Reduced Annexin A1 Expression Associates with Disease Severity and Inflammation in Multiple Sclerosis Patients

Inflammation is a fundamental host defense process mounted in response to chemical, physical lesions, or microbial infection, which results in edema formation and leukocyte trafficking into the injured site (1). As acute inflammation is accompanied by destruction of the surrounding host tissue, the high-tuned regulation of the duration and severity of the proinflammatory phase is necessary for organ/tissue regain of function (2). However, there are some pathological conditions in which inflammation fails to resolve and persists overtime leading to chronic inflammation, like those occurring in immune-mediated diseases of the CNS, such as multiple sclerosis (MS) (3).

MS is an autoimmune disorder characterized by multifocal areas of demyelination, axonal damage, activation of glial cells, and immune cell infiltration. Although several genetic, environmental, and hormonal factors have been implicated in pathogenesis of MS, the perturbation of endogenous regulatory mechanisms represents an important factor for initiation, maintenance, and lack of resolution of brain inflammation (3). Several cytokines, NO, free radicals, and the abundance of proinflammatory cells such as CD4⁺ Th1 and Th17 targeting CNS self-antigens have been implicated in MS pathogenesis (4, 5). Despite that many different disease-modifying treatments have become available to reduce frequency of relapses and slow disease progression, glucocorticoids are recommended as the standard of care for acute MS flares, thanks to their proven effectiveness to reduce clinical disability and promote the recovery (6, 7). This is also secondary to a direct effect on T cells and blood brain barrier (BBB) during MS (8, 9). Annexin A1 (ANXA1) is considered a primary endogenous mediator of the anti-inflammatory and immunosuppressive actions of glucocorticoids (10). It is expressed abundantly in monocytes and neutrophils, playing an important role in the resolution of inflammation through the control of leukocyte migration, macrophage phagocytosis, and neutrophil apoptosis (11, 12). After cytoplasmic ANXA1 protein is externalized, it interacts with negatively charged phospholipids of membrane and its receptor N–formyl peptide receptor 2 (FPR2) (13), activating downstream anti-inflammatory signaling in an autocrine and paracrine manner (inhibition of phospholipase A₂ activity, downregulation of MAPK, and NF-κB activation) (14–16). Despite its well-described inhibitory effect on innate immune responses in vivo and in vitro, its role on adaptive immunity is extremely controversial (17–22). ANXA1 is also expressed in several brain cell types, in particular in microvascular endothelial cells, where it regulates BBB integrity (23). Recently, a selective loss of ANXA1 in the plasma and cerebrovascular endothelium of MS subjects has been reported (23). All these findings pinpoint ANXA1 as a critical component at the interface between the anti-inflammatory regulation of the immune system and the control of the BBB integrity in MS.

In this study, we evaluated how ANXA1 levels correlated with disease clinical score and number of relapses in a cohort of naive-to-treatment relapsing/remitting MS (RRMS) subjects before they underwent standard glucocorticoid administration. This is particularly relevant, as steroids can impact on ANXA1 production, thus affecting our analyses. We also investigated the role of ANXA1 in adaptive immunity by measuring its production in T cell subsets with different roles in MS pathogenesis, including CD4⁺ conventional T (Tconv) cells, CD4⁺ Forkhead-box-p3 (FOXP3)⁺ regulatory T (Treg) cells, and proinflammatory RAR-related orphan receptor γ (RORγt⁺) (Th17) cells, also relating the findings to migratory cell capacity in an in vitro BBB model. Finally, we explored whether ANXA1 had a direct anti-inflammatory role on human T cells and whether this process could be altered in MS. We found reduced systemic and T cell–derived expression of ANXA1 that correlated with disease severity in RRMS subjects. At molecular level, ANXA1 affected T cell activation in healthy subjects through a direct interference with T cell priming, and this pathway was hampered in RRMS.

The study was approved by the Institutional Review Board of the Università degli Studi di Napoli “Federico II” and by the Oxfordshire Research Ethics Committee. RRMS subjects were enrolled by our clinician collaborators at Università degli Studi di Napoli “Federico II,” Azienda Ospedaliera di Rilievo Nazionale “Antonio Cardarelli” Napoli, Istituto di Ricovero e Cura a Carattere Scientifico Istituto Neurologico Mediterraneo Neuromed, Pozzilli (Isernia), and Centre for Neuroscience and Trauma, Queen Mary University of London (London, U.K.). Peripheral blood was obtained from RRMS and healthy subjects after they signed a written informed consent approved by institutional review boards. All blood samples were collected at 9:00 am into heparinized Vacutainers (BD Biosciences) and were processed within the following 4 h. Demographic and clinical characteristics of study cohorts are shown in Table I. All subjects were naive to treatment and with definite RRMS diagnosis, according to the revised McDonald criteria (24). Healthy subjects were matched for age, body mass index, and sex and had no history of inflammation, endocrine, or autoimmune disease, as shown in Table I. The ethnic distribution among the groups was comparable, with all participants being Caucasian.

