Functional consequences of acute tryptophan depletion on raphe nuclei connectivity and network organization in healthy women

Journal Pre-proof Functional consequences of acute tryptophan depletion on raphe nuclei connectivity and network organization in healthy women Karl-Jürgen Bär, Stefanie Köhler, Feliberto de la Cruz, Andy Schumann, Florian D. Zepf, Gerd Wagner PII:

S1053-8119(19)30953-X

DOI:

https://doi.org/10.1016/j.neuroimage.2019.116362

Reference:

YNIMG 116362

To appear in:

NeuroImage

Received Date: 10 July 2019 Revised Date:

9 November 2019

Accepted Date: 13 November 2019

Please cite this article as: Bär, Karl.-Jü., Köhler, S., Cruz, F.d.l., Schumann, A., Zepf, F.D., Wagner, G., Functional consequences of acute tryptophan depletion on raphe nuclei connectivity and network organization in healthy women, NeuroImage (2019), doi: https://doi.org/10.1016/ j.neuroimage.2019.116362. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.

Functional consequences of acute tryptophan depletion on raphe nuclei connectivity and network organization in healthy women

Authors: Karl-Jürgen Bära,1, Stefanie Köhlera, Feliberto de la Cruza, Andy Schumanna, Florian D Zepfb, Gerd Wagnera,1 a

Psychiatric Brain and Body Research Group, Department of Psychiatry and Psychotherapy, Jena University Hospital, Jena, Germany

b

Department of Child and Adolescent Psychiatry, Psychosomatic Medicine and Psychotherapy, Jena University Hospital, Friedrich Schiller University, 07743 Jena, Germany

1

Corresponding authors:

Prof. Dr. Karl-Jürgen Bär Department of Psychiatry and Psychotherapy, Jena University Hospital Philosophenweg 3, 07743 Jena, Germany Email: [email protected]

Dr. Gerd Wagner Department of Psychiatry and Psychotherapy, Jena University Hospital Philosophenweg 3, 07743 Jena, Germany Email: [email protected]

Highlights •

Acute tryptophan depletion (ATD) reduces FC of the 5-HT synthesizing raphe nucleus.

•

Changes in this particular FC correlate with the magnitude of plasma tryptophan availability.

•

ATD increases FC between the right amygdala and the ventromedial prefrontal cortex.

•

ATD increases FC in multiple neural networks compared to tryptophan supplementation.

•

ATD affects the rich-club organization compared to tryptophan supplementation. 1

Abstract Previous research on central nervous serotonin (5-HT) function provided evidence for a substantial involvement of 5-HT in the regulation of brain circuitries associated with cognitive and affective processing. The underlying neural networks comprise core subcortical/cortical regions such as amygdala and medial prefrontal cortex, which are assumed to be modulated amongst others by 5-HT. Beside the use of antidepressants, acute tryptophan depletion (ATD) is a widely accepted technique to manipulate of 5-HT synthesis and its respective metabolites in humans by means of a dietary and non-pharmacological tool. We used a double-blind, randomized, cross-over design with two experimental challenge conditions, i.e. ATD and tryptophan (TRP) supplementation (TRYP+) serving as a control. The aim was to perturb 5-HT synthesis and to detect its impact on brain functional connectivity (FC) of the upper serotonergic raphe nuclei, the amygdala and the ventromedial prefrontal cortex as well as on network organization using resting state fMRI. 30 healthy adult female participants (age: M = 24.5 ± 4.4 yrs.) were included in the final analysis. ATD resulted in a 90% decrease of TRP in the serum relative to baseline. Compared to TRYP+ for the ATD condition a significantly lower FC of the raphe nucleus to the frontopolar cortex was detected, as well as greater functional coupling between the right amygdala and the ventromedial prefrontal cortex. FC of the raphe nucleus correlated significantly with the magnitude of TRP changes for both challenge conditions (ATD & TRYP+). Network-based statistical analysis using time series from 260 independent anatomical ROIs revealed significantly greater FC after ATD compared to TRYP+ in several brain regions being part of the default-mode (DMN) and the executive-control networks (ECN), but also of salience or visual networks. Finally, we observed an impact of ATD on the rich-club organization in terms of decreased rich-club coefficients compared to TRYP+. In summary we could confirm previous findings that the putative decrease in brain 5-HT synthesis via ATD significantly alters FC of the raphe nuclei as well as of specific subcortical/cortical regions involved in affective, but also in cognitive processes. Moreover, an ATD-effect on the so-called rich-club organization of some nodes with the high degree was demonstrated. This may indicate effects of brain 5-HT on the integration of information flow from several brain networks.

2

1. Introduction Previous research on the serotonergic (5-HT) mechanisms and function in the central nervous system provided multiple evidence for its substantial involvement in a variety of cognitive and affective processes (Evers et al., 2010). Abnormal 5-HT neurotransmission has been found in various psychiatric disorders such as major depression (MDD) or obsessivecompulsive disorder (Baudry et al., 2019; Hahn et al., 2014; Kim et al., 2019). For instance, the clinically relevant monoamine-deficiency theory postulates that the underlying pathophysiology of MDD is associated with a deficiency in the function of the monoamine neurotransmitter 5-HT, but also with changes in brain norepinephrine and/or dopamine function, thus impacting several psychopathological symptoms (Willner et al., 2013). Alterations in neurotransmitter function and synthesis are thought to be associated with persistent low mood, anhedonia or anxiety symptoms (Husain and Roiser, 2018; Lowry et al., 2005). There is strong evidence to assume that neural circuitries for emotion regulation and social cognition comprising central subcortical/cortical regions such as amygdala or ventral medial prefrontal cortex (VMPFC) are modulated and partly mediated by 5-HT (Canli and Lesch, 2007; Jasinska et al., 2012). Moreover, increasing evidence from human neuroimaging studies suggests that genetic variations in the5-HT system impact on 5-HT1A binding potentials across several brain areas like the raphe nuclei (David et al., 2005; Kautzky et al., 2017) and can result in abnormal amygdala activation during processing of negative emotions (Pezawas et al., 2005). Regarding the anatomy of the serotonergic system, large ventral ascending serotonergic connections arise mainly from the dorsal raphe nucleus (DRN) and the nucleus centralis superior (NCS) (Nieuwenhuys, 1985). These largest groups of serotonergic neurons are located in the upper brainstem. Animal studies showed that serotonergic fibers project into the amygdala, the hippocampus, the thalamus and cortical regions, including the VMPFC and the anterior cingulate cortex (ACC) (Azmitia and Segal, 1978). Conversely, the upper 5-HT raphe nuclei receive inputs from some of its projection regions, including the amygdala and the VMPFC (Lowry et al., 2008). Amat et al. (2005) clearly demonstrated in rats the inhibition of DRN activity by VMPFC during controllable stress. Furthermore, a modulatory effect of 5HT neurons on amygdala activity has been observed in mice, showing that that 5-HT neurons co-release 5-HT and glutamate onto basal amygdala neurons in a cell-type-specific and frequency-dependent manner (Sengupta et al., 2017). In humans, combining positron 3

emission tomography (PET) imaging, structural MRI and resting state (rs)-fMRI, Beliveau et al. (2015) were able to define DRN and NCS in healthy women based on serotonin transporter (5-HTT) binding and demonstrated significant resting state functional connectivity (RSFC) of both raphe nuclei to the VMPFC and rostral ACC as well as to subcortical regions including amygdala, hippocampus, insula, thalamus and striatum. Beside the use of antidepressants (Harmer, 2008), acute tryptophan depletion (ATD) is a widely accepted technique to manipulate the central nervous synthesis of 5-HT and its metabolites in humans (Young, 2013). By reducing tryptophan (TRP) availability in the periphery of the human body, 5-HT synthesis rate in the brain can be diminished since the essential amino acid TRP serving as the physiological 5-HT precursor but not 5-HT can pass the blood-brain barrier (Pardridge, 1979). In humans, Nishizawa et al. (1997) demonstrated by means of PET a reduction in brain 5-HT synthesis after ATD in different cerebral areas as well as marked sex differences regarding the mean 5-HT synthesis rate. In rats Ardis et al. (2009) showed that ATD specifically altered 5-HT concentration throughout the cortex and hippocampus (except for the striatum), while concentrations of dopamine (DA) or noradrenalin (NA) were not affected. In a further animal study, Cahir et al. (2007) observed a reduction in 5-HT1A autoreceptor binding in raphe nuclei after ATD, while not altering postsynaptic 5-HT1A or 5-HT2A cortico-limbic heteroreceptors. The authors interpreted these results in terms of an intrinsic 'homeostatic response' associated with a reduced serotonergic feedback in dorsal raphe projection areas. Very few human studies investigated the effect of ATD on raphe nuclei fMRI signal. One of these studies reported increased spectral power within an ROI corresponding to the DRN as well as reduced RSFC between DRN and the anterior thalamus after ATD in remitted patients with depression (Salomon et al., 2011). Weinstein et al. (2015) demonstrated lower ATDassociated DRN functional connectivity with the bilateral thalamus and lower DRN functional connectivity with the right rostral ACC in a specific frequency band in contrast to TRYP+ condition in depressed patients. The latter FC separated remitters from non-remitters to antidepressant treatment. Furthermore, the effects of ATD on affective and cognitive processing was reviewed in previous human neuroimaging studies (Evers et al., 2010). Some studies demonstrated an altered BOLD response after ATD administration in the orbitofrontal cortex (OFC), dorsomedial prefrontal cortex (DMPFC) and ACC during response control and negative feedback processing. The authors additionally reported a consistent ATD-related increase in 4

amygdala activation during affective tasks. For instance, Passamonti et al. (2012) demonstrated that ATD significantly modulated connectivity between the amygdala and rostral ACC during processing of angry faces. Beside task-based fMRI studies suggesting an involvement of specific monoamine-producing midbrain/brainstem nuclei in cognitive control processes (Köhler et al., 2016; Köhler et al., 2018; Schumann et al., 2018), functional connectivity (FC) analysis of rs-fMRI data is an established, powerful method to investigate the network architecture of the human brain (Cole et al., 2014). Recently, our group used rs-fMRI in healthy subjects to elucidate patterns of FC and network organization using the graph theoretical (GT) approach that reflects the interaction between neocortex and the monoamine-producing midbrain/brainstem nuclei (Bär et al., 2016). We demonstrated the integration of the serotonergic upper raphe nuclei, i.e. regions inclusive of the DRN and NCS in the default mode network (DMN). In addition, we recently observed disparate functional connectivity of the DRN depending on the class of antidepressant medication in patients with MDD (Wagner et al., 2017). We further recently detected a significantly greater RSFC from the NCS to the frontopolar cortex (FPC) in patients with Borderline Personality Disorder (BPD), which negatively correlated with degree of self-rated aggression (Wagner et al., 2018). The latter has been associated with the abnormal functioning of the brain 5-HT system and altered interactions between prefrontal cortex and amygdala (Passamonti et al., 2012). Since the MR tissue contrast for the raphe nuclei is limited, the raphe ROIs were often defined based on the known anatomical landmarks and available atlases of the human brainstem in various previous studies (Naidich et al., 2009b; Paxinos and Huang, 1995). Previous studies using rs-fMRI and the ATD challenge showed the ATD-related impact on the DMN activation mainly in the OFC and precuneus (Helmbold et al., 2016; Kunisato et al., 2011) in addition to changes in the spectral power within DRN (Salomon et al., 2011). However, the extent to which ATD procedure specifically alters the functional connectivity of the raphe nuclei and of anatomically densely connected cortical/subcortical regions as well its impact on brain organization is still unclear. We therefore conducted a nutritional, non-pharmacological study applying ATD to reduce brain 5-HT synthesis and used rs-fMRI to elucidate its impact on the brain’s FC and network organization. We firstly aimed to investigate whether our anatomical definition of the upper serotonergic ROIs reflects the functionally meaningful regions contributing to the 5

