Obesity disrupts ovarian hemodynamics during the preovulatory and luteal phases in mice†.

Both AvY and HFD groups weighed significantly more and had higher adiposity than normal-weight controls ( Figure 2A and 2B ). The increased adiposity can be visualized in Figure 1B , where obese groups had increased POAT size which pushed the kidney further to the left of the images. Obese mice from both groups had reduced ovulation rates compared to normal-weight mice ( Figure 2C ). There was no difference in ovulation rate between the AvY and HFD groups. Figure 3A presents raw blood flow velocity measurements to compare numerical differences between groups at each preovulatory time point, showing how obesity affects ovarian blood flow at specific stages. Figure 3B shows velocity fold-change relative to baseline (0 h post-hCG), emphasizing the pattern of change in response to hCG within each group. This normalization reduces baseline variability linked to overall physiological differences, enabling clearer interpretation of temporal trends. Thus, all panels in Figure 3B illustrate within-group changes over time. In the OA ( Figure 3A (i) ), at 1 h post-hCG, blood velocity in normal-weight mice trended lower than in HFD mice ( p = 0.094) but showed no difference compared to AvY mice ( p = 0.63). However, HFD mice exhibited higher blood velocity compared to AvY mice ( p = 0.024). At 11 h post-hCG, velocity in normal-weight mice trended lower than in HFD mice ( p = 0.093), but not AvY ( p = 0.15) ( Figure 3A (i) ). Finally, at 24 h post-hCG, normal-weight mice displayed lower ovarian arterial blood velocity than HFD ( p = 0.0002) and AvY mice ( p = 0.008) ( Figure 3A (i) ). In the MV ( Figure 3A (ii) ), at 0 h post-hCG, there was no difference in velocity between normal-weight and AvY mice, although velocity trended higher in HFD mice compared to both groups ( p = 0.072 vs. control; p = 0.057 vs. AvY). At 1 h post-hCG, normal-weight mice exhibited higher flow velocity in MV than AvY mice ( p = 0.001) and a trend of higher velocity than HFD mice ( p = 0.085) ( Figure 3A (ii) ). In the CV ( Figure 3A (iii) ), at 1 h post-hCG velocity was higher in normal-weight controls compared to both AvY ( p = 0.0001) and HFD mice ( p = 0.006) ( Figure 3A (iii) ). No differences in velocity were observed between groups until 24 h post-hCG, when HFD mice showed higher velocity than both normal-weight ( p = 0.0079) and AvY mice ( p = 0.0004). Figure 3B (i) shows velocity fold-change in OA at each time point compared to 0 h post-hCG. In normal-weight mice, a trending increase in velocity fold-change was observed between 0 h and 1 h post-hCG ( p = 0.083), while no change was detected in both obese groups during this period ( Figure 3B (i) ). The fold-change in velocity trended lower at 8 h post-hCG compared to 1 h post-hCG in normal-weight mice ( p = 0.097), a pattern not observed in the obese mice. Velocity fold-change in normal-weight mice remained stable until 12 h post-hCG, at which point it was lower than at 1 h ( p = 0.035) ( Figure 3B (i) ). A similar pattern of velocity change was observed in MV. In normal-weight mice, the fold-change in flow velocity tended to increase between 0 h and 1 h post-hCG ( p = 0.064), followed by a significant decline between 1 h and 8 h ( p = 0.009), 11 h ( p = 0.012), and 12 h post-hCG ( p = 0.002) ( Figure 3B (ii) ). Velocity also declined significantly between 0 h and 11 h ( p = 0.028) and 12 h ( p = 0.046) post-hCG. In contrast, the obese groups (HFD and AvY) showed no significant changes in MV velocity across the time course ( Figure 3B (ii) ). In the CV, a similar upward trend was observed between 0 h and 1 h post-hCG ( p = 0.067) in normal-weight controls, whereas AvY mice exhibited a significant decrease in velocity between 0 h and 12 h ( p = 0.0004) ( Figure 3B (iii) ). HFD mice did not display significant changes in CV velocity. Overall, normal-weight controls exhibited significant shifts in velocity over time including a consistent upward trend between 0 h and 1 h post-hCG in all vessel types. In contrast, obese mice exhibited little to no change in velocity fold-change over time, indicating a markedly more stable—and blunted—hemodynamic response. To investigate the potential contributors to changes in hemodynamics within ovarian vessels during the preovulatory phase, we measured the resistance index for each vessel type at various time points. The log fold-change in RI increased in normal-weight mice within the OA between 1 h and 8 h post-hCG, which was not observed in the obese groups ( Figure 3C (i) ). In the medullary vessels, HFD mice showed multiple differences in RI between time points (0 h, 1 h, 11 h, and 12 h) and 24 h post-hCG ( Figure 3C (ii) ), which were not observed in normal-weight or AvY mice. RI was not calculated for cortical vessels, as their velocity often fell below the detectable threshold, preventing an accurate assessment. We also measured OA diameter using color Doppler. Our results showed no change in the OA diameter fold-change across all groups and time points ( Figure S2 ). However, upon further analysis, we concluded that Doppler ultrasonography lacks the spatial resolution necessary to accurately measure the diameter of the OA, as discussed in more detail in the Discussion section. To assess the functional relevance of the rise in flow velocity from 0 h to 1 h (as observed in normal-weight mice in all three vessel types) in ovulation success as quantified by the rate of ovulation, we analyzed the correlation between these two biological processes with the fold-change now plotted on a linear scale ( Figure 4 ). In the OA, there was no correlation between 0 h–1 h velocity fold-change and ovulation rate ( Figure 4 (i) ); however, this correlation was strong ( R = 0.709, p = 0.0009) in MV ( Figure 4 (ii) ) and moderate ( R = 0.409, p = 0.046) in the CV ( Figure 4 (iii) ). On average, normal-weight mice exhibited the highest ovulation rates and the greatest fold-change in velocity. To illustrate the increase in ovarian perfusion between 0 h and 1 h post-hCG, we present Doppler images of ovarian blood at these time points for each experimental group ( Figure 5 ). At 0 h post-hCG, images by both color and power Doppler showed detectable perfusion across all groups ( Figure 5A ). Between 0 h and 1 h, normal-weight mice exhibited increased perfusion, evident from the enlarged OA, MV, and CV ( Figure 5B ). Power Doppler images—highlighting blood flow in yellow—also showed increased brightness throughout the ovary suggesting enhanced perfusion ( Figure 5A (ii) and 5B (ii) ). In contrast, AvY mice showed no change in