https://orcid.org/0000-0003-2944-0475
Abstract
Hyperandrogenemia and polycystic ovary syndrome are a result of the imbalance of androgen levels in females. Androgen receptor (Ar) mediates the effect of androgen, and this study examines how neuronal Ar in the central nervous system mediates metabolism under normal and increased androgen conditions in female mice. The neuron-specific ARKO mouse (SynARKO) was created from female (Ar fl/wt; synapsin promoter driven Cre) and male (Ar fl/y) mice. A glucose tolerance test revealed impaired glucose tolerance that was partially alleviated in the SynARKO-dihydrotestosterone (DHT) mice compared with Con-DHT mice after 4 months of DHT treatment. Heat production and food intake was higher in Con-DHT mice than in Con-veh mice; these effects were not altered between SynARKO-veh and SynARKO-DHT mice, indicating that excess androgens may partially alter calorie intake and energy expenditure in females via the neuronal Ar. The pAkt/Akt activity was higher in the hypothalamus in Con-DHT mice than in Con-veh mice, and this effect was attenuated in SynARKO-DHT mice. Western blot studies show that markers of inflammation and microglia activation, such as NF-kB p-65 and IBA1, increased in the hypothalamus of Con-DHT mice compared with Con-veh. These studies suggest that neuronal Ar mediates the metabolic impacts of androgen excess in females.
Androgens, the major sex hormones in males exerting physiological actions through androgen receptors (ARs), are normally present in very low levels in females. In women, hyperandrogenism with reproductive dysfunction is seen in polycystic ovary syndrome (PCOS), which also features obesity and metabolic dysfunction.
To model the effects of androgen excess, several animal models have been developed in which hyperandrogenemia and overweight/obesity coexist. These models include high-dose dihydrotestosterone (DHT) [1‐3], letrozole treatment [4‐6], and dehydroepiandrosterone [7, 8] as well as prenatal androgen exposure [9]. Our low-dose DHT mouse model [10, 11] displays normal body mass/composition, which enables us to study the effects of hyperandrogenemia (2-fold DHT of controls) unencumbered by the effects of obesity.
Research on the effects of androgen on glucose metabolism and insulin sensitivity in women is of crucial importance, considering the increasing incidences of metabolic syndrome in women with hyperandrogenemia [12‐15]. In females of different species, including humans, hyperandrogenism is associated with glucose intolerance, insulin resistance, and increased risk of metabolic syndrome, with altered insulin signaling in the peripheral organs, like the liver, adipose tissue, and pancreas [16‐22]. It has been well recognized that, besides the peripheral tissues, the hypothalamus has a major role in the development of insulin resistance and metabolic syndrome [23‐26]. In males, androgens, via the neuronal AR play a protective role to maintain insulin sensitivity and glucose homeostasis [26]. Recent studies address the role of neuronal Ar in rodents using various Cre lines [27‐29]. Global knockout (KO) of Ar in females rescued the DHT-associated metabolic phenotype [28]. Double KO of adipose and brain Ar with FAPB-4–driven Cre rescued adiposity, adipose hypertrophy, and hepatic steatosis associated with DHT implants in female mice starting at 3 weeks of age [27]. CAMKII-driven Cre induced neuronal Ar KO prevented DHT-induced increased body weight (BW) and dyslipidemia and partially alleviated fatty liver, but maintained the impaired fasting glucose [29]. This prevention of the effects in the absence of neuronal Ar in this model could be from either the direct effects of Ar or the secondary effects due to the lower BW of Ar KO. To understand the role of neuronal AR in glucose homeostasis and energy expenditure and eliminate the secondary effects by BW, we treated mice with low dose DHT and used a neuron-specific Ar KO mouse under the control of synapsin1 promoter (SynARKO). ARKO mice using Synapsin 1 Cre as promoter were previously studied in male metabolism [26, 30].
