A total of 30 Fisher 344 rats were used across three experimental groups: young (n = 10), OVX (n = 10), and aged (n = 10). Fisher-344 rats are the considered the “standard” model for investigations of aging given their long lifespan and consistency across colonies (Holmes, 2003). Further, F344 rats are bred by the National Institute on Aging (National Institute on Aging, n.d.) specifically for use in aging-related studies. Here, female Fischer-344 rats were procured from two sources. The aged rats were acquired from the NIA Aged Rat Colony, and rats in the remaining groups (young proestrus and OVX) were purchased from Charles River (Kingston, NJ) which is also the source for the NIA F344 colony. Rats were housed in an AAALAC accredited vivarium; each cage contained two rats with ad libitum access to food and water. Light cycles were regulated on a 12:12 h light-dark cycles, shifting phases at 6:00 and 18:00 h. Rats in the young and OVX groups were 3–5 months old and randomly assigned to their groups. The aged rats were >20 months old and determined to be acyclic (i.e., in persistent diestrus; details below) to model a post-menopausal state (Sengupta, 2013). All experimental protocols were approved by the Institutional Animal Care and Use Committee at the University of Minnesota (approval no. 2304-40947A).

Surgical procedures were detailed previously (Dougherty et al., 2017; Barok et al., 2021; Miller et al., 2022). Two hours prior to the start of surgery, sustained-release buprenorphine (1 mg/kg; Wedgewood Pharmacy LLC, Laramie, WY, United States) was injected subcutaneously for pain management. Rats were sedated using isoflurane (Piramal, Telangana, India) initially in a closed chamber, and then transferred to a nose cone (2–3% isoflurane in 100% O₂) for the duration of the surgery. An adequate plane of anesthesia was confirmed by the absence of both toe pinch and eye blink reflexes. Rats were placed prone on a heated surgical table, shaved of fur over the surgical area, and skin scrubbed clean (chlorhexidine scrub). Using a dorsal approach, 10–12 mm bilateral incisions were made through the skin and muscle revealing the ovarian fat pads. Ovaries were exteriorized, removed via cautery, and muscle layers were sutured using 4-0 absorbable suture (Covidien, Mansfield, MA). Skin incisions were sealed using 9 mm wound clips (MikRon, Gardena, CA). Rats recovered in their home cages and were monitored for 72 h for surgical complications. Data collection occurred a minimum of 10 days after OVX.

Cytology was used to determine estrous cycle for the young and aged groups. Daily vaginal swabs were performed on all rats for identification of estrous cycle stage. Vaginal cells were examined under light microscopy and cell characteristics and cycle stage determined by two trained judges (Marcondes et al., 2002; Goldman et al., 2007; Dougherty et al., 2017). Data for the young group were gathered during the proestrus phase of the rodent estrous cycle, when circulating estrogen levels are at their highest (Levine, 2015). Both OVX and aged rats were staged to verify cessation of the estrous cycle. Cessation of the estrous cycles was classified as >2 cycles (10 days) of persistent diestrus (Barok et al., 2021). Circulating estrogen levels were subsequently confirmed using an ELISA assay (described below).

Breathing data was assessed in unanesthetized, freely behaving rats using whole body barometric plethysmography (Buxco, Data Sciences International, St Paul, MN, United States). Bias air flowed at a rate of 2.5 L/min through the 4.0 L chambers, and the pressure was continuously sampled at 500 Hz, temperature, humidity, and barometric pressure were sampled at 10 Hz (Data Sciences International, St Paul, MN, United States). Equipment was calibrated daily per the manufactures instructions to account for fluctuations in barometric pressure (McIntosh and Dougherty, 2019; Barok et al., 2021; Miller et al., 2024). All data was collected during the same time of day (0900-1300) to minimize circadian effects (Stephenson et al., 2001; Marciante et al., 2023). The plethysmography protocol consisted of 1 hour of acclimation, followed by baseline data collection in normoxia (20.9% O₂, balance N₂), followed by three 15-minute intervals of stepwise hypoxia, totaling 45 min of hypoxia for each rat (Morgan et al., 2014). The hypoxia steps were as follows: 15% O₂, balance N₂, 12% O₂, balance N₂, 9% O₂, balance N₂. A customizable, gas mixer (GSM-3; CWE, Ardmore, PA, United States) was used to automate the protocol and minimize time between steps. From those traces respiratory frequency (ƒR), inspiratory time (Ti), and expiratory time (Te) were directly recorded. Data was analyzed using Ponemah data acquisition software (Data Sciences International, St Paul, MN, United States). Gas volumes (e.g., tidal volume) were calculated according to the Drorbaugh and Fenn equation, which incorporates chamber and body temperature. Tidal volume (VT) and VE were calculated by Ponemah using temperatures of both the rat and chamber, humidity, and normalized weight per 100 g (Mortola et al., 1994). Body temperature was taken via rectal thermometer just prior to placing the rats in the chambers and immediately following the last hypoxia. Baseline body temperature was used for baseline, FIO₂ 0.15 and 0.12 calculations; final body temperature was used to for FIO₂ 0.09 analyses. Five minutes of baseline breathing was selected during a period of rest (not moving, eyes closed) for analysis by an independent analyzer blinded to experimental group designation. Two-minute segments of hypoxic breathing were selected for analysis near the end of each step of hypoxia. During this period of hypoxia, rats were typically awake but holding stable postures. Segments were selected to avoid gas mix transitions and large movement artifacts. Additional, smaller artifacts from sniffing, grooming, or movement were excluded by a trained judge (McIntosh and Dougherty, 2019; Barok et al., 2021; Miller et al., 2024).

