- Email: [email protected]
Daytime melatonin levels in saliva are associated with inflammatory markers and anxiety disorders
Journal Pre-proof Daytime Melatonin Levels in Saliva are Associated with Inflammatory Markers and Anxiety Disorders Isak SundbergInvestigation Data curation Formal analysis Writing - original draft Writing - review and editing), Annica J. RasmussonMethodology Data curation Formal analysis Writing - original draft Writing - review and editing), Mia RamklintInvestigation Writing review and editing), David JustData curation Formal analysis Writing - review and editing), Lisa EkseliusInvestigation Writing - review and editing), Janet L. Cunningham (Conceptualization) (Methodology) (Funding acquisition) (Project administration) (Investigation) (Data curation) (Formal analysis) (Writing - original draft) (Writing - review and editing)
PII:
S0306-4530(19)31255-7
DOI:
https://doi.org/10.1016/j.psyneuen.2019.104514
Reference:
PNEC 104514
To appear in:
Psychoneuroendocrinology
Received Date:
7 May 2019
Revised Date:
3 September 2019
Accepted Date:
12 November 2019
Please cite this article as: Sundberg I, Rasmusson AJ, Ramklint M, Just D, Ekselius L, Cunningham JL, Daytime Melatonin Levels in Saliva are Associated with Inflammatory Markers and Anxiety Disorders, Psychoneuroendocrinology (2019), doi: https://doi.org/10.1016/j.psyneuen.2019.104514
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.
Sundberg et al
Daytime Melatonin Levels in Saliva are Associated with Inflammatory Markers and Anxiety Disorders Isak Sundberg a, Annica J. Rasmusson b, Mia Ramklint a, David Just a, Lisa Ekselius a, Janet L. Cunningham a
Department of Neuroscience, Psychiatry, Uppsala University, Uppsala, Sweden
b
Department of Medical Sciences, Clinical Pharmacology and Osteoporosis, Uppsala
ro of
a
University, Uppsala, Sweden
-p
Corresponding author: Janet L. Cunningham MD, PhD, Associate Professor
Department of Neuroscience, Psychiatry Uppsala University Hospital, Entrance 10, Floor 3B,
lP
re
751 85, Uppsala, Sweden, E-mail: [email protected]
Highlights
na
We analyzed daytime saliva melatonin in relation to levels of 72 inflammatory markers Melatonin levels at 11:00 were correlated with CD 5 Melatonin after lunch was correlated with MCP-1, MIP-1α and VEGF-A Melatonin after lunch, VEGF-A and MIP-1α was associated with anxiety disorders
ur
ABSTRACT
Jo
Background
The bidirectional interaction between melatonin and the immune system has largely gone unexplored in a clinical context and especially in a psychiatric population. This study explored the association between melatonin during the day and inflammatory cytokines in young adult patients seeking psychiatric care.
Sundberg et al
Methods Samples and data were collected from 108 young adults (mean age 21, SD=2) at an outpatient clinic for affective disorders. Daytime saliva melatonin levels were analyzed with enzyme-linked immunosorbent assay (ELISA) in relation to normalized serum expression levels of 72 inflammatory markers in a proximity extension assay (PEA). In a post hoc analysis the markers associated with melatonin were tested in a generalized linear model to see whether there is a relationship to anxiety disorder or depression.
ro of
Results
After Bonferroni correction for multiple testing, melatonin levels at 11:00 were positively correlated with CD5 (p=4.2e-4). Melatonin levels after lunch were correlated with CCL2/MCP-1 (p=4.2e-4),
-p
CCL3/MPI-1α (p=6.5e-4) and VEGF-A (p=5.3e-6). In the generalized linear model, positive
associations were found for the presence of any anxiety disorder with melatonin after lunch (p=0.046),
re
VEGF-A (p=0.001) and CCL3/MPI-1α (p=0.001).
lP
Conclusion
Daytime saliva levels of melatonin were related to several inflammatory markers in young adults with
na
psychiatric disorders. This observation likely reflects the bidirectional relationship between melatonin production and the immune system. These findings may have relevance for the understanding of
ur
psychiatric disorders and other conditions associated with low-grade inflammation.
Jo
1. Introduction
The circadian release of melatonin from the pineal gland helps to regulate the rhythm of bodily functions hierarchically synchronized by the suprachiasmatic nucleus (Buhr and Takahashi, 2013). Melatonin is additionally produced by all mitochondria-containing cells in a non-circadian manner and exerts a wide diversity of autocrine and paracrine effects (Acuna-Castroviejo et al., 2014). Arguably,
Sundberg et al
melatonin’s most important function is as an antioxidant protecting cells from oxidative damage (Manchester et al., 2015).
A complex and fascinating interaction between melatonin and the immune system that extends well beyond antioxidative effects is summarized in a recent review (Hardeland, 2018). Melatonin is most known for its anti‐ inflammatory properties; however, depending on various conditions, including cell types, level and duration of inflammation, melatonin can have proinflammatory effects (Hardeland, 2018). Proinflammatory actions of melatonin enhance the resistance against pathogens but may be
ro of
detrimental in autoimmune diseases (such as rheumatoid arthritis) (Galbo and Kall, 2016). The anti‐ inflammatory actions of melatonin are of particular interest, however, because they are observed in high‐ grade inflammation (e.g., sepsis, ischemia/reperfusion and brain injury) and also in chronic
-p
low‐ grade inflammation during aging and in neurodegenerative diseases, with suggested relevance also in mood disorders (Hardeland, 2018). The role of melatonin in mood disorders has been
lP
mood disorders remains to be fully elucidated.
re
researched (Anderson, 2017) but the connection between daytime melatonin and inflammation in
The immune system is under circadian control, which is reflected in many ways. For instance, immune
na
cell response to antigen exposure is modulated by central circadian mediators and differs depending on the time of day (Guerrero-Vargas et al., 2014; Rahman et al., 2015). There is an endogenous circadian
ur
regulation of inflammatory cytokines such as C-X-C motif chemokine ligand 8 (CXCL8)/interleukin 8 (IL-8) and C-C motif chemokine ligand 2 (CCL2)/monocyte chemoattractant protein-1 (MCP-1) in the
Jo
presence of bacterial lipopolysaccharide (LPS) in humans (Rahman et al., 2015). Intravenous administration of melatonin suppresses markers of inflammation and oxidative damage in a human daytime endotoxemia model but these effects are dependent on the time of administration (Alamili et al., 2014). Suprachiasmatic nucleus lesion leads to a 10-fold increase in LPS-induced inflammatory cytokine release in animals (Guerrero-Vargas et al., 2014). LPS may also be involved in chronic low‐ grade inflammation. Lipopolysaccharide binding protein (LBP) has been suggested as a marker for low grade inflammation related to LPS (Citronberg et al., 2016; Gonzalez-Quintela et al., 2013). When
Sundberg et al
LPS levels are high, LBP production is induced in the liver and has a long half-life in serum. As a recent example of this, LBP was found to be higher in couples in hostile marital interactions and a history a mood disorder may contribute to this association (Kiecolt-Glaser et al., 2018).
Peripheral proinflammatory cytokines, such as tumor necrosis factor alpha (TNF)-α, shut down pineal melatonin production, contributing to the decreased pineal circadian melatonin synthesis across many medical conditions (da Silveira Cruz-Machado et al., 2012). This regulation can be mediated by TNFα directly in pinealocytes or indirectly via toll-like receptor-4 activation in microglia around the pineal
ro of
gland, which drives local TNF-α production and release (da Silveira Cruz-Machado et al., 2010;
Markus et al., 2017). Reduction in the nocturnal melatonin surge is necessary to allow leukocyte
migration to sites of injury in that circulating melatonin regulates endothelial tight junctions restricting
-p
leukocyte migration to sites of injury (Man et al., 2016).
re
During immunological challenge, immune cells are influenced by both autocrine and paracrine melatonin. While immune activation reduces melatonin production by pinealocytes, the production of
lP
melatonin is upgraded peripherally given that activated macrophages/microglia synthesize melatonin in a nuclear factor kappaB (NF-κB)-dependent manner at the site of the inflammatory response
na
(Markus et al., 2017). In cultured macrophages, NF-κB signaling pathways induce the transcription of the arylalkylamine N-acetyltransferase (AA-NAT) gene needed to catalyze the N-acetylation of
ur
serotonin into melatonin (Muxel et al., 2012). In turn, melatonin enhances phagocytosis through autocrine action and Luzindole, a competitive antagonist of melatonin receptors, decreases
Jo
macrophage phagocytic activity (Muxel et al., 2012). This bidirectional relationship between melatonin and the immune system has been termed the immune-pineal axis (Markus et al., 2017).
Few studies have investigated potential associations between endogenous daytime levels of melatonin and cytokine levels in clinical settings. Morning serum levels of melatonin were lower and TNF-α higher when patients with newly diagnosed multiple sclerosis were compared with healthy controls, but not significantly correlated with each other (Farhadi et al., 2014). In patients with chronic renal
Sundberg et al
failure, melatonin levels were lower both day and night and inversely correlated with TNF-α (Pinto et al., 2016). In contrast, patients with polymyalgia rheumatica had higher levels of melatonin compared with controls, with the relative difference being most pronounced during the day (Galbo and Kall, 2016). These studies indicate an interaction between melatonin and serum inflammatory markers but the studies differ in sampling times, which may explain some of the discrepancies.
Disruption in endocrine and inflammatory response systems are pathophysiological mechanisms implicated in the etiology of depression, anxiety and sleep disorders (Haapakoski et al., 2016;
ro of
Leighton et al., 2018; Miller and Raison, 2016; Young et al., 2014), with possible relevance for
prognosis and treatment response (Jha and Trivedi, 2018; Kappelmann et al., 2016). Previously, we have identified a negative correlation between bedtime melatonin and depressive symptoms in the
-p
present cohort of young adult patients seeking psychiatric care (Sundberg et al., 2016). The
relationship between melatonin and inflammation in psychiatric patients, however, is largely
re
unexplored.
lP
This study aimed to explore the interaction between daytime levels of melatonin in saliva taken at six time points and levels of a large panel of inflammatory cytokines in young adults with psychiatric
na
morbidity. Based on data showing a correlation between depression and evening levels of melatonin in the same cohort of patients (Sundberg et al., 2016), we hypothesized that melatonin levels in the
ur
evening would correlate with levels of inflammatory markers. A secondary hypothesis was that melatonin levels taken mid-day would correlate with inflammatory markers. In addition, we
Jo
investigated the potential relationship between inflammatory markers during the day, melatonin levels and the presence of the clinical variables: depression or anxiety disorder.
Sundberg et al
2. Materials and methods 2.1. Ethical approval and patient consent The study was approved by the Regional Ethics Committee in Uppsala (Dnr. 2012/81, Dnr. 2012/81/1 and Dnr. 2013/219) and all participants gave written consent to participate in the study.
2.2. Data cohort The material and data used in this observational, cross-sectional cohort study, which has been
ro of
previously described (Sundberg et al., 2016), originate from the Uppsala Psychiatric Patient Samples (UPP) cohort. All data were collected from patients seeking care between 2012 and 2014 at an outpatient clinic for young adults of the Section for Affective Disorders at Uppsala University
-p
Hospital, Uppsala, Sweden. This outpatient clinic provides care for patients aged 18-25 years, with mainly mood and anxiety disorders but also neuropsychiatric and personality disorders. Of 722
re
consecutive patients, 300 agreed to participate in the UPP study. Of these, 125 completed saliva sampling. Six patients were excluded because they did not fulfill criteria for any DSM-IV Axis 1
lP
diagnosis. For 11 patients, blood sampling was missing or incomplete. Thus, the final sample comprised 108 patients with a DSM-IV Axis 1 disorder. Saliva and blood samples for all 108 patients
na
were included in this study. No patients were taking antidepressant medication targeting melatonin
ur
receptors MT1/MT2 (agomelatine or ramelteon).
2.3. Clinical data
Jo
Demographic characteristics of the study participants are shown in Table 1. The procedure for assessment of psychiatric diagnoses, symptoms and physical health examination upon entering the UPP has been described in detail elsewhere (Cunningham et al., 2017). Diagnoses are based on the structured interviews used: the Swedish version of the M.I.N.I. International Neuropsychiatric Interview (M.I.N.I. 6.0) or the Structural Clinical Interview for DSM IV axis I disorders (SCID-I). All interviews were performed by trained doctors or psychologists. Physical examination included the measurement of body mass index (BMI) using the formula BMI=kg/m2.
6
Sundberg et al
2.4. Melatonin analysis Melatonin in saliva was analyzed with competitive enzyme-linked immunosorbent assay (ELISA). This procedure has been described previously (Sundberg et al., 2016). In short, participants were instructed to collect saliva samples at six time points during the waking hours of one day: when waking up, 30 minutes after waking up but before breakfast, at 11.00 hours, 30 minutes after lunch, at 22.00 hours and just before going to bed. Time points were chosen to capture the rise of melatonin in the evening before sleep and variation in daytime melatonin and specifically after meals. Saliva was
ro of
collected using inert polymer cylindrical swabs (10 mm x 30 mm) that were subsequently placed in a storage tube consisting of a large outer tube with a small insert and snap cap (swabs and tubes from
Salimetrics Europe Ltd., Suffolk, UK) and kept in the refrigerator until delivery to the lab within 48
-p
hours. Participants were instructed not to eat or drink 30 minutes before sampling. The participants
documented collection times. To ensure compliance the research assistant verified collection times and
re
sampling method with the patient upon receipt (samples not collected as instructed were excluded). Of maximally 648 (6x108) hormone measurements, 31 (4.8%) were missing because of mistakes in saliva
lP
sampling or insufficient saliva volume. Total assay variability was <11%. Upon receipt, tubes were centrifuged and stored at -20°C until analysis. Salivary melatonin was measured with competitive
na
ELISA (Direct Salivary Melatonin Elisa EK-DSM, Bühlmann Laboratories AG, Schönenbuch. Switzerland). Analyses were performed at the routine laboratory of the Department of Clinical
ur
Chemistry at Uppsala University Hospital, Uppsala, Sweden. The laboratory is certified by a Swedish
Jo
government authority (Swedac).
