In this study we analyzed the effects of ω-3 supplementation on anxiety, plasma corticosterone levels, and learning of chronically stressed rats. First, we investigated whether our stress protocol was effective in triggering stress responses. Stressed rats of all experimental groups had less body weight gain than unstressed control group rats (Figure
2A,B). This demonstrates that the stress protocol used was effective and that ω-3 supplementation did not prevent this effect (Figure
2A,B). Comparable results have been reported using similar stress paradigms and corticosterone administration [
44‐
46]. Diets enriched with PUFAs, in particular the ω-3 family, decreased both the adipose tissue mass and plasma leptin levels in rats [
47]. Leptin is released from adipocytes and regulates food intake and body weight by binding to leptin receptors to the hypothalamus [
48,
49]. Thus, it is probable that the ω-3 supplementation used in our experiments decreased the body weight of the rats compared to that of controls (Figure
2A,B).
Effects of restraint stress and ω-3 fatty acid on anxiety and corticosterone plasma levels
Stressed control and vehicle groups rats had significantly lower percentages of entries into the open arms of the EPM and spent less time in the center of the open field test than unstressed rats (Figure
3A, Table
2). These behaviors were related to an anxiogenic effect induced by restraint stress and vehicle treatment (Figure
3A). Interestingly, supplementation with ω-3 had an opposite effect on the stressed rats. Indeed, ω-3 supplementation increased the number of entries into the open arm of the EPM, which was related to an anxiolytic effect (Figure
3A) that was not associated with locomotor impairments due to restraint stress and vehicle (Table
2).
Comparable results were obtained using another chronic restraint stress paradigm and ω-3 supplementation [
50]. Anxiety is mainly regulated by the basolateral amygdala and the bed nucleus of the stria terminalis (BNST) [
51‐
53]. In some chronic stress paradigms, such as chronic unpredictable stress or immobilization, enhanced anxiety has been correlated with dendritic hypertrophy in the basolateral amygdala and BNST [
18,
54,
55]. It is possible that the chronic stress protocol and vehicle treatment in our study induced hyperactivation of the basolateral amygdala and/or BNST by plastic neuronal changes, which significantly increased anxiety, HPA axis activity, and plasma corticosterone levels (Figure
3A,B). Another brain area that modulates BNST neuronal excitability is the dorsal raphé nucleus (DRN) [
56]. Axons of the serotoninergic neurons located in the DRN are sent to the BNST and serotonin is released to the synaptic space, which in turn inhibits the neuronal excitability in the BNST by activation of the 5-HT
1A and 5-HT
1B receptors [
56,
57]. Serotonin levels are reduced in the brain of stressed rats [
58,
59], and thereby the 5-HT
1A and 5-HT
1B receptors in the BNST are not activated. This alteration may contribute to increasing neuronal excitability in the BNST and anxiety in the stressed rats (Figure
3A).
On the other hand, ω-3 supplementation significantly decreased corticosterone levels in the unstressed and stressed rats (Figure
3A, right side). This suggests that ω-3 supplementation prevents hyperactivation of the HPA axis induced by chronic stress, decreasing the effects of corticosterone on dendritic morphology and neuronal activity of the basolateral amygdala and BNST. Supplementation with ω-3 increases serotonin levels in the brain of stressed rats [
60], which in turn may reduce BNST neuronal excitability by activation of the 5-HT
1A and 5-HT
1B. Thus, ω-3 supplementation decreases anxiety of the stressed rats.
Figure
3A and Table
2 show that ω-3 supplementation had an anxiogenic effect on the unstressed rats and significantly reduced corticosterone levels compared to the level in unstressed rats treated with vehicle (Figure
3B). This was unexpected and could be explained by the effects of both ω-3 and serotonin on the HPA axis and neuronal activity in the BNST, respectively. Oral administration of vehicle was a stressor for the rats due because it increased corticosterone levels in the unstressed rats. However, this effect was prevented by ω-3 supplementation (Figure
3B, right side). It is possible that ω-3 prevents stress-induced dendritic hypertrophy in the amygdala of unstressed rats and this decreases corticosterone levels compared to that of unstressed rats treated with vehicle (Figure
3B, right side). Supplementation with ω-3 may increase serotonin levels in the brain of rats [
60]. We suggest that to counteract this effect, the expression of the 5-HT
1A and 5-HT
1B receptors was down-regulated in the BNST of unstressed rats supplemented with ω-3. In this context, neuronal excitability in the BNST may increase because the inhibitory control of serotonin over the BNST is lost. As result, anxiety is enhanced in unstressed rats supplemented with ω-3 (Figure
3A).
