Mitochondrial DNA enhance innate immune responses in neuromyelitis optica by monocyte recruitment and activation

Although recent studies indicate the involvement of monocytes in accelerating the lesion formation of neuromyelitis optica spectrum disorder (NMOSD), the precise mechanism of the innate immune system activation remains elusive. Thus, in this study, we aimed to clarify the mechanisms of NMOSD pathogenesis from the viewpoint of innate immunity activation. We established anti-AQP4 recombinant autoantibodies (Ab) from plasmablasts in NMOSD patient’s CSF. Human astrocytes treated with anti-AQP4 Ab produced a significant amount of CCL2 and contributed to the efficient recruitment of monocytes. Moreover, mitochondrial DNA (mtDNA), which activated monocytes via Toll-like receptor 9 (TLR9), was released from astrocytes treated with anti-AQP4 Ab. MtDNA further enhanced CCL2 production by monocytes, and it was demonstrated that mtDNA concentration correlated with the efficiency of monocyte recruitment in the CSF of NMOSD patients. In conclusion, these observations highlight that mtDNA which was released from astrocytes damaged by anti-AQP4 Ab has a central role in establishing the inflammatory loop of monocyte recruitment and activation via an innate immunity pathway.

However, the precise relationship between the astrocytic injury and monocyte involvement in NMOSD pathogenesis remains elusive. Recently, Yamashita et al. revealed that extracellular mitochondrial DNA (mtDNA) was increased in NMOSD patient CSF compared with multiple sclerosis (MS) and other neurological diseases, and the mtDNA released from astrocytes which AQP4-Ab damaged in a CDC-dependent manner enhanced microglia to secrete IL-1β via Toll-like receptor 9 (TLR9) 9 . Another report showed macrophages were activated and expressed NOD-like receptor family, pryin domain containing 3 (NLRP3) in NMOSD lesion 10 . Therefore, we hypothesized that mtDNA released from damaged astrocytes might serve as a bridge between acquired immunity cytotoxicity and innate immunity activation in NMOSD.
In this study, we aimed to reveal the pivotal role of mtDNA to recruit and activate peripheral monocytes in NMOSD pathogenesis.

Results
Pathogenic anti-AQP4 Abs derived from patient plasmablasts. For the investigation of direct effects of anti-AQP4 Ab in the pathogenesis of NMOSD, we first aimed to establish pathogenic monoclonal AQP4 Ab derived from the patients. When we examined CSF plasmablasts characterized by CD3 − CD19 int CD138 + phenotype among six NMOSD patients (Fig. 1A), the CSF derived from patient 4 (arrow head) showed the highest percentage of CD3 -CD19 int CD138 + cells (Fig. 1B), and was further used for generating monoclonal AQP4 Ab by single-cell sorting. The DNA sequences of V regions of each clone showed interclonal diversity, represented by somatic hypermutations in the CDR regions of IgH, and either Igκ or Igλ (Fig. 1C).
Of the 29 clones generated, eight showed positive immunoreactivity to AQP4-expressing HEK293 cells in FACS analysis and immunohistochemistry ( Fig. 1D and 1E). Of the eight clones, clones GK15, 41, 74 and 89 showed remarkable capacity to bind AQP4 (Fig. 1D). To identify the clone that possesses the highest capacity of CDC, LDH release was examined among eight clones utilizing AQP4-expressing HEK cells. The clone GK89 was the most potent to induce CDC in the LDH release assay (Fig. 1F) and was, thus, selected to be used in the following studies (arrow head).
Chemokine signature of human astrocytes exposed to anti-AQP4 Ab. Human astrocytes exposed to GK89 showed significant upregulation of CCL2 (control; 1.16 ± 0.077, GK89; 1.82 ± 0.039, p = 0.0018) among potent chemokines which are known to be increased in NMOSD CSF, such as CXCL8 and CCL20 ( Fig. 2A). Among the immune cells observed in inflammatory lesions, macrophages serve as an enhancer to form destructive lesion. Therefore, chemokines previously recognized to be involved in monocyte recruitment were further assessed. The expression levels of CCL3, CCL4, CCL5, CCL7, and CX3CL, however, did not show a significant difference between GK89-treated and control groups (Fig. 2B). These results suggest that human astrocytes targeted by anti-AQP4 Abs have the potentiality to attract monocytes by secreting CCL2.
Phenotypic change of monocyte by AQP4-Ab conditioned medium. In addition to the enhanced recruitment, recruited monocytes can also be activated in the inflammatory milieu. Therefore, we further assessed for the phenotypic change of human monocyte exposed to GK89-treated ACM. CD86 expression level was enhanced among the activation markers, such as CCR2, HLADR, and CD80 ( Supplementary Fig. S1). Quantitative PCR analysis also showed that CD14 + monocytes exposed to GK89-treated ACM exhibit enhanced expression of adhesion molecules such as CD11b (p = 0.049) and CD11c (p = 0.049) (Fig. 3A).

