Short-term moderate caloric restriction in the rhesus macaque attenuates markers of ovarian aging in select populations

A nonhuman primate model of short-term moderate CR

To investigate the impact of CR and age on ovarian aging parameters, we acquired a subset of archived ovarian tissue samples from a prior study involving short-term moderate CR in the rhesus macaque in two age cohorts (Figure 1). In the “young” cohort, animals ranged in age from 7–10 years of age at the start of the study, and in the “old cohort” they ranged from 16–23 years of age at the start of the study (Figure 1). The animals were either maintained on a control diet (CON) or one of 30% CR (CR) for a period of three years after which necropsy was performed to harvest organs, including ovaries, for downstream analyses. At the time of necropsy, young animals were 12.3 ± 1.5 (control diet, N = 4) and 13.0 ± 0 years of age (CR diet, N = 4), and the old animals were 24 ± 2.5 (control diet, N = 8) and 23.7 ± 2.6 years of age (CR diet, N = 5) (Table 1). There was no significant difference in age at the time of necropsy based on diet group. The period of the study spanned adolescence and reproductive adulthood in the young cohort and reproductive adulthood, perimenopause, and menopause in the old cohort (Figure 1). All young animals in both the CON and CR groups had regular cycles. Old animals were classified as either regular, irregular, or not cycling based on their hormone profile (Figure 1 and Table 1). There were no regularly cycling animals in the old CR group (Figure 1 and Table 1). No differences in estradiol (E2), progesterone (P4), follicle-stimulating hormone (FSH) or luteinizing hormone (LH; relative to the LH peak) were noted among animals with regular cycles between young and old cohorts (data not shown). Thus, it appears that the CR diet permits normal functioning of the reproductive axis in young regular cyclers. By year 3 of the diet, the relative numbers of irregularly cycling and non-cycling animals increased in both the old CON and old CR groups due the animals entering perimenopause (Figure 1 and Table 1). Initiation of CR later in life did not appear to be beneficial to ovarian cyclicity as similar impairments occurred with age in both the old CON and old CR groups (Figure 1 and Table 1), however the small sample limits our ability to draw definitive conclusions.

The age-associated decrease in follicle number is not mitigated by CR, but CR exhibits a more youthful follicle class distribution in aged animals

To examine the effect of age and CR on ovarian aging parameters in the nonhuman primate, ovaries were harvested at the time of necropsy. The ovaries from young animals weighed more than those from old animals likely due to the greater number of follicles, but there was no difference within age cohort based on diet (Table 1). Consistent with the age-dependent difference in ovarian weight, we were able to obtain more histologic tissue sections (a surrogate marker for ovarian volume) following serial sectioning of young ovaries relative to old (Table 1). In mammals, the widely held dogma is that females are born with a finite and nonrenewable pool of primordial follicles which comprise the ovarian reserve and dictate reproductive lifespan [30]. These follicles are gradually depleted as females age, and at the time of menopause (cessation of ovarian cycling), very few follicles remain. When examining representative histological sections of the ovaries across experimental cohorts at a gross level, there was an obvious age-dependent decline in the number of visible follicles as expected (Figure 2).

To gain further insight into how follicle dynamics in the rhesus macaque are impacted by age and diet, we classified and counted all the major follicle stages with normal morphology in the ovary (Figure 3). These included the quiescent pool of primordial follicles (Figure 3A) as well as those that had activated and were growing, including transitional primordial, primary, transitional primary, secondary, multilayer, and antral follicles (Figure 3B–3G). In addition to healthy follicles, we also classified and counted multi-oocytic follicles, secondary and multilayer follicles with abnormal morphological features, and atretic antral follicles (Figure 3H–3K). For each animal across experimental cohorts, we performed follicle counts in three histologic sections that spanned the ovary (Figure 3L, 3M). Follicle numbers were normalized to the area of the histological section, and each data point represented the average follicle number per area for each animal (n = 3 histological sections per animal) (Figure 4A–4H and Supplementary Figure 1A–1D). Overall, there was a decrease in total follicle number as a function of age for CON animals which is consistent with the gross histology (Figure 4A and Supplementary Table 1). For CON animals, there were significant decreases in primordial, transitional primordial, primary, transitional primary, secondary, and multilayer follicles between young and old animals (Figure 4B–4G). For CR animals, there were statistically significant decreases in primary follicles and secondary follicles between young and old animals (Figure 4D, 4F). Pairwise comparisons could not be made for antral follicles as there was no statistical significance for either of the main effects in the two-way ANOVA. Although there was a significant effect of age on follicle number across all follicle classes (except antral), there was no effect of diet on follicle number based on analysis of sources of variation using two-way ANOVA (Supplementary Table 1).

