Quantitative and morphological analyses reveal microglia-specific neuroinflammation in brain regions linked to cognitive function in ovariectomized and chronic intermittent hypoxia-exposed female rats

Abstract

Women affected by obstructive sleep apnea (OSA) face an increased risk of cognitive impairment and mood disorders, with emerging evidence suggesting that neuroinflammation plays a significant role in the pathophysiology. OSA causes intermittent hypoxia and activates the immune response in the brain. Additionally, menopause is a separate risk factor that can put women with OSA at an even higher risk of cognitive decline. Although neuroinflammation occurs in OSA, the underlying mechanisms of the neuroinflammatory response are not well understood, and it remains unknown whether ovarian hormones modify these mechanisms. To examine the impact of OSA and hormone status on neuroinflammation, we used a 7-day chronic intermittent hypoxia (CIH) and ovariectomy (OVX) to model OSA and hormone status in female rats, respectively. To examine neuroinflammation, we investigated changes in brain microglia and astrocyte morphology and the number of reactive cells. We used a comprehensive three-dimensional reconstruction and analysis of microglia and astrocytes to study the changes in the morphology of these cells. We focused on brain regions associated with cognitive function (CA1 of the dorsal hippocampus, medial prefrontal cortex – mPFC, caudate and putamen – CP). Specifically, immunofluorescence of Iba1 (microglia marker) and GFAP (astrocyte marker) was conducted, along with measuring indicators of glial reactivity (branching, complexity, cell size, and number of reactive cells). Our results found that there were CIH and hormone effects on neuroinflammation in these brain regions. Microglia activation was impacted by an interaction between CIH and hormone status in all regions examined. In contrast, astrocytes showed no reactivity in all regions examined, regardless of CIH or hormone status. These findings suggest that menopause and OSA may impact microglia remodeling in brain areas associated with cognitive function. Microglia-specific neuroinflammation may be part of early mechanisms that lead to the cognitive impairments observed in CIH and hormone loss in females.

Highlights

• •Chronic intermittent hypoxia (CIH) and ovarian hormone loss are associated with microglial activation in in female rats.
• •Astrocytes showed no reactivity to either CIH or hormone loss in all examined brain regions.
• •Hormonal status influenced the effects of CIH on microglia.
• •Microglia activation may contribute to cognitive impairments observed in CIH and hormone loss in female rats.
• •These findings highlight the importance of considering sex hormones in understanding neuroinflammation associated with OSA.

1. Introduction

Obstructive sleep apnea (OSA) affects nearly 1 billion people worldwide (Benjafield et al., 2019). Patients with OSA face an increased risk of cardiovascular, metabolic, and neurodegenerative disorders (Y. Gao et al., 2023; Guo et al., 2021; Yeh et al., 2016). The recurrent episodes of hypoxia and reoxygenation associated with OSA trigger oxidative stress and inflammation, which can result in neurological consequences, including cognitive impairments (Liu et al., 2020; Seda et al., 2021; Yang et al., 2022).

Biological sex plays a significant role in the occurrence and symptoms of OSA. OSA is more common in men and postmenopausal women than in premenopausal women. Postmenopausal women experience OSA at rates three to four times higher than premenopausal women (Peppard et al., 2013; Young et al., 2003). The observed sex disparity in OSA prevalence has often been attributed to the protective effects of ovarian hormones (Peppard et al., 2013; Young et al., 2003). However, premenopausal women with OSA commonly present with neurological symptoms, including depression, morning headaches, and mood disturbances, and are less likely to exhibit symptoms such as snoring and excessive daytime sleepiness, which are more frequently seen in men (Geer and Hilbert, 2021; Levartovsky et al., 2016; Nigro et al., 2018; Shepertycky et al., 2005). This distinctive presentation frequently results in misdiagnosis, underdiagnosis, and an underestimation of OSA cases among women, which obscures the significant cognitive risks they face. Studies indicate that women affected by OSA may be at an even higher risk of cognitive damage compared to men, and this risk seems to increase during menopause (Chang et al., 2013; Lal et al., 2016; Legault et al., 2021; Mosconi et al., 2018; Ramos et al., 2015). Moreover, in some studies in OSA patients, only women exhibited significant changes in brain regions involved in cognition, which included cortical thinning (Macey et al., 2018), changes in hippocampal volume (Macey et al., 2018), and decreased white matter integrity (Macey et al., 2012). Additionally, research has established a connection between premenopausal bilateral oophorectomy and mild cognitive impairment, suggesting that the loss of ovarian hormones alone can contribute to cognitive decline in women (Rocca et al., 2021). Despite this clinical pattern, scientific literature highlights a significant research gap in OSA-related cognitive impairment in females.

Most studies, including those investigating OSA using the established chronic intermittent hypoxia (CIH) animal model, which simulates the arterial hypoxemia associated with OSA, have primarily focused on male subjects (Becker et al., 2016; Kane et al., 2018; Mauvais-Jarvis et al., 2017). For example in male rats, CIH has been shown to induce inflammation and increase oxidative stress in brain regions associated with neurodegeneratvie disorders (Snyder et al., 2017). Although CIH has significant implications for brain health, its interaction with ovarian hormones and its relationship to neuroinflammation in females are still not fully understood. Therefore, further research is necessary to understand how different ovarian conditions influence the response to neuroinflammation in females exposed to CIH, and the mechanisms associated with these responses.

A recent study indicates that CIH induces inflammation and impairs cognitive and motor function in females (Mabry et al., 2024). Additionally, CIH dysregulates caspase-3 activity within critical memory-associated regions, such as the CA1 region of the hippocampus in females (Mabry et al., 2024, 2025). Early neuroinflammation is a potential mechanistic link between CIH and cognitive impairment in females (Yang et al., 2022), and loss of ovarian hormones may further worsen this neuroinflammation. Our previous study examining the impact of CIH on autonomic brain regions in female rats showed that CIH and OVX have separate effects on neural activity in the brain (Appiah et al., 2025). In terms of cognitive function, CIH exposure in INT and OVX females resulted in learning and memory deficits, but the mechanisms are not completely understood (Cheung et al., 2025; Mabry et al., 2024). Other studies also suggest that the depletion of ovarian hormones alone has detrimental effects on cognitive function (Bohm-Levine et al., 2020; Djiogue et al., 2018; Korol and Pisani, 2015; Rashidy-Pour et al., 2019; Wallace et al., 2006; Zhou et al., 2021). While neuroinflammation appears to mediate the effects of CIH and depleted ovarian hormones on cognitive function (Au et al., 2016), it is currently unknown how CIH and ovarian hormones interact to modulate neuroinflammation in cognitive brain regions in females, and this needs to be addressed.

