Sex-dependent effects of chronic intermittent hypoxia: Implication for obstructive sleep apnea

Abstract Background Obstructive sleep apnea (OSA) affects 10–26% of adults in the United States with known sex differences in prevalence and severity. OSA is characterized by elevated inflammation, oxidative stress (OS), and cognitive dysfunction. However, there is a paucity of data regarding the role of sex in the OSA phenotype. Prior findings suggest women exhibit different OSA phenotypes than men, which could result in under-reported OSA prevalence in women. To examine the relationship between OSA and sex, we used chronic intermittent hypoxia (CIH) to model OSA in rats. We hypothesized that CIH would produce sex-dependent phenotypes of inflammation, OS, and cognitive dysfunction, and these sex differences would be dependent on mitochondrial oxidative stress (mtOS). Methods Adult male and female Sprague Dawley rats were exposed to CIH or normoxia for 14 days to examine the impact of sex on CIH-associated circulating inflammation (IL-1β, IL-4, IL-6, IL-10, TNF-α), circulating OS, and behavior (recollective and spatial memory; gross and fine motor function; anxiety-like behaviors; and compulsive behaviors). A subset of rats was implanted with osmotic minipumps containing either a mitochondria-targeting antioxidant (MitoTEMPOL) or saline vehicle 1 week prior to CIH initiation to examine how inhibiting mtOS would affect the CIH phenotype. Results Sex-specific differences in CIH-induced inflammation, OS, motor function, and compulsive behavior were observed. In female rats, CIH increased inflammation (plasma IL-6 and IL-6/IL-10 ratio) and impaired fine motor function. Conversely, CIH elevated circulating OS and compulsivity in males. These sex-dependent effects of CIH were blocked by inhibiting mtOS. Interestingly, CIH impaired recollective memory in both sexes but these effects were not mediated by mtOS. No effects of CIH were observed on spatial memory, gross motor function, or anxiety-like behavior, regardless of sex. Conclusions Our results indicate that the impact of CIH is dependent on sex, such as an inflammatory response and OS response in females and males, respectively, that are mediated by mtOS. Interestingly, there was no effect of sex or mtOS in CIH-induced impairment of recollective memory. These results indicate that mtOS is involved in the sex differences observed in CIH, but a different mechanism underlies CIH-induced memory impairments.

compulsive behaviors).A subset of rats was implanted with osmotic minipumps containing either a mitochondria-targeting antioxidant (MitoTEMPOL) or saline vehicle 1 week prior to CIH initiation to examine how inhibiting mtOS would affect the CIH phenotype.

Results
Sex-speci c differences in CIH-induced in ammation, OS, motor function, and compulsive behavior were observed.In female rats, CIH increased in ammation (plasma IL-6 and IL-6/IL-10 ratio) and impaired ne motor function.Conversely, CIH elevated circulating OS and compulsivity in males.These sex-dependent effects of CIH were blocked by inhibiting mtOS.Interestingly, CIH impaired recollective memory in both sexes but these effects were not mediated by mtOS.No effects of CIH were observed on spatial memory, gross motor function, or anxiety-like behavior, regardless of sex.

