Unmodified or modified forms of fibrinogen (FIB) were used as antigens of interest in this study based on our previous work demonstrating macrophage activation in response to modified FIB antigens (45) and their role in mediating RA-related inflammation (27, 32). MAA modified and/or citrullinated (CIT) FIB antigens were prepared as previously described (45, 46). For MAA modification, native FIB (Cayman Chemical, Ann Arbor, MI, USA) was incubated at a 2:1 ratio of malondialdehyde (2 mM) to acetaldehyde (1 mM) in phosphate buffer (pH 7.2) at 37°C for three days and dialyzed against phosphate buffer for 24-h at 4°C. MAA adduction was confirmed by measuring fluorescence of the dihydropyridine ring structure (excitation 398 nm and emission 460 nm) using a Turner Biosystems (Sunnyvale, CA, USA) LS-5B spectrofluorometer.

For CIT modifications, native or MAA-modified FIB was incubated with rabbit skeletal peptidyl arginine deaminase-2 (PAD2; Sigma Aldrich, Saint Louis, MO, USA) and surplus PAD2 was precipitated out using soy bean trypsin inhibitor (Thermo Fischer, Waltham, MA, USA) (46). CIT modification was verified by ELISA and Western Blotting.

For dual modifications with MAA and CIT, FIB was first MAA- and then CIT-modified based on preliminary data demonstrating optimal cellular responses with modifications generated in this order (46). All modified antigens were sterile filtered through a 0.2µm filter and assessed for endotoxin contamination with levels universally below the detectable limit (46).

U-937 cells (human monocytic cell line, ATCC, Manassas, VA, USA) were maintained at 1x10⁶ cells/mL in RPMI medium at 37°C in a 5% CO₂ incubator. For all experiments, U-937 cells were first differentiated to M0 macrophages (Mϕ) by incubating overnight with 100ng/mL phorbol 12-myristate 13-acetate (PMA) (Sigma Aldrich), followed by a 24-h incubation in RPMI medium. PMA-activated cells were stimulated at 37°C with 25 μg/mL of FIB protein that was unmodified (FIB); MAA-modified (FIB-MAA); citrullinated (FIB-CIT); or MAA-modified and citrullinated (FIB-MAA-CIT). Post 24- and 48-h stimulation, macrophage supernatants (Mϕ-SN) were collected for further evaluation and HFLS cellular stimulations as detailed below.

Peripheral blood was collected from healthy donors (n=3). Informed consent was obtained from all participating individuals for these studies, which were approved by the University of Nebraska Medical Center Institutional Review Board (IRB#:608-22-EP). Peripheral blood mononuclear cells (PBMCs) were obtained using SepMate™ PBMC isolation tubes as per manufacturer’s protocol (StemCell™ Technologies, Vancouver, Canada) and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 2mM L-glutamine, 1% penicillin-streptomycin, and 0.05mM 2-mercaptoethanol. PBMCs were then differentiated into M0 macrophages (MP) using 100ng/mL of macrophage colony stimulating factor (M-CSF; Thermo Fischer) for 7 days (47). These differentiated macrophages were stimulated with unmodified and modified FIB antigens as outlined above. At 24- and 48-h, the supernatants (MP-SN) were collected for subsequent multiplex immunoassay analysis and HFLS cellular stimulations as outlined below.

Primary HFLS cells from healthy controls (HFLS control), osteoarthritis synovium (HFLS-OA), and RA synovium (HFLS-RA) were purchased from Cell Applications, Inc (San Diego, CA, USA). HFLS cells were grown to confluency in Synoviocyte Growth Medium (Cell Applications). Two independent sets of stimulations were performed on HFLS cells as outlined below.

