Inflammation is implicated in the etiology and pathophysiology of several brain pathologies (e.g., major depression [1–3], Alzheimer’s disease [4], and post-stroke depression [5]), as well as in cognitive aging [6] and mortality [7]. Specific markers of systemic inflammation such as cytokines have been identified as important mediators of neurodegenerative [8] and neuroplastic [9, 10] processes relevant to neuropsychiatric disorders. Some of these proteins (e.g., interleukin 1 beta, interleukin 6, tumor necrosis factor) play a critical role in physiological CNS processes, such as cognitive function under immunologically unchallenged conditions [2, 6].

Interleukin 6 (IL-6) is a cytokine that has demonstrated both neurodegenerative [4] and neuroprotective [11, 12] properties. For example, increased levels of IL-6 are associated with neuropsychiatric conditions, such as depression [13] and Alzheimer’s disease [14]. In addition, first studies on the association between IL-6 and brain volume suggested a role of increased serum levels of IL-6 in brain atrophy during normal aging in conjunction with other cytokines [15]. In addition, an association between IL-6 levels and decreased hippocampal gray matter volume in middle-aged adults has recently been reported [16].

However, the hypothesis that IL-6 is mainly proinflammatory and neurodegenerative is challenged [17] with results supporting that this cytokine has several anti-inflammatory and immunosuppressive activities that may play a downregulating role in inflammatory conditions [17]. In addition, IL-6 may act as a developmental neurotrophic factor [18, 19], and it has been shown to improve survival in vitro of several classes of neurons [20–22]. Moreover, it is suggested that IL-6 predominantly plays a protective role by improving survival of neurons in culture [21, 23, 24], protecting neurons from excitotoxic and ischemic insults [25–28], and promoting the growth of axons and consequently the number of synapses in a region [29–32]. Additionally, evidence shows that IL-6 may play a major role in promoting synaptic plasticity, LTP, and memory consolidation [33–35]. Furthermore, IL-6 is found to regulate survival of differentiated neurons and the development of astrocytes [36, 37]. Overall, these findings from previous studies suggest that higher IL-6 levels may play a dual role with both neurodegenerative and neuroprotective biological functions.

The current evidence in humans relies on measures of IL-6 in serum and CSF, whereas limited research on the influence of genetic variants of IL-6 on brain pathology has been published. The IL-6 gene is located on chromosome 7p21, and the GG genotype of the frequently studied IL-6 promoter −174 C/G variation relates to higher levels of IL-6 compared to the CC genotype [38]. Although this single nucleotide polymorphism (SNP) has received a lot of attention in research in aging and longevity, the findings are inconclusively showing an association between the numbers of G alleles either with increased [39] or decreased [38] longevity depending on the study design, ethnicity, lifestyle, and cultural differences. Additional single nucleotide polymorphisms (SNPs), such as rs1800796, influence IL-6 expression (G allele carriers increase IL-6 plasma levels) [40] and are influenced by the presence of other polymorphisms (e.g., rs2069833, rs2069840) at this chromosomal locus [41]. However, these other genetic variants of IL-6 have hardly been studied in brain function yet.

Further clarification of the biological role of genetic variants of IL-6 in the human brain is needed to describe its multifunctional effects. In this study, we investigate the role of the IL-6 gene in brain function and brain morphology by investigating the association between several genetic variants of interleukin 6 and brain morphology in healthy adult individuals. While this analysis is conducted in a whole-brain fashion, we expect genetic effects particularly in the hippocampus (HC) since this brain region has a critical role in normal brain function and several neuropsychiatric disorders. The HC region is a highly important structure for memory consolidation, and it has shown a strong susceptibility to stress and response to cytokines [42]. Specifically, several studies have shown that depression, post-traumatic stress disorder (PTSD), and childhood maltreatment are associated with smaller hippocampal volumes [42–44]. The role of genetic inflammatory biomarkers, such as IL-6, in these relationships is unclear.

