Adalimumab regulates intracellular TNFα production in patients with rheumatoid arthritis

Introduction Adalimumab is a fully human anti–tumor necrosis factor α (anti-TNFα) monoclonal antibody that specifically blocks the interaction of TNFα with its receptors. It binds both soluble and transmembrane TNFα. We hypothesized that blocking these TNFα signals regulates the altered TNFα production in rheumatoid arthritis (RA) patients. Methods We compared, by flow cytometry, Toll-like receptor induction levels of membrane and intracellular TNFα in monocytes (iTNFα + CD14+ cells) from 12 patients before and after adalimumab treatment with those from 5 healthy donors. Results Before starting the treatment, the percentage of iTNFα+ CD14+ cells in the RA patients was significantly lower than that in healthy donors (mean ± SEM = 33.16 ± 4.82% vs 66.51 ± 2.4%, P < 0.001). When we added in vitro TNFα to healthy donor culture cells, levels of iTNFα+ CD14+ cells decreased, suggesting that the TNFα signal was responsible for the iTNFα+ CD14+ cell downregulation observed in the RA patients. After 2, 6 and 12 adalimumab injections, we observed significant blocking of membrane and soluble TNFα and a progressive increase in iTNFα+ CD14+ cells in ten patients with a good to moderate response as defined by the European League Against Rheumatism (EULAR) criteria. Levels of iTNFα+ CD14+ cells after 12 injections in these 10 patients were comparable to levels in healthy donors. In two patients, iTNFα+ CD14+ cell upregulation was not observed, and their EULAR-defined responses had not improved. The first patient developed antiadalimumab antibodies, explaining why adalimumab was not able to block membrane and soluble TNFα. In the second patient, adalimumab was discontinued because of adverse effects, which led to a decrease in iTNFα+ CD14+ cells to levels measured before treatment. Conclusions Our findings suggest that adalimumab treatment in RA patients can return iTNFα levels to those of healthy donors. This effect was not observed in the presence of neutralizing antiadalimumab antibodies.


Introduction
Tumor necrosis factor α (TNFα) is a proinflammatory cytokine produced mainly by activated monocytes, macrophages, T lymphocytes and natural killer (NK) cells. In target cells, this cytokine plays a key role in apoptosis, cell survival, immunity and inflammation [1][2][3]. TNFα is initially synthesized and expressed as a transmembrane protein. Its extracellular proportion is released in the form of a soluble 17 kDa molecule when it is cleaved by a metalloproteinase TNFα-converting enzyme (TACE) [4].
The level of TNFα in the synovial fluid in rheumatoid arthritis (RA) is high [5,6]. In the synovia, TNFα contributes to joint destruction by attracting leukocytes, inducing inflammatory cytokines, upregulating adhesion molecules on endothelial cells and activating the synthesis of metalloproteinases in synovial macrophages, fibroblasts and chondrocytes [7][8][9][10]. TNFα also plays a role in osteoclastic bone resorption. It stimulates osteoclastogenesis by differentiating progenitor cells and enhancing the expression of receptor activator of nuclear factor κB ligand [11].
In view of these findings, it is not surprising that TNFα has become a strategic target in the treatment of RA patients. Adalimumab is a fully human neutralizing anti-TNFα monoclonal antibody that specifically blocks the interaction of TNFα with p55 and p75 cell-surface TNFα receptors [12]. By blocking TNFα, adalimumab can attenuate cartilage and bone destruction partially through the downregulation of matrix metalloproteinases [13]. Moreover, adalimumab can reduce acute-phase reactants of inflammation [14], inflammatory cytokines [10] and adhesion molecules responsible for leukocyte migration [13]. Studies have shown that adalimumab is effective in preventing joint damage in early RA [15] and improving clinical and laboratory parameters, emphasizing the pivotal role of TNFα in this pathology. However, not all RA patients treated with adalimumab show this clinical response [16].
Although adalimumab has been found to block soluble and transmembrane TNFα [17], whether adalimumab can regulate the production of intracellular TNFα (intracellular TNFα) on monocytes is unknown. Our aim in this study was to determine whether blocking TNFα signals regulates the TNFα production in patients with RA.

