Human Herpesvirus 6B Greatly Increases Risk of Depression by Activating Hypothalamic-Pituitary -Adrenal Axis during Latent Phase of Infection.

Little is known about the effect of latent-phase herpesviruses on their host. Human herpesvirus 6B (HHV-6B) is one of the most ubiquitous herpesviruses, and olfactory astrocytes are one of the most important sites of its latency. Here, we identified SITH-1, an HHV-6B latent protein specifically expressed in astrocytes. Mice induced to produce SITH-1 in their olfactory astrocytes exhibited olfactory bulb apoptosis, a hyper-activated hypothalamic-pituitary-adrenal (HPA) axis and depressive symptoms. The binding of SITH-1 to the host protein calcium-modulating ligand (CAML) to form an activated complex promoted the influx of extracellular calcium. The serum antibody titers for depressive patients with respect to this activated complex were significantly higher than for normal controls (p = 1.78 × 10-15), when the antibody positive rates were 79.8% and 24.4%, respectively, and the odds ratio was 12.2. These results suggest that, in the latent phase, HHV-6B may be involved in the onset of depression.

Introduction

Herpesviruses have the property to maintain a state of latent infection in hosts throughout their lives and are an important component of the viral microbiome (virome), which occasionally increases the risk of various diseases (Virgin and Todd, 2011, Virgin et al., 2009).

Human herpesvirus 6B is a ubiquitous, neurotrophic virus that is widespread in many parts of the world, including the USA, Europe, and Japan. It causes roseola in the initial infection and afterward establishes latency. The virus is reactivated in response to immunosuppression in transplant patients and sometimes causes encephalitis (Ablashi et al., 2014).

Recent postmortem research revealed that HHV-6B and HHV-6A, a close relative of HHV-6B, had been reactivated from the latent state and had caused productive infection in the cerebellum in patients with mood disorders (major depressive disorder and bipolar disorder [BPD]) (Prusty et al., 2018). Although many aspects of the relationship between HHV-6A/B and these diseases were unknown, the authors stated the importance of conducting research on the mechanism by which latent HHV-6A/B are reactivated and infect the brain. Other past research reported detection of HHV-6B DNA in the orbital frontal cortex of patients with BPD (Conejero-Goldberg et al., 2003).

Proposed sites of HHV-6B latency are the tonsils and adenoids, and the virus has been shown to be actively shed in saliva. The shed virus enters the olfactory pathway and establishes latency in the astrocytes in the olfactory bulb/tract and nasal endothelium, other sites of HHV-6B latency (Ablashi et al., 2014, Donati et al., 2005, Harberts et al., 2011).

The olfactory bulb (OB) is not only a site of HHV-6B latency but also an immune organ that prevents virus in the saliva from invading the brain (Durrant et al., 2016). So, it is possible that the state of the OB is strongly associated with the reactivation of HHV-6B and invasion into the brain. OB dysfunction and a decrease in its volume have been reported in depressive patients (Negoias et al., 2010, Rottstaedt et al., 2018). We therefore considered that the effect of HHV-6B infection on the OB and other parts the olfactory system would differ between depressive patients and normal controls (NCs). As it has been reported that HHV-6B infection in olfactory tissues is mainly latent infection (Harberts et al., 2011), our aim was to study the effects of latent infection.

However, the effects of herpesviruses in the latent state on the host have been clarified in only a few cases, examples of which are oncogenetic herpesviruses, such as Epstein-Barr virus and Kaposi's sarcoma-associated herpesvirus. As reasons that the effects of these viruses are clearly known: first, the proteins produced by them during latency (latent proteins) have been identified; second, the functions of their latent proteins are well understood (Jha et al., 2016). Therefore, we thought that analysis of latent proteins produced by HHV-6B might clarify its effects on the host during latency. Astrocytes are considered to be the cells in the olfactory system latently infected by HHV-6B (Donati et al., 2005, Harberts et al., 2011), so we first attempted to identify a latent protein produced by HHV-6B in astrocytes during latency.