We performed an additional analysis of follicle number divided by total slide number per ovary, as a proxy for ovarian volume. As mentioned previously, we noted a large range of ovarian size across individual animals within an experimental group, as well as differences between groups, with ovarian mass significantly lower in old animals compared to young animals (p = 0.0016, Table 1 and Supplementary Table 2). Standardized follicle number was calculated by summing the follicle number per follicle class in each individual animal across all three tissue sections and dividing by the total number of histologic slides generated for that animal. Similar age-related declines in follicle number were observed when comparing these standardized follicle numbers (Supplementary Figure 2). No effect of diet on standardized follicle number was observed in the main effects analysis (Supplementary Table 3). Raw follicle number data is presented in Supplementary Table 4.

We next evaluated the distribution of follicles across the different developmental classes for each of the four experimental groups (Figure 4I–4L). The follicle class distribution for young CON, young CR, and old CR animals was grossly similar, with the vast majority being primordial or transitional primordial follicles (Figure 4I, 4K, 4L). By contrast, the follicle distribution in old CON animals was substantially different, with less than half of follicles classified as primordial or transitional primordial (Figure 4J). When compared to young animals, the follicle class distribution for old control animals is significantly different (X² (6, N = 467) = 234.6, p < 0.0001). By comparison, there was no statistically significant difference in follicle class distribution between old calorie-restricted animals and young animals (X² (6, N = 336) = 12.54, p = 0.0509). We then plotted the follicle class distribution on the level of individual animals, which further demonstrates a similar follicle class distribution across the young CON, young CR, and old CR groups and confirms that these patterns are not due to particular animals (Figure 4M). Notably, this analysis revealed that the old CON animals appear quite distinct from the others given that 3/8 animals had no follicles and one had only transitional primary follicles in the analyzed histological sections (Figure 4M). Overall, these data suggest that although CR does not prevent the overall age-dependent loss in follicles, it does maintain a distribution of follicles across developmental classes that is more similar to young animals, with a similar proportion of primordial and transitional primordial follicles observed between old CR animals and young animals irrespective of CON or CR status (Figure 4I–4M).

Menstrual cycle status is associated with follicle number

In our study, the cohort of old animals included one animal with regular cycles (n = 1 CON), nine with irregular cycles (n = 5 CON, n = 4 CR), and three non-cycling animals (n = 2 CON, n = 1 CR). We, therefore, performed an additional analysis of how follicle number per area is affected by diet and menstrual cyclicity (Figure 5 and Supplementary Figure 3). For all follicle classes, there was a clear pattern of decreasing follicle number per area with increased menstrual cycle irregularity, with the lowest numbers of follicles seen in non-cycling animals for both CON and CR animals (Figure 5A–5H and Supplementary Figure 1E–1H). This pattern was also seen for standardized follicle number (Supplementary Figure 3). Statistically significant decreases in follicle number per area of irregularly cycling or non-cycling control animals compared to regularly cycling control animals were seen for all follicle classes, except antral follicles (Figure 5). The two-way ANOVA revealed that there was no statistically significant interaction between the effects of diet and menstrual cyclicity for total follicle number per area as well as follicle number per area for all normal and abnormal follicle classes (Supplementary Table 5). Simple main effects analysis showed that diet did not have a significant effect on total follicle number per area as well as follicle number per area for all normal and abnormal follicle classes (Supplementary Table 5). However, simple main effects analysis did show that menstrual cyclicity had a significant effect on the total follicle number per area, primordial follicle number per area, transitional primordial follicle number per area, primary follicle number per area, transitional primary follicle number per area, secondary follicle number per area, multilayer follicle number per area, multi-oocytic follicle number per area and atretic antral follicle number per area (Supplementary Table 5). Similar results in the main effects analysis were seen when looking at standardized follicle counts (Supplementary Table 6).

