Restoration of Default Blood Monocyte-Derived Macrophage Polarization With Adalimumab But Not Etanercept in Rheumatoid Arthritis

Introduction We previously reported a specific defect of rheumatoid arthritis (RA) monocyte polarization to anti-inflammatory M2-like macrophages related to increased miR-155 expression in all RA patients except those receiving adalimumab (ADA). In this longitudinal study, we examined whether different tumor necrosis factor inhibitors were able to restore monocyte polarization to M2-like macrophages and their effect on the transcriptomic signature. Methods M2-like polarization induced by human serum AB was studied in 7 healthy donors and 20 RA patients included in the ABIRA cohort before and 3 months after starting ADA or etanercept (ETA). The differential gene expression of M2- and M1-related transcripts was studied in macrophage-derived monocytes after differentiation. Results At baseline, RA monocytes showed a defect of polarization to M2-like macrophages as compared with healthy donor monocytes, which was negatively correlated with disease activity. M2-like polarization from circulating monocytes was restored only with ADA and not ETA treatment. The transcriptomic signature demonstrated downregulation of M2-related transcripts and upregulation of M1-related transcripts in active RA. In patients receiving ADA, the transcriptomic signature of M2-related transcripts was restored. Conclusion This longitudinal study demonstrates that ADA but not ETA is able to restore the M2-like polarization of monocytes that is defective in RA.


INTRODUCTION
Monocytes/macrophages are key players in the pathogenesis of rheumatoid arthritis (RA) (1, 2) by secreting tumor necrosis factor a (TNF-a) among other inflammatory cytokines. Synovium tissue macrophages are the most common resident immune cells in the healthy synovial membrane. In active RA with myeloid and lymphoid synovitis, the synovial membrane is leucocyte-rich, including an increased number of proinflammatory macrophages. Indeed, besides resident macrophages, blood monocytes can differentiate into monocyte-derived macrophages (MDMs) that may join the synovium and have different phenotypes and functions (3)(4)(5).
The "classically activated M1 macrophage phenotype" is considered to be pro-inflammatory, and the "alternatively activated M2 macrophage phenotype" is considered to be regulatory and anti-inflammatory in tissues. Actually, there is a continuum from pro-inflammatory to anti-inflammatory macrophages, with high plasticity between the different states. Classically activated macrophages contribute to RA pathogenesis by secreting pro-inflammatory cytokines and are the main producers of TNF.
We previously showed that RA patients had defective monocyte polarization toward an M2-like macrophage phenotype (CD11b Lo -CD71 Lo ; CD206 Lo ; CD163 Lo with decreased interleukin 10 [IL-10] secretion) in favor of an M1-like phenotype (inducible nitric oxide synthase + , interferon regulatory factor 5 + , and increased levels of pro-inflammatory cytokines such as IL-1b-IL-6-macrophage inflammatory protein 1a) (6). This defect was specific to RA because it was not found in healthy donors (HDs) or those with other inflammatory diseases (such as Sjogren's syndrome (SS) and spondyloarthritis (SpA) patients). Moreover, we have also found this specific defect in M2-like polarization of monocytes in RA patients receiving etanercept (ETA) but not adalimumab (ADA).
The objectives of this study were to longitudinally study the roles of the different TNF inhibitors (TNFis) for modifying monocyte polarization in RA patients and to study in detail the macrophage population before and after initiation of ADA treatment.

Patients
All RA patients fulfilled the 2010 American College of Rheumatology and European League Against Rheumatism RA criteria (7). The Disease Activity Score in 28 joints (DAS28) was used to assess the disease activity of RA patients. Because of the impact of corticosteroids on macrophages, patients receiving corticosteroid therapy ≥10 mg per day were excluded (8 To compare patients included in the ABIRA cohort, HD PBMCs were frozen under the same condition. Briefly, PBMCs were frozen at 6 to 10.10 6 /ml with autologous plasma-10% dimethyl sulfoxide (DMSO); then cells were placed at −80°C for 48 h and then transferred to −150°C.

