Indoxyl sulfate, a gut microbiome-derived uremic toxin, is associated with psychic anxiety and its functional magnetic resonance imaging-based neurologic signature

ol.:ȋͬͭͮͯͰͱͲͳʹ͵Ȍ

Scientific Reports

| (2021) 11:21011

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports

Indoxyl sulfate, a gut

microbiome‑derived uremic toxin,

is associated with psychic anxiety

and its functional magnetic

resonance imaging‑based

neurologic signature

Christopher R. Brydges

, Oliver Fiehn

, Helen S. Mayberg

2,3

, Henry Schreiber

Siamak Mahmoudian Dehkordi

, Sudeepa Bhattacharyya

, Jungho Cha

2,3

, Ki Sueng Choi

2,3

W. Edward

Craighead

7,8

, Ranga R. Krishnan

, A. John Rush

5,10,11

, Boadie W. Dunlop

3,15

Rima Kaddurah‑Daouk

5,12,13,14

& the Mood Disorders Precision Medicine Consortium

It is unknown whether indoles, metabolites of tryptophan that are derived entirely from bacterial

metabolism in the gut, are associated with symptoms of depression and anxiety. Serum samples

(baseline, 12

weeks) were drawn from participants (n

= 196) randomized to treatment with cognitive

behavioral therapy (CBT), escitalopram, or duloxetine for major depressive disorder. Baseline indoxyl

sulfate abundance was positively correlated with severity of psychic anxiety and total anxiety and

with resting state functional connectivity to a network that processes aversive stimuli (which includes

the subcallosal cingulate cortex (SCC

‑FC), bilateral anterior insula, right anterior midcingulate

cortex, and the right premotor areas). The relation between indoxyl sulfate and psychic anxiety was

mediated only through the metabolite’s effect on the SCC

‑FC with the premotor area. Baseline indole

abundances were unrelated to post

‑treatment outcome measures, and changes in symptoms were not

correlated with changes in indole concentrations. These results suggest that CBT and antidepressant

medications relieve anxiety via mechanisms unrelated to modulation of indoles derived from gut

microbiota; it remains possible that treatment

‑related improvement stems from their impact on other

aspects of the gut microbiome. A peripheral gut microbiome‑derived metabolite was associated with

altered neural processing and with psychiatric symptom (anxiety) in humans, which provides further

evidence that gut microbiome disruption can contribute to neuropsychiatric disorders that may

OPEN

West Coast Metabolomics Center, University of California, Davis, CA, USA.

Department of Neurology and

Neurosurgery, Icahn School of Medicine at Mount Sinai, New York, NY, USA.

Department of Psychiatry and

Behavioral Sciences, Emory University School of Medicine, Atlanta, GA, USA.

Division of Biology & Biological

Engineering, California Institute of Technology, Pasadena, CA, USA.

Department of Psychiatry and Behavioral

Sciences, Duke University School of Medicine, Durham, NC, USA.

Department of Biological Sciences, Arkansas

Biosciences Institute, Arkansas State University, Jonesboro, AR, USA.

Department of Psychiatry and Behavioral

Sciences, Emory University School of Medicine, Atlanta, GA

30329, USA.

Department of Psychology, Emory

University, Atlanta, GA, USA.

Department of Psychiatry, Rush Medical College, Chicago, IL, USA.

Department

of Psychiatry, Health Sciences Center, Texas Tech University, Permian Basin, TX, USA.

Duke-National University

of Singapore, Singapore, Singapore.

Department of Medicine, Duke University, Durham, NC, USA.

Duke

Institute of Brain Sciences, Duke University, Durham, NC, USA.

Duke University Medical Center, DUMC 3903,

Blue Zone South, Durham, NC, USA.

Emory University College of Medicine, 12 Executive Park Dr. NE, Room 347,

Atlanta, GA

30329, USA.

A list of authors and their affiliations appears at the end of the paper.

email: bdunlop@

emory.edu

; kaddu001@mc.duke.edu

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

require different therapeutic approaches. Given the exploratory nature of this study, findings should

be replicated in confirmatory studies.

Clinical trial NCT00360399 “Predictors of Antidepressant Treatment Response: The Emory CIDAR”

https:// clini caltr ials.

gov/

ct2/

show/

NCT00 360399

The gut microbiota impacts numerous aspects of human health and

disease

, including neuropsychiatric disor

ders. The “microbiota–gut–brain axis” refers to a bidirectional communication pathway that connects the central

nervous system (CNS), the gut, and the microbial community that inhabits the gastrointestinal

tract

. Within

this axis, the gut microbiota modulates central processes through the activation of neuronal pathways (e.g., the

vagus nerve) as well as through the production of microbial metabolites and immune mediators that can trigger

changes in neurotransmission, neuroinflammation, and

behavior

–

Disruptions to the gut microbiome have been correlated with several neurological disorders, including Par

kinson’s disease, autism spectrum disorder, schizophrenia, and major depressive disorder (MDD)

–

, though

the specific mechanisms that underlie the role of the gut microbiota in these diseases is not fully understood.

However, research in preclinical rodent models shows that the gut microbiota is sufficient to alter host behavior,

as shown by the increase in anxiety- and depressive-like behaviors in rodents after fecal microbiota transfer

from humans with depression relative to those that received transfer of fecal microbiota from demographic

controls

. Further, transferred microbes resulted in altered metabolic states in the recipient mice that displayed

depressive-like

symptoms

. These data implicate the gut microbiota as direct contributors to behaviors associated

with depression and anxiety through their metabolic effects. In this study, we explore gut microbiota-associated

tryptophan metabolism and correlate levels of metabolites to clinical symptoms and severity of depression and

anxiety in humans.

Tryptophan is an essential amino acid that can be metabolized in the gastrointestinal tract via the serotonin,

kynurenine, and indole metabolic pathways (Fig.

), which have been associated with human maladies including

autoimmunity, inflammatory diseases, metabolic syndrome, and neurological diseases including depression and

anxiety

disorders

. Strikingly, the gut microbiota is exclusively responsible for the conversion of tryptophan to

indole and indole derivatives, as there are no detectable levels of these molecules in gnotobiotic mice that lack a

gut

microbiome

. Analysis of biosynthetic pathways found that the genes necessary to make indole and indole

derivatives, such as indole-3-propionic acid (IPA), indole-3-acetic acid (IAA), and indole-3-lactic acid (ILA), are

found exclusively in the gut microbiome but not in mammalian

genomes

(Fig.

). Indoxyl sulfate (IS), results

from the sulfonation of bacterially-derived indole via sulfotransferases in human

liver

; however, the rate of

indoxyl sulfate production is driven by the presence of indole

derivatives

, which are produced exclusively by

gut

microbes

. Importantly, these indoles can have immunomodulatory effects and are potent agonists for aryl

hydrocarbon

receptors

(AHRs), which regulate host immunity and barrier function at mucosal

sites

Indole derivatives can also affect immune status in the brain, as some indole derivatives (e.g., IPA and IAA)

have anti-inflammatory effects on neurodegenerative diseases in the experimental autoimmune encephalomy-

elitis (EAE) mouse model of multiple

sclerosis

as well as in a cell line model of Alzheimer’s

disease

. Several

studies have examined the role of the brain gut axis, and indoles specifically, for their impact on physical health,

cognition, and neurological

disorders

–

. High levels of the uremic toxin IS, which is normally cleared via the

kidneys and excreted in the urine, is associated with diminished cognitive function in renal dialysis

patients

IS is also associated with both neurodevelopmental and neurodegenerative diseases, as levels of IS are increased

in patients who have an autism spectrum

disorder

or Parkinson’s

disease

. Although the mechanistic role

of IS in these diseases is unknown, IS can cross the blood–brain

barrier

and can increase levels of oxidative

stress and pro-inflammatory cytokine signaling in astrocytes and mixed glial cells in vitro

, which suggests that

inflammation and reactive oxygen species may be involved.

