Increased frequency of IL-6-producing non-classical monocytes in neuromyelitis optica spectrum disorder

Twenty patients with NMOSD fulfilling 2015 diagnostic criteria for NMOSD [18] and 20 patients with relapsing-remitting MS fulfilling the revised McDonald criteria [19] were enrolled from the Department of Neurology, National Cancer Center, South Korea. None of the patients had received high-dose steroids within 2 months preceding blood draws. Twenty age- and sex-matched healthy controls were also recruited for blood donation. Demographic and clinical characteristics of three groups and basic information of each patient are summarized in Tables 1 and 2, respectively. Peripheral blood was obtained by venipuncture and processed immediately for monocyte purification as described below.

Human peripheral blood mononuclear cells (PBMCs) were obtained using a Ficoll density gradient (GE Healthcare, Pasching, Austria) of buffy coats from healthy donors, NMOSD, and MS patients. Monocytes were purified using a CD14⁺ magnetic separation system (MACS, Miltenyi Biotec, Sunnyvale, CA, USA). The purity and viability of purified cells were assessed by flow cytometry and PI staining, respectively. All monocyte and PBMC cultures were performed in RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum (HyClone Laboratories, Logan, UT, USA) and 1% penicillin/streptomycin (GE Healthcare, Pasching, Austria).

For cytokine production and surface molecule analysis, CD14⁺ purified monocytes (5.0 × 10⁵ cells/mL) were incubated for 24 h at 37 °C in complete medium. Monocytes were left unstimulated or were stimulated with recombinant human IFN-γ (100 ng/mL, R&D systems, Minneapolis, MN, USA), or recombinant human CD40L (1.0 μg/mL, Enzo Life Sciences, Farmingdale, NY, USA) or both, to mimic encounters with activated T cells. Supernatants were collected for cytokine detection, and cells were detached for flow cytometry analysis. For ICS, CD14⁺ purified monocytes (5 × 10⁶ cells/mL) were cultured in complete medium, with or without CD40L, for 6 h.

The supernatant from CD14⁺ monocytes cultured media were collected and stored at − 80 °C before use. Cytokine measurement was performed within 4 weeks after the collection of monocytes cultured media and aliquoted to appropriate amounts to avoid repeated freeze/thaw cycles. For measurement of cytokines, standard ELISA kits for IL-10, IL-6, IL-1β, TNFα (all from BioLegend, San Diego, CA, USA), and IL-23 (eBioscience, Vienna, Austria) were purchased and used according to the manufacturer’s instructions.

Purified CD14⁺ monocytes cultured for 24 h, with or without stimulation, were labeled with primary antibodies directed against human CD80, CD86, HLA-DR, ICAM-1, or CD16 (all from BD Biosciences, San Jose, CA, USA). For ex vivo staining of inflammatory surface molecules, PBMCs were stained with CD3, CD14, CD16, CD56, CD66b, CD80, CD86, HLA-DR, and ICAM-1. For gating of the three monocyte subsets, we used a fixed set of CD14, CD16, CD3, CD19, CD56, and CD66b (all from BD Biosciences) on PBMCs for negative selection. For intracellular cytokine staining (ICS), PBMCs and monocytes were isolated using a MACS pan-monocyte isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany) and were incubated with Bredfeldin A (GolgiPlug; 1 μg/mL; BD Biosciences) and CD40L for 6 h. After being stained for surface markers, cells were fixed and made permeable according to the manufacturer’s instructions (BD Biosciences). IL-6, IL-10, TNFα, and IL-1β antibodies against cytokines were from BD Biosciences and IL-23 from Biolegend (San Diego, CA, USA). The appropriate Ig isotypes were included as negative controls. Data was acquired using a FACSVerse (BD Biosciences) and were analyzed with Flowjo software (Treestar, Ashland, OR, USA).

Spanning-tree Progression Analysis of Density-normalized Events (SPADE) is an automated clustering algorithm to confirm manual gating [20]. SPADE analyzes raw flow cytometry data to agglomerate different types of surface molecules into clusters, which allows for easy visualization of rare events. To implement this analysis, we fixed the default settings of 100 nodes and 5 × 10⁴ cells per sample.

The AQP4-IgG serostatus was confirmed by an in-house cell-based assay using the M23-AQP4-transfected Human Embryonic Kidney 293 (HEK-293) cell line, generated by an internal ribosome entry site (IRES) vector, as previously reported [21].