PBMCs from healthy and naive-to-treatment RRMS subjects were isolated after Ficoll–Hypaque gradient centrifugation (GE Healthcare). Peripheral CD4⁺ cells were purified (90–95% pure) from PBMCs of healthy and naive-to-treatment RRMS subjects by magnetic cell separation with Dynabeads Untouched Human CD4 T Cells Kit (Thermo Fisher Scientific). Peripheral Treg (CD4⁺CD25⁺CD127⁻) and Tconv (CD4⁺CD25⁻) cells were purified (90–95% pure) from PBMCs of healthy and naive-to-treatment RRMS subjects by magnetic cell separation with a Regulatory CD4⁺CD25⁺ T Cell Kit (Thermo Fisher Scientific). For isolation of monocytes, we stained PBMCs from healthy subjects with APC-conjugated anti-human CD14 (RMO52) (Beckman Coulter) and sorted monocytes by FACSJazz sorter (BD Biosciences) (cell purity > 95%). For proliferation assays, PBMCs were cultured for 60–72 h (1 × 10⁵ cells per well) in flat-bottom, 96-well plates (Corning Falcon) with RPMI 1640 medium supplemented with penicillin (100 UI/ml), streptomycin (100 μg/ml) (Thermo Fisher Scientific), and 5% autologous serum (AS) in the presence or in the absence of anti-CD3 mAb (OKT3) (0.1 μg/ml), treated or not with recombinant human ANXA1 (r-hANXA1) (GTP Technology) (20 μg/ml). Treg and Tconv cells were cultured (1 × 10⁴ cells per well) for 60–72 h in round-bottom, 96-well plates (Corning Falcon) in RPMI 1640 medium supplemented with 5% AS and stimulated with Dynabeads coated with mAb to CD3 plus mAb to CD28 (anti-CD3 + anti-CD28 mAbs) (0.5 beads per cell; Thermo Fisher Scientific), in the presence or in the absence of r-hANXA1 (20 μg/ml). After 60–72 h, [³H]thymidine (0.5 μCi/well; Amersham Pharmacia Biotech) was added to the cell cultures, and cells were harvested 12 h later. Radioactivity was measured with a β cell plate scintillation counter (Wallac). For cytofluorimetric analysis, PBMCs from healthy and naive-to-treatment RRMS subjects were cultured (2 × 10⁶ per well) in flat-bottom, 24-well plates (Corning Falcon) in RPMI 1640 medium supplemented with 5% AS for 12 and 36 h in the presence of anti-CD3 mAb (OKT3) (0.1 μg/ml), treated or not with r-hANXA1 (20 μg/ml). For immunofluorescence and confocal microscopy, CD4⁺ cells from healthy and naive-to-treatment RRMS subjects were cultured (2 × 10⁶ per well) in flat-bottom, 24-well plates (Corning Falcon) in RPMI 1640 medium supplemented with 5% AS for 12 h, in the presence of anti-CD3 + anti-CD28 mAbs (0.2 bead per cell; Thermo Fisher Scientific). For immunoblot analysis, monocytes were pretreated or not with r-hANXA1 for 1 h (2 × 10⁵ cells per well) in round-bottom, 96-well plates (Corning Falcon), with RPMI 1640 medium supplemented with penicillin (100 UI/ml) and streptomycin (100 μg/ml) (Thermo Fisher Scientific), and then stimulated for 1 h with LPS (1 μg/ml) (Sigma-Aldrich).

ANXA1 levels were determined by ELISA (Human Annexin A 1 ELISA Kit [CSB-E12155H]). Briefly, samples were diluted 1:2 and used on a precoated ELISA plate as per manufacturer’s instructions. Absorbance was measured using a microplate reader (model 500; Bio-Rad Laboratories, Hercules, CA) and sample readings were extrapolated against a concurrently run standard curve. The values were output as ANXA1 nanograms per milliliter.

Fluorescent bead technology was used to assay 50 μl of supernatants from PBMCs culture (24 h) from healthy and naive-to-treatment RRMS subjects. Cytokines levels (IL-1β, IL-10, IL-13, IL-6, IL-12, RANTES, Eotaxin, IL-17A, MIP-1α, GM-CSF, MIP-1β, MCP-1, IL-15, IL-5, IFN-γ, IFN-α, IL-1RA, TNF-α, IL-2, IL-7, IP-10, IL-2R, MIG, IL-4, and IL-8) were measured by using Human Cytokines Magnetic 25-Plex Panel (Novex, Life Technologies). Fluorescence was detected by using Luminex 200 System (Luminex, Austin, TX), and the resulting data were evaluated by using xPONENT Software Version 3.1 (Luminex).

Freshly isolated and OKT3-stimulated PBMCs from healthy and naive-to-treatment RRMS subjects were surface stained with the following mAbs: APC-H7–conjugated anti-human CD4 (RPA-T4) and PE-Cy7–conjugated anti-human CD25 (M-A251) (both from BD Biosciences). Thereafter, cells were washed, fixed, and permeabilized (Human FoxP3 Buffer Set; BD Pharmingen) and stained with following mAbs: unconjugated anti-human ANXA1, followed by incubation with secondary Ab Alexa Fluor 488–conjugated anti-rabbit IgG (H chain + L chain) (from Thermo Fisher Scientific), PE-conjugated anti-human FOXP3 (259D/C7), and Alexa Fluor 647–conjugated anti-human RORγt (Q21-559) (all from BD Biosciences). To evaluate ANXA1 surface levels, we stained PBMCs with anti-human ANXA1, followed by incubation with secondary Ab Alexa Fluor 488–conjugated anti-rabbit IgG (H chain + L chain) (from Thermo Fisher Scientific), APC-H7–conjugated anti-human CD4 (RPA-T4), and PE-Cy7–conjugated anti-human CD25 (M-A251) (from BD Biosciences). Thereafter, cells were washed, fixed, and permeabilized (Human FoxP3 Buffer Set; BD Biosciences) and were stained as previously described. For surface staining of PBMCs from healthy and naive-to-treatment RRMS after 36 h of OKT3 stimulation, we used the following mAbs: PE-Cy7–conjugated anti-human CD80 (L307.4), PE-Cy7–conjugated anti-human CD83 (HB15e), PE-conjugated anti-human CD86 (2331) (all from BD Biosciences); FITC-conjugated anti-human CD54 (MCA1615F) (from Bio-Rad Laboratories); and FITC-conjugated anti-human CD19 (J3-119) and PE-conjugated anti-human HLA-DR (Immu-357) (both from Beckman Coulter). For staining of PBMCs from healthy and naive-to-treatment RRMS with anti–human formyl peptide receptor–like 1 (FPRL1) (GM-1D6) from Aldevron, we incubated PBMCs with anti-human FPRL1 in the presence of Fc block (BD Biosciences) for 45 min at 4°C. Then, cells were washed with PBC buffer (PBS plus 0.15% BSA plus 1.32 mM CaCl₂) and incubated with secondary Ab, goat anti-mouse Alexa Fluor 488 (from Thermo Fisher Scientific) for 30 min at 4°C diluted 1:300 in PBS. Cells were analyzed with FACSCanto II (BD Biosciences) and FlowJo software (Tree Star). For staining of PBMCs from healthy and naive-to-treatment RRMS subjects after transmigration assay, we fixed cells for 30 min at room temperature with Staining Buffer (PBS plus 2% BSA) containing Alexa Fluor 648–conjugated anti-human CD4 (OKT4) and PE-Cy7–conjugated anti-human CD25 (M-A251), all from BD Biosciences. Then, cells were washed, centrifuged, and collected for the analysis (FACS Fortessa; BD Biosciences), followed by FlowJo software (Tree Star) analysis.