Neuroinflammation is characterized by microglial activation and astrocyte reactivity, indicated by proliferation and morphological transformation of microglia and astrocytes (Althammer et al., 2020; Prinz et al., 2019; Zhao et al., 2024). During this process, microglia, which are the principal brain immune cells, change from a resting state characterized by thin, distal arborizations and small soma, to a reactive, proinflammatory state. Reactive microglia exhibit thickened, shortened processes and somatic hypertrophy (Althammer et al., 2020; Prinz et al., 2019). Similarly, reactive astrocytes are characterized by morphological alterations, including changes in branching and complexity (Schiweck et al., 2018; Zhao et al., 2024). Microglia and astrocytes lose their structural complexity, exhibit reduced branching (i.e., become deramified), and release proinflammatory cytokines in response to stress (Althammer et al., 2020; Prinz et al., 2019; Zappa Villar et al., 2018). Therefore, proliferative and morphological alterations serve as reliable indicators of neuroinflammation.

In the present study, we tested the effects of CIH and ovarian hormone depletion on neuroinflammation in key regions related to cognitive and motor function. We focused on the prelimbic, anterior cingulate (ACC) and infralimbic (IL) subregions of the medial prefrontal cortex (mPFC), as well as the CA1 region of the dorsal hippocampus, and caudate and putamen (CP) in female rats. The mPFC is responsible for executive functions such as working memory, decision-making, attention, and emotional regulation (Euston et al., 2012; Kidder et al., 2024; Mavrych et al., 2025). The CA1 region of the hippocampus plays a vital role in spatial learning and memory consolidation, especially in encoding and retrieving long-term episodic memories (Nakazawa et al., 2004). The CP, which are parts of the dorsal striatum, are important for motor control, habit formation, and the integration of cognitive and motor information (Balleine et al., 2007; Cox and Witten, 2019). We analyzed these regions because they are associated with the functions that are impaired in OSA, express estrogen and progesterone receptors, and are sensitive to both hypoxia and ovarian hormones (Almey et al., 2014; Beeson and Meitzen, 2023; Hart et al., 2007; Shansky et al., 2010; Sharma and Tan, 2007; Takahashi et al., 2024; Willett et al., 2020).

We hypothesized that exposure to 7 days of CIH would trigger neuroinflammation in female rats, as evidenced by microglial and astrocyte reactivity, and that this inflammatory response would be amplified in the absence of ovarian hormones. To test this hypothesis, we subjected INT and OVX female rats to either continuous normoxia (CON) or 7 days of CIH. We used a novel morphological analysis of microglia and astrocytes that combines Huygens object analysis and segmentation with ImageJ Simple Neurite Tracer (SNT) (Arshadi et al., 2021; Ferreira et al., 2014) to identify and characterize how neuroinflammatory processes influence glia activation in these regions.

2. Materials and methods

2.1. Animals

The adult female Sprague Dawley rats (250–300 g, 12 weeks old) used in this study were purchased from Charles River Laboratories (Wilmington, MA). All rats were housed in a temperature-regulated room and maintained on a 12:12-h light-dark cycle with unlimited access to chow and water. All animal procedures received approval from the Institutional Animal Care and Use Committee (IACUC) at the University of North Texas Health Fort Worth (IACUC Protocol # 2024-0017). These procedures were carried out following the National Institutes of Health's Guide for the Care and Use of Laboratory Animals, as well as the ARRIVE guidelines. Rats were randomly assigned to one of the following treatment groups: Gonadally intact normoxia control (INT CON), gonadally intact chronic intermittent hypoxia-exposed (INT CIH), ovariectomized normoxia control (OVX CON), and ovariectomized chronic intermittent hypoxia-exposed (OVX CIH). Each group included three to five nulliparous female rats. The experimental timeline is illustrated in Fig. 1. The brain sections used for these studies came from a subset of rats used in a previously published study (Appiah et al., 2025).

Fig. 1.

The schematic shows the experimental timeline of the study. Rats were allowed to acclimate for a week upon arrival (day 1). After acclimation, rats were subjected to either bilateral ovariectomy or remained gonadally intact. Rats were subsequently exposed to 7-day chronic intermittent hypoxia (CIH) or continuous normoxia. All rats were then euthanized on the morning of the 8th day following the 7-day CIH protocol. The brains were removed and processed for immunohistochemistry and glial analysis. The figure was created with biorender.com.

2.2. Bilateral ovariectomy

Rats in the OVX groups were bilaterally ovariectomized as previously described (Appiah et al., 2025). Ovariectomy was performed at approximately 13 weeks of age. The surgical procedure began with the careful anesthesia of the rats using a mixture of 2–3 % isoflurane and 100 % oxygen, ensuring their comfort and safety throughout the process. The flanks of the rats were shaved and cleaned using Betadine and 70 % ethanol. Surgical incisions were made to reach the retroperitoneal area. The ovaries were located, and a hemostat was used to clamp them at the distal end of the uterine horn. The ovaries were then removed, and the remaining oviducts and uterine horns were cauterized to minimize bleeding before being placed back into the retroperitoneal cavity. The flank incisions were subsequently closed using absorbable sutures. Three weeks post-surgery, these rats were relocated to plexiglass chambers (Oxycycler model A42OC, BioSpherix, NY, USA) for habituation before exposure to the 7-day CIH protocol.