vessel size or brightness (yellow signal), indicating stable perfusion ( Figure 5A (ii) and 5B (ii) ). High-fat-diet ovaries exhibited a modest increase in vessel size within the ovary, but minimal change in OA size or brightness ( Figure 5A (ii) and 5B (ii) ). These visual data support an increase in ovarian perfusion, particularly in the MV of normal-weight mice between 0 h and 1 h post-hCG, which is minimal in two groups of obese mice. To determine the impact of obesity on ovarian hemodynamics in physiological conditions, blood flow velocity was assessed in naturally cycling normal-weight controls, HFD, and AvY mice during the first and second days of diestrus and during estrus ( Figure 6A ), when cycle staging by vaginal smear is most reliable. In the OA, velocity was higher in AvY mice on the first day of diestrus compared to normal-weight controls ( p = 0.03). On the second day of diestrus, HFD mice showed increased OA blood flow velocity relative to controls ( p = 0.03). At estrus, OA velocity did not differ between groups. Additionally, no differences in velocity were observed in the MV or CV at any of the time points. Overall, when cycling normally without superovulation stimulation, blood flow velocity in OA is higher in obese groups than the normal-weight group during diestrus, echoing observations at 24 h post-hCG following superovulation ( Figure 3A (i) ). Resistive index was also measured in naturally cycling mice in the OA and MV ( Figure 6B ). While RI did not differ between groups in the OA ( Figure 6B (i) ), it was significantly increased in MV of AvY mice relative to normal-weight controls on the second day of diestrus ( p = 0.02), and relative to both controls ( p = 0.007) and HFD ( p = 0.002) mice during estrus ( Figure 6B (ii) ). Analysis of estrous cycles across experimental groups showed that AvY mice tended to exhibit a higher incidence of impaired cycles compared to controls ( p = 0.08; Figure S3A ). The proportion of time spent in each estrous stage over the 15-day period did not significantly differ between groups ( Figure S3B ). However, all AvY mice displayed irregular cycling patterns, and several HFD mice remained in diestrus for up to two consecutive weeks—an outcome not observed in controls ( Figure S3C (i – iii )). To assess whether and how obesity impacts the structure of ovarian vasculature during the preovulatory phase, ovaries with blood vessels labeled by Alexa 649–conjugated lectin were collected at 0 h post-hCG for whole-mount imaging ( Figure 7A ). To quantify the ovarian vasculature too small to be detected by Doppler ultrasonography, we performed Hessian Tubeness analysis, which identifies and quantitively characterizes tubular structures like blood vessels ( Figure 7B ). This analysis provides three key metrics of blood vessel structures: relative volume (the abundance of vessels of a certain size), average intensity (brightness), and total intensity (weighted measure of vascular brightness and abundance at each vessel size). The results revealed no differences between control and obese groups but did reveal an increase in the abundance of 5- μ m capillaries in HFD ovaries compared to AvY ovaries ( p = 0.02) ( Figure 7B ). To explore the cellular and molecular basis of impaired ovulation in obese mice, we performed bulk RNA-seq on isolated GCs and OST from normal-weight control and HFD mice ( Table S2 —differentially expressed genes (DEGs), Table S3 —DEGs within relevant biological processes). In GCs, while principal component analysis did not reveal clear clustering between experimental groups ( Figure S4 ), the heatmaps of DEGs showed distinct transcriptomic differences in both GCs and OST between HFD and normal-weight controls ( Figure 8A and 8B ). A total of 364 DEGs were identified in GCs, with 149 upregulated and 215 downregulated in the HFD group compared to the control group ( Figure 8C ). In OST, 635 DEGs were identified, with 259 upregulated and 376 downregulated in HFD mice ( Figure 8D ). Gene Ontology analysis of the upregulated DEGs in GCs revealed their involvement in pro-inflammatory processes, including the icosanoid and leukotriene metabolic processes ( Pla2g5 , Cyp4f18 , and Mgst2 ) and extracellular matrix (ECM) organization ( Mmp2 , Adamts4 , and Tnxb ) ( Figure 8E ). The downregulated biological processes in HFD GCs included cell–cell adhesion, which contained vascular function regulators such as cellular communication network factor 1 ( Ccn1 ), and thrombospondin 1 and 4 ( Thsb1 and Thsb4 ) ( Figure 8E ) [ 37 – 40 ]. Genes regulating vasculature development were also downregulated including Ccn1 , Thsb1 , semaphorin 5A ( Sema5a ), and Ras homolog gene family, member B ( Rhob ). This is in line with the disrupted vascular function observed in Figures 3 , 5 , and 6 . In OST, upregulated genes were linked to ECM organization and metabolic processes, including fatty acid metabolism ( Adh7 and Prxl2b ) [ 41 ]. Downregulated biological pathways included female gamete generation ( Gdf9 , Oog1 ) and negative regulation of innate immune response ( Mmp12 and Pparγ ). Several pathways involving ECM remodeling including muscle tissue development ( Ccn2 and Tnni3 ) and connective tissue development ( Bmp2 , Bmpr1b , and Cnmd ) were also downregulated ( Figure 8F ). We further examined the relative mRNA levels of DEGs within all three groups (normal-weight controls, HFD, and AvY) using RT-qPCR. In GCs ( Figure 8G ), consistent with bulk RNA-seq data, genes within the cell–cell adhesion (apolipoprotein A-IV ( Apoa4 )), positive regulation of miRNA transcription (fos proto-oncogene ( Fos )), vasculature development (early growth response 1 ( Egr1 ), and Ccn1 ), and leukotriene metabolic process pathways (period circadian regulator 2 ( Per2 ), also a key regulator of circadian rhythm) were dysregulated in HFD mice. Some of these markers ( Fos and Ccn1 ) were also disrupted in AvY mice, whereas others ( Egr1 , Apoa4 , and Per2 ) showed no differences between AvY and controls. In OST ( Figure 8H ), the dysregulated markers were involved in inflammation, vascular function, and ovulation ( Fos , Apoa4 , and alipoprotein A-I ( Apoa1 )), as well as ECM remodeling (adam metallopeptidase domain 8 ( Adam8 )) [ 42 – 48 ] ( Figure 8H ). Several tested markers exhibited trending differences ( p < 0.1) between normal-weight and HFD mice. The discrepancy between RT-qPCR and bulk-sequencing results may stem from high variability in stromal tissue composition, especially in normal-weight mice. For example, variations in the number of CL from previous estrous cycles could contribute to differences in gene expression.