2.5. Proseek analysis of relative plasma levels of inflammatory proteins Blood samples, collected during office hours, were obtained from non-fasting patients and kept in 80°C at Uppsala Biobank. The relative levels of 91 inflammatory proteins were analyzed in 108 plasma samples from the study cohort using the Proseek Multiplex Inflammation panel (Olink Bioscience, Sweden), a proximity extension-based assay (PEA). The 91 proteins belonged to a preset
7
Sundberg et al inflammatory panel, initially including 92 proteins; however, one protein (Beta-nerve growth factor) was excluded by the manufacturer (see Supplementary Table 2). In short, PEA technology allows amplification and quantification of antibody-coupled, proximity-dependent DNA templates, reflecting relative protein levels in the sample (Assarsson et al., 2014). With detection sensitivity down to fg/mL, the assay can be used to compare relative protein values between groups but is not an absolute quantification. All samples were analyzed with the same batch of reagents at the Clinical Biomarker Facility at the SciLife Lab in Uppsala. DNA amplification and quantification were carried out using
ro of
the BioMark™ HD real-time PCR platform (Fluidigm, South San Francisco, CA, USA).
Plasma was isolated from blood samples and stored at -80°C within <60 minutes after isolation. After thawing on ice for the first time, samples were transferred to 96-well plates with 90 samples and 6
-p
controls (three inter-plate controls and three negative controls [buffer]) evenly distributed over the
plates. Incubation mixes (3μL) containing 92 pairs of unique DNA oligonucleotides each labeled to a
re
corresponding antibody were combined with plasma samples (1μL) and incubated at 4°C overnight. PEA enzymes and PCR reagents were added in a 96 μL extension mix, after which samples were
lP
incubated at room temperature for 5 minutes. The plate was then transferred to the thermal cycler to perform an initial DNA extension for 20 minutes at 50°C and subsequent DNA amplification in 17
na
cycles. Following the manufacturer’s instructions, a 96.96 Dynamic Array IFC (Fluidigm, South San Francisco, CA, USA) was prepared and primed while the sample mixture (2.8 μL) was mixed with
ur
7.2 μL detection mix in a new plate. The right side of the primed 96.96 Dynamic Array IFC was loaded with 5 μL of this mixture and the left side with the unique primer pairs for each protein. Using
Jo
the instructions for Proseek, the protein expression program was then run in the Fluidigm Biomark reader.
Each sample was spiked with one detection control, one extension control and two incubation controls (green fluorescent protein and phycoerythrin). Data normalization was performed using GenEx software in the Olink Wizard. The quantification cycle (Cq) values generated in the real-time PCR were used to calculate the normalized protein expression (NPX) in three steps. First, technical
8
Sundberg et al variations were corrected by subtracting the extension control from the Cq value of every sample (∆Cq sample = Cq sample – Cq extension control). The inter-plate control was then subtracted to compensate for potential variations between runs (∆∆Cq = ∆Cq sample – ∆Cq inter-plate control). The NPX was finally calculated by normalization against a calculation correction factor (NPX = correction factor – ∆∆Cq sample). The NPX data were presented on a Log2 scale.
The samples were analyzed together, randomly distributed across eight plates. The samples were normalized for plate differences using a median normalization, applied separately for each protein. For
ro of
this normalization, we considered only the 72 proteins with detectable values in at least 50% of all the plasma samples (see Supplementary Table 1). Across all 92 assays, the mean intra-assay and inter-
-p
assay variations observed were 7% and 18%, respectively.
2.6. Analysis of plasma protein levels with multiplex electrochemiluminescence analysis
re
As a validation step, two of the plasma proteins found to be significantly associated with melatonin
lP
levels after correction for multiple analyses were quantified using an electrochemiluminescence sandwich immunoassay from the Meso Scale Discovery (Rockville, MD, USA) multiplex platform.
na
In brief, plasma samples were applied to a primary antibody-coated 96-well plate. After subsequent dilution, incubation and washing, the captured proteins were incubated with secondary antibodies
ur
labelled with an electrochemiluminescence tag. The plate was then inserted into a Sector Imager 2400 (MSD, Gaithersburg, MD), where a voltage applied to the plate electrodes causes the captured labels
Jo
to emit light. The MSD Discovery Workbench Software was used to convert the intensity of emitted light to protein concentrations using interpolation from log calibrator curves corresponding to the protein of interest. The validation data were only available for two of the significant markers, namely CCL2/MCP-1 and vascular endothelial growth factor A (VEGF-A) (K15047D-1 and K15050D-1, MSD, Gaithersburg, MD). According to the manufacturer, the inter-assay variation for highest/lowest
9
Sundberg et al standards of CCL2/MCP-1 and VEGF-A has a span of (8.9-3.0) and (2.0-5.3%), respectively. Internal standards showed an inter-assay variation of 10%. Further we have measured total LBP (lipopolysaccharide binding protein) as a proxy for LPS in all 108 patients using the same multiplex electrochemiluminescence analysis (K151IYC-1, MSD, Gaithersburg, MD). Our internal standards showed an inter-assay variation of 10% or less for all three assays.
ro of
2.7. Statistical analyses Statistical analyses were performed using the software SPSS (version 23.0) and R version 3.4.2.
Results from the PEA are presented as NPX values, a relative measurement on the log2 scale. The
samples were distributed over eight plates in the PEA experiment. The plate effect was removed using
-p
a median normalization based on all plasma samples analyzed together. The plate normalized NPX
values were used in all analyses involving PEA data. Limit of detection (LOD) was strictly defined as
re
three standard deviations above background. Lower limit of quantification and upper limit of
lP
quantification were defined as lowest and highest concentrations measured with %CV<30 and relative error (RE) <30%. Values below the LOD were replaced by the LOD. The melatonin values were not
na
normally distributed and therefore log10-tranformed before all analyses.
ur
The association between protein level (NPX) and log10 (melatonin) was assessed using linear regression with the protein value (NPX) as the dependent variable and melatonin and relevant
Jo
covariates as independent variables. The association adjusted for covariates was assessed using a likelihood ratio test. The covariates were sex, BMI, use of antidepressant medication and use of oral contraceptives. Of the 91 proteins on the PEA array, we excluded those with more than 50% of the values below LOD before further analyses. There was no missing data for reasons other than low signal. Thus, 72 markers remained after this exclusion criterion was applied. Three proteins (IL-17A, IL-17C and FGF-5) with more than 20% of the values below LOD were discretized as above LOD (1) or below LOD (0) and the analyses were performed using logistic regression instead of linear
10
Sundberg et al regression. All tests were corrected using the Bonferroni method with the significance level set at error α=0.05 divided by the number of proteins used to analyze the data (72); thus, a p-value below 6.9e-4 was considered significant for the main analysis.
The relationship between melatonin values at each time point and inflammatory markers was analyzed in the whole sample. In a post hoc analysis the markers that had shown a correlation to saliva melatonin were studied in relationship to the presence of depressive episode or anxiety disorder in a
ro of
generalized linear model, controlling for sex, antidepressant medication, anticonception and BMI.
Levels of inflammatory markers VEGF-A and CCL2/MCP-1 that were measured with PEA were validated using multiplex electrochemiluminescence analysis. The correlation between plate
re
assessed using Spearman's rank-order correlation.
-p
normalized NPX values from PEA and log2-transformed values (pg/ml) from the multiplex assay were
3. Results
lP
3.1. Associations between saliva melatonin levels and inflammatory markers. The associations between protein level and melatonin are summarized in heat maps using Euclidean
na
distance and complete linkage hierarchical clustering: positive associations are shown in red and negative associations in blue (Figure 1.). Analyses were corrected for covariates sex, BMI, use of
ur
antidepressant medication and use of oral contraceptives. Supplementary Table 2 presents a list of
Jo
associations and p-values for all proteins.
There were no significant correlations between bedtime melatonin and inflammatory markers in the Proseek Multiplex Inflammation panel after correction for multiple testing (Figure 1 and Supplementary Table 2). Thus, the primary hypothesis that melatonin levels in the evening would correlate with levels of inflammatory markers was not confirmed.
11
Sundberg et al For daytime melatonin values and inflammatory markers, postprandial melatonin was positively correlated with VEGF-A (p= 5.8e-6), CCL2/MCP-1 (p=4.2e-4) and CCL3/monocyte inflammatory protein-1α (MIP-1α) (p=6.5e-4). Melatonin at 11:00 was positively correlated with cluster of differentiation 5 (CD5) (p=4.2e-4). For these markers, all values were within the level of quantification and well above LOD. Adding bedtime melatonin to the respective models did not influence the associations. BMI was positively correlated to VEGF-A (p = 0.011), CCL2/MCP-1 (p = 0.001) and CCL3/MIP-1α (p = 0.028) (Spearman). The regression analysis, however, showed that the
ro of
associations between melatonin and these markers of inflammation independent of BMI.
To study if the daytime/nighttime melatonin ratio, in addition to the daytime melatonin, is associated to protein level, melatonin at bedtime was added to the two models predicting protein level for 11:00
-p
and post prandial melatonin. A model improvement would suggest that melatonin at bedtime in
combination with the respective daytime levels would better predict protein level than each timepoint
re
alone. This was, however, not the case and we can conclude that the daytime/nighttime ratio is not
lP
associated to protein level.
3.2. Post hoc analysis of the association between melatonin levels, inflammatory markers and
na
psychiatric disorders
In the generalized linear model, melatonin at 11:00 was not associated with anxiety disorder (p=0.73)
ur
or depression (p=0.72). Melatonin after lunch was positively associated to the presence of any anxiety disorder (p=0.046) and this association remained when adding depression to the model (p=0.04).
Jo
Melatonin after lunch was not associated with depression (p=0.47). VEGF-A was associated with anxiety (p=0.001) but not with depression (p=0.6); the same findings occurred for CCL3/MIP-1α for anxiety (p=0.001) and depression (p=0.4). CCL2/MCP-1 was not linked to anxiety (p=0.18) but negatively associated with current depression (p=0.042). Finally, CD5 was neither associated with anxiety (p=0.18) nor with depression (p=0.4).
12
Sundberg et al 3.3. Validation of PEA data Plate-normalized NPX values from PEA and log2-transformed meso scale values (pg/ml) were correlated with each other: VEGF-A (r=0.56, p=6.72e-10), CCL2/MCP-1 (r=0.60, p=1.83e-11).
3.4. Post hoc analysis of association between melatonin levels and inflammatory markers in women. When looking at only women (n=94), melatonin at 11:00 am was correlated to LAP-TGFB1 (p=9.3e4), CD5 (p=1.5 e-5), CD6 (p=1e-5) and CXCL8/ IL-8 (p=7.4e-4). Daytime post-prandial values of
ro of
melatonin were correlated to IL-10RB (p=8.2e-4), CCL3/MIP1- α (p=2.6e-4) and VEGF-A (p=9e-6) (see Figure 2).
-p
3.5. Analysis of association between LBP, levels of melatonin and inflammatory markers.
Meso scale values of LBP (ng/ml) correlated with VEGF-A (r=0.397, p=7.9e-5) and CCL3/MIP-1α
re
(r=0.281, p<0.00397). No associations were found for LBP with BMI, sex, medication, the diagnosis of depressive episode or any anxiety disorder (Mann Whitney). For both VEGF-A and CCL3/MIP-1α,
lP
the associations with melatonin were independent from LBP and adding LBP to the models did not
Jo
ur
na
influence the strength of these associations (data not shown).
13
Sundberg et al
4. Discussion The main finding in this study is that daytime saliva levels of melatonin, but not bedtime melatonin levels as hypothesized, were related to inflammatory markers in blood in young adults with psychiatric morbidity. Specifically, after correction for multiple testing, daytime postprandial melatonin levels were associated with levels of VEGF-A, CCL2/MCP-1 and CCL3/MIP-1α and melatonin levels at 11:00 were associated with levels of CD5. Elevated postprandial levels of melatonin and cytokines (VEGF-A and CCL3/MIP-1α) were more
ro of
common in patients with anxiety disorders, whereas CCL2/MCP-1 was negatively associated with depression. The results demonstrate an association between melatonin and the immune
-p
system in a clinical context of psychiatric patients.
The associations between daytime melatonin levels and inflammatory markers and anxiety
re
disorders are novel findings but find support in previous work. The three markers associated with daytime postprandial melatonin levels in this study (VEGF-A, CCL2/MCP-1 and
lP
CCL3/MIP-1α) have been linked to depression and anxiety (Gaspersz et al., 2017; Leighton et al., 2018). Production capacity of several cytokines, including CCL2/MCP-1, was positively
na
associated with severity of depressive and, in particular, anxiety symptoms, even while taking lifestyle and health factors into account (Vogelzangs et al., 2016). Several studies report high
ur
peripheral VEGF-A levels in patients with depression compared with healthy controls (Carvalho et al., 2015; Sharma et al., 2016) and there is evidence for elevated VEGF-A as a
Jo
potential predictor of treatment response in major depression (Clark-Raymond et al., 2016). A majority of studies report higher concentrations of CCL2/MCP-1 in depressed patients compared with controls (Eyre et al., 2016; Kohler et al., 2017), although some studies have reported the opposite result (Lehto et al., 2010; Myung et al., 2016). A recent meta-analysis found that antidepressant treatment decreased CCL2/MCP-1 (Kohler et al., 2018). Higher levels of CCL3/MIP-1α are found in depressed patients compared with controls (Leighton et al., 2018; Merendino et al., 2004) and CCL3/MIP-1α has been suggested as a marker of
Sundberg et al oxidative stress in depression (Oglodek, 2018). The relationship between inflammation and depression is influenced by gender and generally more pronounced in females (Birur et al., 2017). In the subgroup analysis of female patients in this study, additional associations between daytime melatonin and inflammatory markers IL-8, CD5, CD6 and IL-10RB were identified.
