We enrolled 40 patients with RA who were followed at Nagasaki University Hospital between January and July 2016. Their RA was diagnosed based on the ACR/EULAR 2010 RA classification criteria (18). We collected demographic, clinical, and laboratory characteristics of the RA patients and treatments for RA at the time of the monocyte subset analyses. We also examined the patients’ clinical disease activities of RA as disease activity score (DAS) 28-erythrocyte sedimentation rate (ESR) and DAS 28-C reactive protein (CRP) at the time of the monocyte subset analyses. We defined the rheumatoid factor (RF)-positive status as >15.0 U/mL (upper normal value, measured by a latex-enhanced immunonephelometric assay; Dade Behring, Marburg, Germany).

We also defined the anticitrullinated protein antibody (ACPA)-positive status as >4.5 U/mL [upper normal value, measured by enzyme-linked immunosorbent assay (ELISA); DIASTAT Anti-CCP; Axis-Shield, Dundee, UK].

We evaluated the radiographic severity of peripheral joint damage at entry according to Steinbrocker stage (19). In addition, we defined Steinbrocker stage I or II as non-erosive disease and Steinbrocker stage III or IV as erosive disease.

The patients’ characteristics are summarized in Table 1. We also enrolled 20 healthy donors to compare to the RA patients. This study was performed in accordance with the Declaration of Helsinki and was approved by the Investigation and Ethics Committee at Nagasaki University Hospital (approval number: 12072361). Patients and healthy donors gave their informed consent to be subjected to the protocol.

For the identification and quantification of circulating monocyte subsets in whole peripheral blood samples, we performed fluorescence-activated cell sorting (i.e., a FACS analysis). Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood using Lymphoprep (Cosmo Bio, Tokyo). PBMCs were stained for the examination of the expression of CD14, CD68, CCR2, CX3CR1, CD163, and CD206 using PE-conjugated anti-CD14 (BD Biosciences, San Jose, CA, USA), FITC-conjugated anti-CD68 (DakoCytomation, Carpinteria, CA, USA), Alexa Fluor 647-conjugated anti-CCR2 (BD Biosciences), FITC-conjugated anti-CX3CR1 (BioLegend, San Diego, CA, USA), and Alexa Fluor 647-conjugated anti-CD163 and APC-conjugated anti-CD206 (all from BD Biosciences) monoclonal antibodies. FITC-conjugated mouse IgG2b, FITC-conjugated rat IgG2b, AlexaFluor647-conjugated mouse IgG2b, AlexaFluor647-conjugated mouse IgG1 and APC-conjugated mouse IgG1 (all from BD Biosciences) were used isotype controls.

We defined positive CD14, CD68, and CCR2 monocytes as M1 monocytes (Figure 1A), and in separate tubes, we defined positive CD14, CX3CR1, and CD163 or CD206 monocytes as M2 monocytes (Figure 1B). First, we gated monocytes in a forward scatter/sideward scatter (FSC/SSC) dot-plot. Second, we gated CD14-positive monocytes within monocytes. Finally, we gated M1 or M2 monocytes using isotype controls within CD14-positive monocytes. At the same time, PBMCs were stained with a PE-conjugated anti-CD14 and a PerCP-Cy5.5-conjugated anti-CD16 (BD Biosciences) monoclonal antibodies. We also quantified three types of monocytes: classical CD14⁺⁺CD16⁻, intermediate CD14⁺⁺CD16⁺, and nonclassical CD14⁺CD16⁺ monocytes (Figure 1C) (17).

The relative frequency of these monocyte subsets is expressed as percent within the CD14-positive cell gate. Experiments were performed using a FACS Canto II Flow Cytometer and FACS Diva software (BD Biosciences).

At the same time as the FACS analysis, we purified CD14-positive cells from whole peripheral blood from the RA patients and healthy donors using CD14 M-pluriBead (pluriSelect, Spring Valley, CA, USA). We cultured CD14-positive cells (1 × 10⁵ cells per well) in a 96-well dish with Octeoclast Precursor BulletKit™ Medium containing l-glutamine, penicillin/streptomycin, M-CSF, and RANKL (Lonza, Walkersville, MD, USA) so that the cells would differentiate to OCs. We changed the medium three times a week. At day 12, the cells were fixed in 4% paraformaldehyde and stained with tartrate-resistant acid phosphatase (TRAP) using a TRAP staining kit (Cosmo Bio).

We defined OCs as TRAP staining-positive cells with >3 nuclei. We counted the OCs in the whole wells. Three wells were used to count OCs from one patient or one healthy donor, and we defined the average of the numbers of OCs of three wells as “the numbers of OCs.”

Pit formation assays were performed using Dentin Slice (from Ivory, Thin Type for pit formation assay, Wako Pure Chemical Industries, Osaka, Japan.). CD14-positive monocytes were cultured with Octeoclast Precursor BulletKit™ Medium on the dentin slices for 17 days. Resorbed lacunae were visualized by Mayer’s hemalum solution (Merck Millipore, Darmstadt, Germany) staining, and the relative resorbed area was measured under a microscope (BZ-9000, Keyence, Tokyo). Resorbing lacunae were quantified using Image J software (US National Institutes of Health, Bethesda, MD, USA). The relative area of resorbing lacunae is expressed as the percent of the area to the entire dentin slice.