2.3. CIH protocol

INT and OVX rats were exposed to 7 days of CIH or room air as previously described (Appiah et al., 2025; Marciante et al., 2020; Shell et al., 2019). During the CIH exposure, the oxygen concentration was cycled from 3 min normoxia (21 % O2) to 3 min hypoxia (10 % O2) for 8 h (0800–1600 h) during the light phase (0700–1900 h) using an Oxycycler (BioSpherix, model A42OC, NY, USA), which regulates the infusion of nitrogen and oxygen gases into the chambers. For the remaining 16 h (1600 - 0800 h), normoxic room air was continuously injected into the chamber. The rats underwent CIH exposure for a total of 7 days, while control (CON) rats were continuously exposed to 21 % O₂ and kept in the same room, thereby exposing them to the same ambient conditions as the CIH rats. On the 8th day, following the 7-day CIH protocol, all rats were euthanized, and the brains were collected for analysis.

2.4. Immunohistochemistry

All INT and OVX rats were anesthetized with Inactin (100 mg/kg, administered intraperitoneally, Sigma, USA) and then transcardially perfused with a 0.1M phosphate-buffered saline (PBS) solution. This was followed by a perfusion of 4 % paraformaldehyde in PBS. This procedure took place on the morning of the eighth day after a 7-day exposure to either CIH or continuous normoxia (CON). Three separate sets of 40 μm coronal serial sections were cut at −20 °C using a cryostat and then cryopreserved at the same temperature for later immunohistochemical processing. One set of sectioned brains was processed for Ionized Calcium Binding Adaptor Molecule 1 (IBA1) and Glial Fibrillary Acidic Protein (GFAP) immunohistochemistry. IBA1 and GFAP are markers for activated microglia and reactive astrocytes, respectively (Bennett and Viaene, 2021; Gao et al., 2014; Zemtsova et al., 2011). The regions analyzed for staining included the mPFC, CP, and the CA1 region of the dorsal hippocampus.

Free-floating sections were stained for IBA1 and GFAP using goat anti-IBA1 (Abcam, ab5076, dilution 1:500) and mouse anti-GFAP (Sigma, dilution 1:500) primary antibodies, respectively. First, the sections were washed in PBS. They were then incubated with the primary antibodies diluted in PBS solution containing 0.3 % Triton-X and 2.5 % normal horse serum for 2 days at 4 °C. Following this incubation period, donkey anti-goat Alexa Fluor 488 (Abcam, ab150129, 1:1000 dilution) and Cy3-conjugated donkey anti-mouse IgG (1:1,000, Cat. No. 715-165-151, Jackson ImmunoResearch Laboratories, PA) were employed to visualize IBA1 and GFAP, respectively. The sections were placed on slides coated with gelatin, and Prolong Diamond Antifade mountant (Ref. No: P36961, Thermo Fisher Scientific, USA) was applied before coverslipping. The slides were allowed to dry for 2 days before imaging.

2.5. Microscopy and image analysis

2.5.1. Widefield microscopy

Images of each region of interest were captured using an Olympus BX41 microscope, and an Olympus DP70 digital camera system with DP manager software version 3.3.1.222 and DP controller version 3.3.1.992 (Olympus Corporation) as previously described (Appiah et al., 2025). For all photomicrographs, consistent microscope settings were employed. The images were uniformly adjusted for brightness and contrast in ImageJ before further analysis.

The mPFC, subdivided into the anterior cingulate cortex (ACC), prelimbic and infralimbic subregions, was identified using the forceps minor of the corpus callosum and claustrum as neuroanatomical landmarks. The lateral ventricle and corpus callosum were used as landmarks for the caudate putamen, while the dorsal third ventricle was used as an additional anatomical guide for the CA1 region of the hippocampus. For each region, we obtained six images consisting of two bilateral images from three comparable sections across the groups for analysis.

The images were then processed in ImageJ to tally the counts of both IBA+ (green) and GFAP+ (red) cells. GFAP images were subtracted from the corresponding IBA1 images before counting was done on both the resulting image and the GFAP image to eliminate any crosstalk from the GFAP (red) channel on the IBA1 (green) channel. To minimize bias, cell counts for IBA+ and GFAP+ were conducted in a blinded manner. All IBA + cells within each photomicrograph were counted and averaged to determine the average number of microglia per section for that specific brain region. In contrast to the comprehensive photomicrograph counts for microglia, the analysis of GFAP images was approached differently. To avoid counting blood vessels and artifacts, a grid of 12 equal boxes was created over the image using ImageJ. Six boxes were then randomly selected from the grid. To be included in the analysis, the box could not include artifacts or blood vessels. The number of positively stained profiles in the boxes was manually counted and totaled for each section. The average number of positively stained profiles per section for each region was subsequently compiled and used for statistical analysis.

2.5.2. Confocal microscopy

A Zeiss confocal laser scanning microscope 880 (Carl Zeiss Microscopy GmbH) was used to capture images of the prelimbic mPFC, CP, and CA1 region of the hippocampus. Images were captured with C-Apochromat 40X/1.2 W objective with sampling rate of 99 nm on x, 99 nm on y and, 500–600 nm on z which is close to recommendations by Huygen's software for deconvolution, 3D reconstruction and object segmentation of microglia and astrocytes. 6 images (2 bilateral images per section, 3 sections) were obtained for analysis. The images collected were from similar areas for each region of interest, with consistent microscope settings. The raw z-stack images were loaded into Huygens Essential Software (version 24.04) for processing before morphometric analysis.

2.5.3. Huygens essentials – deconvolution, object segmentation and analysis

Optical distortions inherent in microscopy can result in blurring and a loss of contrast that obscure important details of an image. We performed Image deconvolution to enhance image resolution using theoretical point spread function from image metadata and manual input for background reduction to improve signal-to-noise ratio (SNR). Fig. 2 presents an image of microglia (Fig. 2A) showcasing spatial orientations of the cells at different tissue depths (Fig. 2B) and a detailed 3D projection of the selected microglia cell (Fig. 2C-F).

Fig. 2.