Unstressed and stressed rats of control group that were not stimulated had similar corticosterone levels, suggesting that the rats adapted to 21 days of restraint stress (Figure
3B, left side). Previous studies have shown that 3 or 6 hours per day of restraint stress significantly increased corticosterone plasma levels during the first week, while in the second and third weeks of restraint stress the increases of corticosterone levels were less pronounced [
61,
62]. Therefore, if the effects of 21 days of restraint stress on HPA axis activity and corticosterone levels had been lost, the rats that were subjected to restraint stress and unstressed rats would have had comparable plasma corticosterone levels after exposure to a new uncontrollable stressor (acute swimming). However, control group rats subjected to restraint stress had significantly higher plasma corticosterone levels than unstressed rats following one minute of swimming (Figure
3B). This suggests that one day after the restraint stress ended, unstressed and stressed rats had similar HPA axis activity in an environment without stressors. On the other hand, stressed control group rats still showed higher levels of the HPA axis activity than unstressed rats exposed to a new uncontrollable stressor. This neuroendocrine alteration, which induces maladaptive responses to stressors, is characteristic of stressed animals [
42,
50].
Corticosterone plasma levels increased for approximately the first seven days of restraint stress [
10]. However, the long-term impact of the chronic stress on the neuronal morphology of the lateral amygdala and on anxiety levels were measured after twenty-one days of stress-free recovery [
55]. Therefore, chronically stressed rats may have had enhanced anxiety and hyperactivity of the HPA axis at the same time as they were subjected to new stressors like the EPM and swimming in a water maze (Figure
3A,B).
In our experiments, supplementation was applied by oral administration and this method resulted in higher corticosterone levels after acute swimming in unstressed vehicle group rats than those of unstressed control group rats (Figure
3B, right side). This suggests that vehicle treatment, which was applied from weaning to the end of the stress period, was sufficient to induce short-term stress in the supplemented rats. Handing could be comparable to oral administration of vehicle applied before chronic restraint stress protocol. A longer period of handling has gradually less inhibitory effects on the HPA axis activity and significantly decreases the animal’s sensitivity to the restraint stress [
63], while a shorter period of handing before applying acute restraint stress results in significantly lower corticosterone and adrenocorticotropic hormone plasma levels than those of rats without handling [
63].
The method used to apply the supplementation in our experiments could have induced more profound desensitization of the mechanisms involved in inducing the HPA axis response to restraint stress. This in turn could have resulted in lower corticosterone plasma levels in the stressed rats supplemented with vehicle than in animals that were not supplemented in the control group, after acute swimming (Figure
3B, right side). Desensitization of the HPA axis might involve the loss of CRH receptors in the anterior pituitary, which in turn may induce the corticotrophs to become refractory to CRH hypersecretion during restraint stress.
The effects of restraint stress and ω-3 fatty acid supplementation on learning
Restraint stress and oral administration of vehicle were stressful for the rats given that the two treatments increased plasma corticosterone levels (Figure
3B). In addition, these treatments impaired learning during the conditioned trials (Figure
4C,D). Lesion studies in the main nuclei of the auditory system that regulates learning, the inferior colliculus (IC, auditory mesencephalon) and the medial geniculate nucleus (MG, auditory thalamus), have demonstrated that the two brain structures are key factors for acquiring aversive memories to auditory cues during fear conditioning in rats [
64]. As well, restraint stress induces dendritic atrophy in the IC, MG, and primary auditory cortex, and affects auditory processing [
27,
65,
66]. A recent study using micro Positron Emission Tomography supports these findings, in that chronic mild stress induced a significant decrease in glucose metabolism in the IC, but not in the superior colliculus (visual mesencephalon) [
67]. In our study, learning impairment could have been due to stress-induced dendritic atrophy in the IC and/or MG. In support of this idea, unstressed and stressed rats supplemented with vehicle showed significant less learning than animals without supplementation. However, ω-3 supplementation prevented this effect (Figure
4C,D,E), possibly by preventing the stress-induced impairment in the IC and/or MG. In fact, ω-3 fatty acid deficiency impairs active avoidance and decreases the polyunsaturated fatty acid composition in the cellular and subcellular fractions [
38]. As well, learning alterations associated with maternal deficiency of α-linolenic acid are prevented by α-linolenic acid supplementation after weaning [
68]. Likewise, DHA supplementation prevents learning impairments in rats induced by ω-3 deficiency in rats [
30].