These associations may be linked to physiological reactions to oxidative stress as both CCL2/MCP-1 and CCL2/MCP-1 are produced in response to oxidative stress (Deshmane et
ro of
al., 2009). Melatonin appears to counteract oxidative stress and reduce levels of CCL2/MCP1 in animal and cell models (Brazao et al., 2015; Wang et al., 2013). Furthermore, treatment with melatonin in rats subjected to strenuous exercise resulted in higher levels of VEGF-A,
-p
whereas muscular oxidative stress and inflammation were reduced (Borges Lda et al., 2015).
re
As far as we know, a link between the lymphocyte marker CD5 and melatonin has not previously been described. The association was strongest for melatonin at 11:00 but also
lP
present with post-prandial melatonin before correction for multiple testing. CD5 are receptors present on the surface of T cells, but also exist on a subset of B cells and as soluble receptors
na
circulating in serum (Calvo et al., 1999; Domingues et al., 2016; Sarrias et al., 2004; Sigal, 2012). The CD5 receptor is an important regulator of lymphocyte selection and immune
ur
tolerance (Raman, 2002). Increased expression of CD5 on either T cells or B cells seems to protect against autoimmunity (Dalloul, 2009) and experimental studies have shown better
Jo
survival in sepsis models when soluble CD5 has been injected into recipient rats (Sarrias et al., 2007; Vera et al., 2009). Elevated levels of soluble CD5 are linked to aging (Paltsev et al., 2016), sepsis (Aibar et al., 2015) and primary Sjögren's syndrome (Ramos-Casals et al., 2001). In one study, reduced CD5 surface expression on transitional B cells was associated with severe depression and normalized exclusively in clinical responders, possibly indicating
Sundberg et al a B cell-dependent process in the pathogenesis of depression (Ahmetspahic et al., 2018). Further investigation of this marker in the context of psychiatric disease is motivated.
The gut is a possible source of daytime melatonin. Locally, melatonin in the gut can be 400fold higher than levels released by the pineal gland, being highest after food intake (Bubenik, 2002). In a previous study we reported a relationship between post-prandial melatonin levels and gastrointestinal symptoms in the same cohort of patients (Soderquist et al., 2018). One possible underlying mechanism explaining the association between inflammatory markers and
ro of
melatonin in this study could be related to intestinal permeability (Martin-Subero et al.,
2016). In this study, daytime melatonin and LBP are not correlated with each other but both have independent correlations with VEGF-A and CCL3/MIP-1α. We could not confirm
-p
associations between LBP and BMI shown previously in patients with metabolic syndrome
(Sun et al., 2010). The young adults in this study are generally lean and somatically healthy
lP
interplay between these factors.
re
and longitudinal studies may provide interesting insight in understanding the trajectory and
Postprandial melatonin may also contribute to glycemic control. Several studies have shown
na
that individuals with poor glycemic control have higher serum levels of VEGF, CCL2/MCP1 and CCL3/MIP-1α (Chiarelli et al., 2002; Murphy et al., 2015; Zhang et al., 2018). In this
ur
study, we also noted a positive association between VEGF-A, CCL2/MCP-1 and CCL3/MIP1α and BMI: (BMI vs VEGF-A (rho 0.247, p=0.011), BMI vs CCL2/MCP-1 (rho 0.309,
Jo
p=0.001) and BMI vs CCL3/MIP-1α (rho 0.216, p=0.028). In this regard, it can be noted that melatonin has emerged as a hormone important for glucose regulation (Mayo et al., 2018). A potential relationship between daytime postprandial levels of melatonin and insulin resistance merits further exploration.
Strengths of this study include the measurement of selected cytokines using a separate method to validate the findings. Also, confirmation of inflammatory markers previously found related
Sundberg et al to psychiatric disorders provides support for the validity of the findings and against type 1 errors. The study approach maximizes the power of the dataset by examining the range of biological variation within the whole population. The relationships between diagnosis and the biological variation are explored post hoc analysis. Our data supports the hypothesis that distinct putative pathoaetiological drivers with respect to inflammation exist and that they may be more or less prevalent within different diagnostic categories. The data, however, also supports the rationale for analyzing the whole population: that underlying biological
ro of
processes seem to be both diverse within each DSM psychiatric disorder and transdiagnostic.
This study has several limitations. First, blood sampling for inflammatory markers was performed during office hours the day after saliva sampling, but not on a fixed time.
-p
Secondly, while the analyses in this study are adjusted for sex, BMI, antidepressant
medication and use of oral contraceptives, complete data were not available for other lifestyle
re
factors, such as exercise, alcohol use and smoking, all of which may influence levels of inflammatory markers (Vogelzangs et al., 2016). Thirdly, the limited amount of saliva did not
lP
allow for serial dilutions for concentrations above 50 ng/L. Fourthly, the small sample size may have limited the power of this study to identify the full extent of the relationship between
na
inflammatory markers and melatonin levels. Finally, the cross-sectional design of this study cannot provide information on the causal relationship between melatonin and inflammation
ur
over the course of disease.
Jo
In conclusion, this study demonstrated, in a psychiatric population, an association between daytime melatonin and several inflammatory markers that have previously been associated with psychiatric disorders. Evening levels of melatonin were not associated with inflammatory markers in this study. Furthermore, we saw a relationship between anxiety disorders, postprandial melatonin levels and VEGF-A and CCL3/MIP-1α. In contrast, CCL2/MCP-1 was negatively associated with depression but not anxiety disorders. These findings may have implications for understanding melatonin in the context of oxidative stress,
Sundberg et al metabolic dysregulation and inflammation and potentially have relevance in identifying biological mechanisms of psychiatric disease.
Author contributions JC conceived, designed and financed the study. MR, LE, JC contributed to acquisition of the patients and IS collected clinical data. AR and JC conducted the analysis of inflammatory markers and JC, IS and AR interpreted the data. DJ and AR conducted the analysis of LBP.
revision and approved the final version of the manuscript.
-p
Conflicts of interest
CrediT Author Statement
re
None Declarations of interest: none
ro of
IS, AR and JC drafted the first version of the manuscript. All authors contributed to critical
Writing – review & editing
lP
Isak Sundberg: Investigation, Data curation, Formal analysis, Writing – original draft,
na
Annica Rasmusson: Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing
ur
Mia Ramklint: Investigation, Writing – review & editing Lisa Ekselius: Investigation, Writing – review & editing
Jo
David Just: Data curation, Formal analysis, Writing – review & editing Janet Cunningham: Conceptualization, Methodology, Funding acquisition, Project administration, Investigation, Data curation, Formal analysis, Writing – original draft, Writing – review & editing
Sundberg et al Acknowledgements The authors wish to thank Ulla Nordén for her excellent clinical research assistance and Eva Freyhult and Hans Arinell for their exceptional statistical advice. Samples were managed in collaboration with Uppsala Biobank and methylation profiling was performed by the SNP&SEQ Technology Platform in Uppsala, which is part of the Science for Life Laboratory at Uppsala University. SciLife Laboratories are supported by the Swedish Research Council as a part of the national infrastructure. The work in this study was supported by Märta och Nicke Nasvells fund, the Apotekare Hedbergs Fund, the Erik, Karin and Gösta Selanders
ro of
Stiftelse, Fredrik and Ingrid Thurings Stiftelse, Stiftelsen Söderström-Königska sjukhemmet,
Jo
ur
na
lP
re
-p
the Swedish Society of Medicine and ALF Funds from Uppsala University Hospital.
Sundberg et al
References
Jo
ur
na
lP
re
-p
ro of
Acuna-Castroviejo, D., Escames, G., Venegas, C., Diaz-Casado, M.E., Lima-Cabello, E., Lopez, L.C., Rosales-Corral, S., Tan, D.X., Reiter, R.J., 2014. Extrapineal melatonin: sources, regulation, and potential functions. Cell Mol Life Sci 71, 29973025. Ahmetspahic, D., Schwarte, K., Ambree, O., Burger, C., Falcone, V., Seiler, K., Kooybaran, M.R., Grosse, L., Roos, F., Scheffer, J., Jorgens, S., Koelkebeck, K., Dannlowski, U., Arolt, V., Scheu, S., Alferink, J., 2018. Altered B Cell Homeostasis in Patients with Major Depressive Disorder and Normalization of CD5 Surface Expression on Regulatory B Cells in Treatment Responders. J Neuroimmune Pharmacol 13, 90-99. Aibar, J., Martinez-Florensa, M., Castro, P., Carrasco, E., Escoda-Ferran, C., Fernandez, S., Butjosa, M., Hernandez, C., Rinaudo, M., Lozano, F., Nicolas, J.M., 2015. Pattern of soluble CD5 and CD6 lymphocyte receptors in critically ill patients with septic syndromes. J Crit Care 30, 914-919. Alamili, M., Bendtzen, K., Lykkesfeldt, J., Rosenberg, J., Gogenur, I., 2014. Melatonin suppresses markers of inflammation and oxidative damage in a human daytime endotoxemia model. J Crit Care 29, 184 e189-184 e113. Anderson, G., 2017. Linking the biological underpinnings of depression: Role of mitochondria interactions with melatonin, inflammation, sirtuins, tryptophan catabolites, DNA repair and oxidative and nitrosative stress, with consequences for classification and cognition. Prog Neuropsychopharmacol Biol Psychiatry. Assarsson, E., Lundberg, M., Holmquist, G., Bjorkesten, J., Thorsen, S.B., Ekman, D., Eriksson, A., Rennel Dickens, E., Ohlsson, S., Edfeldt, G., Andersson, A.C., Lindstedt, P., Stenvang, J., Gullberg, M., Fredriksson, S., 2014. Homogenous 96-plex PEA immunoassay exhibiting high sensitivity, specificity, and excellent scalability. PLoS One 9, e95192. Birur, B., Amrock, E.M., Shelton, R.C., Li, L., 2017. Sex Differences in the Peripheral Immune System in Patients with Depression. Front Psychiatry 8, 108. Borges Lda, S., Dermargos, A., da Silva Junior, E.P., Weimann, E., Lambertucci, R.H., Hatanaka, E., 2015. Melatonin decreases muscular oxidative stress and inflammation induced by strenuous exercise and stimulates growth factor synthesis. J Pineal Res 58, 166-172. Brazao, V., Colato, R.P., Santello, F.H., Filipin Mdel, V., Toldo, M.P., do Vale, G.T., Tirapelli, C.R., do Prado Junior, J.C., 2015. Interleukin-17, oxidative stress, and inflammation: role of melatonin during Trypanosoma cruzi infection. J Pineal Res 59, 488-496. Bubenik, G.A., 2002. Gastrointestinal melatonin: localization, function, and clinical relevance. Dig Dis Sci 47, 2336-2348. Buhr, E.D., Takahashi, J.S., 2013. Molecular components of the Mammalian circadian clock. Handb Exp Pharmacol, 3-27. Calvo, J., Places, L., Espinosa, G., Padilla, O., Vila, J.M., Villamor, N., Ingelmo, M., Gallart, T., Vives, J., Font, J., Lozano, F., 1999. Identification of a natural soluble form of human CD5. Tissue Antigens 54, 128-137. Carvalho, A.F., Kohler, C.A., McIntyre, R.S., Knochel, C., Brunoni, A.R., Thase, M.E., Quevedo, J., Fernandes, B.S., Berk, M., 2015. Peripheral vascular endothelial growth factor as a novel depression biomarker: A meta-analysis. Psychoneuroendocrinology 62, 18-26.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Chiarelli, F., Cipollone, F., Mohn, A., Marini, M., Iezzi, A., Fazia, M., Tumini, S., De Cesare, D., Pomilio, M., Pierdomenico, S.D., Di Gioacchino, M., Cuccurullo, F., Mezzetti, A., 2002. Circulating monocyte chemoattractant protein-1 and early development of nephropathy in type 1 diabetes. Diabetes Care 25, 1829-1834. Citronberg, J.S., Wilkens, L.R., Lim, U., Hullar, M.A., White, E., Newcomb, P.A., Le Marchand, L., Lampe, J.W., 2016. Reliability of plasma lipopolysaccharide-binding protein (LBP) from repeated measures in healthy adults. Cancer Causes Control 27, 1163-1166. Clark-Raymond, A., Meresh, E., Hoppensteadt, D., Fareed, J., Sinacore, J., Garlenski, B., Halaris, A., 2016. Vascular endothelial growth factor: Potential predictor of treatment response in major depression. World J Biol Psychiatry, 1-11. Cunningham, J.L., Zanzi, M., Willebrand, M., Ekselius, L., Ramklint, M., 2017. No regrets: Young adult patients in psychiatry report positive reactions to biobank participation. BMC Psychiatry 17, 21. da Silveira Cruz-Machado, S., Carvalho-Sousa, C.E., Tamura, E.K., Pinato, L., Cecon, E., Fernandes, P.A., de Avellar, M.C., Ferreira, Z.S., Markus, R.P., 2010. TLR4 and CD14 receptors expressed in rat pineal gland trigger NFKB pathway. J Pineal Res 49, 183-192. da Silveira Cruz-Machado, S., Pinato, L., Tamura, E.K., Carvalho-Sousa, C.E., Markus, R.P., 2012. Glia-pinealocyte network: the paracrine modulation of melatonin synthesis by tumor necrosis factor (TNF). PLoS One 7, e40142. Dalloul, A., 2009. CD5: a safeguard against autoimmunity and a shield for cancer cells. Autoimmun Rev 8, 349-353. Deshmane, S.L., Kremlev, S., Amini, S., Sawaya, B.E., 2009. Monocyte chemoattractant protein-1 (MCP-1): an overview. J Interferon Cytokine Res 29, 313326. Domingues, R.G., Lago-Baldaia, I., Pereira-Castro, I., Fachini, J.M., Oliveira, L., Drpic, D., Lopes, N., Henriques, T., Neilson, J.R., Carmo, A.M., Moreira, A., 2016. CD5 expression is regulated during human T-cell activation by alternative polyadenylation, PTBP1, and miR-204. Eur J Immunol 46, 1490-1503. Eyre, H.A., Air, T., Pradhan, A., Johnston, J., Lavretsky, H., Stuart, M.J., Baune, B.T., 2016. A meta-analysis of chemokines in major depression. Prog Neuropsychopharmacol Biol Psychiatry 68, 1-8. Farhadi, N., Oryan, S., Nabiuni, M., 2014. Serum levels of melatonin and cytokines in multiple sclerosis. Biomed J 37, 90-92. Galbo, H., Kall, L., 2016. Circadian variations in clinical symptoms and concentrations of inflammatory cytokines, melatonin, and cortisol in polymyalgia rheumatica before and during prednisolone treatment: a controlled, observational, clinical experimental study. Arthritis Res Ther 18, 174. Gaspersz, R., Lamers, F., Wittenberg, G., Beekman, A.T.F., van Hemert, A.M., Schoevers, R.A., Penninx, B., 2017. The role of anxious distress in immune dysregulation in patients with major depressive disorder. Transl Psychiatry 7, 1268. Gonzalez-Quintela, A., Alonso, M., Campos, J., Vizcaino, L., Loidi, L., Gude, F., 2013. Determinants of serum concentrations of lipopolysaccharide-binding protein (LBP) in the adult population: the role of obesity. PLoS One 8, e54600. Guerrero-Vargas, N.N., Salgado-Delgado, R., Basualdo Mdel, C., Garcia, J., GuzmanRuiz, M., Carrero, J.C., Escobar, C., Buijs, R.M., 2014. Reciprocal interaction between the suprachiasmatic nucleus and the immune system tunes down the inflammatory response to lipopolysaccharide. J Neuroimmunol 273, 22-30.