Enhancement of CD154/IL4 proliferation by the T follicular helper (Tfh) cytokine, IL21 and increased numbers of circulating cells resembling Tfh cells in chronic lymphocytic leukaemia

Authors


Correspondence: Dr Simon D. Wagner, MRC Toxicology Unit, Hodgkin Building, University of Leicester, Room 323, Lancaster Road, Leicester LE1 9HN, UK.

E-mail: sw227@le.ac.uk

Summary

Chronic lymphocytic leukaemia (CLL) cells encounter T-cells and proliferate in response to T-cell signals in the lymph node microenvironment. In this report we determined interleukin 21 (IL21) function in CLL and showed that IL21 and interleukin 4 (IL4) act co-operatively to promote leukaemic cell proliferation without apoptosis or differentiation We further show that IL21 increased side population (SP) cells, which are associated with resistance to chemotherapy and increased self-renewal capacity in CLL. IL21 and IL4 are the major cytokines produced by the recently described CD4+ T follicular helper (Tfh) cell subset. Determination of Tfh cells in peripheral blood showed that patients had significantly increased numbers as compared to normal subjects although no association was found between Tfh numbers and IGHV gene mutational status or clinical stage. Our data suggests that the Tfh cytokines, IL4 and IL21, contribute to driving leukaemic cell proliferation in the lymph node microenvironment, and may contribute to the specific production of cells resistant to conventional chemotherapy. We suggest that increased circulating Tfh cells is a component of T-cell dysregulation in CLL. Our findings have implications for the therapeutic use of IL21.

Interleukin 21 (IL21) and interleukin 4 (IL4) are produced by a recently described CD4+ T-cell subset, T follicular helper cells, Tfh (Chtanova et al, 2004; Nurieva et al, 2007; Bauquet et al, 2009; Reinhardt et al, 2009; Vijayanand et al, 2012). IL21 acts directly on B-cells in the normal germinal centre reaction (Linterman et al, 2010) and is essential for normal germinal centre responses, (Linterman et al, 2010). It also has roles in regulating the differentiation of B-cells into antibody secreting cells (Bryant et al, 2007) and is required for generation of Tfh (Nurieva et al, 2008). IL4 promotes the survival and proliferation of chronic lymphocytic leukaemia (CLL) cells in vitro (Fluckiger et al, 1992; Kay & Pittner, 2003; Steele et al, 2010) but IL21, when administered alone, appears to be pro-apoptotic (de Totero et al, 2006, 2008; Gowda et al, 2008), which prompted speculation that it might be an effective therapy for this disease (Kolb, 2008). The effects of this combination of cytokines on CLL cells have not previously been determined.

Abnormalities of T-cell number (Catovsky et al, 1981; Gonzalez-Rodriguez et al, 2010; D'Arena et al, 2011; Nunes et al, 2012) and function (Whelan et al, 1982; Riches et al, 2010) are well-described features of CLL. These perturbations may contribute to the immune dysfunction associated with this disease but they may also contribute to the growth and survival of the leukaemic cells, especially in the lymph node microenvironment in which the CLL cells encounter T cells.

T follicular helper cells cells are a subset of memory T cells. Central memory (Tcm) and effector memory (Tem) subsets differ in their homing ability, proliferative responses to antigen or cytokines and effector functions. Surface marker expression defines Tcm and Tem populations. CCR7, L-selectin and the integrin LFA-1 (ITGAL) characterize Tcm, which together with naïve T cells are found in the T zones of secondary lymphoid organs. These cells have limited capacity to secrete cytokines. By contrast, Tem do not express CCR7, secrete cytokines and home to inflammation within non-lymphoid tissue. Tfh cells are a CD4+ helper T-cell subset that lack expression of CCR7, specifically differentiate under the influence of the transcription factor BCL6 and show high expression of the chemokine receptor CXCR5 and inducible T-cell co-stimulator (ICOS) (Breitfeld et al, 2000). More recently, Tfh have been shown to express surface PD1 (PDCD1) at high level (Rasheed et al, 2006; Haynes et al, 2007; Linterman et al, 2009a). Circulating Tfh numbers have not specifically been determined in CLL.

