ORIGINAL ARTICLE High Expression of Several Tyrosine Kinases

High expression of several tyrosine kinases and activation of the PI3K/AKT pathway in mediastinal large B cell lymphoma reveals further similarities to Hodgkin lymphoma

C Renne´1, K Willenbrock1, JI Martin-Subero2, N Hinsch1,CDo¨ring1, E Tiacci3, W Klapper4,PMo¨ller5,RKu¨ppers3, M-L Hansmann1, R Siebert2 and A Bra¨uninger1

1Senckenberg Institute for Pathology, University of Frankfurt, Frankfurt, Germany; 2Institute of Human Genetics, University Hospital Schleswig-Holstein, Campus Kiel, Kiel, Germany; 3Institute for Cell Biology (Tumor Research), University of Duisburg-Essen, Essen, Germany; 4Institute of Pathology, Lymph Node Registry Kiel, University Hospital Schleswig-Holstein, Campus Kiel, Kiel, Germany and 5Department of Pathology, University of Ulm, Ulm, Germany

Mediastinal large B-cell (MBL) and classical Hodgkin lymphoma lymphoma (cHL).8 Both entities frequently affect young indivi- (HL) have several pathogenic mechanisms in common. As we duals and nsHL is also often localized in the mediastinum. recently observed aberrant tyrosine kinase (TK) activities in HL, Morphologically, a more or less extensive sclerosis is a common we now analysed also MBL for such activities. Indeed, MBL and feature of both entities and, owing to the presence of HL were the only B-cell lymphomas where elevated cellular phospho-tyrosine contents were typical features. Three TKs, interspersed small lymphocytes and eosinophils in MBL, some 1 JAK2, RON and TIE1, not expressed in normal B cells, were MBL infiltrates may resemble cHL. each expressed in about 30% of MBL cases, and 75% of cases Furthermore, molecular analyses indicated common patho- expressed at least one of the TKs. Among the intracellular genetic mechanisms for MBL and nsHL. Comparative genomic pathways frequently triggered by TKs, the PI3K/AKT pathway hybridization of primary tumor cases revealed that gains of was activated in about 40% of MBLs and essential for survival chromosomes 2p13–16 and 9p24 were the most frequent of MBL cell lines, whereas the RAF/mitogen-activated protein 8–14 kinase pathway seemed to be inhibited. No activating mutations genomic imbalances in both entities. Gains in 2p encompass were detected in the three TKs in MBL cell lines and primary the nuclear factor-kB (NF-kB) family member REL, and cases. RON and TIE1 were each also expressed in about 35% constitutive activation of the NF-kB pathway is usually observed and JAK2 in about 53% of HL cases. JAK2 genomic gains are in both lymphoma types.15–17 The 9p24 gains encompass the frequent in MBL and HL but we observed no strict correlation of JAK2 tyrosine kinase (TK), which activates STAT6, and JAK2 genomic status with JAK2 protein expression. In conclu- constitutive STAT6 activation is another common feature of sion, aberrant TK activities are a further shared pathogenic 18,19 mechanism of MBL and HL and may be interesting targets for both lymphoma types. In two global gene expression therapeutic intervention. analyses sets of genes discriminating MBL from other DLBCL Leukemia (2007) 21, 780–787. doi:10.1038/sj.leu.2404594 were identified, and a substantial fraction of these genes was Keywords: mediastinal large B cell lymphoma; Hodgkin lymphoma; regulated in a similar manner in MBLs and cHL cell lines as tyrosine kinase; PI3K/AKT pathway; RAF/MAPK pathway compared with DLBCLs.20,21 A pathogenetic mechanism frequently observed in tumor- igenesis is the aberrant activation of TKs. We recently described the aberrant expression and activation of six different receptor tyrosine kinases (RTK) in cHL as compared with the non- neoplastic counterparts of the tumor cells, which was most Introduction pronounced in nsHL and indicated that aberrant TK activity may play an important role in the pathogenesis of nsHL.22 Given the Mediastinal large B-cell lymphoma (MBL) is a distinctive similarity of nsHL and MBL regarding pathogenetic mechan- subtype of diffuse large B-cell lymphoma (DLBCL). Patients isms, we investigated in the present study if aberrant TK activity are usually young, often female and present with localized bulky and activation of intracellular signalling pathways triggered by mediastinal masses, which may contain thymic remnants. The TKs could also contribute to MBL pathogenesis. tumor cells express CD19, CD20, CD22, CD79A and PAX5, thus resembling mature B cells.1,2 However, despite the presence of functionally rearranged immunoglobulin V-genes Materials and methods they do not express any immunoglobulin detectable by 3,4 immunohistochemistry (IHC), and downregulation of the Tissue samples and cell culture internal IgH enhancer activity likely contributes to this Tissue samples were retrieved from the files of the Department phenomenon.5 Because of the expression of MAL and CD23 of Pathology of the University of Frankfurt and were originally in most cases, a derivation from medullary thymic B cells, which submitted for diagnostic purposes. Paraffin sections of six cHL express both proteins, is assumed.6,7 with known JAK2 genomic status as detected by fluorescence in Over the last years, several studies revealed similarities of situ hybridization (FISH) were retrieved from the Institute of MBL to the nodular sclerosis (ns) subtype of classical Hodgkin Pathology, Lymph Node Registry Kiel, University Hospital Schleswig-Holstein Campus Kiel. Altogether, 38 MBL samples Correspondence: Dr A Braüninger, Department of Pathology, Uni- and, in addition to the HL cases analysed in our previous versity of Frankfurt, Theodor-Stern-Kai 7, Frankfurt 60590, Germany. 