Conclusions
Our results indicate that the impact of CIH is dependent on sex, such as an in ammatory response and OS response in females and males, respectively, that are mediated by mtOS.Interestingly, there was no effect of sex or mtOS in CIH-induced impairment of recollective memory.These results indicate that mtOS is involved in the sex differences observed in CIH, but a different mechanism underlies CIH-induced memory impairments.
Patients diagnosed with OSA are typically treated with continuous positive airway pressure (CPAP) machines, the gold standard for therapy [20][21][22].Effective CPAP therapy can reduce in ammation, OS, and cognitive dysfunction [21,[23][24][25].However, adherence to CPAP therapy is a major issue, as 15-40% of patients are non-compliant with prescribed machine usage [20][21][22].Even patients that are adherent to CPAP therapy may not be receiving proper treatment, due to anatomical variations or ineffective machine air pressure titration [20,22].This lack of e cacy has led to increased research to determine OSA mechanisms with the goal of developing conjunctive pharmaceutical therapies for OSA [26][27][28].Current drugs that are approved for OSA treatment are focused on symptom therapy, such as reducing daytime sleepiness and improving attention [27,28] rather than OSA pathophysiology as a result of OSA hypoxia.
Hypoxia induced by OSA is observed to increase many circulating markers of OS and in ammation [13][14][15][16][17][18].This can induce a feed-forward cycle, in which chronic in ammation increases OS [29,30], and chronic OS can increase in ammation [31,32].In addition, sex also plays a role in OS and in ammation [33,34].Women exhibit higher baseline levels of OS compared to men [35][36][37].Furthermore, women experience greater levels of in ammation and autoimmunity compared to men [38][39][40].Based on these ndings, sex differences observed in OSA may be related to basal sex differences in in ammatory and OS status.Since the mitochondria are one of the primary generators of OS within the cell [41,42] and mitochondrial dysfunction has been observed in OSA [43][44][45], mitochondrial OS (mtOS) may mediate sex differences in OSA.
To examine the impact of sex on OSA-associated symptoms (in ammation, OS, memory, motor, anxiety, compulsivity), we utilized an experimental rat model of OSA called chronic intermittent hypoxia (CIH).CIH exposure replicates the fragmented episodes of hypoxia observed in OSA [46].Further, to examine the mechanisms of mtOS on sex differences in CIH, we implanted osmotic pumps subcutaneously in a subgroup of rats to continuously administer a mitochondria-targeting superoxide dismutase mimetic, MitoTEMPOL (MT), to block mtOS.The in ammatory, OS, and behavioral phenotypes of CIH in males are clearly de ned.CIH increases in ammation [6,47,48], increases OS [6, 49,50], impairs recollective memory [51][52][53], impairs spatial learning and memory [47,[54][55][56], and increases anxiety-like behavior [51,52,57,58].Studies examining sex as a biological variable in CIH are scarce.Most of these studies have been conducted only in mice, showing CIH increased circulating OS in male but not female mice [49] and impaired spatial learning and memory in ovariectomized female mice but not intact females [59].Given the limited data on the effects of CIH in females, we predicted that CIH would produce sexdependent increases in in ammation and OS, alongside cognitive dysfunction.Further, we predicted that inhibiting mtOS would prevent the CIH phenotype of increased in ammation, increased OS, and cognitive dysfunction.

Animals
All experiments were conducted using adult virgin Sprague Dawley male and female rats (aged 3-4 months, Charles River, Wilmington, MA).Male and female rats were housed in separate rooms in our animal facility on a 12-hour (hr) reverse light cycle (lights were off at 09:00).Reverse lighting allowed behavioral testing to be conducted during the active phase of the circadian cycle.Food and water were provided ad libitum.Rats (n = 7-13/group) were randomly assigned to either normoxia (room air) or CIH treatment conditions.A subgroup of rats (n = 6-10) were given MT.This resulted in four groups/sex: Normoxic Vehicle, Normoxic MT, CIH Vehicle, CIH MT.To acclimatize the rats to operator handling and reduce stress responses during behavior testing, rats were handled daily, beginning one week prior to the start of behavior testing.At the conclusion of behavior testing, the rats were anesthetized with 2-3% iso urane and euthanized via decapitation during the active phase of the circadian cycle (09:00-11:00).All experiments were conducted in agreement with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the ARRIVE guidelines.These protocols were approved by the Institutional Animal Care and Use Committee of the University of North Texas Health Science Center.

Osmotic minipump implantation
One week prior to the initiation of CIH, all rats were instrumented with an osmotic minipump (Alzet Mini-  [65,69,70].To ensure proper osmotic function as indicated by bubbling present on the surface of pumps, all pumps were incubated in a 37°C water bath for at least 24 hours prior to implantation.Surgeries were performed using aseptic techniques with iso urane (2-3%) anesthetic.

Chronic intermittent hypoxia protocol
One week prior to the initiation of the CIH protocol, the home cages (clear plastic containers) of the rats were placed into Oxycycler chambers (76.2 x 50.8 x 50.8cm,BioSpherix, Lacona, NY, USA) to acclimatize the rats to the chambers under normoxic conditions.During their sleep phase of the circadian cycle, CIH was performed for 8 hrs starting at 21:00.The CIH protocol consisted of intermittent oxygen reduction from 21% (room air) to 10% in 6-minute cycles per hour (i.e., 10 cycles/hr) over 8 hrs/day for a period of 14 days, as previously described [6,46,71,72].10 CIH cycles per hour results in an AHI of 10, which is consistent with mild sleep apnea in humans [25,72].

