Common and differential chemokine expression patterns in rs cells of NLP, EBV positive and negative classical hodgkin lymphomas



Hodgkin lymphoma (HL) is characterized by a minority of neoplastic cells, the so-called Reed-Sternberg (RS) cells and a vast majority of reactive cells. RS cells produce chemokines that can attract subsets of peripheral blood cells into HL tissues. To gain insight in the chemokines involved in HL, 16 chemokines were selected based on their ability to recruit different subsets of cells. Five HL, 5 non-HL-derived cell lines, 22 HL, 5 non-HL and 3 control tissues were analyzed by reverse transcriptase-polymerase chain reaction (RT-PCR). Products for 13 of these 16 chemokines were detected in 1 or more of the cell lines tested. No or only very faint signals were obtained in HL for CXCL12, CCL7 and CCL8, but CXCL10, CCL5, CCL13, CCL17 and CCL22 were highly or differentially expressed in HL cell lines and tissues. Immunohistochemistry was performed with antibodies reactive with the latter 5 chemokines on paraffin sections of 21 cases of HL. CCL17 and CCL22 had the highest signals in RS cells at gene expression and at protein levels. CCL17 was specific for the classic HL subtypes, whereas CCL22 also had low signals in NLP samples, as well as in some non-HL. CXCL10 was expressed in a large proportion of HL cases with a predominant expression in EBV-positive cases. The results indicate that RS cells produce a complex pattern of chemokines that are involved in the recruitment of reactive cells and contribute to the paradox of an extensive but ineffective host immune response. © 2002 Wiley-Liss, Inc.

Hodgkin lymphoma (HL) is characterized by a minority of neoplastic cells, the so-called Reed-Sternberg (RS) cells, that have clonal immunoglobulin gene rearrangements1, 2 and a vast majority of infiltrating cells. These reactive elements consist of T lymphocytes, eosinophils, histiocytes and plasma cells. It has been hypothesized that the presence of these cells in the involved tissues results from the production of specific chemokines. Chemokines (cytokines with chemoattractant properties) are a growing group of about 40 small molecules (8–14 kDa) that play a major role in leukocyte trafficking.3–5 In HL, the T cells that directly surround the RS cells have immunophenotypic features of TH2 cells, e.g., CD4+, CD45RO+ and CD45RBdim and have the capacity to produce TH2-type cytokine IL-4 in vitro.6, 7 The presence of these TH2-like cells may be explained by the high production of CCL17 (TARC) by the RS cells,8–10 which specifically attracts CC receptor 4 (CCR4, the receptor for CCL17)11 positive TH2 cells.

Besides CCL17, other chemokines have also been demonstrated in Reed-Sternberg cells.12–15 These chemokines may account for the presence of other subsets of infiltrating cells in tissues involved with HL. The first publications on chemokine expression in HL reported presence of CXCL8 (IL-8) in HL cell lines after phorbol ester treatment16–18 and in the sera of HL patients.19, 20 However, in HL tissues, CXCL8 expression was found predominantly in the infiltrating cells and not, or only rarely, in RS cells.9, 21, 22 More recent gene and protein expression studies13–15 revealed CXCL9 (Mig), CXCL10 (IP-10), CCL3 (MIP-1α), CCL5 (RANTES), CCL11 (Eotaxin) and CCL 22 (MDC) in HL tissues. One of these studies13 demonstrated a correlation between the presence of Epstein-Barr Virus (EBV) and a high expression of CXCL9, CXCL10 and CCL11 mRNA (messenger RNA) in Hodgkin tissues as well as the presence of CXCL 9, CXCL 10 in RS cells by immunohistochemistry. Tedla et al.12 reported the presence of CCL5 mRNA and protein in RS cells. Jundt et al.23 demonstrated that the high levels of CCL11 in HL tissues are probably secondary to high expression of tumor necrosis factor alpha (TNFα by the RS cells that can stimulate cocultured fibroblasts to produce this chemokine.

