The secret behind peripheral inammation and depression: The hypothalamus

In recent years there has been increasing evidence of an inammatory component due to overstimulation of the hypothalamic-pituitary-adrenals (HPA) in depression. The glutamate metabolites (glutamate and glutamine) are important metabolites that are involved in this stimulation. The aim of this study was to determine the concentrations of hypothalamic glutamate metabolites in depression and to investigate their relationship to peripheral inammation. Participants with diagnosed depression (DE; n = 24) and control subjects without depression (HC; n = 25) were investigated. Hypothalamic glutamate metabolites were recorded using in vivo magnetic resonance spectroscopy. Peripheral cytokines (IL-6, TNF-α, IFN-γ and IL-1β) were assessed using Enzyme-Linked Immunosorbent Assay. For statistical analysis, generalized mixed models were computed using Poisson distributions and a log link function. The results show overall higher hypothalamic glutamate metabolites in DE compared to HC. High TNF-a and IL-1ß concentrations are associated with high hypothalamic glutamate metabolites in DE. These results provide initial evidence that, in depression, increased HPA axis activity is associated with peripheral inammation favored by hypothalamic glutamate metabolites. the HPA axis in depression, the present study focuses on two goals. The rst goal is to determine the differences in the concentration of 1H-MRS hypothalamic glutamine metabolites (Glx, Glu and Gln) in people with depression and community healthy controls. The second goal involves the investigation of the relationship between proinammatory cytokine concentrations in peripheral plasma (IL-6, TNF-α, IFN-γ and IL-1β) and concentrations of hypothalamic glutamine metabolites (Glx, Glu, and Gln) in patients with depression. Gln concentration in the hippocampus in patients with MDD 33 . On the other hand, the study by Bhagwagar et al. reported increased Glu and Gln concentrations in the occipital cortex in participants with MDD 34,35 . The overall higher concentrations of hypothalamic glutamate metabolites found in the present study show that there are no negative changes in neuronal activity such as apoptosis. In various diseases such as hepatic encephalopathy, a higher and sustained increase in a glutamate metabolite (Glu) is observed in the CNS 36 . In such a case, increased levels of Glu cause apoptosis of the neurons, which leads to cognitive and consciousness disorders 36 . The overall effect on the three hypothalamic glutamate metabolites could also suggest that overexpressed astrocyte activity in the hypothalamus may be associated with depression. concentrations between DE and HC: The results of this study showed that people with depression had positive associations between high TNF-α concentrations and Glx/tCr, between high IL-1β concentrations and Glx/tCr and between high IL-1β concentrations and Gln/tCr. The effects of possible confounding factors such as age, BMI and gender were taken into account. IFN-α The study raises other questions can seriously contribute to understanding the relationship between depression, inammation, and hypothalamic activity. (1) The investigation of the above mentioned possible synergistic effect of Gln and Glu with TNF-α, (2) the correlation between quinolinic acid from CSF with glutamate-glutamine metabolites and inammation parameters, (3) investigation of patients with early onset depression in a longitudinal study or cohort to monitor changes in inammation and glutamate-glutamine metabolites in the CNS over time to investigate the possibility that inammatory cytokines affect the hypothalamic structure and function of depression, or (4) the study of other MRS metabolites in the CNS in the context of inammation, HPA axis and depression, for instance energy metabolites. Other inammatory agents such as IL-17 or IL-34 would also be of interest, (5) as well as the effects of ACTH and cortisol on the glutamate-glutamine metabolites in the hypothalamus. Conclusions: In summary, this study shows hypothalamic glutamate-glutamine differences between people with depression and people without depression. A positive correlation between the pro-inammatory cytokine TNF-α and the Glx concentration was found in depression. However, there was no signicant correlation between TNF-α and Glu or Gln concentrations. This could show that Gln and Glu play a synergistic role in the relationship with TNF-α, and could indicate that the higher the TNF-α concentration, the greater the simultaneous effects on neuronal metabolism (i.e. Gln) and on excitatory neurotransmission (i.e. Glu). Finally, IL-1β positively correlated with Glx and Gln levels in depression, which could indicate the effect of IL-1β on the metabolism of hypothalamic neurons. With this in mind, future studies are needed to clarify the role of metabolism and hypothalamic function in peripheral systemic anti-inammatory.


