, it is thought that Drosophila glia secretes Glaz (homologous to apolipoprotein ApoD) to communicate with the fat body during starvation (McMullen et al., 2023); a similar mechanism likely exists in the locust CNS, considering the central role of fatty acids in powering flight muscles (Haunerland, 1997).In the context of anoxic SD recovery, glycolysis block coupled with increased b oxidation may help prevent the oversupply of TCA cycle substrates and preserve the stoichiometric balance of intermediates.As such, the probability of RET and ROS production during reoxygenation would likely decrease, agreeing with the trend of reduced protein carbonylation in 2DG group.
A peculiar effect of 2DG is the increased magnitude of the post-SD positivity.It is worth noting that a similar form of DC positivity is observed following seizures, SD, and periods of heightened neuronal activities in a number of mammalian studies (Gutnick et al., 1979;Haglund and Schwartzkroin, 1990;Kawasaki et al., 1990;D'Ambrosio et al., 2002).
There are two potential mechanisms that could underlie the origin of this positive shift of TPP following recovery.First, the DC positivity may be the consequence of higher Na 1 /K 1 ATPase activity during SD recovery.The loss of ion homeostasis during SD could trigger an increase in pumping rates to restore ion gradients.However, the heightened pump activity may persist even after the restoration of physiological level of [K 1 ] o , leading to an undershoot that is reflected as a slight positive shift of TPP.The subsequent return of [K 1 ] o and TPP to the preanoxic baseline level is thought to be mediated by glial K 1 extrusion mechanisms involving KIR channels (D'Ambrosio et al., 2002).Given the glycolytic preference of insect glia (Rittschof and Schirmeier, 2018), 2DG may interfere with these processes, resulting in greater post-SD positivity.
An alternative explanation of the DC positivity is the differential recovery rates of the component membrane potentials of TPP.The insect TPP represents the potential difference between two polarities of the barrier-forming perineurial cell layer; commonly denoted as the basolateral potential (V b , intracellular vs hemolymph) and the adglial potential (V a , intracellular vs CNS interstitium, see Extended Data Fig. 5-1; Schofield and Treherne, 1984;Robertson et al., 2020).Accordingly, TPP recovery trajectory is dependent on the opposing exponential recoveries of V a and V b during reoxygenation (with time constants t a and t b ).Importantly, when V a recovers faster than V b , the TPP develops a transient positive shift that diminishes after both V a and V b returns to preanoxic baseline level (Extended Data Fig. 5-1).It is likely that the effects of 2DG are primarily driven by accelerated V a recovery and reduced t a , resulting in a faster TPP recovery rate and greater post-SD positivity.2DG may shorten t a by inhibiting uncontrolled excitations, which conserves energy and accelerates the return of CNS ion homeostasis.This effect may also explain the faster re-establishment of ventilatory motor patterning from un-patterned excitations in 2DG group.
Incidentally, 2DG's anticonvulsant effects are well-documented (Wijayasinghe et al., 2022;Sutula and Fountain, 2023), although it is unclear whether the insect neuromuscular hyperexcitability during SD onset and recovery is analogous to mammalian epileptiform bursts.In any case, the mechanism behind 2DG's impact on recovery trajectory warrants further investigation.

Merit of low metabolism during reoxygenation
The CNS is particularly vulnerable to anoxic damage in two major ways: (1) energy depletion during anoxia leads to necrotic cell death and subsequent apoptosis (Callizot et al., 2019;Lemale et al., 2022;Tuo et al., 2022), and (2) reactivation of the stalled electron transport chain (ETC) during reoxygenation causes oxidative damage, worsening tissue injury (Wu et al., 2020;Jaganjac et al., 2022).In the current experiments, supplying a metabolic fuel source (glucose) and an inhibitor (2DG) both improved recovery rates, but only glucose increased tissue damage and CO 2 output during early reoxygenation.Given such observations, we propose that acute glycolysis block by 2DG during anoxic SD recovery could be neuroprotective, and that high glycolytic flux and OXPHOS activity during reoxygenation may contribute to tissue injury in the insect CNS.
The protective effects of 2DG and carbohydrate restriction in the context of energy stress are not novel; however, as far as we are aware, our results represent the first documented acutely induced neuroprotection by this compound during anoxic SD 2DG has long been used to mimic the effect of dietary restriction on ischemic stroke and neurodegenerative diseases (Yu and Mattson, 1999;Pajak et al., 2019).Nonetheless, past studies often employed treatment periods lasting from days to weeks, that incur global changes in metabolic and gene expression profiles.In contrast, the effects observed in our study are likely to be induced by glycolysis inhibition alone, given the brief pretreatment periods (20-30 min) and high dose (50 mM or 820 mg/dl).
Given that the return of neural functioning precedes muscle activation, the first 20 min of whole-animal CO 2 production likely reflects the pattern of CNS metabolism well.Therefore, it is possible that blocking glycolysis may allow a slower and perhaps "smoother" reactivation of mitochondrial ETC during reoxygenation, through modulating substrate availability and thus the thermodynamic driving forces of respiration.Whether this was achieved through glycolysis inhibition alone or involves glial b oxidation requires further verification.Indeed, the ability to down-regulate glycolysis during oxygen reperfusion constitutes an important molecular response to aid the turtle's survival during anoxia (Bundgaard et al., 2019).As noted above, the elevated oxidative damage and cell death in glucose-treated animals are likely because of the greater respiratory activity than tissue demands, which counterintuitively jeopardizes animal survival despite providing a temporary energy boost during reoxygenation.