Materials and Methods
Generation and Genotyping of SynArKO Mice
Neuron-specific Ar KO mice were previously created using Cre recombinase under the control of the Synapsin1 promoter and were verified [26]. Briefly, a male floxed ar mouse [31] was mated with a female Synapsin1 Cre mouse (Stock#: 003966, Jackson Laboratory, ME, USA) to generate a Synapsin Cre female (ar fl/wt; SynCre+/−). Then a Synapsin Cre female (ar fl/wt; SynCre+/−) was mated with a floxed ar male (ar fl/y, Cre–) to obtain SynArKO mice (ar fl/fl, Syncre+/−). Female littermates with genotypes (fl/fl, SynCre–; fl/wt; SynCre–) were used as controls. Genomic DNA isolation and primers to detect the ar gene were described previously [31]. Mice that spontaneously showed a KO band in the tail of the ar gene were eliminated and not used in any studies. Synapsin1 promoter–driven Cre primers used were forward: GTTTACGCTACCCCGTGCTC and reverse: CATCTTCAGGTTCTGCGGGAAACC. Validation of Ar KO in central nervous system is performed with Western blot and immunohistochemistry, which are described in detail in the following methods.
KO Ar using the synapsin1 promoter leads to deletion of AR in the neurons of brain. This study mainly focuses on the hypothalamus, the primary brain region controlling metabolism and reproduction, which is disturbed by hyperandrogenemia.
Female mice were maintained on normal chow (Pico lab Rodent Diet 20 # 5053; nutrients: protein, 20%; fat, 4.50%; crude fiber, 6.00%; ash, 7.00%; moisture, 12.0%; gross energy, 4.11 kcal/g) and water ad libitum under a 12-hour/12-hour light/dark cycle. All procedures were performed with the approval of the Institutional Animal Care and Use Committee at Temple University and Johns Hopkins University School of Medicine.
Hyperandrogenemic Mouse Model
Two-month-old female mice were subcutaneously implanted with a 4-mm pellet with or without DHT. Production of the DHT pellets has been previously described [10, 11]. Dow Corning Silastic tubing (0.04 mm inner diameter and 0.085 mm outer diameter; Fisher Scientific, Hampton, NH) was filled with 4-mm lengths of DHT powder and then sealed with medical adhesive silicone (Factor II, Lakeside, AZ). Pellets were incubated in saline for 24 hours at 37 °C for equilibration before insertion. Pellets were replaced monthly to maintain a constant level of androgen excess.
Grouping of Mice
To understand the role of neuronal AR in female metabolism under normal and hyperandrogenemic conditions, mice were genotyped and then littermates were randomly divided into 4 groups: control-vehicle (Con-veh), control-DHT (Con-DHT), Syn ARKO-vehicle (SynARKO-veh), and Syn ARKO-DHT (SynARKO-DHT).
The series of experiments carried out on the mice is discussed in detail in the following sections and a timeline of the study is shown in Fig. 1A.
Timeline of study and confirmation of neuron-specific disruption of the androgen receptor. (A) Timeline of the study. The study involved 2 conditions: control and SynARKO mice under physiological androgen (veh) and DHT. The DHT pellet was inserted at 60 days of age and replaced every 4 weeks. (B) A representative Western blot of Ar in the hypothalamus, ovary, pituitary, and liver lysate in control and SynARKO mice. Beta actin was used as a loading control. (C) Densitometric analysis of Ar protein in the hypothalamus, ovary, pituitary, and liver tissues. Statistical analysis was performed using the Student t-test with 2-tailed comparison (P < .05). n = 7. Data are mean ± SEM.
Body Weight and Magnetic Resonance Imaging Body Composition Measurement
Body composition of the control and SynARKO mice, vehicle and DHT treated (3.5 months following DHT treatment), was measured as described previously [31]. Briefly, each mouse was loaded into an Echo magnetic resonance imaging (EchoMedical Systems, USA) system to measure the percentage of fat, lean, and water mass. All measurements were conducted in the morning. The lean mass and body composition was done on a separate cohort of mice shown in Fig. 1A.
Glucose, Pyruvate, and Insulin Tolerance Test
Glucose and pyruvate tolerance tests were used to measure whole-body glucose utilization and hepatic gluconeogenesis (conversion of pyruvate to glucose), respectively. The insulin tolerance test (ITT) and homeostatic model assessment for insulin resistance (HOMA-IR) were done to measure whole-body insulin resistance.