As a general model of human in vitro BBB, we used human brain microvascular endothelial cell line (hCMEC/D3), between 26th and 31st passage (25). hCMEC/D3 cells were cultured in tissue culture flasks coated for 2 h at 37°C in rat tail collagen type I solution in EBM-2 medium (Lonza) containing 2.5% (v/v) FBS, as recommended by the manufacturer. Cultures were maintained at 37°C in 5% CO₂, replaced with fresh medium every 3 d, and split when they reached 100% confluence. hCMEC/D3 cells were grown for 72 h on transwell polycarbonate filters (membrane diameter 6.5 mm and 5-μm porosity; Corning) previously coated with calf skin collagen type I (Sigma-Aldrich) and bovine plasma fibronectin (Sigma-Aldrich). EBM-2 medium, containing 5% of AS, was added to the lower compartment of transwell 30 min prior transmigration assay. Then, PBMCs (1 × 10⁶ cells) were suspended in EBM-2 medium containing 5% AS and added to the upper compartment of transwell in contact with hCMEC/D3 monolayer. In some experiments, PBMCs were stimulated for 16 h with anti-CD3 mAb (OKT3) (0.1 μg/ml) and further applied in 5% AS to the endothelial cells for 4 h. The transmigration assays were run during 4 h at 37°C at 5% CO₂. Transmigrated cells were collected from the bottom compartment by aspirating the medium. Cells were fixed with 2% paraformaldehyde for 10 min at room temperature, and staining was performed.

Freshly isolated Treg and Tconv cells from naive-to-treatment RRMS and age- and sex-matched healthy subjects were individually stimulated with anti-CD3 + anti-CD28 mAbs (0.1 beads per cell) for 36 h and subsequently pulled before protein extraction. PBMCs from naive-to-treatment RRMS and age- and sex-matched healthy subjects were individually stimulated with OKT3 from 30 min to 12 h, in the presence or in the absence of r-hANXA1 (20 μg/ml), and subsequently pulled before protein extraction. Monocytes were pretreated or not with r-hANXA1 (20 μg/ml) (1 h) and then stimulated for 1 h with LPS (1 μg/ml) (Sigma-Aldrich). Total cell lysates were obtained incubating cells for 20 min at 4°C in Radioimmunoprecipitation assay buffer (R0278; Sigma-Aldrich), plus SIGMAFAST Protease Inhibitor (S8820; Sigma-Aldrich) and Sigma Phosphatase Inhibitor (P5726; Sigma-Aldrich), and immunoblot analyses were performed as described (26). The mAbs used were the following: anti-ANXA1 (1:5000 dilution; Thermo Fisher Scientific), anti–phospho-AKT (Ser473), and AKT (1:1000 dilution; all from Cell Signaling Technology), anti–phospho-ZAP-70 (Tyr49)/Syk (Tyr526) (1:500 dilution; Cell Signaling Technology) and ZAP-70 (1:500 dilution; Santa Cruz Biotechnology), anti–phospho-STAT3 (3E2) and anti-STAT3 (79D7) (1:1000 dilution; all from Cell Signaling Technology), anti–phospho-LCK (Tyr 505) (1:1000 dilution; Cell Signaling Technology), and anti-ERK 1/2 (H-72) (1:1000 dilution; Santa Cruz Biotechnology) to normalize the amount of loaded protein. All filters were quantified by densitometry as described (26). Results were calculated as the densitometry of phosphorylated protein normalized to that of its total form (for AKT, ZAP-70, and STAT3) or normalized to densitometry of ERK 1/2 (for ANXA1 and LCK) and are presented relative to results obtained for either freshly isolated Treg and Tconv cells and 24 h TCR–stimulated Tconv cells from healthy subjects, PBMCs stimulated with OKT3 alone from healthy or naive-to-treatment RRMS subjects, or monocytes treated with LPS alone.