22 E-mail: [email protected] study, 46 cHL samples were used for TIE1 and JAK2 stainings Received 7 July 2006; revised 20 December 2006; accepted 29 and 25 further nsHL and 65 further mixed cellularity (mc) HL December 2006 cases for the pan-phospho-tyrosine-specific stainings. All tissue High expression of several tyrosine kinases and activation C Renne´ et al 781 samples were studied in accordance with national ethical exemplarily verified for each antibody with at least two MBL principles. The MBL derived cell lines Karpas1106P and MedB-1 cases by preincubation with the peptides used for immunization were grown in RPMI1640 supplemented with 20% fetal-calf or pretreatment of sections with Lambda protein phosphatase serum (FCS) for Karpas1106P and 10% FCS and 50 mM Hepes (Upstate, Lake Placid, NY, USA; 1250 U per section in 200 ml for MedB-1, and the HL cell lines L1236, L428 and KMH2 were buffer (50 mM Hepes pH 7.5, 1 mM ethylenediaminetetraacetic grown in Roswell Park Memorial Institute media (RPMI)-1640 acid, 20 mM manganese chloride and 5 mM Dithiothreitol) for with 10% FCS and HDLM2 in RPMI-1640 with 20% FCS.23,24 3 h at 371C), which in all instances abolished or significantly reduced staining. For detection of bound primary antibody the ABC system with horseradish peroxidase and 3.30-diaminoben- Analysis of global gene expression data zidine as substrate was used (Dako). Global gene expression data sets of 76 MBL and 34 DLBCL obtained using U133A and B microarrays of Affymetrix were retrieved from the Supplementary Data of Savage et al. Fluorescence in situ hybridization (www.genome.wi.mit.edu/cancer/mediastinal.html.).21 The data- Fluorescence in situ hybridization (FISH) analyses for the set was loaded in the GeneSpring software (Version 7.2, Agilent detection of chromosomal rearangements of the JAK2 locus in 12 Technologies, Bo¨blingen, Germany) using per chip and per gene 9p24 were performed using a previously published probe and 26 normalisation as recommended. A supervised comparison of protocol. Slides were analyzed using a Zeiss Axioskop-2 MBL and DLBCL samples was performed to identify TKs (all TKs fluorescence microscope (Zeiss, Go¨ttingen, Germany) equipped from the human kinome (www.stke.org) for which probe sets with appropriate filter sets (AHF, Tu¨bingen, Germany) and were identified on the Affymetrix homepage (www.affymetrix. documented using an ISIS imaging system (MetaSystems, com/analysis/index.affx), i.e., 90 TKs) with a more than twofold Altlussheim, Germany). Nuclei from Hodgkin-Reed/Sternberg differential expression between MBL and DLBCL groups. For (HRS) cells were identified by their large size and frequent generation of a heat map of all TKs the same dataset was hyperploid genomic status. The ratio between the median analysed using the Cluster and Tree View software,25 performing number of JAK2 hybridization signals per case and its estimated median normalization and centering of genes. ploidy level determined as median of number of signals for 12 different loci investigated by FISH, (data not shown) was used to determine the JAK2 genomic status. Ratios between 1.3 and Immunohistochemistry 1.99 were considered as gains whereas ratios equal or higher For Immunohistochemistry (IHC) 5 mm sections of the formalin- than two were considered as amplifications. fixed paraffin-embedded tissues were prepared. After antigen m retrieval, sections were blocked with normal goat serum (4 g/ 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium ml; Santa Cruz Biotechnology, Heidelberg, Germany) for 15 min bromide assay and incubated overnight with the primary antibodies in Tris- Cells were grown in 96 well tissue culture plates (starting with buffered saline (TBS) at room temperature. For detection of 1 Â 105 cells/ml) with various concentrations of LY294002 primary antibodies the Envision system (Dako, Hamburg, (3–50 mM). After an incubation time of 24 h 3-(4,5-dimethylthia- Germany) with either alkaline phosphatase and Fast Red or zol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Calbiochem, horseradish peroxidase with 3.3-diaminobenzidine (both from Darmstadt, Germany) was added to a final concentration of Dako) as substrates were used. Antigen retrieval procedures, 0.5 mg/ml. Cells were incubated for further 4 h and then lysed in suppliers and concentrations at which antibodies against 5% sodium dodecyl sulfate (SDS) for 12 h at 371C. Absorbance PDGFRA, DDR2, EPHB1, RON, TRKB, TRKA and phospho- was measured with a EL311SX microplate reader (Biotek tyrosine (p-Y)(4G10) were used as described previously.22 For Instruments,Winooski, VT, USA) at 550 nm with the reference anti-TIE1 and anti-JAK2 antibodies (Santa Cruz Biotechnology, wavelength set to 690 nm. Each value was measured as six sc-342 and sc-278, respectively), sections were cooked for replicates in three independent experiments. Data were 15 min in 600 ml citrate buffer (10 mM sodium (Na) citrate pH analysed using the GraphPad Prism 4.0 software (GraphPad, 6.0) in a microwave oven at maximum power for antigen San Diego, CA, USA). retrieval. The anti-TIE1-antibody was diluted 1:100 and the anti- JAK2-antibody 1:50 in Tris-buffered saline. Specificity of stainings was tested by pre-incubation of the antibodies with Real-time reverse transcriptase–polymerase chain 20-fold amounts of the peptides used for immunization. For reaction analysis both, TIE1 and JAK2, this pre-incubation strongly reduced or RNA was isolated from Karpas1106P, MedB-1, L1236, L428, completely abolished staining. Specificity of the TIE1 staining KMH2 and HDLM2 cultures with the RNAeasy kit of Qiagen was further confirmed using another TIE1-specific antibody (Qiagen, Hilden, Germany). One microgram of each RNA was (AF619, R&D, Wiesbaden, Germany) at a 1:50 dilution with used with random hexamers for first strand complementary DNA 15 min cooking in citrate buffer for antigen retrieval. From the (cDNA) synthesis using the Roche first strand cDNA synthesis kit cases analysed, the same staining patterns were observed with (Roche, Mannheim, Germany) and analysed for RON and TIE1 both anti-TIE1 antibodies. expression with premade real-time reverse transcriptase–poly- Intracellular signalling was analysed with antibodies specific merase chain reaction (RT–PCR) Assays-on-Demand of Applied for phospho-Ser473-AKT, phospho-Thr202/Tyr204-p44/42- Biosystems (Hs00234013_m1, Hs00892696_m1, Applied Bio- MAPK and phospho-Ser259-RAF (all from Cell Signalling, systems, Weiterstadt, Germany). B2M was used as reference gene Beverly, MA, USA no. 4051, no. 4376 and no. 9421). For all (4326319E of Applied Biosystems). three antibodies antigen retrieval was performed by cooking the section for 15 min at maximum power in a microwave oven in 600 ml 10 mM Na citrate pH 6.0. Incubation with primary Western blot analysis and immunoprecipitation antibody was performed overnight in TBS with 1:200, 1:100 and Karpas1106P and MedB-1 cells were lysed by boiling for 10 min 1:50 dilutions, respectively. Specificity of the antibodies was in Laemmli buffer. Lysates corresponding to approximately