serotonergic neurotransmission in the cortex and to confirm previous findings. In answering this question, we aim to study psychiatric patients with assumed abnormal serotonergic neurotransmission in future studies, such as patients with MDD and to define potential biomarkers. Our second goal was to elucidate the impact of ATD on specific brain regions, i.e. amygdala and VMPFC strongly interacting with the upper raphe nuclei. We specifically hypothesized that the connectivity pattern of anatomically pre-defined upper serotonergic brainstem nuclei would change due to ATD. Based on previous findings of ATD-associated changes in the emotional processing and its impact of the underlying neural circuitry (Evers et al., 2010), we also expected to find specific changes in the FC between upper raphe nuclei, VMPFC and amygdala in terms of an altered connectivity in this particular circuitry. In the second step, and to circumvent some limitations associated with an ROI-based approach, we applied a recently developed statistical method called network-based statistic (NBS) to examine potential changes in the whole brain network connectivity after ATD (Zalesky et al., 2010). Due to the central role of the both raphe nuclei and VMPFC within the DMN and the amygdala within the brain’s salience network (Buckner et al., 2008; Menon, 2015), we additionally expected to find differences in FC in these networks related to decreased 5-HT synthesis. Moreover, since serotonergic projections to cerebral cortex form a widely distributed system (Beliveau et al., 2017), altered interactions with the raphe nuclei may critically influence brain functions and behavior. The so-called rich-club coefficient, a graph-theoretical metric, has been assumed to provide information about global organizational properties of functional and structural brain networks and to indicate the overall brain communication and resilience (Fornito et al., 2016b; van den Heuvel and Sporns, 2011b). In complex networks, hub regions, i.e., nodes with a high number of connections compared to other nodes of the network, tend to be densely connected, forming the so-called “rich club”. Rich-club organization appears to be a fundamental feature of the mammalian brain (Harriger et al., 2012; van den Heuvel and Sporns, 2011a). Furthermore, investigation of the rich club organization in the present study is particularly relevant because growing evidence indicates that psychopathology appears to be associated with abnormal functioning of hub regions (Crossley et al., 2014). For instance, abnormal structural and functional rich-club organization has been found in various 6

psychiatric disorders including schizophrenia (Collin et al., 2014; van den Heuvel et al., 2013b), late-life depression (Mai et al., 2017) and in patients with bipolar disorder (Wang et al., 2019). Very recently, we demonstrated abnormal and specific rich-club organization in MDD patients with suicidal behavior (Wagner et al., accepted), which is associated with altered serotonergic neurotransmission (Ernst et al., 2009). Thus, since alterations in the serotonergic system contribute to specific disorder characteristics, we also expected to find an effect of ATD on the rich-club connectivity.

7

2. Methods 2.1 Study participants Thirty-five female healthy volunteers were recruited. Only women were investigated due to previous studies showing a higher behavioural response in females to ATD (Ellenbogen et al., 1996) as well as due to previously reported greater biochemical effects of ATD on brain 5-HT synthesis in females in several cortical/subcortical brain areas (Nishizawa et al., 1997).

Five participants were excluded due to nausea and associated intolerance of MRI scanning. In all of these five subjects (ATD: n=3; TRYP+: n=2) this was the first point in time of the investigation and was due to the intolerance of the used amino acid mixture. Thus, the final sample consisted of 30 female participants (age: M=24.5 ± 4.4 years). All participants were screened for the absence of any psychiatric disorders using the German version of the MiniInternational Neuropsychiatric Interview (MINI) (Sheehan et al., 1998) and a semi-structured interview for recording past and current neurological as well as psychiatric symptoms and disorders. Subjects’ mean total score on the Beck Depression Inventory (BDI) was before the experiment 4.4 ± 3.8 and 2.1 ± 2.2 on the Hamilton Rating Scale for Depression (HRSD). In addition to the BDI, we used the Self-Assessment Manikin (SAM) to measure changes in mood and arousal ratings due to the challenge intervention (Lang, 1980), which is a very popular self-reporting tools to measure affective states. The SAM is a non-verbal pictorial affective rating system that directly measures mood and arousal levels. SAM ranges from a smiling, happy figure to a sad figure when representing the valence dimension, and ranges from an excited, wide-eyed figure to a relaxed, sleepy figure for the arousal dimension (Bradley and Lang, 1994). SAM has been used to measure emotional responses in a variety of situations and stimuli in healthy subjects as well as in patients (Bradley and Lang, 1994). The 9-point rating scale for mood ratings (valence) ranges from -4 to +4, whereas the 9-point rating scale for arousal levels ranges from 1 to 9. The scale was successfully used in previous experiments applying mood induction procedures by our group (Wagner et al., 2009). The screening procedure also included the assessment and exclusion of subjects with firstdegree relatives with psychiatric disorders. The participants were all right-handed according to the modified version of the Annett’s handedness inventory (Briggs and Nebes, 1975). Informed written consent was obtained in accordance with the protocols approved by the local 8

Ethics Committee, and all subjects received allowance reward of 50 Euros in return for their participation.

2.2 Study design The study made use of a double-blind, randomized, crossover design. Each participant was tested under two respective challenge conditions, i.e. the TRYP+ and the ATD condition. The assignment of participants to both conditions was randomized and counterbalanced. Thus, subjects were randomly assigned to both points in time with respect to ATD or TRP loading conditions to control for unspecific time effects or sequence effects. In this way, 14 subjects were allocated to the first measurement point in time to the ATD condition and 16 to the TRYP+ condition. The two conditions were separated by a 7-day period to rule out potential carry-over effects.

2.3 Procedure On the day before the ATD or TRYP+, participants were required to fast from 7 p.m. until the following day except for the consumption of water. Immediately before the ATD or TRYP+ challenge, participants completed the Self-Assessment Manikin (SAM, Lang, 1980), a blood sample was taken (10 ml) for baseline TRP levels and directly centrifuged. Participants were then administered the respective amino acid (AA) beverage (ATD or TRP+). 100g of the powdered AAs was dissolved in 300 ml of water a few minutes prior to oral administration. We used a fixed collagen-based AA mixture following established methods by Evers et al. (2005) (see Table S1 for the detailed composition of this particular mixture). This mixture was supplemented by L-TRP (10.3g) according to Young et al. (1985). To be comparable with previous studies, which used a weight-adapted dosing regime for the AA beverage, the average weight of study participants was 62.3 kg ± 9.6. Two fMRI experiments, i.e. a resting state scan and a subsequent presentation of an affective paradigm (not reported here) were performed 4h after the administration of the AA beverages. This 4 h latency period was chosen to enable a maximum depletion effect (Williams et al., 1999). Immediately before the fMRI experiment, participants were again administered the

9

SAM and BDI as well as were required to give another blood sample (10 ml) in order to establish respective plasma TRP levels after ATD / TRP+ intake.

2.4 Biochemical Analysis Venous blood samples were separated by centrifugation for 10 min at a relative centrifugal force (RCF) of 2500g. Concentrations of free AAs TRP, tyrosine (TYR), phenylalanine (PHE), valine (VAL), leucine (LEU), and isoleucine (ILE) in plasma were analyzed by Biochrom 30 Plus amino acid analyzer (Biochrom Ltd., UK). VAL, LEU, and ILE levels were analyzed to calculate the ratio of plasma TRP to other large neutral amino acids (LNAAs). 5HT depletion magnitude was calculated using the ratio of TRYP/ΣLNAAs before ATD at baseline minus T1 (after 4 h, immediately before fMRI experiment) (ATDbaseline: TRYP/ΣLNAAs − ATDT1: TRYP/ΣLNAAs). The 5-HT supplementation magnitude was calculated using the ratio of TRYP/ΣLNAAs after TRYP supplementation at baseline minus T1 (TRYP+baseline: TRYP/ΣLNAAs – TRYP+ T1: TRYP/ΣLNAAs).

2.5 MRI Procedure The functional and structural MR data were collected on a 3T whole body system equipped with a 12-element head matrix coil (MAGNETOM TIM Trio, Siemens). The whole measurement consisted of a resting state scan followed by a structural scan and afterwards by an affective paradigm. Participants were instructed to keep their eyes closed during the whole rs-fMRI scan. To reduce head movement during the scan, the NoMoCo memory foam pillow support system (http://www.nomocopillow.com/) was used. NoMoCo is a 12-piece pillow support system kits including a head strap to restrain the forehead. T2*-weighted images were obtained using a gradient-echo EPI sequence accelerated by parallel imaging using GRAPPA (TR = 2520 ms, TE = 30 ms, flip angle = 90°, inter-slice gap = 0.625 mm, GRAPPA factor = 2) with 45 contiguous transverse slices of 2.5 mm thickness covering the entire brain and including the lower brainstem. The matrix size was 88 × 84 pixels with an in-plane resolution of 2.5 × 2.5 mm² corresponding to a field of view of 220 × 210 mm. A series of 240 whole10

brain volume sets were acquired in one session. High-resolution anatomical T1-weighted volume scans (MP-RAGE) were obtained in sagittal orientation (TR = 2300 ms, TE = 3.03 ms, TI = 900 ms, flip angle = 9°, FOV=256 mm, matrix = 256 mm x 256 mm, number of sagittal slices=192, acceleration factor (PAT=2) with an isotropic resolution of 1 × 1 × 1 mm³.

2.6 Physiological recordings during fMRI Photoplethysmogram (PPG) and respiration were recorded throughout MR image acquisition using an MR-compatible polygraph MP150 (BIOPAC Systems Inc., Goleta, CA, USA). Respiratory activity was assessed by a strain gauge transducer incorporated in a belt that was tied around the chest, approximately at the level of the processus xiphoideus. The PPG sensor was attached to the proximal phalanx of the left index finger. All signals were sampled at 500 Hz and amplified in a frequency range of 0.05–3 Hz for PPG and 0.05-10 Hz for respiration. The PPG signal was smoothed over 50 samples and differentiated. Pulse-wave onsets were automatically extracted by detecting peaks of the temporal derivative. The quality of the peak detection was visually inspected.