Results

Identification of SITH-1, a New HHV-6B Latent Infection Protein

In order to identify astrocyte-specific HHV-6B latent protein, we used the latent protein of human cytomegalovirus (HCMV), a close relative of HHV-6B, as a reference. In our previous research, we identified two cytomegalovirus latent transcripts (CLTs) that code for latent proteins in HCMV. One of these CLTs (CLT ORF94) codes for latent protein open reading frame (ORF) 94, which consists of 94 amino acids, and the other (CLT ORF152) for ORF152, consisting of 152 amino acids (Kondo et al., 1996). In HHV-6B, we had already identified HHV-6 latent transcript (H6LT) type I and H6LT type II, homologs of CLT ORF94. They are expressed in macrophages and have the capacity to regulate reactivation of HHV-6B (Kondo et al., 1991, Kondo et al., 2003), which suggests that they are related to HHV-6B latency in tonsils and adenoids.

In the present research, we looked for a homolog of CLT ORF152. This led to identifying a new HHV-6B latent transcript that codes for a protein consisting of 159 amino acids, and we named it Small protein encoded by the Intermediate stage Transcript of HHV-6-1 (SITH-1) (Figures 1A and S1A–S1D). This protein had amino acid homology (20% identity, 74% similarity) with ORF152 (Kondo et al., 1996). We studied SITH-1 expression using macrophage (MΦ) cell lines (THP-1 and HL-60) and astrocyte cell lines (U373 and A172). Although no HHV-6B replication was observed (data not shown) in these cell lines, expression of latency-associated genes U94 (Rotola et al., 1998) and/or H6LT was detected (Figures S1E and S1F) so we considered them to be latency-like gene expressing cells. SITH-1 mRNA was mainly expressed in astrocyte cell lines (U373 and A172) (Figure 1B). Also, SITH-1 protein was produced in HHV-6B-infected U373 cells without production of late proteins (glycoprotein [g]B and gH) (Figure 1C). Based on these properties, SITH-1 was considered to be the latent protein produced during HHV-6B latency in olfactory astrocytes that we were looking for in this study.

Figure 1

Identification and Characterization of SITH-1

(A) Structures of HHV-6B DNA, H6LT, and SITH-1 mRNA. R1, R2, and R3 indicate repeat regions.

(B) SITH-1 mRNA expression in macrophage (MΦ) cell lines (THP-1 and HL-60) and astrocyte cell lines (U373 and A172). Ratios of SITH-1 to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are shown.

(C) SITH-1 protein production in HHV-6B-infected U373 cells. Confocal imaging of HHV-6B infected and mock infected U373 cells. The upper six panels show staining with anti-SITH-1 antibodies and anti-gB antibodies and the lower six panels staining with anti-SITH-1 antibodies and anti-gH antibodies. Green, SITH-1; red, gB or gH; blue, DAPI.

(D) Detection of apoptosis in OB using TUNEL staining. Green, TUNEL; red, propidium iodide (PI).

(E) Numbers of TUNEL-positive apoptotic cells in each section.

(F) Enhancement of OB apoptosis detected by increase in Bax/Bcl-2 ratio

(G) Prolonged immobility in TST. One-way ANOVA with Fisher's post hoc test; ∗, p < 0.05; n.s., not significant.

(E and F) Values are means ± SEM. Unpaired t test; ∗, p < 0.05; ∗∗, p < 0.01.

Scale bars, 100 μm (C and D).

Characterization of SITH-1 Expressing Mice

We examined the effects of SITH-1 expression in olfactory astrocytes in mice. We created the mouse model (SITH-1 mouse) through intranasal inoculation with an adenovirus vector (SITH-Ad) possessing the SITH-1 open reading frame (ORF) downstream of the glial fibrillary acidic protein (GFAP) promoter, which is specifically expressed in astrocytes. In the SITH-1 mouse, SITH-1 was produced in olfactory ensheathing cells (OECs), a type of olfactory astrocyte (Figure S2A), and apoptosis occurred in the olfactory bulb (Figures 1D–1F). The apoptosis in the olfactory bulb occurred mainly in astrocytes (Figures S2B–S2D). As impairment of the olfactory bulb has been reported to be associated with depression (Morales-Medina et al., 2017, Rottstaedt et al., 2018), we subjected SITH-1 mice to the tail suspension test (TST) to see if they exhibited depressive symptoms. There was an increase in immobility time, indicating depression-like behavior, which was suppressed by an anti-depressant (SSRI) (Figure 1G). This suggests that the SITH-1 mice were presenting depressive symptoms.

