Immunogenic and Inflammatory Responses to Citrullinated Proteins Are Enhanced Following Modification with Malondialdehyde-Acetaldehyde Adducts

Protein antigens, modifications, and validation of co-modification.

Human serum albumin (HSA; Talecris Biotherapeutics, Research Triangle Park, NC) was citrullinated using rabbit skeletal peptidylarginine deiminase-2 (PAD2; Sigma, St. Louis, MO) and excess PAD2 removed using soy bean trypsin inhibitor as previously described (25). Citrullination was verified by coating ELISA plates with the modified antigen and using a monoclonal mouse anti-citrulline IgM antibody (F95; Millipore, Temecula, CA). Antibody detection was done using an HRP-tagged goat anti-mouse IgM antibody (Jackson ImmunoResearch, West Grove, PA). After washing the plates, an HRP substrate was added. MAA-adduction was completed by exposing purified HSA (2 mg) to 2 mM of MDA and 1.0 mM of AA in 0.1 M phosphate buffer (pH 7.2) at 37°C for 3 days, followed by dialysis against 3 changes of 0.1 M sodium phosphate buffer for 24 hours at 4°C (26). MAA adduction was determined by fluorescence of the dihydropyridine ring structure at an excitation wavelength of 398 nm and emission wavelength 460 nm using a Turner BioSystems LS-5B spectrofluorometer (Turner BioSystems, Sunnyvale, CA). Co-modified antigens were first MAA-modified and then citrullinated based on preliminary data demonstrating optimal immune responses (data not shown). As such, HSA that was co-modified contained the same number of MAA modifications as HSA that was MAA modified in isolation. As post-translational modifications can potentially result in aggregation, which alone can initiate immune responses, all proteins were sterile filtered through a 0.2 micron filter to eliminate aggregates (19). Although technically a foreign protein, unmodified HSA was used in all experiments as a negative control based on our previous work that has consistently shown negligible in vitro or in vivo responses following exposure to this antigen (19, 23).

To confirm that HSA was co-modified with MAA and citrulline, a capture assay was developed (Supplemental Figure 1). Briefly, an in-house monoclonal mouse anti-MAA IgG1 antibody was coated on 96-well plates at a concentration of 400 ng/ml and incubated overnight at 4°C. HSA modified with both MAA and citrulline was added at 2.5 μg/ml and serially diluted. The modified protein was then removed, washed and incubated with a monoclonal mouse anti-citrulline IgM antibody (F95; Millipore). Antibody detection was done using an HRP-tagged goat anti-mouse IgM antibody (Jackson ImmunoResearch, West Grove, PA) in the presence of HRP substrate. Optimal capture with this assay occurred at a protein concentration of approximately 1 μg/ml. To optimize the detection antibody concentration, a second experiment was completed whereby the wells were coated as previously described with anti-MAA antibody and a constant concentration of HSA-MAA-Cit (1 μg/ml) was added to each well. As above, the modified protein was removed, washed and incubated with different concentrations of monoclonal mouse anti-citrulline IgM antibody. For this assay, optimal detection occurred at an anti-citrulline antibody concentration of approximately 250 μg/ml.

Animal studies.

To examine autoantibody and T cell responses in relation to immunizations with modified antigen, male DBA/J1 mice (n=5 per group; Jackson Laboratories; Bar Harbor, ME) were given intraperitoneal injections once weekly for 5 weeks with 25 μg of: 1) HSA (control); 2) MAA-modified HSA (HSA-MAA); 3) citrullinated HSA (HSA-Cit); or 4) HSA that was MAA-modified and citrullinated (HSA-MAA-Cit). At week 6, animals were euthanized, serum collected and tested for ACPA (AU/ml) using two different ELISAs, the first plated with citrullinated mouse serum albumin (MSA; to mitigate any reactivity related specifically to HSA). To measure antibody responses directed only against the modification, antibody responses to unmodified MSA were subtracted to eliminate background reactivity. The second ACPA measure used a commercially available cyclic citrullinated protein (CCP) kit (Axis-Shield, Abbott, Chicago, IL), using an anti-mouse IgG detecting antibody (27). Recognizing the proprietary nature of the latter assay, we were not able to subtract out reactivity to unmodified peptides/proteins used in the anti-CCP antibody kit.

To quantify T cell responses generated against citrullinated antigen in the same groups of mice, splenic CD4⁺ T cells were isolated using the Miltenyi magnetic bead assay (Miltenyi Biotec, Bergisch Gladbach, Germany) and viability (>95%) was confirmed using a Cellometer with AOPI staining (Nexcelom Bioscience, Lawrence, MA). ELISpot plates were coated overnight at 4°C with 10 μg/ml of IFN-γ capture antibody diluted in 0.1 M NaHCO₃, pH 9.5. Plates were washed with sterile PBS and blocked with 200 μl/well of complete RPMI for 2 hours at 37°C. Citrullinated MSA (25 μg/ml; rather than HSA-Cit; to avoid measuring T cell responses specifically targeting HSA) was added to the T cells (3 × 10⁵ cells/well) in the presence of dendritic cells (2 × 10⁵ cells/well). After 18 to 24 hours of co-incubation at 37°C, T cells and dendritic cells were removed, plates washed with PBS/0.1% Tween-20, and 100 μl of biotinylated detection antibody to IFN-γ in PBS/0.1% BSA added for 2 hours at 37°C. Subsequently, plates were washed, and 100 μl of streptavidin-AP in PBS/0.1% BSA was added for 1 hour at 37°C. After washing, the assay was developed until the spots were visible using BCIP/NBT alkaline phosphatase substrate (Sigma). The number of spot-forming cells was enumerated in a blinded fashion using the ImmunoSpot ELISpot Reader (Cellular Technologies Limited, Cleveland, OH).