RT-PCR was used to evaluate the effects of Mϕ-SN (or MP-SN) vs. direct antigen stimulations on HFLS cells. This approach was chosen as experiments using Mϕ-SN or MP-SN are confounded by the presence of soluble proteins in supernatants, prohibiting the attribution of cell source (Mϕ vs. HFLS) for these analytes. Therefore, we focused these studies on the measurement of gene expression rather than protein to minimize such confounding and to facilitate interpretation. RNA from HFLS cells was extracted using RNeasy® Mini Kit (Qiagen, Hilden, Germany), quantified using Nanodrop 2000c (Thermo Fischer). HFLS gene expression was assessed for 763 genes up to 8-h post-stimulation (as detailed above) using the Nanostring® Human Fibrosis Panel (Nanostring Technologies, Seattle, WA, USA). The 8 genes with the highest mRNA fold increase vs. FIB were then evaluated using real-time (RT)-PCR at 8- and 24-h post-stimulation of HFLS cells. For RT-PCR, extracted RNA was reverse transcribed to cDNA using a High-capacity RNA-to-cDNA™ kit (Applied Biosystems, Waltham, MA, USA). The following primers for the 8 genes (italicized below) were used in RT-PCR analysis; interleukin (IL)-1β, IL-6, type II collagen (COL2A1), metalloproteinase (MMP)-9, MMP-10, MMP-12, transforming growth factor (TGF)-β, and vimentin (VIM). The genes were categorized based on their role in fibroblast activation as pro-inflammatory (IL-1β, IL-6); pro-fibrotic (TGF-β, VIM); pro-invasive (MMP-9, MMP-10, MMP-12); and pro-chondrogenesis (COL2A1) (16, 48–54). The amplification reaction assays contained cDNA, TaqMan™ Gene Expression Master Mix (Applied Biosystems), and Fluorescein Amidite (FAM) tagged primers (Applied Biosystems). The 7500 Fast RT-PCR System (Applied Biosystems) software was used for amplification and analysis. Data were normalized to native FIB and presented as relative mRNA quantity (Rq). Examination of these soluble proteins (IL-1β, IL-6, MMP-9, MMP-10, MMP-12, TGF-β) was not performed due to interference/confounding of the soluble proteins present in Mϕ-SN or MP-SN as outlined above. In addition, extracellular matrix protein deposition of vimentin (gene: VIM) and type II collagen (gene: COL2A1) were examined further as outlined below, as these are only deposited from HFLS cells. Stimulation with media alone served as a negative control and demonstrated negligible upregulation of the genes examined (data not shown).

To quantify ECM protein deposition and cell-culture morphological changes following stimulation with unmodified and modified FIB, formalin-fixed HFLS cells were examined using fluorescent immunohistochemical (IHC) staining for the presence of type II collagen and vimentin. Although, type II collagen is recognized to be primarily found in cartilage, others have reported expression of this osteochondrogenic marker in aggressive phenotype of synovial fibroblasts and was thus included in these experiments (51, 52, 55). Additionally, vimentin is well recognized as an intracellular scaffold protein, it is deposited into the ECM under pathologic conditions such as those posed by RA (49, 56, 57) and was thus included in these experiments.

Briefly, HFLS cells were cultured in 24-well flat-bottom glass plates (Cellvis, Mountain View, CA, USA). The cells were then incubated for 48-h with unmodified/modified FIB antigens or Mϕ-SN as described above. Stimulation with media alone served as a negative control and demonstrated negligible effects on expression of ECM proteins (data not shown). Cells were then washed with PBS, blocked with 2% goat serum, and incubated overnight with a rabbit polyclonal Alexa Flour 594-conjugated antibody to collagen II (Bioss, Woburn, MA, USA) or a polyclonal rabbit Cy5-conjugated antibody to vimentin (Bioss). The cells were again washed, incubated with DAPI for nuclear staining for 2 minutes, mounted in fluoromount-G® (SouthernBiotech, Birmingham, AL, USA), and subjected to imaging using a Zeiss 710 Meta confocal laser-scanning microscope. Native FIB treated cells were used to determine microscope settings and all treatment group images were acquired using these identical parameters. Images were analyzed using Zen 2012 software (Zeiss, Jena, Germany) and quantified as mean pixel density (MPD) using ImageJ (National Institutes of Health, Bethesda, MD, USA).