(A) Photomicrograph showing IBA1-stained microglia. (B) Same image in (A) rotated at a different twist and tilt angle to show the 3D projection of IBA1-stained microglia oriented in space at different depths within the tissue. (C) A magnified microglia cell. (D) 3D projection of the IBA1-stained microglia. (E) 3D projection of segmented microglia. (F) 3D projection of the traced segmented image. Scale bar (A and B) = 50 μm; (C–F) = 40 μm

The deconvolved images were processed using crosstalk correction and exported as individual channels with red (GFAP+) and green (IBA1+) 16-bit TIFF files. Consistent settings were used for each channel during the deconvolution, background removal, and crosstalk correction. Object segmentation was performed by thresholding, seeding, and applying a Gaussian filter. Each object was meticulously inspected to ensure that it corresponded accurately with the original image, guaranteeing a precise representation of all cellular features. Any object in contact with the x, y, or z borders was removed to ensure that only fully intact cells were analyzed. The segmented images were exported as 8-bit TIFF images for further analysis in ImageJ/FIJI.

2.5.4. Image J morphometric assessment

To further reduce noise and connect separated structures because of staining artifacts, 3D Gaussian blur filtering was applied. The pre-processed images were analyzed using SNT version 4.2.1, a plugin within ImageJ/Fiji designed for semi-automated tracing and analysis of neuronal and glial cells. Autotrace functionality was employed with the following parameters: (1) Maximum connection distance of 1.0 pixel: This parameter enables the algorithm to connect adjacent components that are separated by gaps of up to 1.0 pixel, compensating for minor discontinuities in fluorescence intensity typical in fine cellular processes. (2) Connect adjacent components option enabled: This ensures that closely positioned process segments are recognized as belonging to the same cellular structure. (3) Length threshold of 50 pixels: This threshold was implemented to automatically discard small components (<50 pixels in length) that likely represent noise or cellular debris.

Thinning iterations were performed during automated tracings to reduce the complex 3D cellular morphology to a skeleton representation. This step is important for accurate quantification of morphological parameters, as it standardizes all processes to a uniform width regardless of original fluorescence intensity variations. Fragments of processes, cells without visible soma, and fused cells that could not be physically separated were excluded from subsequent analyses to maintain data integrity and biological relevance. To ensure the accuracy of the traced representations, the following validation procedure was implemented: (a) Opening the corresponding original images, (b) Generating maximum intensity projections of the original z-stacks, (c) Overlaying the thinned traced cells with these maximum intensity projections, (d) Visually inspecting the overlay to confirm an accurate representation of cellular morphology, (e) Making manual corrections where necessary. These steps serve as crucial quality control, verifying that the automated tracing accurately represents the actual cellular structures visible in the original images.

Following the tracing, a comprehensive 3D analysis was conducted using a set of morphological parameters in SNT, which included metrics for branching and ramification (such as total number of branches, cable length, number of tips or terminal branches), size metrics (including convex hull: size), and complexity measure (fractal dimension). Upon completion of the analysis, all tracings were saved in SNT's native format (∗.traces), ensuring that the 3D coordinates of all traced points and the connectivity information between segments were preserved. The raw measurement data were exported to a spreadsheet format for subsequent statistical analysis and comparison between experimental groups.

2.6. Statistical analysis

Statistical analyses and graphing were performed using GraphPad Prism version 10.1.1 (GraphPad Software, San Diego, CA). We conducted a two-way analysis of variance (ANOVA) to evaluate the effects of the independent variables, CIH and hormonal status, and used Šídák's post-hoc test for follow-up analyses. In our statistical analyses of morphological parameters, we calculated the average value for each parameter across all cells from each individual animal. This method ensured that the final sample size, denoted as “n,” represented the total number of animals tested to normalize the data and account for differences in cell number. Subsequently, we used these average values for statistical analysis. A P value of less than 0.05 was considered statistically significant for all comparisons. All quantified data are reported as mean ± standard error of the mean (SEM).

3. Results

To determine glial responses to CIH exposure and ovarian hormone depletion in cognitive and motor-associated brain regions in female rats, we performed a quantitative analysis of IBA+ and GFAP + profiles and a detailed 3D assessment of microglia and astrocyte morphology. The various morphometric measures that were included in our analysis are summarized in Table 1. The 3D structural analysis was carefully carried out to ensure an accurate representation of the cells to evaluate the changes in branching, size, and complexity, which reflect glia reactivity and neuroinflammation in the brain, as shown in Fig. 3.

Table 1.

Summary of cell morphology measures.

METRIC	MEASURE	UNIT	INTEPRETATION
Cable length	Summed branch length	μm/cell	Cell ramification
Branches	Number of branches	Number/cell	Cell ramification
Branch points	Number of branch points/junctions	Number/cell	Cell ramification
Terminal branches	Number of terminal endpoints/tips	Number/cell	Cell ramification
Convex hull: size	Minimum size of the 3D polyhedron enclosing the cell	μm3/cell	Cell size
Branch fractal dimension	Summed branch fractal pattern detail per scale	Integer/fraction	Cell complexity

Table 1 shows morphological parameters, including definitions, measurement units, and biological interpretation, used to assess microglial and astrocyte reactivity. These metrics capture different aspects of glial cell structure, including ramification (cable length, branches, branch points, terminal branches), size (convex hull), and complexity (branch fractal dimension).

Fig. 3.

Image processing workflow showing (A) Raw 3D confocal image, converted into a z-stack, and (B) deconvolved to eliminate blurring. (C) The image was subjected to Huygens object analysis and segmentation for microglia. (D) The segmented image is overlaid on the IBA1-stained microglia. (E) Objects touching the borders were removed, followed by (F) Tracing of the cells using simple neurite tracing (SNT). (G) Overlay of traced image and deconvolved raw image to show that the processing did not alter the features of the cells in the raw image. Cells that do not have an overlay were excluded because they either touched the border (and were not complete cells) or were fused and could not be separated. Only microglia are illustrated in (B–G), although astrocytes underwent the same steps. Scale bar = 50 μm

3.1. Prelimbic medial prefrontal cortex: CIH and hormone status increase reactivity of microglia but not astrocytes

We assessed microglia IBA + profiles, branching, size, and complexity in the prelimbic mPFC (Fig. 4). We observed significant interactions between CIH and OVX on the number of reactive microglia. Compared with CON, CIH increased the number of IBA + profiles in INT (CIH x hormone: F(1, 11) = 5, P = 0.0461; Šídák's, P < 0.05) but not in OVX rats, as the OVX-induced elevation was not increased further by CIH (Fig. 4A & C). Two-way ANOVA analysis of microglial branching, size, and complexity, including cable length, branches, branch points, convex hull size, branch fractal dimension, and cell complexity index in the mPFC, revealed a significant effect of CIH (Fig. 4B & C).