Limitations
A major limitation of the current results is the discrepancies between the effects of 2DG and MIA, which requires further experimentation to resolve.Despite potential offtarget effects, iodoacetates are historically reliable glycolytic inhibitors (Fuhrman and Field, 1943;Cano-Ramírez et al., 2012).We cannot exclude the possibility that MIA's effect mainly arises from GAPDH inhibition, or indeed if 2DG exhibited unknown off-target effects.Additionally, it is essential to note that our experiments do not account for necrotic cell death during anoxia.It is entirely possible that anaerobic glycolysis reduced uncontrolled cell lysis in the depolarized state by maintaining a small degree of ion homeostasis (Campbell et al., 2018).Indeed, we observed an attenuated SD amplitude in the glucose group only.As the negative shift of TPP during SD corresponds well with the surge of [K 1 ] o (Robertson et al., 2020), a lower SD amplitude is suggestive of reduced [K 1 ] o , potentially facilitated by anaerobic ion pumping (Armstrong et al., 2012).Future experiments should directly compare how glycolytic capacity affect neural necrosis during anoxia.

Conclusion and prospects
In summary, we present evidence that glycolysis is not a critical metabolic component in either anoxic SD onset or recovery in the locust CNS; we further propose the detrimental role of heightened glycolysis during reoxygenation that may lead to oxidative damage.Correspondingly, acute glycolysis inhibition through 2DG hastens SD recovery and potentially protects against anoxia-reoxygenation damage.Nonetheless, the precise physiological effects elicited by glycolytic inhibition at the cellular level remain unclear and warrant further investigation.Moreover, it will be useful to assess how animal glycolytic capacity affects the long-term behavioral outcomes and fitness following anoxia.
Elucidating how metabolic flux is controlled during the critical period of anoxia-reoxygenation is necessary to improve our understanding of strategies and mechanisms of insect anoxia tolerance.Also, similar experiments could be performed with mammalian ischemic SD models.Considering the prolific use of 2DG and its derivatives in human imaging studies, its effect on CNS glycolysis may have clinical significance in improving neurologic outcomes following stroke or cerebral ischemia.

Figure 1 .
Figure 1.Experimental setup.Ai, Diagram of semi-intact animal setup for electrophysiology, showing chamber and electrode arrangements.Aii, Enlarged dorsal view of the metathoracic ganglion with electrode placements.DC electrode measures transperineurial potential (TPP) within dashed areas, while AC (alternating current) suction electrode measures ventilatory nerve rhythm on the median nerve (MN).Peripheral nerves (N1, N3, N4, and N5) serve as guiding features.B, Not to scale.Schematic of flow-through respirometry system.Solid arrows indicate gas-tight vinyl tubing connections and gas flow direction; dashed arrow represents electrical connection for data acquisition.Refer to Materials and Methods for details.
Data analysisSigmaPlot 13 (SPSS Inc.) was used to analyze data and produce figures.a Was set at p ¼ 0.05.Data normality

Figure 2 .
Figure 2. SD onset and recovery times.A, Representative recording showing TPP and nerve activity.SD onset (F TPP ) is defined as the time when TPP reaches half-maximal amplitude of the abrupt negative shift during anoxia, while SD recovery (R TPP ) is defined as the time when TPP reaches its maximal value during reoxygenation.ReO 2 : Reoxygenation period.Bi, Bii, Biii, Glycolytic capacity has no effect on F TPP .Ci, Cii, Ciii, Glucose and 2DG significantly shorten R TPP , while MIA has no significant effect.Plotted data represent medians and IQR.Asterisks (*) indicate p , 0.05.Refer to Results for sample sizes and p-values.