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Haapakoski, R., Ebmeier, K.P., Alenius, H., Kivimaki, M., 2016. Innate and adaptive immunity in the development of depression: An update on current knowledge and technological advances. Prog Neuropsychopharmacol Biol Psychiatry 66, 63-72. Hardeland, R., 2018. Melatonin and inflammation - story of a double-edged blade. J Pineal Res, e12525. Jha, M.K., Trivedi, M.H., 2018. Personalized Antidepressant Selection and Pathway to Novel Treatments: Clinical Utility of Targeting Inflammation. Int J Mol Sci 19. Kappelmann, N., Lewis, G., Dantzer, R., Jones, P.B., Khandaker, G.M., 2016. Antidepressant activity of anti-cytokine treatment: a systematic review and metaanalysis of clinical trials of chronic inflammatory conditions. Mol Psychiatry. Kiecolt-Glaser, J.K., Wilson, S.J., Bailey, M.L., Andridge, R., Peng, J., Jaremka, L.M., Fagundes, C.P., Malarkey, W.B., Laskowski, B., Belury, M.A., 2018. Marital distress, depression, and a leaky gut: Translocation of bacterial endotoxin as a pathway to inflammation. Psychoneuroendocrinology 98, 52-60. Kohler, C.A., Freitas, T.H., Maes, M., de Andrade, N.Q., Liu, C.S., Fernandes, B.S., Stubbs, B., Solmi, M., Veronese, N., Herrmann, N., Raison, C.L., Miller, B.J., Lanctot, K.L., Carvalho, A.F., 2017. Peripheral cytokine and chemokine alterations in depression: a meta-analysis of 82 studies. Acta Psychiatr Scand 135, 373-387. Kohler, C.A., Freitas, T.H., Stubbs, B., Maes, M., Solmi, M., Veronese, N., de Andrade, N.Q., Morris, G., Fernandes, B.S., Brunoni, A.R., Herrmann, N., Raison, C.L., Miller, B.J., Lanctot, K.L., Carvalho, A.F., 2018. Peripheral Alterations in Cytokine and Chemokine Levels After Antidepressant Drug Treatment for Major Depressive Disorder: Systematic Review and Meta-Analysis. Mol Neurobiol 55, 4195-4206. Lehto, S.M., Niskanen, L., Herzig, K.H., Tolmunen, T., Huotari, A., Viinamaki, H., Koivumaa-Honkanen, H., Honkalampi, K., Ruotsalainen, H., Hintikka, J., 2010. Serum chemokine levels in major depressive disorder. Psychoneuroendocrinology 35, 226-232. Leighton, S.P., Nerurkar, L., Krishnadas, R., Johnman, C., Graham, G.J., Cavanagh, J., 2018. Chemokines in depression in health and in inflammatory illness: a systematic review and meta-analysis. Mol Psychiatry 23, 48-58. Man, K., Loudon, A., Chawla, A., 2016. Immunity around the clock. Science 354, 999-1003. Manchester, L.C., Coto-Montes, A., Boga, J.A., Andersen, L.P., Zhou, Z., Galano, A., Vriend, J., Tan, D.X., Reiter, R.J., 2015. Melatonin: an ancient molecule that makes oxygen metabolically tolerable. J Pineal Res 59, 403-419. Markus, R.P., Fernandes, P.A., Kinker, G.S., da Silveira Cruz-Machado, S., Marcola, M., 2017. Immune-pineal axis - acute inflammatory responses coordinate melatonin synthesis by pinealocytes and phagocytes. Br J Pharmacol. Martin-Subero, M., Anderson, G., Kanchanatawan, B., Berk, M., Maes, M., 2016. Comorbidity between depression and inflammatory bowel disease explained by immune-inflammatory, oxidative, and nitrosative stress; tryptophan catabolite; and gut-brain pathways. CNS Spectr 21, 184-198. Mayo, J.C., Aguado, A., Cernuda-Cernuda, R., Alvarez-Artime, A., Cepas, V., Quiros-Gonzalez, I., Hevia, D., Sainz, R.M., 2018. Melatonin Uptake by Cells: An Answer to Its Relationship with Glucose? Molecules 23. Merendino, R.A., Di Pasquale, G., De Luca, F., Di Pasquale, L., Ferlazzo, E., Martino, G., Palumbo, M.C., Morabito, F., Gangemi, S., 2004. Involvement of fractalkine and macrophage inflammatory protein-1 alpha in moderate-severe depression. Mediators Inflamm 13, 205-207.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Miller, A.H., Raison, C.L., 2016. The role of inflammation in depression: from evolutionary imperative to modern treatment target. Nat Rev Immunol 16, 22-34. Murphy, M.S., Tayade, C., Smith, G.N., 2015. Evidence of inflammation and predisposition toward metabolic syndrome after pre-eclampsia. Pregnancy Hypertens 5, 354-358. Muxel, S.M., Pires-Lapa, M.A., Monteiro, A.W., Cecon, E., Tamura, E.K., FloeterWinter, L.M., Markus, R.P., 2012. NF-kappaB drives the synthesis of melatonin in RAW 264.7 macrophages by inducing the transcription of the arylalkylamine-Nacetyltransferase (AA-NAT) gene. PLoS One 7, e52010. Myung, W., Lim, S.W., Woo, H.I., Park, J.H., Shim, S., Lee, S.Y., Kim, D.K., 2016. Serum Cytokine Levels in Major Depressive Disorder and Its Role in Antidepressant Response. Psychiatry Investig 13, 644-651. Oglodek, E.A., 2018. Changes in the concentrations of inflammatory and oxidative status biomediators (MIP-1 alpha, PMN elastase, MDA, and IL-12) in depressed patients with and without posttraumatic stress disorder. Pharmacol Rep 70, 110-118. Paltsev, M.A., Polyakova, V.O., Kvetnoy, I.M., Anderson, G., Kvetnaia, T.V., Linkova, N.S., Paltseva, E.M., Rubino, R., De Cosmo, S., De Cata, A., Mazzoccoli, G., 2016. Morphofunctional and signaling molecules overlap of the pineal gland and thymus: role and significance in aging. Oncotarget 7, 11972-11983. Pinto, A.R., da Silva, N.C., Pinato, L., 2016. Analyses of melatonin, cytokines, and sleep in chronic renal failure. Sleep Breath 20, 339-344. Rahman, S.A., Castanon-Cervantes, O., Scheer, F.A., Shea, S.A., Czeisler, C.A., Davidson, A.J., Lockley, S.W., 2015. Endogenous circadian regulation of proinflammatory cytokines and chemokines in the presence of bacterial lipopolysaccharide in humans. Brain Behav Immun 47, 4-13. Raman, C., 2002. CD5, an important regulator of lymphocyte selection and immune tolerance. Immunol Res 26, 255-263. Ramos-Casals, M., Font, J., Garcia-Carrasco, M., Calvo, J., Places, L., Padilla, O., Cervera, R., Bowen, M.A., Lozano, F., Ingelmo, M., 2001. High circulating levels of soluble scavenger receptors (sCD5 and sCD6) in patients with primary Sjogren's syndrome. Rheumatology (Oxford) 40, 1056-1059. Sarrias, M.R., Farnos, M., Mota, R., Sanchez-Barbero, F., Ibanez, A., Gimferrer, I., Vera, J., Fenutria, R., Casals, C., Yelamos, J., Lozano, F., 2007. CD6 binds to pathogen-associated molecular patterns and protects from LPS-induced septic shock. Proc Natl Acad Sci U S A 104, 11724-11729. Sarrias, M.R., Gronlund, J., Padilla, O., Madsen, J., Holmskov, U., Lozano, F., 2004. The Scavenger Receptor Cysteine-Rich (SRCR) domain: an ancient and highly conserved protein module of the innate immune system. Crit Rev Immunol 24, 1-37. Sharma, A.N., da Costa e Silva, B.F., Soares, J.C., Carvalho, A.F., Quevedo, J., 2016. Role of trophic factors GDNF, IGF-1 and VEGF in major depressive disorder: A comprehensive review of human studies. J Affect Disord 197, 9-20. Sigal, L.H., 2012. Basic science for the clinician 54: CD5. J Clin Rheumatol 18, 8388. Soderquist, F., Sundberg, I., Ramklint, M., Widerstrom, R., Hellstrom, P.M., Cunningham, J.L., 2018. The Relationship Between Daytime Salivary Melatonin and Gastrointestinal Symptoms in Young Adults Seeking Psychiatric Care. Psychosom Med. Sun, L., Yu, Z., Ye, X., Zou, S., Li, H., Yu, D., Wu, H., Chen, Y., Dore, J., Clement, K., Hu, F.B., Lin, X., 2010. A marker of endotoxemia is associated with obesity and
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
related metabolic disorders in apparently healthy Chinese. Diabetes Care 33, 19251932. Sundberg, I., Ramklint, M., Stridsberg, M., Papadopoulos, F.C., Ekselius, L., Cunningham, J.L., 2016. Salivary Melatonin in Relation to Depressive Symptom Severity in Young Adults. PLoS One 11, e0152814. Vera, J., Fenutria, R., Canadas, O., Figueras, M., Mota, R., Sarrias, M.R., Williams, D.L., Casals, C., Yelamos, J., Lozano, F., 2009. The CD5 ectodomain interacts with conserved fungal cell wall components and protects from zymosan-induced septic shock-like syndrome. Proc Natl Acad Sci U S A 106, 1506-1511. Vogelzangs, N., de Jonge, P., Smit, J.H., Bahn, S., Penninx, B.W., 2016. Cytokine production capacity in depression and anxiety. Transl Psychiatry 6, e825. Wang, Y., Liu, X., Wang, W., Song, W., Chen, L., Fang, Q., Yan, X., 2013. The expression of inflammatory cytokines on the aorta endothelia are up-regulated in pinealectomized rats. Inflammation 36, 1363-1373. Young, J.J., Bruno, D., Pomara, N., 2014. A review of the relationship between proinflammatory cytokines and major depressive disorder. J Affect Disord 169, 1520. Zhang, Q., Fang, W., Ma, L., Wang, Z.D., Yang, Y.M., Lu, Y.Q., 2018. VEGF levels in plasma in relation to metabolic control, inflammation, and microvascular complications in type-2 diabetes: A cohort study. Medicine (Baltimore) 97, e0415.
Sundberg et al Figure text.
b
na
lP
re
-p
ro of
a
Figure 1: Main analysis of all patients. a) Heat map summarizing the associations between
ur
protein (PEA) data and melatonin levels. The heat map is based on all samples and colored according to -sign(beta) x log10(p), where beta is the log10-melatonin coefficient in the linear
Jo
regression model predicting protein level (adjusted for sex, BMI, antidepressant medication and anticonception) and p is the LRT p-value. Hence, all negative associations are colored blue and all positive associations red. The p-values are printed for all associations significant after Bonferroni correction. Mel1: Melatonin when waking up; Mel2: Melatonin 30 minutes after waking up; Mel3: Melatonin at 11:00; Mel4: Melatonin 30 minutes after lunch; Mel5: Melatonin at 22:00; Mel6: Melatonin before going to bed. b) This heat map shows only
Sundberg et al proteins with any association to at least one melatonin value with a p<0.05 before correction for multiple testing b
ur
na
lP
re
-p
ro of
a
Figure 2: Post hoc analysis of only female patients. a) Heat map summarizing the
Jo
associations between protein (PEA) data and melatonin levels. The heat map is based on all samples and colored according to -sign(beta) x log10(p), where beta is the log10-melatonin coefficient in the linear regression model predicting protein level (adjusted for sex, BMI, antidepressant medication and anticonception) and p is the LRT p-value. Hence, all negative associations are colored blue and all positive associations red. The p-values are printed for all associations significant after Bonferroni correction. Mel1: Melatonin when waking up; Mel2: Melatonin 30 minutes after waking up; Mel3: Melatonin at 11:00; Mel4: Melatonin 30
Sundberg et al minutes after lunch; Mel5: Melatonin at 22:00; Mel6: Melatonin before going to bed. b) This heat map shows only proteins with any association to at least one melatonin value with a
Jo
ur
na
lP
re
-p
ro of
p<0.05 before correction for multiple testing.
Sundberg et al Tables Table 1. Characteristics of 108 patients included in the study. Characteristic
All
Females
Males
BMI, mean (SD)
24.5 21 (5.8)
24.8 21 (6.0)
22.7 21 (3.5)
MADRS-S, mean (SD)
23.6 (9.0)
24.3 (8.5)
19.0 (11.2)
Anti-depressant
55 (51 %)
48 (51 %)
7 (50 %)
Anti-psychotic
6 (6 %)
Mood stabilizing
7 (6.5 %)
Oral anti-contraceptives
N/A
Age, mean
ro of
Medication
2(14 %)
5 (14 %)
2 (5 %)
27 (29 %)
N/A
-p
4(4 %)
Diagnoses
62 (57 %)
54 (57 %)
8 (57 %)
63 (68 %)
9 (64 %)
100 (93 %)
88 (94 %)
12 (86 %)
Melatonin 30 minutes after waking up
100 (93 %)
89 (95 %)
11 (79 %)
Melatonin at 11:00
108 (100 %)
94 (100 %)
14 (100 %)
Melatonin 30 minutes after lunch
104 (96 %)
90 (96 %)
14 (100 %)
Melatonin at 22:00
104 (96 %)
91 (97 %)
13 (93 %)
Melatonin before going to bed
101 (94 %)
88 (94 %)
13 (93 %)
re
Current depression
Melatonin samples n (%)
Jo
ur
na
Melatonin when waking up
72 (67 %)
lP
Any anxiety disorder
PII:
S0306-4530(19)31255-7
DOI:
https://doi.org/10.1016/j.psyneuen.2019.104514
Reference:
PNEC 104514
To appear in:
Psychoneuroendocrinology
Received Date:
7 May 2019
Revised Date:
3 September 2019
Accepted Date:
12 November 2019
Please cite this article as: Sundberg I, Rasmusson AJ, Ramklint M, Just D, Ekselius L, Cunningham JL, Daytime Melatonin Levels in Saliva are Associated with Inflammatory Markers and Anxiety Disorders, Psychoneuroendocrinology (2019), doi: https://doi.org/10.1016/j.psyneuen.2019.104514
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.