With respect to psychiatric disorders, IS has been associated with behavioral defects in preclinical models of

anxiety and depression. The administration of IS into rodents’ drinking water results in increased concentrations

of IS in the brain and increased blood–brain barrier permeability in an AHR-dependent manner, with accom

panying increases in anxiety and cognitive

deficits

. Monocolonization experiments with indole-producing

Escherichia coli

and isogenic mutants have shown that indole production by gut bacteria is sufficient to drive

increases in anxiety- and depressive-like behavior in

rats

. Although interest in the role of the brain gut axis

for psychiatric disorders is growing, there are few human studies, with almost no work examining indoles

specifically

Taken together, the preclinical and clinical data indicate that indole derivatives provide excellent models to

study the microbiota-gut-brain axis given their connection to central immune regulators (i.e., AHR), their link

to human neurological diseases, and the exclusivity of indole production to gut microbes.

To date, the effects of peripheral metabolic concentrations on neural functioning have received little study,

likely due to the paucity of datasets that contain concurrently collected metabolomic and neuroimaging meas

ures. Such research is crucial for determining how changes in peripheral systems may yield alterations in brain

function that can produce clinically relevant symptoms such as depression, anxiety, or cognitive impairment.

Metabolomic data may prove to have clinical value by providing biological markers for identifying treatment-

relevant subtypes of MDD, as an objective marker of change during treatment, and as a predictor of the longer-

term course of illness. Beyond enhancing pathophysiological understanding (which could identify novel treat

ment targets), concurrently analyzing metabolomic and neuroimaging data may enable the development of

easier-to-obtain peripheral blood markers that can act as surrogate markers for brain states relevant to disease

pathology and personalization of

treatments

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

Given the importance of some tryptophan metabolites for symptoms of depression and their response to

treatment

–

, we focused on the effects of indoles, which derive from the integrated metabolism of tryptophan

by the gut microbiome and host. Our approach was exploratory, using corrections for multiple comparisons,

due to the paucity of prior work examining the impact of indoles in major depression or anxiety that could have

meaningfully informed a priori hypotheses. Using blood samples stored from the Prediction of Remission in

Depression to Individual and Combined Treatments (PReDICT) study, which was a large study of treatment-

naïve patients with MDD, we measured levels of four indole derivatives (IPA, IAA, ILA, IS) to address the fol

lowing questions:

1.Do levels of indoles and their ratios at baseline prior to treatment correlate with depression and anxiety

severity at baseline?

2.Do levels of indoles and their ratios at baseline correlate with specific individual symptoms of depression?

3.Can symptom change after treatment with duloxetine, escitalopram, or cognitive behavioral therapy

(CBT) be predicted by baseline levels of indoles, and does symptom change correlate with changes in

levels of indoles after treatment?

4.Are there relationships between baseline peripheral metabolic concentrations of indoles and brain rest-

ing state functional connectivity as determined using functional magnetic resonance imaging (fMRI).

Materials and methods

Study design.

The PReDICT study

protocol

, clinical

results

and initial neuroimaging

analyses

have

been published previously. The study was conducted through the Mood and Anxiety Disorders Program of

Emory University from 2007 to 2013. The study was approved by the Emory Institutional Review Board. All

research was performed in accordance with relevant guidelines and regulations. All patients provided written

informed consent to participate.

PReDICT was designed to identify predictors and moderators of outcomes to three randomly assigned first-

line treatments for MDD: duloxetine, escitalopram, or CBT. The study enrolled treatment-naïve adult outpatients,

aged 18–65 years, who had current MDD without psychotic symptoms. To be eligible for randomization, partici-

pants had to score

≥ 18 at screening and

≥ 15 at baseline on the 17-item version of the Hamilton Depression Rating

Scale (HAM-D)

. The screening period ranged from 7 to 28 days. Key exclusion criteria included the presence

Figure 1.

Tryptophan human gut bacterial co-metabolism leading to production of indoles including IPA, IAA,

ILA and IS.

3-HAA

3-hydroxyanthranilic acid,

3H-KYN

3-Hyroxykynurenine,

5-HTP

5-hydroxytryptophan,

AAAD

aromatic amino acid decarboxylase,

AANAT

aralkylamine N-acetyltransferase,

acdA

acyl-CoA

dehydrogenase,

AraT

aromatic amino acid aminotransferase,

ASMT

Acetylserotonin O-methyltransferase,

fldBC

phenyllactate dehydratase,

fldH

phenyllactate dehydrogenase,

indole acrylic acid,

IAA

indole acetic

acid,

IAAld

indole-3-acetaldehyde,

IAld

indole-3-aldehyde,

IAM

indole-3-acetamide,

IDO

indolamine

2,3-dioxygenase,

ILA

indole-3-lactic acid,

IPA

indole-3-propionic acid,

I PYA

indole-3-pyruvate,

K AT

Kynurenine aminotransferase,

KMO

kynurenine 3-monooxygenase,

KYNU

kynureninase,

MAO

monoamine

oxidase,

NAD

nicotinamide adenine dinucleotide,

porB

C: pyruvate : ferredoxin oxidoreductase B and C,

TDO

tryptophan 2,3-dioxygenase,

TMO

tryptophan 2-monooxygenase,

TNA

tryptophanase,

TpH

tryptophan

hydroxylase,

Tr D

tryptophan decarboxylase.

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

of any medically significant or unstable medication condition that could impact study participation, safety, or

data interpretation, as assessed through a medical history, physical exam, electrocardiogram, and screening

laboratory testing. Psychiatric exclusion criteria included a lifetime history of bipolar disorder, primary psychotic

disorder, or dementia, or a diagnosis in the 12 months prior to baseline of obsessive–compulsive disorder, eating

disorder, or dissociative disorder. Patients were also excluded if they met DSM-IV criteria for substance abuse

within 3 months, substance dependence within 12 months of the randomization visit, or if their urine tested

positive for drugs of abuse. Treatment was provided for 12 weeks with duloxetine (30–60 mg/day), escitalopram

(10–20 mg/day), or CBT 16 individual 1-h sessions.

Symptom assessments.

At the baseline visit, participants were assessed by trained interviewers using the

HAM-D and the HAM-A. The HAM-A is a 14-item measure that consists of two subscales, “psychic anxiety”

(items 1–6 and 14), and “somatic anxiety” (items 7–13)

. Psychic anxiety consists of the symptoms of anxious

mood, tension, fears, depressed mood, insomnia, impaired concentration, and restlessness. Somatic anxiety con-

sists of physical symptoms associated with the muscular, sensory, cardiovascular, respiratory, gastrointestinal,

genitourinary, and autonomic systems. Participants also completed the QIDS-SR, which assesses the nine diag-

nostic symptom criteria for

MDD

. The HAM-D, HAM-A, and QIDS-SR were repeated at the Week 12 visit.

Blood sampling.

Participants who met all eligibility criteria at the baseline visit underwent an antecubital

phlebotomy, without regard for time of day, diet, or fasting/fed status. Sampling was repeated at the week 12 visit.

Collected samples were allowed to clot for 20 min and then centrifuged at 4 °C to separate the serum, which was

frozen at − 80 °C until being thawed for the current analyses.

Neuroimaging.

