6796 Vol. 10, 6796–6806, October 15, 2004 Clinical Research

Featured Article Variability in the Degree of Expression of Phosphorylated I␬B␣ in Chronic Lymphocytic Leukemia Cases With Nodal Involvement

Antonia Rodrı´guez,1 Nerea Martı´nez,1 changes in the expression profile (mRNA and ex- Francisca I. Camacho,1 Elena Ruı´z-Ballesteros,2 pression) and clinical outcome in a series of CLL cases with 2 1 lymph node involvement. Activation of NF-␬B, as deter- Patrocinio Algara, Juan-Fernando Garcı´a, ␬ ␣ 3 5 mined by the expression of p-I B , was associated with the Javier Mena´rguez, Toma´s Alvaro, expression of a set of comprising key genes involved in 6 4 Manuel F. Fresno, Fernando Solano, the control of B-cell receptor signaling, signal transduction, Manuela Mollejo,2 Carmen Martin,1 and and , including SYK, LYN, BCL2, CCR7, BTK, Miguel A. Piris1 PIK3CD, and others. Cases with increased expression of ␬ ␣ 1Molecular Pathology Program, Centro Nacional de Investigaciones p-I B showed longer overall survival than cases with lower Oncolo´gicas, Madrid, Spain; 2Department of Genetics and Pathology, expression. A Cox regression model was derived to estimate 3 Hospital Virgen de la Salud, Toledo, Spain; Department of some parameters of prognostic interest: IgVH mutational Pathology, Hospital General Universitario Gregorio Maran˜o´n, ␬ ␣ 4 status, ZAP-70, and p-I B expression. The multivariate Madrid, Spain; Department of Hematology, Hospital Nuestra Sen˜ora analysis disclosed p-I␬B␣ and ZAP-70 expression as inde- del Prado, Talavera de la Reina, Toledo, Spain; 5Department of Pathology, Hospital Verge de la Cinta, Tortosa, Spain; pendent prognostic factors of survival. and 6Department of Pathology, Hospital Central de Asturias, Conclusions: A variable degree of activation of NF-␬B, Oviedo, Spain as determined by the expression of p-I␬B␣, is an identifiable event in CLL, and is correlated with changes in the expres- sion profile and overall survival. ABSTRACT Purpose: Based on previous preliminary observations, we hypothesize that the molecular and clinical variability of INTRODUCTION chronic lymphocytic leukemia (CLL) reflects differences in Chronic lymphocytic leukemia/small lymphocytic lym- the degree of nuclear factor (NF)-␬B activation, as deter- phoma (CLL/SLL) is a lymphoproliferative disorder character- mined by the expression of phosphorylated I␬B␣ (p-I␬B␣). ized by the slow proliferation and accumulation of monoclonal Experimental Design: The expression profile (mRNA mature CD5-positive B lymphocytes in peripheral blood, bone and protein expression) was analyzed with the Centro Na- marrow, lymph nodes, and other tissues. Ninety percent of cional de Investigaciones Oncolo´gicas Oncochip, a cDNA chronic lymphocytic leukemias are CLLs, and this is the most microarray containing 6386 cancer-related genes, and a tis- frequent adult leukemia in Western countries. sue microarray (TMA). The results were correlated with the Although CLL is a disease that is considered to be incur-

IgVH mutational status, ZAP-70 expression, cytogenetic al- able with currently available therapy, its clinical course is het- terations, and clinical outcome. erogeneous; some patients have a more stable disease and die Results: We found correlations between the presence of after many years from unrelated causes, whereas others progress p-I␬B␣, a surrogate marker of NF-␬B activation, and very quickly and die within a few years. This variability has stimulated the search for prognostic markers with which to predict the outcome of patients and to allow treatments to be adapted to the specific risk. In addition to the Rai and Binet Received 4/19/04; revised 7/2/04; accepted 7/9/04. clinical staging systems (1, 2), specific cytogenetic [11q and 17p Grant support: This study was supported by grants from the Comu- deletions, 12 trisomy, 13q14 deletion (3)], molecular events

nidad Auto´noma de Madrid (CAM 08.1/0011/2001.1), the Ministerio de [IgVH mutational index (4–6)], and the expression level of Sanidad y Consumo (FIS 01–0035), and the Ministerio de Ciencia y CD38 and ZAP70 (5–11) have been found to be associated with Tecnologı´a (SAF 2001–0060), Spain. F. Camacho is supported by an Olivia Roddom grant from the Spanish Association for Cancer Research different survival probabilities. (AECC). A. Rodrı´guez is supported by a grant from the Ministerio de The nuclear factor (NF)-␬B protein family, p105/p50, Sanidad y Consumo (FIS 02–9355). p100/p52, Rel A (p65), c-Rel, and Rel B play an important role The costs of publication of this article were defrayed in part by the in differentiation and survival in normal B cells and presumably payment of page charges. This article must therefore be hereby marked also in neoplastic B cells. NF-␬B are present in the advertisement in accordance with 18 U.S.C. Section 1734 solely to ␬ ␣ indicate this fact. cytoplasm bound to I B proteins, which are inhibitory mole- Note: Supplementary data for this article can be found at Clinical cules that sequester NF-␬B dimers in the cytoplasm. After Cancer Research Online (http://clincancerres.aacrjournals.org) B-cell activation and subsequent phosphorylation, ubiquitina- Requests for reprints: Miguel Angel Piris, Molecular Pathology Pro- tion and proteasome degradation of I␬B␣, NF-␬B translocate to gram, Centro Nacional de Investigaciones Oncolo´gicas, C/Melchor Fer- na´ndez Almagro 3, E-28029 Madrid, Spain. Phone: 34-91-224-69-00; the nucleus, making possible the transcription of its target genes. Fax: 34-91-224-69-23; E-mail: [email protected]. NF-␬B activation after induction by CD40-CD154 ligation ©2004 American Association for Cancer Research. has been described in CLL cells (12) and is correlated with

