Background
The earliest stages of rheumatoid arthritis (RA) are characterized by the presence of RA-specific autoantibodies such as rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPAs) years before the manifestation of clinical disease [
1]. In the Amsterdam health care region, ACPA-positive individuals with arthralgia have an approximately 50% chance of developing RA within 3–4 years [
2,
3]. Interestingly, during this at-risk phase synovial inflammation as determined by immunohistochemistry seems absent, suggesting that infiltration of the synovial tissue by inflammatory cells occurs in a later stage [
4,
5]. Because systemic autoimmunity seems to precede synovial tissue inflammation, other, as yet unidentified immune processes, possibly outside synovial tissues, are altered and contribute to disease development.
To effectively mount an adaptive immune response, secondary lymphoid tissues are essential. Animal models have shown phenotypic changes in the cellular compartment of peripheral lymph nodes (LNs) before the onset of arthritis [
6]. Recently, we detected altered frequencies of B cells, T-cell subsets and innate lymphoid cell subsets in LN biopsies of subjects with RA risk and patients with early-stage RA when compared with healthy control subjects [
7‐
10], indicating that LN activation was already present during the RA risk phase. Studies in mouse models revealed that lymph node stromal cells (LNSCs) play an important role in the regulation of T- and B-cell responses [
11,
12]. LNSCs physically construct the LN, and through production of chemokines and adhesion molecules, they guide immune cells within the LN [
13‐
15]. In addition, LNSCs produce cytokines important for lymphocyte activation, differentiation and survival [
16]. In mouse models, LNSCs have been found to induce peripheral T-cell tolerance by direct antigen presentation and clonal deletion as well as maintenance of regulatory T cells [
17‐
19]. Furthermore, during immune responses they are capable of suppressing T-cell proliferation independently of antigens [
19‐
21]. Accordingly, LNSCs are key players in immunity and tolerance. We hypothesise that malfunctioning of LNSCs leads to a microenvironment where immune responses are not properly controlled, which may lead to the activation of (autoreactive) lymphocytes and production of autoantibodies. LNSCs have been studied mainly in animal models, because so far human LNSCs have been obtained either from whole LNs removed during surgery or from deceased individuals [
22‐
24]. Isolating and sorting sufficient numbers of LNSCs directly ex vivo is technically challenging [
24]. We therefore aimed to develop an experimental model to allow research on human LNSCs during health and RA and to lay the foundation for further research on these immune-shaping cells.
Methods
Study subjects and lymph node biopsy sampling
Individuals with arthralgia and elevated immunoglobulin M (IgM)-RF and/or ACPA levels but without any evidence of arthritis upon examination were included (individuals with RA risk phase C/D) [
25]. Median follow-up of individuals with RA risk was 20.3 months (IQR 12.9–33.2), and none of the individuals with RA risk developed arthritis during this period. In addition, patients with RA with established disease based on fulfilment of the American College of Rheumatology/European League Against Rheumatism 2010 criteria [
26] and as assessed by the rheumatologist, as well as healthy control subjects without any joint complaints and without elevated IgM-RF and/or ACPA levels, were included. To be eligible, the healthy control subjects could not have an active viral infection or any history of autoimmunity or malignancy and no present or previous use of disease-modifying anti-rheumatic drugs, biologics or other experimental drugs. IgM-RF was measured using an IgM-RF enzyme-linked immunosorbent assay (ELISA) (upper limit of normal [ULN] 49 kU/L [kilo Unit/L]; HYCOR Biomedical, Garden Grove, CA, USA). ACPA were measured using the CCPlus anti-cyclic citrullinated peptide 2 ELISA (ULN 25 kAU/L [kilo arbitrary Unit/L]; Euro Diagnostica, Malmö, Sweden). The study was performed according to the principles of the Declaration of Helsinki and approved by the institutional medical ethical review board of the Academic Medical Centre, and all study subjects gave written informed consent. All study subjects underwent an ultrasound-guided inguinal LN needle core biopsy as previously described [
27]. Table
1 shows the demographics of the included subjects.