The metabolic profile was evaluated in PBMCs from healthy and naive-to-treatment RRMS subjects stimulated for 12 h with OKT3 (0.1 μg/ml), in the presence or in the absence of r-hANXA1 (20 μg/ml). Real-time measurements of extracellular acidification rate (ECAR) were made with an XFe96 Extracellular Flux Analyzer (Seahorse Bioscience). PBMCs were plated in XF96 plates (Seahorse Bioscience) at a density of 4 × 10⁵ cells per well and were cultured in RPMI 1640 medium supplemented with 5% AS. ECAR was measured in XF medium in basal conditions and in response to glucose (10 mM), oligomycin (4 μM), and 2-deoxy-d-glucose (2DG) (100 mM) (all from Sigma-Aldrich). Experiments with the Seahorse system were done with the following assay conditions: 3 min of mixture, 3 min of waiting, and 3 min of measurement. Metabolic parameters were then calculated from ECAR profile: basal glycolysis (without addition of any compound), basal glycolysis after glucose injection (after glucose addition), maximal glycolysis (after oligomycin addition), and glycolytic capacity (calculated as the difference between oligomycin-induced ECAR and 2DG-induced ECAR).

Confocal microscopy was performed on either freshly isolated or stimulated CD4⁺ cells with anti-CD3 + anti-CD28 mAbs (0.2 beads per cell; Thermo Fisher Scientific) for 12 h. Fixed cells were treated as described elsewhere (26) and incubated overnight at 4°C with primary Abs: mouse anti-human FOXP3 (236A/E7) from eBioscience and anti-human ANXA1 (71-3400; Thermo Fisher Scientific) diluted 1:100 in PBS. Then cells were washed and incubated for 1 h at room temperature with the secondary Abs: goat anti-mouse Alexa Fluor 488 and goat anti-rabbit Alexa Fluor 594, respectively (from Thermo Fisher Scientific), diluted 1:200 in PBS. For the assessment of ANXA1 surface levels, we stained CD4⁺ cells with anti-human ANXA1 diluted 1:100 in PBS for 20 min at 4°C, washed and incubated with the secondary Ab goat anti-rabbit Alexa Fluor 594 (from Thermo Fisher Scientific), diluted 1:200 in PBS for 1 h at 4°C. Subsequently, fixed cells were treated as described (26) and incubated overnight at 4°C with primary Ab: mouse anti-human FOXP3 (236A/E7) (from eBioscience), diluted 1:100 in PBS, washed in PBS, and incubated with the secondary Ab, goat anti-mouse Alexa Fluor 488 (from Thermo Fisher Scientific) for 1 h at room temperature diluted 1:200 in PBS. Nuclei were stained with Hoechst. A total of 1.5 × 10⁴ cells were seeded on multitest slide (MP Biomedicals). Immunofluorescence analyses were carried out on an inverted and motorized microscope (Axio Observer Z.1) equipped with a 63×/1.4 Plan-Apochromat Objective. The attached laser-scanning unit (LSM 700 4× pigtailed laser 405-488-555-639; Zeiss, Jena, Germany) enabled confocal imaging. For excitation, 405-, 488-, and 555-nm lasers were used. Fluorescence emission was revealed by Main Dichroic Beam Splitter and Variable Secondary Dichroic Beam Splitter. Triple staining fluorescence images were acquired separately using ZEN 2012 software in the blue, green, and red channels at a resolution of 1024 × 1024 pixels, with the confocal pinhole set to 1 Airy unit and then saved as TIFF files.

Statistical analyses were performed using GraphPad program (Abacus Concepts). Results were expressed as mean ± SEM or SD; a p value ≤ 0.05 was indicative of statistical significance. The nonparametric Wilcoxon matched-pairs signed-rank test and the Mann–Whitney U test were used. We used two-tailed tests for all analyses. To evaluate the relationship between the dependent variables (Expanded Disability Status Scale [EDSS] score and number of relapses prediagnosis) and ANXA1 levels, we computed the nonparametric Spearman rank correlation.

We measured ANXA1 plasma levels in n = 71 naive-to-treatment RRMS patients at diagnosis (Table I) to assess whether and how this molecule correlated with the clinical disease status. We evaluated two clinical parameters reflecting disease severity and activity: the Kurtzke EDSS score at baseline and the total number of relapses prediagnosis. Circulating ANXA1 plasma levels ranged from 0.252 to 11.45 ng/ml (median = 1.8; mean ± SD = 2.3 ± 1.9). Its levels were higher in subjects with a lower clinical score (EDSS score < 2) and decreased progressively in those with a higher score (EDSS > 3), reflecting a stronger disability in RRMS. We confirmed the inverse correlation between EDSS score and ANXA1 plasma levels by using the nonparametric Spearman rank correlation (R = 0.239, p = 0.039) (Fig. 1A). To further consolidate this finding, we evaluated the distribution of ANXA1 plasma levels in the same cohort of RRMS subjects stratified into three groups according to the number of relapses prediagnosis. We found that ANXA1 inversely correlated with the number of relapses prediagnosis, as patients with lower levels showed higher number of relapses (Fig. 1B). Taken together, these results suggest that a systemic reduction of ANXA1 secretion could be part of the inflammatory state of RRMS subjects.