The glucose tolerance test (GTT) and pyruvate tolerance test (PTT) were performed at 2 and 4 months after DHT treatment. In general, GTT and PTT were performed 5 days apart. For GTT, animals were fasted for 16 hours, and dextrose 2 g/kg was injected intraperitoneally. For PTT, pyruvate 2 g/kg was injected intraperitoneally in mice fasted for 13 hours. ITT was performed in 4-month DHT-treated animals after GTT and PTT. Insulin (0.3 IU/kg) was injected into 6-hour fasted mice. Glucose levels following the above tests in blood were measured at 0, 15, 30, 60, 90, and 120 minutes with a glucometer (One touch Ultra, Johnson and Johnson, NJ, USA). The glucose levels from the above experiments are presented in mg/dL and the area under the curve was calculated for each animal [10]. The HOMA-IR index was calculated using fasting blood glucose and insulin levels in a separate cohort of animals.
Comprehensive Laboratory Animal Monitoring Systems Study
To measure whole-body energetics, in other words, oxygen consumption, carbon dioxide production, respiratory exchange rate (RER; ratio between the volume of CO2 being produced and the amount of O2 consumed), heat production (energy expenditure: heat [calorific value × VO2] calculated using the Oxymax software), ambulatory activity, and food intake, mice were subjected to indirect calorimetry Comprehensive Laboratory Animal Monitoring Systems (CLAMS) study (Columbus Instruments, Columbus, OH). Mice were acclimated for 1 day in the metabolic cages and then monitored for next 48 hours under a 12-hour light/12-dark cycle in a CLAMS unit [32]. While in the CLAMS unit, mice were fed with a ground version of the Pico Lab Rodent Diet 20 (#5053, gross energy, 4.11 kcal/g). Food intake was converted into kcal/48 hours, based on the calorie specifications of our diet. To adjust for BW, an analysis of covariance was performed on heat production [33]. Data are expressed as light, dark, or total phase (both dark and light).
Immunohistochemistry
Mice were perfused with 4% paraformaldehyde; their brains were collected and fixed in 4% paraformaldehyde solution for 24 hours, after which they were transferred to a 30% sucrose solution. The entire brain was cryosectioned at 40 µm thickness and used for further immunostaining and analysis. To see the localization of Ar and Ionized calcium binding adaptor molecule 1 (IBA1) in the brain tissue, immunocytochemistry was performed on the perfused brain cryosections. After washing the cryosections with phosphate-buffered saline, permeabilization was done with 1% triton (Cat. T8787 Sigma Aldrich, MO) for 10 minutes followed by 20 minutes of bloxall (to inactivate endogenous peroxidase, pseudoperoxidase, and alkaline phosphatase) (Cat. P-6000 Vector Laboratories, CA) incubation. Cryosections were blocked with rabbit serum for 1 hour and incubated with anti-Ar (Ab227678, Abcam, MA, RRID:AB_2833098) and anti-IBA1 (019-19741, Wako chemicals, VA, RRID:AB_839504) antibodies overnight at 4 °C in a 1:100 and 1:500 dilution, respectively. Next morning, tissue sections were kept at room temperature for 30 minutes and incubated with biotinylated antibody (PK4001, Vector Laboratories, CA, RRID:AB_2336810) using Avidin-Biotin Complex Kits (ABC) according to the company's protocol. 3,3′-Diaminobenzidine (Cat. SK4100, Vector Laboratories, CA) was used to perform the color reaction and counterstained with hematoxylin (Cat. 51275 Sigma Aldrich, MO). Finally, brain sections were mounted using Vectamount Mounting medium (Cat. H-5000, Vector Laboratories, CA) and imaged under a light microscope (Nikon Eclipse E800). We used the free “Fiji” version of ImageJ from http://fiji.sc for the quantification analysis of IBA1. Briefly, we used the color deconvolution function followed by the quantification of the 3,3′-diaminobenzidine image according to a previously published protocol [34].