Monocyte Selection, Peripheral Blood Mononuclear Cell Thawing, and Differentiation Into Macrophages
For PBMC thawing, cells were placed in a water bath at 37°C and rapidly transferred to complete Roswell Park Memorial Institute (RPMI). Cells were then washed and counted for monocyte staining and polarization to M2 macrophages, we remove samples under 1.10 6 PBMCs and viability under 80% of trypan blue. For M2 differentiation with human serum AB (SAB), fresh blood monocytes were isolated by using a pan-monocyte negative selection according to the manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany) to achieve purity of 90% (for RNA-seq) or thawed PBMCs (from patients included in the ABIRA cohort) and then cultured at 1.10 6 cells/ml for 6 days in hydrophobic Teflon dishes (Lumox; Duthsher) in a macrophage medium (RPMI 1640 medium supplemented with 200 mM of L-glutamine, 100 U of penicillin, 100 mg of streptomycin, 10 mM of HEPES, 10 mM of sodium pyruvate, 50 mM of b-mercaptoethanol, 1% minimum essential medium vitamins, and 1% nonessential amino acids) containing 15% of heat-inactivated human serum AB as a natural source of M-CSF, IL-10, IL-4, and IL-13. After 6 days of culture, MDMs were then harvested and suspended in 100% fetal bovine serum to avoid cell death, scratched, and washed with phosphate-buffered saline (PBS) 1×.

RNA-Sequencing and Bioinformatics Analysis
For RNA-seq, after blood monocyte isolation and differentiation into M2-like macrophages with SAB, mRNAs from macrophages were isolated by using the GeneJet RNA purification Kit (Life Technologies, Carlsbad, CA, USA) and QIAshredder (Qiagen, Valencia, CA, USA). The quality of the samples (RNA integrity number [RIN]) was assessed on the Agilent 2100 Bioanalyzer following the manufacturer's instructions. To construct the libraries, 250 ng of high-quality total RNA (RIN > 8) was processed by using the TruSeq Stranded mRNA kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. Briefly, after purification of poly-A-containing mRNA molecules, mRNA molecules are fragmented and reverse-transcribed with random primers. Replacement of dTTP by dUTP during the second-strand synthesis achieved strand specificity. The addition of a single A base to the cDNA was followed by ligation of Illumina adapters.
Libraries were quantified by a combination of Qubit concentration values and library average size from profiles obtained with the DNA High Sensitivity LabChip kit on the Bioanalyzer 2100. Libraries were sequenced on an Illumina Nextseq 500 instrument using paired-end 75 base-length reads (chemistry v2.5). The average number of reads per sample was 36.7 ± 4.4 million. After sequencing, a primary analysis based on AOZAN software (automated post-sequencing data-processing pipeline) was used to demultiplex and control the quality of the raw data (based on bcl2fastq v2.20.0.422 and FastQC v0.11.5). Obtained fastq files were then aligned by using the STAR algorithm (v2.7.1a). Reads were counted by using RSEM, and the statistical analyses of the read counts involved using the DESeq2 package v1.22.2 (R v3.5.3) to determine the proportion of differentially expressed genes between two conditions. The ingenuity pathway analysis (IPA) database was used to examine the potential functions and regulatory mechanisms of differentially expressed genes in RA-MTX patients versus HDs or RA-ADA versus RA-MTX patients by "upstream regulators" analysis.

Statistical Analysis
The data were analyzed by using Graph Pad Prism V9.0.1. Data were tested by the Mann-Whitney for two groups and the Kruskal-Wallis test with Dunn's multiple comparisons for multiple groups, expressed as mean ± SD or SEM with plot individual values. Correlation between variables was tested with two-tailed nonparametric Spearman's correlation, with a 95% CI. p < 0.05 was considered significant.

Patients
Twenty RA patients with failure of MTX treatment were included and analyzed before and at 3 months after TNFi initiation (n = 10 ADA; n = 10 ETA). We also included 7 HDs. Nine RA patients were recruited at Bicetre hospital, and 4 HDs were used for transcriptomic analysis. Their characteristics are in Table 1.