HFLS-RA cells were cultured in a 6-well plate at 37°C and 5% CO₂ until reaching 90% confluency. The confluent cell monolayer was scratched with 200 μL sterile pipette tip followed by washing of cells with PBS to remove nonadherent cells. Thereafter, cells were treated with either unmodified/modified FIB antigens or Mϕ-SN as described above. The images were captured at t = 0h to record the initial area of the scratch and at t = 24h to evaluate effects of different treatments. Cell migration was monitored by collecting digitized images and 24 hours was selected as our experimental time point as this timeframe provided sufficient cell migration for detecting biological differences without entirely filling up the scratched area. Digitalized images were captured using the light microscopy setting on a Zeiss 710 Meta microscope. Images were analyzed using Image J software to measure the scratched or decellularized area (A). Cell migration was quantified as wound confluence using the following calculation (58): Wound confluence = [A(t₀) – A(t₂₄)]/[A(t₀)].

Given that the phosphorylated signaling pathways were similar to that reported by others following stimulation with platelet derived growth factor (PDGF)-BB (60–63), additional studies were undertaken to quantify this analyte in Mϕ-SN and MP-SN. Mϕ-SN (U-937 derived) and MP-SN (PBMC-derived) were assayed for the secretion of PDGF-BB isoform using a commercial immunoassay platform (Meso Scale Discovery, Rockville, MD, USA; lower limit of detection of 0.29 pg/ml). PDGF-BB isoform was examined in these studies due to its role in RA pathogenesis and activation of signaling mechanisms in fibroblasts as observed in other studies (64, 65). The assays were performed according to the manufacturer’s protocol and analyzed on the Meso QuickPlex SQ 120 imager (Meso Scale Discovery).

Detection of PDGF-BB in Mϕ-SN led to a separate set of experiments to assess its effects on HFLS-RA cells. The cells were stimulated with either 10ng/mL or 5pg/mL of PDGF-BB (Thermo Fisher). Post-stimulation, HFLS-RA cells were examined for ECM protein expression using fluorescent IHC and for activation of signaling pathways using Western Blot as described above.

All statistical analyses were completed using GraphPad Prism 9.0.0 Software (San Diego, CA, USA). Normality was confirmed using Shapiro-Wilk test (for n>3). Experimental data are presented as the mean ± standard error of the mean (SEM) based on ≥ 3 replicate experiments. Statistical differences among groups were completed using one-way analysis of variance (ANOVA) with Tukey’s post-hoc test (for n>3) or the Kruskal-Wallis test with Dunn’s multiple comparisons test (for n=3). Mann Whitney non-parametric test (for n=3) was used for comparisons between two treatment groups (MΦ-SN vs. direct stimulation). Differences were considered statistically significant when P values were less than 0.05.

Mϕ-SN collected post 48-h stimulation demonstrated greater mRNA fold change in HFLS-RA markers compared to 24-h collection of Mϕ-SN (data not shown). Hence Mϕ-SN collected at 48-h were used for all HFLS cell stimulations.