To gain further insight in the chemokines involved in the recruitment of infiltrating cells in HL tissues, we analyzed the expression of 16 chemokines. These include CXCL8 (IL-8: interleukin 8), CXCL9 (Mig: monokine induced by interferon-γ), CXCL10 (IP-10: interferon-γ inducible protein-10), CXCL12 (SDF-1α: stromal derived factor 1α), CXCL13 (BLC: B-lymphocyte chemoattractant), CCL1 (I309), CCL5 (RANTES: regulated on activation, normal T cell expressed and secreted), CCL7 (MCP-3: monocyte chemoattractant protein 3), CCL8 (MCP-2: monocyte chemoattractant protein 2), CCL11 (Eotaxin), CCL13 (MCP-4: monocyte chemoattractant protein 4), CCL17 (TARC: thymus and activation related chemokine), CCL19 (ELC: EBV-induced gene 1 ligand chemokine), CCL20 (MIP3α: macrophage inflammatory protein 3α), CCL21 (SLC: secondary lymphoid tissue chemokine) and CCL22 (MDC: macrophage-derived chemokine). These chemokines were selected based on their abilities to attract different subsets of cells. We analyzed 5 HL and 5 non-HL lymphoma-derived cell lines and 30 selected frozen tissues involved by HL (22), non-HL (5) as well as 1 reactive lymph node and 2 tonsils. Chemokines with a high or differential expression in HL cell lines and/or tissues were selected for an immunohistochemical analysis in paraffin-embedded tissue sections involved by HL.


Cell lines

The L428, L540, L591 and L1236 Hodgkin lymphoma-derived cell lines24, 25 were made available to us by Dr. Volker Diehl and coworkers (Cologne, Germany). DEV was originally published as derived from a case of NS Hodgkin lymphoma,26 but this case was subsequently retyped as nodular lymphocyte predominance HL. The RAY and POP lymphoblastoid cell lines were initiated by transforming peripheral blood B cells with EBV. Large B-cell non-Hodgkin lymphoma cell lines, VER and Rose were established in our laboratory (unpublished data). VER was derived from a large cell B-cell lymphoma with t(8;14) secondary to HL. Rose was derived from a large cell B-cell lymphoma with t(14;18) derived from a transformed follicular lymphoma. Anaplastic large-cell lymphoma cell line KARPAS 299,27 leukemic T-cell line JURKAT and RAJI, a Burkitt's lymphoma cell line, were obtained from American Type Culture Collection (Rockville, MD).


Frozen tissue specimens and cell suspensions of lymph nodes involved by Hodgkin and non-Hodgkin lymphomas were retrieved from the tissue bank of the Department of Pathology and Laboratory Medicine, University Hospital Groningen. Twelve cases of nodular sclerosis (NS) [6 EBV-positive (cases 1–6) and 6 EBV-negative (cases 7–12)], 4 cases of mixed cellularity (MC) [2 EBV-positive (cases 13 and 14) and 2 EBV-negative (cases 15 and 16)] and 6 cases of nodular lymphocyte predominance (NLP) HL (cases 17–22) were used for RT-PCR. Two cases of ALK + (anaplastic lymphoma kinase) anaplastic large-cell lymphoma T/0 (ALCL) and 3 cases of diffuse large B-cell lymphoma (DLCL) with anaplastic morphology and CD30 positivity were used for comparison. All cases were classified according to the World Health Organization classification of neoplastic diseases of the hematopoietic and lymphoid tissues.28 The tissue selection was based on the presence of good-quality RNA in the frozen tissue specimens. As control tissues, we included 1 reactive lymph node and 2 tonsils.

Paraffin-embedded tissues from 9 NS HL cases [4 EBV-positive (cases 1, 2, 4 and 5) and 5 EBV-negative (cases 7, 9, 11, 12 and 23)], 5 MC HL cases [4 EBV-positive (cases 24–27) and 1 EBV-negative (case 15)], 7 NLP HL (cases 19, 20, 21, 28, 29, 30 and 31) and 2 tonsils were submitted to immunohistochemical analysis. These cases were retrieved from the tissue bank of the Department of Pathology and Laboratory Medicine, University Hospital Groningen, The Netherlands, except for 4 MC cases, which were retrieved from the Department of Medical Pathology, Federal University of Paraná, Brazil.