Behavioral Tasks
Behavioral studies were conducted between days 8 and 14 of CIH from 09:45 to 17:00 during the active phase of the circadian cycle.The order of the behavior tests was randomized.Male and female rats were behaviorally tested in separate cohorts to prevent potential confounding effects of pheromones on behavior [73,74].All testing equipment (e.g., marbles, arenas, tanks) were thoroughly cleaned with 70% ethanol between each rat.All behavior studies were conducted under red lighting and recorded for later analysis by an investigator blinded to treatment groups.Behavior tests were used to assess compulsive behaviors (marble burying test) [75,76], ne motor function (footfalls, rearing behavior) [50], anxiety-like behaviors (center entries and center duration in an open eld) [77,78], spatial learning and memory (Morris water maze) [77,79] 50,71].To increase the di culty of locomotion, a wire mesh platform elevated 2 cm above the oor was placed in the arena.The rats were allowed 10 minutes to explore the arena.Fine motor function was classi ed by distance traveled, rearing behavior (assisted, unassisted, total), and footfalls past the elevated wire mesh.Rears were separated into assisted or unassisted when the animal stood on its hindlimbs with (assisted) or without (unassisted) its forelimbs bracing against the wall [46].

Open eld: Gross motor function and anxiety-like behavior
Using a large novel open eld arena (60.96 x 60.9 x 38.1 cm), we assessed gross motor function and anxiety-like behavior during a 5-minute trial.Behaviors were recorded using ANY-maze software (v.5.14, Stoelting CO.).This open eld duration has been previously observed to be sensitive to differences in anxiety-like behavior and stress in rats [85-87].Gross motor function was examined by distance traveled in the open eld.Anxiety-like behaviors were tracked by number of entries (frequency) to the center of the eld and time spent (duration) within the center of the open eld [50,71].In addition, this task allowed habituation to the large open eld arena in order to conduct the novel object recognition task.

Novel object recognition task: Recollective memory
Following habituation to the open eld test using the large arena (60.96 x 60.9 x 38.1 cm), we conducted the novel object recognition task [50].In this task, two identical objects (building blocks) were placed in adjacent corners of the arena, and the rats were given 5 minutes to interact with the objects and then removed from the arena.One hr later, one of the objects was replaced with a novel object (spherical toy ball), and the rats were given 3 minutes to explore the arena and interact with the objects.Contacts with the novel object and latency to the novel object were recorded as measures of short-term recollective memory retention [50,80,81].

Morris water maze: Spatial learning and memory
To examine spatial memory, the Morris water maze test was used according to our published protocols [50,71].Behaviors were recorded using ANY-maze software (v.5.14, Stoelting Co.).Over a period of 4 days, rats were trained to nd a submerged platform in a pool lled with opaque water (23-25°C) and remain on the platform for 20 seconds until removed by the operator.Day 1 of training consisted of visible platform pre-training where the platform was moved on each trial (3/day).On days 2-4, the platform was submerged for spatial training and remained in the same position for all trials.An unweighted learning index (LI) and a weighted learning index (WLI) were generated using the latency data from days 2, 3, and 4 of spatial training.The unweighted LI was generated as the average latency to the target of all trials of all training days [71,88].The WLI was calculated using the average of normoxic control rats/sex [88,89].Speci cally, WLI is calculated by obtaining a multiplier based on initial performance on day 2: day 2 multiplier (X̄D ay 2 Control Latency /X̄D ay 2 Control Latency ); day 3 multiplier (X̄D ay 2 Control Latency /X̄D ay 3 Control Latency ); and day 4 multiplier (X̄D ay 2 Control Latency /X̄D ay 4 Control Latency ).Then, all latencies for each day were multiplied by their respective multiplier (day 2 latency*day 2 multiplier; day 3 latency*day 3 multiplier; day 4 latency*day 4 multiplier) to generate a weighted latency score (WLS).The nal WLI used for analysis was the sum total of each WLS over days 2, 3 and 4 for each rat.The unweighted LI and WLI scores were used as indicators of spatial learning where lower unweighted LI and WLI scores indicated greater spatial learning ability [88,89].Spatial memory was assessed following 4 days of training, in which each rat was administered a probe trial (the underwater platform was removed).Latency and pathlength to the target during the probe trial were used as indicators of spatial memory.

