Nasal Staphylococcus aureus carriage promotes depressive behaviour in mice via sex hormone degradation

Abstract

The human microbiome has a pronounced impact on human physiology and behaviour. Despite its unique anatomical connection to the brain, the role of the nasal microbiome in neurological diseases is understudied. Here, using human data and experiments in mice, we show that nasal Staphylococcus aureus is linked to depression. Nasal microbiome analyses revealed a positive correlation between depression scores and S. aureus abundance among patients with depression and healthy controls. Metabolomics of the nasal cavity showed decreased sex hormones, estradiol and testosterone in patients with depression versus controls. Nasal microbiota transplants from patients reproduced depression-like behaviour in mice with differential abundance of S. aureus. Further homology and mutational analysis uncovered an S. aureus sex hormone-degrading enzyme, 17b-hydroxysteroid dehydrogenase (Hsd12), which degraded testosterone and estradiol in mice, leading to lower levels of dopamine and serotonin in the murine brain. These findings reveal a nasal commensal that influences depressive behaviour and provides insights into the nose-brain axis.

Fig. 1. Nasal S. aureus carriage is associated with depression.

a–g, Results from 16S rRNA microbiome sequencing. a, The Shannon (left) and Simpson (right) methods were used to calculate the α-diversity in the nasal microbiomes of healthy individuals and (n = 118) and patients with depression (n = 100) at the species level. b, β-Diversity comparison of the composition of the nasal microbiomes of the two study cohorts at the species level. Shaded areas show the 95% confidence interval. c, Per-individual and overall relative abundances of the most abundant species in the nasal microbiomes of healthy individuals (n = 118) and individuals with depression (n = 100). d, Multivariate analysis adjusted for age, sex, body mass index, education, income, family relationships, adverse childhood experiences, available supportive friends and batch covariates. Fields marked with + and − signs represent significant associations (adjusted false discovery rate (FDR) < 0.05). A genus-level analysis is shown in Extended Data Fig. 1a,b. e, Actual abundance of S. aureus in the two cohorts (left) as well as among the males (middle) and females (right) among those cohorts. f, Correlation of S. aureus abundance with PHQ-9 depression score in males and females among all individuals. The line shows simple linear regression; r, correlation coefficient. Shaded areas show the 95% confidence intervals. An analysis of the GAD-7 anxiety score is provided in Extended Data Fig. 1f. e,f, The S. aureus abundance was determined as the log-transformed number of nuc copies per μl of DNA. g, S. aureus carrier status in the two study cohorts. An individual was defined as a S. aureus carrier if at least one of 24 randomly selected bacterial clones in the culture-based analysis was found to be S. aureus using matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF-MS) analysis. The number of individuals in the different groups are shown in the bars. Statistical analysis was performed using a two-sided Mann–Whitney test (a,e), permutational multivariate analysis of variance (PERMANOVA; b), Spearman’s correlation (f) or Fisher’s exact test (g). Data are the mean ± s.d. Source data

Fig. 2. Nasal S. aureus carriage induces depressive behaviour in mice.

a, Set-up of the experiment to investigate the impact of nasal microbiome transplantation on anxiety and depression-like behaviour in mice. b, Anxiety-like behaviour of mice in the OFT test (left) as well as depression-like behaviour in FST (middle) and TST (right) tests. Male and female mice received nasal transplants from male and female donors, respectively (one donor per one mouse); n = 16 mice (n = 8 male and n = 8 female; sex-specific analysis in Extended Data Fig. 3a) per group. Light blue and orange symbols depict S. aureus-positive (by culture analysis) transplanted microbiomes. c, Multivariate analysis (covariate-adjusted analysis using MaAsLin2) adjusted for sex. d, Set-up of the experiment to investigate anxiety and depression-like behaviour in female mice nasally colonized with S. aureus or S. epidermidis and control mice. e, Anxiety-like behaviour of mice in the OFT (left) as well as depression-like behaviour in FST (middle) and TST (right) tests. Data for the corresponding experiment with male mice in Extended Data Fig. 6. f, Set-up of the experiment to investigate anxiety and depression-like behaviour in male mice nasally colonized with S. aureus or S. epidermidis and control mice with CUMS treatment. Bacteria were administered seven times at 48-h intervals, including during the seven-day CUMS protocol. g, Anxiety-like behaviour of mice in the OFT (left) as well as depression-like behaviour in the FST (middle) and TST (right) tests. e,f, n = 18 (control and SE) or 20 (SA) mice per group. Data are the mean ± s.d. Statistical analysis was performed using a two-sided unpaired Student’s t-test (b), or two-sided one-way analysis of variance (ANOVA) or Kruskal–Wallis test (depending on the normality of distribution in the groups) and Tukey’s and Dunn’s post-tests, respectively (e,g). Colonization levels and evidence for absence of disease and inflammation in Extended Data Figs. 4 and 5. SA, S. aureus; SE, S. epidermidis. Credit: mouse illustrations in a,d,f, creazilla under a Creative Commons license CC0 1.0. Source data