Samples
Heparinized blood obtained from healthy donors (HDs) (n = 5), patients with active RA who had never received biological therapy (n = 12), RA patients in remission or with low activity who were being treated with methotrexate (n = 3) and similar patients treated with infliximab (n = 3) was collected in BD Vacutainer tubes (BD Pharmingen, Franklin Lakes, NJ, USA). Diagnosis of RA was based on the American College of Rheumatology criteria [18]. Disease activity was measured using the Disease Activity Score in 28 joints erythrocyte sedimentation rate (DAS28-ESR) [19]. Table 1 shows demographic data, clinical parameters and laboratory values of patients with active RA who were receiving adalimumab treatment every 2 weeks. We collected blood from RA patients after two (4 weeks), six (12 weeks) and twelve (24 weeks) injections of adalimumab. To determine the effectiveness of adalimumab treatment, we evaluated clinical improvement using the European League Against Rheumatism (EULAR) response criteria [20]. We also collected blood from three methotrexate-treated RA patients over the course of more than 3 years (ESR (mean ± SD) = 36.33 ± 16.65 mm/h, Creactive protein (CRP) = 18.13 ± 15.15 mg/L) and three RA patients who had been treated with infliximab for more than 5 years (ESR = 42 ± 30.34 mm/h, CRP = 9.17 ± 3.32 mg/L). Written informed consent was obtained from the participants, and ethical approval of the study was granted by the institutional ethics committee of the Hospital de la Santa Creu i Sant Pau.

Determination of tumor necrosis factor α levels in cultures supernatants
Whole blood of HDs and RA patients was cultured with medium, LPS (1 μg/ml) or LTA (10 μg/ml) at 5% CO 2 and 37°C. After 24 hours in culture, supernatants were collected. Soluble TNFα levels were determined using a specific enzyme-linked immunosorbent assay (ELISA) kit according to the manufacturer's instructions (BD Biosciences). TNFα was quantified with standard curves, and the limit of detection was 7.5 pg/ml. Adalimumab injections blocked soluble TNFα, and only free soluble TNFα was detected.

Determination of adalimumab levels and antiadalimumab antibodies in serum
Serum from RA patients was collected and kept at −80°C until use. Adalimumab levels and antiadalimumab antibodies were determined using a specific ELISA kit according to the manufacturer's instructions (Proteomika, Derio, Spain). The limit of detection was 0.15 ng/ml for adalimumab and 0.4 AU/ml for antiadalimumab antibodies.

Flow cytometry analysis
iTNFα and mTNFα were analyzed on CD14+ monocytes, and iTNFα was analyzed in neutrophils and lymphocytes gated according to forward-scatter and side-scatter parameters in CD14− cells using a Beckman Coulter FC500 flow cytometer (Beckman Coulter, Barcelona, Spain). We calculated the percentages of positive cells (% cells) and geometric mean fluorescence intensity (GMFI) of each marker using CXP Software 2.2 (Beckman Coulter). The integrated geometric mean fluorescence intensity (iGMFI) was used to determine the amount of mTNFα produced by CD14+ cells and was calculated by multiplying the percentage of positive cytokine-producing cells by the GMFI [21].

Statistical analysis
Statistical analyses were performed using paired t-tests, t-tests and Spearman's correlation in GraphPad Prism 5 software (GraphPad Software, La Jolla, CA, USA). Data are presented as mean and ± SEM. P-values <0.05 were considered statistically significant.

Results
Tumor necrosis factor α production in rheumatoid arthritis patients iTNFα and mTNFα levels in CD14+ cells and TNFα levels in supernatants of LPS (TLR4 ligand) and LTA (TLR2 ligand) of HDs were compared with those of RA patients. For this purpose, whole-blood cells obtained from HDs and RA patients were cultured with TLR ligands before adalimumab therapy was started.
The percentages of mTNFα+ CD14+ cells and the levels of supernatant TNFα in RA patients did not differ from those of HDs after TLR ligand cultures ( Figure 1B to D).
To determine whether the increases in iTNFα+ CD14+ cells observed in the RA patients were due to adalimumab treatment or a consequence of the remission state, we analyzed RA patients in methotrexate-induced remission. Percentages of iTNFα+ CD14+ cells in RA patients in methotrexate-induced remission were significantly lower than those in RA patients after 12 injections of adalimumab treatment (28.25 ± 10.5% and 60.68 ± 5.87%, respectively, P < 0.05) (Figure 2A). These low percentages of iTNFα+ CD14+ cells in methotrexate-treated patients were not restored by overnight culture with adalimumab (23.89 ± 13.94%).
Adalimumab was not effective in two patients. Patient 11 presented a lower percentage of iTNFα+ CD14+ cells after LPS and LTA cultures (data not shown) at all postinjection time points than HDs ( Figure 2E). Another patient, patient 7, received the third injection after a 4-week delay due to an infection. The patient then presented adverse effects (allergy), and adalimumab was discontinued. In this patient, the increased percentage of iTNFα+ CD14+ cells and higher EULAR-defined response seen after the first two injections were not observed after the third injection ( Figure 2D). Consistent with our observations, DAS28 levels and ESR were inversely correlated with the percentage of iTNFα+ CD14+ cells along treatment (DAS28-ESR = R = −0.445, P = 0.0095; ESR = R = −0.563, P < 0.0001) in good to moderate EULARdefined response patients.