Defective M2 Polarization of Rheumatoid Arthritis Monocytes at Baseline Correlated With Disease Activity
At baseline, we confirmed that in RA patients, macrophages differentiated from monocytes with human SAB showed significantly fewer cells with the pan-macrophage markers CD11b and CD71 (p = 0.0005) as compared with HDs ( Figure 1A) and fewer cells with expression of M2-like markers such as CD206 (p = 0.008) and CD163 (p = 0.0003) ( Figures 1B, C). We found a negative correlation (Spearman = −0.313) between disease activity, measured by the DAS28, and the number of M2-like CD206 + macrophages ( Figure 1D), Thus, patients with a higher monocyte defect of polarization in M2-like macrophages had higher disease activity.

Defective M2 Polarization of Rheumatoid Arthritis Monocytes Is Restored After 3 Months of Adalimumab But Not Etanercept
Longitudinal follow-up of these patients at 3 months after ADA treatment showed that the M2-like polarization was restored to levels similar to HDs for pan-macrophage markers and CD206 (Figures 2A, B). ADA was only partially effective in restoring CD163 levels in macrophages ( Figure 2C). Conversely, ETA was unable to reverse the M2-like polarization defect (Figures 2D-F). Unfortunately, we are not able to demonstrate any correlation between M2-like markers at baseline (such as CD206 or CD163) and clinical or EULAR response at 1, 3, 6, or 12 months after anti-TNF treatment.

Monocyte Subpopulations and Activation Status Do Not Account for Adalimumab-Induced Rescue of the Polarization Defect
The polarization defect of monocytes to macrophages might be due to the abnormal distribution of monocyte subpopulations or their degree of activation. However, longitudinal follow-up of m o n o c y t e s ( C D 1 4 h i C D 1 6 − , CD14 h i C D 1 6 h i a n d CD14 dim CD16 high) ) did not significantly differ before and after treatment with ETA or ADA or between ETA and ADA, which suggests that the correction of the defect of M2-like polarization by ADA was not due to a change in the distribution of the monocyte subsets ( Figure 3A). The expression of HLA-DR was also unchanged after TNFi therapy ( Figure 3B).   Results are shown as symbols, lines, and mean ± SEM and were compared by Kruskal-Wallis test with Dunn's multiple comparisons. *p < 0.05, **p < 0.01, and ***p < 0.001. ns, not significant. difference in miR-155 expression between MTX-treated and ADA/IFX-treated RA patients ( Figure 4A).

RNA-Sequencing Analysis
Then, we performed a supervised analysis looking at genes associated with macrophage polarization. The terminology and markers to describe macrophage activation are heterogeneous; to address obstacles in describing macrophage activation, we selected the papers of Murray et al., Locati et al., and Sica et al. to choose M2 and M1 markers (10)(11)(12).
Finally, IPA of RA patients receiving ADA/IFX and those receiving no TNFi confirmed, according to the top diseases and functions, a decrease in rheumatic disease (activation score −2.2, 89 molecules related) and inflammation of joints (activation score −2.3, 69 molecules related) (Supplementary Figure 2B).