Fig. 3. Reduced nasal sex hormone levels are associated with depression.

a,b, The nasal metabolomes of individuals in the two cohorts (n = 40 each; randomly selected) were analysed using PCoA with Bray–Curtis distance-based PERMANOVA. R² proportion of variance explained by the principal coordinate (a) and OPLS-DA analyses, with the model fit parameters (R²x, R²y) and predictive ability (Q²) shown (b). c, Significantly altered metabolic pathways by KEGG enrichment. d,e, Nasal estradiol (d) and testosterone (e) levels in the same randomly selected cohort subsets. Data are the mean ± s.d. Serum levels are provided in Extended Data Fig. 6c. f,g, Correlation between the levels of nasal estradiol (f) and testosterone (g), and PHQ-9 depression scores. GAD-7 anxiety score analysis in Extended Data Fig. 6d. a,b,f,g, Lines show simple linear regression and shaded areas the 95% confidence intervals. Statistical analysis was performed using a two-sided Mann–Whitney test (d,e) or Spearman’s correlation (f,g). Source data

Fig. 4. Nasal S. aureus degrades sex hormones and is associated with decreased dopamine and serotonin levels in the brain.

a, Testosterone and estradiol biosynthesis. Note the HSD-catalysed conversion of androstenedione/testosterone and estradiol/estrone. Enzyme names are in green font; hydroxyl and keto groups affected by the HSD enzymes are in red. b, Correlation between nasal testosterone (right) and estradiol (left) concentrations, and nasal S. aureus abundance in combined selected cohort subsets (n = 80). The line shows the simple linear regression. Shaded areas show 95% confidence intervals. c, In vitro conversion of estradiol to estrone (left) and testosterone to androstenedione (right) by S. aureus culture filtrate within 24 h (n = 3 per group). d, Conversion of estradiol to estrone (left) and testosterone to androstenedione (right) by culture filtrates from different bacteria. Tryptone soya broth (TSB) control, n = 4; n = 3 for all other groups. e, Nasal and midbrain levels of testosterone and estradiol in mice colonized with S. aureus or S. epidermidis, and in controls. Estradiol was measured in female mice (n = 6 per group) and testosterone in male mice (n = 8 per group). f,g, Transcriptome analysis of midbrain gene expression in male and female mice nasally colonized with S. aureus compared with S. epidermidis. Red, increased in S. aureus-colonized mice; green, increased in S. epidermidis-colonized mice. Dopamine and serotonin biosynthesis enzymes are labelled; n = 3 per group. e–g, Experimental set-up as depicted in Fig. 2d,f. h, Dopamine and serotonin biosynthesis. Enzyme names are in green font. i, Concentrations of serotonin (left) and dopamine (right) in the midbrains of S. aureus-colonized, S. epidermidis-colonized and control mice; n = 7 mice (n = 4 male and n = 3 female) mice per group. The GABA and glutamate concentrations are in Extended Data Fig. 9f. c–e,i, Data are the mean ± s.d. Statistical analysis was performed using a two-sided one-way ANOVA or a Kruskal–Wallis test, depending on the normality of distribution in the groups, and Tukey’s and Dunn’s post-tests, respectively (d,e,i), or Spearman’s correlation (b). SA, S. aureus; SE, S. epidermis. Steroid formulae in a and h created with BioRender.com. Credit: mouse illustrations in e–g, creazilla under a Creative Commons license CC0 1.0. Source data

Fig. 5. An S. aureus HSD degrades sex hormones and induces depressive behaviour in mice.

