Endocrine, metabolic, and endocannabinoid alterations in rats exhibiting high anxiety-related behaviors

Anxiety disorders are major risk factors for obesity. However, the mechanisms accounting for this susceptibility remain unclear. Animal models have proved to be useful tools for understanding the role of emotional functioning in the development and maintenance of metabolic alterations implicated in obesity. Here we sought to determine the predictive value of behavioral indices of anxiety for hormonal and metabolic imbalances in rats. Adult Lewis rats were screened on the elevated plus maze (EPM). K-means clustering was used to divide the rats into two groups based on their anxiety index in the EPM: low (LA) and high anxiety (HA) rats. This proxy of anxiety combines individual EPM parameters and accepted ratios into a single score. Four weeks later, we measured markers of endocrine and metabolic function. We found that relative LA rats, the HA rats exhibited reduced latencies to exit a modified light-dark conflict test. Our results show that the HA rats displayed increased corticosterone levels when compared to LA rats. Furthermore, the HA rats weighted more and exhibited an enhanced glycemic response to exogenously administered glucose during the glucose tolerance test, indicating glucose intolerance. Notably, when compared to LA rats, the HA rats showed higher circulating levels of the endogenous cannabinoid, 2-arachidonoyl-sn-glycerol (2-AG). Together, these data indicate that patterns of emotional reactivity associated with anxiety may share common pathological pathways with metabolic complications implicated in obesity. Uncovering metabolic risk factors for anxiety disorders have the potential to strongly impact how we treat mental illnesses. Objective To test the hypothesis that indices of anxiety in the elevated plus maze have predictive value for alterations implicated in obesity, which include heightened emotional reactivity, and dysregulated corticosterone, glucose metabolism, and circulating endocannabinoid levels.


Introduction
Anxiety disorders are the most common of mental disorders, affecting more than 30% of adults at some point in their lives (1). Anxiety has been defined as "a psychological, physiological, and behavioral state induced in animals and humans by a threat to wellbeing or survival, either actual or potential" (2). Anxiety is characterized by the activation of the autonomic and neuroendocrine systems, heightened arousal, and specific behavioral responses, including escape, avoidance, and overall readiness to respond to danger. Although anxiety is a normal reaction to threat or perceived environmental challenges and essential for survival, excessive and maladaptive anxiety can become pathological. Classical theories of causation support that pathological anxiety manifests in vulnerable individuals and can be triggered by psychological factors (3). However, more integrative views of anxiety disorders are also considering the critical contribution of biological factors to the etiology of pathological anxiety.
Anxiety disorders are highly comorbid with metabolic disorders. There is growing recognition for a reciprocal, bidirectional link between anxiety disorders and proinflammatory diets (4), and obesity and related metabolic diseases (5; 6). Studies in humans and rodents have shown that obesity and the consumption of obesogenic diets are risk factors for the development of anxiety-related psychopathology (7)(8)(9)(10).
Interestingly, a new study suggests that these differences in emotional and behavioral responses related to anxiety appear with changes in body weight in obesity-prone rats (11). Other studies demonstrate that individuals exposed to psychosocial stress and diagnosed with anxiety disorders are susceptible to increases in body mass index (BMI) and obesity (12; 13). Although the mechanisms accounting for the link between obesity and anxiety remain speculative, this bidirectionality suggests common pathophysiological processes between physical and mental illnesses. Thus, this experiment was undertaken to examine whether anxiety-like phenotypes are predictive for pathogenic processes implicated in obesity and related metabolic disorders. We hypothesized that rats exhibiting higher anxiety-like behaviors will share endocrine and metabolic traits comparable to obesity.