C57BL/6J and AvY (AvY/a) mice were housed at the Center for Animal Resources and Education within the Cornell School of Veterinary Medicine, following a 14-h light and 10-h dark cycle. They had unrestricted access to food and water and were maintained in strict accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals, with ethical approval from the Institutional Animal Care and Use Committee at Cornell University. Two models of obesity—one diet-induced and the other epigenetic—were used to assess how obesity affects ovarian hemodynamics during the preovulatory and early luteal phases. The AvY mice, characterized by a mutation upstream of the transcription start site of the Agouti gene, served as the epigenetic model [ 23 ]. In the hypothalamus, Agouti-related protein acts as a potent antagonist to the MC4R melanocortin receptor, a critical regulator of food intake [ 23 , 24 ]. Ectopic expression of the Agouti gene in AvY mice increases MC4R antagonism, leading to hyperphagia, yellow fur, and adult onset obesity and diabetes [ 23 , 24 ]. In the AvY mice, the degree of CpG methylation of the Agouti promoter is inversely correlated with its expression, producing a range of coat colors and varying obesity levels, with yellow mice (unmethylated) being the most obese, mottled mice being decreasingly obese relative to the amount of dark fur present (i.e., slightly mottled mice are more obese than heavily mottled mice), and pseudo-Agouti mice (methylated) remaining mostly lean and having a dark brown coat color [ 23 , 24 ]. For this study, all data on AvY mice were collected at 10 weeks of age from slightly mottled and mottled AvY mice and age-matched black (a/a—non-AvY) controls from the same litters. For the diet-induced obese group, C57BL/6J mice were divided into two groups at 5 weeks of age: one receiving a standard chow diet with 16% of energy from fat and the other a HFD diet (D12492i; Research Diets, New Brunswick, NJ) with 60% of energy from fat. After a 10-week dietary regimen, mice from the HFD and normal-weight control groups were analyzed. Body composition was determined by magnetic resonance imaging (MRI) on a General Electric 3.0-Tesla scanner (Waukesha, WI) at the Cornell MRI Facility. The volume fraction of fat was assessed with a T2-weighted fast spin echo (FSE) sequence with and without fat suppression, and a Dixon sequence to obtain fat- and water-only images. Both data sets were analyzed with customized MATLAB software to compare T2-weighted images with and without fat suppression, and Dixon based fat- and water-only images. Estrous cycle monitoring and stage classification were conducted following established protocols [ 25 ]. The stages of the estrous cycle were determined based on observations of leukocytes, nucleated epithelial cells, and cornified epithelial cells by vaginal smears. Estrous cycles were monitored to identify the optimal stage for ovulation synchronization in studies with superovulation (all mice received PMSG injections during diestrus). For Doppler data collection in randomly cycling females, mice were tracked daily for 15 days to establish estrous cyclicity before imaging during diestrus (2 days) and estrus. Females were considered normally cycling if they completed two full cycles within the 15-day period. Standard superovulation stimulation was applied to adult female mice at 15 weeks of age (HFD mice and their normal-weight controls) and 10 weeks of age (AvY mice and age-matched controls) with an intraperitoneal injection of 10 IU PMSG at diestrus (NATE-0969; Creative Enzymes), followed by 10 IU hCG (9002–61-3; Sigma Aldrich) 48 h later. Agouti viable Yellow mice and age-matched controls were used for experiments at 10 weeks of age because at this age they had similar weight and adiposity as mice in the HFD group at 15 weeks of age. Response to hormonal stimulation was confirmed at the time of tissue collection at 24 h post-hCG when ovaries and oviducts were dissected to confirm ovulation as indicated by the presence of cumulus-oocyte complexes (COCs) in the oviducts and the formation of CL in the ovary. The ovaries collected at 24 h post-hCG were saved but unused in this manuscript. Mice that did not respond to hormonal stimulation were excluded from further analysis. All Doppler imaging experiments were performed as previously described [ 26 ]. Briefly, the MS550D transducer (22–55 MHz; 15 mm maximum imaging depth; 40–80 μ m axial/lateral resolution) at Cornell’s Biotechnology Resource Center Imaging Facility was positioned in the transverse plane relative to the mouse and secured using a stage clamp [ 26 ]. Ultrasonic gel was applied to a previously shaved area, and the transducer was lowered until a clear B-mode image was obtained ( Figure 1 ). Mice were maintained under anesthesia at 2% isoflurane (vol/vol) for the duration of data collection at each timepoint (10–15 min on average). Imaging and data collection were performed using the Visual Sonics Vevo 2100 ultrasound system (VisualSonics Inc.) [ 26 ]. The ovary was identified with reference to nearby anatomical structures, including the kidney (a gray oval structure ~5 mm in length and 6 mm in width) and the periovarian adipose tissue (POAT), a white structure between the ovary and the kidney ( Figure 1 ). The ovary appeared as a gray circular structure to the right of the ovarian fat pad, connected to a tubular structure representing the oviduct (Ov.D). The size of the ovary was influenced by hormonal stimulation. Each image was taken on the same