Other brain nuclei that are key for acquiring and evoking auditory avoidance conditioned responses are the lateral (LA) and basal amygdala [
26]. Therefore, another possible explanation for our results in the 2-AA is that restraint stress and ω-3 supplementation had opposite effects on these nuclei, that restraint stress induced dendritic hypertrophy in the LA and this dendritic change enhanced anxiety-like behaviors by the BNST [
54]. On the other hand, ω-3 supplementation may have prevented these morphologic alterations and produced anxiolitic effects in stressed rats. This, in turn could have improved learning, as has been seen with anxiolitic drugs such as midazolam, which facilitates avoidance retrieval in rats [
69].
Possible cellular mechanisms underlying the anti-stress effects of ω-3 fatty acids supplementation
A growing body of evidence suggests that ω-3 PUFA levels in the brain modulate the reactivity and sensitivity to stress [
70]. In addition, chronic stress reduces the DHA content in the brain phospholipids and prevents the incorporation of supplemental-DHA in the neuronal membranes [
60,
71]. We suggest that restraint stress decreases DHA content in the phosoholipid membranes of glutamatergic neurons at the amygdaloid complex, whereas it increases arachidonic acid (AA) content. In support of this idea, studies with monkeys have shown that chronic stress is associated with a higher phosphatidylethanolamine ω-6/ω-3 ratio, suggesting lower ω-3 fatty acid status in stressed animals [
72]. AA is released from the phospholipid membranes to the cytoplasm by cytoplasmic phospholipase A2 (cPLA
2) activity and is transformed into endocannabinoid (eCb), which in turn inhibits GABA release from presynaptic neurons [
73‐
75]. Through this mechanism, chronic stress may increase excitatory neuronal activity in the amygdala and enhance anxiety. On the other hand, ω-3 supplementation may increase DHA content in the phospholipid membranes of excitatory neurons; which in turn decrease AA levels in the cytoplasm of neurons. Thus, inhibitory transmission could be reduced in the amygdala; decreasing anxiety and plasma corticosterone levels in the stressed rats (Figure
3A,B).
The anxiolytic effect of ω-3 supplementation may be related to increased serotonin levels in the brain of chronically stressed rats. Serotonin has a key role in the regulation of anxiety-like behaviors [
76]. In the case of our study, ω-3 PUFA supplementation could have enhanced the serotonin level in the brain [
76,
77].
The positive effect of ω-3 supplementation on the learning could be related to a direct effect of ω-3 on the auditory brain nuclei that modulates fear learning, such as the MG and IC, which are affected in the stressed rats [
65,
67]. Chronic stress may have decreased proplastic protein levels in the brain nuclei that produce dendritic atrophy. Proplastic proteins are implicated in neurite extension, cell survival and synaptic plasticity [
78]. Alternatively, ω-3 supplementation may increase the level of the proteins that prevent dendritic atrophy in the MG and IC of stressed rats. On the other hand, the positive effects of ω-3 fatty acids on learning may have been by a direct effect in the LA, a brain area key for fear learning [
26]. Long-term potentiation studies show that auditory fear learning depends on AMPA receptor insertion in the plasmatic membrane of LA neurons [
79]. Thus, chronic stress may impair this process in the LA, while ω-3 supplementation could prevent this effect. As well, a mixture of the two mechanisms may be associated with the positive effects of ω-3 fatty acids on the learning of stressed rats.