Caloric restriction maintains the primordial follicle pool in a subpopulation of animals with irregular cycles

Previous studies have shown that the timing of CR interventions can modulate their impact [31, 32]. Although many studies focus on early (pre- or peri-pubertal) intervention [33–35], those aimed at extending the fertile window are likely to be to be more clinically relevant if initiated in later adulthood, when fertility is beginning to decline. Given the physiologic differences between aged animals with and without menstrual cycles, we performed a more focused analysis on the subgroup of irregularly cycling animals to determine the effect of CR in this clinically relevant population (Figure 6 and Supplementary Figure 1I–1L). Despite the lack of statistically significant effect of diet on follicle number in the cohort of all aged animals, it was physiologically reasonable to perform this subgroup analysis, given the significant physiological differences between cycling and non-cycling animals, and the statistically significant effect of menstrual cyclicity on follicle number. In the subgroup of irregularly cycling old animals, CR animals had significantly higher total follicle numbers per area than control (Figure 6A) which was primarily attributed to the significant increase in the number of primordial and multilayer follicles per area (Figure 6B, 6G), while other follicle classes did not differ (Figure 6C–6F, 6H). Despite the small follicle numbers in the abnormal follicle groups, which limits the power of the study, a significant increase in the secondary AMF follicle number per area in CR females was observed (Supplementary Figure 1J). However, a similar trend toward increasing follicle number per area with CR was seen for multilayer AMF follicles and atretic antral follicles (Supplementary Figure 1K, 1L). There was no effect of diet on multi-oocytic follicle number per area (Supplementary Figure 1I). Standardized follicle number for total follicles, follicles with normal morphologies, and follicles with abnormal morphologies are presented in Supplementary Figure 4. There was a statistically significant increase in only primordial follicle number with CR using this analysis (Supplementary Figure 4B), while total follicle number and other follicle classes did not differ (Supplementary Figure 4A, 4C–4L). Thus, a CR intervention later in life when an individual is experiencing irregular menstrual cyclicity may have a positive impact by preserving the primordial follicle pool and thereby potentially prolonging reproductive function.

Age-related ovarian fibrosis is attenuated by CR

Distinct changes in the ovarian extracellular matrix occur with advanced reproductive age, including an increase in collagen and a decrease in hyaluronic acid which are associated with increased tissue stiffness [4, 36, 37]. To investigate whether CR had an impact on age-related changes in ovarian extracellular matrix, we evaluated collagen and hyaluronic acid (HA). We performed Picrosirius Red staining which is a histologic stain that can be used to detect collagen I and III fibers, and we used quantitative digital pathology to quantify pixels of high, medium, and low intensity in ovarian sections across experimental groups, based on our prior work (Figure 7) [38]. As expected with age, there was an increase in collagen content in old CON ovaries (Figure 7B, 7F) relative to young CON ovaries (Figure 7A, 7E) as evidenced qualitatively by the increase in red staining and quantitatively by the age-dependent increase high intensity pixels (Figure 7D, 7H, 7K) with no CR effects in young animals (Figure 7C, 7G). However, this age-dependent increase in ovarian collagen was abrogated in ovaries from old CR animals (Figure 7K). With increased age, there was a decrease in medium intensity pixels in control animals, suggesting a change in the collagen matrix (Figure 7J). No statistically significant differences between groups were seen for low intensity pixels (Figure 7I). In the aging mouse and human ovary, the increase in collagen is accompanied by a decrease in hyaluronic acid [4, 36, 39]. Using a fluorescent hyaluronic acid binding protein (HABP) assay to localize hyaluronic acid within ovarian sections, we noted that the loss of hyaluronic acid with age is also conserved in the rhesus macaque in CON samples (Figure 8A, 8B). However, this age-related difference was not significant in CR animals (Figure 8C–8E). Together these results suggest that CR may attenuate age-related fibrosis in the ovary.