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 782 1 Â 105 cells per lane were separated by SDS–polyacrylamide lymphoma, follicular lymphoma, Burkitt lymphoma, chronic gel electrophoresis (7.5%), blotted onto Polyvinylidene fluoride lymphocytic leukemia, DLBCL and lymphocyte-predominance membranes (BioRad, Munich, Germany), incubated overnight at and cHL) with a pan-p-Y-specific antibody revealed that among 41C with 1:1000 dilutions of anti-TIE1 and anti-RON (Santa B-cell non-Hodgkin lymphomas (B-NHL) only minor fractions of Cruz Biotechnology, sc-342 and sc-322) and visualized using cases (0–21%) contained aberrantly high amounts of p-Y the ELC plus system (Amersham Biosciences, Freiburg, Ger- (Table 1 and22). Distinct from the B-NHLs, the majority of nsHL many). For immunoprecipitation cells were incubated for 2 h cases showed elevated p-Y contents in the HRS cells, and further 7 with 2 mM Na3VO4 to enrich p-Y-proteins and 1 Â 10 cells analysis showed that this was due to the aberrant expression and were lysed in mammalian protein extraction (M-PER) buffer activation of several RTKs.22 (Pierce, Bonn, Germany). Lysates were precleared with Pansor- In the present study, we analysed 37 cases of MBL for cellular bin (Calbiochem, Darmstadt, Germany) and incubated with p-Y content by IHC and extended our analysis of nsHL cases 20 mg 4G10 (or normal mouse Ig as control) antibody bound to (Table 1, Figure 1). Fifteen of the 37 (40%) of the MBL, and 34 of Pansorbin at 41C overnight. Precipitates were washed four times 64 (53%) of the nsHL cases had aberrantly high cellular p-Y in M-PER buffer and cooked for 5 min in 100 ml Laemmli buffer. contents (Table 1). MBL was thus beside nsHL the only entity For Western blot analysis, 10 ml aliquots were used. among mature B-cell lymphomas with elevated cellular p-Y content in a large fraction of cases. This observation indicated that aberrant TK activity might contribute also to MBL Analysis for mutations in JAK2, RON and TIE1 pathogenesis, and we thus tried to identify the aberrantly To analyse for the presence of mutations in the TKs in the cell activated TKs in MBL. lines, RNA from Karpas1106P and MedB-1 was extracted using the Qiagen RNAeasy kit (Qiagen) and used for cDNA synthesis with the Roche first strand cDNA synthesis kit and oligo-dT Aberrant expression of the TKs TIE1, JAK2 and RON primers (Roche, Mannheim, Germany). For the amplification of in MBL the complete open reading frames, 1 ml aliquots of the reactions To identify TKs aberrantly expressed in MBL as compared with were used as templates in PCRs with each primer (sequences in their proposed normal counterparts, that is, thymic B cells, we the Supplementary Data) at 0.125, 2 mM dNTPs, 2 mM performed IHC for candidate TKs on MBLs and normal thymi. magnesium chloride (MgCl2) and 3.5 U of Expand High Fidelity Because of the several similarities of MBL and cHL, all six RTKs in 1 Â Expand PCR buffer (Roche). Each PCR was run for 45 previously identified as aberrantly expressed in HL were cycles of 30 s 951C, 30 s 611C and 6 min 681C after an initial considered as such candidate TKs.22 In addition, we reanalysed denaturation of 2 min at 951C. published global gene expression data of MBL obtained with For the analysis of presence of JAK2 V617F mutation in Affymetrix U133 gene expression microarrays to identify further primary MBLs, RNA was extracted either from 10 mm sections of candidate TKs.21 Owing to a lack of data for normal B cells in frozen (12 cases) or formalin-fixed paraffin-embedded tissues this data set, these analyses were restricted to the identification (10 cases) using Trizol reagent (Invitrogen, Karlsruhe, Germany) of TKs more strongly expressed in MBL than in other DLBCLs.21 followed by clean-up with RNAeasy Minikit columns (Qiagen). In a supervised analysis of 34 MBLs and 76 DLBCLs using the cDNA was synthesised using 1 mg total RNA, random hexamer GeneSpring software no TK with an at least twofold stronger primers and the Roche first strand cDNA synthesis kit (Roche). expression in the MBL group as compared with the DLBCLs was One microlitre of cDNAs was used as template in PCRs with identified. A visual inspection of a heat map of TK expression in 0.125 mM each primer (sequences in the Supplementary Data), both lymphoma groups generated with the Cluster and Tree- 2mM dNTPs and 2 mM MgCl2 and 45 cycles of 30 s 951C30s View software indicated, however, that JAK2 was strongly 611C and 30 s 721C after an initial denaturation of 2 min at 951C. expressed in 15 of 34 and TIE1 in 12 of 34 MBL cases To search for mutations in TIE1 and RON, genomic DNA from formalin-fixed paraffin-embedded primary cases was extracted with the Qiagen DNA mini kit. A portion of 200–800 ng DNA Table 1 Analysis of cellular p-Y content in B-cell lymphomas using was used as template in PCRs with 0.125 mM each primer a pan-p-Y-specific antibody (sequences in the Supplementary Data), 2 mM dNTPs and 2 mM 1 1 1 MgCl2 and 45 cycles of 30 s 95 C, 30 s 61 C and 45 s 72 C after Lymphoma Positive cases/ % of positive an initial denaturation of 5 min at 951C. analysed cases cases All PCR products were visualized on agarose gels, DNA from bands was extracted with the Qiagen gel extraction kit (Qiagen) MBL 15/37 40 and sequenced with the corresponding primers on an ABI 3100 nsHL 34/64 53 automated sequencer with the Big Dye Terminator cycle mcHL 15/125 12 Lymphocyte-predominance HL 1/23 4 sequencing kit (Applied Biosystems). Mantle cell lymphoma 4/19 21 CLL 0/15 0 FL 3/23 13 Results BL 0/9 0 Diffuse large B cell lymphoma 2/20 10 Elevated cellular phospho-tyrosine content is a shared Abbreviations: BL, Burkitt’s lymphoma; CLL, chronic lymphocytic feature of MBL and nsHL among mature B-cell leukemia; FL, follicular lymphoma; MBL, mediastinal large B cell lymphomas lymphom; mcHL, mixed cellularity Hodgkin lymphoma; nsHL, nodular In tumor cells with aberrant TK activity as compared with their sclerosis Hodgkin lymphoma. Analysis of B-NHLs except MBL as well as lymphocyte-predominance normal counterparts cellular phospho-tyrosine (p-Y) content is and part of mc and ns HL cases has already been described in Renne´ often elevated and can be detected by IHC with pan-p-Y- et al.22 The frequency of pan-p-Y positive cases is significantly higher 27 specific antibodies (Supplementary Information, Figure 3). in MBL compared with all B-NHL cases (Po0.001 using Pearson’s w2 Our previous analysis of several B-cell lymphomas (mantle cell test).