Marble burying test: Compulsive behavior
To examine compulsive behaviors, the marble burying test was conducted following published protocols [71,75,76].To conduct this test, the oor of the testing arena (50 x 25 x 30 cm) was thoroughly covered with 1 cm of rodent bedding litter to allow the rats to easily bury marbles.Twenty marbles of similar color and size (1.5 cm) were spaced evenly in a 4 x 5 grid on one side of the arena base.Each rat was given 10 minutes to explore and interact with the marbles.Operators visually monitored the experiment and manually recorded behaviors.After 10 minutes, the rat was removed and the number of marbles buried (75% or more) was photographed and quanti ed.

Sample collection
At the conclusion of behavior tests, rats were anesthetized with iso urane (2-3%) and decapitated during the rst 2 hrs of the rats' active phase of the circadian rhythm to collect tissue and plasma samples [6,46,50,71].Trunk blood was collected in EDTA tubes, and then centrifuged at 2,000 x g for 10 minutes at 4°C to collect plasma.Plasma was stored at -80°C until assayed for circulating in ammation and oxidized proteins.

High throughput multiplex in ammation panel
A MILLIPLEX® rat cytokine/chemokine magnetic bead panel (Sigma Millipore, Cat # RECYTMAG-65K) utilizing antibodies against IL-1β, IL-4, IL-6, IL-10, and TNF-α was used to quantify circulating in ammatory cytokines.IL-6, TNF-α, and IL-1β were selected as common pro-in ammatory cytokines [90][91][92], while IL-10 and IL-4 were selected as common anti-in ammatory cytokines [93,94].An 18-hr overnight incubation was performed according to manufacturer instructions.All samples were diluted 1:2 in assay buffer prior to running the assay according to manufacturer's instructions.Samples were run in duplicate and cytokines were measured on Luminex® 200™ using xPONENT® software version 4.3 (Luminex Corporation, Austin, TX).Quality control values for each cytokine were within the range provided by the manufacturer.The ratio of IL-6/IL-10 was used to investigate the overall relationship between pro-and anti-in ammatory cytokines [95].Due to limited plasma availability, no analyses of normoxic MT plasma were performed.

Advanced oxidation protein products assay
Circulating plasma OS was assayed using OxiSelect Advanced Oxidation Protein Products (AOPP) kit (Cell Biolabs, Inc., San Diego, CA) according to our previously published protocols [6,46].The micromolar (µM) concentration of oxidized proteins in the plasma were measured relative to a known standard.
Chloramine in the kit reacts with oxidized proteins to produce a color change, which is read at 340 nm.To account for colorimetric interference from the plasma samples, a background correction was performed for all samples.All samples were diluted 1:2 in assay buffer for analysis.

Statistical analysis
Statistical analyses were conducted in IBM® SPSS® (SPSS® v. 29.0.0,IBM®, 2022).Normality of data distribution was tested using the Shapiro-Wilk test.Data with non-Gaussian distribution were normalized by log base 10 transformation (x = lg10(x)) or by square root transformation (x = sqrt(x)) through the SPSS® transformation function.Outliers greater than 2 standard deviations from the mean were removed from analysis.3-way ANOVAs were conducted using the factors of CIH, drug treatment, and sex, with the exception of 2-way ANOVAs used for in ammatory cytokine analyses.For all analyses, we provide the F values, degrees of freedom, p-values, and η 2 (measure of effect size).Following ANOVA testing, a Fisher's LSD post-hoc test was used to determine speci c group differences for signi cant results.Results are presented as mean ± S.E.M. unless otherwise indicated.Signi cance was de ned as p ≤ 0.05.We report non-signi cant trends as p ≤ 0.055.Post-hoc signi cance is indicated in gures and tables by unique letters (p ≤ 0.05), in which shared letters are not signi cantly different from each other.

