Alterations of the serum N-glycan profile in female patients with Major Depressive Disorder

Journal of Affective Disorders 234 (2018) 139–147

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research paper

Alterations of the serum N-glycan profile in female patients with Major Depressive Disorder

T

⁎

Christina Boecka, ,1, Sophia Pfistera,1, Alexander Bürkleb, Valerie Vanhoorenc,d, Claude Libertc,d, Juan Salinas-Manriquee, Detlef E. Dietrichf,g, Iris-Tatjana Kolassaa, Alexander Karabatsiakisa a

Clinical & Biological Psychology, Institute of Psychology and Education, Ulm University, Ulm, Germany Molecular Toxicology, Department of Biology, University of Konstanz, Box 628, Germany c VIB Center for Inflammation Research, Ghent, Belgium d Department of Biomedical Molecular Biology, Ghent University, Belgium e AMEOS Clinic for Psychiatry and Psychotherapy Hildesheim, Hildesheim, Germany f Burghof-Klinik, Rinteln, Germany g Department of Mental Health, Hannover Medical School, Hannover, Germany b

A R T I C LE I N FO

A B S T R A C T

Keywords: Childhood sexual abuse Inflammation Major Depressive Disorder Serum N-glycan profile DSA-FACE

Background: Glycans are short chains of saccharides linked to glycoproteins that are known to be involved in a wide range of inflammatory processes. As depression has been consistently associated with chronic low-grade inflammation, we asked whether patients with Major Depressive Disorder show alterations in the N-glycosylation pattern of serum proteins that might be linked to associated changes in inflammatory processes. Methods: In a study cohort of 21 female patients with an acute depressive episode and 21 non-depressed female control subjects aged between 50 and 69 years, we analyzed the serum N-glycan profile by DNA Sequencer Adapted-Fluorophore Assisted Carbohydrate Electrophoresis (DSA-FACE) and assessed the serum levels of interleukin (IL)− 6, tumor necrosis factor (TNF)-α and C-reactive protein (CRP) by chemiluminescence immunoassays and nephelometry. Results: Compared to controls, MDD patients showed significant differences in the serum levels of several Nglycan structures. Alterations in the serum N-glycan profile were associated with depressive symptom severity and exploratory analyses revealed that they were most pronounced in MDD patients with a history of childhood sexual abuse. Furthermore, MDD patients showed higher levels of IL-6 and a trend for higher CRP levels, which were also associated with similar alterations in the serum N-glycan profile as those characteristic for MDD patients. Limitations: The relatively small sample size and the presence of potential confounders (e.g., BMI, smoking, medication). Conclusion: The results offer the first evidence that specific differences in the N-glycosylation pattern of serum proteins constitute a so far unrecognized level of biological alterations that might be involved in the immune changes associated with MDD.

1. Introduction Depression is a severe disease that affects more than 300 million individuals worldwide (World Health Organization, 2017). On a psychological level, experiences of chronic and traumatic stressors constitute major risk factors for the development of depression. In particular traumatic experiences during early life (e.g., childhood sexual abuse) seem to influence the course of the disease: individuals with a history of adverse early life experiences are not only at a higher risk to

⁎

1

develop depression, but are also twice as likely to be resistant to antidepressant treatment (Nelson et al., 2017). Besides the typical depressive phenotype (e.g., fatigue, depressed mood, and difficulties concentrating), depressed individuals also often show increased behavioral risk patterns such as smoking, reduced physical activity, and a higher body-mass-index, which may be one reason for the higher frequency of physical diseases observed among depressed patients (De Hert et al., 2011). On a biological level, depression itself has, however, also been consistently associated with immune alterations and a phenotype of

Correspondence to: Clinical & Biological Psychology, Institute for Psychology and Education, Ulm University, Albert-Einstein-Allee 47, 89081 Ulm, Germany. E-mail addresses: [email protected], [email protected] (C. Boeck). These authors contributed equally to this work.

https://doi.org/10.1016/j.jad.2018.02.082 Received 20 October 2017; Received in revised form 1 February 2018; Accepted 25 February 2018 Available online 27 February 2018 0165-0327/ © 2018 Elsevier B.V. All rights reserved.

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

account for the known gender-specific differences in the serum Nglycan profile (Ding et al., 2011), we recruited only female MDD patients and controls. The MDD group consisted of individuals with a diagnosis of Major Depressive Disorder according to DSM-IV (American Psychiatric Association, 2000) who received an inpatient treatment at the AMEOS Clinic in Hildesheim, Germany. Acute depressive symptom severity was assessed by the Beck Depression Inventory II (BDI-II, selfreport) (Hautzinger et al., 2006). Controls were recruited via public advertisement by posters in public institutions (e.g., supermarkets, gyms). Prior to study participation, the study personnel asked the control subjects whether they had experienced any depressive disorders in their lifetime as well as about a history of depression in two generations before (parents, grandparents). The healthy control group also completed the BDI-II to exclude any individuals with acute symptoms of depression. Study participants further completed the Essener Trauma Inventory (ETI, self-report) (Tagay et al., 2007) for diagnosis of comorbid PTSD. Data on the ETI was missing for two MDD patients. Based on the ETI, the subjects were categorized as positive for a history of childhood sexual abuse if they reported to have personally experienced one of the following two items: 1) “Sexual abuse by a family member during childhood or adolescence”, 2) “Sexual abuse by a stranger during childhood or adolescence”. Furthermore, we assessed covariates with known influences on inflammation, such as smoking, physical activity (both measured by a dichotomous item with the answer categories “yes” and “no” in self-report), and the body mass index (BMI). Exclusion criteria for both groups were any history of severe head trauma, neuro-psychiatric diseases including Parkinson's disease, Alzheimer's disease, and schizophrenia and any other clinically relevant neurologic or psychiatric disorders. Moreover, subjects with anemia, severe immune alterations, autoimmune diseases, and cancer, as well as subjects reporting the current intake of medication with known effects on the immune system (e.g., immunosuppressors, cytostatic agents, recent vaccines) were excluded from the study. Additionally, signs for a clinically relevant acute inflammatory response led to the exclusion from this study, which was the case in one MDD patient (IL6: 13.0 ng/l, CRP: 36.3 mg/l, TNF-α: 8.6 ng/l) and one control subject (IL6: 17 ng/l, CRP: 41.7 mg/l, TNF-α: 12.0 ng/l). Therefore, the final study cohort for all statistical analyses consisted of N= 21 MDD patients and N= 21 control subjects.