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 783 (Supplementary Information, Figure 1), and both were thus addition, in 10 of the 13 cases, which reacted with the pan-p-Y considered as candidate TKs. antibody at least one of the three TKs was expressed. IHC for the eight candidate TKs revealed that three were expressed in signiﬁcant fractions of MBLs but not in normal thymic B cells and, in line with the RNA expression data, in TIE1 and JAK2 are also aberrantly expressed in HRS much smaller fractions of DLBCL (Figure 1, Supplementary data cells of cHL Figure 2, Table 2). TIE1 was expressed in 13 of 35 MBLs (37%), Given the similarities between MBL and cHL we analysed JAK2 in 10 of 28 MBLs (36%) and RON in 12 of 38 MBL cases whether, beside the already known RON expression in cHL (in analysed (31%). The other ﬁve RTKs (PDGFRA, DDR2, EPHB1, 36% of cases), also TIE1 and JAK2 were detectable by IHC in TRKB and TRKA) were only expressed in minor fractions of cases cHL (Figure 1, Table 2). TIE1 expression was observed in 13 of (3–11%) (Table 2). In 75% of all MBL, analysed at least one of 40 cHL cases (32%), with a preferential expression in the ns the three TKs TIE1, JAK2 or RON was expressed, and in about subtype, where 10 of 20 (50%) cases were positive, whereas 30% of cases, two of the three TKs were coexpressed. In only three of 20 mc cases (15%) showed positivity. JAK2

Figure 1 Immunohistochemical stainings of MBLs. Immunohistochemical stainings of a MBL case with the pan-phospho-tyrosine-speciﬁc antibody (4G10, Upstate), (a) and examples of MBL cases with immunohistochemical stainings for RON, TIE1 and JAK2, respectively, are shown (b–d). Immunohistochemical stainings of cHL cases for TIE1 and JAK2, respectively, are shown in (e) and (f). Stainings of MBLs for p-AKT and p-RAF-Ser259 are shown in (g) and (h), respectively. Primary antibodies were visualized using horseradish peroxidase with 3.3-diaminobenzidine and in (a–c, g, h) and alkaline phosphatase and Fast Red as substrate in (d–f).

Figure 2 Sensitivity of Karpas1106P and MedB-1 towards the PI3K inhibitor LY294002. Karpas1106P, MedB-1 and for comparison also the LY294002 sensitive T-cell leukemia-derived cell line Jurkat were incubated at 1 Â 105 cells/ml with various concentrations of LY294002 for 24 h. Survival was monitored with a MTT assay (based on the activity of the mitochondrial respiratory chain) (a). Western blot analysis with a phospho- AKT-speciﬁc antibody (b) showed that treatment of the three cell lines with LY294002 for 24 h before lysis signiﬁcantly reduced the amount of phosphorylated AKT, indicating that the effect of LY294002 on the three cell lines is largely owing to inhibition of PI3K.