We next studied the SITH-1 mice to see if they manifested the pathological state associated with depression. Since it has been found that depression is induced by stress (Belmaker and Agam, 2008, Keller et al., 2017), we examined an association between SITH-1 mouse behavior and stress. When the mice were subjected to mild stress by tilting their cages, they exhibited decreased sucrose preference, a symptom of depression (Figure 2A). Moreover, the expressions of corticotropin-releasing hormone (CRH) (Keller et al., 2017) and FK506 binding protein 5 (FKBP5) (Scharf et al., 2011) were increased, indicating that the hypothalamic-pituitary-adrenal (HPA) axis, the major component of the stress response, was enhanced (Figures 2B and 2C). Expression of Regulated in development and DNA damage response 1 (REDD1) (Ota et al., 2014), which is considered essential to depression induced by HPA axis activation, was also increased (Figure 2D). Steroidogenic acute regulatory protein (StAR) (Otawa et al., 2007), the rate-limiting factor for corticosteroid production in the adrenal gland, was also up-regulated (Figure 2E).

Figure 2

Depression-like Changes in SITH-1 Mice

(A) Enhanced stress in SITH-1 mice. Vector control and SITH-1 mice were caged with (+) or without (−) tilt, and sucrose preference was calculated as % sucrose intake.

(B–E) Up-regulation of indicated mRNAs related to HPA axis activation. (B–D) Up-regulation of indicated mRNAs related to HPA axis activation in the brain. (E) Up-regulation of StAR mRNA in the adrenal gland.

(F) DCX staining of hippocampal dentate gyrus (DG). Green, DCX; blue, DAPI. Scale bar, 100 μm.

(G) Numbers of DCX-positive cells in subgranular zone (SGZ) in each section. Comparison of vector control and SITH-1 mice. Unpaired t test; ∗∗∗, p < 0.001.

(A, B, D, and E) Bars represent median values; Mann-Whitney U-test. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001; n.s., not significant. (C) Values are means ± SEM. Unpaired t test; ∗, p < 0.05.

A decrease in hippocampal neurogenesis has been reported to be a cause of HPA axis activation in depression (Snyder et al., 2011). Through staining with doublecortin (DCX), a marker of immature neurons, we found that DCX-positive cells in the hippocampus of the SITH-1 mice were decreased (Figures 2F and 2G), indicating a decrease in hippocampal neurogenesis. In subsequent experiments using bromodeoxyuridine (BrdU)-positive cells or DCX and BrdU double-positive cells, similar decreases in neurogeneration were observed (Figures S2E–S2H). Also, it has been reported that hippocampus neurogeneration was decreased in a model of depression in which olfactory bulb function was impaired by bulbectomy (Morales-Medina et al., 2017). Considering these observations together, we believe that SITH-1 suppresses hippocampus neurogeneration via olfactory bulb impairment and, by doing so, activates the HPA axis, which results in the development of depressive symptoms. At this time, as there was no enhancement of inflammatory cytokine expression in the brain or olfactory bulb of the SITH-1 mice, the phenomena observed were considered not to be due to an immune response to SITH-1 (Figures S3A–S3J). Also, to exclude the possibility that SITH-1 expression outside the olfactory system causes depression-like phenomena, we administered SITH-Ad at the same titer as for intranasal inoculation orally and intraperitoneally and examined behavior as well as gene expression. No depression-like behavior or increase in depression-related gene expression was observed (Figures S4 and S5).