Table 1
Demographic data of study subjects
Female sex, n (%) | 9 (64) | 20 (87) | 17 (70) |
Median age, years, (IQR) | 29 (26–37) | 49 (35–57) | 56 (44–61) |
IgM-RF-positive, n (%) | 0 (0) | 10 (43) | 20 (3–107) |
IgM-RF level, kU/L, median (IQR) | – | 20 (3–107) | 131 (31–309) |
ACPA-positive, n (%) | 0 (0) | 13 (57) | 18 (75) |
Median ACPA level, kAU/L (IQR) | – | 43 (4–177) | 115 (21–924) |
IgM-RF and ACPA both positive, n (%) | 0 (0) | 0 (0) | 14 (58) |
Median DAS28, (IQR) | – | – | 5 (1–10)c |
Median ESR, mm/h (IQR) | – | 7 (2–10) | 11 (5–27)b |
Median CRP, mg/L (IQR) | 0.5 (0.3–1.2)c | 1.6 (0.9–3.2) | 4.6 (1.4–13)d |
Median TJC68 (IQR) | 0 (0) | 1.5 (0–4.5) | 9 (4–20)e |
Median SJC68 (IQR) | 0 (0) | 0 (0) | 5 (1–10)d |
Treatment, n (%) | | | 9 (39) |
Corticosteroids | | | 6 (26) |
NSAIDs | | | 4 (17)f |
DMARDs | | | 5 (22) |
Failed TNF inhibitor therapy | | | 5 (22) |
Lymph node stromal cell culture
After depletion of lymphocytes through a 70-μm cell strainer (Corning, Landsmeer, the Nederlands), the remaining stromal tissue of a freshly collected LN needle core biopsy was plated on a 6-well culture dish (CELLSTAR®; Greiner Bio-One/VWR, Alpen a/d Rijn, the Nederlands) (passage 0; P0). Complete cell culture medium was added. It consisted of DMEM, low glucose (Thermo Fisher Scientific, Landsmeer, the Netherlands) supplemented with 0.1% penicillin (Astellas Pharma Inc., Leiden, the Netherlands), 0.1% streptomycin, 0.05 mg/ml gentamicin, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer, and 2 mM l-glutamine (all from Thermo Fisher Scientific), as well as 10% FCS (GE Healthcare, Zeist, the Netherlands). Upon reaching confluence of > 80% cells, were passaged to a T75 tissue culture flask (P1) or into two T225 flasks (P2; both Corning® Costar®; Corning). Before being harvested, cells were washed with sterile warm PBS (Fresenius Kabi, 's-Hertogenbosch, the Netherlands) and incubated with 0.05% trypsin/5 mM ethylenediaminetetraacetic acid (Thermo Fisher Scientific) in PBS for 7 min at 37 °C. Subsequently, an equal amount of complete medium was added, after which the cell suspension was collected and centrifuged for 10 min at 1000 rpm at 4 °C. Cells were resuspended in cold complete medium and counted using trypan blue (Sigma-Aldrich, Zwijndrecht, the Nederlands) in a BRAND® Bürker Türk chamber (Sigma-Aldrich). Human LNSCs (passages 4 to 8) were seeded in a 24-well plate (30,000 cells/well) and stimulated with tumour necrosis factor-α (TNF-α) (5 ng/ml; Life Technologies, Landsmeer, the Nederlands) plus lymphotoxin α1β2 (50 ng/ml; R&D Systems, Abingdon, UK).