Western Blot
Mice were sacrificed, followed by tissue extraction of the hypothalamus, pituitary, ovary, and liver. All samples were immediately frozen in liquid N2. Extraction of protein from tissues, measurements of protein concentration, and Western blot analysis of protein expression were performed as described previously [35]. Briefly, samples were homogenized with RIPA buffer and whole cell extracts were obtained. Ar, IBA1, and β-actin (MA5-15739, Invitrogen, MA, RRID:AB_2537664) antibodies were used. In another group, hypothalamus/brain was collected from the respective groups of fasted mice and incubated ex vivo with insulin (100 nM for 15 minutes) at 37 °C. Tissues were pulverized under liquid nitrogen and the cytoplasmic and nuclear fractions were separated and collected by using Nuclear and Cytoplasmic Extraction Reagents (Cat # 78833, Thermofisher, MA). Protein concentrations were determined, and equal amounts of proteins were loaded for sodium dodecyl sulfate polyacrylamide gel electrophoresis. Immunoblotting was done for P-Akt Serine 473 (9271, Cell signaling, MA, RRID:AB_329825); and Akt(pan) (4691, Cell signaling, MA, RRID:AB_915783); β tubulin (MA5-16308, Invitrogen, MA, RRID:AB_2537819) was used as a loading control. Anti-NF-kβ p-65 (# 8242, cell signaling, MA, RRID:AB_10859369) immunoblotting was done on nuclear extract, using HDAC1 (#ab68436, Abcam, MA, RRID:AB_1860585) as loading control. Respective secondary antibodies were used, and imaged with an Odyssey Imager CLx (LICOR Biosciences, USA)
Statistical Analysis
Statistical analysis was done using the unpaired Student's t-test when comparing 2 groups of data. A 2-way analysis of variance (2-way ANOVA) was used to determine the effects of DHT, genotyping (G), and interaction between the 2 factors (DHT × G); post hoc analysis was done with Tukey's test. All results are expressed as mean ± SEM and P ≤ .05 was considered to be statistically significant using GraphPad Prism 9.0 software (San Diego, CA, USA).
Results
Ar KO in the Neurons
Disrupted Ar expression was determined by Western blot and immunostaining. Ar protein levels were examined by Western blot (Fig. 1B and 1C) and the protein levels were significantly reduced to 78% in the brain of SynARKO mice compared with those in control mice (Fig. 1B and 1C). Other tissues such as pituitary, ovary, and liver have equal levels of AR expression between the Con-veh and SynARKO-veh (Fig. 1B and 1C). Ar immunostaining was performed on the brain tissue sections, and representative Ar staining around the third ventricle was visually reduced in SynARKO-veh mice (Fig. S1 [36]).
Body Weight and Body Composition Was Not Changed by Neuronal ARKO in Basal and Hyperandrogenemic Conditions
BW was compared by 2-way ANOVA among the 4 groups of mice and did not show an effect of DHT or genotyping (SynARKO) or an interaction between the 2 above factors on BW (Fig. 2A). The body composition of the mice, which includes fat mass and lean mass, was determined. There was no significant difference in fat mass or lean mass among any of the groups (2-way ANOVA, no main effect of DHT, genotype or the interaction of the 2) (Fig. 2B), reinforcing the lean nature of our mice.
Deletion of the neuronal Ar gene does not impact body weight and body composition in mice with or without DHT. (A) Body weight of Con-veh, SynARKO-veh, Con-DHT, and SynARKO-DHT mice (n = 7-12). (B) Body composition (fat and lean mass; n = 4-10). Statistics were performed using 2-way ANOVA followed by Tukey post hoc analysis. NS, nonsignificant. Values are mean ± SEM.