Introduction
The hypothalamus is involved in the regulation of many physical and psychological processes, including human homeostasis 1,2 along the hypothalamicpituitary-adrenal axis (HPA). The HPA axis regulates hormone and immune functions and plays an important role in in ammation physiology 3 . The role of the HPA axis in regulating stress responses and their effects on mood and emotions has been well described 4,5 . In recent years, the importance of the HPA axis for the pathophysiology of various mood disorders such as depression [6][7][8] has been emphasized. Models that support the in uence of the HPA axis on depression postulate chronically dysregulated activation of the HPA axis, which is also associated with dysregulation of immune function [8][9][10][11] . Various animal studies have shown the relationship between depression and in ammation [12][13][14][15] . In a human study, Pedraz-Petrozzi recently showed correlations between the severity of depression, assessed with the BDI-FS 16 , and the pro-in ammatory cytokine TNF-α in people with depression. Overactivation of the HPA axis is also known to affect immune function and cortisol metabolism, ultimately leading to symptoms associated with depression 17,18 .
Human studies based on the relationship between metabolism and brain activity are required to con rm the hypothesis of overactivation of the HPA axis in depression. In recent years, there have been many suggestions related to imaging and metabolic research for depression. One of these is the examination of brain endophenotypes for depression using in vivo magnetic resonance spectroscopy (MRS) 19,20 .
Magnetic resonance spectroscopy enables the non-invasive detection of the concentrations of various brain metabolites using strong magnetic elds 21 . The concentrations of metabolites or neurotransmitters can be determined using MRS 22 , including glutamine metabolites such as glutamate (Glu) and glutamine (Gln) 23 or the metabolite Glx, which is the sum of Gln and Glu 24 . Glutamate metabolites are mainly associated with central nervous system (CNS) excitatory activity 23,24 . It is known that glutamic activity in the CNS plays a fundamental role in brain metabolic activity, glucose uptake, and excitatory electrical activity in the CNS 19 . Regarding MRS research, the number of publications on glutamate metabolites in depression has increased recently. The brain regions targeted related mainly to cognition and emotion 25 , such as the hippocampus 26,27 , the anterior cingulate gyrus 28,29 and the prefrontal cortex 30 . Despite a signi cant number of studies on MRS in depression, there are few studies that involve the hypothalamus or the HPA axis. Only one study found differences between Glx levels in depression 31 . However, this study included multiple sclerosis patients with depression, but not primarily depressed participants without comorbid illness. Aside from this study, there is no study that evaluates hypothalamic activity in depressed patients without other comorbid conditions using MRS.
To further elucidate the role of the HPA axis in depression, the present study focuses on two goals. The rst goal is to determine the differences in the concentration of 1H-MRS hypothalamic glutamine metabolites (Glx, Glu and Gln) in people with depression and community healthy controls. The second goal involves the investigation of the relationship between proin ammatory cytokine concentrations in peripheral plasma (IL-6, TNF-α, IFN-γ and IL-1β) and concentrations of hypothalamic glutamine metabolites (Glx, Glu, and Gln) in patients with depression.

Results
General sample description General sample description, including baseline clinical and laboratory characteristics of the 49 participants included in the study are listed in table 1.
Descriptive data regarding hypothalamic metabolites are also represented in Association between the peripheral pro-in ammatory cytokines and 1 H-MRS hypothalamic glutamate metabolites The second issue was to determine whether there were signi cant associations between the measured proin ammatory peripheral cytokines and the glutamate hypothalamic metabolites. Pro-in ammatory cytokine concentrations were classi ed as low or high by using the median split method. The focus was on the main effects of the cytokine (low, high) and the interaction effects of the cytokine x group. Corresponding odds ratio, con dence intervals and pvalues are listed in table 2 and are shown in gure 2.
First goal -Group differences with regard to 1H-MRS hypothalamic glutamate metabolites: The analysis revealed a group effect, but no interaction effect (metabolite x group). This means that participants with depression overall had higher hypothalamic glutamate metabolites than the participants without depression.