This study aims at an improved understanding of the genetic background of the dual role of IL-6 in brain morphology and the hippocampal structure in particular. We hypothesize that IL-6 polymorphisms are related to brain gray matter volumes, specifically in the hippocampus. The analysis will inform future studies in clinical psychiatric populations on the possible role and selection of genetic variants of IL-6 for the study of hippocampal function in neuropsychiatric disorders.

Genotyping of four tagging IL-6 SNPs was carried out following published protocols applying the multiplex genotyping assay iPLEX™ for use with the MassARRAY platform [55], yielding an overall genotyping completion rate of 98.9% [4/297 genotyping failures for rs1800795 and rs2069833 (99.0%), 5/297 for rs1800796 and rs2069840 (98.7%)]. Genotypes were determined by investigators blinded for the study.

Hardy-Weinberg equilibrium was fulfilled for all four SNPs, according to the program Finetti provided as an online source (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl; Wienker TF and Strom TM) (exact test: rs1800795, p = 0.33; rs1800796, p = 1; rs2069833, p = 0.33; rs2069840, p = 0.80).

rs1800795 (−174 C/G): The whole-brain analysis yielded a strong main effect of genotype [43 CC vs. 150 CG vs. 100 GG), x = 24, y = −10, z = −15; F(2,286) = 8.54, p_uncorrected = 0.0002; p_{AlphaSim-corrected} = 0.002; cluster size k = 577, effect size f = 0.23 (Figure 1)]. According to the automated anatomical labeling, this cluster was located in the right hippocampus head, extending to the parahippocampal gyrus and the dorsal parts of the right amygdala. The Anatomy toolbox yielded similar localizations (peak effect was found in the cornu ammonis and subiculum area, extending to the laterobasal amygdala). There were no other areas in the entire brain surviving our corrected statistical threshold. Repeating this analysis with smoothing kernels of 6 mm or 10 mm still would yield significant findings.

According to post-hoc t-contrasts, subjects homozygous for the G-allele had significantly larger hippocampal gray matter volumes compared to heterozygous subjects, x = 24, y = −10, z = −15; t(286) = 3.88, p_uncorrected < 0.0001; p_{AlphaSim-corrected} < 0.001; cluster size k = 1210. Again, there were no other brain regions surviving the statistical threshold. However, there were no significant differences between heterozygous subjects and CC carriers in this model. Nonetheless, testing for allele-dose effects via regressing the number of rs1800795 (−174 C/G) G-alleles (0, 1, 2) on whole brain gray matter volume (again, including age, gender, education, and scanner gradient system as nuisance regressors) also yielded a significant cluster at a similar location in the right hippocampus, x = 21, y = −12, z = −15; t(287) = 3.82, p_uncorrected < 0.0001; p_{AlphaSim-corrected} = 0.001; cluster size k = 853, r = 0.22 (see Figure 1).

We further checked for interactions of the rs1800795 genotype and age as well as gender by modeling the interaction term in the three-group ANOVA model and the allele-dose regression. However, none of the interactions reached even a trend level of significance. Thus, the observed genotype effect on hippocampal gray matter volumes was comparable in men and women, and found across the entire age range.

rs1800796, rs2069833, rs2069840: No significant effects of these SNPs on hippocampus morphometry could be discerned in the whole-brain analysis.

This imaging genetics study investigated the association between the IL-6 gene and brain morphology in a large cohort of healthy adult participants in a whole-brain analysis approach. Carriers of the G-allele of the IL-6 genetic variant rs1800795 (−174 C/G) showed a significant association with larger hippocampal volumes on the right side in healthy subjects. This genotype effect was remarkably specific to the hippocampus, with no other structure surviving our statistical threshold corrected for the entire brain. The findings are suggestive of a neuroprotective role of the IL-6 gene [rs1800795 (−174 C/G)] on hippocampal morphology. The IL6 genotype effect was found lateralized to the right. However, at a more lenient uncorrected statistical threshold, a similar genotype effect in the same direction could also be detected in the left hippocampus (p = 0.007, uncorrected, in the allele-dose model). Therefore, we discuss the observed effects for the hippocampus in general. The other investigated three SNPs showed no significant association with gray matter volume in our study. Since the SNPs 2069840 has been related to reduced IL-6 plasma levels, the lack of association in our study can be interpreted as consistent with the assumption that reduced plasma levels do not exert neuroplastic, neuroproliferative, or neuroprotective effects. In contrast, the marker rs1800796 showed no association with gray matter volume in our study, although, in a previous study, the G allele of this SNP has also been associated with higher IL-6 plasma levels [40]. Since these findings were derived from a clinical cohort of patients with diabetic nephropathy without data on brain morphometry, a direct comparison with our study is precluded.