Fig. 4.

(A) A brain schematic depicting the area identified as the prelimbic (PRL) cortex and representative images showing IBA1+ microglia in the PRL of gonadally intact normoxic control (INT CON), gonadally intact chronic intermittent hypoxia (INT CIH), ovariectomized control (OVX CON), and ovariectomized chronic intermittent hypoxia (OVX CIH). Scale bar: 200 μm. (B) Representative photomicrographs showing reactive (IBA1+) microglia in the PRL cortex and their respective overlay produced from simple neurite tracing (SNT) and raw microglia images. (C) presents data on IBA1+ microglia in the PRL cortex. Graphs show the number of IBA1 immunoreactive microglia and individual values (dot-plots) of microglial features in the PRL cortex, including cable length, the number of branches, the number of branch points, the number of terminal branches, convex hull size, and the summed branch fractal dimension. The data were collected from INT CON rats (n = 3 rats, 70 cells), INT CIH (n = 4 rats, 185 cells), OVX CON (n = 3 rats, 149 cells), and OVX CIH (n = 4 rats, 151 cells). The respective p-values for comparisons are indicated in the graphs. Data analysis was conducted using a two-way ANOVA, followed by Šídák's post hoc test. Cell morphometric measures were averaged per animal. Animal averages were used for statistical comparisons. fmi, forceps minor of the corpus callosum; Cl, Claustrum

In prelimbic mPFC, the treatments did not affect the number of reactive astrocytes (Fig. 5). Neither CIH nor hormone depletion was associated with morphological changes in astrocytes. (Fig. 5).

Fig. 5.

The graphs show the number of GFAP + cells and morphological measures of astrocytes in the prelimbic cortex. Individual values of astrocyte features in the prelimbic cortex, including cable length, number of branches number of branch points, number of terminal branches, convex hull size, and summed branch fractal dimension from INT CON rats (n = 3 rats, 162 cells), INT CIH (n = 4 rats, 235 cells), OVX CON (n = 3 rats, 152 cells), and OVX CIH (n = 4, 245 cells) are shown in the graphs. The p-values for comparisons are indicated in the graphs. Cell morphometric measures were averaged per animal. Actual animal averages were used for statistical analysis. Data were analyzed using a two-way ANOVA, followed by Šídák's post hoc test.

3.2. Anterior cingulate and infralimbic cortex: CIH and hormone status increase microglia reactivity

We observed that the ACC subregion exhibited an increase in the number of IBA1+ profiles in the INT but not OVX rats following exposure to CIH (CIH x hormone: F(1, 11) = 8.497, P = 0.0141, Šídák's, P < 0.05; Fig. 6A & B). OVX in normoxic conditions increased the number of IBA1+ microglia in the ACC region. However, this increase in OVX rats was not further enhanced by the addition of CIH. Similarly, an interaction between CIH and hormone status on the elevation of the number of reactive microglia was observed in the IL subregion (CIH x hormone: F(1, 11) = 5.116, P = 0.0449, Šídák's, P < 0.05; Fig. 6D & E). Specifically, CIH increased reactive microglia in INT rats. However, no effects of CIH were noted in OVX rats. Additionally, neither CIH nor OVX increased the number of GFAP + astrocytes in these two subregions (Fig. 6C & F).

Fig. 6.

(A) Brain scheme depicting the anterior cingulate cortex (ACC) region that was captured for analysis and representative images showing IBA1+ microglia in the same region in gonadally intact normoxic control (INT CON), gonadally intact chronic intermittent hypoxia (INT CIH), ovariectomized control (OVX CON), and ovariectomized chronic intermittent hypoxia (OVX CIH)-exposed females. Graphs (B) and (C) show the mean number of IBA1+ microglia and GFAP + astrocytes, respectively, in the ACC. (D) A brain schematic depicting the area identified as the infralimbic (IL) cortex and representative images showing IBA1+ microglia in the IL region of INT CON, INT CIH, OVX CON, and OVX CIH. Scale bar: 200 μm) Graphs (E) and (F) show the mean number of IBA1+ microglia and GFAP + astrocytes in the IL, respectively. The p-values for comparisons are indicated in the graphs. fmi, forceps minor of the corpus callosum; Cl, Claustrum. Data analysis was conducted using a two-way ANOVA, followed by Šídák's post hoc test.

3.3. CA1 region: CIH and hormone status increase activation of microglia but not astrocytes

Together, CIH and hormone status significantly increased IBA1+ microglia (CIH x hormone: F(1, 11) = 14.36, P = 0.0030; Fig. 7A & C) in dorsal CA1. Specifically, CIH increased activated microglia in the CA1 region of INT but not OVX females (Šídák's, P < 0.05). The interaction between CIH and hormone status was also associated with significant morphological alterations of microglia in the CA1 region. Post hoc analysis revealed that CIH significantly reduced microglial branching and complexity in INT but not OVX rats, as demonstrated by decreased number of branches, branch points, and branch fractal dimension (Fig. 7B & C). Interestingly, this pattern of decreased microglial branching and complexity in the INT + CIH group was reversed in the OVX rats exposed to CIH. Thus, in OVX + CIH rats, microglia appeared ramified, whereas in INT + CIH rats, they were deramified. Hence, CIH's effect on microglia structure was only observed in INT rats but not in OVX rats.

Fig. 7.

(A) A brain schematic depicting the area identified as the CA1 region of the hippocampus and representative images showing IBA1+ microglia in the same region. Scale bar: 200 μm. (B) Representative images showing IBA1 immunoreactive microglia in the CA1 region of the hippocampus and their respective overlay created from simple neurite tracing (SNT) and raw microglia images. (C) presents data on IBA + microglia in the CA1 region of the hippocampus. Graphs show the number of IBA1 immunoreactive microglia and the individual values of microglial features in the CA1, which include cable length, number of branches, number of branch points, number of terminal branches, convex hull size, and summed branch fractal dimension. The data were collected from INT CON rats (n = 3 rats, 53 cells), INT CIH (n = 4 rats, 129 cells), OVX CON (n = 3 rats, 108 cells), and OVX CIH (n = 4 rats, 145 cells). The morphometric measurements of cells were averaged for each animal. Animal averages were used for statistical analysis. The p-values for comparisons are shown in the graphs. Data were analyzed using a two-way ANOVA, followed by Šídák's post hoc test.