To explore associations between indole metabolites and brain function, we used the rest-

ing state fMRI (rs-fMRI) scans collected during the week prior to baseline, the details of which have previously

been

published

. Briefly, eyes-open scanning was performed for 7.4 min in a 3-T Siemens TIM Trio (Siemens

Medical Systems, Erlangen, Germany). Image analysis was conducted using

AFNI

. [Analysis of Functional

NeuroImages] software package. The standard preprocessing pipeline implemented in AFNI package was used

for processing rs-fMRI data. The time series of rs-fMRI data were despiked, and then corrected for motion and

slice-time acquisition. Scans with head motion

> 2 mm in any direction were excluded from the analysis. The

remaining effects of the noise signal, including residual head motion inferences, signal from the CSF and local

white matter, were also corrected. Subsequently, data were applied a band-pass filter and smoothed using an

isotropic Gaussian kernel of 8 mm full width at half maximum. The imaging anatomical and functional data

sets were co-registered and normalized to standard Montreal Neurological Institute (MNI) 1-mm voxel space.

Consistent with our prior

analyses

, we used a region-of-interest seed-based approach to assess the resting state

functional connectivity (RSFC) of the SCC. The SCC volume was defined using the Harvard–Oxford

Atlas

and the SCC was thresholded at 50% probability centered on MNI coordinates 66, 24, –11. The seeds comprised

two 5-mm radius spheres, with a final volume of 485 mL each. Utilizing

3dNetCorr

, the mean time course of

the bilateral seed was correlated voxel-wise with the rest of the brain. The voxelwise correlation coefficients were

then z-scored by calculating the inverse hyperbolic tangent, yielding the seed-based RSFC maps for analysis.

Metabolomics data acquisition.

Metabolomics data focused on primary and polar metabolites using

gas chromatography—time of flight mass

spectrometry

. Briefly, 30 μL of plasma was extracted at − 20 °C with

1 mL degassed isopropanol/acetonitrile/water (3/3/2). Extracts were dried down, cleaned from triacylglycerides

using acetonitrile/water (1/1), and derivatized with methoxyamine and trimethylsilylation. Samples (0.5 μL)

were injected at 250 °C to a 30 m rtx5-SilMS column, ramped from 50 to 300 °C at 15 °C/min, and analyzed by

− 70 eV electron ionization at 17 spectra/s. Raw data were deconvoluted and processed using ChromaTOF vs.

4.1 and uploaded to the UC Davis BinBase

database

for data curation and compound

identification

. Result

data were normalized by SERRF software to correct for drift or batch

effects

Statistical analyses.

Indole abundance and ratios of each indole pair were included in all analyses. In

order to investigate the role of indoles in depression and anxiety symptomology at baseline, partial Spearman

rank correlations were conducted between the baseline abundance/ratio of each indole and HAM-D 17-item

total score, HAM-A total score, HAM-A Psychic and Somatic subscores, QIDS-SR 16-item total score, and each

individual QIDS-SR item after accounting for age, sex, and body mass index (BMI). Spearman correlations were

also conducted between baseline indole abundance/ratio and participant demographic factors (age, BMI, height,

and weight). Additionally, sex differences in baseline indole abundance/ratio were tested using Mann–Whitney

tests and fold changes in median abundance/ratio between groups.

To investigate the potential effects of treatment on indoles, changes in indole abundance from pre- to post-

treatment were tested using Wilcoxon signed-rank tests and fold changes. Partial Spearman rank correlations

were conducted between post-treatment indole abundance/ratio and post-treatment HAM-D 17-item total score,

HAM-A total score, and HAM-A Psychic and Somatic subscores, QIDS-SR 16-item total score, and each indi

vidual QIDS-SR item, after accounting for age, sex, and baseline BMI. The same analyses were also conducted

with change from pre- to post-treatment scores of all psychiatric variables, and also with fold changes from pre- to

post-treatment for each indole. We also evaluated categorical outcomes at week 12 for four outcome groups, as

defined

previously

: “Remitter” (HAM-D

≤ 7); “Response without remission” (≥

50% reduction from baseline

HAM-D, but not reaching remission threshold); “Partial response” (30–49% reduction from baseline HAM-D);

and “Treatment failure” (<

30% reduction from baseline HAM-D score).

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

Differences in indole post-treatment abundance/ratio and fold change were investigated between each pair

of treatment response outcome

groups

(treatment failure; partial response; response; remission) by conducting

Mann–Whitney

tests. This analysis was also repeated with baseline indole abundances/ratios to investigate

whether baseline levels of indoles may be associated with treatment outcome. Although statistical power was

reduced by splitting the sample, we also explored whether there were differential changes in indoles between

patients treated with CBT versus antidepressant medication and their within-treatment outcomes. All reported

-values were adjusted for multiple comparisons using the Holm method within each psychometric measure (i.e.,

Holm correction was applied to each statistical test within each measure). This correction was applied in order

to decrease the Type I error rate, whilst also maintaining a low enough Type II error rate given the exploratory

nature of the

hypotheses

Neuroimaging analyses were conducted using

AFNI

and jamovi (

www. jamovi. org

). Of 122 participants who

had an adequate quality of resting-state fMRI

data

, 80 had metabolomic measurements and clinical scores.

Voxel-wise linear regression analyses were performed to examine the relationship between SCC-FC and IS or

psychic anxiety scores (uncorrected

< 0.005 and

> 250 voxels cluster size). A conjunction analysis identified

overlapping areas between the SCC-FC of IS and the SCC-FC of psychic anxiety scores. Subsequently, mediation

analyses were performed using the Medmod

module

in jamovi. Three regions identified by the whole brain

linear regression analysis between SCC-FC and IS were used for the mediation analysis. We explored each region,

and combinations of the three regions, in the mediation models. For each model, the direct and indirect effects

were estimated using bootstrapping with 5000 samples.

Results

Of the 344 patients randomized in PReDICT, 196 had metabolomic measures available for analysis at baseline

and 124 were available at week 12 (

= 34 CBT;

= 44 duloxetine;

= 46 escitalopram). The demographic and

clinical characteristics of the 196 participants are presented in Table

Baseline associations.

Associations of indole metabolites with demographic variables.

Supplemental

Fig.

shows a heat map of correlations between baseline indole abundance/ratio and participant demographic

variables. Abundance of ILA was positively associated with age, height, and weight (all

s > 0.18, all

s < 0.040).

The ratios of IAA/ILA (negative associations) and ILA/IS (positive associations) were also significantly associ-

ated with height and weight (all

< 0.034). For sex differences, abundance of IAA (Fold Changes (FC)

= 1.19,

= 0.039) and ILA (FC = 1.37,

–9

), and ratios of ILA/IPA (FC

= 1.40,

= 0.004) and ILA/IS (FC = 1.16,

= 0.013) were all found to be significantly higher in men than in women.

Table 1.

Subject demographic and clinical characteristics.

HAM-A

Hamilton Anxiety Rating Scale,

HAM-D

Hamilton Depression Rating Scale,

QIDS-SR

Quick Inventory of Depressive Symptomatology, Self-report.

Va r i a b l e

N (%) (N = 196)

Sex (Female)

122 (62.2%)

Race

White

73 (37.2%)

Native American

58 (29.6%)

Black

38 (19.4%)

Asian

2 (1.0%)

Multiracial

14 (7.1%)

Unknown/not reported

11 (5.6%)

Hispanic ethnicity (%)

77 (39.3%)

Mean (SD)

Age (years)

39.11 (11.77)

Body mass index

28.73 (6.27)

HAM-D 17-item total score (baseline)

19.76 (3.77)

QIDS-SR total score (baseline)

14.27 (3.83)

HAM-A total score (baseline)

16.28 (5.37)

HAM-A psychic anxiety subscale score (baseline)

10.79 (2.69)

HAM-A somatic anxiety subscale score (baseline)

5.48 (3.73)

HAM-D 17-item total score (post-treatment)

7.19 (6.11)

QIDS-SR total score (post-treatment)

5.05 (4.25)

HAM-A total score (post-treatment)

6.64 (5.82)

HAM-A psychic anxiety subscale score (post-treatment)

4.08 (3.62)

HAM-A somatic anxiety subscale score (post-treatment)

2.56 (3.12)

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

Associations of indole metabolites with depression and anxiety.