It has been previously reported that innate immune cells, such as neutrophils and monocytes, release ANXA1 (27, 28); however, only few controversial data have been reported on T cells (17–22). To assess whether T cell–derived ANXA1 could be involved in the regulation of adaptive immunity, we evaluated its expression in different CD4⁺ T cell subsets of healthy individuals, in comparison with naive-to-treatment RRMS patients at diagnosis. We performed a double immunostaining which allowed to measure in parallel total (intracellular plus cell surface) or cell surface ANXA1 levels (29) in two different CD4⁺ T cell populations, distinct on the basis of their CD25 expression. Cumulative data of the flow cytometric analysis of freshly isolated PBMCs showed that CD4⁺CD25⁺ and CD4⁺CD25⁻ Tconv cells were enriched for total ANXA1, whereas they expressed low cell surface levels both in healthy and RRMS subjects; also, we observed that total ANXA1 levels were higher in Tconv cells, whereas its cell surface expression was higher in the CD4⁺CD25⁺ T cell subset (Fig. 2A, 2B, 2E, 2F, Supplemental Fig. 1A). Interestingly, both total and cell surface ANXA1 levels were significantly reduced in RRMS subjects when compared with healthy donors (Fig. 2A, 2B, 2E, 2F, Supplemental Fig. 1A) being the impairment more significant in the CD4⁺CD25⁺ T cell compartment (Fig. 2A, 2E, Supplemental Fig. 1A). Because the CD4⁺CD25⁺ T cell subset could represent a mixed cell population, we further analyzed ANXA1 expression by dissecting its levels in two divergent subsets specifically involved in the pathogenesis of MS, namely CD4⁺FOXP3⁺ Treg cells and proinflammatory T CD4⁺RORγt⁺ (Th17) cells (30, 31). Intriguingly, we observed that, among all the freshly isolated T cell subsets analyzed, the highest expression of ANXA1 was observed in the Th17 compartment, but this was significantly impaired in RRMS (Fig. 2C, Supplemental Fig. 1B). When we measured ANXA1 expression on cell surface, we observed that it was significantly higher in Th17 cells of healthy individuals as compared with RRMS subjects (Fig. 2G, Supplemental Fig. 1B). Analysis of Treg cells showed that total ANXA1 levels were higher in CD4⁺FOXP3⁻ cells, whereas its cell surface expression was higher in Treg cells (Fig. 2D, 2H, Supplemental Fig. 1C). Furthermore, ANXA1 expression was significantly reduced in both cellular compartments in RRMS subjects when compared with healthy donors (Fig. 2D, 2H, Supplemental Fig. 1C). These data were also confirmed by Western blot analysis performed in freshly isolated Treg (CD4⁺CD25⁺CD127⁻) and Tconv (CD4⁺CD25⁻) cells from RRMS subjects and healthy controls (Supplemental Fig. 2A, 2B). Confocal microscopy of CD4⁺ T cells freshly isolated from PBMCs of RRMS and healthy subjects, stained for ANXA1 and FOXP3 also confirmed that the expression of ANXA1 was reduced both in CD4⁺FOXP3⁻ and Treg cells during MS (Supplemental Fig. 1D). Taken together, our results suggest that a reduction of T cell–derived ANXA1 could be part of the proinflammatory phenomenon observed in MS.

Evidence for physiological function of ANXA1 in modulating inflammation emerged from studies involving ANXA1-null mice and ANXA1 neutralization strategies (19, 22, 32–34). Because the production of anti-inflammatory mediators during the early phases of TCR engagement has a crucial role in T cell fate decision (35), we measured ANXA1 production in CD4⁺CD25⁺, Th17, and Treg cells, to evaluate whether an impairment of the endogenous production could occur at this stage in RRMS. Briefly, freshly isolated PBMCs from healthy and RRMS subjects were activated in vitro with mAb anti-CD3 (OKT3, 0.1 μg/ml) and cultured in the presence of 5% AS to preserve their physiological milieu. Twelve hours after activation, CD4⁺ T cells were analyzed through flow cytometry gating strategy on the basis of their cell surface CD25 expression (reflecting the activation grade) as CD25-high (CD25^High), CD25-intermediate (CD25^Int) and CD25-low cells to evaluate how the activation status correlated with ANXA1 expression in health and in autoimmunity (Supplemental Fig. 3A). Cumulative data show that, in healthy conditions, the CD4⁺ T cell subset expressing highest levels of CD25 had strongest expression of ANXA1 (both as cell surface and total levels) (Fig. 3A, 3D, Supplemental Fig. 3A); the same was not true for RRMS subjects, as the CD25^High subset expressed significantly lower ANXA1 levels that were comparable to the CD25^Int subset (Fig. 3A, 3D, Supplemental Fig. 3A upper and middle panels). Moreover, after TCR engagement, the expression of ANXA1 in the two subsets representing the recently activated T cell pool (CD25^High and CD25^Int) was impaired in RRMS subjects as compared with healthy individuals (Fig. 3A, 3D, Supplemental Fig. 3A upper and middle panels). We also evaluated whether this could correspond to a different overall activation, evaluated as CD25 expression levels, but we did not observe significant differences in terms of CD25 expression in RRMS compared with healthy subjects (data not shown). Western blot analysis performed with Tconv cells stimulated with anti-CD3/CD28–coated beads for 24 and 36 h also confirmed a substantial reduction in the expression of ANXA1 in RRMS subjects (Supplemental Fig. 2C). All together our data confirmed an impaired T cell–derived ANXA1 production before and after TCR engagement during MS. We next asked whether a defect in ANXA1 production could occur in Th17 cells of RRMS subjects. TCR-stimulated PBMCs (12 h) showed significant impairment of ANXA1 expression in Th17 cells from RRMS subjects when compared with healthy individuals (Fig. 3B, 3E, Supplemental Fig. 3B). FACS analyses of the Treg cell compartment further confirmed the data obtained in freshly isolated Treg cells; indeed, we observed that total ANXA1 levels were higher in CD4⁺FOXP3⁻ cells, whereas its cell surface expression was higher in Treg cells (Supplemental Fig. 3C). Moreover, ANXA1 expression was significantly reduced in Treg cells of RRMS subjects when compared with healthy donors (Fig. 3C, 3F, Supplemental Fig. 3C). Confocal microscopy of TCR-stimulated CD4⁺ T cells stained for ANXA1 and FOXP3 also confirmed that the expression of ANXA1 was reduced in TCR-stimulated CD4⁺FOXP3⁻ and Treg cells of RRMS subjects (Supplemental Fig. 3D). Overall, these data suggest a profound impairment of ANXA1 production in RRMS.