In addition, skin conductance was measured during the fMRI scan continuously (constant voltage technique) at the left hand’s palm with Ag/AgCl electrodes placed at the thenar and hypothenar eminence. The signal was amplified below 10 Hz, median filtered (150 samples) and temporally smoothed (250 samples) to reduce the influence of noise. Skin conductance changes are mainly dependent on the activity of sweat glands innervated by the sympathetic system and can be used to index sympathetic arousal (Bach et al., 2010). We estimated the overall level of the skin conductance (SCL) by averaging the skin conductance signal. Spontaneous skin conductance fluctuations (SCF) were identified by a pattern matching algorithm using the characteristic shape of an SCF described mathematically by Lim and colleagues (Lim et al., 1997). This approach is robust in the presence of noise and less dependent on the individual electrode contact resistance than other threshold-based approaches (Lim et al., 1997).

2.7 rs-fMRI preprocessing Prior to fMRI preprocessing, physiological fluctuations synchronized with cardiac and respiratory cycles were removed from the fMRI signal using the RETROICOR approach (Glover et al., 2000). To account for respiration related signal changes, five regressors 11

defining the respiration volumes per time (RVT) were created and regressed out from the fMRI data (Birn et al., 2008). Since the exact onset time of respiratory response may vary between subjects, Birn et al. (2008) found that the five time-shifted versions of the RVT function (five RVT regressors) with delays of 0, 5, 10, 15, and 20 seconds are sufficient to properly correct for respiratory confounds. This is the default setting in the AFNI’s RetroTS.m script and the same number of RVT regressors was used in previous studies (Davis et al., 2016; Falahpour et al., 2013; Power et al., 2017). Further regressors were eight low-order Fourier time series (4 based on the cardiac phase and 4 on the respiratory phase), which were created using the RETROICOR algorithm. All regressors were generated on a slice-wise basis by AFNI's RetroTS.m script implemented in MATLAB 2016b, which takes the cardiac and respiratory time series synchronized with the fMRI acquisition as input. By combining the RETROICOR with RVT regressors, we ensured a “cleaned” BOLD signal from almost all respiratory-induced fluctuations.

Further standard preprocessing steps of the whole brain were performed using the SPM12 (http://www.fil.ion.ucl.ac.uk/spm) and AFNI (https://afni.nimh.nih.gov/). The first ten images were discarded to obtain steady-state tissue magnetization. Preprocessing using SPM12 included 3D motion correction, i.e. rigid body realignment to the mean of all images. To compare both conditions with regard to differences in head motion, the framewise Euclidean norm of the six-dimension motion derivatives was computed (Jones et al., 2010). No significant differences (p = 0.41) were found between conditions in this metric. Subsequently, a slice timing correction was performed to ensure that the data on each slice corresponded to the same point in time. Afterwards, a within-subject registration was performed between functional and anatomical images using SPM12. The co-registered anatomical images were segmented and functional images were then spatially normalised to the MNI space using the deformation field created during the segmentation process. The whole-brain data were smoothed using a Gaussian filter of 6 mm FWHM for the univariate functional connectivity analyses only. We did not smooth the functional data for the NBS or the rich-club analyses. Preprocessing using AFNI consisted of further additional steps: (i) removal of linear and quadratic trends, (ii) temporal band-pass filtering, retaining frequencies in the 0.01-0.08Hz band, (iii) removal by multiple regression of several sources of variance, i.e. head-motion parameters, CSF as well as white matter signal. Due to the controversial interpretation of the functional connectivity results using global signal regression, we avoided this step in the preprocessing of the functional data (Murphy et al., 2013). 12

To improve the normalization of the brainstem/cerebellum and to more precisely define both raphe ROIs for subsequent fMRI time-series extraction, we used the SUIT toolbox (SUIT, version 3.1) (Diedrichsen, 2006), available as an open source SPM-toolbox. The toolbox contains a high-resolution template of the human brainstem and cerebellum. This procedure was performed with RETROICOR and slice time corrected as well as realigned whole-brain images. The following preprocessing steps were applied using SUIT toolbox: (i) segmentation of the co-registered T1 image as implemented in SPM12, (ii) cropping of the T1 image, retaining only the cerebellum and brainstem, (iii) normalization using the DARTEL (diffeomorphic anatomical registration through exponentiated lie algebra) engine (Ashburner, 2007) that uses gray and white matter segmentation maps produced during cerebellar isolation to generate a flowfield using Large Deformation Diffeomorphic Metric Mapping (LDDMM, Beg et al., 2005), and (iv) reslicing the functional data using the deformation found in the previous step to a voxel size of 2 × 2 × 2 mm³. Due to the small size of brainstem/midbrain nuclei and their close anatomical location, we did not spatially smooth the normalized images. Using AFNI (http://afni.nimh.nih.gov/afni/), linear and quadratic trends were removed. The data were filtered with a frequency-based band-pass filter (AFNI 3dBandpass), retaining frequencies in the 0.01-0.08 Hz band. Multiple regressors of several sources of variance, i.e. head-motion parameters, CSF as well as white matter signal, but not the global signal were regressed out from the data.

2.8 Definition of the seed regions To study our first hypothesis regarding the impact of ATD on the resting state functional connectivity of the serotonergic neurotransmitter system, two upper seed raphe-ROIs of 4 mm radius were defined using the available atlases of the human brainstem (Naidich et al., 2009b) and as defined in our previous study (Bär et al., 2016). These were the nucleus raphes dorsalis (DRN, B7, MNI-coordinates, x = 2, y = -26, z = -18) and the nucleus centralis superior (B6+B8, MNI-coordinates, x = 0, y = -32, z = -24), from which the ascending serotonergic connections mainly arise to cortical/subcortical regions (Nieuwenhuys, 1985). Both raphe ROIs were drawn in the axial plane based on anatomic slices provided by the Duvernoy's Atlas of the Human Brain Stem and Cerebellum (Section 11, pages 628-643; and figures 11.19– 11.23; (Naidich et al., 2009a)). In this atlas, the DRN and NCS are depicted across two adjacent axial slices of 2 mm thickness (section XI in Duvernoy's Atlas). Using 13

this information and given that both raphe nuclei are commonly represented as ellipsoids in the literature (see Duvernoy's Atlas on page 111), we considered a sphere of 4 mm as adequate to represent both raphe ROIs. However, the ROIs were finally generated as a cube of 6 x 6 x 6 mm3, as shown in Figure S1. The reason for this mismatch is due to the voxel dimension (2 x 2 x 2 mm3) of the brainstem image, using which the raphe ROIs were drawn. Given this resolution it is only possible to generate cubic ROIs, when a sphere of 4 mm radius is specified. Furthermore, given the present fMRI resolution of 2.5 × 2.5 × 2.5 mm³, we considered the generation of a cubic ROI with dimensions of 6 x 6 x 6 mm3 more appropriate to included brainstem voxels containing DRN or NCS nuclei, even if we are aware that we are extending beyond the dorsal raphe and NCS volumes.

We also tested for the specificity of the definition of both raphe ROIs by performing an intervoxel coherence analysis using average time series from the DRN and NCS. A coherence map was generated for each subject between the average time series extracted from the DRN and NCS ROIs and time series from all voxels in the midbrain/brainstem and cerebellum. As illustrated in Figure S2, significant coherence was only observed between average time series from both raphe ROIs and voxels in the according raphe ROIs. Furthermore, there were only a few voxels with overlapping coherence between DRN and NCS.

The cortical VMPFC-ROI was drawn as a sphere of 10mm radius, centered at MNIcoordinates, x = 0, y = 44, z = -14, as defined in our previous study (Bär et al., 2016). The left and right amygdala ROIs were created using the WFU Pick Atlas tool for SPM (Maldjian et al., 2003). In the figure S1 the position of the used ROIs is projected on the whole brain template.

2.9 Univariate Functional Connectivity Analysis Functional connectivity analyses were carried out by correlating the regional time course, which was extracted from both brainstem ROIs inclusive of the DRN and NCS as preprocessed with the SUIT toolbox, against all other voxels within the whole brain. The time series from these ROIs were extracted from the unsmoothed brainstem functional data (as preprocessed with SUIT toolbox) to avoid overlap between ROIs due to smoothing.

14

Functional connectivity was obtained by computing Pearson’s correlation coefficients. After application of Fisher z-transformation to the correlation maps, an ANOVA was set up with one within-subjects factor (ATD vs. TRYP+) and individual FC maps. This within-subjects ANOVA design was used to investigate differences in RSFC between a-priori defined ROIs inclusive of the DRN and NCS, amygdala and VMPFC. The statistical comparisons were thresholded at an uncorrected voxel-level significance of p < 0.001 and an FDR corrected cluster-level significance of p < 0.05, which was used due to the results of the influential paper by Eklund et al. (2016).

2.10 Network-based statistics Potential ATD-induced differences in whole-brain FC were examined using the framework of the Network-Based Statistic (NBS) introduced by Zalesky et al. (2010). NBS is a validated non-parametric method to avoid the multiple comparison problems due to mass univariate significance testing in network connectivity. 260 independent anatomical ROI were defined based on the coordinates from extensively validated parcellation system provided by Power et al. (2011) and covering the whole cerebral cortex, cerebellum and including both raphe ROIs inclusive of the DRN and NCS. Short-distance (less than 20 mm) correlations were discarded following Power et al. (2011) to avoid possible shared signal between nearby nodes. The first step was the calculation of the connectivity matrices for both conditions, i.e. ATD and TRYP+. Secondly, a primary component-forming threshold (with a conservative threshold of p < 0.001) was applied to identify all edges displaying potential differences in connectivity strength using a paired t-test design. Thirdly, all sub-threshold edges were assessed for mutual connections forming topological clusters that may point towards the existence of non-chance clusters. Permutation testing with randomized interchange of group assignment within one person was then applied to compute p-values for every component previously identified. To this end, the first three steps were repeated for each of the 10,000 random permutations of group assignments (i.e. ATD or TRYP+), with noting the maximum cluster sizes of components resulting in a null distribution for the largest component size. For better illustration and representation, components were color-coded according to wellknown resting-state networks (modules) (Bär et al., 2016). To compute these networks, we used the graph theoretical approach. Firstly, an adjacency matrix (Aij) was built by retaining 15% of the strongest connections of the average positive connectivity matrix. Secondly, Aij 15

was subsequently partitioned using the spectral approach based on the leading eigenvector of the modularity matrix (Newman, 2006). This graph partitioning led to four functionally distinct modules: the executive-control, the DMN, the salience and visual networks. We used these networks as a reference to label the regions in the NBS component. NBS analysis was conducted using the Brain Connectivity Toolbox (Rubinov and Sporns, 2010), while graph partitioning and component visualization used in-house programs written in Python.