Our study shows the strongest association between the IL-6 genetic variant and HC volume, which has a number of critical functions under healthy and pathological conditions. It is part of a brain network including the dorsomedial and dorsolateral prefrontal cortex, the anterior cingulate cortex, and the amygdala dysregulated in major depression [61]. The HC is central to memory impairment, as seen in non-clinical samples [62] as well as in MDD [63]. Because the HC is a highly stress-sensitive brain region [64] and stress (psychological or psychosocial stress) is related to structural changes in the HC [65–67], atrophy of the HC has been described in imaging studies as a pathological neurobiological feature of depression associated with stress [68]. A meta-analysis of hippocampal volumes in patients with MDD confirmed that patients had hippocampal volumes approximately 4–6% smaller than matched control subjects in the left and right HC [69, 70].

Although the possible role of IL-6 in brain morphology has not been extensively studied yet, our findings are in contrast with previous reports. These show associations between increased IL-6 plasma levels and reduced hippocampal volume in a relatively small study (N = 76) of middle-aged, relatively healthy individuals [16] in a study on first-episode psychosis [71], and in two studies investigating various brain areas and total brain volume, respectively, during aging [72, 73]. Except in one study in relatively healthy individuals, these previous studies investigated individuals with underlying neuropsychiatric conditions. Variation in findings between these studies may be due to other methodological differences, such as the location of gray matter volume changes. While volume changes were located in the left HC in the study by Marsland et al. [16], and in various brain areas and total brain volume in the above-mentioned studies on aging [72, 73], our results were specific to the HC. Another important difference between studies is the biological model of IL-6 effects in the brain. While those previous studies explain their findings using an inflammatory model in which it is proposed that IL-6 plays a proinflammatory role, the explanation of our study builds on the proven anti-inflammatory and immunosuppressive effects of IL-6 according to the well-established dual role of IL-6 [74]. A possible mechanistic explanation to support our finding that IL-6 was associated with increased HC volumes relates to the previously reported neuroproliferative effects of IL-6. For example, it has been shown that cytokines, including IL-6, despite being large molecules not freely passing through the blood–brain barrier, can enter the brain via various pathways (humoral, cellular, neural) [75] to exert their biological effects in the brain even under physiological conditions. More specifically, it has been shown that IL-6 primarily exerts its biological effects through a hexameric receptor ligand complex including the gp130 receptor [11] and the IL-6 receptor [76]. Distinct regions of gp130 activate specific signal-transduction pathways, such as the Janus kinase (JAK) signal transducer and activator of transcription (STAT), mitogen-activated protein kinase (MAPK)/cAMP responsive element-binding protein (CREB), Ras-MAPK, and PI-3 kinase (for review [77]). These pathways are related to neural plasticity by their ability to induce processes of neurogenesis, such as gliogenesis, neuronal differentiation, cAMP response element binding (cAMP), neural progenitor proliferation, and neuronal survival [77–79], and to enhance synaptic plasticity, LTP, and memory consolidation [33–35]. Through activation of these pathways, IL-6 has the ability to exert neuroprotective and neuroproliferative effects. In addition, IL-6 has been found to regulate survival of differentiated neurons and the development of astrocytes [36, 37]. Some in-vitro studies show IL-6 release by activated microglia is a key inhibitor of neurogenesis by approximately 50%; others show IL-6 promoting differentiation of neural stem cells (NSCs) [10, 80–82]. NSCs derived from rodent spinal cord show that IL-6 induces NSC proliferation via the JAK2/STAT3 and MAPK pathways [83]. Supporting a role of IL-6 in neuroproliferation is an in-vivo study showing that IL-6 knockout mice have reduced proliferating NSCs specifically in the HC, hence underlining the importance of IL-6 in cell proliferation and cell survival [84].