Leukocytes infiltration into the brain is a hallmark of MS disease and several reports suggest a key role for ANXA1 in the regulation of immune cell migration (12, 36). We investigated the behavior of CD4⁺CD25⁻ (Tconv), CD4⁺CD25⁺, Th17, and Treg cells in contact with polarized human brain microvascular endothelial cell line (hCMEC/D3), an in vitro model of BBB (25). Freshly isolated PBMCs from healthy or naive-to-treatment RRMS subjects were put in contact with hCMEC/D3 cells for 4 h in 5% AS. We observed a significant increase of Tconv cell migration on the endothelium in RRMS as compared with healthy subjects (Fig. 4A). Noteworthy, we did not find a significant difference in CD4⁺CD25⁺ T cell migration (data not shown). The same trend was also maintained after 16 h of TCR stimulation (Supplemental Fig. 4A, 4B). Because Th17 cell migration into the CNS represent a major event in MS pathogenesis (37), we better explored this process and found increased migration of Th17 and reduced migration of Treg cells from RRMS subjects across the in vitro model of BBB (Fig. 4B, 4C). These findings suggest that impaired ANXA1 expression levels correlated with increased Tconv and Th17 cell migration in RRMS.

We next addressed whether ANXA1 could affect T cell function and activation. To this purpose, we performed a proliferation assay of human PBMCs stimulated with OKT3 in the presence or in the absence of r-hANXA1 (20 μg) in culture medium supplemented with 5% AS to better preserve the in vivo conditions. We observed that addition of exogenous r-hANXA1 to the culture significantly reduced the proliferation of TCR-engaged PBMCs in healthy subjects (∼50% of inhibition), whereas the reduction was not statistically significant in naive-to-treatment RRMS individuals (Fig. 5A). Because it is well known that T cells activate a glycolytic program to sustain their own proliferation (38), we evaluated the ECAR, reflecting the glycolytic activity of PBMCs isolated from healthy and naive-to-treatment RRMS subjects upon TCR stimulation (OKT3) in the same culture conditions (5% AS). First, we confirmed that naive-to-treatment RRMS patients had a lower glycolytic engagement compared with healthy controls (39), as testified by reduced glycolytic activity (Fig. 5B, 5C gray versus black lines and columns). Moreover, we found that ANXA1 addition affected glycolysis in TCR-stimulated PBMCs of healthy subjects, whereas it had no effect in RRMS subjects (Fig. 5B, 5C red versus black and blue versus gray lines and columns, respectively). We dissected this molecularly. We performed a time course stimulation (from 30 min to 12 h) for the kinetics of the main pathways related to TCR activation and cytokine/growth-factor signaling and inflammation. We found that r-hANXA1 treatment differently impacted on TCR activation in healthy (n = 6) versus RRMS (n = 6) subjects (Supplemental Fig. 4C, 4D). Indeed, after 30 min of OKT3 stimulation, r-hANXA1 reduced the activation of AKT in PBMCs from healthy, but not RRMS, subjects, where an increased induction was detected (Supplemental Fig. 4C). We also dissected the early events occurring during TCR-activation signaling and found that 2 h treatment with r-hANXA1 of PBMCs from healthy subjects corresponded to a reduction of ZAP-70 activity secondarily to a higher degree of phosphorylation of an inhibitory tyrosine of LCK (Tyr505). In contrast, r-hANXA1 treatment in PBMCs from RRMS subjects led to the induction of ZAP-70 activation because of the lack of LCK phosphorylation (Supplemental Fig. 4D). These results were also paralleled by a slight reduction in the secretion of proinflammatory cytokines (GM-CSF, IL-17A, IFN-α, IL-1RA, IP-10, MCP-1, RANTES, and TNF-α) in PBMCs of healthy individuals, but not in RRMS subjects (data not shown). We asked whether ANXA1 could directly affect the proliferation of purified Tconv and Treg cells, isolated from healthy and RRMS subjects and stimulated with anti-CD3/CD28–coated beads for 60–72 h in culture medium supplemented with 5% AS. As shown in Fig. 5D, 5E, ANXA1 did not affect the proliferation of TCR-stimulated Tconv and Treg cells, both in healthy and RRMS individuals. We also observed that ANXA1 did not affect Treg cell function, evaluated as their ability to suppress Tconv cell proliferation in coculture experiments, both in healthy and RRMS individuals (Fig. 5F). These results suggest that ANXA1 reduced T cell activation in healthy individuals, but this phenomenon was impaired in subjects with RRMS.