Figure

shows a heat map of correlations

between baseline indole abundance/ratio and baseline levels of the HAM-D total score, Hamilton Anxiety Rat-

ing Scale (HAM-A)

total score, and HAM-A psychic and somatic subscores. Greater abundance of IS was asso-

ciated with higher scores on the HAM-D 17-item total score (

= 0.21,

= 0.018), HAM-A total score (

= 0.26,

= 0.002), and HAM-A psychic subscore (

= 0.31,

= 0.0001), but not on the HAM-A somatic subscore. Addi-

tionally, the ratios of ILA/IS and IPA/IS were negatively correlated with HAM-A total and psychic scores (all

s > -0.20, all

< 0.033), for which a negative correlation indicates that increasingly severe symptoms are associ-

ated with a relative increase in IS and/or a relative decrease in ILA or IPA. Additionally, IPA/IS was negatively

correlated with HAM-D total score (

= − 0.24,

= 0.001).

The same analyses were conducted with the sample stratified by sex (see Supplemental Figs.

and

). The pat-

tern of correlations between groups was similar, although there were more significant correlations in the female

sample, likely due to greater statistical power. In particular, IS was positively correlated with HAM-A psychic

subscore in the females (

= 0.32,

= 0.002) and the males, but the correlation in the male group did not reach

statistical significance (

= 0.28,

= 0.071).

Associations of indole metabolites with individual symptoms of depression.

Correlations between Quick Inven-

tory of Depressive Symptoms-Self-Report (QIDS-SR)

items, and total scores and indole abundances/ratios

are presented in Fig.

. Of note, IS positively correlated with items 4 (hypersomnia;

= 0.22,

= 0.016) and 6

(decreased appetite;

= 0.20,

= 0.034), and the IPA/IS ratio negatively correlated with QIDS-SR total score

(

= -0.21,

= 0.027). The same analyses were conducted with the sample stratified by sex (see Supplemental

Figs.

and

). In the females, IAA/IS and IPA/IS negatively correlated with item 5 (feeling sad;

s = − 0.28 and

− 0.25,

= 0.009 and 0.030, respectively) and QIDS-SR total score (

s = − 0.23 and − 0.25,

s = 0.049 and 0.029,

respectively). In the males, item 13 (general interest) was positively correlated with IS (

= 0.33, p = 0.021), and

item 15 (feeling slowed down) was positively correlated with IAA (

= 0.33,

= 0.026).

Figure 2.

Heat map of Holm-corrected partial Spearman rank correlations between baseline indole abundance/

ratio and Hamilton Anxiety scores and Hamilton Depression scores, after accounting for age, sex, and BMI.

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

Treatment effects.

Compound abundance significantly increased from pre- to post-treatment for ILA

(FC = 1.05,

= 0.006), but not for any other compound/ratio (all

> 0.12). This indicates that the treatments had

limited overall effect on indole composition and levels.

For post-treatment indole abundances and ratios, there were significant correlations between IAA/IS and

QIDS-SR item 15 (feeling slowed down;

= 0.30,

= 0.007). Additionally, post-treatment ILA abundance was sig-

nificantly higher in men than in women (FC

= 1.24,

= 0.00001), as was the ILA/IPA ratio (FC

= 1.54,

= 0.002).

Conversely, the IPA/IS ratio was lower in men than in women (FC

= 0.80,

= 0.048). No other significant post-

treatment effects were observed.

For fold changes, change in IAA/IS ratio correlated with post-treatment scores of QIDS-SR item 5 (feeling

sad;

= 0.27,

= 0.032). No significant associations were observed when correlating indole fold changes with

any other post-treatment scores, or with change in depression/anxiety scores (all

s >

0.10). Additionally, no

sex differences were observed for fold changes (all

s > 0.07), and no differences in fold changes were observed

between response outcome groups (all

s > 0.12).

Baseline levels of indoles and their ratios did not significantly correlate with changes in symptoms for any

measure or item (all

s < 0.15,

s > 0.59). Comparison of categorical response outcomes also showed no mean

ingful differences in baseline indole abundances or ratios. These analyses indicate that pre-treatment indole

compound abundances are not predictive of eventual treatment outcomes.

Treatment-specific effects.

Exposure to CBT or medication (i.e., regardless of outcome) was associated with

differential changes in indole measures. Medication-treated patients demonstrated increases in IPA (FC

= 1.28,

= 0.005) and ILA (FC = 1.09,

= 0.007) and decreases in the IAA/IS ratio (FC

= 0.95,

= 0.035) and IAA/IPA

ratio (FC = 0.91,

= 0.002). In contrast, CBT-treated patients had an increase in the IAA/IPA ratio (FC

= 1.56,

= 0.030).

No baseline indole measures emerged as statistically significant moderators to predict differential outcomes

for the two treatments. There were also no statistically significant differences between treatments in terms of the

strength of the correlations between changes in rating scale scores and individual symptom items with changes in

indole measures (all p

> 0.19). However, examining the within-treatment categorical outcomes found that remit

ters to medication had a significant increase in IPA (FC

= 1.32, p

= 0.005) and decrease in ILA/IPA (FC

= 0.78,

p = 0.049), whereas remitters to CBT had a significant decrease in IPA/IS (FC

= 0.54, p

= 0.049).

Associations of indoxyl sulfate with brain resting state functional connectivity.

Given the

observed association between IS and psychic anxiety, relationships of Subcallosal Cingulate Cortex-Functional

Connectivity (SCC-FC) with IS and with psychic anxiety scores were investigated (see Fig.

). IS abundance was

positively correlated with SCC-FC with the bilateral anterior insula, anterior midcingulate cortex (aMCC), sup-

Figure 3.

Heat map of Holm-corrected partial Spearman rank correlations between baseline indole abundance/

ratio and QIDS-SR items and total score, after accounting for age, sex, and BMI.

QIDS-SR

16-item Quick

Inventory of Depressive Symptomatology-Self-Rated.

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

plementary motor area (SMA), and right premotor area (Fig.

A). Psychic anxiety scores showed a significant

positive correlation with SCC-FC with the left aMCC, right precuneus, and right premotor area; there was a neg-

ative correlation with SCC-FC with the ventromedial prefrontal cortex, right orbitofrontal, and left Brodmann

Area 47 (Fig.

B). The conjunction analysis identified one overlapping area: the right premotor region (Fig.

C).

The mediation analyses explored whether the association of IS with psychic anxiety was mediated through its

effects on SCC-FC. Figure

A shows the overall association between IS and psychic anxiety (z

= 1.976,

= 0.048).

Figure

B shows that the identified overlapping area in the SCC-FC analyses—the right premotor region—medi-

ated the association between IS and psychic anxiety (indirect pathway: z

= 2.138,

= 0.033). Because our whole

brain SCC-FC analyses had also found IS concentrations to be significantly associated with two other regions

previously identified in neuroimaging studies of anxiety (the right anterior insula and the aMCC, Fig.

A), we

conducted further mediation analyses incorporating these two regions along with the right premotor region.

Even though the three regions were highly correlated with each other in their functional connectivity to SCC

(Fig.

C), only the right premotor region mediated the relationship between IS and psychic anxiety scores when

all three regions were included in the model (Fig.

D, indirect pathway: z

= 1.991,

= 0.046).