ovary (right side of the mouse). The primary source of blood supply to the ovary is the ovarian artery (OA), a major blood vessel that originates from the abdominal aorta and enters the ovary through the hilum region. Utilizing Doppler imaging, the OA was identified as the prominent vessel entering the ovary through periovarian adipose tissue ( Figure S1A ). To confirm the identification of the ovary and visualize blood vessels, power Doppler imaging (with high sensitivity to blood flow of low velocity) was employed ( Figure S1B (i) ) [ 26 ]. It is important to note that the OA is often intertwined with the ovarian vein (OV), which carries deoxygenated blood and hormones away from the ovary. To differentiate these vessels, color Doppler imaging was used to distinguish them by colors, with one vessel appearing in blue and the other in red ( Figure S1B (ii) ). The blue and red colors in color Doppler represent the direction of blood flow relative to the ultrasound transducer, not necessarily arteries versus veins. Blood velocity is lower in the vein compared to the parallel artery [ 27 , 28 ]. In a separate group of immature mice—selected to establish the Doppler ultrasonography methodology because of their ease of access and robust hormonal response—we measured blood flow velocity in the red and blue vessels entering the ovary. The red vessel consistently exhibited higher velocity at all time points, identifying it as the ovarian artery ( Figure S1C ). However, because this pattern may not always hold, the ovarian artery and vein were ultimately distinguished based on relative blood flow velocity, and all subsequent data refer specifically to the vessel with the higher velocity (ovarian artery). Measurements of arterial velocity and diameter were obtained at the location marked in Figure S1A and S1B , immediately outside the entrance to the ovary. In this study, the medullary vessels (MV) were identified using power Doppler as the point at which OA enters the ovary and branches into several vessels supplying the cortex ( Figure S1B (i) ) [ 26 ]. This entry and branching point were subsequently confirmed using color Doppler. Measurements were taken at the point where these vessels branched ( Figure S1A and 1B ) [ 26 ]. As the vessels penetrate the ovarian cortex, they extend along the surface of preovulatory follicles, encircling them from the base to the apex. As ovulation approaches (12 h post-hCG in mice), the apical vessels become too small (10 μ m or less) to be detected by ultrasonography [ 17 ]. Therefore, throughout all time points, “cortical flow” refers specifically to vessels located at the base of preovulatory follicles ( Figure S1A and 1B ). Ultrasonic parameters were adjusted to detect slow blood flow within all blood vessels based on preset options following the approach by Migone et al. [ 17 ]. These parameters included Doppler gain (32–55 dB), sensitivity (5), dynamic range (15 DR), and velocity (1 kHz). All images and pulse waveforms were saved as cineloops. Pulse-wave Doppler mode (PW) was utilized to quantify blood flow within each vessel of interest. The three highest peak-systolic (PSV) and end-diastolic velocities (EDV) were recorded and averaged to determine the average velocity (AV): PSV + EDV 2 ( Figure S1D ). Pulse waveforms lacking three individual waves with consistent peak-systolic and end-diastolic velocities were excluded [ 26 ]. This was done to ensure that only measurements that were consistent across multiple cardiac cycles were included to ensure measurement accuracy. For velocities below the 10 mm/s detection threshold, a default value of 5 mm/s was assigned to retain these vessels in the study and enable comparison with those exhibiting detectable flow. Logarithmic transformation was applied to fold-change velocity data to approximate a normal distribution for statistical analysis. The resistive index (RI) was calculated by normalizing the difference between PSV and EDV to PSV: PSV − EDV PSV . However, RI was not calculated for vessels with flow <10 mm/s, as assigning a default value of 5 mm/s to both PSV and EDV resulted in an RI of 0, which did not accurately reflect vascular resistance [ 26 ]. Finally, ovarian arterial diameter was measured using the Vevo LAB measurement tool (VisualSonics Inc.) in color Doppler. All hemodynamic measurements and blood vessel properties were obtained at various time points spanning the preovulatory and early luteal phases, including 0 h, 1 h, 8 h, 11 h, 12 h, and 24 h post-hCG. These time points were selected to capture hemodynamic parameters across distinct stages of the ovulatory phase: immediately before and after the LH surge (0 h and 1 h post-hCG), at the midpoint of the preovulatory window when vascular regulators are typically upregulated (8 h post-hCG), just prior to ovulation (11 h and 12 h post-hCG), and during the early luteal phase (24 h post-hCG). Ovarian hemodynamics were also measured in naturally cycling mice during the first and second days of diestrus, and at midnight during estrus after proestrus was observed in the morning. For each tissue collection, mice were euthanized by CO 2 inhalation (1.3 L/min). Ovarian collection was performed by making an incision through the skin and muscle to expose the ovary attached to the uterine horn, followed by transection below the oviduct and at the level of the POAT beneath the kidney. Any remaining extraneous tissues (POAT, oviduct, or bursa) were carefully removed from the ovary under a dissection microscope. For bulk