Expired gas exiting the chamber was sampled to measure carbon dioxide production (VCO₂) production. For this, a sampling tube was connected to the expired port of the plethysmograph chamber to divert ∼300 mL/min of gas to a respiratory gas analyzer (Gemini, CWE, Ardmore, PA) via subsampler pumps (Sable Systems, North Las Vegas, NV) and an inline multiplexer (Sable Systems, North Las Vegas, NV). Expired gas data were integrated into Ponemah software to enable real-time VCO₂ measures throughout the plethysmography sessions. Chambers were not equipped for SpO₂ acquisition. As such, the current study deviated from Morgan et al. (2014) by using FIO₂ as a proxy for SpO₂.

Serum estradiol levels were confirmed in a subset of animals from each group (young n = 7, OVX n = 6, aged n = 7). Blood samples were collected after completion of plethysmography data collection. Blood was stored for 24 h at −4°C then centrifuged at 3,000 rpm for 10 min. The serum was stored at −20°C until analysis. Serum levels of estradiol were quantified using ELISA kits from Biovendor (Ashville, rat-estradiol, catalogue no. RTC009R, sensitivity 2.5–1,280 pg/mL). If a sample’s concentration was found to be below the level of detection, a value of 2.5 was entered for analysis. Assays were completed according to manufacturer’s instructions and samples were analyzed in duplicate using a microplate reader (absorbance 450 nm). Concentrations were interpolated from standard curves run in each assay using the microplate reader software (Molecular Devices, San Jose, CA 95134).

All statistics were completed using Graphpad Prism software (Dotmatics, San Diego, CA). One-way ANOVAs were conducted to examine differences between group ages, weights, circulating estradiol (E2), and body temperature. One-way ANOVAs were also used to compare baseline measurements of frequency (ƒR), VT, VE, VO₂, and VCO₂, neural drive (VT/Ti), and minute ventilation controlling for metabolism (VE/VCO₂). Responses to hypoxia were analyzed using three methods. In the first approach, two-way ANOVAs were performed for between and within group comparisons of ventilation response (ƒR, VT, VE, VO₂, VCO₂, VT/Ti, VE/VCO₂, and changes to body temperature) across different levels of hypoxia. In the second approach VE, VT/Ti/VCO₂, and VE/VCO₂ for each rat were graphed against FIO₂. Individual slopes were calculated, and the grouped slopes were compared using one-way ANOVA. In the final analysis, the percentage change of output between baseline and maximal hypoxia (FIO₂ 0.09) was calculated for each animal, and groups were compared using one-way ANOVA. Tukey’s multiple comparisons post hoc analyses were conducted when significance was found. Results were considered significant if p-values were less than or equal to 0.05.

No significant group differences were observed in baseline frequency (Table 2; Figure 1B; p = 0.26), VT (Figure 1C; p = 0.33), or inspiratory time (Figure 1E; p = 0.78). Groups differed significantly in baseline VE (Table 2; Figure 1D; p = 0.03). The young group showed a significantly higher VE compared to the aged group (p = 0.04), and the OVX group had a higher mean VE than the aged group that did not reach significance (p = 0.06). There was a significant group interaction effect for VCO₂ (p > 0.0001), and post-hoc analyses revealed that the aged group had a significantly decreased VCO₂ during normoxia compared to both the young (p = 0.0001) and OVX groups (Table 2; Figure 1F; p > 0.0001). We reanalyzed minute ventilation controlling for CO₂ production (VE/VCO₂) and found a significant group effect in basal ventilation (p = 0.005); aged rats had significantly higher baseline VE/VCO₂ compared to the young (p = 0.0024) and OVX groups (Table 2; Figure 1G; 0 = 0.0012). There were no group differences in baseline VT/Ti, which is used as a proxy for respiratory neural drive (Table 2; Figure 1H; p = 0.31). However, when analyzed relative to metabolic rate, a significant group interaction effect emerged. (VT/Ti/VCO₂; Table 2; Figure 1I; p = 0.0004). Rats in the aged group showed a significantly higher VT/Ti/VCO₂ compared with young (p = 0.003) and OVX (p = 0.0006) groups.