2.11 Rich-club analysis We also used Graph theory to examine the topology of the brain networks and focused on the investigation of the rich-club organization. The rich-club coefficient Φ(k) at each level of degree k was computed as the ratio of the number of connections between nodes within the kth subgraph and the total number of possible connections between them based on group averaged networks (Fornito et al., 2016a; van den Heuvel et al., 2012). The network density was set to include the top 7% of the strongest positive connections to reduce the number of potentially spurious connections. A similar network density was used in previous studies investigating rich-club organization (Grayson et al., 2014; Ray et al., 2014). We also tested the stability of the rich-club organization across a range of different network densities varying from 5%, 7% and 9%, which showed a stable behavior. The rich-club coefficient was additionally compared and normalized to sets of “equivalent” random networks Φnorm(k). We generated a thousand random networks having equal size and degree distribution. The network is considered to have rich-club organization, when Φnorm(k) is greater than 1 for a continuous range of k (Fornito et al., 2016a). Permutation test was used to compute the significance of the rich-club curves. The generated thousand random networks produced a null distribution of rich-club coefficients (Φrand(k)). Using this distribution, a p-value was assigned to Φnorm(k) as the percentage of random (null) values that exceeded Φrand(k) (p < 0.05, one-tailed). Differences in the rich-club organization between both conditions (TRYP+ vs. ATD) were additionally tested for significance. For each degree k of the random network of one condition Rcond1(k) and the random network of the other condition Rcond2(k), the difference between the rich-club coefficients for Rcond1(k) and for Rcond2(k) produced a null distribution of a thousand random differences. Using this distribution, a p-value was assigned to each observed difference Φcond1(k)–Φcond2(k) (p < 0.05, two-tailed).

16

3. Results 3.1 Plasma tryptophan levels The non-parametric Wilcoxon signed-rank test revealed a significant and marked decrease in TRP concentrations for the ATD condition (Z = -4.8, p<0.001), when compared to baseline. Plasma TRP concentration decreased for the ATD condition (mean ± S.D. concentration: 4.5 ± 3.7 µmol/l) by 90.4% relative to baseline (mean ± S.D. concentration: 46.9 ± 11.7 µmol/l). For TRYP+, a significant increase (Z = -4.4, p < 0.001) in TRP plasma concentration was observed due to the supplementation of TRP compared to baseline (plasma TRP concentrations were 702.9 ± 442.9 µmol/l (mean ± S.D.) at T1, and 52.7 ± 13.8 µmol/l at baseline). Furthermore, regarding the ratio of plasma TRP to other LNAAs, a significant reduction (by 93%) in the ratio of TRP/∑LNAAs was detected relative to baseline in the ATD condition (Z=-4.8, p<0.001). Regarding the ratio of plasma TRP to other LNAAs, for the TRYP+ condition a significant increase (by 916%) in the ratio of TRP/∑LNAAs was detected relative to baseline (Z = -4.4, p < 0.001). Together these data show that plasma TRP concentrations were successfully influenced by the ATD challenge in comparison to TRYP + and baseline.

3.2 Affective ratings We did not detect any significant changes in the BDI total score after ATD (ATD: mean 3.9 ± 4.5 vs. baseline: mean 3.6 ± 4.6) as well as after TRP supplementation (TRYP+: mean 4.2 ± 3.4 vs. baseline: mean 3.9 ± 3.7) compared to baseline assessment. There were no significant differences in the BDI total score between the respective ATD and TRP+ conditions. Wilcoxon’s test did not reveal any significant differences in SAM mood or arousal ratings between the TRYP+ and ATD condition at baseline (T0) as well as immediately before the fMRI experiment (4h later, T1). A highly significant reduction in SAM mood ratings from T0 (mean ± S.D. 1.9 ± 1.3) to T1 (mean ± S.D. 1.1 ± 1.7) was detected in both the ATD (Z = 3.2, p < 0.001) and the TRYP+ conditions (T0: mean ± S.D. 2.1 ± 1.1; T1: mean ± S.D. 0.7 ± 17

1.6; Z = -3.7, p = 0.001), but not for SAM arousal ratings. At both points in time, subjects rated their mood higher than zero (neutral), thus slightly pleasant. No significant interactions were observed when comparing the pre-post (T0-T1) differences in mood and arousal ratings between both conditions. We did not detect any significant differences between ATD and TRYP+ conditions regarding both skin conductance measurements, i.e. SCF and SCL measures, which can be used to index sympathetic arousal (Bach et al., 2010).

3.3 Univariate RSFC analysis 3.3.1 Serotonergic nuclei Comparing ATD with TRYP+, a significantly lower RSFC was detected from the serotonergic NCS to two clusters located in the right cerebellum with the MNI coordinates: x = 26, y = -80, z = -26, t = 4.8, cluster extent = 141, p < 0.05 FDR corrected, and in the left frontopolar cortex (FPC, x = -8, y = 72, z = 4, t = 4.6, cluster extent = 72, p<0.05 FDR corrected), as depicted in Figure 1. There was no significantly greater RSFC from the NCS after ATD compared to TRYP+. We did not detect any significant differences between ATD and TRYP+ regarding RSFC from the DRN at the FDR corrected threshold.

18

Figure 1: Significant differences in RSFC of the ROI inclusive of the serotonergic NCS is depicted comparing ATD to TRYP+ conditions. Significant differences in RSFC from the NCS as seed region to the frontopolar cortex (FPC) comparing TRYP+ with ATD (p < 0.001 uncorrected, p < 0.05 FDR cluster-level corrected). The box plots depict FC for both conditions separately, as extracted from the significant FPC cluster. The error bars represent the standard deviation. The right scatterplot depicts a significantly positive correlation (r = .36, p < 0.05) between the FC of the NCS to FPC and the magnitude of TRP depletion for the ATD condition, indicating a stronger reduction in the ratio of plasma TRP to other LNAAs after ATD compared to baseline associated with lower RSFC. The left scatterplot depicts a significantly positive correlation (r = .47, p < 0.01) between the FC of the NCS to FPC and the magnitude of TRP supplementation in the TRYP+

19

condition, indicating stronger TRP/∑LNAA increase compared to baseline associated with greater RSFC.

3.3.2 Amygdala In contrast to lower NCS-FC, a significantly greater RSFC was detected comparing ATD with TRYP+ condition from the right amygdala to one cluster located in the VMPFC with the MNI coordinates: x = 8, y = 36, z = -18, t = 5.00, cluster extent = 66, p < 0.05 FDR corr. (Figure 2). This cluster slightly overlapped with the VMPFC-ROI, which was a-priori defined and used as seed region in the present study. We did not detect any significant changes in the RSFC from the left amygdala at the chosen statistical threshold. However, relaxing the cluster-level threshold, we also observed significantly decreased RSFC from the left amygdala to the VMPFC (x = -4, y = 32, z = -8, t= 5.00, cluster extent = 13).

20

Figure 2: Significant difference in the RSFC of the right amygdala comparing ATD and TRYP+ Significant differences in RSFC from the right amygdala to the ventromedial prefrontal cortex (VMPFC) comparing ATD with (TRYP+ (p < 0.001 uncorrected, p < 0.05 FDR cluster-level corrected). The box plots depict FC for both conditions separately, as extracted from the significant VMPFC cluster. The error bars represent the standard deviation. The scatterplot depicts a significantly positive correlation (r = .38, p < 0.05) between the Self-Assessment Manikin (SAM) mood rating immediately before the ATD-fMRI session and greater relative RSFC for the ATD condition from the right amygdala to the VMPFC compared to TRYP+.

3.3.4 VMPFC A significantly greater RSFC was detected for ATD compared to TRP+ from the VMPFCROI to one cluster comprising right precuneus and cuneus (x = 22, y = -88, z = 44, t = 5.34, cluster extent = 743, p < 0.05 FDR corr.), to a second cluster located in left precuneus (x = 16, y = -66, z = 46, t = 4.74, cluster extent = 133, p < 0.05 FDR corr.) and to a third cluster located in left superior temporal lobe (x = - 58, y = -4, z = -8, t = 4.28, cluster extent = 76, p < 0.05 FDR corr.).

3.4 Correlation analysis As shown in Figure 1, a significantly negative correlation was detected between the functional connectivity of the NCS to FPC for the ATD condition and the TRP depletion magnitude (r = -.36, p < 0.05) as well as between the FC of the NCS to FPC in the TRP+ condition and the TRP supplementation magnitude (r = -.47, p < 0.01). Based on the notion that VMPFC and the amygdala are centrally involved in processing of affective information, RSFC between these two regions was correlated with the SAM valence ratings immediately before the according fMRI session. We assumed that the ratings before the rs-fMRI session best describe the affective state of study participants after ATD or after TRP supplementation. As shown in Figure 2, we observed a significant correlation between SAM mood rating immediately before the ATD-fMRI session and relatively greater RSFC in 21

the ATD condition from the right amygdala to VMPFC compared to TRYP+ (r = .38, p < 0.05). We did not observe any further significant correlations between VMPFC-amygdala RSFC in the according conditions and mood ratings. We also correlated RSFCs between NCS and FPC and between the VMPFC and the right amygdala to investigate the relationship these FCs. However, this correlation was not significant.

3.5 Network-Based Statistics (NBS) of FC NBS analysis revealed a single network of increased FC in the ATD condition as compared to the TRYP+ condition (p = 0.04, family-wise error rate [FWER]). The network (Figure 3 and Table S1) comprised a total of 81 nodes connected by 86 edges. Based on the modularity obtained by the spectral partition algorithm, 19 nodes were located in the executive-control network (blue color), including the dorsolateral and ventrolateral PFC, the right thalamus, bilateral putamen and dorsal ACC. 31 nodes were labeled as belonging to the DMN (red color) and consisted of the medial prefrontal regions, i.e. VMPFC, DMPFC and orbitofrontal cortex, of the middle temporal gyrus, of the precuneus, right angular gyrus and bilateral inferior parietal lobe. 14 nodes were labeled as belonging to the salience network (green color), consisting of bilateral primary motor (M1), premotor and left somatosensory (S1) cortices, of the right posterior insula, of the superior temporal gyrus and the midcingulate cortex. Finally, 17 nodes located in the occipital cortex belonged to the visual network (pink color).

22

Figure 3: Results of the network-based analysis of the whole-brain network based on the time series extracted from the 260 ROIs as provided by Power et al. (2011). Comparison between both conditions (ATD vs TRYP+) in FC matrices using Network-Based statistics (NBS). The depicted component shows nodes with significantly (p = 0.04, FWE corr.) greater connectivity in the ATD condition compared to the TRYP+ condition. These connections formed a single connected network and comprised a total of 81 nodes connected by 86 edges. Colors indicate the resting-state network where the node belongs to, according to a community structure obtained using Graph theory and the spectral partition algorithm (Newman, 2006). Magenta, visual network; green, salience-network; red, default mode network; blue, executive-control network. Abbreviations: DMPFC; dorsomedial prefrontal cortex; ACC: anterior cingulate cortex; MCC: midcingulate cortex; PCC: posterior cingulate cortex; OFC: orbitofrontal cortex; MFG: middle frontal gyrus; IFG: inferior frontal gyrus; IC: insular cortex; Thal: thalamus; Put: putamen; STG: superior temporal gyrus; MTG: middle temporal gyrus; ITG: inferior temporal gyrus; AnG: angular gyrus; PreC: precuneus; SPL: superior parietal cortex; IPL: inferior parietal cortex; SMA: supplementary motor area; pM: premotor cortex; M1: primary motor cortex; S1: primary somatosensory cortex; FuG: fusiform gyrus; Parahip: parahippocampal gyrus; OCx: occipital cortex; Cereb: cerebellum. 23

3.6 Rich-club analysis We found a significant rich-club organization (rich-club regime) in both conditions across several levels of k (indicated by the dark gray area in Figure 4B), but with a wider rich-club regime in the TRYP+ condition. Furthermore, we observed significantly lower rich-club coefficients (p < 0.05) across two high levels of k and a trend (p < 0.1) across four levels of k for the ATD condition than for the TRYP+ condition (Figure 4A). This result indicates 5-HT synthesis associated changes in the connectivity of high-degree nodes and thus in the higherlevel network topology.