RT-PCR on HL cell lines and tissues

Total RNA was isolated with Trizol (Gibco BRL, Life Technologies, Paisley, UK) from cell lines and from frozen tissue sections. After isolation of total RNA, a DnaseI (Roche, Mannheim, Germany) treatment was performed to purify the samples from contaminating DNA. Quality of the isolated total RNA was checked on a 1% agarose gel. Only those cases, in which the RNA quality was good and the DNAse treatment was successful (yielding no results), as tested by PCR using a selected genomic primer set, were used for a further analysis. For the cDNA synthesis, we used 3–4 μg of total RNA. The reaction was primed with oligo(dT) in a volume of 40 μl using the protocol provided by the manufacturer (Gibco BRL, Life Technologies). In case genomic sequences were available, primers were selected from different exons to distinguish between RNA and contaminating DNA PCR products. Primer sequences used for the amplification were CXCL8 (f 5′ tgggtgcagagggttgtgg 3′; r 5′ tgggtgcagagggttgtgg 3′), CXCL12 (f 5′ acctcgctggactctcagtc 3′; r 5′ aaacacattatctgggagatgc 3′), CXCL13 (f 5′ aaacacattatctgggagatgc 3′; r 5′ gtgaaagaagcttgagtttg 3′), CCL1 (f 5′ ggaaaccacatggcttcacc 3′; r 5′ caagcagatcctctgtgacc 3′), CCL7 (f 5′ gctggagagctacagaagg 3′; r 5′ ttatacaatacccccatgagg 3′), CCL8 (f 5′ aattcctatccagaggctg 3′; r 5′ cttacaggagcactgattgcc 3′), CCL13 (f 5′ caccaccagcaggtgtccc 3′; r 5′ gggtcagcacagatctcct 3′), CCL17 (f 5′ acctgcacacagagactcc3′; 5′ atctgggccctttgtgccc 3′), CCL19 (f 5′ acagatcctgcacacaccc 3′; r 5′ ggagcagtttactctgac 3′); CCL20 (f 5′ tgtaccaagagtttgctcc 3′; r 5′ ttggacaagtccaggtgagg 3′); CCL21 (f 5′ ccagagagaccgaggaggg 3′; r 5′ gaggtggggtgtactgggg 3′) and GADPH (f 5′ ccatcactgccactcagaagact 3′, r 5′ ttactccttggaggccatgtagg 3′). Primer sequences for CXCL9, CXCL10, CCL5, CCL11 and CCL22 were published elsewhere.13, 14 CXCL12, CCL1 and CCL7 primers yield same-sized PCR products on both mRNA and genomic DNA. Positive and negative controls as well as placental genomic DNA were included in each PCR analysis. PCR for all primer sets was performed on 1 μl of the cDNA synthesis reaction (approximately 100 ng) mix using 1 unit of Taq-polymerase (Amersham Pharmacia Biotech, Buckinghamshire, UK) and the reaction buffer provided by the manufacturer. A semiquantitative PCR was performed for a range of cycles starting from 20 up to 40 cycles. The products were analyzed and demonstrated at the lowest number of cycles that gave products in the majority of primer sets, which was on average 30 cycles. The PCR program consisted of a denaturation step of 30 sec at 94°C, annealing step of 45 sec at 57°C and an extension step of 45 sec at 72°C. The first denaturation step lasted for 5 min and the final extension step lasted for 7 min. PCR products were analyzed on 2% agarose gels containing ethidium-bromide. The intensity of signals was compared to the glyceraldehyde phosphate dehydrogenase (GAPDH) intensity, and was reported to be high when this was equal to or superior to GAPDH, and low if the intensity was lower than the GAPDH intensity.

Epstein-Barr virus encoded RNA (EBER) in situ hybridization

In situ hybridization (ISH) was performed to detect EBER1 and 2 by using a fluorescein-conjugated EBV (EBER) peptide nucleic acid probe (DAKO, Glostrup, Denmark) on paraffin or frozen tissue sections. Alkaline phosphatase-conjugated anti-FITC sheep IgG Fab fragments (Roche) were used as a second detection step. The visualization of the reaction was performed with 5-bromo-4-chloro-3-indolyl-phosphatase and 4-nitrobluetetrazolium (Roche) and MgCl2.


All primary antibodies were applied for 1 hr at room temperature on dewaxed sections employing a heat-induced antigen-retrieval procedure involving microwave incubation at maximum power (750 W) for 8 min. Briefly, tissue sections were pretreated with 2 different buffer solutions depending on the subsequent antibody applied. For CCL13 (MCP-4), CCL17 (TARC) (both diluted at 1:100, goat IgG; R&D Systems, Minneapolis, MN) and CCL22 (MDC) (dilution 1:200, polyclonal rabbit; Pepro Tech, Rocky Hill, NJ) a 10 mM citrate buffer pH 6.0 was used. A TRIS (hydroxymethyl-aminomethane)/EDTA (ethylene-diamine-tetraacetic acid) buffer in a concentration of 50 mM TRIS and 2 mM EDTA pH 9.0 was used for CXCL10 (IP-10) (dilution 1:500, polyclonal rabbit; Pepro Tech) and CCL5 (RANTES) (dilution 1:500, goat, polyclonal IgG, Santa Cruz Biotechnologies, Santa Cruz, CA). All antibody dilutions were prepared with phosphate-buffered saline (PBS), pH 7.4, containing 1% bovine serum albumin (BSA). The second step for CCL5, CCL13 and CCL17 was performed with horseradish peroxidase (HRP)-labeled rabbit-anti-goat Ig antibody (DAKO, Copenhagen, Denmark) followed by a peroxidase-labeled goat-anti-rabbit Ig antibody step (DAKO). For CXCL10 and CCL22, the second step was performed with peroxidase-labeled goat-anti-rabbit Ig antibody (DAKO) and the third step with an HRP-labeled rabbit-anti-goat Ig antibody (DAKO). Peroxidase enzyme staining was performed with diaminobenzidine (DAB) and H2O2, resulting in a dark brown reaction product in positive cells. Sections were counterstained with haematoxylin before dehydration and mounted in nonaqueous mounting medium.


