Thalidomide and its analogues have distinct and opposing effects on TNF-α and TNFR2 during co-stimulation of both CD4+ and CD8+ T cells

Authors


Dr J. B. Marriott, Division of Oncology, St George's Hospital Medical School, Cranmer Terrace, Tooting, London, SW17 ORE, UK.
 E-mail: jmarriot@sghms.ac.uk

Summary

Thalidomide (Thd) is clinically useful in a number of conditions where its efficacy is probably related to its anti-TNF activity. More recently, Thd has also been shown to co-stimulate T cells and second generation co-stimulatory (IMiD™) analogues are currently being assessed in the treatment of cancer patients. However, in contrast to their known suppressive effects during inflammatory stimuli, the effects of Thd/IMiDs on TNF-α and TNF receptors (TNFRs) during T cell co-stimulation are not known. We sought to determine the effect of Thd, two clinically relevant IMiDs (CC-4047, ACTIMID™ and CC-5013, REVIMID™) and a non-stimulatory SelCID analogue (CC-3052) on TNF-α production and on the expression and shedding of TNFRs during co-stimulation. We found that co-stimulation of PBMC with Thd/IMiDs, but not CC-3052, prevented αCD3-induced T cell surface expression of TNFR2 and thereby reduced soluble TNFR2 (sTNFR2) levels. However, there was no effect on total (surface/intracellular) TNFR2 protein expression, suggesting inhibition of trafficking to the cell membrane. The extent of co-stimulation by Thd/IMiDs (assessed by CD69/CD25 expression and IL-2/sIL-2Rα production) was similar for CD4+ and CD8+ T lymphocytes and correlated with TNFR2 inhibition. Co-stimulation, but not the early inhibitory effect on TNFR2, was IL-2-dependent and led to increased TNF-α production by both CD4+ and CD8+ T lymphocytes. The clinical relevance of this observation was confirmed by the elevation of serum TNF-α during REVIMID™ treatment of patients with advanced cancer. Together, these results suggest a possible role for TNF-mediated events during co-stimulation and contrast with the TNF inhibitory effects of Thd and its analogues during inflammatory stimuli.

Introduction

Thalidomide (Thd) is emerging as an important immunotherapeutic drug in a number of clinical situations [1–4]. It is well established as an anti-TNF-α agent in the context of monocyte/macrophage activation in vitro[5,6] and clinically during TNF-α-mediated inflammatory disease [6]. However, other immunomodulatory properties may explain the extensive range of its clinical activity. More recently, it has been shown to inhibit monocyte IL-12 production [7] and also to be able to induce the activation and proliferation of CD8+ T cells stimulated via the T cell receptor in the absence of co-stimulation [8]. Thus, in certain clinical settings, such as in the treatment of cancer patients, Thd may act as an adjuvant to promote T-cell responses. However, Thd treatment has also been known to increase circulating levels of TNF-α[9–12] and IL-12 [13] and it appears likely that the effects of Thd vary, depending on the relative predominance of monocyte/macrophage and T cell activation. Furthermore, the effects of Thd may also vary depending on the activation of particular signalling pathways [14].

Significant side effects such as somnolence and peripheral neuropathy are associated with clinically effective Thd treatment. However, the design and synthesis of Thd analogues has yielded compounds with enhanced activity/toxicity profiles [15,16]. These analogues are currently being characterized and have been shown to segregate into at least two distinct classes. The SelCIDs™ (selective cytokine inhibitory drugs) consist of phosphodiesterase type 4 (PDE4) inhibitors [17]. The IMiDs™ (immunomodulatory drugs) are thought to be mechanistically similar to Thd, although with enhanced potency and therapeutic indices, and act via as yet unknown mechanism(s) [18–24]. Both groups of compounds are potent inhibitors of monocyte/macrophage-derived TNF-α, although T cell co-stimulatory activity is limited to the IMiD group [17].

Thd and its co-stimulatory IMiD analogues, in particular CC-4047 and CC-5013, are currently being assessed in the treatment of patients with advanced multiple myeloma and patients with advanced solid tumours [25–32]. Furthermore, we have shown recently that CC-4047, when co-administered with autologous or allogeneic tumour cell vaccination, is able to enhance protective and long-lasting immunity significantly in a murine model of colorectal cancer [33].