RNA sequencing and RT-qPCR, ovaries were collected 48 h after PMSG treatment (0 h post-hCG). Granulosa cells (GCs) were isolated from follicles by repeated needle puncture, and both granulosa cells and the residual ovarian stromal tissues (OST) were stored at −80°C until RNA extraction. Under anesthesia by isoflurane using the bell jar method, the ovarian vasculature was labeled by retro-orbital injection of 10 μ L DyLight 649-labeled Lycopersicon esculentum (Tomato) Lectin (LEL, TL) ( L32472 ; ThermoFisher) to mice at 0 h post-hCG. Mice were removed from anesthesia after lectin injection to wake up and were euthanized by CO 2 inhalation 5 min after lectin injection. Ovaries were collected and fixed in 4% paraformaldehyde (15710; Electron Microscopy Sciences) at 4°C for 4 h. Fixed ovaries were then washed three times in phosphate buffered saline (PBS, QB-119–069-491; Neta Scientific) for 10 min each before tissue clearing using CUBIC solutions as previously described [ 29 ]. Ovaries were cleared in CUBIC-1 solution by gentle rocking at room temperature for 4–5 days. After washing three times in PBS for 10 min each, ovaries were incubated in 20% sucrose–PBS overnight. Subsequently, cell nuclei were stained with Hoechst 33258 in CUBIC-2 solution for at least 24 h with gentle rocking at room temperature. Lastly, ovaries were washed using CUBIC-2 solution for 4 h and stored in fresh CUBIC-2 solution at room temperature until imaging. Ovaries were imaged using an inverted laser scanning confocal microscope (Zeiss LSM710 microscope). Images were taken at 10- μ m intervals starting from the outer surface of the ovary to generate a 550- to 600- μ m-thick z-stack. Z-stack images were reconstituted into 3D projection images using ImageJ. Features of ovarian vasculature were analyzed and quantified using Hessian Tubeness filters [ 30 ] at specific diameters (d = 5, 10, 20 μ m), generating image stacks of vascular tubes segmented by these different sizes. CL were excluded from quantification because the number and size of CL (some from previous cycles) can vary substantially between mice and introduce variability. The filter sizes were chosen based on typical vessel diameters: capillaries range from 4 to 10 μ m while the average vessel diameter in ovarian stromal tissue is ~20 μ m [ 12 , 31 ]. The relative volume (RV) was calculated as the total volume of vascular tubes at each size (segmented using Otsu’s method [ 32 ]) normalized to the total volume of the corresponding ovary (segmented using the Triangle method [ 33 ]). The average intensity (AI) was determined as the average pixel value in the Tubeness-filtered, Otsusegmented datasets. The total intensity (TI) was then calculated using the formula: T I = R V × A I . All analyses were conducted with scripts written and executed using the macro scripting capabilities within the Fiji version of ImageJ (available upon request). Total mRNA was extracted from GCs and OST from normal-weight and HFD mice using the RNeasy Micro Kit, following the manufacturer’s instructions. Granulosa cells or OST isolated from one ovary were used as one replicate, and three independent biological replicates were used for RNA-sequencing (RNA-seq) library generation. Library preparation and transcriptome sequencing were conducted by Novogene Co., Ltd (Davis, USA). Raw RNA-seq reads were aligned to the mouse genome (GRCm38/mm10) using HISAT2 (version 2.0.5) by default parameters. Raw read counts for each gene were generated using featureCounts (v1.5.0-p3), and transcripts per million (TPM) values were calculated using EdgeR (version 1.3.3). Only genes with TPM >1 in at least one replicate were included. DESeq2 R package (version 1.20.0) was used for differential gene expression analyses with cutoffs absolute fold-change >1.5 and p -values <0.05. Differentially expressed genes (DEGs) were visualized using VolcaNoseR [ 34 ]. Gene Ontology (GO) analyses were performed using the Metascape online tool [ 35 ] and visualized with Hiplot ( https://hiplot.org ) [ 36 ]. Total messenger RNA (mRNA) was extracted from GCs using the Arcturus PicoPure RNA Isolation Kit (KIT0204; ThermoFisher Scientific). For OST, mRNA was extracted using RNeasy Micro Kit (74106; Qiagen). Real-time quantitative PCR (RT-qPCR) was performed using RT 2 SYBR Green ROX FAST Mastermix (330623; Qiagen) on StepOnePlus Real-Time PCR System (4376600; Applied Biosystems). Relative levels of mRNA were calculated using the 2 ( − Δ C t ) method and normalized using housekeeping gene ribosomal protein L19 ( Rpl19) . The primers used are listed in Table S1 . All quantitative data are presented as mean ± standard error of the mean (SEM). Statistical analyses for hemodynamic measurements ( results section 4.2) were done using linear mixed-effects models with mouse as a random effect. Between-group differences at each time point were analyzed in R (lmer), followed by post hoc comparisons using the emmeans function. Differences in velocity fold-change between time points within each experimental group were analyzed with Prism’s mixed-effects REML model. Other statistical analyses were performed using the GraphPad Prism 10 software. For comparison across three groups, one-way ANOVA followed by Tukey multiple comparisons test were used. P value <0.05 was considered statistically significant. Outliers within each experimental group were identified and removed using the interquartile range method. Sample sizes, the statistical tests used, and p -values were described in each figure legend.