RT-PCR analyses of HL- and non-HL-derived cell lines revealed PCR products for 13 of 16 of the chemokines in 1 or more of the cell lines tested. No or only very faint PCR products were obtained for 3 of them, e.g., CXCL8, CCL7 and CCL8. Analysis of CXCL9 and CCL21 did not reveal a Hodgkin-specific expression pattern. For CXCL9 only very faint PCR products were obtained in all cell lines with the exception of L540. Expression of CCL21 was found most abundantly in the non-HL cell lines. For the remaining 11 chemokines, expression was found in (part of) the HL cell lines and variable results were obtained for the non-HL-derived cell lines. A faint PCR product was obtained for CXCL12 in L540, and even fainter PCR products were obtained in L1236 and Karpas. PCR analysis of CCL11 revealed also very faint PCR products in 4 of 5 HL-derived cell lines and in the T/0 cell ALCL cell line KARPAS299, whereas none of the other cell lines revealed a PCR product. A more pronounced expression was seen for the remaining 9 chemokines, which are shown in Figure 1. Expression of CXCL10, CCL1, CCL5 and CCL17 was found predominantly in the HL-derived cell lines. Expression of CCL19 appeared to be specific for L591, an EBV-positive HL cell line. RT-PCR for CCL20 also revealed the most prominent PCR product in L591, but a lower signal was also obtained for L428 and Karpas. Expression of CCL13 was only detected in the mixed cellularity HL-derived cell line L1236. CXCL13 was expressed only in KARPAS. In contrast to these results, expression of CCL22 could be detected in all B-cell lymphomas and in normal B-lymphocytes, but not in the T-cell-derived cell lines.

Figure 1.

RT-PCR (30 cycles) chemokine expression analysis of HL, non-HL lymphoma and 2 EBV-transformed B-lymphoblastoid cell lines. The figure shows an inverted image obtained from agarose gels containing ethidium-bromide. The columns indicate the different cell lines tested. At right, chemokines and the housekeeping gene (GAPDH) are indicated.

A further expression analysis on Hodgkin lymphoma, non-Hodgkin lymphoma and control tissue samples was performed to determine whether or not part of these chemokines might be specific for HL or for specific subtypes of HL. It was previously demonstrated that expression of CCL17 was restricted to the classical subtypes of HL.8–10 No or only very faint signals were obtained for 3 of 16 chemokines tested, i.e., CXCL 12, CCL7 and CCL8 (results not shown). CXCL10 showed signals in control tissues, in 2 of 3 diffuse large B-cell lymphoma with CD 30+ positive cell samples, 3 of 6 NLP-HL, 3 of 4 MC HL and 9 of 12 NS HL. In NS cases, the signals were obtained mainly in EBV-positive cases (6 of 6). Expression of CCL11 was found in several samples, without a specific pattern. Results obtained for CCL20 were similar to signals found in tonsil tissue and no (DLCL and ALCL) or low signals were present in the other samples. CCL5 expression was found in control tissues, in 3 of 4 MC HL, in 8 of 12 NS HL and in 4 of 6 NLP HL tissue samples, but not in non-Hodgkin lymphoma samples. CCL17 expression was highly related to the classic HL samples with no or only low signals in 2 of 6 NLP HL and control tissues. CCL22 showed very high signals in 15 of 16 classic HL cases and in control tissue samples, lower signals in 2 of 3 diffuse large B-cell lymphomas and weak bands in 4 of 6 NLP cases. Parts of these results are demonstrated in Figure 2.

Figure 2.

RT-PCR (30 cycles) chemokine expression analysis on HL and control-tissue samples. The figure shows an inverted image obtained from agarose gels containing ethidium-bromide. The upper portion of the figure indicates the subtype, EBV status and number of HL samples tested, and at right, the chemokine and the housekeeping gene (GAPDH) are indicated.

In addition to the mRNA expression studies, an immunohistochemic analysis was performed on paraffin-embedded samples for chemokines CXCL10, CCL5, CCL13, CCL17 and CCL22, which have differential or high expression in cell lines and tissues. Therefore, we could also assess which cells were probably responsible for the measured chemokine production and, moreover, to detect whether RS cells produce these molecules. All HL specimens were also screened for the presence of EBV by ISH for EBER 1 and 2. The results are summarized in Table I.