To the best of our knowledge, this study is the rst to examine the concentration of hypothalamic glutamine metabolites (Glx, Gln and Glu) in depression with no medical comorbid condition. A study by Kantorová et al. studied Glx levels in patients with multiple sclerosis (MS) and healthy subjects 31  The overall higher concentrations of hypothalamic glutamate metabolites found in the present study show that there are no negative changes in neuronal activity such as apoptosis. In various diseases such as hepatic encephalopathy, a higher and sustained increase in a glutamate metabolite (Glu) is observed in the CNS 36 . In such a case, increased levels of Glu cause apoptosis of the neurons, which leads to cognitive and consciousness disorders 36 . The overall effect on the three hypothalamic glutamate metabolites could also suggest that overexpressed astrocyte activity in the hypothalamus may be associated with depression.
Astrocytes are part of the glutamate metabolism in the CNS by capturing the glutamate produced by the neurons and converting it to glutamine 37 . In addition to the intervention of astrocytes in the glutamate-glutamine metabolism, astrocytes also support neurons with energy and metabolism. In this case, glutamine is an important source of energy for neurons 38,39 . The energy consumption associated with depolarization and excitation in the neurons (glutamate) promotes an immediate glutamine metabolism and, like glucose, supports the neuronal energy requirement 38,39 . This effect could explain the overall effect on the hypothalamic glutamate metabolites in depression, suggesting that the hypothalamus of depressed people has a higher level of energy, metabolism, and nerve activity. This means that an overreaction of astrocytes could correspond to an overactivation of some neurons, which at the same time would correspond to a stronger excitation (glutamate, neurons) and a stronger cell metabolism (glutamine, astrocytes) in the hypothalamus of depressed patients. Further studies in this direction are still missing to con rm such a proposal.
Second goal -Effects of the MRS glutamate metabolites and the peripheral plasma pro-in ammatory cytokine concentrations between DE and HC: The results of this study showed that people with depression had positive associations between high TNF-α concentrations and Glx/tCr, between high IL-1β concentrations and Glx/tCr and between high IL-1β concentrations and Gln/tCr. The effects of possible confounding factors such as age, BMI and gender were taken into account.
The positive relationship between hypothalamic glutamine metabolites and high peripheral proin ammatory cytokine levels in patients with depression but without additional disorders is particularly interesting because few published studies are available to support this correlation. Gonçalves-de-Rezende et al.
reported correlations between the cortisol awakening response and Glx concentrations 29 in patients with postpartum depression, indicating a positive relationship between hypothalamic activity and pro-in ammatory metabolites (e.g. cortisol). Another study on the relationship between glutamate brain metabolites and peripheral in ammation is the work of Haroon and colleagues 40 . This study did not speci cally examine the hypothalamus, but rather changes in cortical and subcortical glutamate metabolites and their relationship to pro-in ammatory markers (e.g. TNF) and depression symptoms during IFNα treatment in patients with hepatitis C. The results showed a relationship between in ammation markers increased glutamate concentrations and reduced motivation 40 .
Another study on this topic is the work of Slavich and colleagues, which showed stronger associations between activity in the dorsal anterior cingulate cortex (dACC) and in the anterior insula with an increase in the TNF-α-soluble receptor 2 41 . These brain regions were also associated with negative affect and distress. The differences between the study by Slavich and colleagues and the present study concern the sample (healthy young adults), the imaging methods and the examined areas of the brain (dACC and anterior insula).