Cellular binding of antigen.

To examine whether co-modification impacts the cellular binding of citrullinated antigen, in vitro studies were completed using Chinese Hamster Ovary (CHO) cells transfected with a single scavenger receptor (SR): type B2 (SR-B2), endothelial cell type 1 (SREC-1) and toll-like receptor 4 (TLR-4). These studies were based on prior work demonstrating that the biologic effects of MAA are mediated primarily though interactions with these select SRs (23, 24). We expanded on this work by exploring whether cellular binding of citrullinated proteins (demonstrated to co-localize with MAA-modified antigen in synovium and lung (4, 22)) might also be facilitated by these receptors, particularly in the presence of MAA modification. CHO cells expressing a single SR were incubated with 25 μg of: 1) HSA; 2) HSA-MAA; 3) HSA-Cit; or 4) HSA-MAA-Cit for 90 minutes on ice. Cells were then washed and stained using a fluorescently labeled antibody (Alexa Fluor 568 rabbit anti-human albumin; Thermo Fisher Scientific, Waltham, MA). In separate experiments, the anti-HSA antibody was shown to bind equally to HSA regardless of the modification(s) present (data not shown). After 1 hour, cells were washed, fixed with paraformaldehyde and subjected to flow cytometry for quantification using FlowJo V10 (BD Biosciences, Palo Alto, CA). Data are presented as the median fluorescence intensity (MFI). The above studies were then repeated using human monocytes (THP-1; ATCC, Bethesda, MD), following activation of cells with phorbol 12-myristate 13-acetate (PMA), as THP-1 cells are known to express all of the aforementioned SRs (28–31). All reagents and antigens used in these experiments were made with lipopolysaccharide (LPS) free water, confirmed using limulus amoebacyte lysate assay testing.

Cellular inflammatory responses to antigen stimulation.

To examine the impact of these protein modifications on pro-inflammatory cellular responses in the absence of other activating factors (e.g. PMA), human THP-1 monocytes were incubated with 25 μg/ml of: 1) HSA; 2) HSA-MAA; 3) HSA-Cit; or 4) HSA-MAA-Cit for 24 hours at 37°C. Tumor necrosis factor (TNF)-α, interleukin (IL)-1β, and IL-10 were measured in culture supernatants by ELISA (BD Biosciences) as these cytokines are well known to be secreted by activated macrophages.

To further examine the effects of these protein modifications on the activation of monocytes/macrophages, studies were performed to evaluate adherence (a sign of maturation) and cell surface molecule expression. Importantly, when THP-1 cells are cultured in the absence of any stimulant, they maintain an immature monocyte phenotype (32). When activated (for example with PMA) these cells can mature to a macrophage phenotype and adhere to plastic, express unique surface molecules important in the removal of foreign materials (i.e. SRs), and even increase the cell surface expression of co-stimulatory molecules.

In initial experiments, THP-1 cells were adjusted to 2 × 10⁵ cells/ml in media and seeded into 25 cm² flasks containing 25 μg/ml of; HSA, HSA-MAA, HSA-Cit, HSA-MAA-Cit or PMA (100 ng/ml; as a positive control). Flasks were incubated at 37°C overnight, at which point non-adherent cells were removed and counted using a Cellometer with AOPI staining (Nexcelecom Bioscience). Flasks were then incubated with Accutase® (STEMCELL Technologies, Seattle, WA) to resuspend the adherent cells and then counted as above. The total number of adherent cells were divided by the total number of cells recovered with data presented as the proportion (%) of adherent cells.

In separate experiments, THP-1 cells were stimulated overnight with antigens and harvested as outlined above in the absence of PMA. Cells were then incubated for 30 minutes at 37°C with monoclonal anti-mouse antibodies to either FITC Mouse anti-human CD36 (SR-B2) or PerCP/Cyanine 5.5 Mouse anti-human CD14 (BD Biosciences), then washed and fixed in paraformaldehyde and subjected to flow cytometry. Antibodies were compensated using BD mouse IgG comp beads (BD Biosciences). Adherent and suspended cell populations were retained and analyzed separately using FlowJo V10. Data is presented as the MFI.

THP-1 cells were again prepared and harvested as above and subjected to flow cytometry to quantify adhesion and co-stimulatory molecules (HLA-DR and CD80) on adherent and non-adherent cells. Monoclonal antibodies used (BUV395 mouse anti-human CD11a, PE-Cy 5 Mouse Anti-Human HLA-DR, APC mouse anti-human CD83, PE-Cy7 mouse anti-human CD80, BV605 Mouse anti-human CD14) were compensated using BD mouse IgG comp beads (BD Biosciences). Data is again presented as MFI.