To examine the effects of Mϕ-SN and direct antigens on HFLS cells derived from patients with rheumatoid arthritis (HFLS-RA), genetic markers of an aggressive phenotype were evaluated at 8-h time point. After 8-h, stimulation of HFLS-RA with Mϕ-SN^FIB-CIT, and Mϕ-SN^FIB-MAA-CIT led to significant mRNA upregulation of the pro-inflammatory genes IL-1β and IL-6 (2- to 7-fold higher vs. Mϕ-SN^FIB; p<0.05) ( Figure 2 ). Additionally, expression of pro-invasiveness genes MMP-9, MMP-10 and MMP-12 were also significantly increased between 2- to 10-fold (p<0.05) following stimulation with Mϕ-SN^FIB-MAA-CIT, which universally yielded the highest mRNA fold increase ( Supplemental Figure 1 ). In contrast, direct stimulation of HFLS-RA with FIB-CIT promoted the highest mRNA fold increase for pro-inflammatory cytokines (IL-1β: 12-fold, p<0.001; IL-6: 26-fold, p<0.001 vs. FIB). Additionally, direct stimulation of HFLS-RA with FIB-CIT and FIB-MAA-CIT promoted increases in mRNA levels for pro-invasiveness and pro-chondrogenesis genes that were lower in magnitude than those observed following stimulation with Mϕ-SN. This included increases in mRNA for MMP-9 (FIB-CIT: 2-fold, p<0.05), MMP-10 (FIB-CIT: 3-fold, p<0.05), and MMP-12 (FIB-CIT: 7-fold, p<0.05; FIB-MAA-CIT: 7-fold, p<0.05) compared to FIB. Exposure of HFLS-RA to FIB-MAA did not significantly change mRNA levels of the genes. Similar expression patterns were observed in HFLS cells derived from patients with osteoarthritis (HFLS-OA) with smaller mRNA fold changes for the above markers ( Figure 2 ; Supplemental Figure 2 ).

To evaluate durability of responses observed after 8-h, additional measurements were completed following 24-h stimulation. After 24-h, COL2A1, MMP-9, MMP-10, MMP-12, TGF-β, and VIM genes were significantly upregulated, between 4- to 35-fold (p < 0.05) following stimulation of HFLS-RA with Mϕ-SN^FIB-MAA-CIT. Further, stimulation of HFLS-RA with Mϕ-SN^FIB-MAA-CIT resulted in the highest fold increases of aggressive phenotype genes based on expression of MMP-9, MMP-12, TGF-β, and VIM ( Figure 2 ; Supplemental Figure 1 ). The relative quantity of mRNA levels for the pro-inflammatory genes related to Mϕ-SN^FIB-CIT and Mϕ-SN^FIB-MAA-CIT stimulation remained elevated, but at a lower magnitude compared to values following 8-h, suggesting a decrease in the inflammatory response following an initial increase. Of note, direct stimulation of HFLS-RA at 24-h yielded smaller fold-changes on mRNA expression for all of the markers in comparison to 8-h. Further, only pro-inflammatory genes remained significantly upregulated in HFLS-RA cells after exposure to FIB-CIT as compared to the 8-h time point. Similar observations were made at 24-h for HFLS-OA cells ( Figure 2 ; Supplemental Figure 2 ). Together, these data suggest that HFLS-RA cells exposed to Mϕ-SN^FIB-MAA-CIT sustained gene upregulation of aggressive phenotype as compared to direct antigen stimulation. Additionally, these responses were lower in magnitude in HFLS-OA cells.

In separate experiments to determine whether freshly isolated human PBMCs reacted similarly to the U-937 cell line utilized above, HFLS-RA cells were stimulated with MP-SN activated with unmodified or modified FIB antigens and showed analogous increases in mRNA levels for IL-6, MMP-9, and MMP-10 as observed following HFLS-RA cell stimulation with the Mϕ-SN (U-937 derived Mϕ-SN) ( Figure 3A ; Supplemental Figure 3A ). Additionally, HFLS control cells from healthy synovium were used as a baseline control cell line. HFLS control cells amplified mRNA levels for IL-6, MMP-9, and MMP-10 in response to treatment with Mϕ-SN^FIB-CIT and Mϕ-SN^FIB-MAA-CIT, but at smaller fold changes compared to HFLS-RA cells ( Figure 3B ; Supplemental Figure 3B ). IL-6 (pro-inflammatory gene) and MMP-9 (pro-invasiveness gene) were chosen for these studies due to the highest Rq values in HFLS-RA cells, while MMP-10 (pro-invasiveness gene) was chosen based on the moderate (range from 3- to 5- fold) Rq amplification pattern.