SynARKO Partially Prevented Disrupted Glucose Homeostasis Under Hyperandrogenemic Conditions
Metabolic tests were conducted to determine the effect of neuronal Ar deletion on glucose metabolism. At 2 months after DHT implantation (Fig. S2A-D [36], there was a trend to improve glucose tolerance upon GTT in the SynARKO-DHT mice; however, this did not reach significance (Fig. S2 [36]). For GTT (Fig. S2 A, B [36]), there was a main effect of DHT (P < .0001), but no effect of genotyping (G) and an interaction between the 2 (DHT × G; P < .05); however, there was no post hoc difference between the Con-DHT and SynARKO-DHT groups. For PTT (Fig. S2 C, D [36], there was an effect of DHT (P < .001), but no effect of genotyping or an interaction of the 2 factors.
In female mice 4 months after DHT implantation, as previously reported [10], blood glucose levels after GTT and PTT were significantly higher in Con-DHT mice (Fig. 3A and 3C) than in Con-Veh mice. With respect to the GTT (Fig. 3A and 3B), the 2-way ANOVA showed a DHT effect (P < .0001), a genotype effect (P < .01), and there was a significant interaction between DHT × G (P < .01). The post hoc Tukey test showed a significant decrease in area under the curve values in the SynARKO-DHT (P < .001) mice compared with Con-DHT (Fig. 3B). With respect to the PTT (Fig. 3C and 3D), as seen in our previous studies, a 2-way ANOVA showed there was a significant increase in area under the curve in the DHT-treated groups compared with the vehicle groups. No genotype effect or interaction (DHT × G) was seen in the PTT.
Greater glucose intolerance upon 4 months of DHT treatment in control vs SynARKO mice. (A, C, E) The GTT, PTT, and ITT, respectively, were performed in control and SynARKO mice with and without DHT (n = 5-8). (B, D, F) Area under the curve of GTT, PTT, and ITT respectively. (G) Basal glucose levels after a 7 hour fast (n = 7-12). (H) HOMA-IR was calculated and compared among groups (n = 5-6). Statistical analysis was performed by 2-way ANOVA followed by Tukey post hoc analysis (P < .05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, nonsignificant). Values are mean ± SEM.
To measure whole-body insulin resistance, the ITT was conducted, and HOMA-IR was calculated in an independent cohort. ITT was compared by 2-way ANOVA (Fig. 3E and 3F). There was a main effect of DHT (P < .0001), but no effect of genotype or an interaction (DHT × G) between the 2 factors. There was a post hoc difference between Con-veh and Con-DHT (P < .001) and between SynARKO-veh and SynARKO-DHT (P < .05). The basal glucose showed a main effect of DHT (Fig. 3G). HOMA-IR (Fig. 3H) showed a main effect of DHT and genotype. The HOMA-IR was significantly increased (P < .0001) in the Con-DHT mice compared with Con-veh mice; however, the significance between the SynARKO and SynARKO-DHT was moderate (P < .05) (Fig. 3H).
Control, but not SynARKO Mice Experience Higher Heat Production and Food Intake Upon DHT Treatment
To determine the impact of central nervous system Ar on whole-body energy expenditure and other metabolic parameters, we performed indirect calorimetry using CLAMS after 4 months of treatment. Real-time monitoring showed no change in VO2 consumption (Fig. S3A-D [36]), VCO2 production (Fig. S3E-H [36]), and RER (Fig. S3I-L [36]) in SynARKO mice compared with control mice, with and without DHT.
Con-DHT mice showed significantly increased heat production (main effect of DHT) (energy expenditure, unadjusted to BW) at both light, dark phase, and total compared with Con-veh (Fig. 4A and 4C, left of columns); however, there was no change between SynARKO-veh and SynARKO-DHT; Con-DHT vs SynARKO-DHT (Fig. 4A-4C). Heat production was also corrected to BW to get adjusted heat production (expressed as kcal/hour/BW) (Fig. S4 [36]). In the light phase (Fig. S4A [36]), there was a significant main effect of DHT (P < .05), no effect of genotype, and a significant interaction effect (DHT × G) (P < .05). Post hoc tests showed a significant change between Con-Veh and Con-DHT. In the dark phase, there was no main effect of DHT, genotype, or an interaction of the 2 (DHT × G) (Fig. S4B [36]). When adjusted for BW, we observed no difference in the total heat production (Fig. S4C [36]), and there was no main effect of DHT, genotype, or an interaction of the 2 (DHT × G).