chronic low-grade inflammation (Berk et al., 2013) as indicated by increased serum levels of the pro-inflammatory cytokines interleukin (IL)− 6, tumor necrosis factor (TNF)-α, and C-reactive protein (CRP) (Dinan, 2009; Miller et al., 2009; Wolkowitz et al., 2010). Almost all molecular processes that are critical for physiological immune functioning (e.g., cell-cell communication, signal transduction, host-pathogen interactions) involve glycoproteins (Ding et al., 2011; Comelli et al., 2006), i.e. proteins which are post-translationally modified through the controlled enzymatic attachment of complex oligosaccharide chains (glycans) to the oxygen atom of serine or threonine amino acid (O-glycosylation) or the nitrogen atom of asparagine amino acid (N-glycosylation) in the protein (Lauc et al., 2014). The attachment of different glycans to the same protein (resulting in different glycoforms) influences its biological function, as it has an impact on the conformation, solubility, antigenicity, and the recognition of the associated glycoprotein (Varki et al., 2009). Alterations in the protein glycosylation pattern thereby allow the organism to fastly adapt its biological functions to changing environments (Lauc et al., 2014). Besides environmental factors, protein glycosylation is further influenced by genetic factors such as gender (Ding et al., 2011) and by the physiological and biochemical status of the individual (Gornik and Lauc, 2008). For instance, inflammatory cytokines can regulate the expression of glycan modifying enzymes such as α1,3-fucosyl-transferase, α2,3sialyl-transferase, and N-acetylglucosaminyl-transferase (Higai et al., 2005) and thereby influence the N-glycosylation pattern of immune proteins. Even minor modifications in the glycosylation pattern of immunoglobulin G (IgG), the most abundant glycoprotein in human serum (Debruyne et al., 2010), are, vice versa, associated with increased cytokine activity (Gornik and Lauc, 2008), which can further activate the immune system and exacerbate the pro-inflammatory status during an inflammatory response (Dall’Olio et al., 2013). Accordingly, diverse inflammatory diseases were already associated with specific alterations in the protein glycosylation profile (Gornik and Lauc, 2008). Furthermore, physiological aging and age-related diseases, namely dementia and the progeroid disease Cockayne syndrome (Vanhooren et al., 2010), have also been associated with specific changes in the N-glycan profile, in particular an increase in an agalactosylated core-α-1,6-fucosylated biantennary N-glycan (NG0A2F) and a decrease in a bigalactosylated, core-α-1,6-fucosylted biantennary N-glycan (NA2F). These findings lead to the development of the GlycoAge Test (=log[NG0A2F/ NA2F]) as an estimate for the biological age of an individual (Vanhooren et al., 2010). The same alterations have recently also been reported for individuals suffering from post-traumatic stress disorder (PTSD), indicating a premature aging phenotype associated with traumatic stress exposure (Moreno-Villanueva et al., 2013). Building on this knowledge, we hypothesized that the serum Nglycan profile would significantly differ in depressed individuals compared to age-matched controls. Furthermore, we hypothesized that alterations in the N-glycosylation pattern were related to an increased inflammatory status, and therefore tested for an association with the serum levels of IL-6, TNF-α, and CRP. Given the impact of adverse early life experiences on MDD disease outcomes (e.g., Nelson et al., 2017), we additionally run exploratory analyses to assess the influence of adverse early life experiences (in particular childhood sexual abuse) on the biological changes associated with MDD.

2.2. Analysis of the serum N-glycosylation profile Whole blood was collected by venous puncture into serum collection tubes (Sarstedt, Nümbrecht, Germany) between 7 a.m. and 2 p.m. Precise data on daytime of blood collection was available for N= 35 study participants. A group comparison showed no difference in mean time of blood collection between MDD patients and control subjects (Table 1). Immediately after blood sampling, whole blood samples were centrifuged at 3000g for 10 min at 4 °C for serum collection. Serum aliquots of 250 µl were immediately frozen and stored at − 80 °C until batch-wise analysis. Frozen serum samples were shipped on dry ice to the VIB Center for Inflammation Research (Ghent, Belgium). Using an ultra-sensitive technique for the quantification and sequencing of Nlinked glycans by “DNA Sequencer Adapted-Fluorophore Assisted Carbohydrate Electrophoresis” (DSA-FACE), the N-glycosylation profile of circulating serum proteins was analyzed in batches as described by Vanhooren et al. (2008). In short, following denaturation of the serum proteins, N-glycans were separated from the proteins through enzymatic digestion. The free N-glycans were then fluorescently labeled, desialylated, and separated on a DNA sequencer (3130 Genetic Analyzer, Applied Biosystems, Foster City, CA, USA). The nine most prominent peaks in the human serum N-glycan profile (Liu et al., 2007) were quantified and the peak sizes, which correspond to the relative concentration of the respective oligosaccharide structures, were normalized to total signal intensity. For analyses, we calculated the GlycoAge Test as the log ratio of peak 1 to peak 6 [log10(peak1/peak6)]

2. Materials and methods 2.1. Study participants In total, 44 individuals – 22 women (age range: 50–69 years) with Major Depressive Disorder (MDD group) and 22 age-matched control subjects – participated in the study. All study participants provided written informed consent. The study was approved by the Ethics Committee of the Hanover Medical School and was conducted in line with the Declaration of Helsinki (World Medical Association, 2013). To 140

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

Table 1 Clinical characteristics of N= 21 control subjects and N= 21 women with MDD. Variables Age, years Mean (SD) Range Gender, Female n (%) BDI-II, sum score Mean (SD) Range Childhood sexual abuse n (%) BMI, kg/m2 Mean (SD) Range Smoker, „yes“ n (%) Physical Activity, „yes“ n (%) Medication, „yes“ n (%) Blood collection time (h)b Mean (SD) Range