Thd and its derivatives are known to suppress TNF-α during the activation of inflammatory pathways by LPS, but it is not known whether this is the case during the direct stimulation of T cell activation pathways. Therefore, in this study we planned to determine the effect of these compounds, plus the non-co-stimulatory analogue CC-3052, on the expression of TNF-α and its receptors during co-stimulation in vitro. Furthermore, we wanted to assess the co-stimulatory effect on CD4+ and CD8+ populations because previous results suggested that CD8+ T cells were activated preferentially. Because signals generated via the TNF system are important in T cell homeostatic control we hoped to provide insights into the T cell co-stimulatory mechanism of these novel compounds. Finally, we hoped to show that any effects on the TNF system were relevant for clinical non-inflammatory disease and to this end we obtained sera from a small phase I study of CC-5013 in the treatment of advanced cancer.

Materials and methods

PBMC and whole blood culture

Thd, two IMiDs (CC-4047 and CC-5013) and a SelCID (CC-3052) (provided by Celgene Corporation, NJ, USA) were dissolved freshly in DMSO to make a 20 mg/ml stock solution. The compounds were then diluted directly into the tissue culture wells at the required concentrations.

Venous blood was collected into sodium heparin vacutainers (Becton Dickinson, Oxford, UK) and PBMC were prepared subsequently by density centrifugation over Ficoll Hypaque (Sigma, Poole, UK). After centrifugation at 700 g for 20 min cells were collected from the interface, washed three times in HBSS (Sigma), counted and resuspended at 1–2 × 106/ml in complete RPMI-1640 medium supplemented with 10% fetal calf serum (FCS), 1% penicillin/streptomycin and 2 mm L-glutamine (all Sigma). PBMC were cultured ± Thd/analogues (1 ng−10 µg/ml; 0·05%v/v final DMSO concentration) at 1 ml/well in 24-well plates, precoated with anti-CD3 mAb (2·5 µg/ml; R&D systems, Minneapolis, USA) in bicarbonate coating buffer pH 9·5. PBMC were incubated for 12 h−7 days at 37°C in 5% CO2. For some experiments PBMC were co-incubated with neutralizing anti-IL-2 mAb (0·005–0·1 µg/ml; R&D systems). Cell-free supernatants were collected and stored in aliquots at −70°C until assayed by ELISA. Cells were harvested and resuspended in PBS/BSA prior to processing as described below.

For lipopolysaccharide (LPS; Sigma, Escherichia coli serotype 0127:B8)-stimulated cultures heparinized venous blood was diluted 1 : 4 in complete RPMI-1640 medium (without FCS) and stimulated in 24-well plates at 1 ml/well with LPS (1 µg/ml). All cultures were incubated at 37°C in 5% CO2 ± Thd/analogues (10 µg/ml). After 24–48 h cell-free supernatants were collected by microcentrifugation and stored in aliquots at −70°C until assayed by ELISA.

Clinical CC-5013 study: patient selection, treatment and sample collection

We collected serum from a small phase I study designed to assess the safety, tolerability and efficacy of REVIMID™ (CC-5013) in the treatment of patients with advanced and heavily pretreated cancer. Of the 20 patients enrolled we obtained samples from 11 subjects, aged between 18 and 75 years, with histologically proven stage IV metastatic melanoma (n = 6), histological or cytological proof of adenocarcinoma of the exocrine pancreas (n = 2) or other malignancy (one each of breast, renal and lung cancer; n=3). Before initiating therapy, patients were subject to a complete medical history, physical examination and baseline evaluation of signs and symptoms, including full blood counts and neurological examination. These were repeated at weekly intervals during the study period and also at the end of the study.

All patients gave written informed consent prior to participation in the study. The protocol was approved by the Wandsworth Local Research Ethics Committee (LREC). All patients were treated with 5 mg oral REVIMID™ daily (tablets were taken every evening) for 1 week with the dose escalating to 10 mg daily at week 2, 25 mg daily at week 3 and 50 mg daily at week at which the treatment was maintained indefinitely depending on tolerance.

Blood was taken at the same time at each visit (and therefore at the same time relative to drug administration), collected into serum separator (SST; Vacutainer, BD) tubes and left to clot for ∼30 min. Tubes were spun at 950 g for 10 min and serum collected. Sera were frozen in aliquots at −70°C until being assayed for TNF-α and sIL-2 receptor α (sIL-2R) by ELISA. Serum was collected at baseline, during the first 1–3 weeks (10–25 mg daily) of treatment and again at weeks 4–5 (50 mg daily).

REVIMID™ tablets were manufactured by Penn pharmaceuticals (Tredegar, Wales, UK) and supplied by Celgene Corporation (Warren, NJ, USA).