Figure 3 .
Figure 3. SD recovery slopes and time constants t .Ai, Ai, Aiii, Representative recordings overlay showing TPP recovery trajectory in glucose, 2DG, and MIA groups, respectively, starting at the onset of reoxygenation.Bi, Bii, Biii, Glucose and 2DG increase the initial slope of TPP recovery, while MIA decreases the slope.Ci, Cii, Ciii, Corresponding to the changes in recovery slope, glucose, and 2DG decrease the time constants (t ) of the exponential recovery trajectory, whereas MIA increases t .Note the different scale in Ciii for clarity.Plotted data represent medians and IQR.Single asterisks (*) indicate p , 0.05, double asterisks indicate p , 0.001.Refer to Results for sample sizes and p-values.

Figure 4 .
Figure 4. Excitability and motorpattern failure and recovery times.Ai, Aii, Representative recordings of nerve activity during anoxia and reoxygenation, respectively, showing excitability failure (F exc ) and recovery (R exc ) defined as the disappearance and re-emergence of randomized spikes, and motor pattern failure (F patt ) and recovery (R patt ) defined as the disappearance and re-emergence of synchronized action potentials.Aii, Trace starts at the onset of reoxygenation.Bi, Bii, Biii, None of the treatments affect F exc .Ci, Cii, Ciii, Glucose and 2DG hasten R exc , while MIA delays the recovery.Note the different scales for clarity.Di, Dii, Diii, None of the treatments affect F patt .Ei, Eii, Eiii, Only 2DG accelerates R patt , while glucose and MIA have no effect.Plotted data represent medians and IQR.Single asterisks (*) indicate p , 0.05, double asterisks indicate p , 0.001.Refer to Results for sample sizes and p-values.

Figure 5 .
Figure 5. Post-SD positivity and SD amplitude.Ai, Aii, Aiii, Representative recordings overlay showing the trajectory of TPP.Horizontal dashed line represents the baseline TPP level before anoxia, which typically persist for a period under anoxia until SD onset.Vertical dashed line indicates the onset of reoxygenation (ReO 2 ).The arrow in Ai highlights attenuated SD amplitude in the glucose group, while arrows in Aii and Aiii show differences in post-SD positivity.Bi, Bii, Biii, Glucose has no effect on DC positivity, 2DG increases its magnitude, and MIA has the opposite effect.Ci, Cii, Ciii, Only glucose reduces SD amplitude.Plotted data represent medians and IQR.Single asterisks (*) indicate p , 0.05, and double asterisks indicate p , 0.001.Refer to Results for sample sizes and p-values.See Extended Data Figure 5-1 for the proposed mechanism of post-SD positivity in insects.

Figure 6 .
Figure 6.Effects on aerobic metabolism.Ai, Representative CO 2 emission trace during the respirometry experiment.Aii, Overlay of representative recordings from control, glucose, and 2DG groups, showing differential CO 2 emission patterns during the initial 30 min of reoxygenation.Glucose induces a rapid increase in emission rate, while 2DG has the opposite effect.Vertical dashed line indicates the 20 min mark, where treatment differences are most pronounced.Bi, Bii, CO 2 emission per gram of body mass during the first 20 min and the total 2-h reoxygenation period, respectively.Glucose animals emit higher CO 2 volume during the initial 20 min, while both glucose and 2DG increase CO 2 release over the 2-h recovery period.Plotted data represent medians and IQR.Refer to Results for sample sizes.

Figure 7 .
Figure 7. Patterns of tissue damage postanoxic SD.Ai, Western blotting detecting protein carbonylation in ganglionic homogenates, with total protein stain included for reference.Each sample (2 ganglia) is loaded in two lanes: C (control, no DNPH) and D (derivatized, with DNPH).Aii, Western blotting quantification shows that glucose and MIA significantly increase protein carbonylation.B, Anoxia significantly increases lipid peroxidation end-product MDA per mg protein content measured in TBARS assay, while glucose reduces MDA levels similar to control.C, Glucose and MIA induce greater neural apoptosis, as measured by Caspase-3 activity per mg protein.Each data point corresponds to a single animal in TBARS and Cas-3 activity assays.CTRL animals represent baseline damage levels, and AX-CTRL are anoxic control animals.Plotted data represent medians and IQR.Refer to Results for sample sizes.See Extended Data Figures 7-1 and 7-2 for Cas-3 activity assay result containing outliers and unaltered Western blotting images, respectively.