a, Conversion of testosterone to androstenedione and estradiol to estrone by putative S. aureus hsd genes expressed in E. coli. The putative hsd7 gene was not included as it could not be expressed, probably due to its large size (n = 3 per group). b, Conversion of testosterone to androstenedione and estradiol to estrone by culture filtrates from S. aureus wild type and its hsd12 isogenic mutant. TSB control, n = 4; other groups, n = 3. c, Set-up of the model to test for induction of anxiety and depression-like behaviour in female mice by nasal colonization with S. aureus wild type, its hsd12 isogenic mutant and controls. d, Anxiety-like behaviour of female mice in the OFT (left) as well as depression-like behaviour in FST (middle) and TST (right) tests (n = 17 per group). e,f, Nose (e) and midbrain (f) estradiol levels in female mice (n = 12 per group). g, Set-up of the model to test for induction of anxiety and depression-like behaviour in male CUMS-treated mice by nasal colonization with S. aureus wild type, its hsd12 isogenic mutant and controls. h, Anxiety-like behaviour of male mice in the OFT (left) as well as depression-like behaviour in FST (middle) and TST (right) tests (n = 17 per group). i,j, Nose (i) and midbrain (j) testosterone levels in male mice (n = 12 per group). k, Relative expression of TH and TPH2 genes in midbrain samples (n = 6 per group; n = 3 females and n = 3 CUMS-treated males). l, Midbrain serotonin (left) and dopamine (right) levels (n = 12 per group; n = 6 females and n = 6 CUMS-treated males). Statistical analysis was performed using a two-sided one-way ANOVA or Kruskal–Wallis test, depending on normality of distribution in the groups, and Tukey’s and Dunn’s post-tests, respectively. Data are the mean ± s.d. SA, S. aureus; SAΔhsd, S. aureus hsd12 isogenic mutant. Credit: mouse illustrations in c,g, creazilla under a Creative Commons license CC0 1.0. Source data

Fig. 6. Depression-inducing capacity of a mouse-adapted S. aureus strain.

a, In vitro sex hormone conversion by S. aureus ST88 culture filtrate (n = 3 per group). b, Sex hormone conversion by culture filtrates from S. aureus ST88 wild type and hsd12 isogenic mutant (n = 3 per group). c, Set-up of model to test for the induction of anxiety and depression-like behaviour in mice by nasal colonization with S. aureus wild type, its hsd12 isogenic mutant and controls. d, Anxiety-like behaviour in the OFT (left) as well as depression-like behaviour in FST (middle) and TST (right) tests (n = 12 per group). e, Set-up of the model to measure degradation of applied sex hormones to the noses of mice. ORX, orchidectomy; OVX, ovariectomy. f, Detected sex hormone levels (estradiol in female and testosterone in male mice) with and without S. aureus application (n = 6 per group). g, Set-up of the model to test for depression and anxiety-like behaviour in female mice after application of sex hormones (in the background of S. aureus colonization). h, Anxiety-like behaviour in the OFT (left) as well as depression-like behaviour in FST (middle) and TST (right) tests (n = 10 per group). i,j, Corresponding set-up (i) of the experiment in g and results in male mice (j; n = 10 per group). Statistical analysis was performed using a two-sided one-way ANOVA with Tukey’s (b,d) post-test with the exception of the male mice TST in d where a two-sided Kruskal–Wallis test with Dunn’s post-test was used, or and unpaired two-sided Student’s t-test (f,h,j) with the exception of female TST and OFT and male FST in h,j where a two-sided Mann–Whitney test was used. Data are the mean ± s.d. SA, S. aureus ST88 wild type; SAΔhsd, S. aureus hsd12 isogenic mutant. Credit: mouse illustrations in c–e,g,i, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 1. Nasal microbiome analysis in healthy versus depressed cohorts—additional data.

a, Per-individual and overall relative abundances of the most abundant genera in the nasal microbiomes of healthy (n = 118) and depressed (n = 100) cohorts as per 16S rRNA sequencing. b, Corresponding multivariate analysis adjusted for age, sex and batch covariates. c, Per-individual and overall abundances of the most abundant species as per culture-based analysis. d, Corresponding multivariate analysis adjusted for age, sex and batch covariates. e, Multivariate analysis of bacterial abundance and PHQ-9 scores adjusted for age, sex, BMI, education, income, family relationships, adverse childhood experiences, available supportive friends, and batch covariates. f,g, Multivariate analysis and correlation plot of bacterial and S. aureus (SA) abundance, respectively, and GAD-7 anxiety scores using the same analytical approach as applied to the PHQ-9. Statistical analysis is by Spearman correlation (f). The lines show simple linear regression. Shaded areas show 95% confidence intervals. h, Sequence-type distribution of randomly selected isolates (one per colonized individual; n = 31, healthy cohort; n = 45, depressed cohort). In multivariate plots (b,d,e,g), fields marked with “+” and “-“ signs represent significant associations (FDR < 0.05). Source data

Extended Data Fig. 2. Nasal transplant experiment—additional data 1.