24

Figure 4. Rich-club organization for the ATD and TRYP+ conditions. The rich-club coefficient values Φ(k) (k is the degree of a node) curve for the conditionaveraged brain network is depicted as a solid line, and the normalized rich-club coefficient Φnorm(k) curve is depicted as a dotted line. Figure 4A shows significant (p<0.05; black star) and a trend for significant differences (p<0.1; red rhombus) in the normalized rich-club coefficient Φnorm(k) between both conditions, computed for each k (permutation test). The rich-club regime for both conditions in figure 4B is indicated by the dark gray area, which is defined by a significant difference of Φnorm(k) from the random null distribution (permutation test, p = 0.05).

25

4. Discussion In the present double-blind, randomized, crossover study, the implemented ATD procedure resulted in a 90% decrease of plasma TRP and in a 93% decrease in the ratio of plasma TRP to other LNAA’s (TRP/∑LNAA) relative to baseline. The latter has been assumed to be a more accurate indicator of brain 5-HT synthesis rate than absolute plasma levels of TRP (Oldendorf and Szabo, 1976; Scholes et al., 2007). This reduction in TRP/∑LNAA ratio is comparable to previous studies (Schmitt et al., 2000; Scholes et al., 2007), demonstrating the efficacy of the implemented ATD procedure in the present study. The putative reduction in brain 5-HT synthesis was associated with lower FC between the serotonergic NCS and the frontopolar cortex as well as the right cerebellum in contrast to TRYP+ condition. Interestingly, this significant change in FC after ATD was only detected for the ROI inclusive of the NCS and not of the DRN, indicating the specificity of the present finding. The coherence analysis (Figure S2) using time series from the DRN and NCS ROIs and voxels from brainstem/cerebellum further supported this interpretation. We could demonstrate a significantly positive association between changes in the TRP/∑LNAA ratio (compared to baseline) and FC between NCS and FPC in both conditions, i.e. in the ATD and TRYP+, additionally pointing towards a specific association between changes in this particular connectivity and changes in TRP concentration and thus in 5-HT synthesis. In both conditions, the observed relationship indicated lower 5-HT availability to be associated with lower NCS-FPC connectivity. Serotonergic fibers running towards the thalamus sprout fibers at the level of the hypothalamus that project into the amygdala, hippocampus and various cortical regions, including the lateral prefrontal cortex, FPC, ACC, VMPFC, posterior cingulate cortex (PCC) as well as cerebellar regions (Azmitia and Segal, 1978; Kerr and Bishop, 1991). The frontopolar cortex, containing the cytoarchitectural area 10 (Ongur et al., 2003), as part of the DMN has been consistently demonstrated to be involved in processes associated with mindwandering, such as self-referential processing or evaluation of self-generated information (Mansouri et al., 2017; Mason et al., 2007). Furthermore, Mansouri et al. (2017) additionally pointed in their review toward a specific role of FPC in managing competing goals by evaluating the significance of current and alternative goals. Thus, one possible explanation for the observed specific FC change to FPC might be the prominent role of the FPC as part of the DMN in the resting state condition. Interestingly, a 26

significantly greater RSFC from the NCS to a similar FPC cluster was recently observed in patients with BPD (Wagner et al., 2018). This abnormal connectivity negatively correlated with degree of self-rated aggression in patients, which has been associated with the abnormal functioning of the 5-HT system (Comai et al., 2016) and altered interactions between prefrontal cortex and amygdala (Passamonti et al., 2012). Thus, present results may deepen our understanding of specific psychiatric symptoms, such as aggression associated with a central nervous serotonergic dysfunction and abnormal neural connectivity. However, we could not replicate findings from previous rs-fMRI studies using ROI inclusive of the upper raphe nuclei. For instance, Salomon et al. (2011) and Weinstein et al. (2015) showed reduced RSFC between DRN, the thalamus and the rACC. Recent study by Anand et al. (2018) demonstrated lower DRN connectivity with the lateral prefrontal and mid-cingulate cortices. One explanation for this inconsistency might be that these studies investigated acutely depressed or remitted patients. We also observed decreased NCS-FC to the right cerebellum, which has been shown to be part of the executive-control network (ECN) (Bär et al., 2016). A further key finding was the observation of a significantly greater FC from the right amygdala to the VMPFC after ATD comparing to TRYP+. This relatively greater FC was significantly correlated with the mood ratings immediately before ATD-fMRI session, indicating greater FC between the amygdala and VMPFC associated with higher mood ratings after ATD. However, investigating the consequences of ATD on mood and arousal ratings, we observed in both conditions a significant reduction of mood after ATD as well as after TRYP+, but no significant interaction effect or significant changes in arousal levels. On the one hand, the lack of specific mood-lowering effects after ATD in healthy subjects is in agreement with previous studies (Deza-Araujo et al., 2019; Ruhe et al., 2007) suggesting a more complex relationship between 5-HT concentrations and mood. On the other hand, the non-significant differences in both conditions in arousal levels, which are thought to have an impact on 5-HT synthesis rate (Young, 2013), make the contribution of potential confounding factors on present results negligible. Confounding effects of physiological arousal were additionally properly controlled by applying state-of-the art physiological-noise correction methods. With respect to FC from the VMPFC, again an increase in FC after ATD was observed to mainly posterior brain regions of the DMN, e.g. precuneus. 27

Previous studies, which manipulated 5-HT availability by means of SSRI administration, consistently demonstrated a reduced fMRI signal during affective paradigms in the amygdala, frontal and cingulate cortices (Murphy et al., 2009; Raab et al., 2016). Interestingly, Harmer et al. (2006) reported during affective face processing not only reduced activation in the amygdala but also in the frontopolar cortex (BA 10) following citalopram (SSRI) treatment. Furthermore, studies using resting state fMRI and SSRI administration reported reduced connectivity between the amygdala and the VMPFC in the SSRI group (McCabe and Mishor, 2011) without significant changes in mood ratings. Klaassens et al. (2015) demonstrated in healthy subjects mostly widespread reductions in functional connectivity in multiple resting state networks after exposure to a single dose of the SSRI sertraline. In agreement with these studies, Deza-Araujo et al. (2019) showed very recently that acute TRP loading reduced significantly DMN FC in the medial prefrontal cortex as well as FC between DMN and putamen, thalamus and occipital regions. In contrast, neuroimaging studies suggest an ATD-mediated amygdala and fronto-cingulate activation increase as well as increased coupling in these circuitries. For instance, (Evers et al., 2010) reported in their review a consistent finding in previous neuroimaging studies of increased amygdala activation during processing of emotional faces as a consequence of TRP depletion. Moreover, Robinson et al. (2013) reported ATD-induced increased ACC/DMPFC amygdala functional connectivity during processing of fearful faces. In sum, previous findings suggest that an increase in 5-HT due to SSRI administration is associated with diminished amygdala and frontal activation as well as decreased resting state connectivity in several functional networks, whereas reduced availability of 5-HT elevates amygdala activation and intensified amygdala-frontal coupling. Thus, the present result of an increased amygdala-VMPFC coupling after ATD is in accordance with this notion. Cahir et al. (2007) observed in rats reduced 5-HT1A autoreceptor binding in raphe nuclei after ATD and no changes in cortico-limbic binding sites. This finding suggests a reduced serotonergic feedback modulation in the raphe nuclei (Cahir et al., 2007), potentially providing a mechanism for increased fronto-limbic connectivity as observed in the present ATD study. Furthermore, as reviewed by Carhart-Harris and Nutt (2017), the serotonergic system has high diversity, which might relate to its capacity for flexibly and adaptably responding to changing environmental conditions and thus also influencing brain connectivity. An alternative interpretation might be that increased 5-HT neurotransmission enhances the responsiveness of 28

specific brain regions, which may result in lower functional connectivity (in terms of higher segregation), but greater response to specific affective/cognitive tasks. Anatomically, the VMPFC lies within a constellation of regions with strong reciprocal connections, e.g., amygdala, anterior insula or hypothalamus (Beckmann et al., 2009). Broadly, these regions are neural substrates for generating emotional states and processing and integrating the representation of motivational information with visceral somatic responses (Critchley et al., 2011). A modulating influence of the VMPFC function on amygdala activation was often demonstrated (Roy et al., 2012). Thus, the significant correlation between the amygdala-VMPFC-FC and the SAM mood rating seamlessly fits into this notion. Our results, therefore, bridge the gap between neurotransmitter changes observed in the serotonergic brainstem region and its relation to psychological states closely associated with the integrity of fronto-limbic regions. Results from the whole brain analysis using NBS suggest additional widespread changes in the FC strength in a network component, which consisted of multiple regions from the four identified brain networks. This finding of FC changes, in terms of ATD-associated increases, between regions of several well-known neural networks is not surprising given the multiple cortical, striatal and limbic serotonergic projection from the raphe nuclei (Jacobs and Azmitia, 1992; O'Hearn and Molliver, 1984). It is also in agreement with the previous study of Klaassens et al. (2015), who showed decreased FC in multiple resting state networks after acute SSRI administration. Previous rs-fMRI studies mainly focused on the ATD-related impact on the DMN connectivity and used the independent component analysis (ICA) (Deza-Araujo et al., 2019; Helmbold et al., 2016). Helmbold et al. (2016) observed in females after ATD enhanced FC with the DMN in the left inferior and middle frontal gyri as well as in the right cerebellum. However, in contrast to Helmbold et al. (2016) and to the present study, Deza-Araujo et al. (2019) found that only TRP supplementation, but not TRP depletion produced significant changes in DMN connectivity. TRP supplementation was associated with decreased DMN connectivity with visual cortices and several brain regions like putamen, subcallosal cortex, thalamus, and frontal cortex, which were also observed in the present NBS analysis, but in terms of increased FC after ADT compared to the TRYP+ condition. In contrast to the present study, the investigation of FC differences was restricted to the spatial ICA-based map representing the DMN. Furthermore, the investigated sample consisted of female and male 29