Enzyme-linked immunosorbent assay (ELISA)

Culture supernatants were assayed for sTNFR1, sTNFR2, sIL-2R using kits provided by R&D Systems (Minneapolis, MN, USA) and for IL-2 and TNF-α using an assay procedure and reagents provided by Pharmingen (BD). Briefly, 96-well flat-bottomed ELISA plates (Nunc) were precoated overnight at 4°C with anticytokine capture mAb in bicarbonate coating buffer, pH 9·5 and washed (with PBS/Tween20) prior to incubation of standards and test (supernatant) samples overnight at 4°C. After washing, plates were incubated with biotinylated anticytokine mAb for 1 h. After further washing, plates were developed using chromogen (tetramethylbenzidine)/hydrogen peroxide for 20–30 min at RT. Standard absorbances (405 nm) of duplicate wells (minus control zero standard) were used to calculate concentration of cytokine/receptor levels which were corrected for any dilution factor. Data are expressed as percentage compared to control (DMSO alone), which is represented as 100%.

Sera from patients were assayed for TNF-α, sTNFR2, IL-2 and sIL-2R using kits provided by R&D Systems. In each case, the manufacturer's instructions were followed exactly. Data are expressed as pg/ml at baseline (pretreatment) compared to follow-up at 1–3 and 4–5 weeks.

Flow cytometric analysis of surface CD25, CD69, TNFR1 and TNFR2

PBMC were harvested from 24-well plates by gentle aspiration, washed and suspended in wash buffer (PBS/0·5% BSA/0·1% sodium azide). Three-colour phenotypical analysis of αCD3-stimulated PBMC was performed by surface staining cells (by incubating cells for 30 min at 4°C) with the following fluorochrome conjugated monoclonal antibodies: anti-CD25 FITC (Becton Dickinson Immunocytometry Systems, BDIS, Oxford, UK), anti-CD69 FITC (L78; BDIS) or anti-TNFR1 FITC (16803·161; R&D Systems) and anti-TNFR2 PE (22235·311; R&D Systems) with either anti-CD4 PerCP (SK3; BDIS) or anti-CD8 PerCP (SK1; BDIS) plus appropriate isotype-matched and compensation controls. Cells were then washed once in 2 ml wash buffer and fixed for analysis in 200 µl Cellfix™ (BDIS) prior to flow cytometric analysis. Upon analysis lymphocytes were gated on forward scatter (FSC) versus side scatter (SSC) properties. PerCP-positive T cell subsets were then selected by gating on SSC versus FL-3 dot-plots and displayed as two-colour dot-plots, with quadrants set according to isotype-matched controls. For each sample 10 000 lymphocytes were acquired on a Becton Dickinson FACScan using CellQuest™ software.

Analysis of intracellular TNFR2, IL-2 and TNF-α

PBMC were stimulated with αCD3 ± CC-4047 for 24–96 h with the protein transport inhibitor Brefeldin A (10 µg/ml; Sigma) present for the last 6 h of culture to allow intracellular accumulation of protein. Cells were harvested, washed and suspended in wash buffer, then fixed and permeabilized using the specially formulated Becton Dickinson FACS lysis and permeabilization solutions as per the manufacturer's instructions. Cells were then stained with anti-TNFR2 FITC, anti-IL-2 FITC (5344·111, BDIS) or anti-TNF-α FITC (6401·1111, BDIS) and anti-CD4 PE or anti-CD8 PE plus appropriate isotype-matched and compensation controls for 30 min before washing once in 2 ml wash buffer followed by a final resuspension in 200 µl Cellfix™. Stained samples were kept in the dark at 4°C prior to two-colour analysis on the permeabilized cells. Upon flow cytometric analysis lymphocytes were gated on forward scatter (FSC) versus side scatter (SSC) properties and displayed as two-colour dot-plots, with quadrants set according to isotype-matched controls. For each sample 10 000 lymphocytes were acquired on a Becton Dickinson FACScan using CellQuest™ software.

Proliferation assay

PBMC (1–2 × 105) were resuspended in complete RPMI-1640 (± Thd/analogues) onto αCD3 precoated 96-well round-bottomed plates (Falcon, BD) to a final volume of 200 µl per well. Cells were cultured at 37°C in 5% CO2 for 5 days with [3H]-thymidine (NEN Life Science Products, Hounslow, UK) added for the last 24 h (2 mCi/well). Cells were harvested onto glass fibre filter paper and counted in a matrix 96 direct beta counter (Canberra Packard). Uptake of [3H]-thymidine was determined as mean cpm of sextuplet culture wells and expressed as percentage change compared to control (DMSO alone).