a, Weight, temperature and sickness scores over the time of the experiment. b, CFU over the course of the experiment. c,d, α-Diversity analyses of the nasal microbiomes of antibiotic-treated (Abx) versus control mice that did not receive antibiotic, measured at day 8 (see Fig. 2a). e, β-Diversity analysis of the nasal microbiomes of antibiotic-treated (Abx) versus control mice that did not receive antibiotic, measured at day 8 (see Fig. 2a). f–h, Corresponding analyses for the lung microbiomes. i–k, Corresponding analyses for the intestinal (fecal) microbiomes. l–n, α- And β-diversity analyses of mouse nasal microbiomes receiving nasal transplants from healthy versus depressed people, measured at day 17 (see Fig. 2a). o–q, α- And β-diversity analyses of mouse nasal microbiomes over the course of the experiment in comparison. n = 12/group (c–n); n = 6/group (a,o–q); n = 12 (b). Statistical analyses are by unpaired, two-sided t-tests (c,i,l), two-sided Mann–Whitney tests (e–g,j,m), two-sided Kruskal–Wallis tests versus control (o,p) and PERMANOVA (d,h,k,n). Shaded areas show 95% confidence intervals (d,h,k,n,q). All error bars show the mean ± standard deviation (SD). Source data

Extended Data Fig. 3. Nasal transplant experiment—additional data 2.

a, Sex-specific behavioural analyses. Male mice received nasal transplants from male and female from female donors. n = 8/group. Statistical analysis is by unpaired, two-tailed t-tests except for female OFT, two-sided Mann–Whitney test. Error bars show the mean ± standard deviation (SD). b, Correlation between anxiety and depression scores in humans with anxiety and depressive-like behaviour tests in nasal transplant-recipient mice. n = 16/group (n = 8 male, n = 8 female). Statistical analysis is by Spearman correlation. Lines show simple linear regression. Shaded areas show 95% confidence intervals. Credit: mouse illustrations in a, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 4. Mouse colonization experiment—additional data 1.

a, Colonizing CFU in the nose at the time of euthanasia (day 21). n = 6/group. b,c, Nose and lung cytokine concentrations in S. aureus (SA) nasally colonized, S. epidermidis (SE) nasally colonized, and control female mice at the time of euthanasia. n = 6/group. d,e, Nose and lung cytokine concentrations in S. aureus nasally colonized, S. epidermidis nasally colonized, and control male (CUMS-treated) mice at the time of euthanasia. n = 6/group. f, Nose and lung histology in S. aureus nasally colonized, S. epidermidis nasally colonized, and control male (CUMS-treated) mice at the time of euthanasia. g, Histological analysis of OMP and Iba1 expression in the olfactory epithelium at the time of euthanasia (day 21). h, Ly6G expression in the olfactory epithelium at the time of euthanasia (day 21). Statistical analysis is by two-sided one-way ANOVA with Tukey’s post-tests (b–e) and two-sided, unpaired t-tests (a). There were no significant differences. All error bars show the mean ± standard deviation (SD). Credit: mouse illustrations in b,d,f, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 5. Mouse colonization experiment—additional data 2.

a,b, Lung cytokine concentrations 6 h post the last bacterial application (day 18) in female (a) and CUMS-treated male mice (b). n = 3/group. Statistical analysis is by two-sided 1-way ANOVAs with Tukey’s post-tests. There were no significant differences. c, Cnet plots of transcriptomic changes (5 top enriched KEGG pathways) in the nasal mucosa 3 h post the last bacterial application (day 18). n = 3/group (male CUMS-treated mice). d–f, Weight, temperature and sickness scores over the time of the experiment. n = 6/group. g, Bacterial loads in the noses, lungs, and faeces measured starting 2 h past the time of the last bacterial application. n = 6/group. All error bars show the mean ± standard deviation (SD). SA, S. aureus; SE, S. epidermidis. Credit: mouse illustrations in a,b, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 6. Induction of anxiety and depression-like behaviour in male mice and metabolome and sex hormone measurement in healthy and depressed cohorts—additional information.

a, Set-up of the experiment (as in Fig. 2d, but using male mice). b, Anxiety-like behaviour of mice in OFT and depression-like behaviour in FST and TST tests. n = 8/group. SA, S. aureus; SE, S. epidermidis. c, Serum estradiol and testosterone levels in female and male cohorts, respectively. n = 19 (male, healthy), n = 34 (male, depressed), n = 24 (female, healthy), n = 42 (female, depressed). Serum levels were determined in all individuals who consented to having blood drawn. d, Correlation of nasal testosterone or estradiol levels with GAD-7 anxiety scores. Statistical analysis is by two-sided one-way ANOVA with Tukey’s post-tests (b), two-sided Mann–Whitney tests (c: serum estradiol), two-sided, unpaired t-test (c: serum testosterone) and Spearman correlation (d). Error bars show the mean ± standard deviation (SD) (b,c). Lines show simple linear regression and shaded areas show 95% confidence intervals (d). Credit: mouse illustrations in a,b, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 7. Nasal S. aureus degrades sex hormones and increases dopamine and serotonin levels in the brain—additional information.