Neither CIH nor hormone status altered the number, branching, size, and complexity of reactive astrocytes in the CA1 region (Fig. 8).

Fig. 8.

Graphs show measures of GFAP + astrocytes in the CA1 region of the hippocampus. These include the number of GFAP + astrocytes, cable length, number of branches, number of branch points, number of terminal branches, convex hull size, and summed branch fractal dimension from INT CON rats (n = 3 rats, 193 cells), INT CIH (n = 4 rats, 240 cells), OVX CON (n = 3 rats, 195 cells), and OVX CIH (n = 4 rats, 266 cells). The morphometric measurements of cells were averaged for each animal and used for statistical analysis. The p-values for comparisons are indicated in the graphs. Data analysis was conducted using a two-way ANOVA, followed by Šídák's post hoc test.

Taken together, our results indicate that CIH and hormone depletion increase activated microglia in the CA1 region, but do not have an additive effect in the combined state. Deramified CA1 microglia were found only in CIH-exposed INT female rats. In contrast, astrocytes did not proliferate or undergo morphological changes under CIH or OVX conditions.

3.4. Caudate and putamen: CIH and OVX increased microglia activation but not astrocytes

CIH significantly increased the number of activated microglia in the CP region (F(1, 11) = 7.020, P = 0.0226; Fig. 9A & C). Two-way ANOVA analysis of microglial branching and complexity index in the CP revealed a significant interaction between CIH exposure and hormonal status (Fig. 9B & C). Post-hoc analyses demonstrated that CIH significantly reduced microglial cable length, number of branches, branch points, terminal branches, size, and branch fractal dimension in INT rats but not OVX rats. Only microglia size was affected by CIH alone. There were no significant differences in morphological parameters between the two OVX rat groups (Fig. 9B & C).

Fig. 9.

(A) A brain schematic depicting the area identified as the caudate and putamen (CP) and representative images showing IBA1+ microglia in the CP of gonadally intact normoxic control (INT CON), gonadally intact chronic intermittent hypoxia (INT CIH), ovariectomized control (OVX CON), and ovariectomized chronic intermittent hypoxia (OVX CIH) Scale bar: 200 μm. (B) Representative photomicrographs illustrating IBA1 immunoreactive microglia in the CP, along with their corresponding overlays created using simple neurite tracing (SNT) and raw microglial images. (C) Graphs show the number of IBA1 immunoreactive microglia and individual values of microglial features in the CP, including cable length, the number of branches, the number of branch points, the number of terminal branches, convex hull size, and the summed branch fractal dimension. The data were collected from INT CON rats (n = 3 rats, 44 cells), INT CIH (n = 4 rats, 171 cells), OVX CON (n = 3 rats, 112 cells), and OVX CIH (n = 4 rats, 148 cells). The morphometric measurements of cells were averaged for each animal. Animal averages were used for statistical analysis. The p-values for comparisons are indicated in the graphs. Data analysis was conducted using a two-way ANOVA, followed by Šídák's post hoc test.

The reactive astrocytes did not exhibit proliferation or morphological changes in the CP region (Fig. 10). Thus, no significant differences were observed in astrocyte cable length, number of branches, branch points, terminal branches, and branch fractal dimension among the groups.

Fig. 10.

Graphs show the number of GFAP + astrocytes and morphological measures in the caudate and putamen (CP). Individual values of astrocyte features in the CP, including cable length, number of branches number of branch points, number of terminal branches, convex hull size, and summed branch fractal dimension from INT CON rats (n = 3, 140 cells), INT CIH (n = 4, 220 cells), OVX CON (n = 3, 127 cells), and OVX CIH (n = 4, 215 cells) are depicted in the graphs. The morphometric measurements of cells were averaged for each animal and used for statistical analysis. The p-values for comparisons are indicated in the graphs. Data were analyzed using a two-way ANOVA, followed by Šídák's post hoc test.

4. Discussion

Currently, not much is known about the effects of OSA in women in general, much less in women with different hormonal statuses, such as in menopause or bilateral oophorectomy. We therefore examined the effects of CIH and hormone status on neuroinflammation (glia alterations) in the mPFC, CA1 region of the hippocampus, and the CP.

The main findings of this study are that CIH and the absence of ovarian hormones were associated with the activation of microglia in all of the regions examined. However, when these treatments were combined, there were no additive or synergistic effects. This suggests that CIH and hormone loss may influence the same inflammatory pathways, such as those related to oxidative stress (Kiernan et al., 2016; Yao et al., 2012; Yin et al., 2015), decreased protective factors such as brain-derived neurotrophic factor (BDNF) (Rashidy-Pour et al., 2019; Yin et al., 2015), or involve NF-κB and toll-like receptor signaling pathways (Liu et al., 2018; Smith et al., 2013; Xu et al., 2016). This overlap may explain the absence of additional effects. This pattern contrasts with astrocyte responses. GFAP staining and morphological analyses showed no astrocyte reactivity to either CIH or hormone loss, indicating astrocytic resilience in the early central nervous system (CNS) responses to these conditions. CIH does not alter ovarian hormones in INT females, nor does it affect the plasma ovarian hormone concentrations (Appiah et al., 2025). Therefore, effects resulting from hormone loss are due to OVX only. Our findings align with previous research indicating that the loss of ovarian hormones in female rats results in increased activated microglia in various brain regions, including the hypothalamus (Baek et al., 2024; Yang et al., 2020), prefrontal cortex (Baek et al., 2024), and the hippocampus (Baek et al., 2024; Sanchez et al., 2023). This confirms that the depletion of ovarian hormones strongly stimulates microglia activation in the brain.