Adalimumab levels and antiadalimumab antibodies in serum of rheumatoid arthritis patients
In view of the findings described above, we next compared adalimumab levels and antiadalimumab antibodies in patients 7 and 11 with those in RA patients with good to moderate EULAR-defined responses. The seven patients with good to moderate EULAR-defined responses presented therapeutic levels of adalimumab (>1 μg/ml) [22] after 12 injections and did not have detectable antiadalimumab antibody levels in serum ( Table 2).
The two patients (patients 7 and 11) who did not show a EULAR-defined response presented different patterns. Although patient 7 showed therapeutic levels of adalimumab after two and three injections, therapy was suspended after the third injection due to allergic reactions. Consequently, at 24 weeks after initiation of treatment (14 weeks from the last adalimumab injection), adalimumab levels were undetectable, coinciding with the downregulation of iTNFα+ CD14+ percentages to levels comparable to those before treatment. Patient 11 presented no clinical response to adalimumab at any moment during follow-up, coinciding with undetectable levels of adalimumab in serum after 2, 6 and 12 injections. These undetectable levels of adalimumab were a consequence of the high titers of antiadalimumab antibodies after the second injection (Table 2). No patients had detectable levels of adalimumab antibodies in serum before adalimumab treatment. (E) This graph shows percentage of iTNFα+ CD14+ cells in LPS culture before and after 2, 6 and 12 injections of adalimumab into patient 11. A paired t-test was used for statistical analysis between RA patients at different time points. A t-test was used for statistical analysis between RA patients and HDs. *P < 0.05, ***P < 0.001.
Regulation of intracellular tumor necrosis factor α production by tumor necrosis factor α signaling We next analyzed how TNFα-TNF receptor signaling downregulated iTNFα production in CD14+ cells. Overnight preincubation of whole-blood cells from HDs with recombinant TNFα (6 ng/ml) decreased the percentage of iTNFα+ CD14+ cells in a 4-hour LPS culture (88.87 ± 6.01% for medium culture vs 11.49+2.11 for TNFα culture; P < 0.01).
To understand the effect of adalimumab on iTNFα regulation by TNFα signaling, we established two conditions. First, we precultured cells with TNFα plus adalimumab (2 μg/ml) overnight before stimulation with LPS plus adalimumab for 4 hours. Adalimumab restored the production of iTNFα in CD14+ monocytes to normal levels (80.59 ± 6.79%). Second, we precultured cells overnight with TNFα before stimulation with LPS in the presence of adalimumab for 4 hours. In this case, adalimumab was not able to restore iTNFα production in CD14+ cells (13.91 ± 0.18%). Preculturing cells with adalimumab did not alter iTNFα+ CD14+ percentages ( Figure 4).

