ERK-dependent T-cell receptor threshold calibration in rheumatoid arthritis¹

Study Population and Cells

T cells from RA patients and demographically matched healthy controls (Table I) meeting the 1988 American College of Rheumatology Criteria for seropositive RA were isolated by negative selection using RosetteSep human T-cell enrichment cocktail (StemCell Tech., Vancouver, Canada). All subjects gave written, informed consent as per the protocol approved by the Emory University IRB. Patients were considered to have active disease when they fulfilled the FDA criteria of morning stiffness >45 minutes, tender joints >6, swollen joints >3, ESR >28 (27). For DC generation, CD14⁺ cells isolated from healthy PBMC using anti-CD14 coated magnetic beads (Miltenyi Biotec Inc. Auburn, CA) were cultured for six days in the presence of IL-4 (1000 IU/ml) and GM-CSF (800 IU/ml). Immature DC were activated with LPS (1.0 μg/ml) for 4 hours just before use.

Antibodies

Anti-CD3-APC Cy7, CD4-PerCP, CD8-PE Cy7, CD45RA-FITC, CD28-PE, CD25-PE, CD69-APC, CD137-APC, CD154-PE and Alexa Fluor 647-conjugated anti- phospho-ERK1/2, phospho-ZAP70, and phospho-NF-κB antibodies were from BD Biosciences, San Jose, CA. In addition, the following antibodies were used: anti-phospho-ERK1/2 and anti-ERK1/2 (Cell Signaling Tech. Beverly, MA), antibodies to N-Ras, K-Ras, Raf-1, and actin (Santa Cruz Biotech, Santa Cruz, CA); anti-SHP-1 (Millipore, Billerica, MA); anti-CD3 (BD Biosciences). Zenon antibody labeling kits with Alexa Fluor 488 and 546 (Invitrogen, Eugene, OR) were used to conjugate primary antibodies.

Flow cytometry

1×10⁶ T cells were stimulated with anti-CD3/CD28 mAb (1 μg/ml each); fixed in BD Cytofix buffer; permeablized by BD Perm Buffer II; and stained for CD3, CD4, CD8, CD28, CD45RA, and indicated signaling molecules. In selected experiments, control and RA T cells were preincubated for one hour with 10 μM of ERK1/2 inhibitor FR180204 (EMD Biosciences, San Diego, CA) followed by activation with anti-CD3/anti-CD28 antibody cross-linking in the continuous presence of the inhibitor. Data were acquired and analyzed on an LSR II flow cytometer (BD Biosciences) with FACS DIVA software. Alternatively, cells were activated using CD3 and CD28 antibody-coated beads for up to 24 hours and stained for activation markers CD25, CD69, CD137, and CD154.

Western Blotting

T cells stimulated as described above were lysed in a cell extraction buffer (Invitrogen) supplemented with 1 mM PMSF and a protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO). 30 μg of total protein were resolved on SDS-PAGE under reducing conditions, transferred to PVDF membranes, and incubated at 4°C with anti-phospho-ERK1/2 or anti-phospho-Raf-1 antibodies followed by washing and incubation with HRP-conjugated secondary antibodies (Santa Cruz Biotechnology). The blots were visualized with an Immobilon Western chemiluminescence detection system (Millipore). Membranes were stripped and re-probed for total ERK1/2 and actin to ensure equal loading.

Confocal Microscopy

T cells from controls and RA patients stimulated and fixed as described above were stained with Alexa Fluor 546-labeled anti-K-Ras or anti-N-Ras antibodies and Alexa Fluor 488-labeled anti-Raf-1 antibody; images were captured by Zeiss confocal laser scanning LSM 510 META Axiovert-200 microscope. Amount and localization of K-Ras, N-Ras, and Raf-1 were quantified by using Image J software. Alternatively, control and RA T cells were stained with Alexa Fluor 546-labeled anti-CD3 antibody and mixed with TSST (1.0 ng/ml)-coated DC at a 5:1 ratio. Following centrifugation at 800 rpm for one minute at 4°C, cells were incubated at 37°C for indicated times. Cells were fixed, permeabilized, and stained with Alexa Fluor 488-labeled anti-SHP-1 antibody.