Flow cytometric analysis
Human LNSCs (passages 3 to 6) were cultured in a 6-well culture dish (100,000 cells/well). To harvest adherent cells, 1 ml of TrypLE™ Select reagent (Thermo Fisher Scientific) was added for 10 min at 37 °C. Subsequently, cells were washed in protein blocking agent (PBA) buffer (PBS containing 0.01% NaN3 and 0.5% bovine serum albumin [Sigma-Aldrich]) and stained for 30 min at room temperature protected from light using the following directly labelled antibodies: CD45 fluorescein isothiocyanate (FITC) (clone HI30; BD Diagnostics, Vianen, the Netherlands), podoplanin Alexa Fluor 647 (clone NC-08; BioLegend, London, UK), CD31 allophycocyanin (APC)-eFluor 780 (clone WM-59; eBioscience, Vienna, Austria), human leucocyte antigen A, B, C phycoerythrin-cyanine 7 (PE-Cy7, clone G46–2.6; BioLegend), or corresponding isotype controls. To examine the expression of podoplanin on LNSCs cultured over different passages, cells were stained for 1 h on ice with unconjugated anti-human podoplanin (clone NZ-1; AngioBio, Huissen, the Nederlands), washed, and subsequently incubated with polyclonal goat anti-rat IgG Alexa Fluor 647 (Thermo Fisher Scientific). Cells were washed in PBA and measured using a FACSCanto II flow cytometer (BD Biosciences, Vianen, the Nederlands). Data were analysed using FlowJo software (FlowJo, Ashland, OR, USA).
Co-cultures containing LNSCs and PBMCs and T-cell proliferation assay
LNSCs (passages 4 to 8) in amounts of 25,000, 10,000, 5000 or 1250 were seeded in duplicates in a 96-well flat-bottomed plate and allowed to rest overnight in DMEM complete culture medium. Subsequently, LNSCs were pre-treated with 50 ng/ml interferon-γ (IFN-γ) (eBioscience) for 48 h or refreshed with DMEM complete medium. Peripheral blood mononuclear cells (PBMCs) that had previously been isolated from healthy donors by using standard density gradient centrifugation and subsequently cryopreserved, were thawed and allowed to rest overnight at 37 °C in RPMI 1640 medium supplemented with 10% FCS (GE Healthcare), 0.1% penicillin (Astellas Pharma), 0.1% streptomycin, 10 mM HEPES buffer and 2 mM l-glutamine (all from Life Technologies). Then, PBMCs were washed and labelled with 2 μl of carboxyfluorescein diacetate succinimidyl ester (CFDA-SE) FITC (clone C1157; Life Technologies) in PBS for 8 min at 37 °C. After removing DMEM complete medium and washing LNSCs once with warm PBS, 50,000 labelled PBMCs in RPMI complete medium per 96-well chamber were added to LNSCs, resulting in ratios of 1:2, 1:5, 1:10 and 1:40 LNSCs to PBMCs. Simultaneously, PBMCs were stimulated with anti-CD3 (1:10,000, clone 1XE; Sanquin, Amsterdam, the Netherlands) and anti-CD28 (0.25 μg/ml, clone 15E8; Sanquin). Cultures were harvested 96 h later, washed with PBA buffer and stained for 30 min at room temperature protected from light using the following directly labelled antibodies: CD45 V500 (clone HI30; BD Biosciences), CD4 PE-Cy7 (clone SK3; eBioscience) and CD8a APC-eFluor 780 (clone SK1; eBioscience). Cells were washed in PBA and measured using the FACSCanto II flow cytometer. Data were analysed using FlowJo software. This methodology was set up by testing PBMCs isolated from four different healthy donors, whereas for the subsequent co-culture experiments, PBMCs from one healthy donor were selected to enable direct comparison between LNSCs from different donors.
qRT-PCR
Messenger RNA (mRNA) was isolated using the RNeasy Mini Kit or the RNeasy Micro Kit (Qiagen, Venlo, the Netherlands) according to the manufacturer’s instructions, including a DNase step to remove genomic DNA. Subsequently complementary DNA (cDNA) was prepared using the RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). qRT-PCR was performed using either TaqMan® Universal PCR Master Mix combined with TaqMan assays or SYBR® Green PCR Master Mix (all from Thermo Fisher Scientific) combined with primers made in-house (Thermo Fisher Scientific). The TaqMan assays and primer sequences are described in Additional file
1: Table S1. For detection, we used the StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific). Values for each gene were normalized to the expression level of 18S ribosomal RNA. An arbitrary calibrator sample was used for normalization. For calculating the relative quantity, the 2
−ΔΔCt comparative cycle threshold method was used for TaqMan assays, and a standard curve method was applied for SYBR Green assays.