Animals

We acquired a subset of archived ovarian tissue samples from a prior study involving short-term moderate CR in the rhesus macaque in two age cohorts (Figure 1). In this study, a cohort of young and old (peri-menopausal) female rhesus macaques (M. mulatta) was selected under a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the Oregon National Primate Research Center. The animals were housed in individual cages with auditory, visual, and olfactory interaction with female conspecifics in a temperature-controlled environment (24°C) under a fixed 12L:12D photoperiod (lights on from 700 h to 1900 h) with ad libitum access to drinking water. Animals were cared for by the Oregon National Primate Research Center in accord with the National Research Council’s Guide for the Care and Use of Laboratory Animals [53], which included daily health checks to ensure normal behavior, food consumption, and waste production. Additionally, routine physical examinations, hematological studies, fecal parasite checks, tuberculin testing, and dental cleaning were performed periodically.

Diet information

Animals in both the young and old cohorts were randomly assigned to control (CON) or caloric restriction (CR) diet. While the original CR studies in rodents performed by McCay et al. [8] use a 40% caloric restriction, when designing this study for non-human primates, a 30% reduction in calories was decided upon based on considerations of safety, compliance, and ability to perform the intervention in juvenile animals still in their growth phase when translation of McCay’s seminal work to a larger animal species. The experimental cohorts for this study included: (1) young/CON (young animals on the control diet; N = 4), (2) young/CR (young animals on the caloric restriction diet; N = 4), (3) old/CON (old animals on the control diet; N = 8), (4) old/CR (old animals on the caloric restriction diet; N = 5) (Figure 1 and Table 1). Animals in the control group were provided a daily allotment of food, adjusted for age and body weight. The average daily value was then divided into two meals. The CR diet was then reduced by 10% per month until a 30% CR was achieved as previously described [31, 32, 54, 55]. CR animals received 30% less food than age- and body weight-matched CON animals. Each animal received a measured portion of specially formulated biscuits (Cargill, Minneapolis, MN, USA) supplemented with daily fresh fruits or vegetables (10–40 cal) and was fed twice daily, at 0800 h and 1500 h. Biscuit composition was 15% protein, 5% fat, and 5% fiber, with a caloric content of ∼3.7 kcal/g. To avoid any deficiencies in essential nutrients in the CR animals, the diet was enriched with a vitamin/mineral mix that was 40% higher than the recommended allowance for rhesus macaques by the National Research Council [53]. These animals were calorie restricted but not malnourished. Animals were maintained on a control (CON) or a moderate CR diet for up to 3 years. Biochemical assays were performed periodically and with every new shipment to ensure diet content and quality [56, 57].

Menstrual cyclicity analysis

We used hormone data from the Endocrine Services Laboratory Core, Oregon National Primate Research Center (Roche, Beaverton, OR, USA) to assign animals into groups of those who had regular cycles and those that did not. Steroid (estradiol (E2), progesterone (P4)) and gonadotropin hormones (follicle-stimulating hormone (FSH), luteinizing hormone (LH)) were retrospectively synchronized relative to the estimated mid-cycle LH peak (designated LH d0). This separates the follicular and luteal phases to permit analyses between comparable physiological states. Based on these data, animals were assigned to one of two groups: regular cyclers and irregular cyclers. Regular cyclers exhibited 3 cycles each demonstrating a mid-cycle E2 peak followed by luteal phase levels of P4 during the 90-day interval. Irregular cyclers typically exhibited at least one cycle within the 90-day interval with a mid-cycle E2 peak followed by basal levels of P4 indicating the absence of a luteal phase. The remaining two cycles within the 90-day interval were characterized by very low levels of E2 and P4 lacking cyclic rises, with no further observation of menses. Non-cycling animals displayed very low hormone activity, with no recorded menses noted and little to no change in both E2 and P4. After the 90-day sampling period (between 1–2 years on CR), menses records were reported every month until the completion of the experiment at the end of three years. Using the definition from Gilardi and colleagues [25], an animal was classified as an irregular cycler if she exhibited 6 consecutive months or more of irregular cycles, excluding the non-breeding season (June through August). Data on hormone levels and response to ovarian stimulation will be presented in an accompanying manuscript.