Data are presented as mean ± standard error of mean (SEM). Prism software was used for unpaired one-way or two-way analysis of variance to assess significance between groups, and Dunnett’s multiple comparison was performed. A p value of less than 0.05 was considered statistically significant and designated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

To evaluate the reactivity of monocytes to inflammatory stimuli, monocytes isolated from PBMCs of healthy control (HC), as well as MS and NMOSD patients (Additional file 1: Figure S1-A), were stimulated with IFN-γ and/or CD40L (n = 15 each) to mimic the inflammatory state, as previously reported [22, 23]. In the inflammatory state, monocytes are recruited to the site of inflammation by IFN-γ and activated through CD40/CD40L ligation by other immune cells such as B and T cells [10, 11]. The data of cytokine ELISA showed that IFN-γ stimulation by itself did not induce an appreciable amount of inflammatory cytokine production from monocytes, whereas it inhibited the CD40L-induced production of IL-10 in HC and MS (p < 0.01) (Fig. 1a). Unlike IFN-γ, CD40L stimulation by itself increased the production of IL-1β (p < 0.0001) and IL-6 (p < 0.0001) from NMOSD monocytes compared to unstimulated (US) monocytes. Production of IL-6 was also greatly increased following CD40L stimulation compared to that in US in MS monocytes (p < 0.001), whereas only a modest increase was seen in IL-10 production in MS. There was no increase in IL-10 production for NMOSD monocytes. Co-stimulation with IFN-γ and CD40L showed different reactivity and diversity of pro-inflammatory cytokine production (Fig. 1a). In NMOSD monocytes, TNFα (p < 0.0001), IL-1β (p < 0.0001), IL-6 (p < 0.05), and IL-23 (p < 0.0001) levels were all increased by co-stimulation compared to that in US. In MS monocytes, TNFα (p < 0.01) and IL-6 (p < 0.01) levels were increased by co-stimulation compared to that in US (Fig. 1a). In HC monocytes, only a modest increase in IL-6 production was observed after co-stimulation (p < 0.05). In contrast, a marked induction of IL-10 production by CD40L stimulation was observed (p < 0.0001). Overall, NMOSD monocytes exhibit a greater inflammatory response to IFN-γ and/or CD40L stimulation than HC and MS monocytes, also showing an increased production of pro-inflammatory cytokines, and an impaired induction of IL-10 following CD40L stimulation, all of which suggest the existence of dysregulated cytokine production in NMOSD monocytes.

To confirm the dysregulated cytokine production in NMOSD monocytes, we then compared cytokine production between HC, MS, and NMOSD monocytes. Even in the US condition, NMOSD monocytes produced a significantly higher amount of TNFα (p < 0.05), IL-6 (p < 0.01), and IL-23 (p < 0.05) than HC monocytes (Fig. 1b). Following IFN-γ stimulation, TNFα production was significantly higher in MS monocytes than in HC monocytes (p < 0.05), as previously reported [5]. Likewise, NMOSD monocytes also showed increased production of TNFα compared to HC monocytes (p < 0.05). Following CD40L stimulation, NMOSD monocytes produced a significantly higher amount of all pro-inflammatory cytokines, most markedly IL-1β (p < 0.0001) and IL-6 (p < 0.0001), and a markedly lower amount of IL-10 (p < 0.0001) compared to HC monocytes. In MS, only IL-6 production was higher than HC monocytes after CD40L stimulation (p < 0.05). This result corresponds to the previous reports on IL-6 and MS, which showed IL-6 production was similar between MS and HC monocytes in US, but monocyte activation led to a significant increase in IL-6 levels in MS [24]. In addition, after CD40L stimulation, IL-1β production was far greater (p < 0.0001) in NMOSD monocytes than in MS monocytes (Fig. 1b). Co-stimulation with IFN-γ and CD40L in NMOSD also showed a significant difference in cytokine production (higher in TNFα, p < 0.0001; IL-1β, p < 0.0001; IL-6, p < 0.01; and IL-23 p < 0.0001, and lower in IL-10, p < 0.0001) compared to that in HC monocytes, but there was no synergistic effect on cytokine production. Co-stimulation resulted in decreases in IL-10 production, compared to CD40L stimulation of HC (p < 0.0001) and MS (p < 0.01) monocytes alone, which corresponds to previous findings that IFN-γ inhibits IL-10 production (Fig. 1a) [25].

To confirm the ELISA data, we conducted ICS on HC, MS, and NMOSD monocytes (n = 15 each), which were purified to obtain pan-monocyte population (Additional file 1: Figure S1-B). For this experiment, we elected to only use CD40L because CD40L stimulation was more effective than IFN-γ at inducing monocyte cytokine production, and co-stimulation was not synergistic. The data showed that in NMOSD, there was an increase in the frequencies of TNFα⁺ (p < 0.01), IL-1β⁺ (p < 0.001), IL-6⁺ (p < 0.001), and IL-23⁺ (p < 0.001) monocytes following CD40L stimulation compared to that in CD40L-stimulated HC monocytes (Fig. 2a, b). In addition, MS also showed increased frequencies of IL-6⁺ (p < 0.05) and IL-23⁺ (p < 0.05) cells following CD40L stimulation compared to CD40L-stimulated HC monocytes. On the other hand, there were fewer IL-10⁺ monocytes observed for NMOSD and MS under both US (MS, p < 0.05; NMOSD, p < 0.01) and CD40L-stimulated conditions (MS, p < 0.01; NMOSD, p < 0.001) (Fig. 2a). Overall, the ICS analysis of NMOSD, MS, and HC monocytes confirmed the ELISA results, showing the increased production of pro-inflammatory cytokines and impaired induction of IL-10 in NMOSD.