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 784 Table 2 Immunohistochemical detection of aberrant TK expression correlation between activation of the PI3K/AKT pathway and in MBL, nsHL and mcHL Ser259 phosphorylation of RAF was observed (of 14 cases for which both, the p-AKT and the p-RAF IHC were informative, Tyrosine kinase Cases positive/cases analysed (%) eight were p-AKT and nine p-RAF positive, and only three of those were positive for both stainings). Together, these findings MBL nsHL McHL indicate that the PI3K/AKT pathway is frequently activated in TIE1 13/35 (37) 10/20 (50) 3/20 (15) MBLs whereas the RAF/MAPK pathway seems to be inhibited. JAK2 10/28 (36) 21/31 (68) 14/35 (40) PI3K is inhibited by LY294002. We used this compound to RON 12/38 (31) 8/18 (44) 6/21 (29) analyse the importance of PI3K activation for survival of the two TRKA 4/38 (11) 7/18 (39) 5/21 (24) cell lines so far established from MBLs, Karpas1106P and MedB-1 PDGFRA 2/38 (5) 17/18 (94) 12/21 (57) (Figure 2). Both cell lines showed sensitivity towards LY294002 DDR2 2/38 (5) 12/18 (67) 5/21 (24) EPHB1 2/38 (5) 11/18 (61) 5/21 (24) at concentrations causing specific inhibition of AKT in other cell TRKB 1/38 (3) 11/18 (61) 1/21 (5) lines (e.g., Jurkat and HRS cell lines) and a LY294002-dependent decrease in AKT phosphorylation,33,34 suggesting that constitu- Abbreviations: MBL, mediastinal large B cell lymphoma; mcHL, mixed tive PI3K signalling is, as for most cell lines, indeed also an cellularity Hodgkin lymphoma; nsHL, nodular sclerosis Hodgkin lymphoma. important survival factor for these MBL-derived cell lines. Analysis of nsHL and mcHL for RON, TRKA, PDGFRA, DDR2, EPHB1 and TRKB expression has already been described in Renne´ et al.22 No activating mutations are present in JAK2, RON and TIE1 Table 3 Intracellular signalling pathways activated in MBL TKs are frequently activated by mutations and we analysed the transcripts encoding JAK2, RON and TIE1 in the two MBL cell Antibody Positive cases/ % positive lines for potentially activating mutations. JAK2 protein analysed cases cases expression in Karpas1106P and MedB-1 has been shown previously,35,36 and our real-time RT–PCR analysis indicated p-Thr202/Tyr204-MAPK 2/27 7 that RON was also expressed in both cell lines and TIE1 in p-Ser473-AKT 10/24 42 p-Ser259-RAF 11/28 39 Karpas1106P (data not shown). In addition, TIE1 transcripts were also detected in the HL cell line KMH2 but not in L1236, Abbreviation: MBL, mediastinal large B-cell lymphoma. L428 and HDLM2 HL cell lines (data not shown). In line with the real-time PCR, Western blot analysis demonstrated TIE1 protein expression in Karpas1106P, whereas significant RON expression was observed in 35 of 66 cHL cases (53%), and protein expression was only observed in MedB-1, although the expression was more frequent in ns subtype than in mc subtype real-time PCR analysis indicated only small differences of RON (21 of 31 (68%) and 14 of 35 (40%) cases, respectively). In RNA expression levels between Karpas1106P and MedB-1 normal lymphatic tissues, (lymph node, tonsil and spleen) (Figure 3). We then used RT–PCRs to amplify the complete neither TIE1 nor JAK2 expression was observed in B cells open-reading frames of JAK2 and RON from both cell lines and (Supplementary Information). All three TKs aberrantly expressed TIE1 from Karpas1106P. The sequence analysis of these RT–PCR in MBL were thus also expressed at comparable frequencies products revealed neither presence of potentially activating in cHL. mutations nor could the reason for the lack of RON protein expression in Karpas1106P be clarified. However, despite the lack of mutations the three TKs were activated in the MBL cell The PI3K/AKT pathway is activated and the RAF/MAPK lines. Activation of JAK2 in both cell lines as determined by pathway seems to be inhibited in MBL p-JAK2 specific antibodies has been shown previously,35,36 and Given the aberrant expression of RON and TIE1 in MBL, we immunoprecipitation with a pan-p-Y-specific antibody and analysed whether intracellular signalling pathways frequently subsequent Western blotting with TIE1 and RON-specific triggered by receptor tyrosine kinase (RTK)s, namely the RAF/ antibodies revealed that both RTKs are also phosphorylated mitogen-activated protein kinase (MAPK) and the Phosphoinosi- (Figure 3). tide 3-kinases (PI3K)/protein kinase B (AKT) pathways, are A constitutively dimerized and activated RON form encoded activated in MBLs.28–30 Activation of the RAF/MAPK and PI3K/ by a splice variant lacking exon 11 has been described in AKT pathways were analysed using antibodies specific for different tumor types.37 Using quantitative RTFPCR, this splice Thr202/Tyr204-phosphorylated MAPK and Ser473-phosphory- variant was also detected in MBLs but also in several lymphatic lated AKT, respectively. Using such antibodies for IHC, we organs with no signs of neoplasia (data not shown), indicating observed that only in two of 27 (7%) MBL cases the MAPK was that presence of the splice variant is not necessarily correlated activated whereas p-AKT was detected in 10 of 24 (42%) MBLs with neoplasia. (Table 3, Figure 1). Of the nine p-AKT positive cases where Recently, several groups described a G to T transversion in the information about TK expression was available, three cases pseudokinase domain of JAK2, which causes constitutive JAK2 expressed none, two cases expressed one, and four cases activation in myeloproliferative diseases.38 Given the aberrant expressed two of the three TKs. JAK2 expression in about a third of MBLs, we analysed 15 cases As AKT can negatively regulate the RAF/MAPK pathway by of MBL by RT–PCR and sequencing for presence of the phosphorylation of RAF in the regulatory domain on described G to T transversion. However, all cases displayed Ser259,31,32 we used an antibody specific for p-Ser259-RAF to the wild-type sequence, in line with a recent analysis of 20 MBL analyse whether activated AKT might inhibit the RAF/MAPK cases, indicating that in MBL V617F replacement in JAK2 does pathway in MBL. In line with the low fraction of p-MAPK not occur.39 positive cases, we observed p-Ser259-RAF positivity in 11 of 28 So far no recurrent activating mutations in RON and TIE1 (39%) cases (Table 3, Figure 1). However, in individual cases no have been observed in human neoplasias. However, as the

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 785 expressing MBLs. In addition, we sequenced exon 20 of TIE1, because the arginine corresponding to the one affected by the R849W mutation causing constitutive activation of TIE2 in vascular dysmorphogenesis is located in this exon.40 Further- more, as two experimentally introduced mutations in exon 18 cause constitutive activation of RON, we also sequenced exon 18 from primary cases.41 However, neither in the seven TIE1 expressing (four of these being p-AKT positive) nor in the six RON expressing cases (one of these being p-AKT positive) analysed mutations were observed.

Genomic JAK2 gains are often but not strictly accompanied by JAK2 expression Gains of the JAK2 gene localized in chromosome band 9p24 are frequently observed in MBL and cHL by comparative genomic hybridization and FISH (about 60 and 30% of cases, respectively),9,12 and these gains might cause aberrant JAK2 expression.9,10,13,14 In addition, constitutive activation of JAK2 in the MBL-derived cell line MedB-1 has been shown.35 To analyse whether genomic gains of the JAK2 locus correlate with increased JAK2 protein expression, IHC for JAK2 was performed with 26 cases of cHL with known JAK2 genomic status as determined by FISH. Although 14 of 16 cases with genomic JAK2 gains showed expression of JAK2 protein, also six of 10 cases without genomic JAK2 gains were positive in JAK2 IHC. These ﬁndings indicate that in addition to the JAK2 genomic gains, also other mechanisms contribute to the high JAK2 expression, at least in cHL.