Mice

SKG mice were obtained from Dr. William E. Seaman (University of California, San Francisco) and housed and bred in the animal facility of Emory University. The experimental protocol was approved by the Emory IACUC. Spleen cells isolated from eight-week-old SKG mice and age-matched BALB/c mice were stimulated with 50 ng/ml of PMA, stained for phospho-ERK1/2, CD3, CD4 and CD8. In the treatment studies, eight-week-old SKG mice (12 per group) were given intra-peritoneal injection of the MEK-1/2 inhibitor U0126 (Promega Corporation, Madison, WI) (1 mg/kg of body weight) twice per week, or vehicle control; joint inflammation was scored weekly by a person blinded to the treatment. The scoring system suggested by Sakaguchi et al (0.1 for a small and 0.5 for a large joint) was used (28). In a parallel treatment group, mice were sacrificed after 4 weeks of treatment; ankle joints of hind limbs were fixed with 10% buffered formalin for 24 h, immersed in 5% EDTA for 2 days for decalcification and embedded in paraffin blocks. The sections were stained with hematoxylin and eosin and evaluated by light microscopy. Joint sections from control and MEK-1/2 inhibitor treated mice were stained with hematoxylin and eosin and leukocyte infiltration into the joint tissue compared between the two groups. Synovitis was scored on a graded scale of 0–3 (0 - no significant change, 1- mild inflammatory infiltration, 2 - pannus formation, and 3 - heavy inflammatory infiltration and bone destruction).

Statistical Analysis

Mean fluorescence intensities (MFIs) of RA and control T cells were compared by two tailed Mann-Whitney U test. The relationship between categorical clinical variables and demographics and pERK levels was examined by Kruskal-Wallis nonparametric one-way analysis of variance; to compare two continuous variables, linear regression analysis was applied. Treatment responses in the SKG mice were assessed by two-way ANOVA.

Increased responsiveness of peripheral RA T cells to TCR stimulation

To screen for differences in T-cell activation thresholds, T cells from RA patients and age-matched controls were stimulated with suboptimal concentrations of anti-CD3/anti-CD28-coated beads, and the induction of the activation markers CD25, CD69, CD137, and CD154 were assessed by flow cytometry at 8 and 24 hours. While the frequency of cells expressing activation markers was not different at either time point, RA CD4 naïve and memory T cells expressed increased cell surface densities of CD69 and CD154 (Fig. 1). Similar differences between RA patients and controls were found for CD69 expression on CD8 T cells. In contrast, cell surface expression of CD25 was not significantly different. Expression of CD137 was about fivefold higher in CD8 than CD4 T cells, but not different in patients and controls.

Activation of the ERK signaling pathway in RA T cells

CD69 is a T-cell activation marker that is very sensitive to the activation of the ERK signaling pathway (29). Also, CD154 expression is dependent on sustained ERK and JNK signaling and sensitive to the activity of the dual specific phosphatase DUSP4 (unpublished observations). We therefore examined RA T cells for their ERK signaling. Initial studies by Western blotting showed increased anti-CD3-induced ERK phosphorylation in RA patients (Fig. 2A). In a subsequent larger study, we used PhosFlow which has the additional advantage of separately analyzing T-cell subsets. T cells from 33 patients with RA and 33 demographically matched controls were purified by negative selection and maintained overnight to exclude effects of cytokines that may be present in RA serum. Basal pERK levels were slightly but significantly increased in RA patients (Fig. 2B, C). The increased baseline pERK was a common feature of all T cells, not limited to in vivo-activated cells and not reduced by overnight culture. The difference in pERK levels was more pronounced in T cells after CD3/CD28 cross-linking (Fig. 2B, C). In contrast to ERK, p38 phosphorylation did not differ at any of the time points analyzed (Fig. 2B, D).