Histological samples

At the end of the study period, animals were sacrificed, and ovaries were harvested. Ovaries were fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) for 24 hours at 4°C, 4% sucrose (Sigma-Aldrich, St. Louis, MO, USA) for 24 h at 4°C, and stored in 70% ethanol (DeCon Labs, Inc., King of Prussia, PA, USA) at 4°C for up to one week prior to dehydration and embedding in paraffin. The ovaries were serially sectioned at 5 μm thickness and mounted onto slides with two sections per slide. As it has been previously demonstrated that the number of primordial follicles did not differ between right and left ovaries in rhesus macaques only one ovary per animal was serially sectioned and used for histological analysis in this study [58]. These slides were used for downstream analyses as described below.

Follicle classification and counting

A subset of histological sections (every 10th slide) was stained with hematoxylin and eosin (H&E) to visualize ovarian follicles and other tissue structures. Samples were deparaffinized in Citrosolv (Decon Laboratories, King of Prussia, PA, USA) and rehydrated in a series of graded ethanol baths (100%, 75%, 70%) and a standard H&E staining protocol was used. H&E-stained slides were imaged using the NanoZoomer Digital Pathology System HT-9600 (Hamamatsu, Hamamatsu City, Japan). Images were then accessed using NDP Viewer Software (Hamamatsu, Hamamatsu City, Japan). Follicle counting was performed in the NDP Viewer software under 40X magnification. To account for variation in ovary size and tissue architecture, three representative sections per animal were selected for analysis at one quarter (1), midway (2), and three quarters (3) through the ovary (Figure 3L). Each section was radially divided into 12 wedge-shaped segments to facilitate counting (Figure 3M). Within each segment, each follicle type was assigned a colored marker based on the classification system described below. Three individual counters counted a section at the same location in the ovary across all animals (e.g., section 1, 2, or 3, Figure 3L). To ensure rigor and reproducibility, for every 12 samples analyzed (top, middle, and bottom section from an animal in each of the four experimental groups), a calibration count was performed as follows. For each section 1, 2, or 3, an individual counter blinded to the original count re-analyzed one wedge-shaped segment from the section. Thus, three wedges per 12 histological samples were re-analyzed. A priori it was established that total follicle counts must be within 5% error, or the entire histological section was reanalyzed by the second counter, and counts were then averaged. No histological sections met the requirement for reanalysis by a second counter. Follicle counts were totaled and reported by follicle class.

The protocol for classifying follicles for counting purposes was adapted for NHP based on categories established during discussions held at the 2021 Ovarian Anatomy Nomenclature Workshop led by the National Institutes of Health [59]. For morphologically normal follicles, a primordial follicle was defined as having an oocyte surrounded by a complete or incomplete layer of squamous granulosa cells (Figure 3A). A transitional primordial follicle was one that has an oocyte surrounded by a complete or incomplete layer of both squamous and cuboidal granulosa cells (Figure 3B). A primary follicle contained an oocyte surrounded by only cuboidal granulosa cells which formed a compact and complete (or nearly complete) layer (Figure 3C). A zona pellucida between the oocyte and granulosa cells was also visible at this stage. A transitional primary follicle consisted of an oocyte surrounded by more than 1, but fewer than 2 complete layer(s) of cuboidal granulosa cells (Figure 3D). Follicles were classified as secondary when the oocyte was surrounded by 2–3 layers of cuboidal granulosa cells and multilayer when surrounded by 3+ such layers (Figure 3E, 3F). Classification was based on the maximum number of layers on any given side of the follicle. Antral follicles were large and contained a fluid-filled antral cavity which spanned multiple sections (Figure 3G). An outer layer of theca cells was typically visible in follicles at the secondary stage and beyond and was distinctly formed in antral follicles. To avoid double counting follicles, antral follicles were not counted in the middle section of each animal. Otherwise, follicles were classified regardless of the presence of an oocyte nucleus, as the sections analyzed were sufficiently far apart.