To determine how loss of circulating steroid hormones impacted the HVR and chemosensitivity in female rats, we exposed all three experimental groups to progressive hypoxic challenges (FIO₂ 0.15, 0.12, and 0.09). Stepwise hypoxia increased ventilation in all experimental groups (Table 2; Figure 2). Breathing frequency showed a significant effect for gas condition (Figure 2A; p < 0.0001) and group (p = 0.014). All groups showed significantly increased rate from baseline to the maximal hypoxic challenge (FIO₂ 0.09; Figure 2A; young p < 0.0001; OVX p = 0.0001 aged p < 0.0001), and the OVX group had a significantly higher frequency than the aged group only at FIO₂ 0.09 (p < 0.01). Inspiratory time (Ti) decreased as O₂ concentration decreased for all groups (Figure 2B; p < 0.0001). There were no between group differences in Ti (p = 0.399).

Significant main effects for gas condition were also observed in measures of tidal volume (Figure 2C; p < 0.0001) with no between group differences (p = 0.58). Post-hoc analyses showed a similar pattern of change across the groups. All three groups demonstrated a significant decline in VT between baseline and FIO₂ 0.15 and 0.12, but at FIO₂ 0.09 VT returned to baseline values. Assessment of minute ventilation (VE) showed a significant main effect in response to progressive hypoxia (Figure 2D; p = 0.02). Gas condition caused significant increases in VE for all groups as FIO₂ declined (p < 0.0001), and a significant group effect was revealed (p = 0.002). Specifically, the aged group displayed smaller VE than the young group and OVX group at baseline (young p = 0.03, OVX p = 0.02) and at FIO₂ 0.09 (young < 0.001, OVX < 0.0001). At FIO₂ 0.12, the aged group had a smaller VE than the OVX group (Figure 2D; p = 0.04).

Neural drive (VT/Ti) significantly increased with exposure to hypoxia for all groups (Figure 2E; p < 0.0001). The effect of group was also significant for VT/Ti (p = 0.05). The aged group showed a subdued response compared to the OVX group at FIO₂ 0.12 (p = 0.04) and compared to both the young and OVX groups during the most severe hypoxia (young p = 0.003, OVX p = 0.0002).

Previous research has strongly suggested that metabolic inhibition occurs during exposure to hypoxia (Gautier and Bonora, 1992; Mortola et al., 1994; Mortola and Maskrey, 2011; Morgan et al., 2014). Our VCO₂ showed a significant group × condition interaction effect (Figure 2F; p = 0.01), a significant effect of gas condition (p < 0.0001), and a significant group effect (p < 0.0001). The young and OVX groups’ VCO₂ declined over the course of hypoxia, but the aged group did not change VCO₂ in response to hypoxia. For the young group, the decline in VCO₂ reached significance by FIO₂ 0.12 (p = 0.01), and the OVX group reached significant decline by FIO₂ 0.09 (p = 0.01). The aged group started with a lower VCO₂ at baseline as compared to the young (p = 0.0008) and OVX (p < 0.0001) groups. Hypoxia-induced declines in VCO₂ observed in the young group brought VCO₂ within range of the aged group at each level of hypoxia (FIO₂ 0.15 p = 0.19, FIO₂ 0.12 p = 0.08, FIO₂ 0.09 p = 0.07). However, VCO₂ in OVX rats remained higher than the aged group at all data points (FIO₂ 0.15 p = 0.001, FIO₂ 0.12 p = 0.006, FIO₂ 0.09 p = 0.0001).

We next calculated both neural drive and minute ventilation relative to VCO₂ (VE/VCO₂ Figure 2G and VT/Ti/VCO₂ Figure 2H). Our analysis of VE/VCO₂ revealed that all experimental groups significantly increased VE/VCO₂ in response to stepwise hypoxia (Figure 2G; p < 0.0001). Although there were baseline differences (noted above, Figure 1C), there were no between group differences over the course of stepwise hypoxia (p = 0.11). All three groups had a significant increase from baseline by FIO₂ 0.12 and similar patterns of change (Figure 2G).