The microenvironment is likely to contribute to resistance to chemotherapy and understanding the mechanisms of resistance might determine choice of future therapies or novel therapeutic routes. The dye Hoechst 33342 emits at two wavelengths (depending on the intracellular location) when excited by ultraviolet light and cells excluding the dye (due to expression of an ATP-binding cassette transporter protein causing dye efflux) have a distinct profile (side populations, SP) on flow cytometry. SP cells are of interest because they have self-renewing or stem cell-like properties in cell lines and primary cells (Hirschmann-Jax et al, 2004, 2005). Previous work demonstrated that a sub-population of circulating CLL cells has the properties of SP cells as defined by exclusion of the dye, Hoechst 33342 (Foster et al, 2010). SP cells are not detectable in mononuclear cells from normal subjects but are found in CLL patients and expand following administration of chemotherapy in vitro, leading to the suggestion that they represent a chemoresistant population (Foster et al, 2010).

A novel role for IL21 was suggested by experiments in which over-expression in transgenic mice produced expanded haemopoietic progenitor cell populations (Ozaki et al, 2006). Resistance to chemotherapy agents can be induced in vitro by culture conditions to mimic the microenvironment in CLL (Willimott et al, 2007a) and we, therefore, determined the effects of cytokines on SP cell numbers.

We show that IL21 enhances IL4 driven proliferation and we suggest that Tfh cells are a biologically important component of the expanded CD4+ T-cell population in CLL.

Materials and methods

Patient samples

All samples from patients with CLL (Table SI), diagnosed by typical immunophenotyping and peripheral blood morphology, were collected after approval from the Local Research Ethics Committee and the Research and Development Office of the University Hospitals of Leicester NHS Trust (UHL09723). Patients were untreated or had not received treatment for more than 3 months and had peripheral blood white cell counts >50 × 109 per l. No patient included in the study had received fludarabine-based treatments. Fifty-four patients were included in the study with a male to female ratio of 36:18. The majority of patients were Binet stage A (40/54). Peripheral blood mononuclear cells were separated from the cell suspension by density gradient centrifugation. Heparinized whole blood [diluted 1:1 with phosphate-buffered saline (PBS)] was gently layered onto a 15 ml Ficoll layer (Histopaque 1077; Sigma Aldrich, St. Louis, MO, USA), without a red cell lysis step, prior to centrifugation (430 g for 30 min) and removal of the mononuclear cell layer at the interphase. Further centrifugation (250 g for 10 min) and removal of supernatant was performed prior to resuspension in RPMI 1640 medium (Invitrogen, Paisley, UK) with 10% fetal calf serum, 2 mmol/l glutamax, and penicillin/streptomycin). IGHV gene mutational status was obtained using standard techniques (van Dongen et al, 2003).

Fresh tonsils were obtained after tonsillectomy. They were transported in RPMI 1640 media on ice and single cell suspensions were produced by pressing tissue through a 70 μm cell strainer (BD Falcon; BD Biosciences, Oxford, UK) before antibody staining and flow cytometry.

Cell culture

Isolated mononuclear cells were incubated at 3 × 106 per ml in RPMI 1640 (Lonza, Slough, UK) with 10% fetal calf serum (Lonza). Soluble CD154 (CD40LG) (6245-CL-050; R&D Systems, Abingdon, UK) and anti-CD40 antibody (EA5) (sc-65264; Santa Cruz Biotechnology, Santa Cruz, CA, USA) were added as described (Jacob et al, 1998) together with either IL4 (10 ng/ml) (204-ML-050; R&D Systems) or IL21, or a combination of the two cytokines. We directly compared IL21 produced in bacteria (Miltenyi Biotec, Bisley, UK) with that produced in mammalian cells to demonstrate greater potency of the mammalian-derived cytokine (Figure S1). Dose response curves for IL21 (5849-ML-025; R&D Systems) were carried out (Figure S2) demonstrating that maximum effect of IL21 was achieved at 20 ng/ml and this concentration was, therefore, employed.