This study provides multiple lines of evidence that obesity disrupts vascular remodeling during the preovulatory and luteal phases in mice. Using Doppler ultrasonography, we show that ovarian hemodynamics is impaired in both naturally cycling and superovulated HFD and AvY obese mice. Notably, the increase in blood velocity fold-change between 0 h and 1 h post-ovulation induction—blunted in obese mice relative to normal-weight controls—was positively correlated with ovulation rate, indicating that this early rise in blood flow may be a predictor of ovulation success. We also observed altered expression of genes involved in pathways essential for ovulation and vascular remodeling. Together, these findings provide new insight into how obesity impairs vascular remodeling and ovulatory function and establish a novel use of Doppler ultrasonography to quantify dynamic ovarian blood flow in vivo. Using Doppler ultrasonography, we established a temporal pattern in ovarian blood flow velocity in normal-weight mice, marked by an increase between 0 h and 1 h post-hCG across the OA, MV, and CV ( Figure 3A and 3B ), paralleling observations reported in other species [ 9 , 18 , 19 , 49 ]. This increase in velocity seen in normal-weight mice was positively correlated with ovulation rate ( Figure 4 ) and was absent in both HFD and AvY obese mice ( Figure 3B ). Similar findings are reported in women undergoing fertility treatments: those who ovulated exhibited an increase in ovarian blood flow 15–180 min after hCG administration, while those who did not ovulate showed no change in blood flow [ 18 ]. These parallels highlight the potential of early hemodynamic responses as non-invasive predictors of ovulatory success. The increase in velocity and correlation with ovulation rate were most robust in the MV ( Figure 3A (ii) and 3B (ii) ), where the disruption of blood flow regulation has been associated with PCOS and ovarian hyperstimulation syndrome [ 50 , 51 ]. These associations suggest a role of MV in ovarian biology, such as in the ovulatory process; further investigation is warranted to define this role. Adult mice exhibited substantial variability in ovarian hemodynamics, which may reflect differences in ovarian composition and or physiological heterogeneity, as observed in humans, where blood pressure and organ perfusion vary both between and within individuals [ 52 – 54 ]. Nevertheless, the consistent increase in velocity from 0 h to 1 h post-hCG, even amid variability, highlights a robust physiological event. While the impact of obesity on ovulatory success is multifaceted—including impaired steroid hormone production, reduced oocyte quality, oxidative stress, and inflammation [ 55 – 57 ]—the correlation presented here, though not experimentally tested as causative, suggests that disrupted ovarian blood flow may contribute to reduced ovulation in obese mice. During the early luteal phase (24 h post-hCG), the superovulated obese groups exhibited increased blood velocity compared to normal-weight controls ( Figure 3A ); in randomly cycling mice, there was a consistent pattern of increased blood velocity on two consecutive days of diestrus in both obese groups ( Figure 6A ). The growing CL requires adequate blood flow to produce and transport progesterone [ 58 ]. Obesity reduced circulating progesterone concentrations and downregulated expression of steroidogenic enzymes in the CL [ 59 ]. In non-obese patients, improved blood flow to the CL is associated with increased serum progesterone levels [ 58 ]. Given the critical role of progesterone and the known impact of obesity on CL function, the disproportionate increase in blood flow velocity observed during the luteal phase may represent a compensatory mechanism to support luteal function. However, future studies are needed to test the functional relationship between increased blood velocity and luteal function in obesity. Together, these data characterize for the first time hemodynamic changes during natural cycles and with superovulation in adult normal-weight and obese mice, showing that obesity, regardless of diet type, disrupts normal ovarian hemodynamics. These data support altered ovarian vascular function as a mechanism by which obesity impacts ovulation. These data also highlight the value of Doppler ultrasonography as a tool for further investigation of murine ovarian hemodynamics. This study compared two obesity models—diet-induced obesity (DIO) through HFD and the AvY model. Both models showed similar disruptions in ovarian blood flow following superovulation, including a blunted increase in velocity at 1 h post-hCG, and elevated velocity in the OA during the luteal phase in superovulated and randomly cycling conditions ( Figure 3A and 3B ; Figure 6A ). However, several model-specific differences also emerged: (1) AvY mice displayed lower ovarian perfusion than HFD mice ( Figure 5 ); (2) RI during estrus was impaired only in AvY mice ( Figure 6B ); (3) HFD mice exhibited greater capillary density than AvY mice ( Figure 7 ); and (4) while some genes ( Fos and Ccn1 ) were altered in both models, others ( Egr1 , ApoA4 , and Per2 ) were dysregulated only in HFD ovaries ( Figure 8 ). These findings indicate that although obesity broadly disrupts ovarian blood flow, the underlying regulatory mechanisms differ between models, suggesting that distinct forms of obesity may impair ovarian function through different pathways. The HFD model is the most widely used approach for studying DIO and recapitulates several features of human obesity, including increased adiposity, brain inflammation, hyperglycemia, and insulin resistance [ 60 – 63 ]. High-fat diet also causes reproductive dysfunctions—such as abnormal estrous cycles, reduced oocyte quality, and lower ovulation rates—that mirror fertility issues in obese women [ 5 , 64 – 68 ]. Diet influences vascular function as well, with high saturated fats and carbohydrates impacting vascular resistance [ 69 , 70 ]. However, human obesity typically arises from a combination of factors—excess caloric intake from both fat and sugar, reduced physical activity, and genetic predisposition—that are not fully captured by HFD alone [ 63 , 71 – 74 ]. Like HFD mice, AvY mice display increased fat mass, hyperinsulinemia, and elevated leptin [ 75 , 76 ]. These phenotypes are due to overeating rather than overconsumption of a specific macronutrient (fat), thus closely representing the overconsumption aspect of human obesity [ 63 , 71 – 78 ]. Unlike other genetic obesity models (e.g., ob/ob mice), AvY mice remain fertile, making them particularly useful for reproductive studies [ 79 – 81 ]. Importantly, MC4R mutations, such as the