Statistical analysis

For analysis of serum data comparisons between groups were examined by the Kruskal–Wallis test with inclusion of the Dunn's multiple comparison test. The statistics were performed using Primer of Biostatistics (3·02) statistical software.

Results

Co-stimulatory IMiD CC-4047 strongly inhibits the secretion of sTNFR2 by αCD3-stimulated PBMC cultures whereas non-co-stimulatory SelCID CC-3052 augments sTNFR2

A single treatment of CC-4047 leads to potent inhibition of sTNFR2 levels in αCD3-stimulated PBMC supernatants at 24 h (Fig. 1a). This effect was consistent in four of four experiments (24 h: mean αCD3 alone, 2676 pg/ml; + CC-4047 = 32·3% ± 23·5% of control). Inhibition was decreased at 48 h (mean αCD3 alone, 4436 pg/ml; + CC-4047 = 52·7% ± 13% of control), although a small inhibitory effect was still apparent after 72 h. Soluble TNFR2 was not consistently detectable in 12 h cultures (data not shown).

Figure 1.

Figure 1.

Co-stimulatory CC-4047 strongly inhibits the secretion of sTNFR2 by αCD3-stimulated PBMC cultures (but not LPS-stimulated cultures) whereas non-co-stimulatory SelCID CC-3052 augments sTNFR2. Divergent effects of CC-4047 (Ο), CC-3052 (□), and Thd (Δ) on the production of sTNFR2, sTNFR1 & TNF-α by PBMC-stimulated with αCD3 (a–c) and whole blood cultures stimulated with LPS (d–f). Cultures were incubated ± Thd/analogues (10 µg/ml) and supernatants collected at the indicated times as per Materials and methods. Results are expressed as percentage change compared to control cultures with DMSO alone. Data presented are from four normal donors used in two separate experiments.

Figure 1.

Figure 1.

Co-stimulatory CC-4047 strongly inhibits the secretion of sTNFR2 by αCD3-stimulated PBMC cultures (but not LPS-stimulated cultures) whereas non-co-stimulatory SelCID CC-3052 augments sTNFR2. Divergent effects of CC-4047 (Ο), CC-3052 (□), and Thd (Δ) on the production of sTNFR2, sTNFR1 & TNF-α by PBMC-stimulated with αCD3 (a–c) and whole blood cultures stimulated with LPS (d–f). Cultures were incubated ± Thd/analogues (10 µg/ml) and supernatants collected at the indicated times as per Materials and methods. Results are expressed as percentage change compared to control cultures with DMSO alone. Data presented are from four normal donors used in two separate experiments.

We found that the non-co-stimulatory analogue, CC-3052, had the opposite effect on sTNFR2 when compared to CC-4047 in all experiments (24 h: + CC-3052 = 222% ± 141% of control). The effect of CC-3052 also decreased at 48 h (+ CC-3052 = 159% ± 50·4%) and a smaller augmentary effect was apparent at 72 h. Thd had very little overall effect on sTNFR2 (for example, at 48 h: + Thd = 102·7% ± 9·3% of control).

In contrast to the effects on sTNFR2 none of the compounds had a consistent or significant effect on the levels of sTNFR1 at 24 h (+ CC-4047, mean 110% ± 10·9%; CC-3052, mean 86·8% ± 13·6%; Thd = 104% ± 11·2%; Fig. 1b). However, CC-3052 did inhibit sTNFR1 at 48 and 72 h.

We also found that CC-4047 and Thd strongly induced the production of TNF-α at 72 h (mean αCD3 alone control, 1443 pg/ml; + CC-4047 = 365% ± 233%; + Thd = 247% ± 122%; Fig. 1c), although there was very little effect at 24 h (αCD3, 2753 pg/ml; CC-4047 = 114% ± 37·8%; Thd = 129% ± 39·8%).

sTNFR2 production by LPS-stimulated whole blood cultures was not affected by CC-4047, Thd or CC-3052

We next determined whether the divergent effects of CC-4047 and CC-3052 on sTNFR2 production seen during T cell co-stimulation with αCD3 were also apparent during LPS activation of monocyte/macrophage cell populations. In contrast to the effect during T cell activation we found that there was no effect of any compound on sTNFR1 or sTNFR2 production by LPS-stimulated whole blood cultures at any time-point (Fig. 1d,e). As expected, CC-3052 and CC-4047 both strongly inhibited LPS-induced TNF-α production (Fig. 1f) and were more effective than Thd in this respect.