Results and Discussion
There is a growing appreciation for the substantial individual differences in anxiety-like behaviors in rats. The elevated plus maze (EPM) have proved to be a successful method to systematically determine individuality in anxiety-like and avoidance behaviors (14)(15)(16).
However, to our knowledge, no studies have investigated whether individual differences in anxiety-like behaviors in the EPM may predict endocrine and metabolic changes cooccurring with obesity in non-obese rats. We began our investigation of the association between anxiety and obesity-like phenotypes by hypothesizing that proxies of anxiety in the EPM may have predictive value for markers of dysregulated metabolism. We used the k-means clustering algorithm to group rats in two groups based on their anxiety-like behaviors in the EPM. As expected, rats in the low anxiety (LA) group exhibited reduced indices of anxiety in the EPM relative to the high anxiety (HA) group. Our analyses revealed that rats from the HA group had a significant 29% increase in anxiety index (AI) when compared to LA group (HA: 0.63 ± 0.02 vs. LA: 0.49 ± 0.01; t(16) = 4.82, p = 0.0002) ( Figure 1B). This finding is confirmed in the track plot shown in Figure 1C, demonstrating that HA rats exhibited high avoidance of the open arms while spending most of the time in the closed arms of the EPM. Although HA rats exhibited higher acoustic startle responses, we found no significant differences between groups (data not shown).
The conflict-avoidance test was used to evaluate others measures of anxiety-like behaviors associated with the rats' innate light aversion (photophobia). In this test, the rats are first trained to escape the aversive bright light chamber and navigate through a 3 mm sharp probe chamber to get to a less aversive dark chamber. This test has been used to evaluate nociception, and cognitive and motivational processing (17). Our analyses showed that HA rats had a significant reduction (54%) in latency (time to exit light chamber after door is opened) when compared to LA rats (LA: 24.63 ± 2.2 vs. HA: 11.24 ± 2.34; t(16) = 4.09, p = 0.0009) (Figure 1D), indicating heightened avoidance behaviors and validating the EPM-based group classification. We used an electronic von Frey anesthesiometer to determine nociceptive thresholds in the LA and HA rats. We reasoned that this test would provide information as to whether differences in escape latencies were related to reduced nociception in the HA rats. Two-way ANOVA analyses revealed no significant group [F(1, 16) = 0.01, p = 0.92] effects on withdrawal threshold in neither limb (data not shown), indicating similar nociceptive thresholds between groups.
Corticosterone is a common marker of anxiety-and stress-related behaviors in rats and implicated in the detrimental effects of obesogenic diets on emotionality (9; 18; 19).
We measured fecal corticosterone levels at the end of the study (30 days after EPM testing). Interestingly, the HA rats exhibited a significant increase (57%) in fecal corticosterone levels when compared to LA (LA: 33.73 ± 1.1 vs. HA: 53.05 ± 5.3; t(7) = 3.17, p = 0.01) ( Figure 1E). Supporting the well-established link between corticosterone, anxiety, and glucose metabolism, we found that HA rats showed impairments in glucose metabolism. Analyses revealed significant group [F (1,9)  effects on glucose clearance ( Figure 1F). Sidak's post-hoc analyses showed that, relative to LA rats, the HA rats exhibited increased glucose levels at 30 min (20% increase; p = 0.044) and at 60 min (26% increase; p = 0.013) following the bolus glucose solution injection. Subsequently, we sought to determine whether proxies of anxiety correlate with changes in body weight. We found increased body mass in HA rats relative to LA rats. Two-way ANOVA analyses revealed significant group [F (1, 16) = 7.14, p = 0.017] and time [F(2, 32) = 2,721, p < 0.0001], while no significant interactions [F(2, 32) = 2.27, p < 0.119] effects on body weight (data not shown). Notably, these differences in body weight emerged at the last time point evaluated (6% increase at PND 170; p = 0.006) and before or during the week that the behaviors were recorded. A feature common to changes in obesity is alterations in leptin, a major orchestrator of metabolism and regulator of emotionality (20). Our analyses revealed that rats from HA group had a significant increase in plasma leptin levels when compared to LA group (HA: 1364 ± 237.4 vs. LA: 638.3 ± 121.4; t(11) = 2.58, p = 0.025) (data not shown). Together, these data support the relationship between body weight, circulating leptin levels, and glucose homeostasis and its potential association with anxiety-related phenotypes.
In search of a unifying biological factor for the above results, we investigated the levels of major circulating endocannabinoids. Several lines of evidence show that the endocannabinoid system influences a wide variety of neurobiological processes including emotionality, cognitive function, and glucocorticoid-mediated stress responses. This system is an attractive area for research into new pharmacotherapeutic targets for anxiety disorders (21). Studies show that both reductions (22) or elevations (23) (30), and hyperphagia associated with diet-induced obesity (33). In humans, hedonic eating was also associated with higher levels of 2-AG in circulation (35; 36). In the present study, we find increases in circulating levels of the endocannabinoid, 2-AG, in rats exhibiting a "high anxiety" phenotype (61% increase relative to LA; LA: 22.81 ± 2.03 vs. HA: 40.54 ± 7.09; t(16) = 2.17, p = 0.046) ( Figure 1G). This result raises the intriguing possibility that increased levels of 2-AG under these conditions may impact feeding and other behaviors. Furthermore, diet-induced changes in 2-AG levels may influence affect and emotionality, resulting in anxiety-related phenotypes. However, a direct test of these hypotheses remains for future studies.

Conclusions
We found significant alterations in obesity-related endocrine and metabolic biomarkers based on individual differences in anxiety-like behaviors. Collectively, our findings confirm that anxiety-like behaviors share common pathogenic pathways with endocrine and metabolic complications of obesity. This model may present new opportunities in the prevention and amelioration of co-morbid obesity and psychiatric disorders.

Limitations
Our results indicate that HA Lewis rats exhibit metabolic, hormonal, and endocannabinoid alterations consistent with obesogenic phenotypes. However, it is difficult to dissociate whether the observed changes enhance susceptibility to an anxiety-like phenotype, or whether the anxiety-like behavioral manifestations in the EPM result in metabolic alterations. Replication in larger cohorts, and with different rat strains and diets is warranted. Considerations for sex as a biological variable should be factored in future studies.