Nitric oxide measurement
Nitric oxide (NO) was measured by evaluating the nitrite content in culture media using modified Griess reagent (G4410; Sigma-Aldrich) according to the manufacturer’s instructions. The co-culture supernatant (100 μl) of healthy individuals with RA risk and of patients with RA was mixed with the same volume of Griess reagent for 5 min, and absorbance was measured at 540 nm. A standard curve with increasing concentrations of sodium nitrite (NaNO2) was constructed in parallel and used for quantitation.
Statistics
Data are presented as median with IQR or mean with SD when normally distributed. Differences between study groups were analysed using the Kruskal-Wallis test followed by Dunn’s post hoc test or two-way analysis of variance followed by Dunnett’s multiple comparisons test, where appropriate. Prism software version 7.01 (GraphPad Software, La Jolla, CA, USA) was used for statistical analysis. P values < 0.05 were considered statistically significant.
Discussion
In this study we set up an experimental model using human LNSCs to allow research on the role of the human LN microenvironment during health and RA. The in vitro expanded human LNSCs express key characteristics as described earlier in mice [
16,
34]. During passaging these markers stay relatively stable, and their expression and induction are largely independent of podoplanin expression, the main marker for FRCs. The frequency of podoplanin-positive cells varied during culturing and between donors, but without any consistent trend towards loss of increase over passages. This variation was observed especially in those LNSC cultures containing high percentages of both FRC and DN cells, therefore probably reflecting a preferential outgrowth of one subset over the other. Mouse studies have shown that DN cells and FRCs resemble each other but have a differential expression of adhesion molecules and IL-7, which in mice are exclusively expressed in FRCs [
16]. Our study showed similarities, because basal IL-7 mRNA levels correlated positively with basal podoplanin mRNA levels, and ICAM-1 induction appeared to be co-regulated with podoplanin induction as also observed in mice [
34]. However, we also demonstrate that LNSC cultures consisting of cells with low podoplanin expression can also express relatively high levels of IL-7. In addition, in human LNSC cultures DN cells are relatively more abundant than FRCs as described before [
24]. Together, these findings suggest a comparable role of DNs and FRCs in humans, although additional studies using isolated LNSC subsets are essential to prove this.
Interestingly, the expression of CXCL12, a B-cell chemoattractant [
35], was significantly lower in LNSCs derived from individuals with RA risk than in healthy control subjects. CXCL12
+ stromal cells derived from both bone marrow and tonsils (LN-like FRCs) of healthy donors can attract malignant B cells and appear to enhance the survival of follicular lymphoma B cells compared with healthy B cells isolated from blood [
36]. Similarly, B-cell survival in the synovium is dependent on IL-6 and CXCL12, which are overexpressed by RA synovial fibroblasts [
37]. The lower CXCL12 expression in LNSCs of individuals with RA risk might reflect an attempt to prevent autoreactive B cells from accessing the LN and impair their survival. We also detected a lower induction of CCL19 and CXCL13 after stimulation with TNF-α plus lymphotoxin α
1β
2 in LNSCs derived from patients with RA. Overall, these data suggest that LNSCs of individuals with RA-specific systemic autoimmunity display an altered chemokine profile, which may lead to disturbed trafficking of lymphocytes within the LN. Future studies are needed to confirm this and to investigate the mechanism by which chemokine production is disturbed in autoimmune LNSCs.