CC-4047 and CC-5013 strongly reduced the surface expression of TNFR2 on αCD3-stimulated CD4+ and CD8+ T cells during PBMC co-stimulation

We then determined whether the inhibitory effect of CC-4047 on supernatant levels of sTNFR2 during co-stimulation was due to reduced surface TNFR2 expression or to reduced receptor shedding. We found that CC-4047 and CC-5013 (and to a lesser extent Thd) reduced the surface expression of TNFR2 and that this correlated with the induction of surface CD25 (IL-2 receptor) (Fig. 2). Furthermore, this was seen equally in both CD4+ (Fig. 2a) and CD8+ populations (Fig. 2b). Results from two dose–response experiments indicated that co-stimulation (gauged by CD69 expression) and inhibition of surface TNFR2 could both be detected at CC-4047 concentrations as low as 10 ng (data not shown).

Figure 2.

Figure 2.

CC-4047 and CC-5013 strongly reduced the surface expression of TNFR2 on αCD3 stimulated CD4+ and CD8+ T cells during PBMC co-stimulation. PBMC were incubated ± Thd/analogues (10 µg/ml) for 48 h. Three-colour flow cytometric analysis was performed by surface staining stimulated PBMC with anti-CD25 FITC, anti-TNFR2 PE and anti-CD4/CD8 PerCP plus appropriate isotype matched and compensation controls as described in Materials and methods. (a) Dot plots represent TNFR2 versus CD25 (IL-2 receptor) gated on (a) CD4 and (b) CD8 T cell subsets. Histograms show mean fluorescent intensity of TNFR2 expression on CD25+ T cells. This effect was highly consistent when performed in 10 separate experiments each using PBMC from two normal donors although data from a single representative experiment are shown.

Figure 2.

Figure 2.

CC-4047 and CC-5013 strongly reduced the surface expression of TNFR2 on αCD3 stimulated CD4+ and CD8+ T cells during PBMC co-stimulation. PBMC were incubated ± Thd/analogues (10 µg/ml) for 48 h. Three-colour flow cytometric analysis was performed by surface staining stimulated PBMC with anti-CD25 FITC, anti-TNFR2 PE and anti-CD4/CD8 PerCP plus appropriate isotype matched and compensation controls as described in Materials and methods. (a) Dot plots represent TNFR2 versus CD25 (IL-2 receptor) gated on (a) CD4 and (b) CD8 T cell subsets. Histograms show mean fluorescent intensity of TNFR2 expression on CD25+ T cells. This effect was highly consistent when performed in 10 separate experiments each using PBMC from two normal donors although data from a single representative experiment are shown.

We were unable to detect the presence of surface TNFR1 either with αCD3 alone or in the presence of the co-stimulatory compounds using cells from a total of four donors in two separate experiments (data not shown).

CC-4047 did not alter the level of total cellular TNFR2 (intracellular plus surface) in T cells during PBMC co-stimulation

We investigated whether the inhibitory effect of CC-4047 on surface TNFR2 was due to the inhibition of TNFR2 protein production. We utilized the protein transport inhibitor Brefeldin A and two-colour flow cytometry in order to measure the total amount of cellular TNFR2 within CD4+ and CD8+ cells during PBMC co-stimulation. The data presented in Fig. 3 show that when total protein was analysed CC-4047 did not appear to exert an inhibitory effect on TNFR2. The data presented suggest strongly that TNFR2 protein production itself is not being affected. Therefore, it is likely that trafficking to the cell surface that is being inhibited. Furthermore, our data also rule out an inhibitory effect on TNFR2 transcription.

Figure 3.

CC-4047 did not alter the level of total cellular TNFR2 (intracellular plus surface) in T cells during PBMC co-stimulation. Unstimulated PBMC and cells stimulated with αCD3 ± CC-4047 (10 µg/ml) for 24 h (TNFR2) (the last 6 h in the presence of the protein transport inhibitor, Brefeldin A) were harvested and processed as described in Materials and methods. Lymphocytes were gated on forward scatter (FSC) versus side scatter (SSC) properties and displayed as two-colour dot-plots, with quadrants set according to isotype-matched controls. No difference was seen in the expression of intracellular TNFR2 during the co-stimulation of CD4+ or CD8+ T cells. The data presented are representative of three separate experiments in which very similar results were obtained.