Sundberg et al
Daytime Melatonin Levels in Saliva are Associated with Inflammatory Markers and Anxiety Disorders Isak Sundberg a, Annica J. Rasmusson b, Mia Ramklint a, David Just a, Lisa Ekselius a, Janet L. Cunningham a
Department of Neuroscience, Psychiatry, Uppsala University, Uppsala, Sweden
b
Department of Medical Sciences, Clinical Pharmacology and Osteoporosis, Uppsala
ro of
a
University, Uppsala, Sweden
-p
Corresponding author: Janet L. Cunningham MD, PhD, Associate Professor
Department of Neuroscience, Psychiatry Uppsala University Hospital, Entrance 10, Floor 3B,
lP
re
751 85, Uppsala, Sweden, E-mail: [email protected]
Highlights
na
We analyzed daytime saliva melatonin in relation to levels of 72 inflammatory markers Melatonin levels at 11:00 were correlated with CD 5 Melatonin after lunch was correlated with MCP-1, MIP-1α and VEGF-A Melatonin after lunch, VEGF-A and MIP-1α was associated with anxiety disorders
ur
ABSTRACT
Jo
Background
The bidirectional interaction between melatonin and the immune system has largely gone unexplored in a clinical context and especially in a psychiatric population. This study explored the association between melatonin during the day and inflammatory cytokines in young adult patients seeking psychiatric care.
Sundberg et al
Methods Samples and data were collected from 108 young adults (mean age 21, SD=2) at an outpatient clinic for affective disorders. Daytime saliva melatonin levels were analyzed with enzyme-linked immunosorbent assay (ELISA) in relation to normalized serum expression levels of 72 inflammatory markers in a proximity extension assay (PEA). In a post hoc analysis the markers associated with melatonin were tested in a generalized linear model to see whether there is a relationship to anxiety disorder or depression.
ro of
Results
After Bonferroni correction for multiple testing, melatonin levels at 11:00 were positively correlated with CD5 (p=4.2e-4). Melatonin levels after lunch were correlated with CCL2/MCP-1 (p=4.2e-4),
-p
CCL3/MPI-1α (p=6.5e-4) and VEGF-A (p=5.3e-6). In the generalized linear model, positive
associations were found for the presence of any anxiety disorder with melatonin after lunch (p=0.046),
re
VEGF-A (p=0.001) and CCL3/MPI-1α (p=0.001).
lP
Conclusion
Daytime saliva levels of melatonin were related to several inflammatory markers in young adults with
na
psychiatric disorders. This observation likely reflects the bidirectional relationship between melatonin production and the immune system. These findings may have relevance for the understanding of
ur
psychiatric disorders and other conditions associated with low-grade inflammation.
Jo
1. Introduction
The circadian release of melatonin from the pineal gland helps to regulate the rhythm of bodily functions hierarchically synchronized by the suprachiasmatic nucleus (Buhr and Takahashi, 2013). Melatonin is additionally produced by all mitochondria-containing cells in a non-circadian manner and exerts a wide diversity of autocrine and paracrine effects (Acuna-Castroviejo et al., 2014). Arguably,
Sundberg et al
melatonin’s most important function is as an antioxidant protecting cells from oxidative damage (Manchester et al., 2015).
A complex and fascinating interaction between melatonin and the immune system that extends well beyond antioxidative effects is summarized in a recent review (Hardeland, 2018). Melatonin is most known for its anti‐ inflammatory properties; however, depending on various conditions, including cell types, level and duration of inflammation, melatonin can have proinflammatory effects (Hardeland, 2018). Proinflammatory actions of melatonin enhance the resistance against pathogens but may be
ro of
detrimental in autoimmune diseases (such as rheumatoid arthritis) (Galbo and Kall, 2016). The anti‐ inflammatory actions of melatonin are of particular interest, however, because they are observed in high‐ grade inflammation (e.g., sepsis, ischemia/reperfusion and brain injury) and also in chronic
-p
low‐ grade inflammation during aging and in neurodegenerative diseases, with suggested relevance also in mood disorders (Hardeland, 2018). The role of melatonin in mood disorders has been
lP
mood disorders remains to be fully elucidated.
re
researched (Anderson, 2017) but the connection between daytime melatonin and inflammation in
The immune system is under circadian control, which is reflected in many ways. For instance, immune
na
cell response to antigen exposure is modulated by central circadian mediators and differs depending on the time of day (Guerrero-Vargas et al., 2014; Rahman et al., 2015). There is an endogenous circadian
ur
regulation of inflammatory cytokines such as C-X-C motif chemokine ligand 8 (CXCL8)/interleukin 8 (IL-8) and C-C motif chemokine ligand 2 (CCL2)/monocyte chemoattractant protein-1 (MCP-1) in the
Jo
presence of bacterial lipopolysaccharide (LPS) in humans (Rahman et al., 2015). Intravenous administration of melatonin suppresses markers of inflammation and oxidative damage in a human daytime endotoxemia model but these effects are dependent on the time of administration (Alamili et al., 2014). Suprachiasmatic nucleus lesion leads to a 10-fold increase in LPS-induced inflammatory cytokine release in animals (Guerrero-Vargas et al., 2014). LPS may also be involved in chronic low‐ grade inflammation. Lipopolysaccharide binding protein (LBP) has been suggested as a marker for low grade inflammation related to LPS (Citronberg et al., 2016; Gonzalez-Quintela et al., 2013). When
Sundberg et al
LPS levels are high, LBP production is induced in the liver and has a long half-life in serum. As a recent example of this, LBP was found to be higher in couples in hostile marital interactions and a history a mood disorder may contribute to this association (Kiecolt-Glaser et al., 2018).
Peripheral proinflammatory cytokines, such as tumor necrosis factor alpha (TNF)-α, shut down pineal melatonin production, contributing to the decreased pineal circadian melatonin synthesis across many medical conditions (da Silveira Cruz-Machado et al., 2012). This regulation can be mediated by TNFα directly in pinealocytes or indirectly via toll-like receptor-4 activation in microglia around the pineal
ro of
gland, which drives local TNF-α production and release (da Silveira Cruz-Machado et al., 2010;
Markus et al., 2017). Reduction in the nocturnal melatonin surge is necessary to allow leukocyte
migration to sites of injury in that circulating melatonin regulates endothelial tight junctions restricting
-p
leukocyte migration to sites of injury (Man et al., 2016).
re
During immunological challenge, immune cells are influenced by both autocrine and paracrine melatonin. While immune activation reduces melatonin production by pinealocytes, the production of
lP
melatonin is upgraded peripherally given that activated macrophages/microglia synthesize melatonin in a nuclear factor kappaB (NF-κB)-dependent manner at the site of the inflammatory response
na
(Markus et al., 2017). In cultured macrophages, NF-κB signaling pathways induce the transcription of the arylalkylamine N-acetyltransferase (AA-NAT) gene needed to catalyze the N-acetylation of
ur
serotonin into melatonin (Muxel et al., 2012). In turn, melatonin enhances phagocytosis through autocrine action and Luzindole, a competitive antagonist of melatonin receptors, decreases
Jo
macrophage phagocytic activity (Muxel et al., 2012). This bidirectional relationship between melatonin and the immune system has been termed the immune-pineal axis (Markus et al., 2017).
Few studies have investigated potential associations between endogenous daytime levels of melatonin and cytokine levels in clinical settings. Morning serum levels of melatonin were lower and TNF-α higher when patients with newly diagnosed multiple sclerosis were compared with healthy controls, but not significantly correlated with each other (Farhadi et al., 2014). In patients with chronic renal
Sundberg et al
failure, melatonin levels were lower both day and night and inversely correlated with TNF-α (Pinto et al., 2016). In contrast, patients with polymyalgia rheumatica had higher levels of melatonin compared with controls, with the relative difference being most pronounced during the day (Galbo and Kall, 2016). These studies indicate an interaction between melatonin and serum inflammatory markers but the studies differ in sampling times, which may explain some of the discrepancies.
Disruption in endocrine and inflammatory response systems are pathophysiological mechanisms implicated in the etiology of depression, anxiety and sleep disorders (Haapakoski et al., 2016;
ro of
Leighton et al., 2018; Miller and Raison, 2016; Young et al., 2014), with possible relevance for
prognosis and treatment response (Jha and Trivedi, 2018; Kappelmann et al., 2016). Previously, we have identified a negative correlation between bedtime melatonin and depressive symptoms in the
-p
present cohort of young adult patients seeking psychiatric care (Sundberg et al., 2016). The
relationship between melatonin and inflammation in psychiatric patients, however, is largely
re
unexplored.
lP
This study aimed to explore the interaction between daytime levels of melatonin in saliva taken at six time points and levels of a large panel of inflammatory cytokines in young adults with psychiatric
na
morbidity. Based on data showing a correlation between depression and evening levels of melatonin in the same cohort of patients (Sundberg et al., 2016), we hypothesized that melatonin levels in the
ur
evening would correlate with levels of inflammatory markers. A secondary hypothesis was that melatonin levels taken mid-day would correlate with inflammatory markers. In addition, we
Jo
investigated the potential relationship between inflammatory markers during the day, melatonin levels and the presence of the clinical variables: depression or anxiety disorder.
Sundberg et al
2. Materials and methods 2.1. Ethical approval and patient consent The study was approved by the Regional Ethics Committee in Uppsala (Dnr. 2012/81, Dnr. 2012/81/1 and Dnr. 2013/219) and all participants gave written consent to participate in the study.
2.2. Data cohort The material and data used in this observational, cross-sectional cohort study, which has been
ro of
previously described (Sundberg et al., 2016), originate from the Uppsala Psychiatric Patient Samples (UPP) cohort. All data were collected from patients seeking care between 2012 and 2014 at an outpatient clinic for young adults of the Section for Affective Disorders at Uppsala University
-p
Hospital, Uppsala, Sweden. This outpatient clinic provides care for patients aged 18-25 years, with mainly mood and anxiety disorders but also neuropsychiatric and personality disorders. Of 722
re
consecutive patients, 300 agreed to participate in the UPP study. Of these, 125 completed saliva sampling. Six patients were excluded because they did not fulfill criteria for any DSM-IV Axis 1
lP
diagnosis. For 11 patients, blood sampling was missing or incomplete. Thus, the final sample comprised 108 patients with a DSM-IV Axis 1 disorder. Saliva and blood samples for all 108 patients
na
were included in this study. No patients were taking antidepressant medication targeting melatonin
ur
receptors MT1/MT2 (agomelatine or ramelteon).
2.3. Clinical data
Jo
Demographic characteristics of the study participants are shown in Table 1. The procedure for assessment of psychiatric diagnoses, symptoms and physical health examination upon entering the UPP has been described in detail elsewhere (Cunningham et al., 2017). Diagnoses are based on the structured interviews used: the Swedish version of the M.I.N.I. International Neuropsychiatric Interview (M.I.N.I. 6.0) or the Structural Clinical Interview for DSM IV axis I disorders (SCID-I). All interviews were performed by trained doctors or psychologists. Physical examination included the measurement of body mass index (BMI) using the formula BMI=kg/m2.
6
Sundberg et al
2.4. Melatonin analysis Melatonin in saliva was analyzed with competitive enzyme-linked immunosorbent assay (ELISA). This procedure has been described previously (Sundberg et al., 2016). In short, participants were instructed to collect saliva samples at six time points during the waking hours of one day: when waking up, 30 minutes after waking up but before breakfast, at 11.00 hours, 30 minutes after lunch, at 22.00 hours and just before going to bed. Time points were chosen to capture the rise of melatonin in the evening before sleep and variation in daytime melatonin and specifically after meals. Saliva was
ro of
collected using inert polymer cylindrical swabs (10 mm x 30 mm) that were subsequently placed in a storage tube consisting of a large outer tube with a small insert and snap cap (swabs and tubes from
Salimetrics Europe Ltd., Suffolk, UK) and kept in the refrigerator until delivery to the lab within 48
-p
hours. Participants were instructed not to eat or drink 30 minutes before sampling. The participants
documented collection times. To ensure compliance the research assistant verified collection times and
re
sampling method with the patient upon receipt (samples not collected as instructed were excluded). Of maximally 648 (6x108) hormone measurements, 31 (4.8%) were missing because of mistakes in saliva
lP
sampling or insufficient saliva volume. Total assay variability was <11%. Upon receipt, tubes were centrifuged and stored at -20°C until analysis. Salivary melatonin was measured with competitive
na
ELISA (Direct Salivary Melatonin Elisa EK-DSM, Bühlmann Laboratories AG, Schönenbuch. Switzerland). Analyses were performed at the routine laboratory of the Department of Clinical
ur
Chemistry at Uppsala University Hospital, Uppsala, Sweden. The laboratory is certified by a Swedish
Jo
government authority (Swedac).