Con-DHT mice had enhanced heat production and food intake, but these were not altered in SynARKO-DHT mice compared with vehicle treatment. Heat production of (A) light phase and (B) dark phase, and (C) total combined dark and light phases in mice was measured by indirect calorimetry. Calorie intake in (D) light phase, (E) dark phase, and (F) total combined dark and light phases. (G) Food intake over 48 hours during the light and dark phases. There was no significant difference between SynARKO-veh compared with SynARKO-DHT. Two-way ANOVA followed by the Tukey post hoc test was used to analyze data among 4 groups (P < .05; *P < .05; **P < .01; ***P < .001; NS, nonsignificant). n = 7-14. Data are means ± SEM.
With respect to calorie intake (kcal/24 hours: 2 × 12-hour light or dark phase cycles), during the light cycle (Fig. 4D), we observed a main effect of DHT (P < .001) on the calorie intake, while there was no effect of genotype; as for the interaction (DHT × G), there was a trend toward significance (P = .05). Post hoc comparison showed a difference between Con-veh and Con-DHT (P < .0001) and a Con-DHT vs SynARKO-DHT (P < .05). Calorie intake during the dark phase (Fig. 4E) showed no difference among the groups. With respect to the total calorie intake, there was a main effect of DHT, but no genotype and interaction effect (DHT × G) (Fig. 4F), with a post hoc difference (P < .05) between the Con-veh and Con-DHT groups. The total food intake is graphically represented in Fig. 4G.
Control but not SynARKO Mice Exhibit Higher Nuclear NF-kβ Expression and IBA1 Upon DHT Treatment
The levels of nuclear NF-kβ p-65 expression was quantified by densitometry in the nuclear fraction of the hypothalamic fraction. Two-way ANOVA showed that there was a significant effect of DHT (P < .05), but no effect of genotype or an interaction effect (DHT × G). Post hoc analysis showed a difference between Con-veh and Con-DHT groups (P < .05) (Fig. 5A and 5B). Insulin signaling was significantly increased through p-Akt/Akt in the hypothalamus of the Con-DHT mice vs Con-veh mice (Fig. 5C and 5D). Two-way ANOVA showed that there was a significant effect of DHT (P < .0001) and an interaction between DHT × G (P < .05). Post hoc analysis showed a significant difference between Con-veh and Con-DHT (P < .0001). There was a decrease between Con-DHT and the SynARKO-DHT (P = .05) (Fig. 5D). Hypothalamic AR expression was also analyzed in these mice (Fig. S5 [36]).
DHT-induced increase in nuclear NF-kβ expression and IBA1 is absent in SynARKO mice. (A, C) A representative western blot of NF-kβ in the nuclear fraction and p-Akt/Akt in the cytosolic fraction of the hypothalamus in mice with or without DHT. (B, D) Densitometric analysis of nuclear NF-kβ, and cytosolic p-Akt/Akt in hypothalamus tissue lysate. (E) Representative western blot of IBA1 in the hypothalamus of mice with and without DHT. (F) Densitometric analysis of IBA1. Two-way ANOVA followed by the Tukey post hoc test were used to analyze data among 4 groups. (G) Representative immunostaining of IBA1 around the third ventricle section of the hypothalamus in Con-veh and Con-DHT mice. The boxed areas are shown at higher magnifications. The arrows point to IBA1 staining. (H) Quantification of IBA1 staining in the hypothalamus from Con-Veh and Con-DHT mice. Two tailed Student's t-test was used for statistics analysis (P < .05; *P < .05; ***P < .001; #P .05; NS nonsignificant). n = 4-7. Data are mean ± SEM.
Inflammation in the brain/hypothalamus is associated with increased activation of microglia. Activated microglia were examined with IBA1 using immunostaining and Western blot. Immunoblotting showed a significant increase in the IBA1 levels in the DHT group compared with the control group (main effect of DHT, P < .05), and an interaction of DHT × G (P < .05). For Con-DHT vs SynARKO-DHT, the post hoc Tukey test showed a tendency to decrease, but did not reach significance (P = .06) (Fig. 5E and 5F). IBA1 staining of the hypothalamus is shown and quantified (Fig. 5G and 5H). For statistical analyses using the Student’s t-test between the Con-veh and Con-DHT groups, there was an increase in IBA1 staining in the Con-DHT group compared with the veh group (P = .05) (Fig. 5H).