Table I. Immunohistochemical analysis for selected chemokines: CXCL-10 (IP-10), CCL5 (RANTES), CCL13 (MCP-4), CCL17 (TARC) and CCL22 (MDC)1
CaseSubtypeEBV StatusCXCL10 (IP-10)CCL5 (RANTES)CCL13 (MCP-4)CCL17 (TARC)CCL22 (MDC)
  • 1

    The scoring system is based on analysis of the RS cells.–; + Positive; −; Negative. Chemokine abbreviations are provided in the text.


CXCL10 was present in lymphocytes, macrophages and endothelial cells in tonsils and HL tissues. RS and Hodgkin cells stained stronger than the infiltrating cells. Overall CXCL10-positive RS cells were more frequently found in EBV-positive cases (7 of 8) and in NLP cases (6 of 7) compared to EBV-negative classic HL (3 of 6) cases. CCL5 staining was also present in macrophages, T lymphocytes and endothelial cells in tonsils and HL samples. Considerable staining variation was observed within and between cases. In 7 of 9 NS, 3 of 5 MC and 5 of 7 NLP cases, the RS cells stained for CCL5. CCL13 was found in macrophages, endothelial and some dendritic cells in tonsils and HL samples. Weakly positive RS cells were observed in 3 of 9 NS cases, 3 of 5 MC cases and in 5 of 7 NLP cases. Immunostaining for CCL17 demonstrated strongly positive RS cells, with a paranuclear (Golgi-like) staining pattern in 13 of 14 classic HL cases, whereas no staining was present in the 7 NLP samples tested. Only sporadic CCL17-positive dendritic cells were identified. CCL22 expression was detected in macrophages, B cells and dendritic cells in tonsils and tumor samples. In classical HL cases, cytoplasmic staining was observed in RS cells in all the samples tested regardless of subtype or EBV association. In L&H (lymphocytic and histiocytic)-type RS cells of NLP, CCL22 could also be noticed but was much weaker than in classical HL cases. Representative immunostaining examples for these chemokines are demonstrated in Figure 3.

Figure 3.

Immunohistochemic analysis of CXCL10 (IP-10), CCL5 (RANTES), CCL13 (MCP-4), CCL17 (TARC) and CCL22 (MDC) in HL tissues. Paraffin-embedded tissue sections were stained with various antibodies. Classical and nodular lymphocyte predominance HL are illustrated. The chemokines tested are marked within the figures. Scale bars = 40 μm.


Our RT-PCR results demonstrate a high expression of several chemokines in HL cell lines and tissues. Expression of CCL22 (MDC) frequently coincides with expression of CXCL10 (IP-10), CCL17 (TARC) and to a lesser extent also with CXCL9 (MIG) and CCL13 (MCP-4). This suggests that in part of the tissue samples analyzed there is an overall activation of chemokine expression. This may be a result of a high expression of NF-kB (nuclear factor kappa B) transcription factor in RS cells,29, 30 which is known to enhance expression of a variety of chemokines because promoters of chemokine genes contain kB binding sites.31, 32 In EBV-related cases, latent membrane protein 1 (LMP1) can stimulate transcription of NF-kB33, 34 and thus enhance expression of chemokines. A general overview from tested chemokines, their cellular sources, main receptors and target cells is provided in Table II.

Table II. Overview of tested chemokines, main cellular sources, receptors, target cells and analysis in HL cell lines and tissues1
ChemokineSynonymMain cellular sourcesReceptorsMain target cellsExpression analysis in HL
  • 1

    The most commonly used names and the main ligands, the cell sources and targets are summarized. Note that some chemokines can also bind to other receptors, can be produced by other cell types and chemoattract different subsets of cells. Although some chemokines share the same receptors, the affinity of the receptor-chemokine interaction will determine which subsets will be involved. References are quoted in the text. Chemokine abbreviations are provided in the text. NDP, no differential pattern; NE, no expression; NLP, nodular lymphocyte predominance; HDE, high differential expression; HE, high expression; LE, low expression.

CXCL8IL-8Monocytes/ macrophages and neutrophilsCXCR 1, 2NeutrophilsNE in HL cell lines and NE or WE in HL tissues
CXCL9MIGMonocytes and macrophagesCXCR 3Stimulated T and NK cellsLE in HL cell lines and no differential pattern in HL tissues
CXCL10IP-10Macrophages, endothelial cellsCXCR 3Stimulated T and NK cellsHE in HL cell lines and in EBV + HL and NLP tissues; IHC positive RS cells
CXCL12SDF-1αStromal cellsCXCR 4Naive and memory T cellsNE in HL cell lines and tissues
CXCL13BLCFollicular stromal cellsCXCR 5Follicular B helper T cells and naive B cellsLE in HL cell lines and NDP in HL tissues
CCL11309Th1 cellsCCR 8Monocytes, TH2 and NK cellsHE in HL and in majority of HL tissues NDP
CCL5RANTEST lymphocytesCCR 1, 3, 5Monocytes, T lymphocytes and eosinophilsHE in HL cell lines and tissues; IHC positive RS cells
CCL7MCP-3SeveralCCR 1, 2, 3Monocytes and effector T cellsNE in HL cell lines and tissues
CCL8MCP-2SeveralCCR 3Monocytes and effector T cellsNE in HL cell lines and tissues
CCL11EOTAXINFibroblastsCCR 3Eosinophils and TH2 cellsNE or LE in HL cell lines; WE in HL tissues with NDP
CCL13MCP-4Monocytes and epithelial cellsCCR 2, 3EosinophilsHE in 1 HL cell line; LE in HL tissues; IHC weakly positive RS cells
CCL17TARCThymic cellsCCR 4TH2 T cellsHDE in classic HL cell lines and tissues; IHC highly positive RS cells only in classic HL
CCL19ELCEndothelial and stromal cellsCCR 7Naive, central memory T cells, and B cellsLE in 1 HL cell line and NDP in HL tissues
CCL20MIP-3αThymic and epithelial cellsCCR 6Immature dendritic and memory T cellsHE in 2 HL cell lines and LE in HL tissues
CCL21SLCHigh endothelial venules in T zones of lymph nodesCCR 7Naive T and dendritic cellsLE in HL and non-HL cell lines HE with NDP in HL tissues
CCL22MDCMacrophages, dendritic cellsCCR 4TH2 T, NK and dendritic cellsHE in B and classical HL cell lines tissues; LE in NLP cases; HE in RS cells in classic HL and LE in RS in NLP by IHC.

No expression of CXCL8 (IL-8) was observed in HL-derived cell lines and only very low signals were obtained in few cases of the tissue samples. This indicates that expression of CXCL8 is not very prominent in HL, although previous studies reported expression of CXCL8 in HL tissues.21, 22 RNA ISH indicated that the production of CXCL8 is most abundant in macrophages and neutrophils, whereas it is not or only rarely demonstrated in the RS cells.9, 21

CXCL9 (MIG) and CXCL10 (IP-10) are members of the CXC chemokine family. In vitro studies have shown that CXCL9 and CXCL10 are chemoattractants for stimulated T and NK cells.35 A positive correlation has been reported for EBV-associated lymphoproliferative disorders and a high expression of CXCL9 and CXCL10.36 In classical HL, a correlation has been found between the presence of EBV and a high expression of CXCL10 and MIP1α(CCL3).13, 14 Our RT-PCR analysis revealed only very low expression for CXCL9 in all cell lines and a homogeneous expression in all the HL samples examined. CXCL10 was found in 4 of 5 HL-derived cell lines tested and no expression was found in non-HL-derived cell lines. Analysis of Hodgkin tissue samples revealed high signals of CXCL10 in 3 of 6 NLP-HL, 3 of 4 MC HL and 9 of 12 NS HL (mainly EBV-related samples). In DLCL tested samples, high signals could also be obtained. Protein localization studies confirmed these results indicating high signals in RS cells and weaker stains in endothelial, lymphocytes and macrophages. These findings corroborate the correlation detected by Teruya-Feldstein et al.13, 36 High expression of CXCL-10, an IFNγ-induced chemokine, can be expected since IFNγ is highly expressed in HL.37

CXCL13 (BLC) is constitutively produced by stromal cells and highly expressed in the follicles of Peyer's patches, spleen and lymph nodes. CXCL13 attracts naive B cells, activated and memory T cells in vitro.38, 39 CXCR-5 (formerly Burkitt's lymphoma receptor 1) has been found to be required for this B-cell migration. CXCL13 was predominantly expressed in the KARPAS299 cell line. We found high signals in almost all tissue samples tested, but no clear-cut correlation could be established since the whole tissue analysis included the lymph node stromal cells.

CCL1 (I309) was mainly observed in HL cell lines and in the majority of HL tissue samples without a correlation between expression and subtype or EBV status. No signals could be obtained in the non-HL tissues. CCL1 is produced mainly by TH1 polarized cells and attracts monocytes and activated NK cells,40, 41 which harbor the CCR8 receptor.

CCL5 (RANTES) is a chemoattractant for monocytes, T lymphocytes, eosinophils and basophils42, 43 and binds to CCR1, CCR3 and CCR5.3 Expression of CCL5 was found in 4 HL cell lines derived from classical type HL. Analysis of HL involved tissues revealed expression of CCL5 mRNA in 3 of 4 MC HL, in 8 of 12 NS HL and 4 of 6 NLP tissues. Comparatively, higher expression was obtained in EBV-related than in EBV-unrelated samples, but the reactive lymph node and tonsils also revealed high RT-PCR products for CCL5. In the non-HL samples tested, no signals were present. Protein expression was observed in RS cells as well as many reactive cells in the majority of samples, demonstrating that CCL5 is possibly an important contributor to the recruitment of reactive cells in HL. These results are in agreement with studies that found high CCL5 expression in HL tissue samples.12–14 Buri et al.44 observed a high expression of CCL5 in reactive cells but not in RS cells sorted from HL samples.

CCL11 (Eotaxin) attracts CCR3-positive eosinophils and, in allergic responses, TH2 cells.45 Jundt et al.23 showed that the high levels of CCL11 expression in HL tissue are a secondary effect of TNF-α stimulation in fibroblasts. They did not find signals in RS cells by ISH nor immunohistochemistry but found CCL11 expression in fibroblasts or macrophages. We also found no or very low CCL11 signals in the HL cell lines. RT-PCR revealed only low signals in HL tissues, and no correlation between subtype and EBV status was present. Our data and those of Jundt et al.23 are in contradiction with those of Teruya-Feldstein et al.13 that indicated a higher expression in the RS cells of NS cases.

CCL13 (MCP-4) has been reported to be chemotactic for eosinophils, monocytes and T lymphocytes.46 Our analysis revealed expression of CCL13 in the L1236, which is derived from a case of MC HL. Analysis of lymphoma tissues showed a high expression of CCL13 mRNA in 14 of 16 classic HL and in 2 of 3 B-cell large-cell lymphomas. Low CCL13 protein expression was also observed in RS cell cytoplasm in 6 of 14 classical HL tissues including both MC and NS cases and in all NLP samples.

CCL 17 (TARC) RT-PCR and protein expression studies in cell lines and tissues confirmed previous results.8–10 The CCL17 RT-PCR expression in all 5 HL cell lines and 14 of the 16 HL classical subtype tissues was confirmed by immunohistochemistry in 13 of 14 cases of classical subtype. Our analysis also showed low expression by RT-PCR in 3 of 6 NLP, but at the protein level, CCL17 was present in none of the 7 samples. No expression was found in non-HL samples. The results indicate that CCL17 plays an important role in recruiting TH2 cells in the classical subtypes but not in recruiting the CD57-positive TH2 cells of NLP HL.

CCL19 (ELC) is produced by endothelial or stromal cells in T- or B-cell areas and regulates the encounters between CCR7-positive dendritic cells, T and B cells.47, 48 An RT-PCR signal was obtained only in L591, an NS HL-derived EBV-positive cell line. Since the analyzed whole-tissue specimens contain large numbers of endothelial and stromal cells, it was not unexpected that high signals were obtained in all lymphoma HL tissues. Although complex chemokine networks are involved in HL-reactive background formation, it is unlikely that this chemokine plays a crucial role in HL.

Expression of CCL20 (MIP 3α) was found in only 2 of the NS HL-derived cell lines, L428 and L591. Analysis of lymphoma and control-tissue samples revealed CCL20 expression in tonsils and HL and non-HL samples. The expression in the 2 Hodgkin-derived cell lines could be an artifact caused by the in vitro culture. CCL20 attracts immature dendritic cells and memory T lymphocytes, which have the CCR6 receptor.49 Based on the levels of CCL20 mRNA in the HL tissues and the absence of its expression in part of the HL cell lines, it is not likely that CCL20 is critically involved in the characteristic infiltrate of HL. However, chemokines may have a high expression in RS cells, but since these cells represent only a minority of the total cell population, it is possible to measure only a low chemokine expression in tissue.

CCL 21 (SLC) directs the migration of CCR7-positive naive T cells and dendritic cells to lymph nodes.39, 50 It is produced by high endothelial venules in the T-cell zones of lymph nodes.39, 50 An important role of CCL21 in HL could not be defined in our study since low expression was obtained in all cell lines.