2.5. Proseek analysis of relative plasma levels of inflammatory proteins Blood samples, collected during office hours, were obtained from non-fasting patients and kept in 80°C at Uppsala Biobank. The relative levels of 91 inflammatory proteins were analyzed in 108 plasma samples from the study cohort using the Proseek Multiplex Inflammation panel (Olink Bioscience, Sweden), a proximity extension-based assay (PEA). The 91 proteins belonged to a preset
7
Sundberg et al inflammatory panel, initially including 92 proteins; however, one protein (Beta-nerve growth factor) was excluded by the manufacturer (see Supplementary Table 2). In short, PEA technology allows amplification and quantification of antibody-coupled, proximity-dependent DNA templates, reflecting relative protein levels in the sample (Assarsson et al., 2014). With detection sensitivity down to fg/mL, the assay can be used to compare relative protein values between groups but is not an absolute quantification. All samples were analyzed with the same batch of reagents at the Clinical Biomarker Facility at the SciLife Lab in Uppsala. DNA amplification and quantification were carried out using
ro of
the BioMark™ HD real-time PCR platform (Fluidigm, South San Francisco, CA, USA).
Plasma was isolated from blood samples and stored at -80°C within <60 minutes after isolation. After thawing on ice for the first time, samples were transferred to 96-well plates with 90 samples and 6
-p
controls (three inter-plate controls and three negative controls [buffer]) evenly distributed over the
plates. Incubation mixes (3μL) containing 92 pairs of unique DNA oligonucleotides each labeled to a
re
corresponding antibody were combined with plasma samples (1μL) and incubated at 4°C overnight. PEA enzymes and PCR reagents were added in a 96 μL extension mix, after which samples were
lP
incubated at room temperature for 5 minutes. The plate was then transferred to the thermal cycler to perform an initial DNA extension for 20 minutes at 50°C and subsequent DNA amplification in 17
na
cycles. Following the manufacturer’s instructions, a 96.96 Dynamic Array IFC (Fluidigm, South San Francisco, CA, USA) was prepared and primed while the sample mixture (2.8 μL) was mixed with
ur
7.2 μL detection mix in a new plate. The right side of the primed 96.96 Dynamic Array IFC was loaded with 5 μL of this mixture and the left side with the unique primer pairs for each protein. Using
Jo
the instructions for Proseek, the protein expression program was then run in the Fluidigm Biomark reader.
Each sample was spiked with one detection control, one extension control and two incubation controls (green fluorescent protein and phycoerythrin). Data normalization was performed using GenEx software in the Olink Wizard. The quantification cycle (Cq) values generated in the real-time PCR were used to calculate the normalized protein expression (NPX) in three steps. First, technical
8
Sundberg et al variations were corrected by subtracting the extension control from the Cq value of every sample (∆Cq sample = Cq sample – Cq extension control). The inter-plate control was then subtracted to compensate for potential variations between runs (∆∆Cq = ∆Cq sample – ∆Cq inter-plate control). The NPX was finally calculated by normalization against a calculation correction factor (NPX = correction factor – ∆∆Cq sample). The NPX data were presented on a Log2 scale.
The samples were analyzed together, randomly distributed across eight plates. The samples were normalized for plate differences using a median normalization, applied separately for each protein. For
ro of
this normalization, we considered only the 72 proteins with detectable values in at least 50% of all the plasma samples (see Supplementary Table 1). Across all 92 assays, the mean intra-assay and inter-
-p
assay variations observed were 7% and 18%, respectively.
2.6. Analysis of plasma protein levels with multiplex electrochemiluminescence analysis
re
As a validation step, two of the plasma proteins found to be significantly associated with melatonin
lP
levels after correction for multiple analyses were quantified using an electrochemiluminescence sandwich immunoassay from the Meso Scale Discovery (Rockville, MD, USA) multiplex platform.
na
In brief, plasma samples were applied to a primary antibody-coated 96-well plate. After subsequent dilution, incubation and washing, the captured proteins were incubated with secondary antibodies
ur
labelled with an electrochemiluminescence tag. The plate was then inserted into a Sector Imager 2400 (MSD, Gaithersburg, MD), where a voltage applied to the plate electrodes causes the captured labels
Jo
to emit light. The MSD Discovery Workbench Software was used to convert the intensity of emitted light to protein concentrations using interpolation from log calibrator curves corresponding to the protein of interest. The validation data were only available for two of the significant markers, namely CCL2/MCP-1 and vascular endothelial growth factor A (VEGF-A) (K15047D-1 and K15050D-1, MSD, Gaithersburg, MD). According to the manufacturer, the inter-assay variation for highest/lowest
9
Sundberg et al standards of CCL2/MCP-1 and VEGF-A has a span of (8.9-3.0) and (2.0-5.3%), respectively. Internal standards showed an inter-assay variation of 10%. Further we have measured total LBP (lipopolysaccharide binding protein) as a proxy for LPS in all 108 patients using the same multiplex electrochemiluminescence analysis (K151IYC-1, MSD, Gaithersburg, MD). Our internal standards showed an inter-assay variation of 10% or less for all three assays.
ro of
2.7. Statistical analyses Statistical analyses were performed using the software SPSS (version 23.0) and R version 3.4.2.
Results from the PEA are presented as NPX values, a relative measurement on the log2 scale. The
samples were distributed over eight plates in the PEA experiment. The plate effect was removed using
-p
a median normalization based on all plasma samples analyzed together. The plate normalized NPX
values were used in all analyses involving PEA data. Limit of detection (LOD) was strictly defined as
re
three standard deviations above background. Lower limit of quantification and upper limit of
lP
quantification were defined as lowest and highest concentrations measured with %CV<30 and relative error (RE) <30%. Values below the LOD were replaced by the LOD. The melatonin values were not
na
normally distributed and therefore log10-tranformed before all analyses.
ur
The association between protein level (NPX) and log10 (melatonin) was assessed using linear regression with the protein value (NPX) as the dependent variable and melatonin and relevant
Jo
covariates as independent variables. The association adjusted for covariates was assessed using a likelihood ratio test. The covariates were sex, BMI, use of antidepressant medication and use of oral contraceptives. Of the 91 proteins on the PEA array, we excluded those with more than 50% of the values below LOD before further analyses. There was no missing data for reasons other than low signal. Thus, 72 markers remained after this exclusion criterion was applied. Three proteins (IL-17A, IL-17C and FGF-5) with more than 20% of the values below LOD were discretized as above LOD (1) or below LOD (0) and the analyses were performed using logistic regression instead of linear
10
Sundberg et al regression. All tests were corrected using the Bonferroni method with the significance level set at error α=0.05 divided by the number of proteins used to analyze the data (72); thus, a p-value below 6.9e-4 was considered significant for the main analysis.
The relationship between melatonin values at each time point and inflammatory markers was analyzed in the whole sample. In a post hoc analysis the markers that had shown a correlation to saliva melatonin were studied in relationship to the presence of depressive episode or anxiety disorder in a
ro of
generalized linear model, controlling for sex, antidepressant medication, anticonception and BMI.
Levels of inflammatory markers VEGF-A and CCL2/MCP-1 that were measured with PEA were validated using multiplex electrochemiluminescence analysis. The correlation between plate
re
assessed using Spearman's rank-order correlation.
-p
normalized NPX values from PEA and log2-transformed values (pg/ml) from the multiplex assay were
3. Results
lP
3.1. Associations between saliva melatonin levels and inflammatory markers. The associations between protein level and melatonin are summarized in heat maps using Euclidean
na
distance and complete linkage hierarchical clustering: positive associations are shown in red and negative associations in blue (Figure 1.). Analyses were corrected for covariates sex, BMI, use of
ur
antidepressant medication and use of oral contraceptives. Supplementary Table 2 presents a list of
Jo
associations and p-values for all proteins.
There were no significant correlations between bedtime melatonin and inflammatory markers in the Proseek Multiplex Inflammation panel after correction for multiple testing (Figure 1 and Supplementary Table 2). Thus, the primary hypothesis that melatonin levels in the evening would correlate with levels of inflammatory markers was not confirmed.
11
Sundberg et al For daytime melatonin values and inflammatory markers, postprandial melatonin was positively correlated with VEGF-A (p= 5.8e-6), CCL2/MCP-1 (p=4.2e-4) and CCL3/monocyte inflammatory protein-1α (MIP-1α) (p=6.5e-4). Melatonin at 11:00 was positively correlated with cluster of differentiation 5 (CD5) (p=4.2e-4). For these markers, all values were within the level of quantification and well above LOD. Adding bedtime melatonin to the respective models did not influence the associations. BMI was positively correlated to VEGF-A (p = 0.011), CCL2/MCP-1 (p = 0.001) and CCL3/MIP-1α (p = 0.028) (Spearman). The regression analysis, however, showed that the
ro of
associations between melatonin and these markers of inflammation independent of BMI.
To study if the daytime/nighttime melatonin ratio, in addition to the daytime melatonin, is associated to protein level, melatonin at bedtime was added to the two models predicting protein level for 11:00
-p
and post prandial melatonin. A model improvement would suggest that melatonin at bedtime in
combination with the respective daytime levels would better predict protein level than each timepoint
re
alone. This was, however, not the case and we can conclude that the daytime/nighttime ratio is not
lP
associated to protein level.
3.2. Post hoc analysis of the association between melatonin levels, inflammatory markers and
na
psychiatric disorders
In the generalized linear model, melatonin at 11:00 was not associated with anxiety disorder (p=0.73)
ur
or depression (p=0.72). Melatonin after lunch was positively associated to the presence of any anxiety disorder (p=0.046) and this association remained when adding depression to the model (p=0.04).
Jo
Melatonin after lunch was not associated with depression (p=0.47). VEGF-A was associated with anxiety (p=0.001) but not with depression (p=0.6); the same findings occurred for CCL3/MIP-1α for anxiety (p=0.001) and depression (p=0.4). CCL2/MCP-1 was not linked to anxiety (p=0.18) but negatively associated with current depression (p=0.042). Finally, CD5 was neither associated with anxiety (p=0.18) nor with depression (p=0.4).
12
Sundberg et al 3.3. Validation of PEA data Plate-normalized NPX values from PEA and log2-transformed meso scale values (pg/ml) were correlated with each other: VEGF-A (r=0.56, p=6.72e-10), CCL2/MCP-1 (r=0.60, p=1.83e-11).
3.4. Post hoc analysis of association between melatonin levels and inflammatory markers in women. When looking at only women (n=94), melatonin at 11:00 am was correlated to LAP-TGFB1 (p=9.3e4), CD5 (p=1.5 e-5), CD6 (p=1e-5) and CXCL8/ IL-8 (p=7.4e-4). Daytime post-prandial values of
ro of
melatonin were correlated to IL-10RB (p=8.2e-4), CCL3/MIP1- α (p=2.6e-4) and VEGF-A (p=9e-6) (see Figure 2).
-p
3.5. Analysis of association between LBP, levels of melatonin and inflammatory markers.
Meso scale values of LBP (ng/ml) correlated with VEGF-A (r=0.397, p=7.9e-5) and CCL3/MIP-1α
re
(r=0.281, p<0.00397). No associations were found for LBP with BMI, sex, medication, the diagnosis of depressive episode or any anxiety disorder (Mann Whitney). For both VEGF-A and CCL3/MIP-1α,
lP
the associations with melatonin were independent from LBP and adding LBP to the models did not
Jo
ur
na
influence the strength of these associations (data not shown).
13
Sundberg et al
4. Discussion The main finding in this study is that daytime saliva levels of melatonin, but not bedtime melatonin levels as hypothesized, were related to inflammatory markers in blood in young adults with psychiatric morbidity. Specifically, after correction for multiple testing, daytime postprandial melatonin levels were associated with levels of VEGF-A, CCL2/MCP-1 and CCL3/MIP-1α and melatonin levels at 11:00 were associated with levels of CD5. Elevated postprandial levels of melatonin and cytokines (VEGF-A and CCL3/MIP-1α) were more
ro of
common in patients with anxiety disorders, whereas CCL2/MCP-1 was negatively associated with depression. The results demonstrate an association between melatonin and the immune
-p
system in a clinical context of psychiatric patients.
The associations between daytime melatonin levels and inflammatory markers and anxiety
re
disorders are novel findings but find support in previous work. The three markers associated with daytime postprandial melatonin levels in this study (VEGF-A, CCL2/MCP-1 and
lP
CCL3/MIP-1α) have been linked to depression and anxiety (Gaspersz et al., 2017; Leighton et al., 2018). Production capacity of several cytokines, including CCL2/MCP-1, was positively
na
associated with severity of depressive and, in particular, anxiety symptoms, even while taking lifestyle and health factors into account (Vogelzangs et al., 2016). Several studies report high
ur
peripheral VEGF-A levels in patients with depression compared with healthy controls (Carvalho et al., 2015; Sharma et al., 2016) and there is evidence for elevated VEGF-A as a
Jo
potential predictor of treatment response in major depression (Clark-Raymond et al., 2016). A majority of studies report higher concentrations of CCL2/MCP-1 in depressed patients compared with controls (Eyre et al., 2016; Kohler et al., 2017), although some studies have reported the opposite result (Lehto et al., 2010; Myung et al., 2016). A recent meta-analysis found that antidepressant treatment decreased CCL2/MCP-1 (Kohler et al., 2018). Higher levels of CCL3/MIP-1α are found in depressed patients compared with controls (Leighton et al., 2018; Merendino et al., 2004) and CCL3/MIP-1α has been suggested as a marker of
Sundberg et al oxidative stress in depression (Oglodek, 2018). The relationship between inflammation and depression is influenced by gender and generally more pronounced in females (Birur et al., 2017). In the subgroup analysis of female patients in this study, additional associations between daytime melatonin and inflammatory markers IL-8, CD5, CD6 and IL-10RB were identified.