Despite the consistent increase in microglial number across regions, morphological responses varied in a region- and hormone-dependent manner. In the mPFC, we selected the prelimbic cortex for detailed morphometric analysis based on several reasons. First, this subregion is vital for executive functions, working memory, and cognitive flexibility, which are particularly affected by SA-related cognitive impairments (Krysta et al., 2017; Marchi et al., 2024). Our data regarding IBA1+ profiles indicated consistent patterns across all mPFC subregions, making the prelimbic cortex the most relevant for understanding these dysfunctions. Second, analyzing a single subregion is necessary due to the significant methodological demands. Lastly, this focused approach allows for more robust statistical analyses with larger sample sizes, enabling the detection of subtle changes in glial morphology. The depth of analysis achievable with this method provides more definitive conclusions about neuroinflammatory processes compared to a broader, shallower analysis across all subregions.

In the prelimbic cortex, CIH had a consistent and significant decrease in ramification (deramification), size, and complexity of microglia independent of the rats' hormonal status. The reduction in cell size resulted from the decreased branching and retraction of microglia processes, indicating an activated state. CIH can damage neurons through oxidative stress and inflammation, activating microglia, which amplify the oxidative stress in the brain, and promote neuronal apoptosis (Appiah et al., 2025; Kiernan et al., 2016). Ovariectomy also showed a trend in shifting microglia to a reactive morphology. Reactive microglia become deramified and less complex during CNS injury (C. Gao et al., 2023; Kolliker-Frers et al., 2021; Morrison et al., 2017; Vidal-Itriago et al., 2022). This change boosts phagocytosis of dead and damaged cells and enables them to release pro-inflammatory cytokines, which act as signaling molecules and help recruit and activate other microglia and astrocytes to protect the CNS from further damage, allowing repairs to take place (C. Gao et al., 2023; Kolliker-Frers et al., 2021; Morrison et al., 2017; Vidal-Itriago et al., 2022). However, when microglia are persistently activated by harmful stimuli, they remain in a pro-inflammatory state, leading to excessive production of reactive oxygen species, elevated levels of pro-inflammatory enzymes, and increased release of pro-inflammatory cytokines that promote cell death and loss of neurons that may contribute to dysfunctions in learning and memory (Appiah et al., 2025; C. Gao et al., 2023; Kolliker-Frers et al., 2021; Morrison et al., 2017; Vidal-Itriago et al., 2022). In this study, microglia activation was observed in both INT and OVX rats, which relates to previous findings showing that CIH negatively impacts learning and memory in both INT and OVX females (Cheung et al., 2025; Mabry et al., 2024). Therefore, CIH-induced microglia alterations could be a potential contributor to impaired cognitive function in both INT and OVX females.

Our observation that hormonal depletion tended to shift microglia towards deramification, a less complex morphology, and reduced size under normal oxygen conditions has potential connections with previous studies that have shown that OVX rats have significant impairments in learning and working memory (Bohm-Levine et al., 2020; Rashidy-Pour et al., 2019; Zeibich et al., 2021; Zhou et al., 2021). Hormone loss increases activated microglia and primes them towards a deramified proinflammatory state. Our observations of changes in microglia in hormone-depleted rats under normoxia suggest that menopause itself may serve as a risk factor for neuroinflammation in the mPFC. The CA1 region of the hippocampus showed notable hormone-dependent differences in microglial morphology. INT females exposed to CIH exhibited a typical neuroinflammatory response, characterized by microglia that were deramified, smaller in size, and less complex. The presence of ovarian hormones did not protect microglia in CA1 of INT rats from the deramifying effects of CIH. Surprisingly, in the absence of ovarian hormones, microglia morphology was not altered and remained ramified, despite their proliferation. The ramified microglia in the OVX rats could suggest a compensatory state. Ramified microglia have demonstrated neuroprotective capacity against excitotoxic damage in the hippocampus (Vinet et al., 2012), suggesting that maintaining ramified morphology could limit proinflammatory cytokine release and subsequent neuronal damage. Additionally, the intriguing responses of CA1 microglia in OVX rats could indicate a different stage of the inflammatory response. Since the hormone depletion in OVX rats began long before they were exposed to CIH, the OVX group may be experiencing inflammatory stages at a different rate in this region compared to the INT + CIH females who experienced CIH as an acute insult. This difference in the timing of insult presentation may have placed these groups at distinct stages of the neuroinflammatory cascade at the 7-day CIH timepoint. The OVX group may have undergone a gradual shift in microglial phenotype toward a ramified protective state over time. Thus, hormone loss may be protective, preventing microglial overactivation. Whether this ramified morphology represents a beneficial adaptation or a maladaptive, impaired inflammatory response remains an important question for future investigation. In the CP, microglial morphological changes mirrored the mPFC and CA1 pattern in INT + CIH females. However, these responses were blunted in OVX animals, suggesting modulation of inflammatory responses in the CP by hormonal status. The CP region has received comparatively little attention in OSA research despite its fundamental role in motor control and accumulating evidence that CIH can exacerbate motor dysfunction, especially in movement disorder models (Kim et al., 2013; Mabry et al., 2024; Oh et al., 2023). The present findings highlight the CP as a potentially vulnerable region in CIH. Changes to microglia in the CP induced by CIH may contribute to these motor impairments. The potential role of microglia in the CP and the link between OSA and worsening neurodegenerative movement disorders highlight the need for focused investigations into the CP.

Across all regions, astrocytes remained unaltered in the number of reactive cells or morphology by either CIH or hormone status.This indicates that microglia are the primary mediators of neuroinflammatory responses in these regions. In contrast, astrocytes may be more resilient and may require more intense or prolonged stress to undergo significant microglia-astrocyte crosstalk and morphological changes (Pereyra et al., 2025; Wu et al., 2025). Alternatively, they might contribute to pathology through non-morphological mechanisms, such as altered metabolic processes or neurotransmitter regulation (Jia et al., 2022; Martinez et al., 2020), which were not investigated in this study. Overall, these results emphasize a microglia-driven, cell-type-specific neuroinflammatory response to CIH and hormone loss.