one driving obesity in AvY mice, are the most common cause of monogenic obesity in humans, further supporting the relevance of AvY mice for modeling obesity-related reproductive dysfunction [ 71 ]. Nevertheless, their use is less established than HFD models, and while some reports describe reproductive alterations such as impaired oocyte quality, reproductive tract hyperplasia, and reduced fertility, much of the impacts of obesity in AvY mice on reproductive function remain to be discovered [ 80 , 82 ]. In this study, we contribute to the reproductive knowledge on AvY mice by reporting decreased ovulation rates, abnormal estrous cyclicity, and impaired ovarian hemodynamics both in hormonally stimulated and naturally cycling conditions. The lethal yellow (Ay/a) mouse model, like AvY, develops obesity through hyperphagia driven by MC4R antagonism and exhibits abnormal estrous cyclicity, reduced mating success, and a lower ovulation rate [ 83 ]. Although the two models are not identical and further characterization of AvY reproductive function is needed, findings from data reported here and Ay/a model suggest that similar reproductive abnormalities may also occur in AvY mice. One limitation of using both models is the difference in the timing of obesity development: AvY mice and their age-matched controls were collected at 10 weeks of age, whereas HFD mice and their controls were collected at 15 weeks. Although this represents a 5-week age gap, all animals were still within the adult stage (rather than immature or aged). Additionally, body weight and adiposity are similar between the two models at their respective age ( Figure 2A and 2B ), supporting their overall comparability. Overall, by incorporating both HFD and AvY models, this study captures complementary aspects of obesity—dietary and genetic/hyperphagiadriven—providing broader insight into how excess nutrition impacts ovarian vascular function, while also contributing new data on reproductive function in AvY mice. The hCG/LH surge induces both genomic and rapid non-genomic enzymatic responses that regulate blood flow, particularly through nitric oxide (NO) and hydrogen sulfide signaling—two potent and fast-acting modulators of vascular tone [ 84 – 89 ]. Disruption of vascular regulators transcripts at 0 h post-hCG may contribute to the abnormal vascular phenotype observed at 1 h post-hCG in obese mice. However, further research is needed to elucidate non-genomic mechanisms underlying hemodynamic regulation and dysfunction. Bulk RNA sequencing on GC from normal-weight and HFD obese mice at 0 h post-hCG showed downregulated vascular development genes including Ccn1 ( Figure 8G ), an activator of NO signaling [ 90 ]. Nitric oxide is a key vasodilator and reduced Ccn1 in obese GCs may limit NO bioavailability, vasodilation, and vessel compliance, potentially impacting ovulation [ 91 – 94 ]. Interestingly, RT-qPCR revealed that genes within the vascular development pathway ( Fos and Ccn1 ) were also dysregulated in the AvY mice, suggesting that some similar vascular dysfunction may be at play. Extracellular matrix organization genes were upregulated in both GCs and OST from HFD mice. Extracellular matrix remodeling is essential for follicular development and ovulation, and obesity increases collagen deposition and disrupts cell–cell and cell–matrix interactions, contributing to anovulation [ 95 – 98 ]. Some ECM genes, such as Adam8 ( Figure 8F )—an LH-regulated ECM protease critical for follicle rupture—were downregulated, reflecting obesity-driven disruption of ECM regulation [ 99 – 101 ]. Such alterations may contribute to impaired ovulation and the abnormal vascular phenotype observed by Doppler ultrasonography, as blood flow changes require coordinated ECM reorganization and inflammatory signaling [ 11 , 102 ]. We show that several ECM regulators are disrupted in ovaries from obese mice at the time of ovulation induction ( Figure 8E and 8F ), identifying potential targets for future studies on obesity-related biomechanical changes during this critical window. We focused on HFD mice for bulk RNA-sequencing because this is the most widely used model, yet the ovarian transcriptome at the onset of the LH surge had not previously been characterized. The absence of comprehensive transcriptomic data for AvY ovaries remains a limitation of this study but also provides an opportunity for future work aimed at elucidating the mechanisms of ovarian function in this model, which closely reflects obesity associated with MC4R mutations in humans. To assess the physiological and structural mechanisms underlying obesity-related impairments in ovarian blood flow velocity, we evaluated whether RI, ovarian arterial diameter, and the architecture and abundance of small vessels were altered in ovaries from obese mice. Changes in vessel diameter via vasoconstriction or dilation are key regulators of blood flow. Smooth muscle contraction and relaxation alter diameter and, in turn, vascular resistance [ 103 , 104 ]. Vasodilation reduces resistance and increases flow, whereas vasoconstriction increases resistance and decreases flow [ 105 , 106 ]. Our analysis of OA diameter did not reveal differences between groups or across time points ( Figure S2 ) that could account for the variability in blood flow velocity. However, Doppler imaging may lack the resolution to detect the small diameter changes reported in murine ovaries (some as little as 5 μ m) which fall below its detection threshold [ 12 ]. Even subtle changes of this magnitude could substantially affect blood flow according to Poiseuille law. This has been the case in other species as well, for example, in newborn lambs, Doppler overestimated tibial artery diameter by 10–27% compared to arteriography [ 107 ]. Further investigation into OA vasodilation and its regulatory mechanisms is essential to understanding the rapid increase in blood flow following the LH surge. To assess vascular resistance, we measured RI and observed an increase between 1 h and 8 h post-hCG in normal-weight mice ( Figure 3C (i) ), paralleling a reduction in blood velocity. In contrast, HFD mice exhibited greater variability in RI across time points, a pattern not seen in normal-weight or AvY mice ( Figure 3C (ii) ). This suggests that obesity may disrupt the normal temporal regulation of ovarian vascular resistance during the preovulatory and early luteal phases in a diet-specific manner. Obesity is generally associated with abnormal systemic vascular resistance, and, while PCOS has been linked to reduced ovarian RI, the specific impact of obesity alone on ovarian RI remains poorly understood [ 108 – 110 ]. In randomly cycling mice, RI