Statistical analyses.

Comparisons were completed using one-way ANOVA with Tukey’s post-hoc test to account for multiple comparisons. P values <0.05 after adjustment for multiple comparisons were considered to be statistically significant. All statistical analyses were completed using Sigma Stat (Systat Software, Inc, San Jose, CA).

Autoantibody responses.

Using the commercially available anti-CCP antibody kit, mice immunized with co-modified antigen (HSA-MAA-Cit) demonstrated significantly higher circulating ACPA concentrations than all other groups at 6 weeks (p < 0.002) (Figure 1). Anti-CCP antibody concentrations did not differ in mice immunized with either HSA-Cit or HSA-MAA compared to immunization with native protein. Using citrullinated mouse serum albumin as the target antigen, mice immunized with co-modified HSA (HSA-MAA-Cit) again demonstrated higher circulating ACPA concentrations than mice immunized with HSA alone (p < 0.0001) or mice immunized with only HSA-MAA (p = 0.042) or HSA-Cit (p = 0.003) (Figure 1). Of note, mice immunized with HSA-MAA demonstrated quantitatively higher serum ACPA concentrations using this assay than mice immunized with HSA-Cit, although this difference did not achieve statistical significance (p = 0.052).

T cell responses.

T cell responses were significantly higher in mice immunized with MAA-modified and/or citrullinated HSA compared to native HSA (p < 0.012) (Figure 1). Of the modified antigens used for immunization, the highest T cell responses to citrullinated protein were generated in mice immunized with HSA-MAA (p < 0.02 vs. all other groups). Of note, T cell responses following immunization with co-modified antigen (HSA-MAA-Cit) were quantitatively higher than responses following immunization with HSA-Cit, although this difference did not reach statistical significance (p = 0.080).

Cellular binding of modified antigen.

Cellular binding of unmodified and modified HSA was conducted using three different lines of CHO cells, each line expressing only a single SR (Figure 2). Cellular binding to unmodified HSA was low regardless of the transfected SR expressed. Binding of HSA-Cit was not significantly different than binding of unmodified HSA regardless of the transfected CHO cell examined. Cellular binding with co-modified antigen (HSA-MAA-Cit) was significantly higher than that observed for HSA-Cit and unmodified HSA in all transfected CHO cells while binding to HSA-MAA was higher than other antigens in all transfected cells. In activated human monocytes (THP-1 cells) expressing all of these SRs, citrullination of HSA increased cellular binding by 26-fold over unmodified HSA (p < 0.001) while MAA modification increased binding by 32-fold (p < 0.001 vs. unmodified HSA). Enhanced cellular binding was further increased with co-modification (p < 0.0001 vs. HSA-Cit or HSA-MAA).

Pro-inflammatory responses and cellular activation.

Human monocytes exposed to HSA-Cit or HSA-MAA demonstrated similar levels of TNFα, IL-1β, and IL-10 as those exposed to unmodified antigen (Figure 3). In contrast, TNFα (> 31-fold), IL-1β (> 2-fold), and IL-10 (> 5.6-fold) concentrations were significantly higher following exposure to co-modified antigen than following exposures with native HSA or HSA harboring only a single modification (p < 0.0001). Cytokine concentrations were universally higher for co-modified HSA than following exposure to PMA (p < 0.05 for all cytokines; data not shown).

Monocyte activation.

We next examined the impact of the different modified antigens on monocyte activation. Again using THP-1 cells, we specifically examined the impact of antigen stimulation on the proportion of cells becoming adherent as well as the expression of co-stimulatory molecules (CD80, HLA-DR) and SRs (CD14, SR-B2), measures indicative of cellular activation and/or macrophage transformation. As expected, THP-1 stimulation with HSA alone led to negligible adherence, a value that was not significantly changed following modification with either MAA or Cit alone (Figure 4). In contrast, the proportion of adherent cells increased markedly following stimulation with HSA-MAA-Cit (p < 0.05 vs. all other groups), although adherence was lower than that observed following PMA stimulation (p < 0.05 vs. all other groups).

In subsequent studies, we examined the effects of protein antigen stimulation on the THP-1 expression of SRs and co-stimulatory molecules, assessing adherent cells and those remaining in suspension separately. CD14 expression was significantly increased in adherent and suspended cells following exposure to HSA-MAA-Cit compared to stimulation with all other antigens (p ≤ 0.001) (Figure 5). SRB2 expression was quantitatively highest following stimulation with co-modified antigen, although differences did not reach statistical significance vs. HSA-MAA in adherent or suspended cells. Likewise, examining co-stimulatory responses separately in adherent and non-adherent cells, HLA-DR expression was significantly higher in magnitude following stimulation of co-modified HSA compared to other antigens (Figure 6). Following stimulation with co-modified HSA, CD80 expression was significantly than observed following HSA stimulation in adherent cells (p = 0.006) and significantly greater than either HSA stimulation (p = 0.004) or HSA-MAA stimulation (p = 0.015) in suspended cells.