In subsequent studies, we investigated the effect of the same HFLS stimulations on changes in cell morphology in addition to quantifying the cellular production of ECM proteins using both fluorescent immunohistochemistry (IHC) and Western Blot. Compared to HFLS-RA cells directly stimulated with FIB-CIT and FIB-MAA-CIT, HFLS-RA cells stimulated with Mϕ-SN^FIB-CIT, Mϕ-SN^FIB-MAA-CIT demonstrated an elongated and round, “spindle-shaped” morphology ( Figure 4 ) in addition to significant increases in the deposition of vimentin (p<0.0001 for Mϕ-SN^FIB-CIT, p<0.0001 for Mϕ-SN^FIB-MAA-CIT vs. direct stimulation with the same antigens) and type II collagen (p<0.0001 for Mϕ-SN^FIB-CIT, p<0.0001 for Mϕ-SN^FIB-MAA-CIT) ( Figures 5A, B ). Changes observed in cellular morphology and in the deposition of vimentin were highest following stimulation of HLFS-RA cells with Mϕ-SN^FIB-MAA-CIT (p<0.0001 vs. other antigens for vimentin). Stimulation with Mϕ-SN^FIB-MAA significantly increased HFLS-RA cellular deposition of vimentin (p<0.0001) and type II collagen (p<0.0001) compared to direct stimulation with the same antigen although yielding less striking changes in cellular morphology. By Western Blot, HFLS-RA cells demonstrated a similar pattern of expression to fluorescent IHC with higher expression of vimentin in treatments with Mϕ-SN compared to direct antigens (p<0.05 for Mϕ-SN^FIB-CIT, Mϕ-SN^FIB-MAA-CIT; Figures 5C, D ). Together these data suggest a unique HFLS-RA cellular morphology and increased protein deposition of vimentin and type II collagen in response to Mϕ-SN. Also, this aligns with RT-PCR data above demonstrating an increase in mRNA levels for vimentin (VIM) and type II collagen (COL2A1) genes.

To examine specificity of HFLS responses, the above experiments were then repeated using HFLS-OA cells. Unlike HFLS-RA, stimulation in HFLS-OA with Mϕ-SN did not result in changes in cellular morphology ( Figure 6 ). Stimulation of HFLS-OA cells with Mϕ-SN (following stimulation with modified FIB) also increased vimentin and type II collagen expression in comparison to direct antigen stimulation ( Figure 7 ). Type II collagen was probed in both HFLS-RA and HFLS-OA lysates with two different antibodies, polyclonal and monoclonal antibodies, and both antibodies to type II collagen were undetectable by Western Blot (data not shown).

To confirm the effect of Mϕ-SN on inducing aggressive phenotype in HFLS-RA cells, cellular migration was evaluated using the wound scratch assay. Stimulation of HFLS-RA with Mϕ-SN^FIB-CIT and Mϕ-SN^FIB-MAA-CIT promoted the highest wound confluence 0.68 ±0.02 and 0.69 ±0.05, respectively. Treatment of HFLS-RA with Mϕ-SN from the modified FIB antigens promoted higher wound confluence compared to direct antigens (p<0.05 for all; Figure 8 ).