was elevated in the medullary vessels of AvY mice compared to normal-weight controls on the second day of diestrus and at estrus ( Figure 6B ). An increased RI typically indicates reduced organ blood flow, suggesting that although flow velocity was unchanged in these vessels, overall perfusion may have been impaired in AvY mice [ 109 , 111 ]. No group differences in RI were observed following hormonal stimulation. It is possible that the disrupted RI pattern seen in naturally cycling AvY mice was masked by the elevated ovarian blood flow induced by superovulation. This finding highlights the value of studying ovarian hemodynamics under both natural and hormonally manipulated conditions. Small vessels at the apex of ovulatory follicles are particularly important as they constrict to facilitate follicle wall breakdown [ 17 ]. We selected 0 h post-hCG rather than 1 h—when abnormal blood flow is observed in obese mice—because structural changes detectable by this method likely require more time to manifest than functional changes such as nitric oxide–mediated shifts in blood flow. This analysis revealed no differences in vessel volume or intensity between obese groups and normal-weight controls. However, diet composition is known to influence vascular function [ 112 , 113 ], and while prior studies on obesity’s impact on ovarian microvasculature are inconsistent [ 114 , 115 ], reports from other organs describe obesity-induced capillary rarefaction after an initial phase of expansion [ 116 , 117 ]. Prolonged HFD exposure may therefore exacerbate ovarian microvascular disruption, underscoring the need for higher-resolution approaches and longer exposure periods in future studies. Despite its resolution limitation, Doppler ultrasonography enables repeated, non-invasive quantification of ovarian hemodynamics without disrupting blood flow, allowing for the characterization of temporal trends over time such as during the preovulatory phase or at multiple time points of the estrous cycle. Using this approach, we demonstrate that obesity disrupts ovarian hemodynamics in mice, with the most pronounced impairment occurring during the first hour following the LH surge and during the luteal phase (Graphical abstract). In parallel, we identify alterations in several key regulators of vascular development and function in obese ovaries (Graphical abstract). Collectively, these findings provide a foundation for future studies aimed at understanding how obesity alters ovarian vascular remodeling and identifying therapeutic targets to improve ovulatory outcomes in women with obesity-related infertility.

The prevalence of obesity has nearly tripled since the 1970s and is projected to affect half of the world’s population by 2035 [ 1 , 2 ]. Obesity is a major risk factor for reduced fertility in both men and women [ 3 – 5 ]. Obese women are particularly at risk for reproductive disorders like polycystic ovary syndrome (PCOS), endometriosis, and uterine fibroids [ 6 ]. Ovulatory disorders are a major cause of infertility, with obese women being three times more likely to experience anovulatory infertility compared to lean women [ 7 , 8 ]. Ovulation is a complex process that requires adequate follicle development under the influence of follicle-stimulating hormone (FSH), followed by the preovulatory luteinizing hormone (LH) surge that triggers follicle wall breakdown and oocyte release. Both follicular and preovulatory phases involve extensive vascular remodeling (dynamic changes in the structure and hemodynamics of the ovarian vasculature) to supply developing follicles with nutrients, hormones, and oxygen [ 9 – 12 ]. Continuous physiological vascular remodeling in adults is largely unique to reproductive tissues such as the ovary and uterus, underscoring its essential role in reproductive function [ 13 – 16 ]. Ovarian vascular remodeling in response to the LH surge encompasses a range of spatially and temporally coordinated processes, including angiogenesis, changes in vascular permeability, vasodilation, vasoconstriction, and modulation of blood flow, which together enable the rupture of the follicle at the right location and time and the subsequent formation of corpora lutea (CL) [ 10 – 12 , 17 – 19 ]. During superovulation protocols, pregnant mare serum gonadotropin (PMSG) is administered to mimic FSH, followed 48 h later by human chorionic gonadotropin (hCG) to simulate the LH surge. In several species, including humans, cattle, and rabbits, there is consistent evidence that the LH surge induces a rapid increase in blood flow velocity within ovarian vessels [ 9 , 18 , 19 ]. In women undergoing fertility treatments, the immediate rise in flow velocity following hCG treatment has been associated with successful ovulation [ 18 ]. The consistently increased blood perfusion and velocity following the LH surge across multiple species suggests that this is a critical physiological response during the preovulatory phase. Disruption of vascular remodeling during this period has been shown to impair ovulation; however, the contribution of preovulatory changes in ovarian blood flow to ovulatory success remains largely untested and could highlight new therapeutic targets for improving ovulation [ 12 , 17 ]. Obesity disrupts several aspects of cardiovascular hemodynamics, including elevated cardiac index (cardiac output normalized for body surface area), increased systemic vascular resistance, and reduced systemic vascular compliance [ 20 , 21 ]. There is limited information on the specific impact of obesity on ovarian vascular function. However, in individuals with PCOS, overweight women had lower ovarian blood flow compared to lean women, suggesting that obesity impacts ovarian blood flow in addition to PCOS [ 22 ]. Given the adverse effects of obesity on systemic blood flow, we hypothesized that obesity disrupts ovarian vascular function during the preovulatory phase, contributing to impaired ovulation. To test this hypothesis, we employed Doppler ultrasonography as a novel, non-invasive technique to evaluate hemodynamics at a series of time points spanning the preovulatory phase, and during natural cycles, in adult normal-weight mice and two obese mouse models: one induced by a high-fat diet (HFD) and the other through epigenetically mediated hyperphagia (Agouti viable Yellow (AvY) mice) [ 23 ]. The impact of obesity on ovarian vasculature was further investigated by analyzing lectin-labeled blood vessel structure. Additionally, we performed bulk RNA sequencing to elucidate the cellular and molecular pathways mediating the effects of obesity on ovarian vascular function.

Supplementary data are available at BIOLRE online.