Evidence for a pathophysiological role of cysteinyl leukotrienes in classical Hodgkin lymphoma

Abstract

Classical Hodgkin lymphoma (cHL) is characterized histologically by a minority of malignant Hodgkin Reed-Sternberg cells surrounded by abundant inflammatory cells, generally believed to be of major importance in the pathophysiology of the disease. Here, we present data that link inflammatory cell-derived arachidonic acid metabolites, the cysteinyl leukotrienes (CysLT), to the pathogenesis of cHL. Two HL cell lines, L1236 and KMH2, were shown to express functional CysLT₁ receptors, responding with a robust calcium signal upon leukotriene (LT) D₄ challenge. LTD₄ stimulated protein release of tumor necrosis factor-α, interleukin-6 and -8 by L1236 cells and interleukin-8 by KMH2 cells. Importantly, all these LTD₄-induced effects were blocked by the CysLT₁ receptor-specific antagonist zafirlukast. Immunohistochemical studies of cHL biopsies and microarray analysis of microdissected cells revealed that the CysLT₁ receptor is expressed also by primary Hodgkin Reed-Sternberg cells. As these cells are surrounded by CysLT-producing eosinophils, macrophages and mast cells, our results suggest the CysLTs as mediators in the pathogenesis of cHL, contributing to the aberrant cytokine network of this lymphoma. © 2008 Wiley-Liss, Inc.

Classical Hodgkin lymphoma (cHL) tumor tissue is characterized by a minority of malignant Hodgkin Reed-Sternberg (H-RS) cells interspersed among an abundant infiltrate of inflammatory cells.1 Histologically, cHL is divided into 4 different subgroups: nodular sclerosis (NS), mixed cellularity (MC), and the less common lymphocyte rich (LR) and lymphocyte depleted (LD). cHL, mainly NS, is increasing among adolescents and young adults in the Western world.2 Most patients with cHL are cured by chemotherapy and radiotherapy. However, patients with very high-risk features may have a poor prognosis.3 Eosinophils and mast cells are usually present within the inflammatory component, and their degree of infiltration has been shown to inversely correlate with prognosis.4, 5 Several lines of experimental data suggest interdependency between the H-RS cells and the bystander cells, contributing to the specific features of the tumor. In this respect, the wide array of cytokines and chemokines produced by the H-RS cells are of particular importance. These molecules not only participate in the attraction of inflammatory cells to the tumor site but also mediate certain cell-specific stimulatory and inhibitory effects on the tumor-associated cells, ultimately leading to a microenvironment favoring persistence of the H-RS cells.6 Furthermore, expression of chemokine/cytokine receptors by H-RS cells constitutes the foundation of autocrine loops promoting tumor survival and progression, exemplified by the interleukin (IL)-13—IL-13 receptor axis.7

Traditionally, leukotrienes have been considered as key players in inflammatory diseases.8 The cysteinyl leukotrienes (CysLTs) LTC₄, LTD₄ and LTE₄ are arachidonic acid-derived lipid mediators implicated in a diverse array of biological responses. Activated eosinophils, basophils, mast cells and macrophages are the major sources of CysLTs.8 Phospholipase A₂, the enzyme initiating the biosynthesis of leukotrienes in response to certain stimuli, catalyzes the liberation of arachidonic acid from membrane phospholipids. In concert with 5-lipoxygenase-activating protein, 5-lipoxygenase oxidizes free fatty acid to 5-hydroperoxy-eicosatetraenoic acid, followed by subsequent dehydration to yield LTA₄.8 This compound can be further metabolized to LTC₄, which is exported extracellularly for conversion to LTD₄ and LTE₄, via glutamyl transpeptidase or dipeptidase, respectively. CysLT signaling has been shown to be mediated mainly by the 7-transmembrane G-protein coupled receptors CysLT₁ and CysLT₂. These receptors are expressed by bronchial smooth muscle cells, mast cells, monocytes, macrophages and eosinophils.8 The CysLT₂ receptor is also expressed by cardiac Purkinje cells, adrenal medulla cells, coronary smooth muscle cells9 and endothelial cells.10 The CysLTs are potent bronchoconstrictors, known to induce granulocytic infiltration, vascular permeability, edema formation and airway remodeling, all important features of asthma.8 Antileukotrienes, clinically used in the treatment of asthma, have been shown to promote bronchodilatation11 and decrease the number of asthma exacerbations.12 A putative role of leukotrienes in carcinogenesis has also been suggested.13–15 Thus, given the inflammatory features of cHL, we investigated the biosynthesis and effects of leukotrienes in this lymphoma entity. This report elucidates the expression and function of the CysLT receptors in H-RS cells.