Flow cytometry

Whole blood (50 μl) diluted 1:1 with PBS was incubated with antibody for 30 min at room temperature in the dark before red cell lysis (FACScan Lysis Buffer; BD Biosciences) and analysis on a FACS Canto II (BD Biosciences, Oxford, UK) using flowjo software (Tree Star Inc., Ashland, OR, USA). Antibodies used included anti-CD19 (catalogue number 561295), anti-CD4 (catalogue number 561841), anti-ICOS (catalogue number 557802), anti-CD23 (anti-FCER2) (catalogue number 332782), anti-IL21R (catalogue number 560331), anti-CD154 (anti-CD40LG) (catalogue number 558988), anti-CD138 (SDC1) (catalogue number 552723) and anti-PD1 (anti-PDCD1) (catalogue number 557860) all from BD Bioscience (San Diego, CA, USA) and anti-CXCR5 (catalogue number 335001; BioLegend, San Diego, CA, USA). Details of the antibodies and their fluorochrome conjugates are shown in Table SII.

Reverse transcription and polymerase chain reaction (PCR)

RNA was produced using an RNeasy kit (Qiagen, Crawley, UK) and reverse transcribed using the ProtoScript® M-MuLV first strand cDNA synthesis kit (New England Biolabs, Ipswich, MA, USA). PRDM1 was PCR-amplified using forward primer 5′-tggacatggaggatgcggatatg and reverse primer 5′ aggtcctttcctttggacgggttg (Shaffer et al, 2000). Reactions were set up with TopTaq™ PCR master (Qiagen) with a final primer concentration of 0·2 μmol/l.

Detection of side-population (SP) cells

5 × 106 cells were cultured in Iscove's Modified Dulbecco's Media (IMDM) supplemented with 25 mmol/l HEPES, 2% fetal calf serum (FCS, Lonza), and penicillin/streptomycin at 37°C for 36 h. Cell concentration was then adjusted to 1 × 106 cells/ml and 5 μl/ml Hoechst dye (Sigma Aldrich) was added followed by incubation at 37°C for 60 min with occasional agitation. Verapamil (Sigma Aldrich) 50 μmol/l was employed to validate the SP phenotype. Cells were washed ready for analysis whilst kept on ice and in the dark. Data acquisition was performed using the FACS Aria II fitted with a UV laser (355 nm). Hoechst blue fluorescence was detected on a band-pass (BP) 455/50 filter and red fluorescence on a BP 660/20 filter.

Immunohistochemistry

Paraffin-embedded CLL lymph node sections were obtained from archived material in the Department of Histopathology, Leicester Royal Infirmary. A total of six lymph nodes were employed in this study. The tissue sections were treated with xylene followed by antigen retrieval in a microwave oven (700 W for 20 min) in pH 9 Tris-EDTA buffer before staining with anti-PD1 antibodies (gift of Dr Giovanna Roncador, Madrid, Spain). Primary antibody was incubated with tissue sections for 3 h at room temperature and detected using Novolink Polymer Detection System (Novocastra, Milton Keynes, UK). Tissue sections were counterstained with haematoxylin, dehydrated, re-paraffinized, and mounted.

[3H] thymidine incorporation

Freshly isolated mononuclear cells from CLL patients (1 × 106/200 μl) were cultured in triplicate in RPMI 1640 media supplemented with FCS), penicillin, streptomycin, and glutamate (Sigma Aldrich), with various combinations of IL4 and IL21 (R&D Systems) for 24 h. [3H] thymidine (37 Bq/well) was added for the following 36 h of culture. Cells were harvested onto filter paper, dried fully, and covered in scintillation fluid before measurement of radioactivity.

BrdU incorporation

Chronic lymphocytic leukaemia cells from the various culture conditions were labelled with bromodeoxyuridine (BrdU) (BD Pharmingen; BD Biosciences). Briefly, 10 μmol/l of BrdU was added to each culture condition for the last 18 h of culture. Cells were collected, fixed, and permeabilized before being resuspended in DNAse (30 μg of DNAse/106 cells) for 60 min at room temperature to expose incorporated BrdU. Samples were then stained with fluorescein isothicyanate (FITC)-conjugated anti-BrdU antibody for 20 min. Finally, total DNA was stained with 7-aminoactinomycin D (7-AAD) prior to resuspension in buffer and flow cytometric analysis.