Functional Analysis of SITH-1

Next, we examined the mechanism by which SITH-1 induced apoptosis in astrocytes of the olfactory bulb. In two-hybrid screening, human calcium-modulating cyclophilin ligand (CAML) was identified as a protein binding to SITH-1 (data not shown). The binding of SITH-1 and CAML was suggested by the mammalian two-hybrid system (Figure 3A). CAML has been found to promote the influx of calcium into cells (Holloway and Bram, 1996). Also, it is believed to manifest this function through binding with other molecules in cells (Yamamoto and Sakisaka, 2015). When SITH-1 was expressed in the human U373 astrocyte cell line, using confocal fluorescence microscopy, we observed the colocalization of SITH-1 and CAML as well as an increase in immunofluorescence intensity, suggesting accumulation of the SITH-CAML complex (Figures 3B and S6). In addition, in the U373 cells in which SITH-1 was expressed, in the presence of extracellular calcium, there was an increase in intracellular calcium concentration (Figures 3C and 3D). In view of this, the complex of SITH-1 and CAML was considered to have the property to induce influx of extracellular calcium into cells. Human CAML and mouse CAML (mCAML) have high homology (88% identity, 98% similarity), and the binding of SITH-1 and mCAML was suggested by the mammalian two-hybrid system (Figure 3E). For further confirmation of the binding of SITH-1 and CAML determined using the mammalian two-hybrid system, we carried out a co-immunoprecipitation assay. Into HEK293T cells, which have hardly any endogenous CAML production, we transfected SITH-1 as well human or mouse CAML and carried out immunoprecipitation using anti-CAML antibodies. SITH-1 and CAML were co-immunoprecipitated, clearly showing that SITH-1 bound with both human and mouse CAML (Figure 3F). In addition, SITH-1 brought about an increase in intracellular calcium concentration in mouse cells (Figure 3G).

Figure 3

Characterization of SITH-1

(A) Binding activity of SITH-1 and CAML.

(B) Colocalization of SITH-1 and CAML in astrocytes. U373 cells expressing SITH-1 (SITH-1[+]) and not expressing SITH-1 (SITH-1[−]) were immunostained. Confocal images are shown. Red, SITH-1; green, CAML. Scale bar, 100 μm.

(C and D) Promotion of intracellular calcium influx by SITH-CAML complex with extracellular calcium (C) and without it (D). U373 cells expressing (SITH-1[+]) or not expressing SITH-1(SITH-1[−]) were stimulated with thapsigargin (TG) at time 0. Values are means ± SEM.

(E) Binding activity of SITH-1 and mCAML determined by a mammalian two-hybrid assay.

(F) Co-immunoprecipitation assay in cells co-expressing SITH-1 and CAML. After immunoprecipitation using anti-CAML antibodies, western blot analyses were performed using anti-CAML antibodies and anti-SITH-1 antibodies. The upper two panels show input cell lysate and the lower two panels show immunoprecipitated samples. hCAML, human CAML; mCAML, mouse CAML; M, molecular weight marker.

(G) Increased calcium concentration in mouse 3T3 cells due to SITH-1 expression. Bars represent median values; Mann-Whitney U-test. ∗∗∗∗, p < 0.0001.

(H) Enhancement of OB apoptosis by calcium ionophore.

(A, E, and H) Unpaired t test; ∗∗, p < 0.01; ∗∗∗, p < 0.001; ∗∗∗∗, p < 0.0001. Values are means ± SEM (H).

In order to confirm whether influx of extracellular calcium induced apoptosis in the olfactory bulb, we administered a calcium ionophore with a Ca2+ transport enhancing action in the cell membrane into the nasal cavity of the mice. This resulted in enhancing apoptosis in the olfactory bulb (Figure 3H). An increase in calcium concentration in OECs has caused ectopic neurovascular coupling in the olfactory bulb via release of ATP and glutamate (Thyssen et al., 2010), and by causing cerebrovascular stenosis (Mulligan and MacVicar, 2004) it could be associated with olfactory bulb cell impairment. In view of this, we consider that the mechanism by which SITH-1 induces apoptosis in the olfactory bulb is based on the promotion of extracellular calcium influx into OECs by the complex of SITH-1 and CAML.