Discussion

Increasing evidence suggests that gut bacteria can complement human metabolism, and that together they

define the metabolome comprised of the collection of small molecules in blood and in different organs. Bac-

teria can further metabolize compounds available through human metabolism, food-intake, and/or human

ingestion of chemicals. Also, humans can further metabolize compounds produced by bacteria, which results

Figure 4.

Resting state functional connectivity of subcallosal cingulate cortex (SCC) associations with

peripheral indoxyl sulfate abundances and psychic anxiety scores. (

) SCC functionally connected regions

showing a significant correlation with indoxyl sulfate abundances. Orange circles identify regions incorporated

into the mediation models. (

) SCC functionally connected regions showing a significant correlation with

psychic anxiety scores. Green circle identifies right premotor region. (

) Conjunction analysis: SCC functionally

connected region showing a significant correlation with both indoxyl sulfate abundances and psychic anxiety

scores. The red circle indicates the only region to emerge in this analysis, the right premotor region.

HAMPSY

Psychic anxiety subscore of the Hamilton Anxiety Rating Scale.

SCC-FC

subcallosal cingulate cortex functional

connectivity.

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

in human-bacteria co-metabolism and the production of a large number of chemicals that can impact human

health, including brain function. Examples include the metabolism of cholesterol and its clearance mediated by

bacteria, which can produce secondary bile acids that we recently implicated in the pathogenesis of Alzheimer’s

disease

. Several compounds produced from the metabolism of phospholipids and choline by gut bacteria

lead to compounds like trimethylamine N-oxide, which have been implicated in cardiovascular diabetes and

CNS

disease

Indoles represent a class of gut bacterially-derived compounds that are produced from tryptophan, an essen-

tial amino acid that can also be converted (through separate pathways) into tryptamine, serotonin, skatol, and

melatonin, among other metabolites involved in CNS functioning and diseases. Mounting evidence suggests

that indoles derived from gut bacterial metabolism exert significant biological effects and may contribute to the

etiology of cardiovascular, metabolic, and psychiatric diseases. To date, research in this area has been mainly

limited to experimental studies in model systems.

In this investigation, we interrogated levels of four indoles produced by gut bacteria and their relationship

to anxiety and depression severity and response to treatment. At baseline, IS abundance was found to positively

correlate with severity of Psychic Anxiety and total anxiety. IPA seems protective, as noted earlier, indicating

Figure 5.

The impact of indoxyl sulfate on psychic anxiety scores is mediated by its effects on the resting state

functional connectivity between the subcallosal cingulate cortex and the right premotor region. (

) Association

between indoxyl sulfate and psychic anxiety scores. (

) Mediation model incorporating the overlapping

area, right premotor region, indicating that the effect of indoxyl sulfate on psychic anxiety is mediated via its

effects on the functional connectivity between the SCC and right premotor region. (

) Significant SCC-FC

correlations between the right anterior insula, right anterior midcingulate cortex, and right premotor region,

which were included in the mediation model shown in (

). (

) Full mediation model incorporating the three

regions showing significant SCC-FC correlations with indoxyl sulfate abundances. Although indoxyl sulfate is

significantly correlated with all three regions, only the pathway through the right premotor region significantly

mediates indoxyl sulfate’s effect on psychic anxiety. Black lines indicate significant associations within the model;

grey lines are insignificant associations. Red line indicates significant mediation of indoxyl sulfate on psychic

anxiety through the indirect pathway of right premotor SCC-FC.

HAMPSY

Psychic anxiety subscore of the

Hamilton Anxiety Rating Scale.

SCC-FC

subcallosal cingulate cortex functional connectivity.

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

that indoles, as a class, can have mixed effects on neuropsychiatric health. Different strains of bacteria can lead

to the production of different indoles; for example, tryptophanase-producing bacteria produce toxic IS while

Clostridium sporogenes

and other bacterial strains produce protective

IPA

. Notably, IS levels did not mean

ingfully change with treatment, and changes in IS were not correlated with improvement in depression or anxi-

ety measures. This suggests that for patients who do not benefit from standard treatments, a novel therapeutic

approach may be to modulate gut microbiome composition and activity.

This association between IS and anxiety may reflect an impact of IS on the functional connectivity between

the SCC and the right premotor region. IS abundances were also associated with greater connectivity of the

SCC with a well-established network for the processing and control of emotionally salient, particularly aversive,

stimuli, comprising the anterior insula and aMCC. The anterior insula, aMCC, and supplementary motor area

form a network that is involved in the attention to, interpretation of, and control of emotional

responses

–

and greater activity in this network correlates positively with subjective

anxiety

. The premotor cortex is

functionally and structurally connected to the SMA and the aMCC, which act together in the preparation and

readiness for voluntary movement in response to internal and external

stimuli

. The aMCC also functions as

a site of integration for the processing of pain and motor

control

. Outputs from this network include projec

tions to the spinal cord and adrenal

medulla

, which may contribute to the sympathetic arousal and heightened

cortisol release under situations of psychic stress. Taken together, an interpretation of these results is that the

co-metabolism of tryptophan by certain gut microbiota that result in the production of IS may induce anxiety

through the activation of established brain networks, and that existing treatments do not specifically resolve this

pathogenetic process when they lead to clinical improvement.

Although the activity of the premotor cortex has not been a major focus in studies of anxiety and depression,

Pierson and

colleagues

, using the electroencephalography measure of contingent negative variation (which

localizes to the premotor

cortex

), demonstrated abnormal activation of this region in anxious MDD patients

compared to MDD patients with psychomotor retardation. Ma and

colleagues

found that patients with gen-

eralized anxiety disorder have increased resting state functional connectivity between the habenula and right

premotor cortex. Others have found abnormal premotor function or connectivity in social anxiety

disorder

Studies of healthy controls revealed that premotor cortex is significantly involved in processing abstract emotional

concepts

and in threat anticipation, perhaps reflecting an unconscious preparation for

action

Our finding of a positive association between IS and connectivity of the SCC and the insula bilaterally is

consistent with the insula’s known involvement in processes relevant to anxiety, including emotional

salience

empathy for others’ pain, and processing of

uncertainty

–

. In contrast, we did not find an association between

IS abundances and somatic anxiety scores, nor was there an association of IS abundances with functional con-

nectivity of the SCC-posterior insula, the insular region involved in sensorimotor integration. This reveals the

specificity of the IS-anterior insula association for psychic anxiety.

The preceding discussion of the significant associations between the peripheral indole concentrations and

brain network activity assumes that the direction of effect is that the metabolites are impacting brain function.

However, because the gut-brain axis is bidirectional we cannot rule out the possibility that the brain activity pat-

terns observed could have been contributing to the environment and activity of the gut to result in the observed

metabolomic profiles.

Conceptualizing psychic anxiety as a chronic aversive stimulus akin to long-term pain may explain the posi-

tive correlation between higher IS levels and the QIDS-SR loss of appetite item. In mice, inflammatory pain is

inhibited in the presence of hunger, mediated by neuropeptide Y signaling in the parabrachial

nucleus

. The

association of higher IS concentrations with both reduction in appetite and increased connectivity between

brain regions involved in pain processing (anterior insula and aMCC) may indicate that the symptom of low

appetite reflects a compensatory response to this chronic anxiety-type pain. The observed positive correlations

between change in change in the IAA/IS ratio items 15 (“Feeling slowed down”) and 5 (“Feeling sad”) also war

rant comment. First, these results indicate that changes in the relative concentrations across the individual indole

metabolites may have greater explanatory value than considering the individual metabolites in isolation. Second,

the positive association of the IAA/IS ratio with item 15 is consistent with our conclusion that IS is particularly

relevant for anxiety states, given that psychomotor slowing is rare in anxious depressed outpatients.