Apoptosis detection

Apoptosis studies were performed using BD Pharmingen Annexin V Apoptosis Detection Kit (BD Biosciences). 1 × 106 cells were taken from each culture condition, washed in PBS, and resuspended in annexin buffer before the addition of 5 μl of FITC Annexin V and propidium iodide and incubated for 15 min on ice in the dark. Four-hundred micro litre of annexin buffer was added to each sample tube ready for flow cytometry analysis.

Western blot

Cells were harvested, washed in PBS and lysed in 30–50 μl radioimmunoprecipitation assay (RIPA) buffer with the addition of phosphatase (1:100) and protease inhibitors (1:50) (Sigma Aldrich) for 10 min on ice. Lysates were then centrifuged at 13 800 g for 10 min at 4°C and supernatants collected and frozen at −20°C. 40–60 μg of protein was added to RNA-free water and 10 μl of loading buffer to a total volume of 30 μl. Each sample was boiled for 5 min before loading into wells of a 7·5% polyacrylamide gel alongside a prestained protein ladder (Hyperladder V; BioRad, Hercules, CA, USA) to allow molecular weight estimation. Gel was run in an electrophoresis tank for 70 min at 100V.Wet transfer of proteins was performed using polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA) soaked in methanol. The following primary rabbit polyclonal antibodies were employed: anti-phospho-STAT1Tyr701, anti-phospho-STAT3Tyr705, anti-phospho-STAT5Tyr694, anti-STAT1, anti-STAT3 and anti-STAT5, all from Cell Signaling Technology (Denvers, MA, USA) and employed at a dilution of 1:1000.

Statistics

All statistical analyses were performed using graphpad prism 4.0 (GraphPad Software; Inc., La Jolla, CA, USA). The nonparametric Mann–Whitney U-test was used to test the differences between mutated IGHV gene (>2%) and unmutated IGHV gene (<2%) cases. Paired 2-tailed Student t test was used for analysis of IL21 receptor expression in resting or activated CLL B cells and CD154 expression on T cells. < 0·05 was considered significant.

Results

IL21 causes proliferation without apoptosis or terminal differentiation

CD154 stimulation increases IL21R expression on CLL cells (de Totero et al, 2006), a result we confirmed (Figure S3) suggesting that the leukaemic cells are potentially capable of responding to this cytokine. Because both IL4 and IL21 are products of Tfh cells we determined the effects of these cytokines on CLL proliferation and apoptosis (n = 21). As anticipated from previous work (Willimott et al, 2007b), sCD154 produced a significant (< 0·001, Mann–Whitney U-test) increase in [3H] thymidine incorporation with the addition of IL4 producing a further increment (< 0·01) (Fig 1A). There were significant differences in the median [3H] thymidine incorporations between sCD154/IL4 and sCD154/IL4/IL21 (< 0·02, Mann–Whitney U-test) and sCD154/IL21 and sCD154/IL4/IL21 (< 0·03) suggesting that the combination of cytokines enhances CLL proliferation. To support this finding we determined BrdU incorporation (n = 6) and showed that cells in S phase increased from a mean of 17% with sCD154/IL4–36% with sCD154/IL4/IL21 (Fig 1B). In order to assess individual differences in response [3H] thymidine incorporation with sCD154/IL4, sCD154/IL21 or sCD154/IL4/IL21 was normalized to the activity with sCD154 alone (Fig 1C). sCD154/IL4 produced >2·5-fold increase in proliferation in four patients and sCD154/IL21 in three patients whereas sCD154/IL4/IL21 produced this level of increase in all patients. To examine specifically the effect of IL21, [3H] thymidine incorporation with sCD154/IL4/IL21 was normalized to sCD154/IL4 (Fig 1D). All but two patients showed a greater than two-fold increase in proliferation. Overall we showed that IL21 enhances proliferation due to sCD154/IL4.

Figure 1.