Next, we compared the cell-surface expression levels of key immune molecules in CD14⁺ monocytes derived from HC, MS patients, and NMOSD patients (n = 15 each). In the US condition, NMOSD monocytes had expression levels of CD80, CD86, ICAM-1, and HLA-DR similar to that in HC and MS monocytes (Fig. 3a). However, IFN-γ stimulation of NMOSD monocytes increased the expression levels of CD80 (p < 0.001), ICAM-1 (p < 0.01), and HLA-DR (p < 0.01) compared to IFN-γ-stimulated HC monocytes. The level of CD80 in MS monocytes were also increased compared to that in HC monocytes (p < 0.05), as previously reported [26]. CD40L stimulation upregulated the expression of CD80 (p < 0.05) and ICAM-1 (p < 0.05) to higher levels in NMOSD monocytes than in HC monocytes. Similar to IFN-γ stimulation, co-stimulation with both CD40L and IFN-γ upregulated CD80 (p < 0.0001), ICAM-1 (p < 0.001), and HLA-DR (p < 0.001) expression in NMOSD monocytes compared to HC monocytes (Fig. 3b).

The level of CD16 was increased in NMOSD monocytes even in the US condition compared to HC monocytes (p < 0.001). However, neither CD40L nor IFN-γ stimulation could further alter the CD16 level; it remained constitutively high in NMOSD monocytes under all stimulated conditions when compared to stimulated HC monocytes. In contrast, MS monocytes only showed a difference in CD16 levels, compared to HC monocytes, following IFN-γ stimulation (p < 0.05) (Fig. 3a, b).

Monocytes can be divided into three subsets based on their expression levels of CD14 and CD16 (FcγRIII): CD14⁺⁺CD16⁺ are referred to as classical monocytes, CD14⁺CD16⁺ are referred to as intermediate monocytes, and CD14⁺CD16⁺⁺ are referred to as non-classical monocytes. The increased level of CD16 in NMOSD and MS monocytes could be due to either an increase in intermediate or in non-classical subsets. This question could not be addressed using the CD14⁺ MACS-separation process, because the CD14⁺ beads were already occupied with CD14 molecules. Thus, to investigate the frequency of the three different monocyte subsets, peripheral blood mononuclear cells (PBMCs) from HC, MS, and NMOSD patients (n = 20 each) were stained for CD3, CD14, CD16, CD19, CD56, and CD66b to gate the monocyte subset (Additional file 1: Figure S2). The data shows that, in MS and NMOSD, the classical monocyte subset was decreased by 10% compared to that in HC monocytes (p < 0.001 for both). In addition, the frequencies of intermediate and non-classical subsets were increased by a percentage equivalent to the decrease in the classical monocyte frequency seen in MS and NMOSD monocytes, respectively (Fig. 4b). In fact, the frequency of non-classical monocytes in NMOSD was four-fold and two-fold higher than that in HC (p < 0.0001) and MS (p < 0.05), respectively. However, the frequency of intermediate monocytes in NMOSD was similar to, or lower than, that in HC (p = 0.4866) and MS (p < 0.05), respectively (Fig. 4a, b). Notably, a remarkable increase in the intermediate monocyte subset frequency was observed in MS (p < 0.05 for both), agreeing with previous reports [17, 24]. Overall, in NMOSD, the frequency of non-classical monocytes increased whereas the frequency of intermediate monocytes did not. This result implies that the constitutively increased level of CD16 observed previously in NMOSD (Fig. 3b) was mainly driven by an increase in non-classical monocytes.

To validate our data on the CD16 levels in HC, MS, and NMOSD, we employed a SPADE analysis. The data showed that NMOSD monocytes do indeed have increased CD16 expression and decreased CD14 expression level in the monocyte clusters compared to HC and MS (Fig. 4c).

From our data, NMOSD monocytes exhibit increased inflammatory cytokine production, increased inflammatory surface molecules, and an increased frequency of the non-classical monocyte subset compared to HC monocytes (Figs. 3 and 4). Based on these results, it was still unclear which monocyte subset was responsible for the cytokine production and inflammatory characteristics. To address this, we purified monocytes using pan-monocyte isolation kit (Additional file 1: Figure S1-B) and stimulated monocytes with CD40L, identical to that used previously (Fig. 2) and compared cytokine production from each monocyte subset between three groups (n = 15 each).