Controls

MDD patients

Statistics

p-value

57.48 (5.65) 50 − 69

58.67 (6.31) 50 − 69

W=200.5

.62

21 (100%)

21 (100%)

2.14 (2.24) 0−7

23.57 (7.35) 7 − 42

t(21.8)= −9.13

< .001

4 (19%)

7 (33%)a

χ2(1)=1.58

.21

24.53 (2.99) 19.89 − 30.47

28.68 (7.35) 19.76 − 47.03

W=144.5

.06

3 (14%)

11 (52%)

χ2(1)=6.86

.01

18 (86%)

13 (62%)

χ2(1)=3.08

.08

13 (62%)

21 (100%)

χ (1)=9.88

.002

11.46 (2.03) 8.00 – 14.15

10.68 (2.10) 7.26 – 14.05

t(33)=1.12

.27

2

Abbreviations: MDD, Major Depressive Disorder; BDI-II, Beck depression inventory II; BMI, Body mass index; SD, standard deviation. Bold p-values indicate significance on an alpha level of .05 (two-sided). a Data on the Essener Trauma Inventar (ETI) was missing for two MDD patients. b Time of blood sampling was calculated as hours from midnight. Data missing from 8 study participants: N(Controls)= 18, N(MDD patients)= 17.

Inflammation markers were assessed in blood serum aliquots by standard assays established in clinical routine (IL-6 and TNF-α by chemiluminescence immunoassays; CRP by nephelometry) and all measurements were performed by accredited clinical laboratories (IL-6 and TNF-α: MVZ Labor Dr. Limbach & Kollegen GbR, Heidelberg, Germany; CRP: MVZ Labor Dr. Hauß and colleagues, Lehrte, Germany). For analyses, frozen serum samples were shipped on dry ice and serum cytokine and CRP levels were analyzed in batches. The serum levels of inflammation markers were particularly low in the control group, yielding values below the lower detection limits for each of the analytes (lower limits of detection: 2 ng/l for IL-6; 4 ng/l for TNF-α; .3 mg/l for CRP).

therefore reanalyzed with permutation tests (Anderson and Legendre, 1999; Freedman and Lane, 1983). Correlations between the level of distinct N-glycan peaks and psychological data were analyzed using Pearson product-moment correlation coefficients and Kendall τ rank correlation coefficients where appropriate. To analyze the influence of a history of childhood sexual abuse, subjects were categorized into four groups (controls without childhood sexual abuse; controls with childhood sexual abuse; MDD patients without childhood sexual abuse; MDD patients with childhood sexual abuse) and ANCOVAs (Tukey post hoc tests) were calculated for the biological variables of interest. As all serum inflammation levels included values below the detection limit, group comparisons and regression analyses of IL-6, CRP, and TNF-α were calculated by Tobit regression on ranks as described by Ballenberger et al. (2012) with the R package “censReg” (Henningsen, 2017). We hypothesized that individuals with an acute episode of MDD would show higher levels of pro-inflammatory markers compared to control subjects, therefore one-tailed tests were applied for group comparisons of IL-6, TNF-α, and CRP levels.

2.4. Statistical analysis

3. Results

Statistical analyses were performed with R 3.3.2 (R Development Core Team, 2016). The alpha level was .05. As it is known that the serum N-glycan profile differs with age (Vanhooren et al., 2010), BMI (Knežević et al., 2010), smoking (Knežević et al., 2010), and might be additionally influenced by the time of blood collection and physical activity, we assessed the association between these potential confounders and the serum N-glycan levels across the whole study cohort by Kendall τ rank correlations or Pearson product-moment correlation for continuous variables and Wilcoxon rank-sum tests for group comparisons as preparatory analyses. Group differences in N-glycan profiles between depressed patients and control subjects were then calculated using ANCOVAs, accounting for potential confounders by separate inclusion as covariates into the statistical models in a second step. The residuals of all models were tested for deviation from normal distribution (Shapiro-Wilk test) and homogeneity of variances (Levene's test). The residuals of peak 1, peak 2, peak 5, peak 6, peak 9 and the level of agalactosylated N-glycans were not normally distributed and

3.1. Clinical characteristics

(Vanhooren et al., 2010), the total level of agalactosylated N-glycans (sum of peak 1, peak 2, peak 3 and peak 4), the total level of bigalactosylated N-glycans (sum of peak 5, peak 6 and peak 7) and the total level of triantennary N-glycans (sum of peak 8 and peak 9). 2.3. Levels of circulating inflammation markers

Clinical characteristics of the study cohort are summarized in Table 1. Of the depressed patients, 33% presented a first episode of depression, while 67% were diagnosed with recurrent MDD. Additionally, three MDD patients were diagnosed with comorbid PTSD based on the ETI. As these individuals did not differ significantly in any of the variables of interest from the remaining MDD patients, they were retained in all statistical analyses. Groups differed in current smoking status with a higher percentage of smokers in the MDD group compared to the control group. Furthermore, depressed individuals showed a trend for an increased BMI and reduced physical activity (Table 1). 76% of the depressed individuals reported the current intake of antidepressant and antipsychotic medication (as an adjunct therapy for antidepressant treatment), while none of the control subjects reported the intake of any psychopharmacological drugs. Additionally, all individuals in the MDD group and 57% of the controls reported the intake 141

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

3.3. N-Glycan profiles in depressed individuals

Table 2 Serum N-glycan levels of N= 21 control subjects and N= 21 depressed subjects. Serum N-glycansa N-glycan

Controls Structure

M (SD)

MDD patients M (SD)

ANOVA analyses revealed that the level of two agalacto core-α-1,6fucosylated biantennary glycans (NG1A2F) represented as peak 3 and peak 4, as well as the total level of agalactosylated N-glycans were significantly reduced in the MDD group compared to the control group (Table 2). In contrast, the level of a non-fucosylated biantennary glycan (NA2) represented as peak 5, the level of peak 9 representing the branching α-1,3-fucosylated triantennary glycan NA3FB, and the total level of triantennary N-glycans were significantly elevated among MDD patients compared to the control group (Table 2). There were no significant group differences in any other N-glycan peak (Table 2). Separate inclusion of age as a covariate had no influence on any of the results.