When metabolism was considered in the context of neural drive, we found a significant main effect (p = 0.04), a significant effect of gas condition (p < 0.0001), and a significant group effect (Figure 2H; p = 0.01). Exposure to stepwise hypoxia increased neural drive for all groups. The increase in VT/Ti/VCO₂ reached significance by FIO₂ 0.12 for all groups (Figure 2H; young p = 0.04; OVX p = 0.02; aged p = 0.008). Post-hoc analyses of the group effect showed that the aged group had increased VT/Ti/VCO₂ relative to the other groups at baseline (young p = 0.02, OVX p = 0.003), and compared to the OVX group at the beginning of stepwise hypoxia (p = 0.0003), but all three groups had similar output as the hypoxia became more severe.

Our fundamental question for the current study was whether sensitivity and dynamic range of the respiratory system was impacted by estrogen loss in two unique experimental models. In other words, we wanted to know whether the rate of change was faster or slower between groups (i.e., sensitivity), and we wanted to know whether the magnitude of change was larger or smaller between groups (i.e., dynamic range). Figure 3 has illustrations based on patterns seen in the rate of change and magnitude of change data. The illustration in Figure 3A shows how two animals could have different rates of change but similar magnitudes of response to hypoxia. The top, red line represents an animal that had immediately increased breathing output in response to increasing hypoxia, whereas the lower, blue line represents an animal that was slower to respond to increasing hypoxia. In other words, the red line shows higher sensitivity to the stimuli than the blue line. This sensitivity is captured by computing the slope of the data (represented as dotted gray lines). For our analyses, the slope (or rate of change) was calculated for each animal and then group analyses were completed. Another aspect of Figure 3A is that both lines begin and end at the same points, showing that even with dissimilar sensitivities to the stimuli, there can still be similar magnitudes of change from beginning to end. Figure 3B illustrates two hypothetical animals that had very similar sensitivity to the stimuli (the slopes of the lines are similar), but the top red line represents an animal that produced a larger amount of change from start to finish (i.e., larger magnitude of change) as compared to the lower blue line. In our analyses, the percent change from baseline to maximal hypoxia was calculated and then group analyses were carried out.

Utilizing the “Gold-Standard” analyses described in Morgan et al. (2014) to assess HVR and chemosensitivity, we defined each animal’s HVR as the slope of VE. Since metabolism is influenced by a variety of factors including age and concentration of sex hormones, and therefore may differ between our groups, we used VCO₂ as a factor in calculations for neural drive and chemosensitivity. The neural drive response to hypoxia was defined as the slope of VT/Ti/VCO₂. Chemosensitivity was defined as the slope of VE/VCO₂ in response to stepwise hypoxia. One-way ANOVAs were used to compare the grouped slopes of respiratory output as the animals were exposed to increasing levels of hypoxia. One-way ANOVA of HVR showed that the rates of change were significantly different between the groups (Figure 4A; p = 0.01). Post-hoc analyses showed that there was a significant difference between the OVX and aged groups (p = 0.01) but not the young and aged groups (p = 0.28). Other results indicated that the aged group showed metabolic differences from the young and OVX groups. Once the individual differences in VCO₂ measures were included, one-way ANOVAs of neural drive sensitivity (VT/Ti/VCO₂ slope) and chemosensitivity (VE/VCO₂ slope) showed no group differences in rates of change (Figure 4B, VT/Ti/VCO₂ slope p = 0.18; Figure 4C, VE/VCO₂ slope p = 0.24). Consistent with prior studies (Lhuissier et al., 2012; MacNutt et al., 2012), the variability in these measures was notable, and we opted to show the individual data points for Figure 4 to highlight this range of responses.

We then compared the magnitude (% change) of responses between baseline and maximal hypoxia (Figure 5). There were no significant group differences in the magnitude of the frequency response to hypoxia (Figure 5A; p = 0.34), VT (Figure 5B; p = 0.81), or VE (Figure 5C; p = 0.2). As suggested by the data in Figure 2F, the magnitude and direction of VCO₂ changes were significantly different between the aged rats and other groups (Figure 5D; p = 0.001). Accordingly, significant group differences were seen in percent change of VE/VCO₂ (Figure 5E; p = 0.0009). Although the groups showed similar sensitivity in their rates of change (Figure 4C), the aged group had a smaller magnitude of change because they started from a higher baseline VE/VCO₂ (Figures 2G, 3C) compared to young (p = 0.002) and OVX (p = 0.005) groups. The magnitude of neural drive change, controlling for metabolism (VT/Ti/VCO₂; Figure 5F) was significantly different between groups (p = 0.003). Post-hoc analyses showed the aged group had a smaller magnitude of VT/Ti/VCO₂ change in response to stepwise hypoxia than the young group (p = 0.003) and the OVX group (p = 0.03). Collectively, the aged rats started at a higher output and remained high, resulting in a smaller magnitude of change than the young and OVX groups.