In addition to morphologically normal follicles, we also classified and quantified follicles with abnormal morphological features. A multi-oocytic follicle contained two or more oocytes within one follicle structure enclosed by a basement membrane (Figure 3H). A secondary or multilayer follicle with abnormal morphological features (secondary AMF, multilayer AMF) contained structural abnormalities in the oocyte and/or granulosa cells, such as the presence of vacuoles in more than 50% of the oocyte (Figure 3I, 3J). Other criteria leading to an AMF classification included a damaged, misshapen, or degenerating oocyte, nucleus, and/or granulosa cells, or the severe shrinkage of the oocyte and/or granulosa cells away from the stroma in a section where most follicles were clearly identifiable suggesting that the phenotype was not a fixation artifact. Atretic antral follicles were defined as those with significant numbers of granulosa cells which had separated from the mural layer and localized within the antral cavity. Atretic antral follicles typically also exhibited other AMF features in the nucleus, oocyte, and/or granulosa cells (Figure 3K). Follicle number per area was calculated by dividing the follicle number by the area of the histologic section, which was measured using Fiji (ImageJ, National Institutes of Health, Bethesda, MD). Standardized follicle number was calculated by summing the follicle number per follicle class in each individual animal across all three tissue sections and normalizing to the total number of histologic slides generated for that animal. To generate the pie charts in Figure 4I–4L, raw follicle number from each follicle class was summed for all animals in a given experimental group and these sums were plotted to show the distribution of follicles from each class. As there are different numbers of animals in each experimental group and different numbers of follicles counted for each animal, each pie chart depicts the follicle distribution for a different number of animals/follicles (for: (I) N = 4 animals, n = 3315 follicles); (J) N = 5 animals, n = 467 follicles); (K) N = 4 animals, n = 3448 follicles; and (L) N = 5 animals, n = 336 follicles). Animals without follicles were not included in pie charts.

Picrosirius red staining and collagen quantification

For assessment of collagen I and collagen III, ovarian sections (N = 4–8 animals/group, N = 6 sections/ovary) were stained with Picrosirius Red (PSR) according to established protocols [36]. Briefly, tissue sections were deparaffinized in Citrosolv (Decon Laboratories) and then rehydrated in a series of graded ethanol baths (100%, 70%, and 30%). Slides were submerged in a PSR staining solution, which was prepared by dissolving Sirius Red F3B (Direct Red 80, Color Index 35780, Sigma-Aldrich, St. Louis, MO, USA) in a 1.3% saturated aqueous picric acid solution (Sigma-Aldrich, St. Louis, MO, USA) at 0.1% w/v. Slides were immersed in the PSR staining solution for 30 minutes at room temperature, then destained with acidified water. Samples were then rapidly dehydrated in 100% ethanol, cleared in xylene, and mounted with synthetic mounting media. Quantitative digital microscopy was used to analyze the collagen content in the PSR-stained tissue sections as previously described [38]. Stained slides were imaged using the NanoZoomer Digital Pathology System HT-9600 (Hamamatsu, Hamamatsu City, Japan). Images were then imported into Visiopharm software (Visiopharm, Hoersholm, Denmark) and analyzed using the Image Analysis Module. Three slides (top, middle, and bottom) with two adjacent sections were analyzed for each animal. Regions of interest (ROIs) were drawn around each ovarian tissue section and assigned to “right” or “left” ovary. Then, these images were converted into grayscale values using the Chromaticity Green feature band. The software labeled each pixel as low, medium, or high degree of staining based on thresholds used in our prior work [38]. Images were processed in batch mode using this algorithm. The area of pixels with low, medium, and high staining intensity was calculated per total ovarian area and averaged between left and right ovarian sections.