F(1,40) p-value

NG0A2F (Peak 1)

10.9 (4.2) 9.3 (3.0)

2.19

.15

NG0A2FB (Peak 2)

2.3 (1.8)

1.8 (.7)

1.28

.26

NG1A2F (Peak 3)

8.1 (1.5)

6.7 (1.3)

10.7

.002

NG1A2F (Peak 4)

5.6 (.8)

5.0 (.8)

6.60

.01

NA2 (Peak 5)

37.8 (4.9) 41.9 (4.0) 9.10

.005

NA2F (Peak 6)

18.3 (2.5) 17.0 (2.8) 2.50

.12

NA2FB (Peak 7)

6.6 (1.3)

.77

3.4. Serum N-glycan levels are associated with depressive symptom severity

6.5 (1.3)

.09

Depressive symptom severity, as assessed with the BDI-II sum score, correlated negatively with the level of peak 3 (r = −.38, p = .01) and as a trend with the total level of agaloctosylated N-glycans (τ = −.21, p = .06), while it showed a positive correlation with the level of peak 5 (τ = .30, p = .007), peak 9 (τ = .26, p = .02), and as a trend with the total level of triantennary N-glycans (r = .29, p = .06) (Fig. 1). The results did not change if one outlier in peak 5 (> 3 SD from the mean) was excluded. Peak 4 was also negatively associated with depressive symptom severity (r = −.15, p = .35), this correlation did, however, not reach statistical significance. 3.5. No changes in the GlycoAge Test in depressed individuals

NA3 (Peak 8)

7.2 (1.9)

7.7 (1.7)

.77

.39

NA3FB (Peak 9)

1.9 (.9)

2.7 (1.2)

6.38

.02

With a mean value of − .28 (SD = .20) for depressed women and a mean value of − .25 (SD= .20) for female control subjects, the two groups did not differ significantly in the GlycoAge Test (t[40]= .55, p = .59). Chronological age was, however, the best predictor for the value in the GlycoAge Test in a linear model in the total cohort (b= .013, standardized β = .41, t = 2.81, p = .008). 3.6. Changes in inflammatory markers in depressed individuals

Total agalactosylated N-glycans Total bigalactosylated N-glycans Total triantennary N-glycans

26.9 (6.3) 22.7 (4.4) 6.21 62.7 (5.2) 65.4 (4.0) 3.68 9.1 (1.7) 10.4 (1.6) 6.55

Preparatory analyses to assess the influence of potential confounders on serum inflammation levels revealed no associations between age, smoking status, physical activity, the time of blood collection and the serum levels of IL-6, CRP, and TNF-α across the whole study cohort. The BMI showed, however, a positive correlation with IL6 (b= 1.07, t = 2.72, p = .007) and CRP (b= .80, t = 4.46, p < .001), but not TNF-α levels (b= .23, t = 1.38, p = .17). The MDD group presented increased serum IL-6 (F[1,39]= 3.08, p = .04) and CRP levels (F[1,39]= 2.62, p = .06) compared to the control group (see Supplementary Table 2 for a detailed overview of inflammation levels). The serum level of TNF-α was not significantly altered between cotnrols and depressed individuals (F[1,39]= .02, p = .44).

.02 .06 .01

Abbreviations: MDD, Major Depressive Disorder; M, Mean; SD, standard deviation. Bold p-values indicate significance on an alpha level of .05 (two-sided). Symbol legend: filled square = N-acetylglucosamine (Glc-Nac); filled circle = mannose; unfilled circle = galactose; filled triangle = fucose. a Given as percent of total serum N-glycan levels.

of other demand medications mainly related to age-associated somatic conditions (see Supplementary Table 1 for detailed overview).

3.7. Serum N-glycan levels are associated with levels of inflammatory markers

3.2. Preparatory analyses assessing the influence of potential confounders on serum N-glycan levels

As graphically illustrated in Fig. 2, serum IL-6 levels were negatively associated with the level of peak 3 (b= −1.02, t = −2.48, p = .01) and peak 4 (b= −1.16, t = −2.74, p = .006). Furthermore, there was a positive association between serum IL-6 levels and peak 9 (b= 1.33, t = 3.15, p = .002). Regression analyses between IL-6 levels and the total level of agalactosylated N-glycans (b= −.61, t = −1.54, p = .12), peak 5 (b= .41, t = 1.03, p = .30), and the total level of triantennary glycans (b= .53, t = 1.36, p = .18) did not yield any significant associations. Serum CRP levels were as a trend also negatively associated with peak 3 (b= −.35, t = −1.62, p = .10) and significantly positively

Across the whole study cohort, serum N-glycan levels were not associated with BMI, time of blood collection, or physical activity. Smoking status only showed a significant effect on the level of glycan peak 8 (t[40]= 2.43, p = .02) with reduced levels among smokers, while there were no differences between smokers and non-smokers in any other glycan peak. Age was correlated with several N-glycan peaks (peak 1: τ = .28, p = .01; peak 2: τ = .24, p = .03; peak 6: τ = −.24, p = .02; agalactosylated N-glycans: τ = .23, p = .04; bigalactosylated N-glycans: τ = −.22, p = .04) and was therefore considered as a covariate in all subsequent analyses. 142

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

Fig. 1. The level of two single agalacto core-α-1,6-fucosylated biantennary glycans (NG1A2F) represented by (A) N-glycan peak 3 and (B) N-glycan peak 4, and (C) the total level of agalactosylated N-glycans were negatively correlated with the BDI-II sum score as a measure for depressive symptom severity, while the level of (D) one non-fucosylated bigalacto biantennary glycan (NA2) represented by N-glycan peak 5, (E) one branching α-1,3- fucosylated triantennary glycan (NA3FB) represented by N-glycan peak 9, and (F) the total level of triantennary N-glycans were positively correlated with the BDI-II sum score in N= 21 patients with Major Depressive Disorder (MDD) and N= 21 controls.