Results
CIH induced a pro-in ammatory state in female rats that was reduced by blocking mitochondrial oxidative stress.
To examine the impact of sex and CIH on in ammatory cytokines, we used 2-way ANOVAs with the factors of sex and CIH in vehicle rats.We observed a signi cant interaction between CIH and sex on IL-6 levels (F 1, 19 =8.420; p = 0.009; η 2 = 0.304; Fig. 1A).Speci cally, CIH increased IL-6 levels only in female rats, and normoxic males had higher IL-6 levels than normoxic females.There was no observed effect of CIH or sex on IL-10 levels in vehicle rats (Table 1).However, there was a signi cant interaction between To examine the impact of sex and mtOS on CIH-induced in ammation, we used 2-way ANOVAs with the factors of sex and MT treatment (vehicle or MT) in CIH exposed rats.We observed that CIH females had higher IL-6 levels than CIH males (F 1, 20 =5.327; p = 0.032; η 2 = 0.210; Fig. 2A).Inhibition of mtOS with MT had no effect on CIH-elevated IL-6 levels (Fig. 2A).Similar to IL-6, CIH female rats had higher TNF-α levels than CIH males, regardless of MT (F 1, 18 =8.406;p = 0.010; η 2 = 0.318; Fig. 2B).There was a signi cant interaction between sex and MT on IL-1β levels in CIH rats, where vehicle males had higher IL-1β levels than vehicle females, and MT males had lower IL-1β levels than vehicle males and MT females (F 1, 19 =7.844; p = 0.011; η 2 = 0.287; Fig. 2C).A signi cant interaction was observed between sex and MT on IL-10 levels in CIH rats, where vehicle males had higher IL-10 levels than vehicle females, and MT increased IL-10 levels only in females (F 1, 20 =7.077; p = 0.015; η 2 = 0.237; Fig. 2D).We observed a main effect of sex on the IL-6/IL-10 ratio in CIH rats, where CIH-exposed females showed higher IL-6/IL-10 ratios compared to CIH-exposed males (F 1, 20 =8.331; p = 0.009; η 2 = 0.237; Fig. 2E).We also observed a signi cant interaction between sex and MT on the IL-6/IL-10 ratio in CIH rats, where MT only reduced the IL-6/IL-10 ratio in females (F 1, 20 =5.817; p = 0.026; η 2 = 0.165; Fig. 2E).No effects of sex or MT were observed on IL-4 levels in CIH rats (Table 2).CIH impaired recollective memory in male and female rats, which was unaffected by inhibiting mitochondrial oxidative stress.
We also examined spatial learning and memory using the Morris water maze.We observed lower WLI in female rats compared to males (F 1, 49 =8.385; p = 0.006; η 2 = 0.144), regardless of CIH or inhibition of mtOS with MT (Fig. 5A).No effects of CIH, sex, or MT were observed on unweighted LI (Fig. 5B).Next, we examined spatial memory by performing a probe trial test on the nal day of the Morris water maze.In contrast to the WLI scores observed, males exhibited a trend towards shorter latency to the probe target in Morris water maze than females (F 1, 52 =3.905; p = 0.053; η 2 = 0.067; Fig. 5C).No effects of CIH or inhibition of mtOS with MT were observed on latency to the probe target.We observed no effects of CIH, sex or mtOS inhibition on pathlength to the probe target in the Morris water maze (Table 3).To examine gross motor function, we used a large novel open eld (60.96 x 60.9 x 38.1 cm).We observed that female rats traveled farther than male rats (F 1, 60 =10.598; p = 0.002; η 2 = 0.136; Fig. 6A), particularly compared to CIH males with MT treatment (p ≤ 0.05).No effects of CIH or mtOS inhibition with MT were observed on distance traveled in the open eld (Fig. 6A).
Next, we examined ne motor function using a novel modi ed open eld (MOF; 40.64 x 40.64 x 38.1 cm) in which a wire mesh platform elevated 2 cm above the oor was placed in the arena.We observed a signi cant effect of CIH on MOF distance traveled (F 1, 64 =4.706; p = 0.034; η 2 = 0.058; Fig. 6B).In addition, we observed a signi cant interaction between CIH, sex, and mtOS with MT on distance traveled in the MOF (F 1, 64 =6.683; p = 0.012; η 2 = 0.083; Fig. 6B).Speci cally, CIH reduced MOF distance traveled in vehicle females, which was blocked by mtOS inhibition using MT (p ≤ 0.05).CIH did not impact ne motor function (distance traveled) in males.However, inhibition of mtOS did increase ne motor function in normoxic males (p ≤ 0.05).No effects of CIH, sex, or MT, were observed on other measures of ne motor function -assisted rears, unassisted rears, total rears, or footfalls in a modi ed open eld (Table 4).CIH increased compulsive behavior in male rats that was prevented by blocking mitochondrial oxidative stress.
We examined the effects of CIH, sex, and mtOS inhibition on compulsivity using the marble burying task.