The relative extent of TNFR2 inhibition by CC-4047, CC-5013 and Thd corresponds to co-stimulatory activity as gauged by the production of IL-2, sIL-2R and proliferation in αCD3 stimulated PBMC cultures

We then investigated whether there was a possible connection between the inhibitory effect of these compounds on TNFR2 surface expression and co-stimulatory activity. The co-stimulatory activity of CC-4047, CC-5013 and Thd was assessed by IL-2 and sIL-2R production and also by the incorporation of tritiated thymidine (to measure proliferation).

We found that IL-2 production was not detected in control (αCD3 alone) cultures. However, after 24 h CC-4047, CC-5013 and, to a lesser extent Thd, strongly induced IL-2 production (Fig. 4a). For CC-4047 this effect was still apparent after 7 days of culture. The increased expression of sIL-2R became apparent after 72 h and followed a very similar pattern (Fig. 4b). Proliferation was assessed at days 5–6 with a consistent pattern of results (Fig. 4c). Therefore, using these three markers of co-stimulation the same relative pattern of effectiveness emerged: CC-4047 > CC-5013 >> Thd.

Figure 4.

The relative extent of TNFR2 inhibition by CC-4047, CC-5013 and Thd corresponds to the degree of co-stimulation by these compounds. PBMC cultures were co-incubated for 24 h (IL-2), 72 h (sIL-2R) and 120 h (proliferation) ± Thd/analogues (10 µg/ml) as described in Materials and methods. Cell culture supernatants were assayed for IL-2 and sIL-2R by ELISA. Proliferation was assessed by [3H]-thymidine incorporation. Results are expressed as percentage change in (a) IL-2, (b) sIL-2R and (c) proliferation compared to control cultures. Data are from four normal donors used in two separate experiments.

In contrast to these effects the non-co-stimulatory SelCID, CC-3052, strongly inhibited IL-2 and sIL-2R although it had no effect on proliferation, as we have shown previously [19].

Co-stimulation by CC-4047 and subsequent TNF-α production, but not early inhibition of TNFR2, is dependent on IL-2-mediated signalling

We next determined the importance of IL-2-mediated signalling for the process of CC-4047-mediated co-stimulation. We wanted to determine whether elevation of TNF-α was a consequence of T cell activation and not a direct effect of CC-4047. We also wanted to confirm that TNFR2 inhibition was not due to co-stimulation itself and raise the possibility that this early effect may be significant in the initiation of the co-stimulatory process.

Inhibition of IL-2 signalling was achieved with the use of a neutralizing anti-IL-2 mAb precoated onto the culture well. Under co-stimulatory conditions anti-IL2 mAb decreased supernatant IL-2 levels in a dose-dependent manner (Fig. 5a) and also decreased the surface expression of IL-2 receptor (data not shown). In these same cultures inhibition of IL-2 prevented the superinduction of TNF-α in a similar dose–response fashion (Fig. 5b) Therefore, co-stimulation-induced TNF-α production is clearly mediated via IL-2 generated signalling. However, in identical studies we found that the inhibition of IL-2-mediated signalling did not prevent TNFR2 inhibition by CC-4047 (data not shown).

Figure 5.

Co-stimulation by CC-4047 and subsequent TNF production, but not early inhibition of TNFR2, is dependent on IL-2-mediated signalling. PBMC cultures were stimulated with αCD3 in the presence of CC-4047 (10 µg/ml) and anti-IL-2 mAb (0·1–0·005 µg/ml) as described in Materials and methods. Results are expressed as percentage values relative to cultures co-stimulated with αCD3 + CC-4047 (0% represents αCD3 alone). Cultures with increasing concentrations of anti-IL-2 mAb show a reduction in (a) supernatant IL-2 levels and (b) supernatant TNF-α levels. Data shown are from four normal donors in two separate experiments.

Co-stimulation-induced IL-2 and TNF-α is derived from both CD4+ and CD8+ T cells

Having established a requirement for IL-2 production and (subsequent IL-2-mediated signalling) during CC-4047- mediated co-stimulation we sought to determine the source of this cytokine. In particular we wished to determine conclusively whether co-stimulation affects both CD4+ and CD8+ T cells. This was achieved by intracellular cytokine specific staining of CD4+ and CD8+ T cells within the PBMC population. Data are shown from one of two separate experiments in which almost identical data were obtained.Figure 6a confirms the lack of IL-2 production observed by ELISA in control (αCD3 alone) cultures because very few cells from either subset produce IL-2. However, upon CC-4047-mediated co-stimulation there are increased numbers of IL-2-producing cells from both T cell subsets. Furthermore, the elevated production of TNF-α, which is secondary to co-stimulation, is also due to both T cell types (Fig. 6b). Thus, co-stimulation by CC-4047 affects CD4+ and CD8+ T cell subsets rather than inducing preferential activation of CD8+ cells, as has been reported previously [8].