= 1.40, p = .18). The same was also true for controls with and without childhood sexual abuse (W= 38, p = .75). Within the group of nondepressed controls, only four individuals reported a history of childhood sexual abuse. Due to the small sample size and as group comparisons between controls with and without childhood sexual abuse revealed no significant differences in any of the sociodemographic or biological parameters (i.e., serum N-glycans and inflammation levels), all control subjects were considered as one group in the follow-up analyses. Group comparisons between controls, MDD patients without childhood sexual abuse and MDD patients with childhood sexual abuse revealed a main effect of group on the serum levels of peak 3 (F[2,37] = 6.95, p = .003), peak 4 (F[2,37]= 2.85, p = .07), the total level of agalactosylated N-glycans (F[2,37]= 3.44, p = .04), peak 5 (F[2,37] = 4.69, p = .02), peak 9 (F[2,37]= 3.27, p = .05), and the total level of triantennary N-glycans (F[2,37]= 2.99, p = .06). Post-hoc analyses showed the most pronounced effects in MDD patients with a history of childhood sexual abuse compared to control subjects, while MDD patients without childhood sexual abuse were positioned in-between these two groups (Fig. 3). With regard to peripheral inflammation levels, accounting for the influence of childhood sexual abuse showed no significant effects on the serum levels of IL-6 (F[2,36]= 1.78, p = .18), CRP (F[2,36]= 1.62, p = .21), and TNF-α (F[2,36]= .35, p = .71). Replication of the analyses under exclusion of the control subjects with experiences of childhood sexual abuse produced the same results as reported above.

associated with peak 9 (b= .63, t = 3.04, p = .002) (Fig. 2). Regressions between serum CRP levels and peak 4 (b= −.23, t = −1.04, p = .30), peak 5 (b= .08, t = .35, p = .73), the total level of agalactosylated N-glycans (b= −.07, t = −.30, p = .77) and the total level of triantennary N-glycans (b= .14, t = .62, p = .54) did not reveal any significant associations. There were no significant associations between serum TNF-α levels and peak 3 (b= −.13, t = −.80, p = .42), peak 4 (b= −.20, t = −1.19, p = .23), the total level of agalactosylated N-glycans (b= −.003, t = −.02, p = .99), peak 5 (b= .15, t = .91, p = .36), peak 9 (b= .10, t = .56, p = .57), and the total level of triantennary glycans (b= −.08, t = −.47, p = .64). 3.8. Exploratory analyses accounting for the influence of childhood sexual abuse on serum N-glycan profiles and inflammation levels As MDD patients with experiences of childhood adversities seem to differ in their depressive illness from MDD patients without adverse childhood experiences (Nelson et al., 2017), we additionally run exploratory analyses to account for the influence of childhood sexual abuse on the depression-associated biological alterations observed in this study. More MDD patients than control subjects (33% vs. 19%) reported to have experienced sexual abuse in childhood or adolescence, the group difference was, however, not significant. MDD patients with childhood sexual abuse did not differ in depressive symptom severity from MDD patients without a history of childhood sexual abuse (t[17] 143

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

Fig. 2. (A) The level of N-glycan peak 3 and (B) N-glycan peak 4 were negatively correlated with the serum IL-6 level, while (C) the level of N-glycan peak 9 was positively correlated with the serum IL-6 level. Similarly, (D) the level of N-glycan peak 3 was negatively and (E) the level of N-glycan peak 9 positively correlated with the serum CRP level. N= 21 patients with Major Depressive Disorder (MDD) and N= 21 controls. Boxplots display the median and the 25th and 75th percentiles as lower and upper hinges with the whiskers extending from the hinge to the largest and smallest value at most 1.5 * the inter-quartile range of the hinge.

4. Discussion

et al., 2008). Within the context of psychological conditions, no other studies so far reported alterations in any of the N-glycan structures that we observed to be significantly altered in association with MDD. Reduced levels of NG1A2F (represented by peak 3 and peak 4) and increased levels of NA2 (peak 5) and NA3FB (peak 9) were, however, already described in individuals with hepatitis-B-related hepatocellular carcinoma (Fang et al., 2010), non-small cell lung cancer (Bartling et al., 2011), colorectal cancer (Zhao et al., 2012), and gastric cancer (Zhao et al., 2014). This overlap in study findings is not surprising, as depression is recognized as a risk factor for the initiation and progression of diverse types of cancer (Currier and Nemeroff, 2014; Reiche et al., 2004). As there is a high comorbidity between cancer and depression (Rasic et al., 2008), future studies investigating N-glycan alterations in cancer patients are warranted that also assess the presence and severity of depressive symptoms in order to investigate whether the changes observed in cancer patients might be indicative of a comorbid depression. It is, however, also possible that the specific alterations in the serum N-glycan profile are attributable to a third variable that is common to both disorders, i.e. chronic inflammation. Chronic inflammation is one of the most prominent common factors that was suggested to be a driving force in the pathophysiology of both depression and malignant diseases (Currier and Nemeroff, 2014; Reiche et al., 2004). Confirming previous results (Dinan, 2009; Miller et al., 2009; Wolkowitz et al., 2010), the group of patients with MDD presented increased serum IL-6 and CRP levels. Interestingly, higher serum IL-6

This is the first study to demonstrate alterations in the serum Nglycan profile of women with MDD compared to female, age-matched control subjects. With increasing severity of depressive symptoms, the observed alterations were more pronounced. Additionally, MDD patients with a history of childhood sexual abuse showed the strongest alterations in the serum N-glycan peaks in exploratory analyses. Depressed individuals presented higher levels in two inflammatory markers, namely IL-6 and CRP, which were also associated with an increasing BMI. Furthermore, IL-6 and CRP levels were associated with specific N-glycan profile alterations that were similar to those observed in MDD patients, pointing towards a potential link between inflammatory processes and alterations in the serum N-glycan profile. Women with a diagnosis of acute MDD presented lower levels of agalactosylated N-glycans compared to controls, which was mainly attributable to a decrease in the level of N-glycan peak 3 (NG1A2F). At the same time, the serum levels of NA2 (peak 5), NA3FB (peak 9) and the total level of triantennary N-glycans were significantly increased in the MDD compared to the control group. Interestingly, the alterations in the N-glycan peak 3, peak 5, and peak 9 were also associated with depressive symptom severity, pointing towards a dose-response relationship. The positive association between depressive symptom severity and the serum level of triantennary N-glycans is in accordance with one previous study that reported a reduction in depression-like behavior in mice lacking tri- and tetraantennary N-glycans (Soleimani 144