Effects of IL21 on [3H] thymidine incorporation. (A) Scatter plot showing [3H] thymidine incorporation for each culture condition (n = 17) (Patients 2, 7, 9, 10, 12, 15, 18, 23, 25–27, 37, 43, 48, 52-54 in Table SI). Horizontal black line represents the median. For statistical analysis proliferation was compared between either sCD154/IL4 or sCD154/IL21 and sCD154/IL4/IL21 (*< 0·05; Mann–Whitney U-test). (B) The proportion of cells within S-phase for each culture condition was measured by BrdU incorporation (n = 6). Mean ± SEM. (C) To demonstrate individual responses the [3H] thymidine incorporation in the sCD154/IL4, sCD154/IL21 and sCD154/IL4/IL21 culture conditions was normalized to sCD154 culture. The dotted horizontal line is set at 2·5 to allow comparison between the different culture conditions. (D) Similarly, [3H] thymidine incorporation in the sCD154/IL4/IL21culture condition was normalized to sCD154/IL4 culture to demonstrate the effect of IL21 for individual patients.

IL21 has been reported to cause mature B-cell apoptosis (Jin et al, 2004; Gowda et al, 2008; Sarosiek et al, 2010) or differentiation (Ozaki et al, 2002; Ettinger et al, 2005). We did not observe apoptosis, or a specific effect of IL21 on the B-cell terminal differentiation markers, CD138 or PRDM1 (Figure S4).

sCD154/IL4 primes leukaemic cells to the effects of IL21

Following a report (Saito et al, 2008) that normal mouse B-cells show enhanced proliferative responses if IL4 is administered before IL21, we developed a protocol for the sequential addition of these cytokines. CLL cells were exposed to sCD154 and IL4 for 24 h prior to the addition of IL21 and [3H] thymidine was added 24 h after IL21 so that direct comparisons of IL21 response with the standard schedule could be made.

The increased time in culture prior to [3H] thymidine addition produced higher proliferative responses in all conditions compared to that seen with the standard schedule (Fig 2A), but comparison of [3H] thymidine incorporation relative to sCD154/IL4 alone demonstrated that sCD154/IL4/IL21 in the delayed condition produced significantly more proliferation than the same combination in standard culture conditions (Fig 2B).

Figure 2.

[3H] thymidine incorporation following pre-incubation with sCD154/IL4 and sequential addition of IL21. (A) Scatter plot showing [3H] thymidine incorporation (n = 8) (Patients 2, 7, 9, 10, 18, 23, 27, 37 in Table SI). (*< 0·05, **< 0·01; Mann–Whitney U-test). (B) Box and whisker plot showing [3H] thymidine incorporation relative to sCD154 alone. Delayed addition of IL21 produced significantly more [3H] thymidine incorporation than simultaneous addition of IL4 and IL21 (< 0·03, Mann–Whitney U-test). (C) Western blot results showing expression of phosphorylated and total STAT1, STAT3 and STAT5. Lysates were obtained from freshly isolated cells (T = 0), and after culture on tissue culture plastic (−) and with sCD154, sCD154/IL4, sCD154/IL21 and sCD154/IL4/IL21.

IL21 can activate STAT1, STAT3 and STAT5. In the culture system, STAT1 was phosphorylated under all culture conditions, sCD154 caused STAT5 phosphorylation whilst only the addition of IL21 caused STAT3 phosphorylation (Fig 2C). Collectively, our data suggests that sCD154 and IL4 prime the leukaemic cells to enhance the proliferative effects of IL21, which are exerted through STAT3.

IL21 induces side population (SP) cells

Side population cells are associated with resistance to chemotherapy in CLL and we determined SP populations following IL21 culture in order to ascertain whether this could be a potential mechanism for chemoresistance. Cells showing the SP phenotype were viable (Fig 3A) and incubation with the calcium channel blocker verapamil reduced their numbers (Fig 3B), confirming that active dye efflux was responsible for this population. An SP population was detectable in freshly isolated CLL cells (n = 7) but, in line with the results of others (Foster et al, 2010), not in peripheral blood mononuclear cells of normal subjects (n = 5) (= 0·01, Mann–Whitney U-test) (Fig 3C). Neither sCD154 nor sCD154/IL4 significantly increased cell numbers. sCD154/IL21 caused increased SP cell numbers in all patients with two patients producing exceptionally high numbers at 0·6 and 8% of leukaemic cells (Fig 3D). SP cell numbers were significantly increased as compared to sCD154 alone (< 0·02, Mann–Whitney U-test) (Fig 3E). Surprisingly, the combination of sCD154/IL4/IL21 suppressed SP numbers.