Next to lymphocyte trafficking and survival, LNSCs play a crucial role in regulating adaptive immune responses. LNSCs prevent extensive T-cell proliferation and thereby dampen immune responses through the release of NO in a tightly regulated and contact-dependent manner [
20,
32,
33]. Our results diverge from murine studies where low numbers of LNSCs already lead to full inhibition and NO plays a crucial role [
20,
32,
33]. We observed incomplete suppression when an LNSC/T-cell ratio of 1:2 was used and found that proliferation was even increased in LNSC/T-cell ratios of 1:5 and 1:10. Even though NO levels are higher in an LNSC/T cell ratio of 1:2, no differences were observed between donor groups. Therefore, the altered suppressive effect observed in RA LNSCs at the ratio of 1:2 is probably not dependent on changes in NO production. Likewise, the expression of IL-7, measured at P2 under homeostatic conditions that might drive T-cell proliferation, was not differentially expressed between donor groups or correlated with T-cell proliferation (data not shown) [
38]. Even though IFN-γ signalling on LNSCs is crucial for NO production, exogenous IFN-γ alone increases only
NOS2, the gene encoding inducible nitric oxidase synthase, but it fails to increase NO or nitrite in culture medium [
32]. Furthermore, Transwell experiments show that LNSC-T-cell contact is needed to induce NO production and consequently T-cell suppression [
20,
32,
33]. Taken together, this points towards additional components in this pathway derived from intimate cell contact to constrain T-cell proliferation, but future research is needed to formally prove this contact dependency in human co-cultures. Furthermore, in contrast to murine studies, which mostly use sorted and autologous cells derived from T-cell receptor (TCR) transgenic mice, we used an allogeneic co-culture system. Missing TCR-major histocompatibility complex (TCR-MHC) interaction might diminish close cell-cell contact, or mismatched TCR-MHC might additionally trigger T-cell proliferation [
39]. However, our observations in healthy LNSCs are in line with data derived from mesenchymal stem cells (MSCs). In mice as well as in humans, using an allogeneic system containing MSCs and T cells, suppression of T-cell proliferation was observed only when relatively high numbers of MSCs were used, whereas relatively low numbers of MSCs supported T-cell proliferation [
40,
41]. Furthermore, co-culture of allogeneic T cells with human MSCs in MSC/T-cell ratios of 1:4 and 1:40 increased the numbers of FoxP3-expressing cells [
40], and maintenance of regulatory T cells by LNSCs has also been observed in mice [
17]. It will be interesting to investigate in future experiments whether human LNSCs also play a role in maintenance of regulatory T cells and whether this process is disturbed in LNSCs from patients with RA. However, these experiments are highly challenging, because for human studies, knowledge is lacking on well-defined self-antigens expressed by human LNSCs and the availability of corresponding autoreactive human T cells.
The suppression of T cells in low LNSC/T-cell ratios (1:2) and the immunostimulatory effect in higher ratios (1:5 and 1:10) was seen mostly in LNSCs from healthy individuals. In this study, we show, for the first time to our knowledge, that this bipolar behaviour depending on LNSC/T-cell ratio is less maintained in LNSCs derived from patients with RA. It is tempting to speculate that reduced inhibition of T cells might result in less inhibition of self-reactive T cells and that reduced proliferation or induction of regulatory T cells leads to loss of tolerance. This and the cellular mechanism behind the potential exhausted state and aberrant function of RA LNSCs remain to be determined in future studies.
Conclusions
Overall, we developed, for the first time to our knowledge, an experimental model to study the role of human LNSCs during the earliest phases of RA. Our exploratory study shows differences between the LN microenvironment of individuals with RA risk, patients with RA and healthy control subjects. To study in detail their immunoregulatory function, in vitro expansion of LNSCs is required. Because it is difficult to obtain LN biopsies from a large cohort of individuals with RA risk and patients with RA, and because the culture of human LNSCs is very time-consuming owing to their slow growth, the number of donors analysed in this study is relatively low. Also, because of the high inter-individual variation in podoplanin expression, the contribution of different LNSC subsets to the findings reported here remains to be further explored. The translation from in vitro results to in vivo relevance should be demonstrated by using mouse models or through targeted intervention studies in patients. However, using this in vitro model, we can start delineating the role of human LNSCs in T-cell-mediated B-cell responses during the earliest phases of RA, which ultimately may lead to the identification of innovative targets for immunomodulation.
Acknowledgements
We thank the study participants in the study; the radiology department at the AMC for lymph node sampling; the flow cytometry facility in the haematology department at AMC, especially J. A. Dobber (Laboratory of Hematology, AMC); and the AMC Clinical Immunology and Rheumatology department, especially M. J. H. de Hair and M. Safy for patient recruitment and G. Rikken and D. Drop for sample processing.