Discussion
The major ndings of this study are 1) CIH-induced OS and in ammation are sex-and mtOS-dependent, whereas 2) CIH-induced behavioral changes are dependent on multiple factors that include sex, mtOS, and other mechanisms (Fig. 8).Speci cally, we found that CIH increased in ammation and ne motor dysfunction in female rats, which was blocked by inhibiting mtOS.Conversely, CIH increased circulating OS and compulsive behavior in males, which was blocked by inhibiting mtOS.However, we observed no sex differences in CIH-induced recollective memory impairment.Both male and female rats exposed to CIH performed worse than normoxic controls in the novel object recognition task, which was unaffected by inhibition of mtOS.
This study is the rst study to examine the effects of CIH on in ammatory cytokines in females, along with the rst study to compare sex differences in CIH-induced in ammatory cytokines.We observed that CIH increased circulating IL-6 levels and IL-6/IL-10 ratio in females.IL-6 is a pleiotrophic cytokine that is associated with many in ammatory diseases [90], whereas IL-10 is an anti-in ammatory cytokine that is linked with downregulating the effects of IL-6 in the immune response [93,96].The IL-6/IL-10 ratio indicates the status of pro-in ammatory and anti-in ammatory processes in circulation [95,96].
Irrespective of CIH, we observed sex differences in cytokine levels.In vehicle rats, normoxic males had higher levels of IL-6 than normoxic females.Few studies have examined sex differences in basal plasma cytokine levels; however, female rats typically have higher levels of IL-6 and TNF-α than males [97,98].In our study, we found no sex differences in any other cytokines, which is consistent with other studies examining sex and cytokine levels in plasma [97,98].
Although prior studies examining the impact of CIH on the immune system have only examined male rodents, they observed increased pro-in ammatory cytokines IL-6, TNF-α, and IL-1β in the circulation in response to CIH [47,48,99,100].In contrast to these studies, our study found that CIH did not impact IL-6, TNF-α, or IL-1β levels in male rats.This negative nding is consistent with our previous publication that showed no effect of CIH on IL-6 levels (Normoxic males: 558.54 ± 778.63 pg/mL; CIH males: 418.66 ± 551.45 pg/mL; mean ± SD) or TNF-α levels (Normoxic males: 5.16 ± 5.41 pg/mL; CIH males: 3.94 ± 4.03 pg/mL; mean ± SD) [6].One possible explanation for this discrepancy in CIH-induced cytokine levels is the CIH protocol.Our studies utilized a CIH protocol with an AHI of 10 (10 hypoxic episodes/hr/over 8 hrs/day) to model mild sleep apnea in humans [25,72].In contrast, other laboratories used different CIH protocols in male rats that range in AHIs of 30-40 with similar CIH protocol duration (8 hrs/day; 14-21 days) [48, 99, 100], which is consistent with severe sleep apnea in humans [25].This could indicate that the in ammatory response to CIH in males is dependent on AHI severity.
Similar to our cytokine data, we observed sex differences in circulating oxidized protein levels (AOPP).Consistent with prior reports [6, 49, 50], we found that CIH increased AOPP concentrations in male rats.
However, this is the rst study to show that CIH-elevated AOPP in male rats was mediated by mtOS.In contrast to male rats, CIH did not increase AOPP in female rats, which is consistent with other studies showing no effects of CIH on circulating OS measures [49,101].We also found sex differences in basal circulating OS, in which normoxic female rats had higher OS levels than male rats.This nding is consistent with clinical data wherein women frequently exhibit higher circulating OS than men [35][36][37].
This study is the rst to examine sex differences in CIH-induced behavior displayed by adult rats.We observed distinct sex differences in CIH-induced ne motor function impairment and compulsivity but not in CIH-induced recollective memory impairments.CIH reduced ne motor function only in females, which was blocked by inhibiting mtOS.Since inhibiting mtOS decreased CIH-induced in ammation in females, we propose that a mtOS-in ammation feed-forward mechanism is involved in CIH-induced ne motor impairment in females (Fig. 8A).Indeed, in ammation has been observed to impair motor behaviors [102,103], which may be related to CIH-induced neuroin ammation in motor behavior-associated brain regions in rats [6,99].