Figure 3.

Side population changes due to IL21. CLL cells were cultured for 24 h and stained with Hoechst 33342. (A) Viability was assessed by propidium iodide (PI) staining. Populations staining positive or negative for PI are present in the ungated cells but only PI negative cells are found in the SP populations. (B) Staining in the presence of 50 μmol/l Verapamil was carried out to determine whether the observed SP phenotype was due to efflux of the Hoechst dye. This drug caused reduction in SP numbers. (C) Comparison of percentage of SP cells in peripheral blood mononuclear cells of normal subjects and patients with CLL. = 0·01, Mann–Whitney U-test. (D) Examples of flow cytometry dot-plots obtained after staining leukaemic cells cultured in various conditions with Hoechst 33342. Results for two patients (upper row Patient 21, and lower row Patient 33) are shown. The upper row demonstrates increased SP cell numbers due to IL21 whereas in the lower row this is caused by the combination of IL4 and IL21. (E) Scatter plots comparing SP cell number changes in CLL patients (n = 7) (Patients 7, 18, 21, 33, 38, 42, 53 in Table SI). The combination of CD154/IL21 produces more SP cells than CD154 or CD154/IL4 (< 0·05, Mann–Whitney U-test).

PD1 and IL21 expressing cells are present in CLL lymph nodes

In order to demonstrate a Tfh population in CLL lymph nodes we stained lymph nodes (n = 6) with anti-PD1(Pangault et al, 2010) (Fig 4). PD1+ cells were readily detected. PD1+ cells could be concentrated in proliferation centres (Fig 4A) or more diffusely distributed (Fig 4B). A tonsil control was carried out for each CLL lymph node in order to ascertain that the level of staining was similar to that present in normal germinal centres (Fig 4C). Tfh cells are the principle producers of IL21 and we also demonstrated IL21 expressing cells in CLL lymph nodes (Fig 4D). Utilizing tonsil sections as a control, we demonstrated IL21 staining in the germinal centre staining with the mantle zone being negative (Fig 4E,F).

Figure 4.

PD1 and IL21 expression in CLL lymph nodes. PD1 expression in CLL lymph nodes. (A) CLL lymph node demonstrating focal staining within proliferation centres and (B) more diffuse staining. Magnification of main picture is ×10 and inset ×40. (C) Anti-PD1 staining in tonsil from a subject with chronic tonsillitis Main picture shows germinal centres (GC) (×10) containing PD1 expressing cells. Inset (×40) shows the border between the germinal centre containing PD1 expressing cells and the surrounding mantle zone. (D) IL21 expression in a CLL lymph node (×40). (E) Low power (×10) view of tonsil stained with anti-IL21. Germinal centres (GC), mantle zones and the interfollicular (T-cell) regions are indicated. An area of these normal germinal centres that stains more heavily with anti-IL21 corresponds to the T-cell rich area.

Increased numbers of cells with the surface characteristics of T follicular helper (Tfh) cells

T follicular helper cells numbers in CLL have not been determined. In order to produce a robust standard for the definition of circulating Tfh, we first analysed tonsils (n = 12) for expression of CXCR5 and ICOS by flow cytometry. We employed a gate with a mean fluorescence intensity above the 70th percentile for both CXCR5 and ICOS and showed that this gave results for Tfh numbers in line with the current literature for tonsil and reactive lymph node (Pangault et al, 2010) and we followed this by applying these gates to peripheral blood from normal subjects and CLL patients (Figure S5). Human peripheral blood Tfh, have been defined as CD4+CD45RO+CXCR5+ (Morita et al, 2011). Therefore, we again utilized the 70th percentile threshold obtained with tonsillar cells to show that the CD4+CXCR5hi population in peripheral blood was CD45RO+ (Figure S6) and we utilized CD4+CXCR5hi as a definition of Tfh cells in order to compare cell numbers directly between patients.