To delineate the signaling mechanisms involved, c-Jun N-terminal protein kinase (JNK), p38 mitogen activated protein kinase (MAPK), p44/42 MAPK (Erk1/2), Rho-family-alpha serine/threonine-protein kinase (Akt), paxillin, and focal adhesion kinase (FAK) were investigated in HFLS-RA cells incubated with Mϕ-SN. These signaling pathways were investigated based on their broad regulation of an aggressive phenotype in HFLS-RA cells (16, 50, 59). Incubation of HFLS-RA cells at 1-h and 4-h were optimal for phosphorylation of signaling molecules, whereas with 15-min and 30-min incubation only p-Erk1/2 was detected, and the band intensities for p-Erk1/2 were lower than 1-h and 4-h incubation (data not shown). At the time points examined, phosphorylation of p38 MAPK, paxillin, and FAK was not detected in HFLS-RA cells (data not shown). At 1-h, stimulations of HFLS-RA cells with Mϕ-SN^FIB-MAA and Mϕ-SN^FIB-CIT yielded 4-fold magnitude increases in Erk1/2 phosphorylation (p-Erk1/2) (vs. Mϕ-SN^FIB), an increase not observed with JNK or Akt activation ( Figures 9A, C ). Exposure of HFLS-RA cells to Mϕ-SN^FIB-MAA-CIT increased phosphorylation (p-) of JNK (3-fold, p<0.05 vs. Mϕ-SN^FIB), Erk1/2 (6-fold, p<0.01), and Akt (2-fold, p<0.05) at 1-h. Similar magnitudes of fold increase in phosphorylation were observed in the positive control (treatment of HFLS-RA with LPS; data not shown). After incubation for 4-h, only Erk1/2 remained phosphorylated (p-Erk1/2) following stimulation of HFLS-RA with Mϕ-SN^FIB-CIT (6-fold, p<0.05) and Mϕ-SN^FIB-MAA-CIT (8-fold, p<0.01) ( Figures 9B, D ). Protein expression of JNK, Erk1/2, and Akt were similar across the treatment groups at 1- and 4-h period ( Figure 9 ). Notably, direct stimulation of HFLS-RA cells with the same antigens did not lead to phosphorylation of JNK, Erk1/2, and Akt at the time points examined above (data not shown). These experiments demonstrated that when exposed to Mϕ-SN, HFLS-RA cells activate signaling primarily through Erk1/2, JNK, and Akt pathways.

Due to the phosphorylation of JNK, Erk1/2, and Akt signaling molecules [pathways involved in PDGF-BB signaling (16, 50)] and transformation to an aggressive phenotype [PDGF-BB playing an established role in promoting this phenotype (60, 64)] in HFLS-RA cells stimulated with Mϕ-SN, we quantified PDGF-BB in Mϕ-SN (U-937 derived SN). Compared to stimulations with Mϕ-SN^FIB, PDGF-BB was increased in Mϕ-SN following Mϕ stimulation with both FIB-CIT (p<0.0001) and FIB-MAA (p<0.0001). The highest concentration of PDGF-BB was observed after stimulations with FIB-MAA-CIT (p<0.0001 vs. all other groups) ( Figure 10A ). A similar observation was made in the MP-SN derived from stimulated PBMCs ( Figure 10B ).

The presence of PDGF-BB in the Mϕ-SN prompted us to investigate the effect of PDGF-BB on HFLS-RA cellular morphology, ECM deposition, and signaling pathway activation. HFLS-RA cells were stimulated with 10ng and 5pg of PDGF-BB based on previous literature (60, 66) and mean PDGF-BB levels detected in Mϕ-SN, as described above. By fluorescent IHC, HFLS-RA cellular morphology appeared “spindle-shaped” only at 10ng PDGF-BB treatment ( Figure 11 ). Additionally, vimentin and type II collagen deposition were significantly upregulated with 10ng PDGF-BB stimulation and to a lesser extent with 5pg in comparison to treatment with media ( Figures 12A, B ). By Western Blot, vimentin expression was significantly upregulated in PDGF-BB treated groups (p<0.05 vs. media; Figures 12C, D ). For signaling pathways, HFLS-RA cells increased expression of p-JNK, p-Erk1/2, and p-Akt at both 10ng and 5pg PDGF-BB treatment concentrations (p<0.05) compared to media at 1-h incubation period ( Figure 13 ). The levels of JNK, Erk1/2, and Akt remained the same among the treatment groups. These results show that PDGF-BB is present in Mϕ-SN following stimulation with modified FIB and activates HLFS promoting an aggressive phenotype.