Conjectures
While we observed distinctive behavioral, endocannabinoid, metabolic, and hormonal profiles in rats exhibiting high anxiety-like behaviors in the elevated plus maze, it is unclear from our data whether the neuroendocrine and metabolic changes accompanying weight gain promote anxiety or vice versa. Future studies are required to test a causal relationship. Importantly, further research is also required to estimate whether the metabolic-related alterations in rats exhibiting high indices of anxiety can be targeted to ameliorate both, anxiety-and obesity-like phenotypes.

Acoustic Startle
The acoustic startle reflex (ASR) were performed using the SR-Lab Acoustic Chambers (San Diego Instruments). An experimental session began with a habituation program, consisting of 5 min of background noise at 68 dB. Following habituation, the initial block consisted of 11 acoustic startle trials was recorded (120 dB; 20 ms). The trials were presented in a randomized manner (intertrial interval = 10 -25 s). The second block had 36 trials presented in a pseudorandomized order. In this block, the trials included pulsealone trials (120 dB) and pre-pulse trials (five trials each: 71, 74, or 80 dB; 40 ms duration). The third block of the session consisted of 5 additional pulse-alone startle trials (120 dB). Subsequently, the rats were returned to their cages and each chamber was cleaned with soap and water and thoroughly dried. Max startle amplitudes during the session were averaged.

Modified Light-Dark Test: Mechanical Conflict-Avoidance System
Conflict-avoidance behaviors were measured using the Mechanical Conflict-Avoidance  5). For testing days, the light of the device was turned off and the rat was placed in the LC. After 10 seconds, the light was turned on and remained on for 20 seconds. The LC door was carefully opened and the time to exit the LC was recorded. Each rat was given 30 seconds to exit LC which was counted as the latency time. After the rats spent 5-10 seconds in the dark chamber, the lid was removed and the rats were returned to their home cages. The apparatus was clean thoroughly between each test. For each experimental session, we repeated this process 3 times per rat with at least 10 minutes between trials.

Von Frey
Sensory thresholds to a mechanical stimulus were measured using an electronic von Frey aesthesiometer (IITC Life Science, Woodland Hills, CA). Each rat was placed in a Plexiglas chamber on top of an elevated mesh stand that provided access to the plantar surface of the paws. The rats were allowed to acclimate to the testing chamber for 30 min prior to testing. Three mock trials were performed halfway through the acclimation period.
Subsequently, a rigid blunt tip that was attached to the aesthesiometer was applied to the plantar surface. The withdrawal threshold or mechanical sensitivity was defined as the average force (g) required for paw withdrawal in five trials separated by at least 2 min interval. The maximum and minimum threshold values were excluded from each paw before analyses.

Intraperitoneal glucose tolerance test (IPGTT)
The IPGTT was performed at PND 170. The rats were fasted for 12 h prior to being anesthetized with pentobarbital (100 mg/kg). Subsequently, the rats received intraperitoneal injections of 40% dextrose solution (2 g of dextrose per kg of body weight;).

Measuring corticosterone and leptin concentrations using ELISA
The rats were placed on new bedding 24 h before feces collection. Fecal samples were collected from each cage at PND 170 and corticosterone extracted following previous methods (40). Briefly, fecal corticosterone was extracted by incubating feces in ethanol overnight at a ratio of 10 mL ethanol per gram of feces. Corticosterone concentration was determined using the Correlate EIA kit (Assay Designs, Ann Arbor, MI). ELISA plates were read at 405 nm (Qcal Biotek Instruments plate reader, Winooski, VT USA). Specific concentrations for each sample were determined as percentage bound using a standard curve of samples ranging from 32 to 20,000 pg/mL. Values were then calculated and reported as picograms of corticosterone per mL.
Leptin plasma levels were measured using a commercial ELISA kit (Abcam).
Plasma samples were diluted with kit assay buffer (1:3 dilutions). ELISA plates were read at 450 nm on a plate reader (Molecular Devices). Specific concentrations for each sample were determined as mean absorbance using a standard curve of samples ranging from 0 to 8000 pg/ml.

Statistics
Statistical analysis was performed using GraphPad Prism 8. We used the K-means clustering method to assign all rats to two groups according to their anxiety index (SPSS Version 25.0; IBM Corp., Armonk, NY). Statistical differences were evaluated using 2tailed unpaired Student's t test or 2-way ANOVA with multiple comparisons, as indicated. P values of less than 0.05 were considered statistically significant.

Funding Statement
This work was partly supported by Seed Grant Funds to Johnny D. (D) Average latency obtained from the modified light-dark box behavioral test for LA and HA rats. HA rats had a significant reduction (54%) in latency (time to exit the aversive light chamber) relative to LA rats (***p < 0.001).
(E) Average levels of corticosterone in feces. HA rats exhibited a significant increase (57%) in fecal corticosterone levels when compared to LA (*p < 0.05).
(G) Average levels of 2-AG in plasma. HA rats had a significant increase (61%) in circulating levels of 2-AG relative to LA rats (*p < 0.05).