We analysed CD4+CXCR5hiICOShi and CD4+CXCR5hi cells by flow cytometry in patients with CLL (n = 51) and age-matched normal subjects (n = 13). Median numbers of CD4+CXCR5hiICOShi cells were significantly increased in patients with CLL (< 0·05, Mann–Whitney U-test) (Fig 5A), and this difference was maintained when numbers of CD4+CXCR5hi cells were compared between the two groups (< 0·03, Mann–Whitney U-test) (Fig 5B). IGHV gene mutational status, serum lactate dehydrogenase, lymphocyte doubling time and gender were not associated with significant differences in median numbers of CD4+CXCR5hi cells (Fig 5C).

Figure 5.

Circulating Tfh cells and correlation with prognostic markers. Comparison of (A) CD4+CXCR5hiICOShi and (B) CD4+CXCR5hi cell numbers between patients (n = 51) (Patients 1–51 in Table SI) and normal subjects (n = 13). Circulating Tfh cells are increased in patients with CLL. There are significant differences between CD4+CXCR5hiICOShi (< 0·05, Mann–Whitney U-test) and CD4+CXCR5hi (< 0·03) in patients and normal subjects. (B) Scatter plots showing lack of association between CD4+CXCR5hi numbers and IGHV gene mutational status (M, mutated or U, unmutated, serum lactate dehydrogenase (greater or less than the median value), lymphocyte doubling time (>12 months or <12 months), clinical stage (by Binet group A, B or C) and gender.

Discussion

The lymph node microenvironment is responsible for proliferation, survival and trafficking of malignant cells. In CLL the relationship between leukaemic cells and the microenvironment is believed to mirror that between normal B-cells and the normal cellular elements of the lymph node, i.e. stromal cells and T cells (Burger et al, 2009). The balance between repressive (Treg) and stimulatory T-cell subsets may contribute to determining clinical outcome in lymphoproliferative disease (Ansell et al, 2001; Lee et al, 2006). Such considerations have suggested that T cells or their cytokine products may be therapeutic targets in both B-cell (Ramsay et al, 2012) and other malignancies (Brahmer, 2012), but the biological complexity of the microenvironment suggests a need for greater understanding of interactions between cellular and growth factor elements in order to introduce immunomodulatory agents in a rational manner.

IL21 has been reported to cause apoptosis of CLL cells and this has led to interest in its therapeutic potential (Kolb, 2008), although a role for IL21 signalling in driving B-cell proliferation has led to the opposite suggestion (Good et al, 2006). Indeed, IL21 blocking agents, i.e. anti-IL21 antibodies, are currently under investigation in animal models of autoimmunity (Young et al, 2007) and phase 2 clinical trials in rheumatoid arthritis. Recently a phase I trial of rituximab and IL21 was reported (Timmerman et al, 2012) which demonstrated some efficacy in CLL.

In this report we have evaluated proliferative responses to the combination of IL4 and IL21. This combination of cytokines was chosen because they are major products of Tfh cells, a CD4+ subset that is essential for normal B-cell maturation in germinal centres (Linterman et al, 2010), which are expanded in autoimmune disease (Simpson et al, 2010; Terrier et al, 2012). IL21 is implicated in driving terminal differentiation of germinal centre B-cells to plasma cells (Linterman et al, 2010; Zotos et al, 2010; Ding et al, 2013) through effects of STAT3 at the PRDM1 (BLIMP1) locus (Ding et al, 2013). Our results in CLL demonstrate that IL21 causes STAT3 phosphorylation at Tyr705 and proliferation without evidence of apoptosis or terminal differentiation, and argue against its therapeutic use. In line with our data, predominantly proliferative effects of IL21 have recently been demonstrated in Waldenstrom macroglobulinaemia (Hodge et al, 2012). Increased IL21R expression correlated with aggressive disease in follicular lymphoma (Wood et al, 2012), again suggesting that IL21 responses associate with proliferation and not apoptosis. In addition, IL21 causes proliferation of normal B-cells (Good et al, 2006) and the role of Tfh, in autoimmunity (Linterman et al, 2009b) including increased serum IL21 in systemic lupus erythematosus (Simpson et al, 2010) is consistent with a B-cell proliferative effect.