In the US condition, inflammatory cytokine-producing monocytes were rarely observed. Interestingly, however, a significant proportion of NMOSD non-classical monocytes constitutively produced IL-6 (p < 0.05) (Fig. 5a). In fact, the mean frequency of IL-6⁺ cells under US conditions, as well as following CD40L stimulation, was 60% of that of the total non-classical monocytes in NMOSD, which is three-fold higher than that of MS and HC monocytes (Additional file 1: Figure S3). In contrast, intermediate monocytes in MS showed an increased frequency of IL-6⁺ cells compared to HC monocytes in the US condition (Fig. 5a, b). Unlike IL-6, other cytokines (TNFα, IL-1β, IL-23, IL-10) were not constitutively increased in NMOSD, MS, and HC monocytes.

After CD40L stimulation, the frequency of the TNFα⁺ classical subset of NMOSD monocytes was increased five-fold relative to that of the same subset in HC monocytes (p < 0.05). There was also a modest increase in TNFα⁺ intermediate monocytes compared to HC monocytes (p < 0.05) (Fig. 5b). In the case of MS, previous reports have shown that the frequency of TNFα⁺ monocytes was similar to that of HC monocytes, but CD16⁺ purified MS monocytes produced a greater amount of TNFα than HC monocytes [17]. In our data, in MS monocytes, the frequency of the TNFα⁺ classical monocyte subset was similar to that of the same subset in HC monocytes, but the TNFα⁺ intermediate monocyte subset was upregulated by CD40L stimulation, being significantly different from the HC intermediate monocyte subset (p < 0.01) (Fig. 5b). This result suggests that the earlier observation on MS CD16⁺ monocytes could be largely influenced by intermediate monocytes. The IL-6⁺ classical monocyte subset of NMOSD monocytes were increased three- and five-fold compared to the IL-6⁺ classical monocyte subset of MS (p = n.s.) and HC monocytes (p < 0.001), respectively. In addition, the IL-6⁺ non-classical monocyte subset of NMOSD monocytes was increased two-fold relative to the same subset in HC monocytes (p < 0.0001). In this study, we found that IL-10 was mainly upregulated by CD40L stimulation in the classical monocyte subset of HC monocytes whereas other subsets were less affected (p < 0.01) (Fig. 5b). This observation has also been previously reported [27], which suggests that CD14⁺ purified monocytes produced greater amounts of IL-10 than CD16⁺ purified monocytes. Both NMOSD and MS monocytes showed impaired IL-10 production, since the frequencies of IL-10⁺ classical monocytes were decreased in both monocytes compared to HC monocytes (MS, p < 0.01; NMOSD, p < 0.001). In addition, the IL-10⁺ intermediate monocyte subset of NMOSD monocytes was significantly decreased compared to the same subset of HC monocytes (MS, p = n.s.; NMOSD, p < 0.05) (Fig. 5a, b).

We also analyzed and compared the expression pattern of cell-surface molecules (CD80, CD86, ICAM-1, HLA-DR) in these monocyte subsets (n = 10 each). The co-stimulatory molecules CD80 and CD86 showed a different pattern of expression in these monocyte subsets, with CD80 having a higher expression only in the intermediate monocyte subset than in the other subsets, but CD86 showed higher expression in both the intermediate and non-classical than in the classical subset (Fig. 5c). Interestingly, the ICAM-1 expression level in the intermediate monocyte subset was high in NMOSD and MS monocytes, but not in HC. For HLA-DR expression, intermediate monocyte subset expressed significantly higher than other subsets in all three groups, as previously reported [28, 29]. A comparison between disease groups showed that the NMOSD monocyte subsets had higher expression of these cell-surface molecules. In classical monocytes, the expression level of ICAM-1 was increased in NMOSD compared to that in HC classical monocytes (p < 0.001). In intermediate monocytes, the expression levels of CD86 (p < 0.01), ICAM-1 (p < 0.0001), and HLA-DR (p < 0.0001) in NMOSD were also increased compared to that in HC intermediate monocytes. In addition, ICAM-1 expression was the highest in NMOSD intermediate monocytes; MS intermediate monocytes also showed increased expression of ICAM-1 compared to HC intermediate monocytes (p < 0.001), but the expression was lower than in NMOSD intermediate monocytes (p < 0.05). In non-classical monocytes, the expression levels of CD86 (p < 0.001) and ICAM-1 (p < 0.0001) in NMOSD were increased compared to that in HC monocytes. Also, MS non-classical monocytes exhibited increased expression of CD86 (p < 0.05) and ICAM-1 (p < 0.05) compared to HC. However, ICAM-1 expression was higher in NMOSD non-classical monocytes than in MS non-classical monocytes (p < 0.05).