Discussion

MBL and nsHL show several similarities, and it is likely that also similar transformation mechanisms contribute to lymphomagen- esis in both entities. Our analysis now revealed a further similarity in this regard as both entities were the only ones among several types of B-cell lymphomas where significant fractions of cases contained elevated cellular p-Y contents indicative of aberrant TK activity. Taking into account that p-Ys are very unstable and often not well preserved even in cases with constitutively activated TKs,27,42 it is likely that in the majority of cases of MBL TKs are aberrantly activated, as has Figure 3 Western blot analysis of RON and TIE1 expression and been previously shown for nsHL.22 activation in MBL and HL cell lines. Lysates corresponding to An IHC analysis for aberrantly expressed TKs in MBL approximately 1 Â 105 cells were loaded per lane and after blotting to membranes probed with the anti-RON (a) and anti-TIE1 (b) identified three TKs, JAK2, RON and TIE1, each of which was antibodies used for IHC. For RON the HL cell lines L428 and L1236 expressed in about 30% of cases. Together, about 75% of MBL were included as positive and KMH2 as negative control. Although cases aberrantly expressed at least one of these TKs and about nearly similar amounts of RON transcripts were observed in 30% of cases showed coexpression of two of the TKs. The vast Karpas1106P and MedB-1 with real-time RT–PCR (data not shown) majority of cases (10 of 13) showing positivity in the pan-p-Y only for MedB-1 strong RON protein expression was observed. TIE1 stainings expressed at least one of the three TKs. This may expression was only observed in Karpas1106P and the HL cell line KMH2, in line with real-time PCR analysis, where among the HL cell indicate that the elevated cellular p-Y content of MBLs is largely lines L1236, L428, KMH2 and HDLM2 only KMH2 showed TIE1 owing to the aberrant expression of these three TKs. However, as expression (data not shown). Loading of equal amounts of proteins was our IHC analysis was largely based on the inspection of the controlled with an anti-actin antibody. For analysis of the activation available global gene expression data and therefore limited to a status of the RTKs, cell lines were incubated for 2 h with 2 mM Na3VO4 comparison of MBL to DLBCL and not to normal B cells, we may before lysis to enrich tyrosine-phosphorylated proteins. P-Y-proteins have missed TKs aberrantly expressed in both entities. were immunoprecipitated from lysates of 1 Â 107 cells with a pan-p-Y For one of the TKs, JAK2, constitutive activation in the MBL- antibody (4G10)(lanes B in (c) and (d) and 1/10 of the lysates were 35 analysed with RON (c) and TIE1 (d) specific antibodies. Normal mouse derived cell line MedB-1 has recently been shown, and here Ig was used as negative control (lanes A in (c) and (d)). we show that also TIE1 and RON are activated in the MBL cell lines. With the currently available reagents, however, we could juxtamembrane regions of RTKs are often affected by activating not analyse the activation status of the three TKs in primary mutations we amplified the genomic regions encoding these MBLs directly. In a search for activating mutations in the TKs in parts (exons 13 and 14) of TIE and RON from TIE1 and RON the MBL cell lines and primary cases so far no mutations were