We demonstrated that pre-incubation with sCD154/IL4, causing increased IL21R expression (de Totero et al, 2006), enhanced the proliferative effects of IL21, supporting the view that IL21 mediated proliferation through engagement with the IL21R. IL21 produced activation of STAT3 (Fig 4C) in CLL. STAT3 activation has been associated with IL21 driven proliferation of naïve and memory mouse B-cells (Good et al, 2006) and we suggest that IL21 exerts its proliferative effects on CLL cells through IL21R-mediated induction of STAT3.

However, IL21 has been reported to cause apoptosis in mature B-cell malignancies including CLL (de Totero et al, 2006; Akamatsu et al, 2007; Gowda et al, 2008; Sarosiek et al, 2010). We noted that recombinant bacterial protein produced results that were quantitatively very different to those produced by a cytokine produced in mammalian cells and we speculate that the source of IL21 could be a potential cause of variability.

Side population cells characterize haemopoietic stem cell populations (Goodell et al, 1997) and have self-renewing or stem cell-like properties (Hirschmann-Jax et al, 2004, 2005). When compared directly in xenograft transplantation studies, SP cells showed much greater ability to initiate cancer than non-SP cells (Hu et al, 2010). SP cells have previously been described in the peripheral blood of CLL patients, but not normal subjects, and are associated with resistance to chemotherapy (Foster et al, 2010). Work from the same group suggested that SP cells in CLL have greater self-renewing capacity than non-SP cells (Gross et al, 2010). Our work raises the possibility that cytokines play a role in producing SP cells in the lymph node microenvironment, suggesting a novel mechanism for chemotherapy resistance in this context.

Abnormalities in circulating T cells in CLL, as compared to normal subjects, have been thoroughly described (Nunes et al, 2012). This is the first report to characterize the numbers of circulating Tfh cells in CLL. Tfh cell numbers are significantly increased in patients although there was no association with some of the important prognostic markers, suggesting that they do not have a role in determining clinical outcome. Mouse models provide evidence that T-cell abnormalities are a consequence of the leukaemia (Hofbauer et al, 2011) but also that T cells are a component of the microenvironment that is required by the leukaemic B-cells (Bagnara et al, 2011). One possibility is that increased Tfh numbers are a consequence of the disease [forming part of the spectrum of T-cell abnormalities reported in CLL (Riches et al, 2010; Nunes et al, 2012)], and provide cytokines necessary for leukaemic cell survival and proliferation. In support of this view, CLL cells have been shown to secrete IL6 (Buggins et al, 2008), which is an essential cytokine for Tfh cell differentiation (Eto et al, 2011).

We have shown that CD154 (CD40LG) and IL4, a cytokine produced by Tfh cells, enhances the effects of IL21, another Tfh cytokine, in driving proliferation of CLL cells. Although trialled in CLL this result suggests that IL21 will have limited clinical utility. We speculate that a previously unrecognized function of IL21 is to induce the production of cells bearing ABC transporter proteins (SP cells) that are resistant to chemotherapy and have been associated with self renewal capacity. The generation of SP cells in the microenvironment may be important to understanding resistance of leukaemic cells within lymph nodes to current chemotherapy.

Acknowledgements

The anti-PD-1 antibody was a gift from Dr Giovanna Roncador, Monoclonal Antibodies Unit, Biotechnology Programme, Spanish National Cancer Research Centre (CNIO), Madrid, Spain. This work was supported in part by grants from the Kay Kendall Leukaemia Fund and Leukaemia and Lymphoma Research to SDW. MJA was supported by an unrestricted grant from Roche (Welwyn Garden City, UK) and by grants from the Leicester Haematology Research Fund.

Author contributions

MJA, SW and LP designed and carried out experiments and analysed results. MJA and SDW designed the study, analysed data and wrote the paper. DBK, FM and MJSD revised the paper.

Ancillary