This is the rst study to examine the impact of CIH on marble burying.We observed that CIH increased compulsive behavior in male rats but not female rats.Further, inhibition of mtOS blocked CIH-induced compulsivity in males.Since inhibiting mtOS decreased CIH-induced OS in our males and CIH increases OS damage to brain regions [6, 57, 104] associated with marble burying behavior (e.g., hippocampus, orbitofrontal cortex, striatum) [77,105,106], mtOS may be one of the primary mechanisms in CIH-induced compulsivity in male rats (Fig. 8A).
We did not observe any sex differences in CIH-induced recollective memory impairment.Consistent with prior studies showing CIH-induced impairments in recollective memory in male rodents [51][52][53], we also found that CIH impaired recollective memory in male rats.CIH also decreased recollective memory in females, and our study is the rst to examine these relationships.Interestingly, inhibiting mtOS did not prevent the effects of CIH on recollective memory.This indicates that a mtOS-independent mechanism induced by CIH is impairing recollective memory in both sexes (Fig. 8B).Although no prior studies have examined the effect of CIH on recollective memory in females, studies have examined sex differences in recollective memory in response to neurotoxins.These ndings on sex differences in recollective memory were equivocal, in which some studies show no sex differences in recollective memory impairment [107,108] or only males were impaired and not females [107,109].
CIH had no effects on spatial learning and memory or anxiety-like behavior in either male or female rats.
Our ndings are consistent with prior reports demonstrating CIH did not impact spatial learning and memory in gonadally intact female mice [59].However, in contrast to prior studies showing CIH impairment of spatial learning and memory in male rats [47,[54][55][56], we did not observe any effects of CIH on spatial learning and memory in males.Studies routinely report males exposed to CIH exhibit increased anxiety-like behaviors [52,57,58], which was not observed in this study.It should be noted that these prior studies in which CIH induced spatial learning and memory impairments and anxiety-like behaviors were conducted with different CIH protocols (AHI ranged from 40-60) [47,52,[54][55][56][57][58] compared to our CIH protocol (AHI = 10).Indeed, a study using a CIH protocol with an AHI of 20 found no effect of CIH on spatial learning and memory in male rats [110], and another study using an AHI of 15 found no effects on anxiety-like behavior in the open eld in male and female mice [59].These ndings indicate that CIH protocols with > 20 AHI is necessary to induce spatial learning and memory impairments and anxiety-like behaviors.
We also observed sex differences in learning and motor behaviors exhibited by our normoxic rats.
Consistent to prior reports showing females exhibit greater recollective memory than males [111][112][113], we also found that females exhibited better recollective memory (shorter latency to novel object) than males.In addition to better recollective memory, females exhibited greater spatial learning ability (lower WLI scores) than males.Prior studies have typically shown either male biases [71,111,114] or no sex differences in spatial learning via Morris water maze [114,115], which is in contrast to our ndings.
Notably, these studies did not use a WLI [71,111,114] or weighted both males and females against male control values [115].As our WLI scores were weighted against within-sex and within-experiment control values, this may have increased the sensitivity of our measure to detect previously unobserved sex differences [88,89].Neither unweighted LI scores nor latency to platform during the probe trial detected signi cant sex differences, further underscoring the importance of using WLI scores as a measure of Morris water maze spatial learning ability [88,89].In addition, our Morris water maze protocol includes a pre-training day on day 1, which has been previously observed to improve female performance [116,117].
Consistent with prior reports, female rats showed greater locomotor activity than male rats in a large novel open eld arena (60.96 x 60.9 x 38.1 cm) and no sex differences in a smaller novel open eld (40.64 x 40.64 x 38.1 cm) [46,118].Unlike these ndings, we did not observe any sex differences in marble burying behavior in normoxic rats.The data on sex differences in marble burying is equivocal, with some groups nding either no sex differences [119,120], a female bias [121], or a male bias [71,122].