Figure 6.

Figure 6.

Co-stimulation-induced IL-2 and TNF-α are derived from both CD4+ and CD8+ T cells. Unstimulated PBMC and cells stimulated with αCD3 ± CC-4047 (10 µg/ml) for 96 h (IL-2/TNF-α) (the last 6 h in the presence of the protein transport inhibitor, Brefeldin A) were harvested and processed as described in Materials and methods. Lymphocytes were gated on forward scatter (FSC) versus side scatter (SSC) properties and displayed as two-colour dot-plots, with quadrants set according to isotype-matched controls. The induction of intracellular IL-2 and TNF-α was detected in both CD4+ and CD8+ T cells. Data shown are from one normal donor and are representative of very similar data obtained from three donors.

Figure 6.

Figure 6.

Co-stimulation-induced IL-2 and TNF-α are derived from both CD4+ and CD8+ T cells. Unstimulated PBMC and cells stimulated with αCD3 ± CC-4047 (10 µg/ml) for 96 h (IL-2/TNF-α) (the last 6 h in the presence of the protein transport inhibitor, Brefeldin A) were harvested and processed as described in Materials and methods. Lymphocytes were gated on forward scatter (FSC) versus side scatter (SSC) properties and displayed as two-colour dot-plots, with quadrants set according to isotype-matched controls. The induction of intracellular IL-2 and TNF-α was detected in both CD4+ and CD8+ T cells. Data shown are from one normal donor and are representative of very similar data obtained from three donors.

Treatment of cancer patients with CC-5013/REVIMID™ leads to increased serum levels of sIL-2R and TNF-α

We wanted to extend our in vitro observations and determine whether the clinical use of a co-stimulatory IMiD (CC-5013/REVIMID™) could lead to elevation of serum TNF-α in the context of T cell activation. We obtained serum from a small phase I study in patients with advanced cancer which had shown that REVIMID™ was well tolerated. The details of this study have been submitted for publication (Marriott et al.). Serum was collected at baseline and twice during treatment (weeks 1–3 and weeks 4–5). Most patients showed evidence of T cell activation with significantly increased serum levels of sIL-2R in 10 of 11 subjects (90·9%), although we were unable to detect circulating IL-2 (detection limit of assay < 7 pg/ml). However, serum TNF-α was also elevated in 10 of 11 patients. Figure 7 shows that both sIL-2R and TNF-α are increased soon after commencing treatment compared to baseline. For sIL-2R: baseline mean, 710 pg/ml (n = 11); weeks 1–3 mean, 1049 (n = 9; n.s. versus baseline); weeks 4–5 mean, 1294 (n = 7; P < 0·05 versus baseline). For TNF-α: baseline mean, 2023 pg/ml; weeks 1–3 mean 3408 (P < 0·05); weeks 4–5 mean 3544 (P < 0·05).

Figure 7.

Treatment of advanced cancer patients with CC-5013/REVIMID™ results in increased TNF-α and sIL-2 receptor levels. Patient serum samples were obtained from a small phase I study as detailed in Materials and methods. Serum was collected at baseline and twice after commencing CC-5013 treatment (weeks 1–3 and weeks 4–5) and assayed for (a) sIL-2R and (b) TNF-α by ELISA. Each symbol represents a single patient.

Discussion

In this study we have established a close correlation between the co-stimulatory ability of Thd/IMiDs with a reduced ability to induce the surface expression of TNFR2. Co-stimulation (as determined by surface CD25/CD69 expression and intracellular IL-2/TNF-α production) and inhibition of TNFR2 was observed with both CD4+ and CD8+ T lymphocytes within a mixed PBMC population. This contrasts with previous data relating to Thd's co-stimulatory activity in two respects. First, we observed co-stimulation within a PBMC population containing accessory cells, and secondly there did not appear to be preferential activation of CD8+ cells [8]. Both of these observations are also supported by data showing the activation of both CD4+ and CD8+ T lymphocytes in the peripheral blood of patients with advanced cancer treated with CC-4047 (our unpublished observations) and CC-5013 (manuscript submitted).

We found that IMiD-induced co-stimulation prevented TNFR2 surface expression and subsequent secretion of sTNFR2. However, total cellular TNFR2 was not affected. Thus, when total (intracellular and surface) TNFR2 expression was analysed there was no effect of CC-4047. Together, these data exclude an inhibitory effect at the transcriptional level or on receptor shedding. The effect on TNFR2 is therefore due to either an inhibitory effect on its transport from the endoplasmic reticulum or storage granules to the cell membrane, although increased receptor endocytosis from the surface cannot be ruled out.