Limitations of this study include the absence of a healthy control comparison group and the inability to con-

trol for diet and fasting status. We did not pre-specify the items on the QIDS-SR to evaluate, so the identified

associations should be considered tentative until replicated. We lacked fecal samples to analyze which would

have allowed for a more direct correlation between specific gut microbiome species and the IS measures. We

could not determine whether IS is the etiological agent of the anxiety because IS also acts to reduce the integrity

of the blood brain

barrier

, thereby creating the possibility that CNS penetration by an alternative molecule in

the periphery is responsible for the observed association between anxiety and IS.

Taken together, our results indicate that increases in IS are associated with greater connectivity within an

established brain network that is involved in the processing and control of aversive stimuli, but that the conscious

experience of anxiety depends upon the degree of IS-related activation of SCC-right premotor cortex functional

connectivity. The absence of an association between psychic anxiety scores and anterior insula/aMCC SCC-FC

(Fig.

B) may indicate that although IS concentrations are associated with level of connectivity of this control

network in all patients, it is only when network function is inadequate that psychic anxiety ensues in conjunction

with premotor activation in preparation for

action

. These analyses reveal the potential of integrated peripheral

metabolomic-neuroimaging analyses to reveal mechanistic pathways that are associated with neuropsychiatric

symptoms, especially for characterizing the pathological impact of specific gut microbiome-derived metabolites.

Although our results require replication before they can be applied in clinical practice, this work emphasizes

several important considerations for investigations into the impact of metabolomics on psychiatric disorders.

Among these considerations are the need to take into account the clinical heterogeneity of the DSM-defined

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

disorders when examining effects of metabolites, the importance of evaluating all components within a meta

bolic pathway rather than single metabolites, and the tremendous potential of integrating neuroimaging and

metabolomics analyses for understanding pathophysiology and developing surrogate markers of brain function

relevant to the diagnosis and treatment of psychiatric disorders.

Received: 6 May 2021; Accepted: 15 September 2021

References

1. Frank, D. N. & Pace, N. R. Gastrointestinal microbiology enters the metagenomics era.

Curr. Opin. Gastroenterol.

, 4–10 (2008).

2. Foster, J. A. & Neufeld, K. A. M. Gut–brain axis: How the microbiome influences anxiety and depression.

Trends Neurosci.

312–315 (2013).

Sherwin, E., Sandhu, K. V., Dinan, T. G. & Cryan, J. F. May the force be with you: The light and dark sides of the microbiota–gut–

brain axis in neuropsychiatry.

CNS Drugs

, 1019–1041 (2016).

4. Kelly, J. R.

et al.

Breaking down the barriers: The gut microbiome, intestinal permeability and stress-related psychiatric disorders.

Front. Cell. Neurosci.

, 392 (2015).

Sampson, T. R. & Mazmanian, S. K. Control of brain development, function, and behavior by the microbiome.

Cell Host Microbe

, 565–576 (2015).

Forsythe, P. & Kunze, W. A. Voices from within: Gut microbes and the CNS.

Cell Mol. Life Sci.

, 55–69 (2013).

Adams, J. B., Johansen, L. J., Powell, L. D., Quig, D. & Rubin, R. A. Gastrointestinal flora and gastrointestinal status in children

with autism–comparisons to typical children and correlation with autism severity.

BMC Gastroenterol.

, 22 (2011).

8. Li, C.

et al.

Gut microbiota differs between Parkinson’s disease patients and healthy controls in northeast China.

Front. Mol. Neu-

rosci.

, 171 (2019).

9. Aizawa, E.

et al.

Possible association of Bifidobacterium and Lactobacillus in the gut microbiota of patients with major depressive

disorder.

J. Affect Disord.

202

, 254–257 (2016).

10.

Xu, R.

et al.

Altered gut microbiota and mucosal immunity in patients with schizophrenia.

Brain Behav. Immun.

, 120–127

(2020).

11.

Kelly, J. R.

et al.

Transferring the blues: Depression-associated gut microbiota induces neurobehavioural changes in the rat.

Psychiatr. Res.

, 109–118 (2016).

12.

Zheng, P.

et al.

Gut microbiome remodeling induces depressive-like behaviors through a pathway mediated by the host’s metabo-

lism.

Mol. Psychiatry

, 786–796 (2016).

13.

Agus, A., Planchais, J. & Sokol, H. Gut microbiota regulation of tryptophan metabolism in health and disease.

Cell Host Microbe

, 716–724 (2018).

14.

Gao, J.

et al.

Impact of the gut microbiota on intestinal immunity mediated by tryptophan metabolism.

Front. Cell Infect. Microbiol.

, 13 (2018).

15.

Wikoff, W. R.

et al.

Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites.

Proc. Natl.

Acad. Sci. U S A

106

, 3698–3703 (2009).

16.

Banoglu, E. & King, R. S. Sulfation of indoxyl by human and rat aryl (phenol) sulfotransferases to form indoxyl sulfate.

Eur. J.

Drug. Metab. Pharmacokinet.

, 135–140 (2002).

17.

Niwa, T., Ise, M. & Miyazaki, T. Progression of glomerular sclerosis in experimental uremic rats by administration of indole, a

precursor of indoxyl sulfate.

Am. J. Nephrol.

, 207–212 (1994).

18.

Schroeder, J. C.

et al.

The uremic toxin 3-indoxyl sulfate is a potent endogenous agonist for the human aryl hydrocarbon receptor.

Biochemistry

, 393–400 (2010).

19.

Lamas, B., Natividad, J. M. & Sokol, H. Aryl hydrocarbon receptor and intestinal immunity.

Mucosal Immunol.

, 1024–1038

(2018).

20.

Rothhammer, V.

et al.

Type I interferons and microbial metabolites of tryptophan modulate astrocyte activity and central nervous

system inflammation via the aryl hydrocarbon receptor.

Nat. Med.

, 586–597 (2016).

21.

Rothhammer, V.

et al.

Microglial control of astrocytes in response to microbial metabolites.

Nature

557

, 724 (2018).

22.

Chyan, Y. J.

et al.

Potent neuroprotective properties against the Alzheimer β-amyloid by an endogenous melatonin-related indole

structure, indole-3-propionic acid.

J. Biol. Chem.

274

, 21937–21942 (1999).

23.

Bliss, E. S. & Whiteside, E. The gut-brain axis, the human gut microbiota and their integration in the development of obesity.

Front.

Physiol.

, 900 (2018).

24.

Pappolla, M. A.

et al.

Indoles as essential mediators in the gut-brain axis. Their role in Alzheimer’s disease.

Neurobiol. Dis.

156

105403 (2021).

25.

Osadchiy, V., Martin, C. R. & Mayer, E. A. The gut-brain axis and the microbiome: Mechanisms and clinical implications.

Clin.

Gastroenterol. Hepatol.

, 322–332 (2019).

26.

Lin, Y. T.

et al.

Protein-bound uremic toxins are associated with cognitive function among patients undergoing maintenance

hemodialysis.

Sci. Rep.

, 20388 (2019).

27.

Gevi, F., Zolla, L., Gabriele, S. & Persico, A. M. Urinary metabolomics of young Italian autistic children supports abnormal tryp

tophan and purine metabolism.

Mol. Autism

, 47 (2016).

28.

Cassani, E.

et al.

Increased urinary indoxyl sulfate (indican): New insights into gut dysbiosis in Parkinson’s disease.

Parkinsonism

Relat. Disord.

, 389–393 (2015).

29.

Banoglu, E. & King, R. S. Sulfation of indoxyl by human and rat aryl (phenol) sulfotransferases to form indoxyl sulfate.