Hyaluronic acid binding protein assay and hyaluronic acid quantification

A hyaluronic acid binding protein (HABP) assay was performed to localize hyaluronic acid within histological sections of ovarian tissue as previously described [39, 60]. In brief, tissue sections were deparaffinized in Citrosolv (Decon Laboratories) and then rehydrated in a series of graded ethanol baths (100%, 95%, and 95%). Slides were incubating in running reverse-osmosis water for 1 minute and washed in 1X phosphate buffered saline (PBS) with gentle rocking for 10 minutes. An Avidin/Biotin Blocking Kit (Vector Laboratories, Burlingame, CA, USA) was used according to kit instructions to block endogenous avidin and biotin within the tissue sections. Avidin was applied for 15 minutes at room temperature. Unbound avidin was rinsed off using 1X PBS, then biotin was applied for 15 minutes at room temperature. Unbound biotin was rinsed off using 1X PBS and slides were subsequently incubated in normal goat serum (Fisher Scientific, Waltham, MA, USA) for 20 minutes. Next, 1 mg/mL hyaluronidase (Sigma-Aldrich, St. Louis, MO, USA) in saline solution was applied to negative control tissue sections and saline alone was applied to experimental sections. Tissue sections were incubated in a humid chamber for 1 hour at 37°C. After rinsing in 1X PBS, biotinylated HABP (Calbiochem, San Diego, CA, USA) diluted in normal goat serum was applied to all sections and incubated for 1 hour at room temperature. Slides were again washed in 1X PBS. To amplify signal, slides were incubated in ABC reagent (Vector Laboratories, Burlingame, CA, USA) for 30 minutes at room temperature, followed by TSA Plus Fluorescein System (Akoya Biosciences, Marlborough, MA, USA) at a 1:400 dilution for 5 minutes at room temperature. All incubations were performed within a humified chamber. Samples were mounted in Vectashield HardSet Antifade Mounting Medium with DAPI (4′,6-diamidino-2-phenylindole; Vector Laboratories, Burlingame, CA, USA) to counterstain cell nuclei. Sections were processed in batches containing one slide from each animal. The HABP assay was performed on three replicates from each animal. Entire ovarian tissue sections were scanned at 10X using a Leica DM6B Fluorescent Microscope (Leica Biosytems, Deer Park, IL, USA). The imaging settings were kept constant for all samples after determining the threshold. As a control, the second tissue section on the slide was treated with hyaluronidase and we confirmed that the hyaluronidase-treated samples did not show a positive signal and used this to help set the threshold. HABP staining intensity analysis was performed using Fiji (ImageJ). Hyaluronic acid staining is reported as a ratio of HABP intensity per area in the total ovarian section. The background was determined using the hyaluronidase-treated sections, and this value was subtracted from the hyaluronic acid positive sections.

Statistical analysis

Data were analyzed using Prism Software version 9.3.1 (GraphPad, La Jolla, CA, USA) and R version 4.4.2. Descriptive statistics expressed as the mean ± standard error of means (SEM). Visual inspection demonstrated follicle number per area and standardized follicle number data to be skewed and contained clusters of zeros. Therefore, the original data set was log transformed. For outcomes with positive values, the natural logarithm was taken directly. For outcomes with zero values, a small constant (half of the smallest positive outcome value) was added before applying the logarithm to avoid undefined values. We used unpaired t-tests to determine significance between two groups and one-way analysis of variance (ANOVA) with a Tukey’s post hoc test to determine significance between three or more groups. To evaluate the effect of two independent variables (diet and age or menstrual cyclicity) on follicle number, two-way ANOVA was used. Levene’s test for homogeneity of variances was performed for all two-way ANOVA analyses; all models met the homogeneity assumption except for one (multilayer follicle number per area), which showed borderline heteroscedasticity (p = 0.042). Tukey’s post hoc test was applied when one or both of the main effects were statistically significant. To compare the follicle distribution pie charts in Figure 4, Chi-square test was used to compare the observed distribution for each experimental group against the average distribution of young control and young calorie-restricted animals. For all statistical analyses, significance was set at unadjusted or adjusted p < 0.05. In all figures, data are plotted untransformed and statistical tests are performed on log transformed data. P-values for main effects and interaction testing from two-way ANOVAs are shown in Supplementary Tables 1, 3, 5, and 6. Statistically significant pairwise comparisons are shown visually in the figures with asterisks signifying level of significance.