a, Conversion of estradiol and testosterone to estrone and androstenedione, respectively, by randomly selected isolates of main isolate STs. n = 3/group. b, Serum estradiol and testosterone levels in mice colonized with S. aureus (SA) or S. epidermidis (SE), or in controls. n = 6–8/group (n = 8, CUMS-treated male; n = 6, female). c, Pregnenolone and progesterone midbrain levels. n = 6/group (n = 3, CUMS-treated male; n = 3, female). d, Expression of serotonin and dopamine biosynthesis genes in the midbrain of mice colonized with S. aureus or S. epidermidis, or in controls. n = 7/group (n = 4, CUMS-treated male; n = 3, female). e, Results of KEGG pathway enrichment. f, GABA and glutamate midbrain concentrations in S. aureus-colonized, S. epidermidis-colonized, and control mice. n = 7/group (n = 4, CUMS-treated male; n = 3, female). Statistical analysis is by two-sided one-way ANOVAs or Kruskal–Wallis tests, depending on normality of distribution in the groups, and Tukey’s and Dunn’s post-tests, respectively (b–d,f). All error bars show the mean ± standard deviation (SD). Credit: mouse illustrations in e, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 8. Location and deletion of the S. aureus hsd12 gene.

a, Genetic context of the hsd12 gene, shown for S. aureus strain USA300 FPR3757. b, Growth comparison of S. aureus wild-type (SA) and isogenic hsd12 deletion (SAΔhsd) strains of strain ST398 in two different media. n = 4/group. c, RNA-Seq transcriptomic comparison of gene expression in ST398 wild-type versus isogenic hsd12 deletion strains grown in TSB to logarithmic growth phase. Significantly differentially expressed genes are coloured (red, up in Δhsd; blue, up in wild-type). d,e, Corresponding analyses for ST88 wild-type versus isogenic hsd12 deletion strains. n = 4/group. All error bars show the mean ± standard deviation (SD). Source data

Extended Data Fig. 9. Absence of inflammation and histology differences in mice nasally colonized by S. aureus wild-type and isogenic hsd12 (ST398) deletion strains.

a–c, Weight, temperature and sickness scores over the time of the experiment. n = 6/group. d, Nose and lung cytokine concentrations in S. aureus (SA) nasally colonized, S. aureus Δhsd (SAΔhsd) nasally colonized, and control female mice. n = 6/group. e, Nose and lung cytokine concentrations in S. aureus nasally colonized, S. aureus Δhsd nasally colonized, and control (CUMS-treated) male mice. n = 6/group. f, Colonizing CFU in the nose at the time of euthanasia. n = 6/group. g, Nose and lung histology in S. aureus nasally colonized, S. aureus Δhsd nasally colonized, and control male (CUMS-treated) mice. Statistical analysis is by two-sided one-way ANOVA with Tukey’s post-tests (d,e) and two-sided, unpaired t-tests (f). There were no statistically significant differences except for where marked. All error bars show the mean ± standard deviation (SD). Credit: mouse illustrations in d,e,g, creazilla under a Creative Commons license CC0 1.0. Source data

Extended Data Fig. 10. ST88 experiments—additional information.

a, Bacterial loads in noses, lungs, and guts (faeces) after second colonization (day 13; time 0 h). n = 6/group. b, Weights, body temperatures, and sickness scores over the time of the experiment. n = 6/group. c, Colonizing CFU in the nose at day 20. n = 6/group. d,e Nose and lung cytokine concentrations in S. aureus nasally colonized, S. aureus Δhsd nasally colonized, and control female (d) and (CUMS-treated, e) male mice. n = 6/group. f, Ex vivo imaging of nasal hydrogel-loaded FITC retention in the nose or excretion. Note prolonged retention in the nose (yellow arrows) in the hydrogel versus liquid control samples. g, In vitro cumulative release of estradiol or testosterone from the nasal hydrogel. Statistical analysis is by two-sided one-way ANOVA with Tukey’s post-tests (d,e). There were no significant differences. n = 3/group. All error bars show the mean ± standard deviation (SD). Credit: mouse illustrations in d,e, creazilla under a Creative Commons license CC0 1.0. Source data