Limitations
Although our study has many strengths ranging from examining a broad spectrum of circulating markers of in ammation and OS and multiple behavioral domains, there are some limitations.Due to limited plasma, we were unable to perform analysis of cytokines in normoxic rats treated with MT.Additionally, we did not assess estrous cycle in our female rats.It is probable that the variability in the female data is related to estrous cycle, as many of the outcomes measured in our study could be in uenced by estrogen status [123][124][125].However, we did not want to introduce a stress response by conducting vaginal smears [126], nor introduce a variable that could not be conducted on the male rats.Lastly, we did not determine the brain concentrations of MT.However, our data showing behavior responses to MT, data showing that MT has similar membrane permeability to TEMPOL [68], and data that TEMPOL penetrates the bloodbrain-barrier [67] indicates that MT has actions at the level of the brain.

Perspectives and signi cance
Our data indicates that mild CIH (AHI = 10) can have signi cant impacts on multiple domains, such as circulating OS and in ammation, recollective memory, ne motor function and compulsivity.Furthermore, most of these CIH effects are sex-and mtOS-dependent with the exception of recollective memory impairment.This is of concern, as patients with sleep apnea with AHIs < 15 are classi ed as having mild OSA [14,16,18], and therefore may not receive treatment due to fewer reported quality of life impairments [4,127,128].This lack of treatment for mild OSA can affect women more than men, as women are more likely to present with mild OSA or be underdiagnosed with OSA [4,5].We also show that many of the sexdependent effects of CIH are mediated through mtOS, which may be a plausible target for therapeutics aimed at 15-40% of OSA patients that are unable to be effectively treated with CPAP machines [20][21][22].

Declarations
Ethics approval: This study was performed under IACUC protocol approved by the University of North Texas Health Science Center.

Osmotic
Pump Model 2002, Durect Corporation, Cupertino, CA) implanted subcutaneously between the scapula [60].Minipumps contained either 0.9% saline vehicle or MitoTEMPOL (0.7 mg/kg/day; Caymen Chemical Company, Ann Arbor, MI; MT) dissolved in saline.MT combines the antioxidant moiety TEMPOL, with a lipophilic cation triphenylphosphonium [61, 62].Triphenylphosphonium increases mitochondrial aggregation of MT by several hundred-fold over TEMPOL alone [62-64].MT has been observed to reduce mtOS in vitro and in vivo [64, 65].Drugs like MT are easily administered with multiple biologically active routes of administration and high blood-brain barrier permeability [66-68].The dose of MT was chosen based on previous in vivo studies

Figures Figure 1
Figures

Figure 2 Impact
Figure 2

Table 1
In ammatory cytokines in vehicle rats Plasma concentrations for pro-in ammatory cytokines TNF-α and IL-1β and anti-in ammatory cytokines IL-10 and IL-4.All cytokines plasma concentrations are reported in pg/mL.All values presented as mean ± SD.Data for in ammatory cytokines was log transformed for analysis.Analyzed by 2-way ANOVA with Fisher's LSD multiple comparisons tests, n = 5-6/group.CIH: Chronic intermittent hypoxia.

Table 3
Pathlength (cm) for rats to nd the target location during the probe trial.All values presented as mean ± SD.Analyzed by 3-way ANOVA with Fisher's LSD multiple comparisons tests, n=5-8/group.CIH: Chronic intermittent hypoxia; MT: MitoTEMPOL.CIH impaired ne motor function in female rats that was reduced by blocking mitochondrial oxidative stress.

Table 4
Fine motor behavior in modi ed open eld duration of entries within the center of the large open eld.No effects of CIH or sex were observed on anxiety-like behaviors -center entries or center duration (Table5).However, inhibition of mtOS did decrease open eld anxiety-like behaviors, as evidenced by decreased center entries (F 1, Rearing behavior and footfalls in an open eld arena (40.64 x 40.64 x 38.1 cm) modi ed with a 2 cm elevated wire mesh platform.Rears were categorized as assisted if the rat used forelimbs to brace against side of open eld arena.All values presented as mean ± SD.Data for unassisted rears was square root transformed for analysis.Analyzed by 3-way ANOVA with Fisher's LSD multiple comparisons tests, n = 6-13/group.CIH: Chronic intermittent hypoxia; MT: MitoTEMPOL.No effects of sex or CIH on anxiety-like behavior in an open eld.We examined anxiety-like behaviors using a large novel open eld (60.96 x 60.9 x 38.1 cm) by quantifying the frequency and

Table 5
Inhibiting mitochondrial oxidative stress increased anxiety-like behavior in an open eld arena Inhibiting mitochondrial oxidative stress reduced center entries and center duration (s) in an open eld arena (60.96 x 60.9 x 38.1 cm).No effects of CIH or sex were observed on center entries or center duration.All values presented as mean ± SD.Analyzed by 3-way ANOVA with Fisher's LSD multiple comparisons tests, n = 6-13/group.ANOVA signi cance for main effect of MT indicated by: ** = p < 0.01; Post-hoc signi cance indicated by unique letters (p ≤ 0.05).Shared letters are not signi cantly different.