Eur. J. Drug

Metab. Pharmacokinet.

, 135–140 (2002).

30.

Adesso, S.

et al.

Indoxyl sulfate affects glial function increasing oxidative stress and neuroinflammation in chronic kidney disease:

Interaction between astrocytes and microglia.

Front. Pharmacol.

, 370 (2017).

31.

Karbowska, M.

et al.

Neurobehavioral effects of uremic toxin–indoxyl sulfate in the rat model.

Sci. Rep.

, 9483 (2020).

32.

Bobot, M.

et al.

Uremic toxic blood-brain barrier disruption mediated by AhR activation leads to cognitive impairment during

experimental renal dysfunction.

J. Am. Soc. Nephrol.

, 1509–1521 (2020).

33.

Jaglin, M.

et al.

Indole, a signaling molecule produced by the gut microbiota, negatively impacts emotional behaviors in rats.

Front.

Neurosci.

, 216 (2018).

34.

Merchak, A. & Gaultier, A. Microbial metabolites and immune regulation: New targets for major depressive disorder.

Brain Behav.

Immunity Health

, 100169 (2020).

35.

Dunlop, B. W. & Mayberg, H. S. Neuroimaging-based biomarkers for treatment selection in major depressive disorder.

Dialogues

Clin. Neurosci.

, 507–518 (2014).

36.

Bilello, J. A.

et al.

MDDScore: Confirmation of a blood test to aid in the diagnosis of major depressive disorder.

J. Clin. Psychiatry

, e199-206 (2015).

Vol:.(1234567890)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

37.

Gabbay, V.

et al.

The possible role of the kynurenine pathway in adolescent depression with melancholic features.

J. Child Psychol.

Psychiatry Allied Discip.

, 935–943 (2010).

38.

Bradley, K. A. L.

et al.

The role of the kynurenine pathway in suicidality in adolescent major depressive disorder.

Psychiatry Res.

227

, 206–212 (2015).

39.

Liu, D.

et al.

Beta-defensin 1, aryl hydrocarbon receptor and plasma kynurenine in major depressive disorder: Metabolomics-

informed genomics.

Transl. psychiatry

, 10 (2018).

40.

Bhattacharyya, S.

et al.

Metabolomic signature of exposure and response to citalopram/escitalopram in depressed outpatients.

Transl. Psychiatry

, 173 (2019).

41.

Dunlop, B. W.

et al.

Predictors of remission in depression to individual and combined treatments (PReDICT): Study protocol for

a randomized controlled trial.

Tr i al s

, 106 (2012).

42.

Dunlop, B. W.

et al.

Effects of patient preferences on outcomes in the Predictors of Remission in Depression to Individual and

Combined Treatments (PReDICT) study.

Am. J. Psychiatry

174

, 546–556 (2017).

43.

Dunlop, B. W.

et al.

Functional connectivity of the subcallosal cingulate cortex and differential outcomes to treatment with

cognitive-behavioral therapy or antidepressant medication for major depressive disorder.

Am. J. Psychiatry

174

, 533–545 (2017).

44.

Hamilton, M. A rating scale for depression.

J Neurol Neurosurg Psychiatry

, 56–62 (1960).

45.

Hamilton, M. The assessment of anxiety states by rating.

Br. J. Med. Psychol.

, 50–55 (1959).

46.

Maier, W., Buller, R., Philipp, M. & Heuser, I. The Hamilton Anxiety Scale: Reliability, validity and sensitivity to change in anxiety

and depressive disorders.

J. Affect. Disord.

, 61–68 (1988).

47.

Rush, A. J.

et al.

The inventory for depressive symptomatology (IDS): Preliminary findings.

Psychiatry Res.

, 65–87 (1986).

48.

Cox, R. W. AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages.

Comput. Biomed. Res.

, 162–173 (1996).

49.

Cox, R. W. & Hyde, J. S. Software tools for analysis and visualization of fMRI data.

NMR Biomed.

, 171–178 (1997).

50.

Desikan, R. S.

et al.

An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions

of interest.

Neuroimage

, 968–980 (2006).

51.

Taylor, P. A. & Saad, Z. S. FATCAT: (an efficient) functional and tractographic connectivity analysis toolbox.

Brain Connect.

523–535 (2013).

52.

Fiehn, O. Metabolomics by gas chromatography–mass spectrometry: Combined targeted and untargeted profiling.

Curr. Protoc.

Mol. Biol.

114

, 30–34 (2016).

53.

Fiehn, O.

et al.

Quality control for plant metabolomics: Reporting MSI-compliant studies.

Plant J.

, 691–704 (2008).

54.

Lai, Z.

et al.

Identifying metabolites by integrating metabolome databases with mass spectrometry cheminformatics.

Nat. Methods

, 53–56 (2018).

55.

Fan, S.

et al.

Systematic error removal using random forest for normalizing large-scale untargeted lipidomics data.

Anal. Chem.

, 3590–3596 (2019).

56.

MahmoudianDehkordi, S.

et al.

Alterations in acylcarnitines, amines, and lipids inform about mechanism of action of citalopram/

escitalopram in major depression.

Transl. Psychiatry

, 153 (2021).

57.

Aickin, M. & Gensler, H. Adjusting for multiple testing when reporting research results: The Bonferroni vs Holm methods.

Am.

J. Public Health

, 726–728 (1996).

58.

Rosseel, Y. Iavaan: An R package for structural equation modeling and more.

J. Stat. Softw.

, 1–36 (2012).

59.

Mahmoudian Dehkordi, S.

et al.

Altered bile acid profile associates with cognitive impairment in Alzheimer’s disease: An emerging

role for gut microbiome.

Alzheimers Dement.

, 76–92 (2019).

60.

Nho, K.

et al.

Altered bile acid profile in mild cognitive impairment and Alzheimer’s disease: Relationship to neuroimaging and

CSF biomarkers.

Alzheimers Dement.

, 232–244 (2019).

61.

Ascher, S. & Reinhardt, C. The gut microbiota: An emerging risk factor for cardiovascular and cerebrovascular disease.

Eur. J.

Immunol.

, 564–575 (2018).

62.

Heianza, Y., Ma, W., Manson, J. E., Rexrode, K. M. & Qi, L. Gut microbiota metabolites and risk of major adverse cardiovascular

disease events and death: A systematic review and meta-analysis of prospective studies.

J. Am. Heart Assoc.

, e004947 (2017).

63.

Wikoff, W. R.

et al.

Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites.

Proc. Natl.

Acad. Sci. USA

106

, 3698–3703 (2009).

64.

Elsden, S. R., Hilton, M. G. & Waller, J. M. The end products of the metabolism of aromatic amino acids by clostridia.

Arch.

Microbiol.

107

, 283–288 (1976).

65.

Morawetz, C., Bode, S., Derntl, B. & Heekeren, H. R. The effect of strategies, goals and stimulus material on the neural mechanisms

of emotion regulation: A meta-analysis of fMRI studies.

Neurosci. Biobehav. Rev.

, 111–128 (2017).

66.

Vogt, K. M., BeckerCJ, W. A. D. & Ibinson, J. W. Human posterior insula functional connectivity differs between electrical pain

and the resting state.

Brain Connect.

, 786–794 (2016).

67.

Strigo, I. A., Simmons, A. N., Matthews, S. C., Arthur, D. & Paulus, M. P. Association of major depressive disorder with altered

functional brain response during anticipation and processing of heat pain.

Arch. Gen. Psychiatry

, 1275–1284 (2008).

68.

Shirer, W. R., Ryali, S., Rykhlevskaia, E., Menon, V. & Greicius, M. D. Decoding subject-driven cognitive states with whole-brain

connectivity patterns.