The results obtained during T cell co-stimulation contrast with those obtained during the activation of monocytes/macrophages by LPS. In these cultures we observed the expected reduction of TNF-α production by Thd and its analogues (both IMiDs and SelCIDS); the analogues are more effective inhibitors than Thd as they have been designed and screened specifically for this property. None of the compounds had any effect on monocyte/macrophage secretion of sTNFR1 or sTNFR2. Hence, the TNFR2 inhibitory effect of these compounds is only apparent using co-stimulatory compounds during the stimulation of T cells.

In contrast to the inhibitory effect in LPS-stimulated cultures, we also observed strongly elevated production of TNF-α during co-stimulation by Thd/IMiDs. Augmentation of TNF-α was due to production by both CD4+ and CD8+ T cells and, like co-stimulation itself, was dependent on IL-2-mediated signalling. Increased TNF-α production was observed from 48 h onwards and was not observed readily at 24 h, the time-point used to assess TNF-α production during the Haslett et al. study [8], although Corral et al. showed modest stimulation of TNF-α by CC-4047 at 24 h [17]. As expected, the complete inhibition of TNF-α by the non-co-stimulatory SelCID, CC-3052, was maintained over the same period. The kinetics of TNF-α induction by co-stimulatory compounds and the requirement for IL-2 signalling support the idea that T cell activation, rather than a direct affect of IMiDs, leads to increased TNF-α production.

The significance of the early effect of Thd/IMiDs on TNFR2 remains unknown. However, signals mediated via TNFR2 are known to play an important role in T cell homeostasis. For example, proliferation of mature peripheral T cells can be achieved by specific activation via TNFR2 [34] and TNFR2 also appears to have a role in the down-regulation of ongoing T cell responses [35]. In this study, we have found that although CC-4047-mediated co-stimulation requires IL-2 the early inhibitory effect on TNFR2 is IL-2 independent. However, it is still possible that TNFR2-mediated events may play a role in the initiation of the co-stimulatory process.

The augmentary effect on TNF-α that we have detected during co-stimulation in vitro has possible implications for the ongoing clinical use of Thd/IMiDs in the treatment of TNF-α-mediated inflammatory disease, although in this context TNF-α suppressive effects are more likely to predominate. However, in one study Thd treatment of patients with toxic epidermal necrolysis led to a further increase in TNF-α production and poor clinical outcome [10]. In this study we have shown that CC-5013/REVIMID™ treatment of patients with cancer leads to elevation of serum TNF-α and sIL-2 receptor levels (compared to pretreatment baseline levels). This supports the concept that T cell co-stimulatory effects may explain the increased TNF-α levels in the serum of patients given Thd for the treatment of non-inflammatory conditions [36].

Pharmacokinetic data for IMiD activity in vivo are not yet available. However, we have noted that CC-4047 is able to exert its co-stimulatory and TNFR2 inhibitory effects in vitro at concentrations as low as 10 ng/ml (our unpublished observation). The previously reported observation [8,37] that plasma Thd levels of 1–2 µg/ml are obtainable using an oral dose of 150 mg/day suggests that these compounds will be clinically useful at far lower doses than the parent compound. This is borne out by the immunostimulatory effects seen in advanced cancer patients at weeks 1–3 after commencing CC-5013/REVIMID™ treatment at a dose of up to 25 mg/day. Furthermore, we have also noted profound immunostimulatory effects of oral CC-4047/ACTIMID™ at 5 mg/day in patients with advanced multiple myeloma (our unpublished observations).

In conclusion, the co-stimulatory activity of Thd and two clinically relevant IMiD analogues correlates with the ability to prevent the induction of surface TNFR2 expression. Subsequent IL-2-dependent co-stimulation of both CD4+ and CD8+ T cells leads to the superinduction of TNF-α. These effects are in contrast to those seen with activated monocytes/macrophages and help to explain the diverse effect of Thd and its analogues on the TNF family. Further studies are required in order to delineate the mechanisms of action of these novel and potentially important compounds and tailor their use to the relevant clinical settings.

Acknowledgements

This work was supported by Celgene Corporation, NJ, USA. We also acknowledge the help of Kneka Belonwu in the collection of clinical samples and Dr Peter Schafer and Dr Laura Corral for helpful comments during the preparation of the manuscript.

Ancillary