Material and methods

Cell lines, biopsies and patients

The human HL cell lines L1236, HDLM2, KMH2, L428 and L591 (kind gifts from Prof. Volker Diehl, Cologne, Germany) were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum, L-glutamine (2 mM), penicillin (100 U/ml) and streptomycin (100 μg/ml) (Gibco, Paisley, Scotland, UK) at 37°C in an atmosphere containing 5% CO₂. The examined cell lines were of B-cell phenotype (L1236, L428 and KMH2) and T-cell phenotype (HDLM2) and of subtype MC (L1236 and KMH2) and NS (L428 and HDLM2). All cell lines, except for L591, were negative for Epstein-Barr virus (EBV). The cHL tumors studied by immunohistochemical staining (Table I) were of subtype NS (n= 8), MC (n = 8), LR (n = 3) and LD (n = 1) according to the WHO classification.16 EBV status of the tumors was assessed by EBER in situ hybridization and/or LMP1 immunohistochemistry. The patient ages ranged from 10 to 81 years with a median of 37 years. The study was approved by the local ethics committee.

Table I. Patient and Sample Characteristics

Patient number	Sex	Age	Clinical stage	HL subtype	EBV status1	H-RS cells/high power field	H-RS cells positive for the CysLT1 receptor (%)2	Arbitrary CysLT1 receptor expression level3
NA, not available. 1 In situ hybridization (EBER) and/or LMP1 immunostaining.^36 2 Immunohistochemistry. 3 The CysLT1 receptor expression of H-RS cells was compared to that of tumor tissue eosinophils by immunohistochemistry and classified as 0 (negative), + (weaker than eosinophils), ++ (equal to eosinophils) or +++ (stronger than eosinophils). 4 Micro arrayed cHL tumors. ND, not done.
1	Female	61	IIIA	NS	Neg	12	13	++
2	Female	31	IIIB	NS	Neg	47	70	+
3	Female	16	IIB	NS	Neg	20	20	+
4	Male	66	IIIA	NS	ND	70	27	+++
5	Male	10	IIB	NS	ND	16	20	+
6	Male	14	IIA	NS	Neg	14	7	+
7	Male	13	IIB	NS	Neg	33	35	+++
8	Male	16	IIA	NS	Neg	2	80	+
9	Female	16	IIA	MC	ND	27	20	+
10	Male	24	IVB	MC	ND	31	0	0
11	Male	37	IIIB	MC	Neg	109	50	++
12	Male	61	IIA	MC	ND	2	0	0
13	Male	17	IIIB	MC	Neg	3	50	+
14	Male	20	IIA	MC	Neg	21	0	0
15	Male	20	IIB	MC	Neg	13	20	+
16	Male	61	IA	LR	Neg	<1	0	0
17	Male	51	IA	LR	Neg	<1	0	0
18	Female	70	IIIA	LR	Neg	>1	0	0
19	Female	58	IIIB	LD	Neg	4	0	0
20	Male	81	IIA	MC	Pos	4	0	0
421	Male	10	IIA	NS	Pos	ND	ND	ND
422	Male	5	IA	NS	Pos	ND	ND	ND
423	Male	5	IIA	NS	Pos	ND	ND	ND
424	Female	15	IIB	NS	Neg	ND	ND	ND
425	NA	NA	NA	NS	ND	ND	ND	ND
426	NA	NA	NA	NS	Neg	ND	ND	ND
427	Male	12	IIB	NS	Pos	ND	ND	ND
428	Female	13	IIIB	NS	Pos	ND	ND	ND
429	Male	5	IIA	NS	Pos	ND	ND	ND

Microdissection, RNA amplification and microarrays

The cHL tumors (n = 9, patients aged 15 years or less) used for H-RS cell microdissection were all of NS subtype and comprised 5 EBV positive tumors, 3 EBV negative tumors and 1 tumor of undetermined EBV status according to LMP1 staining. Microdissection was performed using the PALM Laser Microbeam System (P.A.L.M. Microlaser Technologies GmbH, Bernried, Germany) on frozen HL sections stained with hematoxylin. The system allows contact-free isolation of cells from tissue sections. H-RS cells were identified according to their distinctive morphology. Nonmalignant reactive cells were also microdissected from each tumor. For the purposes of mRNA amplification ∼150–200 H-RS cells and 200–300 nonmalignant cells were picked from each of the HL tumors. For linear T7-based mRNA amplification the ExpressArt mRNA amplification kit17 (Artus, Hamburg, Germany) was used according to the manufacturer's instructions. Three rounds of linear amplification were performed on each sample in this study, in the final IVT reaction the resulting RNA was biotinylated. The resulting yields of amplified RNA were between 30–40 μg, derived from <10 ng of input total RNA. For microarray analysis, biotinylated cRNA was hybridized to HG-U133Plus2 microarrays (http://www.affymetrix.com/products/arrays/). A complete description of procedures is available at http://bioinf.picr.man.ac.uk/mbcf/downloads/GeneChip_Target_Prep_Protocol_CRUK_v_2.pdf. The target cDNA generated from each sample was processed according to the manufacturer's recommendation using an Affymetrix GeneChip Instrument System (http://www.affymetrix.com/support/technical/manual/expression_manual.affx). GCOS (gene chip operating software) signal for the mRNA expression of CysLT₁ and CysLT₂ in microdissected H-RS cells from 9 tumors, in corresponding nonmalignant reactive cells from 3 of the tumors and in purified germinal centre (GC) B-cells from reactive lymphoid tissue (tonsils of 3 patients from Birmingham Children's Hospital) were plotted. The detection of several house-keeping genes in the amplified RNA samples was assessed using GCOS data; both GAPDH and beta-actin mRNAs were called “present” on 10/15 arrays, either GAPDH or beta-actin “absent” on 2/15 arrays and both “absent” on the 3/15 arrays (data not shown).

RNA preparation and quantitative PCR

Total RNA was extracted using RNeasy Mini Kit (Qiagen GmbH, Hilden, Germany), including a DNA digestion step, or Ultraspec (Biotex Laboratory Inc., TX). The RNA was reverse-transcribed with MMLV-RT (Invitrogen, Lidingö, Sweden) and random hexanucleotides as described earlier.18 All reagents and equipment used for the quantitative PCR were purchased from Perkin Elmer Applied Biosystems (Foster City, CA). Samples were analyzed using an ABI Prism 7700 Sequence detector. Taqman Gene Expression Assays for CysLT₁ and CysLT₂ were performed according to the manufacturer's instructions to elucidate the mRNA expression in HL cell lines and primary H-RS cells. β-2-microglobulin was used to calculate the relative CysLT receptor expression. For cytokine mRNA analysis, the Cytokine Gene Expression Plate I was used. Also, Taqman Gene Expression Assays for TNF-α, IL-6, IL-8 and IL-13 were performed according to the manufacturer's instructions. β2-microglobulin and ABL were run as external controls. The system was linked to a Power-Macintosh-7200/120 containing the data analysis software ABI Prism™ 7200/7700 Sequence Detection system. The mRNA levels in LTD₄ stimulated cells were compared with mRNA levels in nonstimulated cells. The mean values of duplicate samples from cells cultured in the presence of LTD₄ were normalized as fold changes of vehicle-treated cells. The relative gene expression was calculated with the comparative ΔΔCT method as previously described.18

Immunohistochemistry

L1236, HDLM2, KMH2 and L428 cells were resuspended in PBS with 10% FCS at a final concentration of 1 × 10⁶cells/ml. The cells were cytocentrifuged and fixed in 4% paraformaldehyde for 10 min. For CysLT₁ and CysLT₂ receptor staining, all antibodies were diluted in Antibody Diluent (Dako, Glostrup, Denmark). Tris (tris (hydroxymethyl)aminomethane) buffer (0.5% Tween 20, pH 7.6) was used in all washing steps. Nonspecific antibody binding was blocked with Protein Block Serum-Free (Dako, Glostrup, Denmark). The sections were incubated in room temperature with polyclonal rabbit antihuman CysLT₁ (Lifespan Biosciences, Seattle, WA) (1:150) or CysLT₂ receptor antibody (Cayman, MI) (1:150), respectively, for 45 min. Subsequently, biotinylated goat anti-rabbit IgG (H + L), alkaline phosphatase conjugated avidin–biotin complex and Vector® Red Alkaline Phosphatase Substrate (Vector Laboratories, Burlingame, USA) were used. The expression of the CysLT₁ receptor and CD30 was investigated in formalin-fixed and paraffin-embedded 1.5-μm thick consecutive sections from cHL lymph node biopsies. Deparaffinized and rehydrated tissues were boiled in 0.01 M citrate buffer, pH 6.0, to unmask epitopes. The stainings were performed exactly as described earlier with the exception of a primary antibody dilution of 1:50. Controls included omission of primary antibody and rabbit serum. For neutralization experiments, the blocking peptide was mixed with the corresponding antibody (peptide: antibody ratio for CysLT₁ 10:1, for CysLT₂ 20:1) and incubated over night at +4°C before application. CD30 stainings were performed in an automated staining machine (Ventana Bio Tek systems, Tucson, AZ). Proteinase K was used for antigen retrieval followed by a monoclonal mouse antihuman CD30 (Clone Ber-H2) antibody (Dako, Glostrup, Denmark). For detection, Iview DAB detection kit (Ventana Medical Systems, Tucson, AZ) was used. The numbers of H-RS cells per high-power field were determined in section areas with the highest number of these cells. The percentage of H-RS cells positive for the CysLT₁ receptor was determined in the same areas. In all but 4 cases (patients 8, 12, 13, 16; see Table I), at least 100 H-RS cells were recorded. In all tumors, eosinophils were used as an internal control for the CysLT₁ receptor expression. The CysLT₁ receptor expression intensity in H-RS cells was compared with that of tumor tissue eosinophils by immunohistochemistry and classified as 0 (negative), + (weaker than eosinophils), ++ (equal to eosinophils) or +++ (stronger than eosinophils). Fibrosis were semiquantitatively assessed and arbitrarily classified using Gordon-Sweet staining as − (negative), + (slight) or ++ (advanced). The numbers of eosinophils and mast cells are given as mean values from 10 high-power fields. The frequency of epitheloid cells was semiquantitatively assessed and arbitrarily classified as low or high by morphology. The frequency of macrophages was also semiquantitatively assessed and arbitrarily classified as low or high by morphology combined with CD68 expression. The neutrophils were semiquantitatively assessed by morphology and CD15 expression. For analysis, a microscope (Olympus BX60, Tokyo, Japan) equipped with a digital camera (Sony DKC-5000, Tokyo, Japan) was used.

Microarray of transient-transfected GC B-cells and EBV-infected KMH2 cells

The EBV infection and microarray analysis of KMH2 cells have been previously described.19 Isolation and transfection of CD10⁺ GC B-cells were performed as described recently.20 RNA from the FACS-sorted transfected GC B-cells was amplified with ExpressArt mRNA Amplification kit protocol (Amptech). Ten micrograms of fragmented cRNA were hybridized to HG-U133Plus2 microarrays. Microarray chips were analyzed using the GCOS Software from Affymetrix Probe level quantile normalization and robust multiarray analysis was performed using the affy package of the Bioconductor project. The expression of CysLT₁ receptor mRNA in LMP1-transfected cells was compared with that in control vector transfected GC B-cells from 3 patients using paired t-test.

Calcium mobilization

Cells were washed once in serum-free medium and transferred to a black, poly-D-lysine-coated 96-well plate (Costar 3667, Corning, NY). Calcein3 fluorescence dye solution (Molecular Devices, CA) was used as fluorofore according to the manufacturer's protocol, and a FLEX station (Molecular Devices, CA) was used as fluorescence reader (excitation wavelength: 485 nm, emission wavelength: 525 nm, cutoff: 515 nm). Methanol and DMSO were used as leukotriene and zafirlukast controls, respectively.

Cultivation of cells in the presence of LTD₄

L1236 and KMH2 cells were pelleted, resuspended in serum-free AIM-V medium (Gibco, Paisley, Scotland, UK) (1 × 10⁶ cells/ ml) supplemented with LTD₄ (1–300 nM) (Biomol, Rungsted Kyst, Denmark) or vehicle (methanol) and cultured for 4, 11 and 25 hr, respectively. After centrifugation, supernatants were collected, and the cell pellets were washed once in PBS.

Cytokine and chemokine detection

L1236 and KMH2 cells were cultured in the presence of LTD₄ (100 nM) for 4 and 11 hr, respectively. One milliliter of supernatant was collected from each sample, and the levels of 60 different cytokines and chemokines were determined using the RayBio® Human Cytokine Antibody Array VI (RayBiotech, Inc., Norcross, GA) according to the manufacturer's instructions. Additionally, the human Cytometric Bead Array kit (Human Inflammation CBA, BD Biosciences, Stockholm, Sweden) was used for quantitative measurements of TNF-α, IL-1β, IL-6, IL-8, IL-10 and IL-12p70 in 25 μl supernatants from L1236 cells cultured in the presence of 1–300 nM LTD₄ or vehicle for 4, 11 and 25 hr, respectively. A FACSCalibur flow cytometer was used, and the results were analyzed by CellQuest Pro Software (BD Biosciences, San Jose, CA) and BD CBA software (BD Biosciences, Stockholm, Sweden). Furthermore, the quantitative sandwich enzyme immunoassay technique (human IL-13, Quantikine, R&D systems, Inc, MN) was used for IL-13 quantification in 100 μl supernatant from L1236 cells cultured in the presence of LTD₄ (100 nM) for 4, 11 and 25 hr, respectively, according to the manufacturer's instructions.

Statistical analyses

ANOVA with Fisher statistics was used for statistical analysis of the data obtained with the cytometric bead array. The pair-wise comparisons were adjusted for multiple comparisons by LSD adjustments.

Results

Expression of CysLT₁ and CysLT₂ receptor mRNA in HL cell lines

CysLT₁ and CysLT₂ receptor mRNA expression in HL cell lines was determined by real-time PCR. L1236 and KMH2 cells were shown to be strongly and weakly positive for the CysLT₁ receptor mRNA, respectively (data not shown). In contrast, L428 cells and HDLM2 cells were negative. Furthermore, L1236 cells and HDLM2 cells were strongly positive for the CysLT₂ receptor mRNA, whereas L428 cells and KMH2 cells were negative for this receptor (data not shown).

Immunohistochemical and microarray analyses of HL cell lines and cHL biopsies

To delineate the protein expression of the CysLT₁ receptor by HL-derived cell lines and H-RS cells in biopsies from cHL patients, immunohistochemical studies were performed using a polyclonal rabbit antihuman CysLT₁ receptor antibody. Cytocentrifuged, paraformaldehyde-fixed L1236 cells were analyzed for CysLT₁ receptor expression by the avidin–biotin alkaline phosphatase method. Most L1236 cells were positive for the CysLT₁ receptor (Supplemental data, Fig. S1A, red color). When the blocking peptide was mixed with the antibody prior to application, the signal was greatly reduced (Supplemental data, Fig. S1B). The controls, rabbit serum and omission of primary antibody, were negative (Supplemental data, Figs. S1C and S1D). KMH2 cells were weakly positive for the CysLT₁ receptor and the signal could be neutralized by the blocking peptide (data not shown). L428 cells and HDLM2 cells did not express the CysLT₁ receptor (data not shown). Staining with the CysLT₂ receptor antibody yielded a strong signal only in L1236 cells, which was only partly quenched in spite of preincubation with excessive amounts of the corresponding peptide (data not shown). We therefore consider the protein expression of the CysLT₂ receptor in the HL cell lines as indefinite.

In primary tumor tissue, H-RS cells positive for the CysLT₁ receptor were detected in NS (8/8) and MC (4/8) HL tumors. The LR (n = 3) and LD (n = 1) tumors under study did not express the CysLT₁ receptor (Table I). Figure 1 shows examples of HL tumors of subtype NS (a), MC (b), LR (c) and LD (d) stained with a CysLT₁ receptor antibody. When the CysLT₁ receptor antibody was preincubated with the corresponding blocking peptide, no signal was noted in NS (e), MC (f), LR (g) or LD (h) HL tumors. The omission of primary antibody or the application of rabbit serum did not yield any signal (data not shown). A very strong expression was noted in 2 tumor samples (Table I). The expression was more frequent and usually stronger in H-RS cells from NS tumors (Table I). Available tumors were assessed for the degree of fibrosis and infiltrating cells. In most cases, only few neutrophils were present, but in some cases numerous neutrophils were found. However, no association was noted between the quantitative CysLT₁ receptor status of the H-RS cells and degree of fibrosis or the numbers of infiltrating eosinophils, mast cells, epitheloid cells, macrophages or neutrophils (Supplemental data, Table SI). In addition, there was no obvious correlation between CysLT₁ receptor status and age, sex, clinical stage, B-symptoms or outcome (Table I and data not shown). To confirm these observations we used microarray data to compare the expression of CysLT₁ receptor in microdissected H-RS cells from 9 cases of NS HL with that in CD10-positive GC B-cells derived from reactive lymphoid tissue (tonsils) of 3 patients. In 5 of 9 cases CysLT₁ receptor mRNA was detectable in H-RS cells with levels higher than that observed in GC B-cells (Fig. 2a), the presumed progenitors of H-RS cells. In keeping with the immunohistochemistry data, CysLT₁ receptor expression was also noted in the microarray analysis of nonmalignant reactive cells microdissected from 1 of 3 tumors. Additionally, quantitative PCR was performed on cDNA derived from both CysLT₁ receptor mRNA-positive (nr 25, 26 and 28) and negative (24, 27 and 29) microarrayed tumors. CysLT₁ receptor mRNA was detected in tumor 25 and 26 with an arbitrary expression level of 4.12 and 0.35 compared with β-2 microglobulin mRNA expression. No CysLT₁ receptor mRNA could be detected in tumor 24, 27, 28 or 29 (data not shown). Unfortunately, none of the CysLT₂ receptor-specific antibodies that were tested (Cayman, Lifespan) worked well on the formalin-fixed paraffin-embedded HL tumors. Analysis of the microarray data showed that CysLT₂ receptor mRNA was “present” in only 1/9 of the primary H-RS cells (Fig. 2b). CysLT₂ receptor mRNA expression was also “absent” in microdissected reactive cells from 3 HL tumors.

Figure 1. Immunohistochemical staining for CysLT₁ receptor expression in cHL tumors. Paraffin sections from formalin-fixed HL tumors were stained with a polyclonal rabbit antihuman CysLT₁ receptor antibody. Shown in the upper panel are NS [(a), nr 4 in Table I], MC [(b), nr 11 in Table I], and LR [(c), nr 16 in Table I] and LD [(d), nr 19 in Table I]. Shown in the lower panel (e–h) are CysLT₁ receptor stainings after preincubation with the corresponding blocking peptide for each tumor subtype. Original magnification ×46.

Figure 2. Microarray analysis of microdissected H-RS cells. GCOS (Gene chip operating software) signal for (a) CysLT₁ and (b) CysLT₂ determined from the microarray analysis of H-RS cells microdissected from 9 primary NS HL tumors (nr 21–29 in Table I) and nonmalignant (NM) reactive infiltrate microdissected from 3 primary NS HL tumors compared with that from purified germinal centre (GC) B-cells from tonsils. CysLT₁ mRNA is called “present” in H-RS cells of 5 tumors, whereas “absent” in other samples including GC B-cells. CysLT₂ mRNA is called “present” in H-RS cells from 1 tumor, whereas “absent” in all other samples. The designations correspond to the numbers in Table I.

Microarray analysis of LMP1-transfected GC B-cells and EBV-infected KMH2 cells

To investigate whether EBV encoded proteins could contribute to the upregulation of CysLT₁ and CysLT₂ expression, LMP1 was expressed in CD10⁺ GC B-cells of 3 patients. Cells were transfected with LMP1 expression vector (pSG5-LMP1) or vector control (pSG5). LMP1 did not induce CysLT₁ expression (p value of paired t-test = 0.699). CysLT₂ was called “absent” in LMP1 expression vector or vector control transfected GC B-cells. We also compared CysLT₁ expression in EBV infected and control KMH2 cells using our previously described microarray data.19 EBV had no significant effect on CysLT₁ expression (p value of unpaired t-test = 0.21). CysLT₂ was called “absent” in EBV infected and control KMH2 cells (data not shown).

Effects of cysteinyl leukotrienes on calcium mobilization in HL cell lines

KMH2, L1236, L428 and HDLM2 cells were tested for their ability to respond to exogenously added leukotrienes by intracellular calcium release. All CysLTs (LTC₄, LTD₄ and LTE₄), but not LTB₄, were able to evoke a calcium response in KMH2 and L1236 cells, whereas L428 and HDLM2 cells did not respond to any leukotriene. To determine the potency of these ligands in KMH2 and L1236 cells, dose-response curves were performed and the concentrations for half-maximal responses (EC50) were established (Figs. 3a and 3b, respectively). In KMH2 cells, the EC50 value of LTD₄-induced intracellular calcium release was lower than those of LTC₄ and LTE₄ (Fig. 3a), whereas the EC50 values were lower for LTC₄ and LTD₄ than for LTE₄ in L1236 cells (Fig. 3b). Preincubation with 100 nM of the CysLT₁ receptor-specific antagonist zafirlukast completely abolished the LTD₄-induced calcium release in L1236 cells (Fig. 3c). The half-maximal inhibitory concentrations (IC50) of zafirlukast and another CysLT₁ receptor-specific antagonist, montelukast, of LTD₄-induced calcium release in KMH2 and L1236 cells were determined after preincubation of the cells for 30 min with different concentrations of antagonist and subsequent analysis of the calcium response to 300 nM LTD₄. Both inhibitors were able to potently block the intracellular calcium release in both KMH2 and L1236 cells (Supplemental data, Table SII).

Figure 3. Effect of leukotrienes on calcium mobilization in L1236 and KMH2 cells. Different concentrations of leukotrienes were added to (a) KMH2 and (b) L1236 cells, followed by detection of free cytosolic Ca²⁺ levels as described in Material and methods. Maximum changes in fluorescence (in relative fluorescence units, RFU) are depicted on the Y-axis for each concentration of ligand, which is depicted on the X-axis. Mean RFU ± SEM are derived from 2 separate experiments, performed in triplicate. EC50 values for each leukotriene, representing ligand concentration required to achieve half-maximal calcium response, are presented in the figure. The EC50 rank order in KMH2 cells is LTD₄ < LTC₄ = LTE₄, and in L1236 cells LTD₄ = LTC₄ < LTE₄. (c) L1236 cells were pretreated for 30 min with 100 nM zafirlukast before addition of different concentrations of LTD₄, and assayed for calcium mobilization as described in Materials and methods. Maximum changes in fluorescence (RFU) after addition of LTD₄ are plotted on the Y-axis. Each value from 1 representative experiment is shown.

Effects of LTD₄ on cytokine and chemokine mRNA expression in L1236 cells

To investigate the effects of LTD₄ on cytokine and chemokine mRNA expression, L1236 cells were cultured in the absence or presence of LTD₄ for different time periods. Total RNA was reversely transcribed and quantitative PCR was performed. As shown in Supplemental data, Table SIII, L1236 cells cultured in the presence of LTD₄ (100 nM) for 4 hr (n ≥ 3) showed increased IL-8, IL-6, TNF-α and IL-13 mRNA expression as compared to control cells. These differences could not be detected after 11 and 25 hr of culture. LTD₄ did not significantly affect the mRNA levels of IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-10, IL12p35, IL12p40, IL-15 or interferon-γ (n = 1, data not shown). Similar results were obtained with 500 nM LTD₄ (data not shown).

Effects of LTD₄ on cytokine and chemokine protein release by L1236 and KMH2 cells

To elucidate whether the secretion of cytokines and chemokines by L1236 cells was influenced by LTD₄, a cytokine antibody array was performed. L1236 cells cultured with and without LTD₄ (100 nM) for 4 and 11 hr secreted, among others, RANTES, monocyte chemoattractant protein-1, TNF-β, TNF-α and particularly high levels of IL-6. Leukotriene D₄ stimulation increased the protein levels of IL-6 and TNF-α (data not shown). Therefore, the effects of LTD₄ on cytokine protein secretion by L1236 and KMH2 cells were further quantified by flow cytometry using the Cytometric Bead Array technique. As shown in Figures 4a–4c, L1236 cells cultured in the presence of LTD₄ (100 nM), significantly increased the secretion of TNF-α (p < 0.001), IL-6 (p < 0.001) and IL-8 (p = 0.012) as compared to control cells. Shown are mean values after treatment for 4 (n = 7), 11 (n = 6) and 25 (n = 5) hr. The LTD₄-induced effect was dose-dependent and blocked by 1 μM zafirlukast (Supplemental data, Figs. S2A and S2B). Also, the secretion of IL-8 by KMH2 cells was increased by LTD₄, and this effect was blocked by 0.1 μM zafirlukast (Supplemental data, Fig. S2C). Furthermore, human IL-13 ELISA was performed to elucidate whether IL-13 protein release by L1236 cells was affected by LTD₄. However, in most analyzed supernatants, the IL-13 levels were below the limit of detection (32 pg/ml).

Figure 4. Effects of LTD₄ on cytokine protein release by L1236 cells. L1236 cells were cultured in serum-free AIM-V medium supplemented with LTD₄ or vehicle for 4 (n = 7), 11 (n = 6) and 25 (n = 5) hr, respectively. Cytometric bead array was used for quantitative measurements. Upon LTD₄ (100 nM) challenge, the protein release of (a) TNF-α (p < 0.001), (b) IL-6 (p < 0.001) and (c) IL-8 (p = 0.012) was significantly increased as compared to vehicle. ANOVA with Fisher statistics was used for statistical analyses. Values represent mean values ± standard error.

Discussion

In this study, we demonstrate the expression of functional CysLT₁ receptors in the HL cell lines L1236 and KMH2 as measured by intracellular calcium mobilization upon CysLT stimulation. The EC50 rank order of the different CysLTs (LTC₄, LTD₄ and LTE₄) obtained in KMH2 cells is typical for a CysLT₁ response,21 whereas the rank order obtained in L1236 cells could be valid also for a CysLT₂ response.22 Therefore it is tempting to speculate that the L1236 cells contain a functionally CysLT₁–CysLT₂ receptor dimer, or alternatively, that the cells also express the recently discovered CysLT receptor GPR17.23 However, the CysLT-induced calcium responses were completely abolished if the cells were pretreated with zafirlukast or montelukast at concentrations selectively inhibiting the CysLT₁ receptor, suggesting that this receptor is involved in the CysLT-mediated response. Although CysLT₂ mRNA could be detected in L1236 cells as well as in primary H-RS cells, we were not, due to technical impediments, able to satisfactorily establish whether the protein is expressed in these cells. Thus, the possible functional relevance of the CysLT₂ receptor in H-RS cells remains obscure. The HL cell lines L428 and HDLM2 did not respond to CysLTs in the calcium assay. The fact that not all tumor specimens showed CysLT₁ receptor expressing H-RS cells may relate to the observed cell line discrepancy in CysLT₁ receptor expression, as these may be derived from both CysLT₁ negative and positive tumors, respectively. Additionally, HL is a heterogeneous disease with pronounced biological and clinical variability. Among the HL cell lines, L1236 has been molecularly proven to be of H-RS cell origin.24 LTB₄, known to mediate its effects via BLT receptors25 and peroxisome proliferator-activated receptor-α,26 respectively, did not prompt any calcium signal.

Notably, CysLT₁ receptor protein expression was also detected by immunohistochemistry in primary H-RS cells in HL tumors of both MC and NS, but not LR and LD, subtypes. In support of this finding, microarray data obtained from laser-captured primary H-RS cells showed substantially higher levels of CysLT₁ receptor mRNA in H-RS cells in comparison to GC B-cells. Additionally, cDNA from the same samples were analyzed by quantitative PCR for the expression of CysLT₁ receptor mRNA. The resulting data, except for 1 tumor (nr 28), was well in line with the previously obtained microarray data. The reason for the discrepancy in this tumor is however likely to be due to the very small amount of available cDNA. There was no obvious correlation between CysLT₁ receptor expression status and clinical characteristics or outcome. However, one should bear in mind the limited number of patients under study and the general good prognosis for cHL.

As ∼30 to 40% of HL tumors are EBV genome-positive, it was of interest to examine the influence of EBV status on CysLT receptor expression. However, microarray analyses of EBV-infected KMH2 cells and LMP-1-transfected GC B-cells revealed no alteration of CysLT₁ or CysLT₂ gene expression (data not shown).

The finding that H-RS cells express functional CysLT₁ receptors together with the fact that eosinophils, macrophages and mast cells are the major sources of cysteinyl-containing leukotrienes,8 implies a potential biological role of CysLTs in cHL. The in vitro data presented herein support such a role. Because cHL is a tumor with specific cytokine features,6 we hypothesized that culturing of the HL cell lines L1236 and KMH2 cells in the presence of LTD₄ may affect the cytokine and chemokine production of these cells. Importantly, we demonstrated significant increase of TNF-α, IL-6 and IL-8 protein release by L1236 cells when cultured in the presence of LTD₄. In contrast to IL-6 and IL-8, the increased secretion of TNF-α protein occurred with a relatively low transcriptional induction (Table SII). These results indicate that the elevated TNF-α protein release was primarily due to post-transcriptional regulation (i.e. increased translation or elevated protein-stability) or increased release of preformed proteins, rather than de novo synthesis. Also, in alveolar macrophages, LTD₄ has been shown to increase the release of TNF-α after LPS-stimulation.27 KMH2 cells also showed a CysLT₁-induced increase in IL-8 secretion upon LTD₄ stimulation, although this increase was less pronounced than in L1236 cells. Regarding IL-13, LTD₄ stimulation of L1236 cells induced gene transcription but not detectable protein release. However, IL-13 has by neutralization experiments been reported to exert important growth stimulatory and antiapoptotic effects in this cell line at concentrations <12 pg/ml,28 which is below the detection level of the ELISA used in this study. Neutralization of IL-13 in our experimental setting would hypothetically lead to downregulated CysLT₁ receptor expression, thus making this a less-attractive approach. We can therefore not rule out the possibility that LTD₄ induces increased IL-13 release at still subdetectable but biological significant levels.