Discussion
Our present findings suggest that, as a result of exposure to TNFα, cells from patients with active RA had a lower percentage of iTNFα+ CD14+ cells compared with HDs. After 12 injections of adalimumab treatment, the percentage of iTNFα+ CD14+ cells increased progressively, reaching levels similar to those of HDs. This effective adalimumab-induced iTNFα+ CD14+ upregulation was observed only in RA patients who showed a EULARdefined response. After the first two injections, TNFα on the membrane and in the culture supernatant was effectively blocked by adalimumab in all but one patient, who developed antiadalimumab antibodies.
TNFα plays a key role in RA physiopathology. However, we did not expect patients with active RA to have a lower percentage of iTNFα+ CD14+ cells than HDs after TLR cultures. This finding suggests that monocytes from patients with active RA have different regulation of iTNFα production. To understand this unexpected finding, we cultured HD cells with exogenous TNFα, which confirmed that TNFα signals induced negative feedback on iTNFα production or increased traffic of iTNFα to the membrane. Other authors have previously observed that TNFα induced tristetrapolin production, which destabilized TNFα mRNA [23]. Our observations therefore suggest that RA cells were exposed to TNFα. However, TNFα levels were undetectable in serum from these RA patients. We propose that alternative sources of TNFα for circulating RA cells could be local soluble or membrane TNFα [5,24]. On the basis of the results of our present experiments, we cannot exclude the potential that there are additional factors involved in the reduction of iTNFα+ CD14+ cells. We observed that ESR and DAS28-ESR were inversely correlated with the number of iTNFα+ CD14+ cells, suggesting that iTNFα+ CD14+ cells and inflammatory activity were robustly associated. We think that the downregulation of iTNFα+ CD14+ cells percentage was a regulatory mechanism to control excessive TNFα and its detrimental consequences. Adalimumab is currently indicated to reduce signs and symptoms in adult patients with moderately to severely active RA. It induces clinical response, inhibits the progression of joint damage and improves physical function [25]. Here we show that levels of iTNFα+ CD14+ cells increased progressively after the first two injections of adalimumab. This effect was not observed in RA patients in remission induced by methotrexate, suggesting that upregulation of iTNFα+ CD14+ cells was adalimumab-dependent. In addition, the highest levels of iTNFα+ CD14+ cells after 12 injections coincided with the greatest clinical improvement (measured by EULARdefined response), suggesting that proper regulation of iTNFα production in RA patients was associated with a good to moderate EULAR-defined response .
In our in vitro experiments, adalimumab was not able to restore normal iTNFα+ CD14+ levels overnight when cells had previously been exposed to TNFα, suggesting that, in vivo, normalization of iTNFα+ CD14+ cells was due to indirect mechanisms. One possible mechanism is that adalimumab blocks TNFα binding to its receptor [12]. Without TNFα signaling [23], interleukin 1 (IL-1), IL-6, IL-8, granulocyte-macrophage colony-stimulating factor [10] and acute-phase reactants of inflammation (CRP, fibrinogen and ESR) [14] are reduced. This reduction implicates the return of RA cells to a basal state similar to that of cells of HDs and the consequent correction of the percentage of iTNFα+ CD14+ cells. Another possible mechanism is that the renovated circulating pool of CD14+ cells has no contact with TNFα, because TNFα is blocked [26]. A third possible mechanism is also worthy of mention. Monocytes and macrophages from RA patients are targets of TNFα signals and the major producing cells of TNFα, which they present in significant amounts in membrane [27]. Adalimumab has cytotoxic effects on mTNFα-expressing cells [28]. However, it takes several days to get rid of all the cells with mTNFα. Once they are all eliminated, new CD14+ cells with regular iTNFα production repopulate the system.
We observed that infliximab, another neutralizing anti-TNFα monoclonal antibody, restored the percentages of iTNFα+ CD14+ cells to normal values after a long period of therapy. However, in a previous study with infliximab, researchers showed that monocytes produced lower amounts of iTNFα after 24 weeks of treatment than before treatment [29]. This discrepancy in results could be due to clinical and methodological differences. The most crucial difference is that the results described in the previous report were expressed as molecules of equivalent soluble fluorescein per cell, whereas ours are expressed as the percentage of CD14+ monocytes with iTNFα. In addition, and in contrast to those authors, we also evaluated iTNFα levels in HDs to establish the normal percentages of iTNFα+ CD14 + -producing cells.
It should be noted here that, although infliximab and adalimumab are both TNFα-blocking antibodies, they are not fully comparable. Infliximab is administered intravenously at 0, 2 and 6 weeks, followed by a maintenance infusion dose every 8 weeks, whereas adalimumab is administered subcutaneously every 15 days. Moreover, infliximab is a chimeric antibody, whereas adalimumab is a fully humanized antibody.
Adalimumab treatments significantly reduced levels of free mTNFα on CD14+ cells and free soluble TNFα in TLR cultures compared to the first injection. This dramatic reduction in mTNFα on cells was due to two activities of adalimumab: (1) its binding blocked TNF signals and (2) it competed with antibodies for flow cytometry detection [17]. A similar conclusion could be reached for soluble TNFα [30]. Unfortunately, our experimental approach does not allow us to establish the fate of mTNFα and soluble TNFα after adalimumab. However, because iTNFα is the source of mTNFα and soluble TNFα, we can speculate that iTNFα, mTNFα and soluble TNFα have comparable kinetics.
Not all patients respond to adalimumab therapy [16]. In our study, patient 11 did not respond, and, consequently, no iTNFα regulation in CD14+ cells was observed. The lack of response in this patient was due to the presence of antiadalimumab antibodies after the second injection. Although patient 7 responded to adalimumab and iTNFα+ CD14+ cells increased accordingly after two injections, treatment was discontinued after the third injection due to allergy. Consequently, the percentage of iTNFα+ CD14+ cells dropped to levels seen before adalimumab therapy was started. We think that, if adalimumab had not been withdrawn, this patient could have had a good or moderate EULAR-defined response.
We hope that our ongoing experiments will allow us to determine whether the increase in iTNFα+ CD14+ cells is RA-specific or could be extended to other inflammatory pathologies in which TNFα plays a crucial role. Studying the regulation of iTNFα+ CD14+ cells in RA patients could provide individualized information about the efficacy of adalimumab treatment.

Conclusions
Our findings suggest that adalimumab treatment in RA patients can return iTNFα levels to those of HDs. This effect was not observed in the presence of neutralizing antiadalimumab antibodies that produced no EULARdefined response. Our future studies of iTNFα in monocytes during treatment with adalimumab therapy in RA patients could provide information about the effectiveness of treatment.