subjects, which may also lead to gender-specific interaction effects with respect to serotonergic neurotransmission in the brain (Nishizawa et al., 1997). Finally, a further new finding of the present study was that the manipulation of 5-HT synthesis has some impact on the rich-club organization, which takes a central position in the brain's network topology. It describes the phenomenon that nodes with high degrees tend to interconnect with themselves and it gives important information about higher-level network topology with respect to the integration of information among different neural subsystems (van den Heuvel and Sporns, 2011b). Previous studies indicated abnormally reduced rich-club organization in patients with schizophrenia (van den Heuvel et al., 2013a) and in patients with bipolar disorder (Wang et al., 2019). Very recently, we demonstrated abnormal rich-club organization in MDD patients with suicidal behavior (Wagner et al., accepted), which is associated with altered serotonergic neurotransmission and increased impulsive and aggressive behavior. Here, we observed a reduced rich-club organization after nutritional 5HT manipulation in healthy subjects, which may have an effect on the efficient integration of information flow in a brain network (Misic et al., 2014; van den Heuvel and Sporns, 2011b). This, in turn, may contribute to the vulnerability for psychiatric disorders such as MDD or suicidal behavior. Finally, Jasinska et al. (2012) proposed a model for how genetically associated variability in 5-HT reuptake might affect the stress response in highlighting the altered between the amygdala, the VMPFC and the DRN. Thus, it will be interesting to stratify subjects based on genetic variability of the 5-HT system and to investigate the differential impact of ATD on this circuitry and associated affective changes in future studies. Several limitations should be noted. Only women were investigated due to previous studies showing a higher behavioural response to ATD (Ellenbogen et al., 1996) and marked differences in the 5-HT synthesis rate in women compared with male subjects (Nishizawa et al., 1997). Furthermore, we did not assess the hormonal status, menstrual phase and sex hormones in the present study as previously performed by Helmbold et al. (2016). Since there is some evidence from former studies for the modulating influence of sex hormones on serotonergic transmission (Barth et al., 2015) and on RSFC (Lisofsky et al., 2016), this might be a limitation of the present study. However, by using a double-blind, randomized, crossover design in the present study and including non-menopausal females, we consider shortterm systematic effects of hormonal status on present results as rather negligible. It is more likely that the putative variability in female participants with respect to cycling period and 30

contraceptive use, may have increased the FC variability and thus reducing the statistical significance of the observed underlying effects of ATD. ATD condition was compared to the TRYP+ condition, which may be different from the baseline condition. As stated above, Deza-Araujo et al. (2019) demonstrated very recently that acute TRP supplementation led to significant FC changes within DMN compared to baseline condition. Thus, the observed significant differences in RSFC between ATD and TRYP+ conditions may be due to the high-dose supplementation of TRP in the present study, corresponding on average to 165mg/kg body weight. The AA beverage in the study of DezaAraujo et al. (2019) contained 70 mg/kg body weight tryptophan and produced significant changes in the FC relative to a balanced drink. Furthermore, the nuclei within the 5-HT nuclei are small and potentially more susceptible to signal distortion and artifacts arising from local tissue interfaces and physiological noise. By using advanced brainstem normalization methods and properly correcting the BOLD signal for respiratory and cardiac noise we addressed this issue. The improved normalization of the brainstem/cerebellum using the SUIT toolbox was originally validated for the cerebellum (Diedrichsen, 2006), but not for the brainstem and midbrain. However, by creating a highresolution atlas template of both cerebellum and brainstem, the normalization of brainstem and midbrain structures is assumed to be improved as well. Furthermore, present voxel resolution of 2.5×2.5×2.5mm³ may introduce a partial volume effect. Thus, future studies with increased spatial resolution are required to confirm our findings.

5. Conclusion Our study demonstrated significant changes in central nervous TRP availability associated with significant differences in FC of 5-HT synthesizing raphe nuclei to the frontopolar cortex. The functional coupling between these regions becomes weaker with reduced TRP availability. We also observed an increased coupling of the amygdala and VMPFC as well as between cortical and subcortical regions of several brain networks as a result of decreased TRP availability. Finally, we showed for the first time a significant impact of ATD on the rich-club organization in terms of decreased rich-club coefficients compared to tryptophan supplementation. Thus, we could demonstrate that ATD and thus putative decrease in 5-HT synthesis significantly impacts FC of a ROI inclusive of the NCS as well as of multiple subcortical/cortical regions centrally involved in affective, but also in cognitive processes. In 31

addition, a significant impact of ATD was observed on the so-called rich-club organization, thus affecting the efficient integration of information flow from several brain networks.

32

References Amat, J., Baratta, M.V., Paul, E., Bland, S.T., Watkins, L.R., Maier, S.F., 2005. Medial prefrontal cortex determines how stressor controllability affects behavior and dorsal raphe nucleus. Nature Neuroscience 8, 365-371. Anand, A., Jones, S.E., Lowe, M., Karne, H., Koirala, P., 2018. Resting State Functional Connectivity of Dorsal Raphe Nucleus and Ventral Tegmental Area in Medication-Free Young Adults With Major Depression. Frontiers in Psychiatry 9, 765. Ardis, T.C., Cahir, M., Elliott, J.J., Bell, R., Reynolds, G.P., Cooper, S.J., 2009. Effect of acute tryptophan depletion on noradrenaline and dopamine in the rat brain. J Psychopharmacol 23, 51-55. Ashburner, J., 2007. A fast diffeomorphic image registration algorithm. Neuroimage 38, 95-113. Azmitia, E.C., Segal, M., 1978. An autoradiographic analysis of the differential ascending projections of the dorsal and median raphe nuclei in the rat. J Comp Neurol 179, 641-667. Bach, D.R., Friston, K.J., Dolan, R.J., 2010. Analytic measures for quantification of arousal from spontaneous skin conductance fluctuations. International Journal of Psychophysiology 76, 52-55. Bär, K.J., de la Cruz, F., Schumann, A., Koehler, S., Sauer, H., Critchley, H., Wagner, G., 2016. Functional connectivity and network analysis of midbrain and brainstem nuclei. Neuroimage 134, 5363. Barth, C., Villringer, A., Sacher, J., 2015. Sex hormones affect neurotransmitters and shape the adult female brain during hormonal transition periods. Front Neurosci 9. Baudry, A., Pietri, M., Launay, J.M., Kellermann, O., Schneider, B., 2019. Multifaceted Regulations of the Serotonin Transporter: Impact on Antidepressant Response. Front Neurosci 13. Beckmann, M., Johansen-Berg, H., Rushworth, M.F., 2009. Connectivity-based parcellation of human cingulate cortex and its relation to functional specialization. J Neurosci 29, 1175-1190. Beg, M.F., Miller, M.I., Trouve, A., Younes, L., 2005. Computing large deformation metric mappings via geodesic flows of diffeomorphisms. International Journal of Computer Vision 61, 139-157. Beliveau, V., Ganz, M., Feng, L., Ozenne, B., Hojgaard, L., Fisher, P.M., Svarer, C., Greve, D.N., Knudsen, G.M., 2017. A High-Resolution In Vivo Atlas of the Human Brain's Serotonin System. Journal of Neuroscience 37, 120-128. Bradley, M.M., Lang, P.J., 1994. Measuring emotion: the Self-Assessment Manikin and the Semantic Differential. J Behav Ther Exp Psychiatry 25, 49-59. Briggs, G.G., Nebes, R.D., 1975. Patterns of hand preference in a student population. Cortex 11, 230238. Buckner, R.L., Andrews-Hanna, J.R., Schacter, D.L., 2008. The brain's default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci 1124, 1-38. Cahir, M., Ardis, T., Reynolds, G.P., Cooper, S.J., 2007. Acute and chronic tryptophan depletion differentially regulate central 5-HT1A and 5-HT 2A receptor binding in the rat. Psychopharmacology (Berl) 190, 497-506. Canli, T., Lesch, K.P., 2007. Long story short: the serotonin transporter in emotion regulation and social cognition. Nat Neurosci 10, 1103-1109. Cole, M.W., Bassett, D.S., Power, J.D., Braver, T.S., Petersen, S.E., 2014. Intrinsic and task-evoked network architectures of the human brain. Neuron 83, 238-251. Collin, G., Kahn, R.S., de Reus, M.A., Cahn, W., van den Heuvel, M.P., 2014. Impaired Rich Club Connectivity in Unaffected Siblings of Schizophrenia Patients. Schizophrenia Bulletin 40, 438-448. Comai, S., Bertazzo, A., Vachon, J., Daigle, M., Toupin, J., Cote, G., Turecki, G., Gobbi, G., 2016. Tryptophan via serotonin/kynurenine pathways abnormalities in a large cohort of aggressive inmates: markers for aggression. Prog Neuropsychopharmacol Biol Psychiatry 70, 8-16. Critchley, H.D., Nagai, Y., Gray, M.A., Mathias, C.J., 2011. Dissecting axes of autonomic control in humans: Insights from neuroimaging. Auton Neurosci 161, 34-42. Crossley, N.A., Mechelli, A., Scott, J., Carletti, F., Fox, P.T., McGuire, P., Bullmore, E.T., 2014. The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain 137, 2382-2395. 33

David, S.P., Murthy, N.V., Rabiner, E.A., Munafo, M.R., Johnstone, E.C., Jacob, R., Walton, R.T., Grasby, P.M., 2005. A functional genetic variation of the serotonin (5-HT) transporter affects 5-HT1A receptor binding in humans. J Neurosci 25, 2586-2590. Deza-Araujo, Y.I., Neukam, P.T., Marxen, M., Muller, D.K., Henle, T., Smolka, M.N., 2019. Acute tryptophan loading decreases functional connectivity between the default mode network and emotion-related brain regions. Hum Brain Mapp 40, 1844-1855. Diedrichsen, J., 2006. A spatially unbiased atlas template of the human cerebellum. Neuroimage 33, 127-138. Eklund, A., Nichols, T.E., Knutsson, H., 2016. Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proceedings of the National Academy of Sciences of the United States of America 113, E4929-E4929. Ellenbogen, M.A., Young, S.N., Dean, P., Palmour, R.M., Benkelfat, C., 1996. Mood response to acute tryptophan depletion in healthy volunteers: Sex differences and temporal stability. Neuropsychopharmacology 15, 465-474. Ernst, C., Mechawar, N., Turecki, G., 2009. Suicide neurobiology. Prog Neurobiol 89, 315-333. Evers, E.A., Tillie, D.E., van der Veen, F.M., Lieben, C.K., Jolles, J., Deutz, N.E., Schmitt, J.A., 2005. Effects of a novel method of acute tryptophan depletion on plasma tryptophan and cognitive performance in healthy volunteers. Psychopharmacology (Berl) 178, 92-99. Evers, E.A.T., Sambeth, A., Ramaekers, J.G., Riedel, W.J., van der Veen, F.M., 2010. The Effects of Acute Tryptophan Depletion on Brain Activation During Cognition and Emotional Processing in Healthy Volunteers. Curr Pharm Des 16, 1998-2011. Fornito, A., Zalesky, A., Bullmore, E., 2016a. Fundamentals of Brain Network Analysis. Academic Press, London. Fornito, A., Zalesky, A., Bullmore, E.T., 2016b. Fundamentals of brain network analysis. 476. Glover, G.H., Li, T.Q., Ress, D., 2000. Image-based method for retrospective correction of physiological motion effects in fMRI: RETROICOR. Magn Reson Med 44, 162-167. Grayson, D.S., Ray, S., Carpenter, S., Iyer, S., Dias, T.G., Stevens, C., Nigg, J.T., Fair, D.A., 2014. Structural and functional rich club organization of the brain in children and adults. Plos One 9, e88297. Hahn, A., Haeusler, D., Kraus, C., Hoflich, A.S., Kranz, G.S., Baldinger, P., Savli, M., Mitterhauser, M., Wadsak, W., Karanikas, G., Kasper, S., Lanzenberger, R., 2014. Attenuated Serotonin Transporter Association Between Dorsal Raphe and Ventral Striatum in Major Depression. Hum Brain Mapp 35, 3857-3866. Harmer, C.J., 2008. Serotonin and emotional processing: Does it help explain antidepressant drug action? Neuropharmacology 55, 1023-1028. Harmer, C.J., Mackay, C.E., Reid, C.B., Cowen, P.J., Goodwin, G.M., 2006. Antidepressant drug treatment modifies the neural processing of nonconscious threat cues. Biol Psychiatry 59, 816-820. Harriger, L., van den Heuvel, M.P., Sporns, O., 2012. Rich Club Organization of Macaque Cerebral Cortex and Its Role in Network Communication. PLoS One 7, e46497. Helmbold, K., Zvyagintsev, M., Dahmen, B., Biskup, C.S., Bubenzer-Busch, S., Gaber, T.J., Klasen, M., Eisert, A., Konrad, K., Habel, U., Herpertz-Dahlmann, B., Zepf, F.D., 2016. Serotonergic modulation of resting state default mode network connectivity in healthy women. Amino Acids 48, 1109-1120. Husain, M., Roiser, J.P., 2018. Neuroscience of apathy and anhedonia: a transdiagnostic approach. Nat Rev Neurosci 19, 470-484. Jacobs, B.L., Azmitia, E.C., 1992. Structure and function of the brain serotonin system. Physiol Rev 72, 165-229. Jasinska, A.J., Lowry, C.A., Burmeister, M., 2012. Serotonin transporter gene, stress and raphe-raphe interactions: a molecular mechanism of depression. Trends Neurosci 35, 395-402. Jones, T.B., Bandettini, P.A., Kenworthy, L., Case, L.K., Milleville, S.C., Martin, A., Birn, R.M., 2010. Sources of group differences in functional connectivity: An investigation applied to autism spectrum disorder. Neuroimage 49, 401-414. 34