Evidence for SITH-1 Expression in Depressive Patients

Next, we developed a method for investigating whether calcium influx due to SITH-1 occurs in vivo in depressive patients. Owing to the high invasiveness and difficulty of directly detecting SITH-1 expressed in the olfactory system, we decided to investigate the production of antibodies to the complex of SITH-1 and CAML. Based on a study using the SOSUI secondary structure prediction system (Hirokawa et al., 1998), the complex was assumed to be one with SITH-1 at the N terminus and CAML at the C terminus (N-SITH-CAML-C) (Figures S7A–S7D). We measured intracellular calcium concentrations for cells expressing each of the structures shown in Figures S7A–S7D. As we had expected, N-SITH-CAML-C was strongest in promoting calcium influx, whereas a complex with CAML at the N terminus and SITH-1 at the C terminus (N-CAML-SITH-C) achieved hardly any calcium influx (Figure 4A). Also, the strongest serum antibody reaction in the SITH-1 mice was with respect to N-SITH-CAML-C (Figures 4B and 4C). Based on these observations, the antibodies produced in vivo in the SITH-1 mice to the SITH-1-CAML complex were anti-N-SITH-CAML-C antibodies. Thus, through measurement of these antibodies, it should be possible to study the formation of the SITH-CAML complex and calcium influx in vivo.

Figure 4

Formation of SITH-CAML Complex in Depressive Patients

(A) Intracellular calcium concentrations in various protein expressing cells.

(B) Antibody response to various antigens in mice. Shows representative IFA images for SITH-1 mice and vector control mice for various antigens. Scale bar, 100 μm.

(C) Comparison of antibody titers for N-SITH-CAML-C, N-CAML-SITH-C and SITH-1in SITH-1 mice.

(D) Antibody reactions to various antigens in depressive patients. Shows representative IFA images for individual antigens in one patient and one normal control (NC). Scale bar, 100 μm.

(E) Comparison of antibody titers for N-SITH-CAML-C, N-CAML-SITH-C, and SITH-1 in depressive patients.

(F) Elevated anti-N-SITH-CAML-C titers in depressive patients.

(G) Comparison of anti-N-SITH-CAML-C antibody titers in normal controls with BDI up to 3 and 4 or greater.

Bars represent median values, and Kruskal-Wallis test with Steel's post hoc test (A, C, and E) and Mann-Whitney U-test (F and G). ∗∗, p < 0.01; ∗∗∗∗, p < 0.0001.

We next measured anti-N-SITH-CAML-C antibodies in depressive patients. Similar to the SITH-1 mouse, the serum antibody reaction in depressive patients was strongest with respect to N-SITH-CAML-C (Figures 4D and 4E). Also compared with normal controls, the anti-N-SITH-CAML-C antibody titer was higher in depressive patients (p = 1.78 × 10−15) (Figure 4F). In analysis using receiver operating characteristic (ROC) curves, the area under the curve (AUC) was 0.8573, the antibody positive rates were 79.8% for depressive patients and 24.4% for healthy subjects, and the odds ratio was 12.2 (cutoff value 1.957) The use of Anti-N-SITH-CAML-C antibodies achieved diagnostic accuracy of 77.7% for depression (Figure S7E). There was no difference between depressive patients and normal controls (NCs) regarding antibody titers for HHV-6 (Figure S7F). Based on these findings, we considered that more SITH-CAML complex was being formed in depressive patients and that calcium influx was enhanced. Compared with NCs who had no depressive symptoms (Beck Depression Inventory [BDI] ≤3), the anti-N-SITH-CAML-C antibody titer was higher in NCs who had a few depressive symptoms (4 ≤ BDI ≤10) (Figure 4G). This finding may support the contribution of SITH-CAML to the risk of depression. There were no significant differences in sex ratio or age between depressive patients and NCs (Table S1).

Discussion

In the present research, we identified the HHV-6B latent protein SITH-1, and our findings in a mouse model in which SITH-1 was expressed in the olfactory astrocytes suggested that its binding with CAML enhanced calcium influx into cells and promoted OB apoptosis.