DISCUSSION
In RA patients with failure of MTX who were enrolled in the ABIRA study, we confirmed specific abnormalities of monocyte polarization in anti-inflammatory macrophages. The proportion of anti-inflammatory macrophages was negatively correlated with disease activity. This defective monocyte polarization in antiinflammatory macrophages was reversed by ADA but not ETA treatment. The transcriptomic analysis confirmed that stimulation in HD orientates to an anti-inflammatory macrophage phenotype, which leads to RA to a few anti-inflammatory macrophages and a majority of pro-inflammatory macrophages, which was reversed by ADA treatment.
The link between the proportion of blood monocyte-derived anti-inflammatory macrophages and disease activity supports the importance of this regulatory population even outside the joint. Previous studies suggested that synovial macrophages (SMs) have two major origins, namely, tissue-resident and monocyte-derived SMs, where the pro-inflammatory-M1 or anti-inflammatory-M2 concept does not reflect the phenotype of SMs (13,14). The Moreover, the amount of CD163 pos CD206 pos STMs is higher in sustained RA remission than in patients with flare (15). This study has some limitations. The number of patients followed up longitudinally was low (10 with each drug), the duration of disease was heterogeneous, and age and sex differ between HDs and RA patients. The transcriptomic analysis was not performed in patients who received ETA. The choice of the M1-and M2-related genes may be the subject of discussion, we used only CD206 and not CD163 for defining M2-like macrophages and we did not provide functional experiments.
Despite these limitations, this study sheds some light on a possible different mechanism of action of the different types of TNFi. With a longitudinal 3-month follow-up, we showed a restoration of M2-like anti-inflammatory macrophages with ADA but not ETA treatment. Monoclonal anti-TNF antibodies (such as ADA/IFX) and the TNF receptor 2 (TNFR2)-immunoglobulin (ETA) can bind soluble TNF and membrane TNF (mTNF), but the stability of the interaction with mTNF is higher, with monoclonal anti-TNF antibody revealing a possible differential mechanism of action of these molecules (16). Our study shows that monocyte polarization could be used as a biomarker to help to choose the best anti-TNF in RA patients. In case of a defect of polarization of monocytes into anti-inflammatory M2-like macrophages, ADA could be preferred over ETA. In addition, the assessment of polarization could be used as a biomarker for predicting and following the efficacy of anti-GM-CSF treatment, a new and very promising drug in RA. Indeed, anti-GM-GSF is able to inhibit the differentiation of monocytes into pro-inflammatory M1-like macrophages and thus can reverse the defect of polarization into anti-inflammatory M2-like macrophages.
Thus, if TNFi in RA acts mainly by inhibiting soluble TNF and if this action is the same between both types of drugs, there might be a supplementary action of monoclonal anti-TNF antibody by binding mTNF, which is mainly expressed by monocytes/macrophages. Some examples of a differential action of TNFi have been published regarding the activation of T regulatory cells specific to the monoclonal anti-TNF antibody via an increase in mTNF level and direct stimulation of T regulatory cells via TNFR2 (17). Recently, Diallo et al. demonstrated in mice transgenic for TFNR1−/−, TNFR2−/−, and tmTNFKI/KI (3TG mice), with canonical TNF signaling abolished and soluble TNF not secreted by monocytes/macrophages, that M1 macrophages polarized from monocytes of bone marrow and treated by an anti-murine-TNF antibody (MP6-XT22) or ETA inhibited the expression of proinflammatory cytokines and inducible nitric oxide synthase mainly by upregulating arginase 1 (18).
The mechanism of the restoration of M2-like macrophage differentiation by monoclonal anti-TNF antibody remains to be explored. Our transcriptomic study clearly showed that monocytes could not differentiate from anti-inflammatory macrophages. The link between the binding to mTNF and this defect is still unknown. In our previous study, we found that miR-155 was overexpressed in RA monocytes and M2 macrophages except in ADA-treated patients and could lead to this defect when miR-155 was introduced in healthy monocytes. In this study, we confirm that miR-155 is increased in M2-like macrophages from MTX-treated RA but also in M2-like macrophages from ADA/IFX-treated patients.
In conclusion, our study demonstrated that ADA but not ETA could restore the M2-like polarization of monocytes that is defective in RA. This is another example of the differential action of the two types of TNFi. The pathway involved in this restoration should be further studied to identify novel therapeutic targets.

DATA AVAILABILITY STATEMENT
The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (19) and are accessible through GEO Series accession number GSE196743 (https:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE196743).

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by CPP SudMé diterrané e V2020-A00509-30 for p a t i e n t s i n c l u d e d i n t h e R N A -S e q A n a l y s i s a n d ClinicalTrials.gov:NCT02116504 for patients included in the ABIRA cohort. The patients/participants provided their written informed consent to participate in this study.