These associations may be linked to physiological reactions to oxidative stress as both CCL2/MCP-1 and CCL2/MCP-1 are produced in response to oxidative stress (Deshmane et
ro of
al., 2009). Melatonin appears to counteract oxidative stress and reduce levels of CCL2/MCP1 in animal and cell models (Brazao et al., 2015; Wang et al., 2013). Furthermore, treatment with melatonin in rats subjected to strenuous exercise resulted in higher levels of VEGF-A,
-p
whereas muscular oxidative stress and inflammation were reduced (Borges Lda et al., 2015).
re
As far as we know, a link between the lymphocyte marker CD5 and melatonin has not previously been described. The association was strongest for melatonin at 11:00 but also
lP
present with post-prandial melatonin before correction for multiple testing. CD5 are receptors present on the surface of T cells, but also exist on a subset of B cells and as soluble receptors
na
circulating in serum (Calvo et al., 1999; Domingues et al., 2016; Sarrias et al., 2004; Sigal, 2012). The CD5 receptor is an important regulator of lymphocyte selection and immune
ur
tolerance (Raman, 2002). Increased expression of CD5 on either T cells or B cells seems to protect against autoimmunity (Dalloul, 2009) and experimental studies have shown better
Jo
survival in sepsis models when soluble CD5 has been injected into recipient rats (Sarrias et al., 2007; Vera et al., 2009). Elevated levels of soluble CD5 are linked to aging (Paltsev et al., 2016), sepsis (Aibar et al., 2015) and primary Sjögren's syndrome (Ramos-Casals et al., 2001). In one study, reduced CD5 surface expression on transitional B cells was associated with severe depression and normalized exclusively in clinical responders, possibly indicating
Sundberg et al a B cell-dependent process in the pathogenesis of depression (Ahmetspahic et al., 2018). Further investigation of this marker in the context of psychiatric disease is motivated.
The gut is a possible source of daytime melatonin. Locally, melatonin in the gut can be 400fold higher than levels released by the pineal gland, being highest after food intake (Bubenik, 2002). In a previous study we reported a relationship between post-prandial melatonin levels and gastrointestinal symptoms in the same cohort of patients (Soderquist et al., 2018). One possible underlying mechanism explaining the association between inflammatory markers and
ro of
melatonin in this study could be related to intestinal permeability (Martin-Subero et al.,
2016). In this study, daytime melatonin and LBP are not correlated with each other but both have independent correlations with VEGF-A and CCL3/MIP-1α. We could not confirm
-p
associations between LBP and BMI shown previously in patients with metabolic syndrome
(Sun et al., 2010). The young adults in this study are generally lean and somatically healthy
lP
interplay between these factors.
re
and longitudinal studies may provide interesting insight in understanding the trajectory and
Postprandial melatonin may also contribute to glycemic control. Several studies have shown
na
that individuals with poor glycemic control have higher serum levels of VEGF, CCL2/MCP1 and CCL3/MIP-1α (Chiarelli et al., 2002; Murphy et al., 2015; Zhang et al., 2018). In this
ur
study, we also noted a positive association between VEGF-A, CCL2/MCP-1 and CCL3/MIP1α and BMI: (BMI vs VEGF-A (rho 0.247, p=0.011), BMI vs CCL2/MCP-1 (rho 0.309,
Jo
p=0.001) and BMI vs CCL3/MIP-1α (rho 0.216, p=0.028). In this regard, it can be noted that melatonin has emerged as a hormone important for glucose regulation (Mayo et al., 2018). A potential relationship between daytime postprandial levels of melatonin and insulin resistance merits further exploration.
Strengths of this study include the measurement of selected cytokines using a separate method to validate the findings. Also, confirmation of inflammatory markers previously found related
Sundberg et al to psychiatric disorders provides support for the validity of the findings and against type 1 errors. The study approach maximizes the power of the dataset by examining the range of biological variation within the whole population. The relationships between diagnosis and the biological variation are explored post hoc analysis. Our data supports the hypothesis that distinct putative pathoaetiological drivers with respect to inflammation exist and that they may be more or less prevalent within different diagnostic categories. The data, however, also supports the rationale for analyzing the whole population: that underlying biological
ro of
processes seem to be both diverse within each DSM psychiatric disorder and transdiagnostic.
This study has several limitations. First, blood sampling for inflammatory markers was performed during office hours the day after saliva sampling, but not on a fixed time.
-p
Secondly, while the analyses in this study are adjusted for sex, BMI, antidepressant
medication and use of oral contraceptives, complete data were not available for other lifestyle
re
factors, such as exercise, alcohol use and smoking, all of which may influence levels of inflammatory markers (Vogelzangs et al., 2016). Thirdly, the limited amount of saliva did not
lP
allow for serial dilutions for concentrations above 50 ng/L. Fourthly, the small sample size may have limited the power of this study to identify the full extent of the relationship between
na
inflammatory markers and melatonin levels. Finally, the cross-sectional design of this study cannot provide information on the causal relationship between melatonin and inflammation
ur
over the course of disease.
Jo
In conclusion, this study demonstrated, in a psychiatric population, an association between daytime melatonin and several inflammatory markers that have previously been associated with psychiatric disorders. Evening levels of melatonin were not associated with inflammatory markers in this study. Furthermore, we saw a relationship between anxiety disorders, postprandial melatonin levels and VEGF-A and CCL3/MIP-1α. In contrast, CCL2/MCP-1 was negatively associated with depression but not anxiety disorders. These findings may have implications for understanding melatonin in the context of oxidative stress,
Sundberg et al metabolic dysregulation and inflammation and potentially have relevance in identifying biological mechanisms of psychiatric disease.
Author contributions JC conceived, designed and financed the study. MR, LE, JC contributed to acquisition of the patients and IS collected clinical data. AR and JC conducted the analysis of inflammatory markers and JC, IS and AR interpreted the data. DJ and AR conducted the analysis of LBP.
revision and approved the final version of the manuscript.
-p
Conflicts of interest
CrediT Author Statement
re
None Declarations of interest: none
ro of
IS, AR and JC drafted the first version of the manuscript. All authors contributed to critical
Writing – review & editing
lP
Isak Sundberg: Investigation, Data curation, Formal analysis, Writing – original draft,
na
Annica Rasmusson: Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing
ur
Mia Ramklint: Investigation, Writing – review & editing Lisa Ekselius: Investigation, Writing – review & editing
Jo
David Just: Data curation, Formal analysis, Writing – review & editing Janet Cunningham: Conceptualization, Methodology, Funding acquisition, Project administration, Investigation, Data curation, Formal analysis, Writing – original draft, Writing – review & editing
Sundberg et al Acknowledgements The authors wish to thank Ulla Nordén for her excellent clinical research assistance and Eva Freyhult and Hans Arinell for their exceptional statistical advice. Samples were managed in collaboration with Uppsala Biobank and methylation profiling was performed by the SNP&SEQ Technology Platform in Uppsala, which is part of the Science for Life Laboratory at Uppsala University. SciLife Laboratories are supported by the Swedish Research Council as a part of the national infrastructure. The work in this study was supported by Märta och Nicke Nasvells fund, the Apotekare Hedbergs Fund, the Erik, Karin and Gösta Selanders
ro of
Stiftelse, Fredrik and Ingrid Thurings Stiftelse, Stiftelsen Söderström-Königska sjukhemmet,
Jo
ur
na
lP
re
-p
the Swedish Society of Medicine and ALF Funds from Uppsala University Hospital.
Sundberg et al
References
Jo
ur
na
lP
re
-p
ro of
Acuna-Castroviejo, D., Escames, G., Venegas, C., Diaz-Casado, M.E., Lima-Cabello, E., Lopez, L.C., Rosales-Corral, S., Tan, D.X., Reiter, R.J., 2014. Extrapineal melatonin: sources, regulation, and potential functions. Cell Mol Life Sci 71, 29973025. Ahmetspahic, D., Schwarte, K., Ambree, O., Burger, C., Falcone, V., Seiler, K., Kooybaran, M.R., Grosse, L., Roos, F., Scheffer, J., Jorgens, S., Koelkebeck, K., Dannlowski, U., Arolt, V., Scheu, S., Alferink, J., 2018. Altered B Cell Homeostasis in Patients with Major Depressive Disorder and Normalization of CD5 Surface Expression on Regulatory B Cells in Treatment Responders. J Neuroimmune Pharmacol 13, 90-99. Aibar, J., Martinez-Florensa, M., Castro, P., Carrasco, E., Escoda-Ferran, C., Fernandez, S., Butjosa, M., Hernandez, C., Rinaudo, M., Lozano, F., Nicolas, J.M., 2015. Pattern of soluble CD5 and CD6 lymphocyte receptors in critically ill patients with septic syndromes. J Crit Care 30, 914-919. Alamili, M., Bendtzen, K., Lykkesfeldt, J., Rosenberg, J., Gogenur, I., 2014. Melatonin suppresses markers of inflammation and oxidative damage in a human daytime endotoxemia model. J Crit Care 29, 184 e189-184 e113. Anderson, G., 2017. Linking the biological underpinnings of depression: Role of mitochondria interactions with melatonin, inflammation, sirtuins, tryptophan catabolites, DNA repair and oxidative and nitrosative stress, with consequences for classification and cognition. Prog Neuropsychopharmacol Biol Psychiatry. Assarsson, E., Lundberg, M., Holmquist, G., Bjorkesten, J., Thorsen, S.B., Ekman, D., Eriksson, A., Rennel Dickens, E., Ohlsson, S., Edfeldt, G., Andersson, A.C., Lindstedt, P., Stenvang, J., Gullberg, M., Fredriksson, S., 2014. Homogenous 96-plex PEA immunoassay exhibiting high sensitivity, specificity, and excellent scalability. PLoS One 9, e95192. Birur, B., Amrock, E.M., Shelton, R.C., Li, L., 2017. Sex Differences in the Peripheral Immune System in Patients with Depression. Front Psychiatry 8, 108. Borges Lda, S., Dermargos, A., da Silva Junior, E.P., Weimann, E., Lambertucci, R.H., Hatanaka, E., 2015. Melatonin decreases muscular oxidative stress and inflammation induced by strenuous exercise and stimulates growth factor synthesis. J Pineal Res 58, 166-172. Brazao, V., Colato, R.P., Santello, F.H., Filipin Mdel, V., Toldo, M.P., do Vale, G.T., Tirapelli, C.R., do Prado Junior, J.C., 2015. Interleukin-17, oxidative stress, and inflammation: role of melatonin during Trypanosoma cruzi infection. J Pineal Res 59, 488-496. Bubenik, G.A., 2002. Gastrointestinal melatonin: localization, function, and clinical relevance. Dig Dis Sci 47, 2336-2348. Buhr, E.D., Takahashi, J.S., 2013. Molecular components of the Mammalian circadian clock. Handb Exp Pharmacol, 3-27. Calvo, J., Places, L., Espinosa, G., Padilla, O., Vila, J.M., Villamor, N., Ingelmo, M., Gallart, T., Vives, J., Font, J., Lozano, F., 1999. Identification of a natural soluble form of human CD5. Tissue Antigens 54, 128-137. Carvalho, A.F., Kohler, C.A., McIntyre, R.S., Knochel, C., Brunoni, A.R., Thase, M.E., Quevedo, J., Fernandes, B.S., Berk, M., 2015. Peripheral vascular endothelial growth factor as a novel depression biomarker: A meta-analysis. Psychoneuroendocrinology 62, 18-26.