The number of CSF monocyte correlates with mitochondrial DNA levels in NMOSD
patients. An immunohistochemical study exhibited that CD68-positive monocytes were detected at the subarachnoid space in NMOSD, where the loss of AQP4 staining was observed ( Fig. 4A (left column)). CD68positive monocytes were rarely observed in the control group ( Fig. 4A (right column)). These results highlight the active monocyte recruitment near the active lesions of NMOSD, where AQP4 expression is diminished. As astrocytes exposed to anti-AQP4 Abs were shown to preferentially recruit monocytes in vitro, we investigated whether CSF monocyte counts correlate with astrocyte damage in NMOSD patients. For this purpose, we investigated the correlation between CSF mtDNA levels and various immune cell subsets in CSF and we confirmed the positive correlation between mtDNA and CD14 + monocytes in the CSF of NOMSD patients. For other cell subtypes, we didn't find any significant correlation (Fig. 4B, C).
These results suggest that CD14 + monocytes are preferentially recruited within the CNS after astrocytic damage by the release of mtDNA in NMOSD patients.

Discussion
In this study, we showed that mtDNA released from astrocyte stimulated by anti-AQP4 Ab serve as a molecular bridge of innate immunity, by enhancing monocyte activation and recruitment via CCL2 in NMOSD pathogenesis.
As for the role of monocytes in the acceleration of lesion formation in NMOSD, a previous report showed the reduction in the lesion size after macrophage depletion with clondronate liposomes in the NMOSD rat model 7 .
In this study, we revealed that CCL2 generated from astrocytes had a critical role in recruiting monocytes. Consistently, previous literature showed that anti-AQP4 Ab stimulated astrocytes to release CCL2 11 and the ablation of CCL2 released from astrocytes reduced the recruitment of peripheral monocytes to CNS and dampened axonal loss and neurological deficit 12 . Another study also reported that accumulation of lactosylceramide (Lac-Cer) via β-1,4-galactosyltransferase6 (B4GALT6) promoted CCL2 production in astrocyte and B4GALT6 was actually upregulated in astrocytes expressing CCL2 in MS lesions 13 . In addition, previous reports showed that CCL2, CCL3 and CCL4, the chemokines for attracting monocytes, were elevated within the CSF of patients with NMOSD 14 . Thus, it is plausible that for the massive infiltration of monocytes in the lesions of NMOSD, CCL2 serves as the critical factor to enhance the innate immune response.
On the other hand, our recent report showed that mtDNA, one of the damage-associated molecular patterns (DAMPs), in CSF was elevated in NMOSD and mtDNA concentration was reduced after immunotherapy 9 . These observations indicated that mtDNA might be a critical factor for forming NMOSD pathology. So we hypothesized that mtDNA released from damaged astrocytes might be involved in monocyte activation in NMOSD. Actually, in this study, we showed that monocytes exposed to ACM containing abundant mtDNA facilitated expression of CD86, CD11b and CD11c, and mtDNA further accelerated