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

Fig. 3. Three-group-comparisons of (A) N-glycan peak 3, (B) N-glycan peak 4, (C) the total level of agalactosylated N-glycans, (D) N-glycan peak 5, (E) N-glycan peak 9, and (F) the total level of triantennary N-glycans between control subjects without childhood sexual abuse (N= 17, open circles), control subjects with childhood sexual abuse (N= 4, filled circles), MDD patients without childhood sexual abuse (N= 12, open squares), and MDD patients with a history of childhood sexual abuse (N= 7, filled squares). # p-value < .10, * p-value < .05, * * p-value < .01. Boxplots display the median and the 25th and 75th percentiles as lower and upper hinges with the whiskers extending from the hinge to the largest and smallest value at most 1.5 * the inter-quartile range of the hinge.

Exploratory analyses indicated that MDD patients with a history of childhood sexual abuse showed the most pronounced alterations in serum N-glycan levels. This finding was not attributable to a higher depressive symptom severity among individuals with childhood sexual abuse and suggests that adverse early life experiences influence the biological alterations associated with MDD. A body of evidence shows that adverse childhood experiences are associated with a broad range of biological alterations that have a long-lasting impact on the stress response and the immune system (e.g., Danese et al., 2009; Nemeroff et al., 2016), which might set the biological trajectory for an increased vulnerability to develop depression in later life and might worsen the clinical course of the disease (Nelson et al., 2017). In our study, we did not see any effects of childhood sexual abuse on the serum N-glycan profile among non-depressed control subjects. These findings should, however, be interpreted with caution as only four control subjects reported a history of childhood sexual abuse. Future studies are warranted to investigate potential differences in serum glycosylation associated with adverse childhood experiences in larger study cohorts that are free of any psychiatric disorders. Limitations of this study include the presence of potential confounders, in particular the higher prevalence of smokers and higher BMI in the MDD group. Both significant differences in BMI and a higher prevalence of smokers are known to be associated with depression (Lasser et al., 2000; De Wit et al., 2009). Therefore, exclusion of these study participants might have led to the recruitment of a clinically less representative cohort of MDD patients. Comparing the presented results

levels were associated with a decrease in NG1A2F (peak 3 and peak 4), as well as with an increase in NA3FB (peak 9), i.e., with a similar serum N-glycan profile that was characteristic for the MDD group. Higher CRP levels were also associated with decreased levels of N-glycan peak 3 and increased levels of N-glycan peak 9. Even though the complex reciprocal influence between N-glycans and inflammation is not yet fully understood, our associative results indicate that the observed changes of the N-glycosylation pattern could be functionally connected to inflammatory processes. As traumatic psychological stress has been associated with ageingspecific alterations in the serum N-glycan profile represented by the GlycoAge Test (Moreno-Villanueva et al., 2013), we hypothesized that depressed individuals might also show the same changes. However, the MDD group did not differ from the age-matched control group with respect to the GlycoAge Test. In line with the literature (Vanhooren et al., 2008, 2010), chronological age was, however, also in the present study cohort the best predictor for the GlycoAge Test. In a previous study, we showed a significant difference in the GlycoAge Test between individuals with PTSD and low-stress control subjects, which was equivalent to an advanced biological age of additional 15 years in PTSD patients (Moreno-Villanueva et al., 2013). In contrast to the present study which recruited only women, this cohort consisted mainly of male study participants. As the serum N-glycan profile differs by gender (Ding et al., 2011), it is possible that the GlycoAge Test might also be more pronounced in male MDD patients, which should be investigated in future studies. 145

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

Acknowledgements

with the alterations in serum N-glycan peaks that were described to be associated with BMI and smoking status in a community sample of N= 2000 healthy men and women (Knežević et al., 2010) showed, however, no significant overlap in the respective serum N-glycan profile changes. Additionally, across our complete study cohort, neither the BMI, nor the smoking status were associated with any of the N-glycan peaks that were related to depression. Together, these results suggest that the observed group differences in the serum N-glycan profile between depressed patients and control subjects may not be attributable to differences in smoking status or BMI. In contrast, the serum IL-6 and the CRP level were not only elevated among MDD patients, but also higher with increasing BMI. As the MDD group presented a significantly higher BMI than the control group, it was not possible to clearly differentiate between the effect of group and the effect of BMI on serum inflammation levels, which is a typical problem of covariance analyses with groups that inherently differ in particular variables (Miller and Chapman, 2001). Future studies are therefore needed to dissect the association between inflammation and glycan alterations in psychiatric and non-psychiatric cohorts and to investigate the potential mediating role of the BMI in this relation. This study is further limited by its relatively small sample size, which precluded more detailed analysis of potential depression-associated heterogeneity in the MDD group (e.g., recurrent vs. first-episode MDD, type and duration of antidepressant treatment). Additionally, as our study cohort included only women at an advanced age (age range 50–69 years), future studies are needed to investigate whether these results can be transferred to males and younger MDD patients. Besides various medications for the treatment of age-related diseases, 77% of the MDD group received some kind of antidepressant medication. While antidepressants have been described to exert potent anti-inflammatory effects (Galecki et al., 2018; Tynan et al., 2012), it is not known whether antidepressant medications may interfere with N-glycan biosynthesis. In this study, we focused on the influence of childhood sexual abuse. It is known that other types of maltreatment (e.g., physical and emotional abuse, neglect) are also associated with long-lasting biological disturbances (DeBellis and Zisk, 2014; Nemeroff, 2016). Future studies are warranted with larger study cohorts that include both men and women and account for the clinical history of the depressed patients (e.g., duration of depressive illness, treatment approaches, and adverse childhood experiences) and the influence of potential confounders to gain more insights into depression-associated alterations in the serum N-glycan profile. Furthermore, as the alterations in specific N-glycan structures correlated significantly with depressive symptom severity, it would be of high interest to investigate whether a reduction in depressive symptom load by antidepressant medication or psychotherapy is accompanied by a normalization of the serum N-glycan profile.