The regional pattern of microglial activation in this study may be facilitated by anatomical connectivity among examined regions. The CA1 projects directly to the mPFC (Aleman-Andrade et al., 2025; Ye et al., 2017), and the mPFC in turn projects to the CP (Mesa et al., 2022). Microglial activation in one region might influence connected brain regions through altered neuronal activity patterns or trans-synaptic inflammatory signaling, although demonstrating such mechanisms requires future investigation. Previous studies in CIH and heart failure models have demonstrated increased angiotensin II release and sensitivity in the brain, as well as the contribution of AT1 receptors to neuroinflammation in males (Althammer et al., 2023; Farmer and Cunningham, 2024). We speculate that angiotensin II might contribute to the effects of CIH in our study, requiring future studies.

Our study has implications for the health of women with OSA and/or menopause. Loss of ovarian hormones during menopause may increase microglia activation and initiate deramification that may lead to impairments in cognitive functions (He et al., 2021). Deficits in executive functions, learning, and memory may help explain symptoms such as “brain fog,” trouble concentrating, and increased forgetfulness, often reported and observed in menopausal women (Conde et al., 2021; Georgakis et al., 2019; Weber et al., 2012). Our findings suggest that OSA acts as a potent neuroinflammatory stimulus that affects women even in the presence of hormones, challenging the notion that premenopausal women are protected from the neurological consequences of OSA due to their hormonal status. Finally, our findings suggest that in the combined state, the effects of mild OSA and hormone loss on glia alterations may not be additive.

4.1. Strengths and limitations

This study presents several notable strengths, including the development of a novel, detailed, and unbiased approach to glial morphological analysis. To the best of our knowledge, this is the first time Huygens image analysis has been combined with SNT for a comprehensive assessment of microglia and astrocytes. This innovative analytic approach allowed us to identify distinct patterns of reactivity in specific brain regions. Furthermore, our extensive evaluation of brain regions associated with learning, memory, and motor function provides insights into region-specific neuroinflammatory responses that correlate with OSA symptoms that are more common in women. However, important limitations must be acknowledged. We used a CIH protocol with an apnea-hypopnea index of 10, indicating the frequency of apneic episodes per hour, which characterizes mild OSA, a more common form of the condition (Cunningham et al., 2021; Schiza and Bouloukaki, 2020). While CIH can induce some of the pathological effects of OSA, it does not fully replicate all the features of OSA, such as changes in negative intrathoracic pressure, hypercapnia, and airway obstruction (Dempsey et al., 2010; Lv et al., 2023). The estrous cycle phase was not controlled in INT animals. We note that this represents a limitation, as hormonal fluctuations across the estrous cycle could introduce variability in the INT group. Our study also focused on the neuroimmune mechanisms related to glial responses within specific brain regions associated with cognition and behavior. However, the absence of behavioral data in our study restricts our ability to establish causal relationships between specific morphological changes and cognitive dysfunction. Again, our single-timepoint analysis, conducted seven days post-CIH exposure, offers a valuable snapshot of early neuroinflammatory responses but does not capture the dynamic nature of glial activation and potential recovery phases. Addressing this would require studies that involve multiple timepoints to determine whether the observed changes represent transient responses or sustained states. Although microglia activation in pathogenic CIH, as used in our study, leads to a persistent proinflammatory microglia phenotype in the cortex and dorsal hippocampus (Sapin et al., 2015; Smith et al., 2013), our study did not include measures of proinflammatory cytokines nor did it demonstrate whether the altered morphology leads to changes in microglial functions, such as synaptic remodeling, phagocytosis, and trophic support, which contribute to neuronal function and health (Bennett et al., 2018; Frost and Schafer, 2016; Prinz et al., 2019). It is possible that, in addition to the increased proinflammatory cytokine release, changes in these supportive functions due to morphological alterations can also contribute to neuronal damage. However, this requires further studies. Our study did not include additional markers for glia proliferation (e.g., Ki67, BrdU) or other markers of activation. IBA1 is expressed in both M1 (proinflammatory) and M2 (anti-inflammatory) microglia. While deramification, decreased cell size, and complexity are generally associated with M1-like activation, future studies incorporating additional markers (e.g., CD86, Arg1) are needed to differentiate polarization states. Our study utilized relatively small sample sizes (n = 3–5 animals per group), which, while sufficient for detecting the robust morphological changes we observed, may limit statistical power for detecting subtle effects. Finally, the use of bilateral ovariectomy to model hormone depletion more closely represents surgical menopause rather than natural menopause. Natural menopause involves a gradual decline in hormone levels, while OVX leads to an abrupt loss of hormones (Pillay and Manyonda, 2022; Page et al., 2024; Xing and Kirby, 2024), as shown in our previous study (Appiah et al., 2025). Therefore, future research will be needed to address these limitations through: [1] longitudinal studies incorporating multiple timepoints and varying severities of CIH, [2] integration of morphological, biochemical, and behavioral assessments, [3] inclusion of additional specific markers for activation such as excitatory amino acid transporter 2 (EAAT2) and [4] comparative studies between surgical and natural menopause models.

5. Conclusion

Our study shows that the female rats demonstrate microglia-specific neuroinflammatory responses to ovarian hormone depletion and CIH. The findings suggest that women experiencing OSA or going through menopause may be facing unrecognized neuroinflammation, which could accelerate cognitive aging and heighten the risk of dementia. Further research is needed to fully understand the mechanisms behind cognitive impairments and how biological sex contributes to these mechanisms. Overall, our study provides evidence of alterations in microglia in response to ovarian hormone depletion and CIH. These alterations may play a role in the early development of cognitive and motor impairments in females and lay the groundwork for future studies investigating neuroinflammation in other brain regions related to the pathology induced by CIH (Appiah et al., 2025).

CRediT authorship contribution statement

Cephas B. Appiah: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Kishor Kunwar: Writing – review & editing, Software, Methodology, Formal analysis, Data curation, Conceptualization. Rebecca L. Cunningham: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. J. Thomas Cunningham: Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Funding

This work was supported by NIH grants R01 HL155977 to JTC and R01 AG085296 to RLC.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: J. Thomas Cunningham reports financial support was provided by National Heart Lung and Blood Institute. Rebecca L. Cunningham reports financial support was provided by National Institute on Aging. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.