monocytes to secrete IL-1β in TLR9 and inflammasome-dependent manners. IL-1β not only activates leukocytes 15,16 but also makes the blood-brain barrier leaky enhancing monocyte migration 17 and IL-1β are shown to have essential roles in forming NMOSD lesions in a rat model 18 . Furthermore, in this study we elucidated the inflammatory loop of monocyte recruitment in NMOSD, where monocytes themselves enhanced the production of monocyte-recruiting chemokine after being exposed to mtDNA. Together, these observations suggest that astrocytes not only play an essential role in recruiting monocytes, but they also promote the activation of monocytes, and mtDNA serve as a critical factor in enhancing the inflammatory conditions. Importantly, we observed that CD68-positive monocytes were preferentially recruited at the subarachnoid space in the active lesion of NMOSD. Three pathways, the entries through circumventricular organs, subarachnoid space and parenchymal vessel, were previously reported to have crucial roles in immune cell invasion to the CNS 19 , and monocyte infiltration through the CNS surface is observed at the early phase in experimental autoimmune encephalomyelitis (EAE) 20 . Thus, our results suggest the importance of the subarachnoid pathway in NMOSD pathology. Moreover, anti-AQP4 antibody-induced astrocyte loss is reported at the pial glial limitans, and microglial revitalization is known to take place at the pila surface of the CNS in NMOSD 21 . This observation is consistent with our results that the degree of monocyte migration correlates with the concentration of mtDNA, the astrocyte destruction marker in NMOSD CSF.
A previous report showed the elevation of DAMPs, such as HMGB1, in NMOSD CSF but did not reveal the association between the increase of DAMPs and NMOSD pathogenesis 22 . By demonstrating that mtDNA concentration within the CSF show positive correlation with monocyte population in NMOSD patients, the clinical relevance of mtDNA as an initial trigger to enhance innate immunity in NMOSD is suggested in our study. A previous report elucidated the amount of CCL2 correlated with the copy number of mtDNA in the CSF of HIV CNS inflammation 23 . In this sense, our observation indicates that CNS damage could be directly related to monocyte migration via pathogenic mtDNA release.
Taken together, mtDNA which was released from astrocytes serve as a molecular bridge of innate immunity in NMOSD by activating and attracting monocytes from peripheral blood, further accelerating the inflammatory loop so as to exacerbate CNS damage (Fig. 5).

Patient information and CSF sample collection. CSF samples were obtained by patients with NMOSD
positive for anti-AQP4 antibody, fulfilling the 2015 NMOSD diagnostic criteria 24 . Each sample was collected at the relapse phase, defined as appearance of new neurological symptoms lasting more than 24 h. Informed consent was obtained from each patient. Human CSF was immediately centrifuged at 400 g for 10 min after lumbar puncture and stocked at − 80 °C. All of the patients' characteristics were shown in Table 1.