Discussion
This study addresses how androgens act on ARs in the neurons to mediate metabolism in females under normal and hyperandrogenemic conditions. Successful ARKO was obtained in the neurons (78% decrease in AR in brain homogenates through immunoblotting). Though Ar is mainly expressed in neurons, Ar is also present in other brain cells including astrocytes [37‐39]. The key brain region mediating metabolism by Ar is the hypothalamus [21, 26], hence molecular studies are focused on the hypothalamus. Neurons innervate the organs of the periphery, including the gut and there is low expression of synapsin in these organs [40]; however, it is not clear to what extent AR is present in gut-innervating neurons, and if it affects glucose metabolism. The major findings of the study are that SynARKO-DHT exhibited partially ameliorated glucose intolerance compared with Con-DHT mice (Fig. 3), independent of changes in BW. Besides, we found that DHT altered/increased energy expenditure and calorie intake in control mice compared with Con-veh mice (Fig. 4). Molecular studies showed increased inflammation markers such as NF-kβ and IBA1 (Fig. 5) in the hypothalamus of Con-DHT mice, which could contribute to neuron-regulated metabolic dysfunction in hyperandrogenemia.
Hyperandrogenemia causes impaired glucose tolerance and insulin resistance in women and in female rodent models [2, 10, 41‐43]. Whole-body AR KO studies show that DHT mediates metabolic dysfunction through Ar receptors [29]. Studies show that deletion of hepatic AR ameliorates the impaired metabolic function (GTT, PTT, ITT) seen under hyperandrogenic conditions [31]. Deletion of Ar in muscle has no effect on glucose metabolism [43]. We observed improved GTT in SynARKO mice compared with control mice after 4 months of DHT treatment (Fig. 3 A and 3B), due to worsening of glucose intolerance in control mice (Fig. 3B), which was not observed at 2 months of DHT treatment (Fig. S2B [36]), suggesting the loss of neuronal AR is protective against chronic worsening of glucose intolerance. While DHT overall significantly increased gluconeogenic capacity (PTT-DHT main effect) as previously seen in our studies, there was no difference between the Con-DHT and SynARKO-DHT, suggesting the increased hepatic glucose production is not improved by the neuronal ARKO, unlike that seen in hepatic ARKO [31]. The increase in HOMA-IR and impaired ITT between the Con-veh and Con-DHT was more intense than between the SynARKO-veh and SynARKO-DHT groups. These results agree with the changes in adipose + brain ARKO mice, suggesting that the neuronal AR plays a role directly or indirectly via peripheral tissues in maintaining glucose homeostasis and insulin sensitivity [27].
There was no significant difference in BW and body composition (fat mass and lean mass) between treated and untreated mice, suggesting that metabolic changes observed in SynARKO-DHT mice is independent of BW.
Disruption of AR and estrogen receptor signaling is associated with defects in energy homeostasis and expenditure [44]. In our female mice, we observed an increase in the calorie intake in Con-DHT mice, which could be contributing to increased energy expenditure (Fig. 4) and explain the lean nature of the model. Energy expenditure is mostly determined by the thermic effect of food and activity-induced energy expenditure [45]. Skeletal muscles in hyperandrogenemia, unlike in other models of insulin resistance, show increased Akt signaling [46‐48], which may lead to enhanced metabolism and energy expenditure. Further, the increased heat production/energy expenditure without change in RER indicates that the increase in energy expenditure is achieved with the same balance of aerobic and anaerobic metabolism [49]. However, female mice exposed to male-like DHT levels with increased BW showed no changes in food uptake but had decreased energy expenditure and may indicate a dose-dependent response to DHT on food intake and energy expenditure [50]. Con-DHT mice showed increased calorie uptake during their rest (light) phase, which was not seen in SynARKO-DHT mice (Fig. 4D), suggesting that the circadian rhythm is regulated by neuronal AR. Studies have shown dysregulated circadian rhythm in women with PCOS [51, 52].