Because the inhibitory effect of ANXA1 was observed only in the presence of a mixed leukocyte population, this led us to hypothesize a direct action on APCs rather than on T cells. For this reason, we evaluated the expression of several activation markers involved in T cell priming and activation in PBLs, in monocytes and B cells, representing the main APCs present in the mixed TCR-stimulated PBMCs culture. We found that ANXA1 treatment reduced the expression of ICAM-1, MHC class II (MHC II), CD80, CD86, and CD83 in PBLs of healthy subjects, but not in RRMS subjects (Fig. 6A, 6B), likely explaining why we did not observe a significant reduction of T cell proliferation in these subjects (Fig. 5A). The same analysis was performed on B cells, but we did not observe any difference upon ANXA1 treatment in this cellular subset (Supplemental Fig. 4E). Further analysis on monocytes (gated on CD14⁺) also confirmed that ANXA1 profoundly reduced the expression of CD80, CD86, and CD83 on monocytes of healthy subjects, but not of RRMS subjects (Fig. 6C, 6D). Biochemical analysis performed on monocytes isolated from healthy subjects and stimulated in vitro with LPS (1 h) revealed an hyperactivation of the STAT3 pathway after an acute pretreatment (1 h) with ANXA1 (Fig. 6E). This suggests that ANXA1 could exert its anti-inflammatory role through a direct effect on STAT3 pathway, already known to impair the expression of costimulatory molecules on APCs with resulting decrease in T cell priming (40, 41). In this context, it has also recently been shown that IL-6 reduces the expression of MHC II and CD86 in human dendritic cells induced from PBMCs of healthy donors via STAT3 activation (42, 43). To this aim, we measured IL-6 production in culture supernatants and found an increased IL-6 secretion after ANXA1 treatment in TCR-stimulated PBMCs from healthy subjects (Fig. 6F); this, together with a direct ANXA1-dependent STAT3 activation, could contribute to sustain the STAT3 hyperactivation, leading to the impairment of monocyte maturation upon ANXA1 treatment. Moreover, we measured ANXA1 receptor, FPR2, to exclude that its reduced expression could account for the differential effect that ANXA1 exerted in healthy versus RRMS subjects. We analyzed PBMCs, CD14⁺, CD4⁺, and CD4⁺CD25⁺ isolated from healthy ad RRMS subjects and found a slight increase of FPR2 expression in RRMS, reaching the statistical significance only in the PBMCs population (Supplemental Fig. 4F–I). All our data reveal that, through a direct control of monocyte maturation, ANXA1 has a key role in the regulation of T lymphocyte activation in human healthy subjects, but this process is impaired during MS.

Pathophysiological mechanisms contributing to the development of MS have been intensively investigated, with a major focus on the role of immune system in the inflammatory process. An imbalance between pro- and anti-inflammatory actions is believed to be important in the development and progression of MS (44).

ANXA1, a glucocorticoid-inducible protein abundantly expressed in neutrophils and innate immune cells, has long been known as a potent anti-inflammatory mediator (11, 45). Several studies suggest that its mechanism of action is mainly related to a direct action on monocytes, involving both transcriptional changes as well as rapid posttranslational effects (28, 29, 46). Glucocorticoids directly attenuate leukocyte extravasation by inhibiting endothelial transcription of E-selectin and VCAM (VCAM-1) but also dampen TCR signaling, resulting in reduced proliferative responses and cytokine production, including reduced IL-2 secretion (47–49). A series of reports have also shown that glucocorticoids induce de novo synthesis as well as translocation of ANXA1 to the cell surface in PBMCs (50, 51). ANXA1 can act dually either as an endogenous signaling molecule or as a secreted mediator (52). The expression of this molecule and its release in the extracellular milieu may thus serve as an indicator of the endogenous anti-inflammatory state. However, only few and controversial studies have investigated whether ANXA1 impacts on human T cell function (17–22).

In this study, we measured circulating ANXA1 levels in a large cohort of naive-to-treatment RRMS subjects and related it to their clinical status. Among all clinical parameters evaluated (gadolinium-enhancing lesions, spinal cord lesions, number of relapses prediagnosis, and EDSS), we found that ANXA1 levels inversely correlated with EDSS and number of relapses (both parameters reflecting disease severity and clinical score) in our RRMS cohort of patients at diagnosis, before disease-modifying treatments. Because EDSS is critical in defining the clinical status of patients, as it includes multiple parameters of neural impairment that correlate with tissue and cell damage; this suggests a protective role in MS. We acknowledge, nonetheless, the limitation that EDSS does not identify mechanistically the individual contributing factors responsible for the overall assessment.

ANXA1 is expressed in human and mouse T cells but to a lower extent than in innate immune cells, such as neutrophils and monocytes (53, 54). We found that ANXA1 expression in different human T cell subsets, namely CD4⁺CD25⁺, CD4⁺CD25⁻ (Tconv), CD4⁺FOXP3⁺ (Treg), and CD4⁺RORγt⁺ (Th17), was significantly lower in RRMS subjects as compared with healthy controls, both as total content and cell surface level. Because ANXA1 is an endogenous anti-inflammatory mediator (and to ascertain that the differences we observed in freshly isolated T cells were not due to an increased recruitment/localization of ANXA1⁺ T cells into the CNS), we measured its expression levels in TCR-stimulated T cells. Compared with healthy controls, RRMS subjects exhibited a significant reduction in ANXA1 production (both as total content and cell surface expression) in Treg cells, Th17 cells, and CD25^High T cells. This, together with data obtained in freshly isolated T cells, confirmed a strong impairment of T cell–derived ANXA1 production during MS.

A key anti-inflammatory property of ANXA1 is its ability to inhibit leukocyte transmigration via a modulation of adhesion, molecule-based, leukocyte-endothelium interactions (12). Activated leukocytes (mainly neutrophils) mobilize their endogenous ANXA1 to the plasma membrane that is then secreted to promote leukocyte detachment from endothelial cells, thus functioning as a negative regulator of the transmigratory process (55, 56). Several studies reported that ANXA1 binds α4β1 integrin on leukocytes to prevent its interaction with VCAM-1 and/or to cause l-selectin shedding from leukocytes in a calcium-dependent manner, preventing the tethering, rolling, and firm adhesion of leukocytes to the endothelium necessary for transmigration (12, 57). Because T cell migration into the CNS is a crucial event for Ag-specific T cell activation in MS (58), we asked whether a reduced expression of ANXA1 could be associated with an increased T cell migration in RRMS subjects. We used an in vitro BBB model represented by human hCMEC/D3 cells polarized on transwell to allow correct tightness (25). Downmodulation of ANXA1 expression in Th17 cells corresponded to a higher degree of migration, likely contributing to facilitate the proinflammatory activity of these cells. On the contrary, downmodulation of ANXA1 in anti-inflammatory Treg cells was not beneficial to migration and capacity to control tissue damage. Being ANXA1, an endogenous molecule involved in the resolution of inflammation (10, 17, 18), these different effects could reflect divergent mechanisms of action of ANXA1 on proinflammatory Th17 and Treg cells. Both the reduced expression of ANXA1 on proinflammatory T cell subsets, together with the loss of ANXA1 in the brain microvascular endothelium previously described in MS subjects (23), could contribute to higher degree of leukocyte infiltration into the CNS.