Generation of monoclonal antibodies by single-cell sorting. After lumbar puncture, CSF samples
were immediately centrifuged at 400 g for 10 min. Cells were stained with PECy7-conjugated CD3 (TONBO biosciences, diluted 1:200), PB-conjugated CD14 (BD pharmingen, diluted 1:200), APC-conjugated CD19 (eBioscience, diluted 1:50) and PE-conjugated CD138 (BD parmingen, diluted 1:50) and single-cell sorting was performed on Aria flow cytometer (BD biosciences), and CD3 -CD19 int CD138 + plasmablasts were isolated. The collected cells were lysed with NP40 and directly reverse-transcripted with ReverTra Ace (TOYOBO), according to the instruction protocol. V regions of IgH, and either IgK or Igλ were amplified by nested PCR from the reconstructed cDNA [25][26][27] . The generated V regions were then ligated into p3XFLAG-CMV-14 expression vectors (Sigma-Aldrich), which harbored constant regions of human IgG1, Igκ or Igλ. After the plasmids were transfected into HEK293 cells, antibodies produced in the supernatant were purified with protein G sepharose (GE Healthcare). The IgG concentration of each sample was determined by ELISA (capture antibody; SouthernBiotech, alkaline phosphatase conjugated detection antibody; SouthernBiotech, p-Nitrophenyl phosphate; SIGMA).   immunocytochemistry. One day after cells were seeded on 16-well chamber slides, 2 µg/ml of generated recombinant Abs were incubated for 30 min on ice, and 1:200 FITC-conjugated anti-human IgG (Southern Biotech) was used as a secondary antibody for 30 min on ice after three washes. Finally these cells were fixed with 4% paraformaldehyde (Nakarai) for 10 min and the cells were observed by BZ-X700 fluorescence microscope (Keyence).
Binding assay of recombinant Abs. 2
For the inhibitory assay of CCL2, 5 × 10 4 THP1 monocytes in RPMI without FCS was placed in the upper chamber of 8 μm-pore two chamber wells (24 wells) (Corning). After 30 min incubation with ACM and anti-CCL2 neutralizing antibody (R&D) at room temperature, the number of migrated cells for 2 h was counted with trypan blue (Gibco).
Quantitative PCR analysis of human monocytes. Human monocytes were isolated with CD14 magnetic beads (Miltenyi Biotec) from human PBMCs, and 1 × 10 5 monocytes were seeded on 96-well plates in the presence of ACM for 24 h. After the incubation, monocytes were collected and stored in IsogenII. The total RNA was extracted and directly reverse-transcribed with SuperScript VILO (Thermo Fisher Scientific) according to the manufacturer's protocol. Quantitative PCR analysis was performed with TaqMan probe against IL-1β assay. 1 × 10 5 human monocytes were isolated with CD14 magnetic beads (Miltenyi Biotec) from human PBMCs. CD14 + monocytes were treated with mtDNA extracted from 80 µl ACM for 24 h after being pretreated for 1 h with either 10 μM MCC950 (inflammasome inhibitor), 2 μM ODN2088 (TLR9 inhibitor) or only medium respectively. On the following day, the plate was centrifuged at 400 g for 10 min, and the supernatants were collected. The IL-1β in supernatants was measured using ELISA kit (abcam) according to the manufacturer's instruction.
CCL2 assay. 1 × 10 5 CD14 + human monocytes from human PBMCs were treated with mtDNA extracted from 20 µl ACM after pretreatment with 2 μM ODN2088 for 1 h. CCL2 in the supernatants was analyzed with an ELISA kit (R&D) according to the manufacturer's instruction.
immunohistochemistry. Brain blocks from six autopsy cases (3 NMOSD and 3 disease controls) were used. NMOSD samples included a 67-year-old (yo) female, a 44-yo female and a 80-yo female. The disease control samples contained a 63-yo male (MS), a 70-yo male (stroke) and a 56-yo male (chronic inflammatory demyelinating polyneuropathy). The formalin-fixed paraffin-embedded blocks were cut into 5 mm thick sections and were de-waxed. The sections were stained with ant-AQP4 rabbit antibody (1:1,000, Proteintech) or the anti-CD68 mouse antibody (undiluted, DAKO), followed by a secondary antibody conjugated with HRP (DAKO). Positive staining was detected by ImmPACT HRP substrate (Funakoshi).
CSF analysis of monocytes. CSF derived from 12 NMOSD patients were immediately centrifuged at 400 g for 10 min after lumbar puncture, and treated with the FcR blocking reagent (Miltenyi Biotec) for 15 min on ice. The cells were then stained with PE-conjugated CD4 (BioLegend), FITC-conjugated CD19 (BD Biosciences), and APC-conjugated CD14 antibodies (BioLegend) for 30 min on ice, and analyzed by Canto II flow cytometer (BD biosciences).
Statistical analysis. P values were calculated by the Mann-Whitney U tests (two-tailed) or T-test using SPSS software. Data are represented as average ± SEM. Statistical significance is reported as not significant; (n.s.), *; p < 0.05, **; p < 0.01 and ***; p < 0.001. ethics approval and consent to participate. All of the methods and experiments were conducted in accordance with the revised Declaration of Helsinki and Good Clinical Practice guidelines and was approved by the ethics committee of Osaka University Hospital (permit number 12091-6). We confirmed that informed consent was obtained by all participants or their legal representative.

Data availability
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.