Kautzky, A., James, G.M., Philippe, C., Baldinger-Melich, P., Kraus, C., Kranz, G.S., Vanicek, T., Gryglewski, G., Wadsak, W., Mitterhauser, M., Rujescu, D., Kasper, S., Lanzenberger, R., 2017. The influence of the rs6295 gene polymorphism on serotonin-1A receptor distribution investigated with PET in patients with major depression applying machine learning. Transl Psychiatry 7, e1150. Kerr, C.W.H., Bishop, G.A., 1991. Topographical Organization in the Origin of Serotoninergic Projections to Different Regions of the Cat Cerebellar Cortex. Journal of Comparative Neurology 304, 502-515. Kim, M., Kwak, S., Yoon, Y.B., Kwak, Y.B., Kim, T., Cho, K.I.K., Lee, T.Y., Kwon, J.S., 2019. Functional connectivity of the raphe nucleus as a predictor of the response to selective serotonin reuptake inhibitors in obsessive-compulsive disorder. Neuropsychopharmacology. Klaassens, B.L., van Gorsel, H.C., Khalili-Mahani, N., van der Grond, J., Wyman, B.T., Whitcher, B., Rombouts, S.A.R.B., van Gervend, J.A., 2015. Single-dose serotonergic stimulation shows widespread effects on functional brain connectivity. Neuroimage 122, 440-450. Köhler, S., Bar, K.J., Wagner, G., 2016. Differential involvement of brainstem noradrenergic and midbrain dopaminergic nuclei in cognitive control. Hum Brain Mapp 37, 2305-2318. Köhler, S., Schumann, A., de la Cruz, F., Wagner, G., Bar, K.J., 2018. Towards response success prediction: An integrative approach using high-resolution fMRI and autonomic indices. Neuropsychologia 119, 182-190. Kunisato, Y., Okamoto, Y., Okada, G., Aoyama, S., Demoto, Y., Munakata, A., Nomura, M., Onoda, K., Yamawaki, S., 2011. Modulation of default-mode network activity by acute tryptophan depletion is associated with mood change: a resting state functional magnetic resonance imaging study. Neurosci Res 69, 129-134. Lang, P.J., 1980. Behavioral treatment and bio-behavioral assessment. . In: Sidowski, J.B., Johnson, J.H., Williams, T.A. (Eds.), Technology in mental health care delivery systems. Ablex., Norwood, NJ, pp. 119–137. Lim, C.L., Rennie, C., Barry, R.J., Bahramali, H., Lazzaro, I., Manor, B., Gordon, E., 1997. Decomposing skin conductance into tonic and phasic components. International Journal of Psychophysiology 25, 97-109. Lisofsky, N., Riediger, M., Gallinat, J., Lindenberger, U., Kuhn, S., 2016. Hormonal contraceptive use is associated with neural and affective changes in healthy young women. Neuroimage 134, 597-606. Lowry, C.A., Hale, M.W., Evans, A.K., Heerkens, J., Staub, D.R., Gasser, P.J., Shekhar, A., 2008. Serotonergic systems, anxiety, and affective disorder: focus on the dorsomedial part of the dorsal raphe nucleus. Ann N Y Acad Sci 1148, 86-94. Lowry, C.A., Johnson, P.L., Hay-Schmidt, A., Mikkelsen, J., Shekhar, A., 2005. Modulation of anxiety circuits by serotonergic systems. Stress 8, 233-246. Mai, N., Zhong, X., Chen, B., Peng, Q., Wu, Z., Zhang, W., Ouyang, C., Ning, Y., 2017. Weight Rich-Club Analysis in the White Matter Network of Late-Life Depression with Memory Deficits. Frontiers in Aging Neuroscience 9, 279. Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage 19, 1233-1239. Mansouri, F.A., Koechlin, E., Rosa, M.G.P., Buckley, M.J., 2017. Managing competing goals - a key role for the frontopolar cortex. Nature Reviews Neuroscience 18, 645-657. Mason, M.F., Norton, M.I., Van Horn, J.D., Wegner, D.M., Grafton, S.T., Macrae, C.N., 2007. Wandering minds: the default network and stimulus-independent thought. Science 315, 393-395. McCabe, C., Mishor, Z., 2011. Antidepressant medications reduce subcortical-cortical resting-state functional connectivity in healthy volunteers. Neuroimage 57, 1317-1323. Menon, V., 2015. Salience Network. In: Toga, A.W. (Ed.), Brain Mapping: An Encyclopedic Reference. Academic Press: Elsevier., Oxford, pp. 597-611. Misic, B., Sporns, O., McIntosh, A.R., 2014. Communication efficiency and congestion of signal traffic in large-scale brain networks. PLoS Comput Biol 10, e1003427. 35

Murphy, K., Birn, R.M., Bandettini, P.A., 2013. Resting-state fMRI confounds and cleanup. Neuroimage 80, 349-359. Murphy, S.E., Norbury, R., O'Sullivan, U., Cowen, P.J., Harmer, C.J., 2009. Effect of a single dose of citalopram on amygdala response to emotional faces. Br J Psychiatry 194, 535-540. Naidich, T.P., Duvernoy, H.M., Delman, B.N., Sorensen, A.G., Kollias, S.S., Haacke, E.M., 2009a. Duvernoy's Atlas of the Human Brain Stem and Cerebellum. Springer, Wien. Naidich, T.P., Duvernoy, H.M., Delman, B.N., Sorensen, A.G., Kollias, S.S., Haacke, E.M., 2009b. Duvornoy's Atlas of the Human Brain Stem and Cerebellum. Springer, Wien. Newman, M.E., 2006. Modularity and community structure in networks. Proc Natl Acad Sci U S A 103, 8577-8582. Nieuwenhuys, R., 1985. Chemoarchitecture of the brain. Springer, Berlin. Nishizawa, S., Benkelfat, C., Young, S.N., Leyton, M., Mzengeza, S., DeMontigny, C., Blier, P., Diksic, M., 1997. Differences between males and females in rates of serotonin synthesis in human brain. Proceedings of the National Academy of Sciences of the United States of America 94, 5308-5313. O'Hearn, E., Molliver, M.E., 1984. Organization of raphe-cortical projections in rat: a quantitative retrograde study. Brain Res Bull 13, 709-726. Oldendorf, W.H., Szabo, J., 1976. Amino acid assignment to one of three blood-brain barrier amino acid carriers. Am J Physiol 230, 94-98. Ongur, D., Ferry, A.T., Price, J.L., 2003. Architectonic subdivision of the human orbital and medial prefrontal cortex. J Comp Neurol 460, 425-449. Pardridge, W.M., 1979. The role of blood-brain barrier transport of tryptophan and other neutral amino acids in the regulation of substrate-limited pathways of brain amino acid metabolism. J Neural Transm Suppl, 43-54. Passamonti, L., Crockett, M.J., Apergis-Schoute, A.M., Clark, L., Rowe, J.B., Calder, A.J., Robbins, T.W., 2012. Effects of acute tryptophan depletion on prefrontal-amygdala connectivity while viewing facial signals of aggression. Biol Psychiatry 71, 36-43. Paxinos, G.T., Huang, X.-F., 1995. Atlas of the Human Brainstem. Academic Press Inc., San Diego. Pezawas, L., Meyer-Lindenberg, A., Drabant, E.M., Verchinski, B.A., Munoz, K.E., Kolachana, B.S., Egan, M.F., Mattay, V.S., Hariri, A.R., Weinberger, D.R., 2005. 5-HTTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nat Neurosci 8, 828-834. Power, J.D., Cohen, A.L., Nelson, S.M., Wig, G.S., Barnes, K.A., Church, J.A., Vogel, A.C., Laumann, T.O., Miezin, F.M., Schlaggar, B.L., Petersen, S.E., 2011. Functional Network Organization of the Human Brain. Neuron 72, 665-678. Raab, K., Kirsch, P., Mier, D., 2016. Understanding the impact of 5-HTTLPR, antidepressants, and acute tryptophan depletion on brain activation during facial emotion processing: A review of the imaging literature. Neurosci Biobehav Rev 71, 176-197. Ray, S., Miller, M., Karalunas, S., Robertson, C., Grayson, D.S., Cary, R.P., Hawkey, E., Painter, J.G., Kriz, D., Fombonne, E., Nigg, J.T., Fair, D.A., 2014. Structural and functional connectivity of the human brain in autism spectrum disorders and attention-deficit/hyperactivity disorder: A rich cluborganization study. Hum Brain Mapp 35, 6032-6048. Robinson, O.J., Overstreet, C., Allen, P.S., Letkiewicz, A., Vytal, K., Pine, D.S., Grillon, C., 2013. The role of serotonin in the neurocircuitry of negative affective bias: Serotonergic modulation of the dorsal medial prefrontal-amygdala 'aversive amplification' circuit. Neuroimage 78, 217-223. Roy, M., Shohamy, D., Wager, T.D., 2012. Ventromedial prefrontal-subcortical systems and the generation of affective meaning. Trends Cogn Sci 16, 147-156. Rubinov, M., Sporns, O., 2010. Complex network measures of brain connectivity: uses and interpretations. Neuroimage 52, 1059-1069. Ruhe, H.G., Mason, N.S., Schene, A.H., 2007. Mood is indirectly related to serotonin, norepinephrine and dopamine levels in humans: A meta-analysis of monoamine depletion studies. Mol Psychiatry 12, 331-359. 36