Christina Boeck was supported by a scholarship of the Carl Zeiss Foundation. Author's contribution A.K. designed the study together with D.E.D., I.T.K., A.B. and C.L.; J.S. conducted the psychological assessments; V.V. performed the biological experiments; C.B. and S.P. undertook the statistical analyses and interpreted the results together with A.K., I.T.K. and A.B., C.B. wrote the manuscript with support from S.P., and all authors revised and approved the final manuscript. Role of the funding source This study was financed by university resources of I.T.K., D.E.D, and C.L. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jad.2018.02.082. References American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.). Washington, DC. Anderson, M.J., Legendre, P., 1999. An empirical comparison of permutation methods for tests of partial regression coefficients in a linear model. J. Stat. Comput. Simul. 62 (3), 271–303. Ballenberger, N., Lluis, A., Von Mutius, E., Illi, S., Schaub, B., 2012. Novel statistical approaches for non-normal censored immunological data: analysis of cytokine and gene expression data. PLOS One 7 (10), e46423. Bartling, B., Vanhooren, V., Dewaele, S., Libert, C., Hofmann, H.S., Haerting, J., Nuding, S., Silber, R.E., Simm, A., Chen, C.C., 2011. Altered desialylated plasma N-glycan profile in patients with non-small cell lung carcinoma. Cancer Biomark. 10 (3, 4), 145–154. Berk, M., Williams, L.J., Jacka, F.N., O’Neil, A., Pasco, J.A., Moylan, S., Allen, N.B., Stuart, A.L., Hayley, A.C., Byrne, M.L., Maes, M., 2013. So depression is an inflammatory disease, but where does the inflammation come from? BMC Med. 11 (1), 200. Comelli, E.M., Head, S.R., Gilmartin, T., Whisenant, T., Haslam, S.M., North, S.J., Wong, N.K., Kudo, T., Narimatsu, H., Esko, J.D., Drickamer, K., Dell, A., Paulson, J.C., 2006. A focused microarray approach to functional glycomics: transcriptional regulation of the glycome. Glycobiology 16 (2), 117–131. Currier, M.B., Nemeroff, C.B., 2014. Depression as a risk factor for cancer: from pathophysiological advances to treatment implications. Annu. Rev. Med. 65, 203–221. Dall’Olio, F., Vanhooren, V., Chen, C.C., Slagboom, P.E., Wuhrer, M., Franceschi, C., 2013. N-glycomic biomarkers of biological aging and longevity: a link with inflammaging. Ageing Res. Rev. 12 (2), 685–698. Danese, A., Moffitt, T.E., Harrington, H., Milne, B.J., Polanczyk, G., Pariante, C.M., Poulton, R., Caspi, A., 2009. Adverse childhood experiences and adult risk factors for age-related disease: depression, inflammation, and clustering of metabolic risk markers. Arch. Pediatr. Adolesc. Med. 163 (12), 1135–1143. De Hert, M., Corell, C.U., Bobes, J., Cetkovich-Bakmas, M., Cohen, D., Asail, I., Detraux, J., Gautam, S., Möller, H.J., Ndetei, D.M., Newcomer, J.W., Uwakwe, R., Leucht, S., 2011. Physical illness in patients with severe mental disordesr. I. Prevalence, impact of medications and disparities in health care. World Psychiatry 10 (1), 52–77. De Wit, L.M., van Straten, A., van Herten, M., Penninx, B.W.J.H., Cuijpers, P., 2009. Depression and body mass index, a u-shaped association. BMC Public Health 9, 14. DeBellis, M.D., Zisk, A., 2014. The biological effects of childhood trauma. Child Adolesc. Psychiatr. Clin. N. Am. 23 (2), 185–222. Debruyne, E.N., Vanderschaeghe, D., Van Vlierberghe, H., Vanhecke, A., Callewaert, N., Delanghe, J.R., 2010. Diagnostic value of the hemopexin N-glycan profile in hepatocellular carcinoma patients. Clin. Chem. 56 (5), 823–831. Dinan, T.G., 2009. Inflammatory markers in depression. Curr. Opin. Psychiatry 22 (1), 32–36. Ding, N., Nie, H., Sun, X., Sun, W., Qu, Y., Liu, X., Yao, Y., Liang, X., Chen, C.C., Li, Y., 2011. Human serum N-glycan profiles are age and sex dependent. Age Ageing 40 (5), 568–575. Fang, M., Zhao, Y.P., Zhou, F.G., Lu, L.G., Qi, P., Wang, H., Zhou, K., Sun, S.H., Chen, C.Y., Gao, C.F., 2010. N-glycan based models improve diagnostic efficacies in hepatitis B virus-related hepatocellular carcinoma. Int. J. Cancer 127 (1), 148–159. Freedman, D., Lane, D., 1983. A nonstochastic interpretation of reported significance levels. J. Bus. Econ. Stat. 1 (4), 292–298. Galecki, P., Mossakowska- Wójcik, J., Talarowska, M., 2018. The anti-inflammatory mechanism of antidepressant – SSRIs, SNRIs. Prog. Neuropsychopharmacol. Biol.