Further, the altered metabolism seen in DHT animals may be through AR-regulated inflammation in the hypothalamus. Inflammation in the hypothalamus indeed plays a role in peripheral metabolism with insulin receptor pathways and other inflammatory pathways implicated [23]. We found an increase in p-Akt/Akt signaling in the hypothalamus under hyperandrogenic conditions. Increased p-Akt/Akt signaling has been associated with the activation of neuroinflammation pathways [53]. Activated microglia use the PI3K-Akt pathway to produce proinflammatory mediators [53]. We observed increased activation of NF-kβ by increased expression of nuclear NF-kβ p65 (Fig. 5) and increased activated microglia expression of IBA1 (Fig. 5). Unlike other models of insulin resistance, hyperandrogenemia may not be involved in decreased Akt signaling in the hypothalamus but rather with neuroinflammation, which could contribute to whole-body metabolic dysfunction. A previous study also showed altered/activated microglia (IBA1) during certain stages of development in the prenatally androgenized mice model [54], and evidence has suggested that PCOS is associated with chronic low-grade inflammation [55, 56]. Although inflammation in crucial hypothalamic regions could lead to energy dysregulation, glucose intolerance, and metabolic dysfunction, further details of specific neuronal populations regulated by Ar and involved in the pathogenesis of hyperandrogenism need to be further examined. Our molecular studies were limited to the hypothalamus only, and further evaluation of the fore and hind brain, which also have a role in metabolism, is needed in future.
The animals were randomly assigned to the vehicle and DHT groups based on their genotyping. Though no experiments were done specially with the Synapsin driven Cre only mice (SynCRE+; Ar wt/wt), we anticipate the Cre itself does not influence the parameters studied in this paper since our SynCre mice (SynCre+; Ar fl/fl) had no difference in BW, body mass, metabolic tolerance tests, CLAMS, and molecular studies from the vehicle mice (SynCre-; Arfl/fl). Further, DHT effects were time dependent, and the protective effects in the SynARKO-DHT group were not apparent until the fourth month of DHT treatment, suggesting the absence of a transgene effect.
Here, using our low-dose DHT hyperandrogenemic model, we have demonstrated metabolic dysfunction, altered energy expenditure, and increased brain inflammation in DHT-treated mice, and show that SynARKO female mice were partially protected from hyperandrogenemia-induced metabolic dysfunction and neuroinflammation. Increased AR activation in neurons could lead to altered Akt signaling in the hypothalamus and activation of the NF-kβ p65 inflammatory pathways along with increased activation of the microglia (IBA1) (Fig. 6).
Schematic representation of the study and major findings. Absence of neuronal AR tempers altered glucose metabolism and energy expenditure under hyperandrogenemic conditions in lean female mice and is associated with lower androgen-induced inflammation in the hypothalamus. DHT via Ar increased inflammation (NF-kB, IBA1) in the hypothalamus with long-term (4-month) treatment. ARKO improved systematic glucose intolerance and showed unaltered energy expenditure, with no increased inflammation (NF-kB and IBA1) in the hypothalamus.
Funding
This work was supported by the National Institutes of Health (grants R00-HD068130 04 and R01-HD09551201A1 to S.W.).
Disclosures
The authors have nothing to disclose.
Data Availability
All data generated or analyzed during this study are included in this published article or in data repositories.
References
Abbreviations
- ANOVA
analysis of variance
- Ar
androgen receptor
- CLAMS
Comprehensive Laboratory Animal Monitoring Systems
- DHT
dihydrotestosterone
- GTT
glucose tolerance test
- HOMA-IR
homeostatic model assessment for insulin resistance
- ITT
insulin tolerance test
- KO
knockout
- PCOS
polycystic ovary syndrome
- PTT
pyruvate tolerance test
- RER
respiratory exchange rate
Author notes
Vaibhave Ubba and Serene Joseph equal contributions to the manuscript.