Although the anti-inflammatory activity of exogenous ANXA1 has been established in different in vivo and in vitro models (19, 23, 34), a global analysis of effects in human T cell subsets have never been assessed. Several investigators demonstrated that ANXA1 exerts an inhibitory effect on adaptive immune response in vivo (59, 60) and during Ag-specific T cell activation in vitro (20, 33). Other studies revealed that ANXA1 promotes Th1 while inhibiting Th2 responses (21). In this study, we evaluated the impact of ANXA1 on human PBMCs, Tconv and Treg cells isolated from healthy or RRMS subjects and in vitro activated through the TCR. We found that ANXA1 impaired engagement of glycolysis and proliferation of in vitro TCR–stimulated PBMCs from healthy, but not naive-to-treatment RRMS, subjects. However, this different behavior in RRMS subjects was not secondary to a reduced expression of ANXA1 receptor (FPR2) on target cells, but to a different effect, ANXA1 exerted on TCR-related signaling and AKT pathway, as testified by reduced TCR/ZAP-70 activity and AKT phosphorylation observed in PBMCs of healthy, but not RRMS, subjects. Strikingly, ANXA1 neither affected the proliferation of TCR-stimulated Tconv and Treg cells nor Treg cell suppressive ability, both in healthy and naive-to-treatment RRMS subjects. Our data revealed that when ANXA1 was added to TCR-engaged PBMCs, there was an impairment of T cell activation and a reduction in the expression of several activation markers (ICAM-1, MHC II, CD80, CD86, and CD83), secondarily to an inhibitory effect on human monocytes, which are major APCs in PBMCs. It affected maturation of monocytes from human healthy subjects, as testified by a reduced expression of several costimulatory molecules, such as CD80, CD86, and CD83 (61, 62). This means that the reduced T cell proliferation observed is secondary to an altered T cell priming, and not to a direct effect on T cells. Interestingly, ANXA1 did not affect the expression of costimulatory molecules on monocytes of RRMS subjects, likely explaining why we did not detect a significant impairment of T cell activation in our cohort.

Treatment of monocytes from human healthy subjects with ANXA1 affected LPS-induced in vitro maturation. This phenomenon was secondary to a hyperactivation of the STAT3 pathway, known to exert an inhibitory effect during monocyte maturation (40, 41).

Our data highlight the significance of ANXA1 as part of the endogenous immunosuppressive circuit and couple its anti-inflammatory action to STAT3 signaling pathway. This ANXA1-dependent STAT3 hyperactivation is also supported by an increased IL-6 production in TCR-stimulated PBMCs cultured in the presence of ANXA1. Several reports indeed suggest that the IL-6/STAT3 pathway leads to the suppression of TLR4 ligand and LPS-induced APC maturation and activation (42, 43). Also, it has been shown that STAT3 hyperactivation suppresses CD86 and MHC II expression on human APCs, thus limiting Ag-specific CD4⁺ T cell activation (41, 63). Our results are in agreement with other reports showing that ANXA1 was able to elicit rapid anti-inflammatory effects by inhibiting neutrophil extravasation and activating a specific gene program that leads to the induction of an anti-inflammatory phenotype (56, 64). Indeed, ANXA1 is an endogenous determinant of proresolving properties of cAMP-elevating agents and cAMP-mimetic drugs (65). Also, thanks to its ability to bind α4β1 integrin (very late Ag 4, VLA4) on the cell surface, ANXA1 acts as an endogenous inhibitor of the leukocyte recruitment into the CNS (12, 36). This pathway is specifically targeted by a very effective second line drug for MS, the mAb natalizumab (66). Therefore, proresolving ANXA1-based therapies could control inflammation using the pathway shared with natalizumab.

Taken together, our results help to elucidate the role of ANXA1 on human T cells, both in healthy conditions and during MS, and pinpoint the key anti-inflammatory mediator ANXA1 as part of the counter-regulatory circuits elicited by interfering with the process of human monocyte activation and leukocyte recruitment.

By unveiling a systemic reduction of ANXA1, particularly evident in RRMS subjects with higher disability, our findings indicate that loss of ANXA1-mediated anti-inflammatory function could represent a novel mechanism of immune dysfunction in the pathogenesis of MS.

We apologize to colleagues whose work could not be cited in this study because of space limitations. We thank M. Montagna and S. De Simone for technical support, K. Schmierer (Blizard Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, U.K., and Clinical Board Medicine, The Royal London Hospital, Barts Health National Health Service Trust, London, U.K.) for supplying patients’ blood samples used in the studies of PBMC migration, and Ornella Moreggia (Dipartimento di Neurologia, Centro Regionale Sclerosi Multipla, Azienda Ospedaliera “A. Cardarelli,” Naples, Italy) for the collection of clinical information of RRMS subjects.

The authors have no financial conflicts of interest.