The results of this study showed no positive associations between high IL-6 or IFN-γ concentrations and glutamine hypothalamic metabolites in patients with depression. In addition, TNF-α did not correlate with the individual Gln or Glu concentrations, but with the combined Gln and Glu concentrations, i.e. the Glx concentration. This indicates a synergistic effect of Gln and Glu, which leads to an increase in the variance when the two substances are combined. TNF-α could therefore affect both substances more directly than either. For example, TNF-α could have a direct impact on the metabolites involved in neuronal metabolism and energy support (Gln), as well as excitatory transmission and toxicity (Glu). However, further evidence is needed to understand the effect of TNF-α on hypothalamic and brain metabolism, cell energy processes and excitatory transmission in the human CNS. Finally, in patients with depression, higher IL-1β levels were associated with both Glx and Gln levels. This indicates that the increase in IL-1β in people with depression depends more on hypothalamic neuron metabolism and energy support (Gln) than on excitatory neuronal activity (Glu). The IL-1 family has been shown to induce the activity of various neurotransmitters in HPA 42 . The induction of various neurotransmitters due to the presence of cytokines from the IL-1 family is mainly related to the activation of cortisol metabolism. The latter is usually associated with depression symptoms and other mood disorders 42 . In contrast to the current data situation, the results of the present study showed that HPA activity in depression is not due to excitatory neural activity, but mainly to an increased cell metabolism (glutamine) in the HPA axis. However, further studies (i.e. the role of IL-1β on the hypothalamus in depression) are needed to con rm these results.
Limitations: The current study has shown interesting results regarding hypothalamic glutamate-glutamine metabolites, in ammatory cytokines and depression. However, some aspects should be considered when evaluating the result. As almost always, a larger sample could improve the generalizability of the study results. However, the estimated power (1-β) for a total sample size of 49 participants is 0.98, which is above the accepted minimum power threshold of 1-β = 0.80. This indicates that the sample size for this study design is su cient to achieve the study goals.
More women than men were included in the study. However, the ratio of women to men was the same in both groups examined and the ratio of female to male participants corresponds to the population of people with depression in Germany 43,44 .
Attention should be drawn to a limitation by the method for non-invasive detection of glutamate metabolites, MRS with chemical shift imaging (CSI). Although all extracted voxels come from a brain slice that is selected to best contain the hypothalamus, there is a theoretical possibility that the signals attributed to the hypothalamus may also come from other neighboring structures (i.e. thalamus, corpus callosum, hippocampus, etc.). As far as we know, the method is currently the best for detecting non-invasive hypothalamic glutamate levels.
Future directions: The study also raises other questions that can seriously contribute to understanding the relationship between depression, in ammation, and hypothalamic activity.

Materials And Methods
Study design and selection criteria A case-control study was carried out. 25 patients who met the ICD-10 criteria for depression or depressive episodes (DE) and 24 healthy controls (HC, community sample) between the ages of 18 and 65 were included in this study. The participants were between 18 and 65 years old. Both samples matched age and gender. The data was collected between June 2019 and September 2019. A sample description with the data on the pro-in ammatory cytokines and the indicator of the severity of the depression is given in table 1.
Volunteers were included in the DE group who had a depressive episode but no accompanying psychotic episode. People with another comorbid psychiatric illness (i.e., bipolar disorder, personality disorder, adaptive syndrome, or post-traumatic stress disorder) have been included as long as the depressive episode has predominated in the past 6 months. Exclusion criteria for both groups were insu cient knowledge of the German language and somatic restrictions that did not allow participation, especially visual or acoustic restrictions. With the exception of depression or depressive episode among the participants in the DE group, any acute or chronic illness was an exclusion criterion for both groups (e.g. u-like symptoms, systemic lupus erythematosus, rheumatoid arthritis, cancer, etc.). In addition, volunteers were not included in the study if contraindications to participating in MRI studies were identi ed (e.g. dental implants, pacemakers, etc.).
Each participant or its legal authorized representative were fully informed about the objectives and procedures of this study and gave his written consent to participate. All procedures complied with the Helsinki Declaration. The approval of the local ethics committee of the medical faculty of the Justus Liebig University (JLU) has been obtained. The study complies with the APA ethical standards.
Data collection Blood sampling 3 mL fasting venous blood samples were collected between 8:00 am and 12:00 pm with EDTA K blood sample tubes (S-Monovette 2.7 mL K3E tube with 1.6 mg EDTA/mL, SARSTEDT AG & Co. KG Nümbrecht, Germany) and then centrifuged at 4°C with 1100 x g for 15 minutes. After centrifugation, 1.5 mL plasma was collected and immediately stored at -20°C. Every 4 weeks, the blood samples were collectively delivered to a university research facility, which was about 30 km away, and stored at -80°C for further use.
Values below the ELISA detection limit was considered to be zero pg/mL and included in the analysis. Maximum plasma cytokine values (indicated by the software as '> Max') were excluded from the analysis.