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 786 detected. RTKs like TIE1 and RON can also be constitutively Acknowledgements activated by their ligands via auto- or paracrine mechanisms, but analysis of the global gene expression data indicated that We thank Yvonne Blum, Sabine Albrecht, Ekaterini Hadzoglou, there is no significant expression of the RON ligand MST1 in Claudia Becher and Dorit Schuster for excellent technical MBL, and the ligand for TIE1 is unknown.21 However, the assistance. Supported by grants from the DFG (RK 1315/2-1 and selection of the tumor cells for aberrant expression of each of the 2-2 to RK and BR 1238/6-1 and 6-2 to AB), the Deutsche three TKs in about 30% of cases, the activation of the TKs in the Krebshilfe Network (70-3173-Tr3 to MLH, PM and RS) and the MBL cell lines and the high cellular p-Y content in primary cases Deutsche Jose´ Carreras Leuka¨mie Stiftung (to ET). strongly argues for an essential role and activation of the TKs in MBL pathogenesis. The aberrant JAK2 activity in MBL seems to be due to two References different genetic aberrations. On the one hand, inactivation of the SOCS1 protein owing to deletions or mutations prevents 1 Banks PM, Warnke RA (eds). Mediastinal (thymic) large B-cell degradation of the JAK2 protein,35,43 and on the other hand lymphoma. IARC Press: Lyon, 2001. 12 2 Barth TF, Leithaüser F, Joos S, Bentz M, Mo¨ller P. Mediastinal JAK2 gene copy number gains may cause a gene dosage effect. (thymic) large B-cell lymphoma: where do we stand? Lancet Oncol For the latter mechanism, however, our data indicate that JAK2 2002; 3: 229–234. gene copy gains in MBL are only in a fraction of cases 3 Kanavaros P, Gaulard P, Charlotte F, Martin N, Ducos C, Lebezu M accompanied by JAK2 protein expression detectable by IHC et al. Discordant expression of immunoglobulin and its associated (60% of cases show JAK2 gene gains in previously published molecule mb-1/CD79a is frequently found in mediastinal large B series,8 but we observed expression only in 36% of cases). A cell lymphomas. Am J Pathol 1995; 146: 735–741. 4 Leithaüser F, Baüerle M, Huynh MQ, Mo¨ller P. Isotype-switched direct correlation of JAK2 expression with JAK2 gene dosage in immunoglobulin genes with a high load of somatic hypermutation cHL revealed that most cases with genomic gains showed also and lack of ongoing mutational activity are prevalent in JAK2 protein expression, but that also the majority of cases mediastinal B-cell lymphoma. Blood 2001; 98: 2762–2770. without genomic JAK2 gains expressed the protein. JAK2 5 Ritz O, Leithaüser F, Hasel C, Bru¨derlein S, Ushmorov A, Mo¨ller P expression is thus not strictly correlated to the JAK2 genomic et al. Downregulation of internal enhancer activity contributes to status. abnormally low immunoglobulin expression in the MedB-1 mediastinal B-cell lymphoma cell line. J Pathol 2005; 205: In most tumors where aberrant TK activity has been observed, 336–348. activation of a single TK (e.g., ALK in anaplastic large cell 6Mo¨ller P, Moldenhauer G, Momburg F, La¨mmler B, Eberlein- lymphoma or BCR-ABL in chronic myeloid leukemia)44 or two Gonska M, Kiesel S et al. Mediastinal lymphoma of clear cell type TKs from the same family of TKs (e.g. KIT or PDGFRA in is a tumor corresponding to terminal steps of B cell differentiation. gastrointestinal stroma tumors)45 were found in the majority of Blood 1987; 69: 1087–1095. cases. It is thus an interesting question whether these three TKs, 7 Copie-Bergman C, Gaulard P, Maouche-Chretien L, Briere J, Haioun C, Alonso MA et al. The MAL gene is expressed in primary which belong to different families of TKs, contribute in the same mediastinal large B-cell lymphoma. Blood 1999; 94: 3567–3575. way to MBL pathogenesis or whether each is relevant only for a 8 Poppema S, Kluiver JL, Atayar C, Berg A, Rosenwald A, Hummel M specific aspect of the MBL phenotype. For JAK2 it seems clear et al. Report: workshop on mediastinal grey zone lymphoma. Eur J that activation of STAT5 and STAT6 is a major conse- Haematol 2005; 75 (Suppl 66): 45–52. quence.19,35 It should, however, be noted that STATs can also 9 Bentz M, Barth TF, Bru¨derlein S, Bock D, Schwerer MJ, Baudis M be activated by several RTKs,28 and that there is thus the et al. Gain of chromosome arm 9p is characteristic of primary mediastinal B-cell lymphoma (MBL): comprehensive molecular possibility that also TIE1 and RON could contribute to the cytogenetic analysis and presentation of a novel MBL cell line. STAT5/6 activation in MBL. For TIE1 and RON activation of the Genes Chromosomes Cancer 2001; 30: 393–401. PI3K/AKT pathway has been described,29,30 and we indeed 10 Joos S, Ku¨pper M, Ohl S, von Bonin F, Mechtersheimer G, Bentz M were able to demonstrate the activation of this pathway in MBL. et al. Genomic imbalances including amplification of the tyrosine Furthermore, JAKs can also activate the PI3K/AKT signalling kinase gene JAK2 in CD30+ Hodgkin cells. Cancer Res 2000; 60: cascade,46 and it is thus conceivable that all three TKs 549–552. 11 Joos S, Menz CK, Wrobel G, Siebert R, Gesk S, Ohl S et al. contribute to MBL pathogenesis by constitutive activation of Classical Hodgkin lymphoma is characterized by recurrent copy the same signalling molecules, namely PI3K/AKT and STATs. number gains of the short arm of chromosome 2. Blood 2002; 99: The PI3K/AKT pathway is constitutively activated in several 1381–1387. different types of tumors, among them also HL,33,47,48 and can, 12 Joos S, Granzow M, Holtgreve-Grez H, Siebert R, Harder L, beside the inhibition of apoptosis, also enhance cell prolifera- Martin-Subero JI et al. Hodgkin’s lymphoma cell lines are tion and growth.49 Constitutive STAT6 activation has so far only characterized by frequent aberrations on chromosomes 2p and 18,50 9p including REL and JAK2. Int J Cancer 2003; 103: 489–495. been observed in HL and prostate cancer. The conse- 13 Martin-Subero JI, Gesk S, Harder L, Sonoki T, Tucker PW, quences of the activation of these signalling molecules in MBL Schlegelberger B et al. Recurrent involvement of the REL and await further investigation. In this regard, our experiments with BCL11A loci in classical Hodgkin lymphoma. Blood 2002; 99: the PI3K inhibitor LY294002 and the MBL cell lines gave first 1474–1477. hints for a role of PI3K activation in promoting survival of MBL 14 Martin-Subero JI, Knippschild U, Harder L, Barth TF, Riemke J, cells. Grohmann S et al. Segmental chromosomal aberrations and centrosome amplifications: pathogenetic mechanisms in Hodgkin Taken together, aberrant TK activities in significant fractions and Reed-Sternberg cells of classical Hodgkin’s lymphoma? of cases are a common feature of MBL and nsHL and distinguish Leukemia 2003; 17: 2214–2219. both entities from other mature B cell lymphomas. Both entities 15 Bargou RC, Emmerich F, Krappmann D, Bommert K, Mapara MY, have the aberrant expression of JAK2, RON and TIE1 in Arnold W et al. Constitutive nuclear factor-kappaB-RelA activation common. Thus, our analysis adds MBL to the list of tumors is required for proliferation and survival of Hodgkin’s disease where aberrant TK activity likely plays an important role in tumor cells. J Clin Invest 1997; 100: 2961–2969. 16 Feuerhake F, Kutok JL, Monti S, Chen W, Lacasce AS, Cattoretti G pathogenesis. Together with the identification of the constitu- et al. NF kappa B activity, function and target gene signatures in tively activated PI3K/AKT pathway these findings may form the primary mediastinal large B-cell lymphoma and diffuse large B-cell basis for novel therapeutic strategies. lymphoma subtypes. Blood 2005; 106: 1392–1399.