CCL22 (MDC) is produced by macrophages, dendritic cells and at least in vitro, in activated B lymphocytes.51–53 It binds the CC chemokine receptor 4 (CCR4)54 and attracts dendritic cells, IL-2-activated NK cells and TH2 cells with CCR4 expression.53In vitro studies demonstrated that IL-4 and IL-13 stimulate the CCL22 production in monocytes and macrophages,53, 55 whereas interferon-γ reverts the stimulatory effect.56 Andrew et al.55 demonstrated an inhibition of CCL22 expression by IL10, a cytokine reported to be produced by RS cells, especially in EBV-positive cases.9, 57 TH2-like cells directly surrounding the RS cells may, by producing IL4 and IL13,55 stimulate the CCL22 production, which leads to an increase of TH2 cells, resulting in an amplification circuit of TH2 cells. In our analysis, expression of CCL22 appeared to be most abundant in B-cell lymphoma-derived cell lines, although the level of expression did vary between different samples tested. Remarkably 18 of 22 samples of HL expressed high signals by RT-PCR, especially in classical forms (15 of 16), but signals were also observed in 4 of 6 NLP HL and in 2 of 3 DLCL samples. No signals were obtained in ALCL samples. This is in agreement with a previous gene expression study in HL cell lines and single RS cells isolated from HL tissues, which demonstrated CCL22 overexpression when compared to germinal center cells.58 Another RT-PCR study showed the presence of CCL22 in whole HL tissues. However, our results do not confirm the correlation found between higher expression of CCL22 and the NS subtype reported by these authors.13, 14 Strong cytoplasmic staining of RS cells (Fig. 3) could be observed in 14 of 14 classical HL cases in agreement with a recent study.15 Although much lower, protein expression was also present in NLP cases, in contrast with the findings of Hedvat et al.,15 who did not detect MDC expression in this HL subtype. Together with CCL17, CCL22 provides an explanation for the predominance of TH2 cells in the lymphocyte infiltrate in HL tissues. However, according to our data, CCL17 remains more specific for classic subtypes of HL than CCL22 that is also expressed in NLP cases and B-cell non-Hodgkin lymphomas.

In recent studies, it has become clear that chemokines may be natural substrates of CD26, a leukocyte differentiation antigen expressed on the cell surface of T lymphocytes and macrophages.59 CD26 is a type II membrane glycoprotein with intrinsic dipeptidyl-peptidase IV (DPP IV) activity, which cleaves the first 2 amino acids from peptides with penultimate proline or alanine residue.60 Several chemokines, including CCL5, CCL8, CCL11, CXCL10 and CXCL12, share a conserved NH2-X-Pro sequence and have been shown to be CD26 substrates.61 In addition, Oravecz et al.61 showed that the truncated form of CCL5 has altered receptor specificity, demonstrating that CD26 is able to modulate the chemotactic effects of CCL5. Truncation of the 2 most NH2-primed amino acids from CCL11 protein by CD26 DPPIV resulted in a reduced chemotactic activity for eosinophils and an impaired binding to CCR3.62 For CCL8, it has been shown that the naturally modified form (amino acids 6–72) was devoid of activity, but that it could block the chemotactic effects of intact CCL8, CCL2, CCL7 and CCL5 completely.59 For CCL22, it has been shown that truncation of the protein resulted in a modified protein consisting of amino acids 3–69 that completely lost its ability to bind its receptor CCR4, indicating that CCL22 (3–69) has altered effects as compared to normal-sized CCL22 (1–69). Together these data demonstrate an important role for CD26 DPPIV activity in the regulation of chemotactic effects.

Expression of CD26 has been associated with T-cell activation.63 In HL, the T cells that directly surround the RS cells have a TH2-like phenotype and do not express CD26 on the cell surface.64 This lack of CD26 on TH2 cells prevents the modification of the chemokines expressed by the RS cells via the CD26 DPPIV activity. Of the chemokines expressed in HL, CCL5, CCL13, and CCL22 have been shown to be cleaved by DPPIV of CD26.59, 60 Whether CD26 DPPIV also functions in CCL17 processing is still unknown.

In summary, our study demonstrated that several chemokines are produced by RS cells and may be involved in the recruitment of different subsets of leukocytes in HL subtypes. These chemokines attract cells with different physiologic properties, resulting in a complex network of interactions between RS cells and the reactive component. CCL17 (TARC) appears to be the most specific chemokine for classical HL subtypes, whereas CCL22 (MDC) expression could also be observed in NLP and non-Hodgkin lymphoma samples.


Our study was supported by The Dutch Cancer Society (RUG, grant numbers 97-1580 to EMM and 01-2414 to JK). Dr. L.F. Bleggi Torres kindly provided part of the patient material from the Department of Medical Pathology, Federal University of Paraná, Brazil.