Cytokine concentrations were calculated using the Tecan Reader and Magellan Reader Software (Tecan Group Ltd., Männedorf, Switzerland). For the parameter calculation, the Marquardt's 4-parameter estimation method was used.

Depression
Depressive episodes (DE) were de ned according to the diagnostic criteria of the International Statistical Classi cation of Diseases and Related Health Problems, 10th version (ICD-10) 45 . The presence of depressive episodes was diagnosed by clinical experts of the psychiatry department from the University Hospital Giessen and Marburg (location -Giessen).
The Beck Depression Inventory -Fast Screening, German version (BDI-FS) 46 was used to assess the severity of depressive symptoms. The BDI-FS delivers values in the range between 0 and 21. The highest value indicates a higher depressiveness. The questionnaire shows good internal consistency (Cronbach's α = 0.84) and convergent validity with the PHQ-9 of r = 0.67, including validation based on a representative German sample (n = 2467). The manual de nes categories as "minimal" (0 to 3 points), "mild" (4 to 8 points), "moderate" (9 to 12 points) and "severe" (13 to 21 points).

MR data acquisition
All participants were examined using a 1

Imaging processing
The data obtained from the 1 H-MRS were evaluated by two blinded researchers (a physicist and a neuroimaging scientist). The data analysis was carried out on the selected layer with TARQUIN 4.3.1. This software was selected based on positive reports in the literature 47,48 . For the current study, three glutamine metabolites from the water-suppressed CSI were quanti ed using SE 1 H -MRS data, namely glutamate (Glu), glutamine (Gln) and glutamate + glutamine (Glx). Total creatinine (tCr) values were also calculated and used as a reference signal for the analysis.
The 1 H-MRS data obtained for the hypothalamic region could contain a different number of voxels for each participant. To solve this problem, the voxels with a signal-to-noise ratio (SNR) for the glutamate metabolites and a reference signal (tCr) of more than 3 were selected. Hypothalamic voxels with an SNR less than 3 for these metabolites were discarded. The glutamate metabolite concentrations (Glx, Glu, Gln) were calculated for these selected voxels and divided by a reference metabolite. Total creatinine (tCr) was chosen as the reference metabolite. The methodological considerations for MRS data analysis, in particular for determining the metabolite concentrations, have been published elsewhere in the literature 49 . Finally, the voxel with the highest concentration was selected for each participant and for each of the metabolites (Glx/tCr, Glu/tCr, Gln/tCr) and fed into the statistical analysis.

Statistical Analysis
The data set in this study consists of the metabolic data of the hypothalamus (Glx/tCr, Gln/tCr, and Glu/tCr) and the plasma peripheral proin ammatory cytokines (IL-6, IL-1β, TNF-, and IFN-). Both showed positively-skewed, non-Gaussian distributions with a few large values. The TARQUIN software expresses the hypothalamic metabolic data (Glx/tCr, Gln/tCr, and Glu/tCr) in ppm (parts per million), which corresponds to count data with a discrete distribution and is best represented by a Poisson distribution. Therefore, generalized mixed models were calculated using the Poisson distribution with log link-function for Glx/tCr, Glu/tCr, and Gln/tCr as dependent variables. The best model was selected according to the AIC criterion.
Differences in hypothalamic metabolites between DE and HC were investigated with the xed effects metabolite (Glx/tCr, Gln/tCr, Glu/tCr) and group (DE, HC). For analyzing differences in hypothalamic metabolites with respect to the pro-in ammatory cytokines, the data from the pro-in ammatory cytokines was dichotomized using the median split method. These dichotomized variables were individually introduced into the model as xed effects together with the group status (DE, HC). A model was calculated for each combination of a hypothalamic metabolite and a cytokine.
For all models, the variables age, BMI and gender were included as covariates to control the possible effects of these variables. Smoking behavior and medication intake explained the variance of the models only minimally, so that they were not included in the models.
Models were computed using the R software v.  Generalized mixed model marginal estimates and 95% con dence intervals for H MRS ratios of Glx/tCr, Gln/tCr, and Glu/tCr, divided into groups DE and HC. Figure 2