5. Conclusion Mounting evidence supports the hypothesis that inflammation plays a major role in the development of depressive disorders and could also account for the high comorbidity of MDD with other, especially inflammation-associated, somatic conditions. This study provides the first evidence for alterations in the N-glycan profile of MDD patients, which was also associated with higher levels of inflammation. Exploratory analyses further suggested that a history of childhood sexual abuse is associated with long-term consequences on MDD-related N-glycan changes, which might help to explain the variance in biological alterations among depressed individuals. Although we do not yet know the functional implications of the observed findings for health and disease outcomes, the specific differences in glycosylation patterns warrant more investigation as they constitute a so far unrecognized level of biological alterations that might be associated with the immune disturbances observed in MDD patients. 146

Journal of Affective Disorders 234 (2018) 139–147

C. Boeck et al.

R-project.org/〉. Rasic, D.T., Belik, S., Bolton, J.M., Chochinov, H.M., Sareen, J., 2008. Cancer, mental disorders, suicidal ideation and attempts in a large community sample. Psycho‐Oncology 17 (7), 660–667. Reiche, E.M.V., Nunes, S.O.V., Morimoto, H.K., 2004. Stress, depression, the immune system, and cancer. Lancet Oncol. 5 (10), 617–625. Soleimani, L., Roder, J.C., Dennis, J.W., Lipina, T., 2008. Beta N-acetylglucosaminyltransferase V (Mgat5) deficiency reduces the depression-like phenotype in mice. Genes Brain Behav. 7 (3), 334–343. Tagay, S., Erim, Y., Stoelk, B., Möllering, A., Mewes, R., Senf, W., 2007. Das essener trauma-inventar (ETI) – ein screeninginstrument zur identifikation traumatischer ereignisse und posttraumatischer störungen. Psychother. Psychosom. Med. Psychol. 1, 75–89. Tynan, R.J., Weidenhofer, J., Hinwood, M., Cairns, M.J., Day, T.A., Walker, F.R., 2012. A comparative examination of the anti-inflammatory effects of SSRI and SNRI antidepressants on LPS stimulated microglia. Brain Behav. Immun. 26 (3), 469–479. Vanhooren, V., Laroy, W., Libert, C., Chen, C., 2008. N-Glycan profiling in the study of human aging. Biogerontology 9 (5), 351–356. Vanhooren, V., Dewaele, S., Libert, C., Engelborghs, S., De Deyn, P.P., Toussaint, O., Debacq-Chainiaux, F., Poulain, M., Glupczynski, Y., Franceschi, C., Jaspers, K., van der Pluijm, I., Hoeijmakers, J., Chen, C.C., 2010. Serum N-glycan profile shift during human ageing. Exp. Gerontol. 45 (10), 738–743. Varki, A., Cummings, R.D., Esko, J.D., Freeze, H.H., Stanley, P., Bertozzi, C.R., Hart, G.W., Etzler, M.E. (Eds.), 2009. Essentials of Glycobiology, 2nd edition. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. Wolkowitz, O.M., Epel, E.S., Reus, V.I., Mellon, S.H., 2010. Depression gets old fast: do stress and depression accelerate cell aging? Depress. Anxiety. 27 (4), 327–338. World Health Organization, 2017. Depression, Fact Sheet N°369. Available at: 〈http:// www.who.int/mediacentre/factsheets/fs369/en/index.html〉. World Medical Association, 2013. Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects. JAMA. 310(20), 2191–2194. Zhao, Y.P., Ruan, C.P., Wang, H., Hu, Z.Q., Fang, M., Gu, X., Ji, J., Zhao, J.Y., Gao, C.F., 2012. Identification and assessment of new biomarkers for colorectal cancer with serum N-glycan profiling. Cancer 118 (3), 639–650. Zhao, Y.P., Xu, X.Y., Fang, M., Wang, H., You, Q., Yi, C.H., Ji, J., Gu, X., Zhou, P.T., Cheng, C., Gao, C.F., 2014. Decreased core-fucosylation contributes to malignancy in gastric cancer. PloS One 9 (4), e94536.

Psychiatry 80, 291–294. Gornik, O., Lauc, G., 2008. Glycosylation of serum proteins in inflammatory diseases. Dis. Markers 25 (4–5), 267–278. Hautzinger, M., Keller, F., Kühner, C., 2006. BDI-II. Beck Depressions Inventar Revision—Manual. Harcourt Test Services, Frankfurt, Germany. Henningsen A., 2017. censReg: Censored Regression (Tobit) Models. R package version 0. 5. 〈http://CRAN.R-Project.org/package=censReg〉. Higai, K., Aoki, Y., Azuma, Y., Matsumoto, K., 2005. Glycosylation of site-specific glycans of alpha1-acid glycoprotein and alterations in acute and chronic inflammation. Biochim. Biophys. Acta 1725 (1), 128–135. Knežević, A., Gornik, O., Polašek, O., Pučić, M., Redžić, I., Novokmet, M., Rudd, P.M., Wright, A.F., Campbell, H., Rudan, I., Lauc, G., 2010. Effects of aging, body mass index, plasma lipid profiles, and smoking on human plasma N-glycans. Glycobiology 20, 959–969. Lasser, K., Boyd, J.W., Woolhandler, S., Himmelstein, D.U., McCormick, D., Bor, D.H., 2000. Smoking and mental illness. A population-based prevalence study. JAMA 284, 2606–2610. Lauc, G., Krištić, J., Zoldoš, V., 2014. Glycans – the third revolution in evolution. Front Genet. 5, 145. Liu, X.E., Desmyter, L., Gao, C.F., Laroy, W., Dewaele, S., Vanhooren, V., Wang, L., Zhuang, H., Callewaert, N., Libert, C., Contreras, R., Chen, C., 2007. N-glycomic changes in hepatocellular carcinoma patients with liver cirrhosis induced by hepatitis B virus. Hepatology 46 (5), 1426–1435. Miller, A.H., Maletic, V., Raison, C.L., 2009. Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biol. Psychiatry 65 (9), 732–741. Miller, G.A., Chapman, J.P., 2001. Misunderstanding analysis of covariance. J. Abnorm. Psychol. 110, 40. Moreno-Villanueva, M., Morath, J., Vanhooren, V., Elbert, T., Kolassa, S., Libert, C., Bürkle, A., Kolassa, I.T., 2013. N-glycosylation profiling of plasma provides evidence for accelerated physiological aging in post-traumatic stress disorder. Transl. Psychiatry 3, e320. Nelson, J., Klumparendt, A., Doebler, P., Ehring, T., 2017. Childhood maltreatment and characteristics of adult depression: a meta-analysis. Br. J. Psychiatry 210 (2), 96–104. Nemeroff, C.B., 2016. Paradise lost: the neurobiological and clinical consequences of child abuse and neglect. Neuron 89 (5), 892–909. R Development Core Team, 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL 〈https://www.

147