Salomon, R.M., Cowan, R.L., Rogers, B.P., Dietrich, M.S., Bauernfeind, A.L., Kessler, R.M., Gore, J.C., 2011. Time series fMRI measures detect changes in pontine raphe following acute tryptophan depletion. Psychiatry Res 191, 112-121. Schmitt, J.A.J., Jorissen, B.L., Sobczak, S., van Boxtel, M.P.J., Hogervorst, E., Deutz, N.E.P., Riedel, W.J., 2000. Tryptophan depletion impairs memory consolidation but improves focussed attention in healthy young volunteers. J Psychopharmacol 14, 21-29. Scholes, K.E., Harrison, B.J., O'Neill, B.V., Leung, S., Croft, R.J., Pipingas, A., Phan, K.L., Nathan, P.J., 2007. Acute serotonin and dopamine depletion improves attentional control: findings from the stroop task. Neuropsychopharmacology 32, 1600-1610. Schumann, A., Köhler, S., de la Cruz, F., Gullmar, D., Reichenbach, J.R., Wagner, G., Bär, K.J., 2018. The Use of Physiological Signals in Brainstem/Midbrain fMRI. Front Neurosci 12, 718. Sengupta, A., Bocchio, M., Bannerman, D.M., Sharp, T., Capogna, M., 2017. Control of Amygdala Circuits by 5-HT Neurons via 5-HT and Glutamate Cotransmission. J Neurosci 37, 1785-1796. Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., Dunbar, G.C., 1998. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry 59 Suppl 20, 22-33;quiz 34-57. van den Heuvel, M.P., Kahn, R.S., Goni, J., Sporns, O., 2012. High-cost, high-capacity backbone for global brain communication. Proc Natl Acad Sci U S A 109, 11372-11377. van den Heuvel, M.P., Sporns, O., 2011a. Rich-Club Organization of the Human Connectome. Journal of Neuroscience 31. van den Heuvel, M.P., Sporns, O., 2011b. Rich-club organization of the human connectome. J Neurosci 31, 15775-15786. van den Heuvel, M.P., Sporns, O., Collin, G., Scheewe, T., Mandl, R.C., Cahn, W., Goni, J., Hulshoff Pol, H.E., Kahn, R.S., 2013a. Abnormal rich club organization and functional brain dynamics in schizophrenia. JAMA Psychiatry 70, 783-792. van den Heuvel, M.P., Sporns, O., Collin, G., Scheewe, T., Mandl, R.C.W., Cahn, W., Goñi, J., Hulshoff Pol, H.E., Kahn, R.S., 2013b. Abnormal Rich Club Organization and Functional Brain Dynamics in Schizophrenia. JAMA Psychiatry 70, 783. Wagner, G., de la Cruz, F., Kohler, S., Bar, K.J., 2017. Treatment Associated Changes of Functional Connectivity of Midbrain/Brainstem Nuclei in Major Depressive Disorder. Sci Rep 7, 8675. Wagner, G., de la Cruz, F., Köhler, S., Pereira, F., Richard-Devantoy, S., Turecki, G., Bär, K.-J., Jollant, F., accepted. Connectomics-Based Functional Network Alterations in both Depressed Patients with Suicidal Behavior and Healthy Relatives of Suicide Victims. Scientific reports. Wagner, G., Koschke, M., Leuf, T., Schlosser, R., Bar, K.J., 2009. Reduced heat pain thresholds after sad-mood induction are associated with changes in thalamic activity. Neuropsychologia 47, 980-987. Wagner, G., Krause-Utz, A., de la Cruz, F., Schumann, A., Schmahl, C., Bar, K.J., 2018. Resting-state functional connectivity of neurotransmitter producing sites in female patients with borderline personality disorder. Prog Neuropsychopharmacol Biol Psychiatry 83, 118-126. Wang, Y., Deng, F., Jia, Y., Wang, J., Zhong, S., Huang, H., Chen, L., Chen, G., Hu, H., Huang, L., Huang, R., 2019. Disrupted rich club organization and structural brain connectome in unmedicated bipolar disorder. Psychol Med 49, 510-518. Weinstein, J.J., Rogers, B.P., Taylor, W.D., Boyd, B.D., Cowan, R.L., Shelton, K.M., Salomon, R.M., 2015. Effects of acute tryptophan depletion on raphe functional connectivity in depression. Psychiatry Res 234, 164-171. Williams, W.A., Shoaf, S.E., Hommer, D., Rawlings, R., Linnoila, M., 1999. Effects of acute tryptophan depletion on plasma and cerebrospinal fluid tryptophan and 5-hydroxyindoleacetic acid in normal volunteers. J Neurochem 72, 1641-1647. Willner, P., Scheel-Kruger, J., Belzung, C., 2013. The neurobiology of depression and antidepressant action. Neurosci Biobehav Rev 37, 2331-2371. Young, S.N., 2013. Acute tryptophan depletion in humans: a review of theoretical, practical and ethical aspects. J Psychiatry Neurosci 38, 294-305. 37

Young, S.N., Smith, S.E., Pihl, R.O., Ervin, F.R., 1985. Tryptophan depletion causes a rapid lowering of mood in normal males. Psychopharmacology (Berl) 87, 173-177. Zalesky, A., Fornito, A., Bullmore, E.T., 2010. Network-based statistic: identifying differences in brain networks. Neuroimage 53, 1197-1207.

Supplementary material Table S1: Composition (in grams) of the collagen-based amino acid mixture (ATD condition). 100 g of this mixture was dissolved in 300 g water. In the TRYP+ condition, this mixture was supplemented by 10.3 g TRP. Alanine Arginine Aspartic acid Glutamic acid Glycine Histidine Hydroxylysine Hydroxyproline Isoleucine Leucine Lysine Methionine Phenylalanine Proline Serine Threonine Tyrosine Valine

8.4 7.7 4.5 10 23.3 0.9 1.5 12.3 1.2 2.6 3.3 0.9 1.6 13.7 3.4 1.9 0.6 2.2 38

Table S2: Significant group differences between ATD and TRYP+ condition in Network-Based Statistics (NBS) analyses at the whole-brain level.

ATD vs. TRYP+ conditions, p = 0.04, FWER Node Cerebellum Precuneus Precuneus Inferior parietal cortex Angular gyrus Posterior cingulate cortex Inferior parietal cortex Angular gyrus Middle temporal gyrus Middle temporal gyrus Fusiform gyrus Middle temporal gyrus Middle temporal gyrus Middle temporal gyrus Inferior temporal gyrus Middle temporal gyrus Middle temporal gyrus Premotor cortex Middle temporal gyrus Inferior frontal gyrus Inferior frontal gyrus Dorsomedial prefrontal cortex Inferior frontal gyrus Inferior frontal gyrus Premotor cortex Inferior frontal gyrus Orbitofrontal cortex Anterior cingulate cortex Middle frontal gyrus Dorsomedial prefrontal cortex Dorsomedial prefrontal cortex Occipital cortex

MNI coordinates x y z -18 -76 -24 -6 -72 42 12 -66 42 38 -64 40 52 -60 36 -6 -54 28 -42 -54 44 46 -50 28 -68 -42 -6 -50 -42 0 -34 -38 -16 64 -30 -8 -58 -26 -14 -60 -22 -18 64 -12 -20 -56 -12 -10 -52 2 -28 -44 2 46 52 6 -30 -30 18 -18 48 22 10 -16 28 54 -46 32 -14 52 32 0 24 34 48 50 36 -12 8 48 -16 -8 50 -2 -34 54 4 6 54 16 6 64 22 -42 -60 -8 39

Network DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN DMN ECN

Superior parietal lobe Middle temporal gyrus Thalamus Putamen Putamen Inferior frontal gyrus Putamen Inferior frontal gyrus Middle frontal gyrus Inferior frontal gyrus Anterior insula Middle frontal gyrus Anterior cingulate cortex Anterior cingulate cortex Orbitofrontal cortex Inferior frontal gyrus Orbitofrontal cortex Medial frontal gyrus Superior temporal gyrus Superior temporal gyrus Primary motor cortex Primary motor cortex Superior temporal gyrus Inferior parietal cortex Midcingulate cortex Superior temporal gyrus Premotor cortex Midcingulate cortex Premotor cortex Premotor cortex Primary somatosensory cortex Posterior insula Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex Occipital cortex

-28 58 12 30 -16 -42 -22 -46 48 34 36 -42 -2 0 24 32 -22 44 -56 66 14 -38 -50 60 -14 58 28 0 -52 20 56 36 -24 -12 18 -24 -26 -14 -40 24 20 36 26 28

-58 -52 -18 -14 4 6 8 10 10 16 22 24 26 30 32 38 40 50 -40 -34 -32 -28 -26 -18 -18 -16 -16 -14 -10 -8 -6 0 -98 -94 -92 -92 -90 -90 -88 -88 -86 -82 -80 -76 40

48 -14 8 2 8 32 -4 24 32 -8 2 30 44 28 -18 -12 -20 -2 14 20 74 70 6 28 40 8 70 46 24 64 14 -4 -12 -12 -14 18 4 32 -6 24 -2 2 -16 26

ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN ECN SN SN SN SN SN SN SN SN SN SN SN SN SN SN VN VN VN VN VN VN VN VN VN VN VN VN

Occipital cortex Occipital cortex Occipital cortex Occipital cortex Parahippocampal gyrus

16 -42 -16 20 26

-76 -74 -72 -66 -40

30 0 -8 2 -12

VN VN VN VN VN

Abbreviations: Family-wise error rate, FWER; DMN, default mode network; ECN, executive-control network; SN, salience-network; VN, visual network.

Figure S1: Position of the seed regions used for the RSFC analyses in the present study: VMPFC (red), the left and right amygdala (blue), DRN (green) and NCS (yellow).

41

Figure S2: The averaged coherence maps between the averaged time series from the ROIs inclusive of the DRN and NCS and all other voxels in the brainstem/cerebellum are illustrated. A coherence map was generated for each subject between the average time series extracted from the DRN and NCS ROIs and time series from all voxels in the midbrain/brainstem and cerebellum. Individual coherence maps were averaged and thresholded at a coherence value of Cij = 0.4 in order to generate a group-averaged map. The DRN coherence map is depicted in fuchsia and the NCS coherence map in blue. The black square represents the edges of the raphe ROIs. Both color scales depict the degree of coherence.

42

KJB: idea of the study, organizing the study, writing the manuscript, data calculation SK: organizing the study, performing the fMRI experiment, FDLC: calculating data, writing the manuscript, figure design AS: calculating data, writing the manuscript, discussing results FDZ: writing and revising the manuscript, discussing results GW: writing the manuscript, data calculation, literature search for ATD, figure design