We also measured antibodies (anti-N-SITH-CAML-C antibodies) to the SITH-1-CAML complex in human serum. Differing from postmortem examination of brains, detection of antibodies is an indirect method of evaluation, but it is minimally invasive and has the advantage of applicability as a test in patients who currently have depression. The detection rate of Anti-N-SITH-CAML-C antibodies in depressive patients was 79.8%, with a diagnostic accuracy of 77.7% for depression. In a previous study on antiphospholipid syndrome (APS), antibodies that react with the structure of active form β2-glycoprotein 1 were reported to be important for detection of APS (Pelkmans and de Laat, 2012). In the present study, the structure recognized by the detected antibodies was the SITH-1-CAML complex, which promotes calcium influx into cells, suggesting the possibility that they recognized a disease-related structure as in the case of APS.

Regarding an association with the OB, in imaging diagnosis, a decrease in OB volume was observed in depressive patients and diagnostic accuracy for depression based on this decrease was reported to be 68.1% (Rottstaedt et al., 2018). Thus, the diagnostic accuracy for depression using N-SITH-CAML-C antibodies in the present study and that using OB volume were close to each other, so it is conceivable that OB apoptosis due to SITH-1 expression similar to that in the mouse model occurs in depressive patients.

Using the mouse model, we found that promotion of OB apoptosis by SITH-1 activated the HPA axis. Since activation of the HPA axis has been observed to suppress host immune function by increasing corticosteroid production, its activation is thought to be advantageous for HHV-6B survival and reactivation. In addition, the OB is reportedly an immune organ that prevents viruses from invading the brain (Durrant et al., 2016) and induction of apoptosis in the OB may help HHV-6B in saliva to reach the brain. In this regard, a postmortem study reported proliferation of HHV-6B in cerebellums from depressive patients (Prusty et al., 2018).

We still know little about the cause of depression, and even in large-scale genetic analyses, no genes with an odds ratio of greater than 1.2 for an influence in depression have been discovered (Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium et al., 2013, Levinson et al., 2014). In the results of the present study, at 12.2, the odds ratio for SITH-1-CAML complex antibody positivity having an influence in depression was extremely high. In view of this, multiple factors could be at work in the influence of HHV-6B and SITH-1 in depression.

HPA axis activation is known to raise depression risk (Belmaker and Agam, 2008, Keller et al., 2017). Also, the immunosuppression due to a hyper-activated HPA axis and decreased immune function caused by OB impairment would promote HHV-6B proliferation in the brain as suggested above. In addition, risks for depression have been reported for pathogens apart from HHV-6B, which include other herpesviruses and chlamydia (Wang et al., 2014). HPA axis activation and OB impairment may increase the risk of depression by facilitating the infiltration of these pathogens into the brain and their subsequent proliferation.

The mechanism by which SITH-CAML complex is produced in greater quantities in depressive patients than in normal controls will need to be studied further in the future. As a possibility, overwork is known to increase salivary HHV-6B (Aoki et al., 2016), and in this case, the increase in HHV-6B could bring about an increase in SITH-1-producing cells among OECs. A study finding that job strain was a risk factor for depression (Madsen et al., 2017) is considered to support this hypothesis.

We believe that the present study is the first to demonstrate an influence of a herpesvirus latent protein in a non-oncological disease. Similar to HHV-6B, other non-oncogenic herpesviruses also express (or likely express) latent proteins, so our methods would be applicable to research on the disease risk of other herpesviruses. We consider that our study suggests new directions for research on causes of depression as well as new perspectives for research on this group of viruses and the virome.

Limitations of the Study

When we examined SITH-1 expression in cells infected with HHV-6A and HHV-7, we were unable to detect SITH-1 mRNA with the structure indicated in this study. Additionally, in PCR-based studies (Okuno et al., 1995, Tanaka-Taya et al., 1996, Yoshikawa et al., 2000a, Yoshikawa et al., 2000b) and an immunological study (Wang et al., 1999), most Japanese were reported to be infected with HHV-6B. Therefore, we consider our present findings are thought to be limited to HHV-6B. However, if a latent protein with a similar function to SITH-1 can be identified for HHV-6A, it may be possible to extend the results of the present study to HHV-6A.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kazuhiro Kondo (kkondo@jikei.ac.jp).

Materials Availability

All unique reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability

The data included in this manuscript have been deposited in GenBank (accession numbers HV763913.1 and HV763914.1) and Mendeley Data (http://dx.doi.org/10.17632/jxxpc9t732.1).

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.