Increased ERK phosphorylation does not correlate with disease activity

Data in Figure 2 were analyzed to address the question of whether pERK in RA T cells correlated with demographic and/or clinical characteristics of the RA population; results are summarized in Table II. Importantly, the disease activity score and the laboratory disease activity markers ESR and CRP failed to correlate with pERK levels at 5 and 10 minutes after activation. Treatment with TNF-α–neutralizing reagents (etanercept, infliximab or adalimumab) correlated with increased pERK levels but inversely and only in CD8 T cells at 5 minutes; patients on anti-TNF-α were even more different from healthy controls consistent with previous studies that TNF-α impairs T-cell function (30, 31). Methotrexate treatment did not influence stimulation of the ERK pathway. Other treatment regimes were too infrequent to analyze. These data suggest that heightened activation-induced pERK is not a disease activity marker. ERK phosphorylation in CD8 T cells correlated with age; however, this correlation was entirely due to a CD8 subpopulation of CD28⁻ T cells that is increased in the elderly (19). The age-related increase in CD28⁻ T cells with age isalso found in healthy controls and is not specific for RA.

Activation of the ERK signaling pathway involves naïve and memory T cells from RA patients

ERK pathway calibration may change with T-cell activation and differentiation, and decline in naïve T cells and expansion of effector T-cell populations have been shown in RA (32, 33). To determine whether the increased ERK activity was due to differential representation of T-cell subsets or a higher frequency of activated T cells, we quantified pERK in naïve and memory CD4 and CD8 T cells. Representative histograms are shown in Supp Fig. 2; data from 33 patients and 33 controls are summarized in Fig. 3. Because the fixation need for PhosFlow interferes with CCR5 and CCR7 staining, we used co-expression of CD45RA and CD28 to identify naïve cells and loss of CD28 to define end-differentiated effector cells which frequently regain CD45RA expression. ERK phosphorylation was more pronounced but less sustained in naïve compared to memory T cells with the exception of CD8⁺CD28⁻ effector T cells which exhibited a strong ERK response (Fig. 3). CD4⁺CD28⁻ T cells were too infrequent in normal individuals to be analyzed. The increased ERK responsiveness in RA was a feature of all CD4 and CD8 T cells including naïve and memory T-cell subsets.

Signal abnormalities in RA map distal from early TCR signaling

To map the difference in signaling events that distinguishes RA from control T cells, we quantified phosphorylation of ZAP70 as a proximal signaling event after TCR engagement which peaks earlier than pERK. ZAP70 phosphorylation was indistinguishable between T cells from RA patients and controls up to 5 minutes after activation (Supp. Fig. 3, Fig. 4A). There was a trend for RA CD4 T cells to have higher pZAP70 at 10 minutes which reached significance at 30 minutes. In CD8 T cells, an increase in ZAP70 phosphorylation in RA reached significance at 10 and 30 minutes. This finding of normal ZAP70 phosphorylation early after activation raised the possibility that the increased ERK phosphorylation is not primarily due to increased TCR signaling, but may induce a positive feedback loop that sustains ZAP70 phosphorylation at later time points in RA T cells. Supporting results for this interpretation were obtained from calcium influx and NF-κB studies. No differences in calcium fluxes were found in the first 10 minutes after stimulation (data not shown). RA patients and controls did not have different baseline levels of pNF-κB (Fig. 4B). There was no evidence for increased NF-κB phosphorylation in RA T cells in the first 10 minutes; a difference was only seen at 30 minutes for CD8 T cells.

Results so far have indicated that the difference between RA patients and normal controls maps to the ERK pathway distal from TCR signaling and that the observed differences in ZAP70 and NF-κB activation are secondary events. To address this issue, we examined ERK phosphorylation in RA patients and controls after PMA stimulation. PMA activates PKC, which controls one amplification loop in ERK phosphorylation and RasGRP, which directly stimulates Ras. As seen with TCR stimulation, pERK was increased in RA patients after PMA stimulation (Fig. 4C). Again, naïve and memory CD4 and CD8 cells were affected (data not shown).