Leukemia High expression of several tyrosine kinases and activation C Renne´ et al 787 17 Jungnickel B, Staratschek-Jox A, Braüninger A, Spieker T, Wolf J, 34 Uddin S, Hussain A, Al-Hussein K, Platanias LC, Bhatia KG. Diehl V et al. Clonal deleterious mutations in the IkappaBalpha Inhibition of phosphatidylinositol 30-kinase induces preferentially gene in the malignant cells in Hodgkin’s lymphoma. J Exp Med killing of PTEN-null T leukemias through AKT pathway. Biochem 2000; 191: 395–402. Biophys Res Commun 2004; 320: 932–938. 18 Skinnider BF, Elia AJ, Gascoyne RD, Patterson B, Trumper L, Kapp 35 Melzner I, Bucur AJ, Bru¨derlein S, Dorsch K, Hasel C, Barth TF U et al. Signal transducer and activator of transcription 6 is et al. Biallelic mutation of SOCS-1 impairs JAK2 degradation and frequently activated in Hodgkin and Reed-Sternberg cells of sustains phospho-JAK2 action in the MedB-1 mediastinal lympho- Hodgkin lymphoma. Blood 2002; 99: 618–626. ma line. Blood 2005; 105: 2535–2542. 19 Guiter C, Dusanter-Fourt I, Copie-Bergman C, Boulland ML, Le 36 Melzner I, Weniger MA, Bucur AJ, Bru¨derlein S, Dorsch K, Hasel C Gouvello S, Gaulard P et al. Constitutive STAT6 activation in et al. Biallelic deletion within 16p13.13 including SOCS-1 in primary mediastinal large B-cell lymphoma. Blood 2004; 104: Karpas1106P mediastinal B-cell lymphoma line is associated with 543–549. delayed degradation of JAK2 protein. Int J Cancer 2006; 118: 20 Rosenwald A, Wright G, Leroy K, Yu X, Gaulard P, Gascoyne RD 1941–1944. et al. Molecular diagnosis of primary mediastinal B cell lymphoma 37 Zhou YQ, He C, Chen YQ, Wang D, Wang MH. Altered identifies a clinically favorable subgroup of diffuse large B cell expression of the RON receptor tyrosine kinase in primary human lymphoma related to Hodgkin lymphoma. J Exp Med 2003; 198: colorectal adenocarcinomas: generation of different splicing RON 851–862. variants and their oncogenic potential. Oncogene 2003; 22: 21 Savage KJ, Monti S, Kutok JL, Cattoretti G, Neuberg D, De Leval L 186–197. et al. The molecular signature of mediastinal large B-cell 38 Frohling S, Scholl C, Gilliland DG, Levine RL. Genetics of myeloid lymphoma differs from that of other diffuse large B-cell lymphomas malignancies: pathogenetic and clinical implications. J Clin Oncol and shares features with classical Hodgkin lymphoma. Blood 2005; 23: 6285–6295. 2003; 102: 3871–3879. 39 Melzner I, Weniger MA, Menz CK, Mo¨ller P. Absence of the JAK2 22 Renne´ C, Willenbrock K, Ku¨ppers R, Hansmann ML, Braüninger A. V617F activating mutation in classical Hodgkin lymphoma and Autocrine- and paracrine-activated receptor tyrosine kinases in primary mediastinal B-cell lymphoma. Leukemia 2005; 20: classic Hodgkin lymphoma. Blood 2005; 105: 4051–4059. 157–158. 23 Nacheva E, Dyer MJ, Metivier C, Jadayel D, Stranks G, Morilla R 40 Vikkula M, Boon LM, Carraway III KL, Calvert JT, Diamonti AJ, et al. B-cell non-Hodgkin’s lymphoma cell line (Karpas 1106) with Goumnerov B et al. Vascular dysmorphogenesis caused by an complex translocation involving 18q21.3 but lacking BCL2 activating mutation in the receptor tyrosine kinase TIE2. Cell 1996; rearrangement and expression. Blood 1994; 84: 3422–3428. 87: 1181–1190. 24 Mo¨ller P, Bru¨derlein S, Stra¨ter J, Leithaüser F, Hasel C, Bataille F 41 Santoro MM, Penengo L, Minetto M, Orecchia S, Cilli M, Gaudino et al. MedB-1, a human tumor cell line derived from a primary G. Point mutations in the tyrosine kinase domain release the mediastinal large B-cell lymphoma. Int J Cancer 2001; 92: 348– oncogenic and metastatic potential of the Ron receptor. Oncogene 353. 1998; 17: 741–749. 25 Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis 42 Mandell JW. Phosphorylation state-specific antibodies: applica- and display of genome-wide expression patterns. Proc Natl Acad tions in investigative and diagnostic pathology. Am J Pathol 2003; Sci USA 1998; 95: 14863–14868. 163: 1687–1698. 26 Martin-Subero JI, Harder L, Gesk S, Schlegelberger B, Grote W, 43 Mestre C, Rubio-Moscardo F, Rosenwald A, Climent J, Dyer MJ, Martinez-Climent JA et al. Interphase FISH assays for the detection Staudt L et al. Homozygous deletion of SOCS1 in primary of translocations with breakpoints in immunoglobulin light chain mediastinal B-cell lymphoma detected by CGH to BAC micro- loci. Int J Cancer 2002; 98: 470–474. arrays. Leukemia 2005; 19: 1082–1084. 27 Pulford K, Delsol G, Roncador G, Biddolph S, Jones M, Mason DY. 44 Haralambieva E, Jones M, Roncador GM, Cerroni L, Lamant L, Ott Immunohistochemical screening for oncogenic tyrosine kinase G et al. Tyrosine phosphorylation in human lymphomas. activation. J Pathol 1999; 187: 588–593. Histochem J 2002; 34: 545–552. 28 Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell 45 Heinrich MC, Corless CL, Duensing A, McGreevey L, Chen CJ, 2000; 103: 211–225. Joseph N et al. PDGFRA activating mutations in gastrointestinal 29 Danilkovitch A, Donley S, Skeel A, Leonard EJ. Two independent stromal tumors. Science 2003; 299: 708–710. signaling pathways mediate the antiapoptotic action of macro- 46 Rane SG, Reddy EP. Janus kinases: components of multiple phage-stimulating protein on epithelial cells. Mol Cell Biol 2000; signaling pathways. Oncogene 2000; 19: 5662–5679. 20: 2218–2227. 47 Morrison JA, Gulley ML, Pathmanathan R, Raab-Traub N. 30 Kontos CD, Cha EH, York JD, Peters KG. The endothelial receptor Differential signaling pathways are activated in the Epstein-Barr tyrosine kinase Tie1 activates phosphatidylinositol 3-kinase and virus-associated malignancies nasopharyngeal carcinoma and Akt to inhibit apoptosis. Mol Cell Biol 2002; 22: 1704–1713. Hodgkin lymphoma. Cancer Res 2004; 64: 5251–5260. 31 Rommel C, Clarke BA, Zimmermann S, Nunez L, Rossman R, Reid 48 Georgakis GV, Li Y, Rassidakis GZ, Medeiros LJ, Mills GB, Younes K et al. Differentiation stage-specific inhibition of the Raf-MEK-ERK A. Inhibition of the phosphatidylinositol-3 kinase/Akt promotes G1 pathway by Akt. Science 1999; 286: 1738–1741. cell cycle arrest and apoptosis in Hodgkin lymphoma. Br J 32 Zimmermann S, Moelling K. Phosphorylation and regulation of Raf Haematol 2006; 132: 503–511. by Akt (protein kinase B). Science 1999; 286: 1741–1744. 49 Vivanco I, Sawyers CL. The phosphatidylinositol 3-Kinase AKT 33 Dutton A, Reynolds GM, Dawson CW, Young LS, Murray PG. pathway in human cancer. Nat Rev Cancer 2002; 2: 489–501. Constitutive activation of phosphatidyl-inositide 3 kinase con- 50 Ni Z, Lou W, Lee SO, Dhir R, DeMiguel F, Grandis JR et al. tributes to the survival of Hodgkin’s lymphoma cells through a Selective activation of members of the signal transducers and mechanism involving Akt kinase and mTOR. J Pathol 2005; 205: activators of transcription family in prostate carcinoma. J Urol 498–506. 2002; 167: 1859–1862.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia