Short-chain fatty acids: microbial metabolites that alleviate stress-induced brain-gut axis alterations

Abstract

Key points: Chronic (psychosocial) stress changes gut microbiota composition, as well as inducing behavioural and physiological deficits. The microbial metabolites short-chain fatty acids (SCFAs) have been implicated in gastrointestinal functional, (neuro)immune regulation and host metabolism, but their role in stress-induced behavioural and physiological alterations is poorly understood. Administration of SCFAs to mice undergoing psychosocial stress alleviates enduring alterations in anhedonia and heightened stress-responsiveness, as well as stress-induced increases in intestinal permeability. In contrast, chronic stress-induced alterations in body weight gain, faecal SCFAs and the gene expression of the SCFA receptors FFAR2 and FFAR3 remained unaffected by SCFA supplementation. These results present novel insights into mechanisms underpinning the influence of the gut microbiota on brain homeostasis, behaviour and host metabolism, informing the development of microbiota-targeted therapies for stress-related disorders.

Abstract: There is a growing recognition of the involvement of the gastrointestinal microbiota in the regulation of physiology and behaviour. Microbiota-derived metabolites play a central role in the communication between microbes and their host, with short-chain fatty acids (SCFAs) being perhaps the most studied. SCFAs are primarily derived from fermentation of dietary fibres and play a pivotal role in host gut, metabolic and immune function. All these factors have previously been demonstrated to be adversely affected by stress. Therefore, we sought to assess whether SCFA supplementation could counteract the enduring effects of chronic psychosocial stress. C57BL/6J male mice received oral supplementation of a mixture of the three principle SCFAs (acetate, propionate and butyrate). One week later, mice underwent 3 weeks of repeated psychosocial stress, followed by a comprehensive behavioural analysis. Finally, plasma corticosterone, faecal SCFAs and caecal microbiota composition were assessed. SCFA treatment alleviated psychosocial stress-induced alterations in reward-seeking behaviour, and increased responsiveness to an acute stressor and in vivo intestinal permeability. In addition, SCFAs exhibited behavioural test-specific antidepressant and anxiolytic effects, which were not present when mice had also undergone psychosocial stress. Stress-induced increases in body weight gain, faecal SCFAs and the colonic gene expression of the SCFA receptors free fatty acid receptors 2 and 3 remained unaffected by SCFA supplementation. Moreover, there were no collateral effects on caecal microbiota composition. Taken together, these data show that SCFA supplementation alleviates selective and enduring alterations induced by repeated psychosocial stress and these data may inform future research into microbiota-targeted therapies for stress-related disorders.

Keywords: Behaviour; Chronic Stress; Gut Microbiota.

Figure 1. Experimental design of the short‐chain fatty acid treatment, psychosocial stress and behavioural assessment

After 1 week of short‐chain fatty acid (SCFA) treatment, animals underwent a 3‐week psychosocial stress protocol containing intermittent social defeat and overcrowding procedures. Mice subsequently underwent behavioural assessment, starting with the least stressful test to the most stressful test. The order of the behavioural tests was as follows. Week 4: social interaction test (SIT) and sucrose preference test (SPT); week 5: three‐chamber sociability test (3‐CT) and female urine sniffing test (FUST); week 6: open field (OF), novel object recognition test (NOR) and marble burying test (MB); week 7: elevated plus maze (EPM), stress‐induced hyperthermia test (SIH) and tail‐suspension test (TST); week 8: fear conditioning (FC) and hot plate test (HPT); week 9: FITC‐dextran intestinal permeability test (FITC); week 10: forced swim test (FST); week 11: sacrifice. SCFA treatment was for the entire duration. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 2. Chronic stress induces aggressor‐specific social avoidance

Social interaction with a CD1 mouse used in the social defeat procedure was assessed in the social interaction test 1 day after the last stressor (A). Mice were additionally assessed for social preference (B) and recognition (C) with a conspecific mouse in the three‐chamber sociability test 1 week post stress. The social interaction test was non‐parametrically distributed and analysed using the Kruskal‐Wallis test. Significant differences are depicted as: ^* P < 0.05; Control compared to Stress. In the three‐chamber sociability test, differences between ‘object’ versus ‘mouse’ and ‘familiar mouse’ versus ‘novel mouse’ were assessed using a Student's unpaired t test. Significant differences are depicted as: ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001. All data are expressed as means ± SEM (n = 9–10). [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 3. SCFAs decrease specific anxiety‐ and depressive‐like behaviours in control but not stressed animals

Anxiety‐like behaviour was assessed in the open field test 2 weeks post stress (A and B), whereas depressive‐like behaviour was determined using the forced swim test 6 weeks post stress (C). The open field test was non‐parametrically distributed and analysed using the Kruskal‐Wallis test, followed by the Mann‐Whitney test. The forced swim test was analysed using a two‐way ANOVA, followed by an LSD post hoc test. Significant differences are depicted as: ^* P < 0.05 and ^** P < 0.01; Control/Control compared to SCFA/Control, ^$$ P < 0.01; SCFA/Control compared to SCFA/Stress. All data are expressed as means ± SEM (n = 9–10). Open circles on each graph represent individual animals. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 4. Psychosocial stress induces long‐term anhedonia, which is absent after SCFA supplementation

A, hedonic and reward‐seeking behaviour was assessed using the female urine sniffing test 2 weeks post stress. B, the expression of genes involved in reward signalling was investigated in the striatum, which were the dopamine receptor D1a (DRD1a), dopamine receptor D2 (DRD2), brain‐derived neurotrophic factor (BDNF Exon IV) and tropomyosin receptor kinase B (TrkB). Data from the female urine sniffing test and BDNF gene expression was non‐parametrically distributed and analysed using the Kruskal‐Wallis test, followed by the Mann‐Whitney test. Gene expression data from all other genes were normally distributed and analysed using a two‐way ANOVA, followed by an LSD post hoc test. Significant differences are depicted as: ^** P < 0.01 and ^*** P < 0.001; Control/Control compared to SCFA/Control or Control/Stress, ^$ P < 0.05; SCFA/Control compared to SCFA/Stress, ^# P < 0.05; Control/Stress compared to SCFA/Stress. All data are expressed as means ± SEM (n = 8–10). [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 5. Neither stress, nor SCFA treatment affected the acquisition, retention and extinction phase data from the fear conditioning test

Cue‐ and context‐associative learning induced by foot shock was evaluated using fear conditioning 4 weeks post stress. At phase 1 (acquisition), mice were presented with a tone, followed by a foot chock. The context‐associative learning was assessed by measuring freezing behaviour in between tones (A), whereas cue‐associative learning was determined during the presentation of the tone (D). At phase 2 (retention), mice underwent the same protocol without the shocks to assess context‐ and cue‐associative retention (B and E, respectively). At phase 3 (extinction), mice were assessed for context‐ and cue‐associative extinction (C and F, respectively). All data are expressed as means ± SEM (n = 9–10). Open circles on each graph represent individual animals. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 6. Psychosocial stress induces increased responsiveness to acute stress, which was ameliorated by SCFAs

Stress responsiveness and HPA‐axis reactivity were assessed using the stress‐induced hyperthermia test 3 weeks after psychosocial stress (A), and by assessing the corticosterone levels in response to acute stress 6 weeks after psychosocial stress (B and C). The stressor used for the latter was the forced swim test. Hypothalamic genes involved in HPA‐axis signalling of which expression was investigated were corticotrophin‐releasing factor (CRF), mineralocorticoid receptor (MR) and glucocorticoid receptor (GR) (D). For the hippocampus, these were corticotrophin‐releasing factor receptor 1 (CRFR1), MR and GR (E). All data were non‐parametrically distributed and analysed using the Kruskal‐Wallis test, followed by the Mann‐Whitney test. Significant differences are depicted as: ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001; Control/Control compared to SCFA/Control or Control/Stress, ^$ P < 0.05, ^$$ P < 0.01; SCFA/Control compared to SCFA/Stress, ^# P < 0.05 and ^## P < 0.01; Control/Stress compared to SCFA/Stress. All data are expressed as means ± SEM (n = 8–10). [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 7. Stress induces intestinal permeability, which was rescued by SCFA treatment

A, in vivo intestinal permeability was assessed 5 weeks post stress. B, colonic genes involved in intestinal permeability of which gene expression was investigated were claudin‐1 (Cldn1), tight junction protein 1 (Tjp1), occludin (Ocln) and mucus 2 (Muc2). C, gene expression analysis of genes involved glucocorticoid signalling were corticotrophin‐releasing factor receptors 1 and 2 (CRFR1 and CRFR2 respectively), mineralocorticoid receptor (MR) and glucocorticoid receptor (GR). All data were non‐parametrically distributed and analysed using the Kruskal‐Wallis test, followed by the Mann‐Whitney test. Significant differences are depicted as: ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001; Control/Control compared to SCFA/Control or Control/Stress, ^$ P < 0.05 and ^$$ P < 0.01; SCFA/Control compared to SCFA/Stress, ^# P < 0.05 and ^## P < 0.01; Control/Stress compared to SCFA/Stress. All data are expressed as means ± SEM (n = 8–10). Open circles on each graph represent individual animals. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 8. Neither stress nor SCFAs affected caecal microbiota diversity

A, microbial alpha‐diversity was determined using Chao1, Simpson, Shannon metrics, as well as the number of species detected. B, beta‐diversity of the overall composition was depicted using principle component analysis. Alpha‐diversity metrics were non‐parametrically distributed and analysed using the Kruskal‐Wallis test. Data are depicted as median with IQR and minimum/maximum values as error bars (n = 9–10). [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 9. Subtle differences in caecum microbiota at family, genus and phylum levels were detected

The 16S sequencing revealed few significant differences in microbiota composition on phylum, family and genus level. Data are organised on stress‐effect (A), concurrent SCFA‐ and stress‐effect (B), and combined SCFA and stress effect (C). Data were non‐parametrically distributed and analysed using the Kruskal‐Wallis test, followed by the Mann‐Whitney test. Significant differences are depicted as: ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001; Control/Control compared to SCFA/Control or Control/Stress, ^$ P < 0.05, ^$$ P < 0.01 and ^$$$ P < 0.001; SCFA/Control compared to SCFA/Stress, ^# P < 0.05 and ^## P < 0.01; Control/Stress compared to SCFA/Stress. Data are depicted as median with IQR and minimum/maximum values as error bars (n = 9–10). [Color figure can be viewed at http://wileyonlinelibrary.com]