Ras-Raf complex formation in RA patients

To monitor Raf activation, we examined Ras-Raf-1 complex formation at the cell membrane of T cells stimulated with anti-CD3/anti-CD28. T cells were fixed at indicated times after TCR stimulation and stained with N-Ras- or K-Ras-(red) and Raf-1 (green)-specific antibodies (Fig. 5A, B). Membrane-close fluorescence was quantified; results for 5 patients and 5 controls are summarized in Fig. 5C-H. K-Ras baseline levels tended to be slightly but not significantly higher in RA. After stimulation, neither K-Ras nor N-Ras recruitment was different between patients and controls. In contrast, membrane-close Raf-1 concentrations which were similar at 0 and 2 min were increased at 5 min in RA T cells, suggesting increased Raf-1 recruitment. The results for Raf-1 were similar regardless of whether it was examined together with K-Ras (Fig. 5D) or N-Ras (Fig. 5G) staining. In Figure 5E and H, we used regression analysis of the red and green pixel fluorescence to determine a coefficient of co-localization. In healthy individuals, the coefficient peaked after 2 minutes. In contrast, co-localization continued to increase in the RA patient over the first 5 minutes. The confocal studies suggested that the difference between RA patients and healthy controls can be mapped to sustained Raf-1 activation in the patient. In support of this model, serine phosphorylated Raf-1 as determined by Western blot was only different at later time points and not immediately after stimulation (Figure 5I).

Increased ERK activation in RA T cells sustains TCR signaling by delaying SHP-1 recruitment

R. Germain and colleagues proposed a model that assigns pERK an important function in regulating TCR threshold (34, 35). In this model, serine phosphorylation of Lck by pERK prevents Lck inactivation and amplifies TCR signaling. To examine whether the increased ZAP70 phosphorylation seen in RA T cells at later time points after activation is a primary event due to increased TCR signaling or is due to a secondary feedback loop initiated by the increased ERK phosphorylation, we used a specific ERK inhibitor FR180204. This inhibitor inhibits the kinase activity of the phosphorylated ERK1/2 and therefore does not inhibit ERK1/2 phosphorylation. Initial induction of ERK phosphorylation was not suppressed by the inhibitor, neither in control subjects nor in RA patients (Fig. 6A). However, ERK activity inhibition reduced pERK in RA patients after 30 minutes to about the same level as in controls. ERK activity inhibition also had an inhibitory effect on ZAP70 phosphorylation in RA patients (Fig. 6B). A trend was already seen 10 minutes after stimulation and reached significance at 30 minutes. These data confirm that proximal TCR signaling is normal and that a positive amplification loop involving pERK sustains ZAP70 phosphorylation in RA patients.

To examine whether SHP-1 recruitment to the TCR signaling complex is delayed in RA, we used confocal imaging of CD4 T cells interacting with TSST-loaded DC. SHP-1 and CD3 fluorescence in the TCR-DC contact region was quantified (Fig. 6C–E). CD3 fluorescence increased, reflecting the TCR clustering. SHP-1 recruitment was delayed in RA patients and significantly reduced in the first minutes after T-cell activation. To determine whether the delayed SHP-1 recruitment is associated with a lowered TCR threshold in RA patients, ERK phosphorylation was quantified after suboptimal stimulation with anti-CD3/anti-CD28. CD4 and CD8 T cells from RA patients responded to lower antibody doses. The dose response curve from RA patients had a more sigmoid pattern and drifted apart from the controls at anti-CD3 concentrations of less than 1 ng/ml (Fig. 6F).

ERK phosphorylation is increased in SKG mice compared to BALB/c mice

Our preliminary analysis did not find an association of increased pERK levels with disease activity (Table II). We, therefore, wanted to explore the possibility that the increased responsiveness preceded disease onset and predisposed for disease. To examine the preclinical situation, we turned to an animal model of RA which has been shown to be T cell dependent. The SKG mouse has a mutation in ZAP70 and develops arthritis similar to RA (28). We examined ERK phosphorylation in T cells at an age before the mice develop disease. Results are shown in Figure 7A. As observed in RA patients, pERK levels were increased in T cells from 2-month-old non-arthritic SKG mice compared to normal BALB/c mice. To examine whether the increased ERK responsiveness predisposes for disease, SKG mice (12 mice/group) were treated with solvent control or a MEK-1/2 inhibitor, U0126 (1 mg/kg per injection, twice per week) starting at the age of 8 weeks. This dose is subtherapeutic; the normal dosing in cancer treatment trials is 25 mg/kg two times daily. Control-injected mice started to develop disease 2 weeks after initiation of treatment. In the MEK-1/2 inhibitor-treated mice, disease onset was delayed and activity was reduced (Fig. 7B). Representative histologies from treated and control mice are shown in Fig. 7C.