Although an issue that has not been addressed in this work, one may hypothesize that the transcription of the CysLT₁ receptor gene in H-RS cells is related to the specific cytokine profile of these cells. In human monocytes and macrophages, CysLT₁ receptor expression has been shown to be induced at the transcriptional level by IL-4 and IL-13,29 through activation of the transcription factor STAT6 and its subsequent binding to a response element in the putative proximal CysLT₁ receptor promoter.30 As a consequence of autocrine IL-13 stimulation, H-RS cells exhibit in vivo and in vitro high levels of phosphorylated (activated) STAT6 with a nuclear localization,31 thus providing a basis for efficient transcription of the CysLT₁ receptor gene. However, since not all HL cell lines expressed the functional CysLT₁ receptor, other mechanisms (e.g. epigenetic) probably contribute to the control of this gene. The fact that the CysLT₁ receptor is expressed only in a subset of HL cells may relate to the local variation of IL-13 levels and variation in the IL-13 receptor expression. Similar observations have been reported, and it has been shown that IL-13 was expressed in less than 20% of H-RS cells in 4/36 primary HL tumors and in 20–50% of H-RS cells in 17/36 tumors.32 The biological implications of these findings, i.e. increased release of TNF-α, IL-6 and IL-8 in cultured H-RS cells upon LTD₄ stimulation, are not clear. In fact, little is known regarding the role of these potent mediators in the pathogenesis of cHL, mainly due to the lack of an appropriate in vitro or animal model. Although IL-6 and TNF-α as well as their cognate receptors have been shown to be expressed to various extents by primary and cultured H-RS cells, in vitro analyses do not support these cytokines to be autocrine growth factor in cHL.33 IL-8 mRNA was detected in primary H-RS cells in 3/33 tumors by in situ hybridization in 1 study,34 but the functional relevance of this cytokine in the setting of cHL has, to our knowledge, not been addressed. However, cytokine-specific primary and secondary effects on receptor-expressing bystander cells would be expected. Release of IL-8 by the H-RS cells may induce chemotaxis of eosinophils into the tumor site. TNF-α is an important mediator of inflammation which, among many other effects, also promotes the synthesis of other proinflammatory compounds by various cell types, such as IL-1 and IL-6 by macrophages.33 Furthermore, TNF-α-induced release of eotaxin by tumor-associated fibroblasts will promote the chemotaxis of eosinophils.35 These cells not only contribute to the inflammatory cHL environment but also constitute a source for additional LTD₄, thus forming the basis for a positive feedback loop. A hypothetical model of CysLT₁ receptor signaling is schematically presented in Fig. 5.

Figure 5. A hypothetical model of cysLT signaling in cHL. The eosinophils, mast cells and macrophages, surrounding the H-RS cells, synthesize LTC₄ that is released extracellularly. Leukotriene C₄ is further metabolized to LTD₄ by glutamyl transpeptidase. Leukotriene D₄ and LTC₄ will bind to the CysLT₁ receptors expressed by the H-RS cells, thereby promoting release of TNF-α, IL-6 and IL-8 by these cells. In an autocrine manner, IL-13 may increase the transcription of the CysLT₁ receptor gene via STAT6 phosphorylation. Speculative biological effects may include increased release of TNF-α and IL-8 that contribute to chemoattraction of inflammatory cells, and TNF-α-induced eotaxin released by fibroblasts contributing to the eosinophil infiltration. H-RS, Hodgkin Reed-Sternberg cell; Eos, eosinophil; CD4⁺, CD4⁺ T-lymphocyte; Mø, macrophage; Neu, neutrophil; solid arch, CysLT₁ receptor.

The discrepancy between L1236 and KMH2 cells in terms of cytokine/chemokine release may be due to the fact that the LTD₄-induced calcium signal was significantly higher in L1236 cells compared with KMH2 cells.

Based on the results from 2 HL cell lines and primary H-RS cells, this study supports a role of CysLTs in the pathogenesis of cHL by contributing to increased cytokine release by the tumor cells. Further studies will reveal whether this finding may have diagnostic or clinical implications.

Acknowledgements

We thank Mrs. Margareta Andersson, Mrs. Margareta Söderqvist and Mrs. Margareta Waern for excellent technical assistance. We also thank Mr. Bo Nilsson for statistical advice and analyses. K.R.N.B. was supported by the Leukaemia Research Fund, UK.