In another set of experiments, CD14+ CX3CR1+ CD163+ (M2 monocyte) cells were isolated from PBMCs derived from healthy donors by FACS. We used the rest of the CD14-positive cells as M1-dominant monocytes because M1 monocytes could not be isolated by FACS due to the intracellular expression of CD68. They were incubated in MEM Alpha medium with 10% fetal bovine serum, 1% penicillin/streptomycin (all from Gibco, Basel, Switzerland) for 24 h in a 12-well dish at 5 × 10⁴ cells per well. Cells were stimulated for 24 h with LPS (100 ng/mL; Sigma-Aldrich, St. Louis, MO, USA) or phosphate-buffered saline as a control, and supernatants were obtained. Proteins in the cultured supernatants were detected by ELISA using an IL-6-specific ELISA kit according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN, USA). Absorption was measured at 450 nm.

Variables are described using frequencies for categorical variables and with the median and interquartile range (IQR) for quantitative variables. We assessed the association between variables using Wilcoxon’s rank sum test for quantitative variables. Correlations were analyzed using Spearman’s correlation coefficient. Univariate and multivariable regression model analyses were performed to determine the factors that contribute to the number of OCs. All tests were two-sided, and a p-value < 0.05 was considered significant. All statistical analyses were performed using JMP Statistical Software, ver. 11 (SAS Institute, Cary, NC, USA) and GraphPad Prism ver. 7.0 (GraphPad Software, San Diego, CA, USA).

As shown in Table 2, we compared RA patients with healthy donors for each monocyte subset. The population of CD206-positive M2 monocytes was too small to examine (0–0.5% within CD14-positive cells). Therefore, we redefined positive CD14, CX3CR1 and CD163 monocytes as M2 monocytes (Figure 1B). The median M1 and M2 monocyte percentages in CD14-positive cells were 16.7 and 40.0% in the RA group and 25.8 and 29.7% in the healthy donor group, respectively, with no significant differences between the RA and healthy groups. The RA and healthy groups had nearly equal M1/M2 ratios (0.60 vs. 0.62, respectively; p = 0.88). Likewise, the distribution of monocyte subtypes divided by CD14 and CD16 expression showed no significant differences between the RA patients and healthy donors. There was no correlation between the M1/M2 ratio and these subsets (Figures 1D–F).

Monocytes are capable of differentiating into OCs. We next investigated the osteoclastogenic ability of M1/M2 monocytes. There was a significantly positive correlation between the M1/M2 ratio and the number of OCs in the RA patients (ρ = 0.81, p < 0.001) (Figure 2A). In contrast, there was no significant correlation between the M1/M2 ratio and the number of OCs in the healthy donors (ρ = 0.30, p = 0.196) (Figure 2B). The numbers of OCs in vitro were significantly correlated with the area percentage of the pit formation area (ρ = 0.74, p = 0.001) (Figure 2C), which demonstrated that the RA patients who had lower M1/M2 ratios and fewer OCs had smaller resorbed areas (Figure 2D) compared to the RA patients who had higher M1/2 ratios and greater numbers of OCs (Figure 2E).

To explore the factors that influence the M1/M2 subsets in RA, we analyzed the influence of RF/ACPA positivity on the M1/M2 subsets. There were no significant differences between the RF-positive RA patients and the RF-negative RA patients regarding the M1/M2 ratio and the number of OC (Figures 3A,B). In contrast, the ACPA-positive patients had significantly higher M1/M2 ratios (0.87 vs. 0.41, p = 0.028) (Figure 3C) and greater numbers of OCs (76 vs. 47 per well, p = 0.005) (Figure 3D). We show an ACPA-negative patient’s OCs in vitro of (Figure 3E) and those of an ACPA-positive patient (Figure 3F).

There were no significant differences between erosive RA patients at entry (n = 13) and the non-erosive RA patients at entry (n = 27) regarding the M1/M2 ratio and the number of OC (Figures 4A,B). In addition, there were no significant differences between RA patients treated with biologics (n = 17, 7 patients treated with TNF inhibitors, 5 patients treated with tocilizumab, and 5 patients treated with abatacept) and RA patients treated without biologics (n = 23) regarding the M1/M2 ratio and the number of OC (Figures 4C,D).

Anticitrullinated protein antibody is a well-known bone erosion promoting factor in RA (20). To elucidate which factor is crucial for osteoclastogenesis, we investigated factors that may contribute to the number of OCs by performing univariate and multivariable analyses (Table 3). In the univariate analysis, the positivities for RF and ACPA and the M1/M2 ratio significantly contributed to the numbers of OCs. However, the multivariable regression analysis revealed that the M1/M2 ratio was the sole significant factor contributing to the number of OCs (β-coefficient: 16.3, p < 0.0001).

An M1/M2 ratio >1 means that there are relatively more M1 than M2 monocytes. We divided the RA patients into two groups: those with an M1/M2 ratio >1 (n = 13) and those with an M1/M2 ratio ≤1 (n = 27). There were no significant differences of CRP levels between M1/M2 ratio >1 group and M1/M2 ratio ≤1 group (Figure 5A). Because tocilizumab is known to normalize CRP levels (21), we excluded patients treated with tocilizumab (n = 5) to evaluate CRP levels. As a result, the CRP levels were significantly higher in the M1/M2 ratio >1 group (0.08 vs. 0.45 mg/dL, p = 0.032) (Figure 5B). In addition, the M1/M2 ratio >1 group, which included patients treated with tocilizumab, had significantly higher ESR values (9 vs. 29 mm/1 h, p = 0.011) (Figure 5C). The M1/M2 ratio >1 patients had significantly higher numbers of OCs (47 vs. 100 cells per well, p < 0.001) (Figure 5D).

To analyze the function of M1/M2 monocytes in inflammation, we extracted M2 monocytes from whole monocytes and evaluated their cytokine production by measuring the IL-6 levels of the culture supernatants. The results demonstrated that the M1-domonant monocytes produced significantly more IL-6 than the M2 monocytes did (n = 5, p = 0.032, Figure 6).