Despite mechanistic evidence and studies in humans for both pro- and anti-inflammatory effects of IL-6 in the brain, the role of IL-6 in the hippocampus remains to be clarified. Specifically, it is questionable that increased levels of IL-6 have purely degenerative effects since proliferative effects of IL-6 in the HC were demonstrated in an exercise study in mice: a wheel-running study in mice over 16 weeks showed that exercise increased IL-6 levels in the HC, whereas other cytokines such as TNF and IL-1ra decreased during exercise [85]. These results suggest that an upregulation of IL-6 could have anti-inflammatory effects and be neuroprotective in the cytokine milieu of the HC, and thereby IL-6 may buffer cognitive decline through exercise-induced changes in the HC milieu.

Translating these findings into a human study, one could argue that peripherally increased IL-6 levels could be interpreted as an anti-inflammatory activity rather than a proinflammatory state. Hence, previously observed correlations between increased plasma levels of IL-6 and decreased HC volumes could alternatively be interpreted as an anti-inflammatory response of IL-6 to other increased cytokines such as TNF and IL-1beta. Indeed, both cytokines have previously been shown to be associated with hippocampal volumes (TNF) [86] and with increased white matter hyperintensities (IL-1beta) [87] in healthy individuals. In such a case, IL-6 would only be a marker of a global inflammatory process, and reduced brain volume might primarily be induced by proinflammatory cytokines such as TNF and IL1-beta.

In light of these studies suggesting effects of IL-6 on various mechanisms subserving neuroproliferation and assuming that the carriers of the G-allele of the IL-6 polymorphism rs1800795 (−174 C/G) in our sample have increased IL-6 levels as previously reported, it can be suggested that in our imaging study, this particular SNP might exert neuroprotective effects on the HC via increased IL-6 levels, hence the observed increased gray matter volume.

Our study has strengths and limitations. We were able to employ the genetic analysis in a large imaging sample using a cohort of carefully selected and well-characterized healthy individuals. For future studies, clinical measures such as hypertension or BMI could be useful covariates when investigating genetic inflammatory biomarkers such as IL-6; however, the relevance of hypertension might be of greater relevance in clinical samples than in our healthy cohort. Our discussion is based on the assumption that larger gray matter values in the hippocampus correspond to better function. Albeit reduced hippocampal volumes are consistently found in neuropsychiatric disorders, the relation of volume and function remains to be established more firmly. Although no protein data were available to validate the well-described upregulation of IL-6 by the SNP rs1800795 (−174 C/G), our study is the first genetic study investigating the association between the IL-6 gene and brain morphometry, and the HC in particular. Future genetic imaging studies would benefit from additional protein data. Moreover, a clinical control group with a psychiatric disorder such as depression or psychosis/schizophrenia might add knowledge on the dual role of the IL-6 gene in health and disease states. Another important consideration for interpreting these results is related to the lack of a cutoff of level of IL-6 defining normal, increased, and decreased peripheral IL-6 levels, limiting the interpretation of physiological and pathological brain conditions. The LD indices indicate complete LD (D’ = 1) for the correlation of all four marker combinations (except D’ = 0.993 for rs1800795 x rs2069833), which indicates that the reported findings are not explained by relevant SNP correlations.

This imaging genetic study suggests the IL-6 genetic variant rs1800795 (−174 C/G) as a biomarker of hippocampal morphometry. This genetic variant may exert neuroprotective effects on hippocampal volume in healthy individuals. Replication in independent and clinical samples is warranted.