Cereb. Cortex

, 158–165 (2012).

69.

Wager, T. D.

et al.

An fMRI-based neurologic signature of physical pain.

N. Engl. J. Med.

368

, 1388–1397 (2013).

70.

Alvarez, R. P.

et al.

Increased anterior insula activity in anxious individuals is linked to diminished perceived control.

Transl.

Psychiatry

, e591 (2015).

71.

Arnold Anteraper, S.

et al.

Hyper-connectivity of subcortical resting-state networks in social anxiety disorder.

Brain Connect.

81–90 (2014).

72.

Cunnington, R., Windischberger, C., Deecke, L. & Moser, E. The preparation and readiness for voluntary movement: A high-field

event-related fMRI study of the Bereitschafts-BOLD response.

Neuroimage

, 404–412 (2003).

73.

Misra, G. & Coombes, S. A. Neuroimaging evidence of motor control and pain processing in the human midcingulate cortex.

Cereb. Cortex

, 1906–1919 (2015).

74.

Dum, R. P., Levinthal, D. J. & Strick, P. L. Motor, cognitive, and affective areas of the cerebral cortex influence the adrenal medulla.

Proc. Natl. Acad. Sci. USA

113

, 9922–9927 (2016).

75.

Pierson, A.

et al.

Loss of control of pre-motor activation in anxious agitated and impulsive depressives. A clinical and ERP study.

Prog. Neuro-Psychopharmacol. Biol. Psychiatry

, 1037–1050 (1994).

76.

Hultin, L.

et al.

Neuromagnetic localization of the late component of the contingent negative variation.

Electroencephalogr. Clin.

Neurophysiol.

, 425–448 (1996).

77.

Ma, Z.

et al.

Identifying generalized anxiety disorder using resting state habenular circuitry.

Brain Imaging Behav.

, 1406–1418

(2020).

78.

Muzzarelli, L., Tettamanti, M., Taddei, M. & Battaglia, M. The emotional processing network at the onset of social anxiety disorder:

A combined diffusion tensor imaging and functional connectivity study.

Eur. Neuropsychopharmacol.

, S328–S329 (2016).

79.

Moseley, R., Carota, F., Hauk, O., Mohr, B. & Pulvermüller, F. A role for the motor system in binding abstract emotional meaning.

Cereb. Cortex

, 1634–1647 (2012).

80.

Drabant, E. M.

et al.

Experiential, autonomic, and neural responses during threat anticipation vary as a function of threat intensity

and neuroticism.

Neuroimage

, 401–410 (2011).

Vol.:(0123456789)

Scientific Reports

| (2021) 11:21011 |

https://doi.org/10.1038/s41598-021-99845-1

www.nature.com/scientificreports/

81.

Seeley, W. W.

et al.

Dissociable intrinsic connectivity networks for salience processing and executive control.

J. Neurosci.

2349–2356 (2007).

82.

Cauda, F.

et al.

Functional connectivity of the insula in the resting brain.

Neuroimage

, 8–23 (2011).

83.

Singer, T., Critchley, H. D. & Preuschoff, K. A common role of insula in feelings, empathy and uncertainty.

Trends Cogn. Sci.

334–340 (2009).

84.

Kurth, F., Zilles, K., Fox, P. T., Laird, A. R. & Eickhoff, S. B. A link between the systems: Functional differentiation and integration

within the human insula revealed by meta-analysis.

Brain Struct. Funct.

214

, 519–534 (2010).

85.

Alhadeff, A. L.

et al.

A neural circuit for the suppression of pain by a competing need state.

Cell

173

, 140–152 (2018).

86.

Nguyen, V. T., Breakspear, M. & Cunnington, R. Reciprocal interactions of the SMA and cingulate cortex sustain premovement

activity for voluntary actions.

J. Neurosci.

, 16397–16407 (2014).

Acknowledgements

We acknowledge the editorial services of Mr. Jon Kilner, MS, MA (Pittsburgh) and the assistance of Ms. Lisa

Howerton (Duke). This work was funded by grant support to Dr. Rima Kaddurah-Daouk (PI) through NIH

grants R01MH108348, R01AG046171 and U01AG061359. Dr. Boadie Dunlop has support from NIH grants

P50-MH077083 (PI Mayberg), R01-MH080880 (PI Craighead), UL1-RR025008 (PI Stevens), M01-RR0039 (PI

Stevens) and the Fuqua Family Foundations. A preprint of this manuscript is available on bioRxiv (

https://

doi.

org/

10. 1101/

2020.

12.

08. 388942

.).

Author contributions

C.R.B. did analysis of data and helped write the manuscript; O.F. and his team generated biochemical data and

wrote its methods and helped with interpretation of findings; H.S.M., W.E.C., J.C., K.S.C. and B.W.D. did analysis

connecting metabolomics data to imaging data and helped with writing of manuscript; S.M.D., S.B. and H.S.

helped with background literature searches and with interpretation of findings; R.R.K., B.W.D. and A.J.R. helped

with interpretation of findings and clinical relevance; R.K.D. is P.I. for project helped with concept development,

study design, data interpretation and connecting biochemical and clinical data, and with writing of manuscript.

Competing interests

Dr. Dunlop has received research support from Acadia, Compass, Aptinyx, NIMH, Sage, and Takeda, and has

served as a consultant to Greenwich Biosciences, Myriad Neuroscience, Otsuka, Sage, and Sophren Therapeutics.

Dr. Rush has received consulting fees from Compass Inc., Curbstone Consultant LLC, Emmes Corp., Holmusk,

Johnson and Johnson (Janssen), Liva-Nova, Neurocrine Biosciences Inc., Otsuka-US, Sunovion; speaking fees

from Liva-Nova, Johnson and Johnson (Janssen); and royalties from Guilford Press and the University of Texas

Southwestern Medical Center, Dallas, TX (for the Inventory of Depressive Symptoms and its derivatives). He is

also named co-inventor on two patents: U.S. Patent No. 7,795,033: Methods to Predict the Outcome of Treat-

ment with Antidepressant Medication, Inventors: McMahon FJ, Laje G, Manji H, Rush AJ, Paddock S, Wilson

AS; and U.S. Patent No. 7,906,283: Methods to Identify Patients at Risk of Developing Adverse Events During

Treatment with Antidepressant Medication, Inventors: McMahon FJ, Laje G, Manji H, Rush AJ, Paddock S. Dr.

Mayberg receives consulting and intellectual property licensing fees from Abbott Neuromodulation. Dr. Krishnan

is a holder of number of patents in the metabolomic and brain computer interface space some of which have

been licensed to Chymia LLC and sublicensed to Psyprotalix. Dr. Kaddurah-Daouk in an inventor on a series of

patents on use of metabolomics for the diagnosis and treatment of CNS diseases and holds equity in Metabolon

Inc. The other authors declare no competing interests.

Additional information

Supplementary Information

The online version contains supplementary material available at

https:// doi. org/

10. 1038/ s41598- 021- 99845-1

Correspondence

and requests for materials should be addressed to B.W.D. or R.K.-D.

Reprints and permissions information

is available at

www.nature.com/reprints

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and

institutional affiliations.

Open Access

This article is licensed under a Creative Commons Attribution 4.0 International

License, which permits use, sharing, adaptation, distribution and reproduction in any medium or

format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the

Creative Commons licence, and indicate if changes were made. The images or other third party material in this

article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the

material. If material is not included in the article’s Creative Commons licence and your intended use is not

permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from

the copyright holder. To view a copy of this licence, visit

http:// creat

iveco

mmons. org/ licen

ses/ by/4. 0/