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Chiarelli, F., Cipollone, F., Mohn, A., Marini, M., Iezzi, A., Fazia, M., Tumini, S., De Cesare, D., Pomilio, M., Pierdomenico, S.D., Di Gioacchino, M., Cuccurullo, F., Mezzetti, A., 2002. Circulating monocyte chemoattractant protein-1 and early development of nephropathy in type 1 diabetes. Diabetes Care 25, 1829-1834. Citronberg, J.S., Wilkens, L.R., Lim, U., Hullar, M.A., White, E., Newcomb, P.A., Le Marchand, L., Lampe, J.W., 2016. Reliability of plasma lipopolysaccharide-binding protein (LBP) from repeated measures in healthy adults. Cancer Causes Control 27, 1163-1166. Clark-Raymond, A., Meresh, E., Hoppensteadt, D., Fareed, J., Sinacore, J., Garlenski, B., Halaris, A., 2016. Vascular endothelial growth factor: Potential predictor of treatment response in major depression. World J Biol Psychiatry, 1-11. Cunningham, J.L., Zanzi, M., Willebrand, M., Ekselius, L., Ramklint, M., 2017. No regrets: Young adult patients in psychiatry report positive reactions to biobank participation. BMC Psychiatry 17, 21. da Silveira Cruz-Machado, S., Carvalho-Sousa, C.E., Tamura, E.K., Pinato, L., Cecon, E., Fernandes, P.A., de Avellar, M.C., Ferreira, Z.S., Markus, R.P., 2010. TLR4 and CD14 receptors expressed in rat pineal gland trigger NFKB pathway. J Pineal Res 49, 183-192. da Silveira Cruz-Machado, S., Pinato, L., Tamura, E.K., Carvalho-Sousa, C.E., Markus, R.P., 2012. Glia-pinealocyte network: the paracrine modulation of melatonin synthesis by tumor necrosis factor (TNF). PLoS One 7, e40142. Dalloul, A., 2009. CD5: a safeguard against autoimmunity and a shield for cancer cells. Autoimmun Rev 8, 349-353. Deshmane, S.L., Kremlev, S., Amini, S., Sawaya, B.E., 2009. Monocyte chemoattractant protein-1 (MCP-1): an overview. J Interferon Cytokine Res 29, 313326. Domingues, R.G., Lago-Baldaia, I., Pereira-Castro, I., Fachini, J.M., Oliveira, L., Drpic, D., Lopes, N., Henriques, T., Neilson, J.R., Carmo, A.M., Moreira, A., 2016. CD5 expression is regulated during human T-cell activation by alternative polyadenylation, PTBP1, and miR-204. Eur J Immunol 46, 1490-1503. Eyre, H.A., Air, T., Pradhan, A., Johnston, J., Lavretsky, H., Stuart, M.J., Baune, B.T., 2016. A meta-analysis of chemokines in major depression. Prog Neuropsychopharmacol Biol Psychiatry 68, 1-8. Farhadi, N., Oryan, S., Nabiuni, M., 2014. Serum levels of melatonin and cytokines in multiple sclerosis. Biomed J 37, 90-92. Galbo, H., Kall, L., 2016. Circadian variations in clinical symptoms and concentrations of inflammatory cytokines, melatonin, and cortisol in polymyalgia rheumatica before and during prednisolone treatment: a controlled, observational, clinical experimental study. Arthritis Res Ther 18, 174. Gaspersz, R., Lamers, F., Wittenberg, G., Beekman, A.T.F., van Hemert, A.M., Schoevers, R.A., Penninx, B., 2017. The role of anxious distress in immune dysregulation in patients with major depressive disorder. Transl Psychiatry 7, 1268. Gonzalez-Quintela, A., Alonso, M., Campos, J., Vizcaino, L., Loidi, L., Gude, F., 2013. Determinants of serum concentrations of lipopolysaccharide-binding protein (LBP) in the adult population: the role of obesity. PLoS One 8, e54600. Guerrero-Vargas, N.N., Salgado-Delgado, R., Basualdo Mdel, C., Garcia, J., GuzmanRuiz, M., Carrero, J.C., Escobar, C., Buijs, R.M., 2014. Reciprocal interaction between the suprachiasmatic nucleus and the immune system tunes down the inflammatory response to lipopolysaccharide. J Neuroimmunol 273, 22-30.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Haapakoski, R., Ebmeier, K.P., Alenius, H., Kivimaki, M., 2016. Innate and adaptive immunity in the development of depression: An update on current knowledge and technological advances. Prog Neuropsychopharmacol Biol Psychiatry 66, 63-72. Hardeland, R., 2018. Melatonin and inflammation - story of a double-edged blade. J Pineal Res, e12525. Jha, M.K., Trivedi, M.H., 2018. Personalized Antidepressant Selection and Pathway to Novel Treatments: Clinical Utility of Targeting Inflammation. Int J Mol Sci 19. Kappelmann, N., Lewis, G., Dantzer, R., Jones, P.B., Khandaker, G.M., 2016. Antidepressant activity of anti-cytokine treatment: a systematic review and metaanalysis of clinical trials of chronic inflammatory conditions. Mol Psychiatry. Kiecolt-Glaser, J.K., Wilson, S.J., Bailey, M.L., Andridge, R., Peng, J., Jaremka, L.M., Fagundes, C.P., Malarkey, W.B., Laskowski, B., Belury, M.A., 2018. Marital distress, depression, and a leaky gut: Translocation of bacterial endotoxin as a pathway to inflammation. Psychoneuroendocrinology 98, 52-60. Kohler, C.A., Freitas, T.H., Maes, M., de Andrade, N.Q., Liu, C.S., Fernandes, B.S., Stubbs, B., Solmi, M., Veronese, N., Herrmann, N., Raison, C.L., Miller, B.J., Lanctot, K.L., Carvalho, A.F., 2017. Peripheral cytokine and chemokine alterations in depression: a meta-analysis of 82 studies. Acta Psychiatr Scand 135, 373-387. Kohler, C.A., Freitas, T.H., Stubbs, B., Maes, M., Solmi, M., Veronese, N., de Andrade, N.Q., Morris, G., Fernandes, B.S., Brunoni, A.R., Herrmann, N., Raison, C.L., Miller, B.J., Lanctot, K.L., Carvalho, A.F., 2018. Peripheral Alterations in Cytokine and Chemokine Levels After Antidepressant Drug Treatment for Major Depressive Disorder: Systematic Review and Meta-Analysis. Mol Neurobiol 55, 4195-4206. Lehto, S.M., Niskanen, L., Herzig, K.H., Tolmunen, T., Huotari, A., Viinamaki, H., Koivumaa-Honkanen, H., Honkalampi, K., Ruotsalainen, H., Hintikka, J., 2010. Serum chemokine levels in major depressive disorder. Psychoneuroendocrinology 35, 226-232. Leighton, S.P., Nerurkar, L., Krishnadas, R., Johnman, C., Graham, G.J., Cavanagh, J., 2018. Chemokines in depression in health and in inflammatory illness: a systematic review and meta-analysis. Mol Psychiatry 23, 48-58. Man, K., Loudon, A., Chawla, A., 2016. Immunity around the clock. Science 354, 999-1003. Manchester, L.C., Coto-Montes, A., Boga, J.A., Andersen, L.P., Zhou, Z., Galano, A., Vriend, J., Tan, D.X., Reiter, R.J., 2015. Melatonin: an ancient molecule that makes oxygen metabolically tolerable. J Pineal Res 59, 403-419. Markus, R.P., Fernandes, P.A., Kinker, G.S., da Silveira Cruz-Machado, S., Marcola, M., 2017. Immune-pineal axis - acute inflammatory responses coordinate melatonin synthesis by pinealocytes and phagocytes. Br J Pharmacol. Martin-Subero, M., Anderson, G., Kanchanatawan, B., Berk, M., Maes, M., 2016. Comorbidity between depression and inflammatory bowel disease explained by immune-inflammatory, oxidative, and nitrosative stress; tryptophan catabolite; and gut-brain pathways. CNS Spectr 21, 184-198. Mayo, J.C., Aguado, A., Cernuda-Cernuda, R., Alvarez-Artime, A., Cepas, V., Quiros-Gonzalez, I., Hevia, D., Sainz, R.M., 2018. Melatonin Uptake by Cells: An Answer to Its Relationship with Glucose? Molecules 23. Merendino, R.A., Di Pasquale, G., De Luca, F., Di Pasquale, L., Ferlazzo, E., Martino, G., Palumbo, M.C., Morabito, F., Gangemi, S., 2004. Involvement of fractalkine and macrophage inflammatory protein-1 alpha in moderate-severe depression. Mediators Inflamm 13, 205-207.
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
Miller, A.H., Raison, C.L., 2016. The role of inflammation in depression: from evolutionary imperative to modern treatment target. Nat Rev Immunol 16, 22-34. Murphy, M.S., Tayade, C., Smith, G.N., 2015. Evidence of inflammation and predisposition toward metabolic syndrome after pre-eclampsia. Pregnancy Hypertens 5, 354-358. Muxel, S.M., Pires-Lapa, M.A., Monteiro, A.W., Cecon, E., Tamura, E.K., FloeterWinter, L.M., Markus, R.P., 2012. NF-kappaB drives the synthesis of melatonin in RAW 264.7 macrophages by inducing the transcription of the arylalkylamine-Nacetyltransferase (AA-NAT) gene. PLoS One 7, e52010. Myung, W., Lim, S.W., Woo, H.I., Park, J.H., Shim, S., Lee, S.Y., Kim, D.K., 2016. Serum Cytokine Levels in Major Depressive Disorder and Its Role in Antidepressant Response. Psychiatry Investig 13, 644-651. Oglodek, E.A., 2018. Changes in the concentrations of inflammatory and oxidative status biomediators (MIP-1 alpha, PMN elastase, MDA, and IL-12) in depressed patients with and without posttraumatic stress disorder. Pharmacol Rep 70, 110-118. Paltsev, M.A., Polyakova, V.O., Kvetnoy, I.M., Anderson, G., Kvetnaia, T.V., Linkova, N.S., Paltseva, E.M., Rubino, R., De Cosmo, S., De Cata, A., Mazzoccoli, G., 2016. Morphofunctional and signaling molecules overlap of the pineal gland and thymus: role and significance in aging. Oncotarget 7, 11972-11983. Pinto, A.R., da Silva, N.C., Pinato, L., 2016. Analyses of melatonin, cytokines, and sleep in chronic renal failure. Sleep Breath 20, 339-344. Rahman, S.A., Castanon-Cervantes, O., Scheer, F.A., Shea, S.A., Czeisler, C.A., Davidson, A.J., Lockley, S.W., 2015. Endogenous circadian regulation of proinflammatory cytokines and chemokines in the presence of bacterial lipopolysaccharide in humans. Brain Behav Immun 47, 4-13. Raman, C., 2002. CD5, an important regulator of lymphocyte selection and immune tolerance. Immunol Res 26, 255-263. Ramos-Casals, M., Font, J., Garcia-Carrasco, M., Calvo, J., Places, L., Padilla, O., Cervera, R., Bowen, M.A., Lozano, F., Ingelmo, M., 2001. High circulating levels of soluble scavenger receptors (sCD5 and sCD6) in patients with primary Sjogren's syndrome. Rheumatology (Oxford) 40, 1056-1059. Sarrias, M.R., Farnos, M., Mota, R., Sanchez-Barbero, F., Ibanez, A., Gimferrer, I., Vera, J., Fenutria, R., Casals, C., Yelamos, J., Lozano, F., 2007. CD6 binds to pathogen-associated molecular patterns and protects from LPS-induced septic shock. Proc Natl Acad Sci U S A 104, 11724-11729. Sarrias, M.R., Gronlund, J., Padilla, O., Madsen, J., Holmskov, U., Lozano, F., 2004. The Scavenger Receptor Cysteine-Rich (SRCR) domain: an ancient and highly conserved protein module of the innate immune system. Crit Rev Immunol 24, 1-37. Sharma, A.N., da Costa e Silva, B.F., Soares, J.C., Carvalho, A.F., Quevedo, J., 2016. Role of trophic factors GDNF, IGF-1 and VEGF in major depressive disorder: A comprehensive review of human studies. J Affect Disord 197, 9-20. Sigal, L.H., 2012. Basic science for the clinician 54: CD5. J Clin Rheumatol 18, 8388. Soderquist, F., Sundberg, I., Ramklint, M., Widerstrom, R., Hellstrom, P.M., Cunningham, J.L., 2018. The Relationship Between Daytime Salivary Melatonin and Gastrointestinal Symptoms in Young Adults Seeking Psychiatric Care. Psychosom Med. Sun, L., Yu, Z., Ye, X., Zou, S., Li, H., Yu, D., Wu, H., Chen, Y., Dore, J., Clement, K., Hu, F.B., Lin, X., 2010. A marker of endotoxemia is associated with obesity and
Sundberg et al
Jo
ur
na
lP
re
-p
ro of
related metabolic disorders in apparently healthy Chinese. Diabetes Care 33, 19251932. Sundberg, I., Ramklint, M., Stridsberg, M., Papadopoulos, F.C., Ekselius, L., Cunningham, J.L., 2016. Salivary Melatonin in Relation to Depressive Symptom Severity in Young Adults. PLoS One 11, e0152814. Vera, J., Fenutria, R., Canadas, O., Figueras, M., Mota, R., Sarrias, M.R., Williams, D.L., Casals, C., Yelamos, J., Lozano, F., 2009. The CD5 ectodomain interacts with conserved fungal cell wall components and protects from zymosan-induced septic shock-like syndrome. Proc Natl Acad Sci U S A 106, 1506-1511. Vogelzangs, N., de Jonge, P., Smit, J.H., Bahn, S., Penninx, B.W., 2016. Cytokine production capacity in depression and anxiety. Transl Psychiatry 6, e825. Wang, Y., Liu, X., Wang, W., Song, W., Chen, L., Fang, Q., Yan, X., 2013. The expression of inflammatory cytokines on the aorta endothelia are up-regulated in pinealectomized rats. Inflammation 36, 1363-1373. Young, J.J., Bruno, D., Pomara, N., 2014. A review of the relationship between proinflammatory cytokines and major depressive disorder. J Affect Disord 169, 1520. Zhang, Q., Fang, W., Ma, L., Wang, Z.D., Yang, Y.M., Lu, Y.Q., 2018. VEGF levels in plasma in relation to metabolic control, inflammation, and microvascular complications in type-2 diabetes: A cohort study. Medicine (Baltimore) 97, e0415.
Sundberg et al Figure text.
b
na
lP
re
-p
ro of
a
Figure 1: Main analysis of all patients. a) Heat map summarizing the associations between
ur
protein (PEA) data and melatonin levels. The heat map is based on all samples and colored according to -sign(beta) x log10(p), where beta is the log10-melatonin coefficient in the linear
Jo
regression model predicting protein level (adjusted for sex, BMI, antidepressant medication and anticonception) and p is the LRT p-value. Hence, all negative associations are colored blue and all positive associations red. The p-values are printed for all associations significant after Bonferroni correction. Mel1: Melatonin when waking up; Mel2: Melatonin 30 minutes after waking up; Mel3: Melatonin at 11:00; Mel4: Melatonin 30 minutes after lunch; Mel5: Melatonin at 22:00; Mel6: Melatonin before going to bed. b) This heat map shows only
Sundberg et al proteins with any association to at least one melatonin value with a p<0.05 before correction for multiple testing b
ur
na
lP
re
-p
ro of
a
Figure 2: Post hoc analysis of only female patients. a) Heat map summarizing the
Jo
associations between protein (PEA) data and melatonin levels. The heat map is based on all samples and colored according to -sign(beta) x log10(p), where beta is the log10-melatonin coefficient in the linear regression model predicting protein level (adjusted for sex, BMI, antidepressant medication and anticonception) and p is the LRT p-value. Hence, all negative associations are colored blue and all positive associations red. The p-values are printed for all associations significant after Bonferroni correction. Mel1: Melatonin when waking up; Mel2: Melatonin 30 minutes after waking up; Mel3: Melatonin at 11:00; Mel4: Melatonin 30
Sundberg et al minutes after lunch; Mel5: Melatonin at 22:00; Mel6: Melatonin before going to bed. b) This heat map shows only proteins with any association to at least one melatonin value with a
Jo
ur
na
lP
re
-p
ro of
p<0.05 before correction for multiple testing.
Sundberg et al Tables Table 1. Characteristics of 108 patients included in the study. Characteristic
All
Females
Males
BMI, mean (SD)
24.5 21 (5.8)
24.8 21 (6.0)
22.7 21 (3.5)
MADRS-S, mean (SD)
23.6 (9.0)
24.3 (8.5)
19.0 (11.2)
Anti-depressant
55 (51 %)
48 (51 %)
7 (50 %)
Anti-psychotic
6 (6 %)
Mood stabilizing
7 (6.5 %)
Oral anti-contraceptives
N/A
Age, mean
ro of
Medication
2(14 %)
5 (14 %)
2 (5 %)
27 (29 %)
N/A
-p
4(4 %)
Diagnoses
62 (57 %)
54 (57 %)
8 (57 %)
63 (68 %)
9 (64 %)
100 (93 %)
88 (94 %)
12 (86 %)
Melatonin 30 minutes after waking up
100 (93 %)
89 (95 %)
11 (79 %)
Melatonin at 11:00
108 (100 %)
94 (100 %)
14 (100 %)
Melatonin 30 minutes after lunch
104 (96 %)
90 (96 %)
14 (100 %)
Melatonin at 22:00
104 (96 %)
91 (97 %)
13 (93 %)
Melatonin before going to bed
101 (94 %)
88 (94 %)
13 (93 %)
re
Current depression
Melatonin samples n (%)
Jo
ur
na
Melatonin when waking up
72 (67 %)
lP
Any anxiety disorder