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Clinical Cancer Research 6797

Bcl-2 expression (13). The presence of activated NF-␬B has feature in most lymphoid malignancies. Here we have investi- been shown to be activated by cytokines (14) and regulated by gated the correlation of the presence of phosphorylated I␬B␣ APRIL and BAFF genes (15). Electrophoretic mobility shift (p-I␬B␣), a marker of NF-␬B activation (17, 18), with changes analysis studies have confirmed the presence of NF-␬B activa- in the expression profile and clinical outcome of CLL cases. tion in a subset of CLL cases and have also revealed a subset of To achieve this, we used the Centro Nacional de Investi- cases lacking this (16). Nevertheless, the real incidence of this gaciones Oncolo´gicas (CNIO) Oncochip, a cDNA microarray, phenomenon and its precise features have yet to be fully eluci- containing 6386 cancer-related genes, and a tissue microarray dated. (TMA), containing 96 cores corresponding to 38 cases and Our previous preliminary observations led us to hypothe- controls. To facilitate simultaneous analysis using a cDNA size that the clinical variability of CLL reflects, at least partially, microarray and a TMA, we restricted the analysis to CLL cases differences in the degree of NF-␬B activation of CLL, a cardinal with nodal involvement, selected for being at Rai stage Ն 1.

Fig. 1 Immunohistochemical staining in the TMA. The two columns on the left show nega- tive or low-level expression. Positive or strong expression for each marker is shown in the right-hand columns. A low- and high-power field is illustrated for each stain (original magni- fications of ϫ4 and ϫ40, re- spectively).

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. 6798 Differences in NF-␬B Activation in CLL

Table 1 Description of the expression of the markers analyzed in mean survival-free disease time was 30 months (12–68 months). the TMA Progression with or without treatment occurred in 28 cases, with a Median Range Visual evaluation mean and median progression-free survival of 30 and 39 months, Syk 27.25 0.31–94.96 Strongly heterogeneous cytoplasmic respectively. The mean overall survival (OS) in this series was 52 expression. months (median, 46 months). This was shorter than that of other Lyn 63.05 20.46–82.61 Cytoplasmic expression, moderate series, perhaps because of the disease being more advanced at the variability. time of its diagnosis. Bcl2 54.86 17.35–77.89 Cytoplasmic, strong expression. p-I␬B␣ 2.63 0.01–65.43 Cytoplasmic and nuclear In most cases, CLL was considered the main cause of expression, highly variable. Two death, although four patients developed secondary gastrointes- major groups defined. tinal and breast carcinoma. Ki67 1.79 0.02–9.3 Nuclear expression, stronger in Tissue Microarray and Immunohistochemistry. A proliferation centers. c-Flip 0.02 0–1.24 Surface expression, low intensity. TMA block was constructed with a Tissue Arrayer device (Beecher IgM 51.76 0.26–82.63 Variable cytoplasmic-surface Instruments, MD) with two 1-mm-diameter cylinders from two expression. selected areas of each case and from appropriate controls. Sample cores were obtained from paraffin-embedded lymph node blocks corresponding to the initial diagnosis, which were available in 38 samples of the study. Paraffin sections were examined to select the involved area. The resulting TMA contained 96 cores from cases This meant that we could also use frozen, unmanipulated diag- and controls. Antibodies against Syk (Santa Cruz, CA), Lyn (Santa nostic lymph nodes, the NF-␬B activation status of which more Cruz Biotechnology), BclII (DAKO, Copenhagen, Denmark), closely reflects the in vivo status. These more advanced CLL p-I␬B␣ (Cell Signaling, Beverly, MA), Ki67 (DAKO), FLIP (Ab- cases are of additional interest because they require further cam, Cambridge, United Kingdom), and IgM (DAKO) were used clinical surveillance and early treatment and because auxiliary to determine protein expression. The immunohistochemical proto- prognostic variables could be identified. col was carried out as described previously (21). All hybridized TMAs were scanned with the BLISS system MATERIALS AND METHODS (Bacus Laboratories, Inc, Lombard, IL), which uses a three charge- Patients and Samples. Cases were selected for study coupled device chip for the three colors red, green, and blue sensor based on the availability of frozen lymph node tissue at the optically coupled to a microscope. Tissue microarrays were quan- moment of diagnosis of CLL. Frozen samples were supplied by titatively scored on a scale of 0 to 100 with TMAscore v.1.0 image the Hospital Virgen de la Salud, Toledo; Hospital Gregorio analysis software (Bacus Laboratories, Inc., Lombard, IL), which Maran˜o´n, Madrid; Hospital Doce de Octubre, Madrid; Hospital uses WebSlide v.1.0 (Bacus Laboratories, Inc.) virtual slides. Central de Asturias, Oviedo; Hospital Verge de la Cinta, Tor- ZAP-70 expression (Cell Signaling) was classified as pos- tosa; and the Tumor Bank Network of the CNIO. Diagnosis of itive or negative after visual inspection and comparison with the CLL was based on the National Cancer Institute-sponsored level of expression of reactive T cells. Working Group guidelines and the World Health Organization Microarray Assays. RNA isolation and amplification Classification criteria (19, 20). for microarray experiments were performed as described previ- Lymph node biopsies were done to assess CLL involve- ously (22). Universal Human Reference RNA (Stratagene) was ment at diagnosis. Lymph node tissue was frozen and paraffin- used as a reference. From each amplified RNA product, 2.5 ␮g embedded following standard procedures. Expert pathologists was labeled with cyanine 3-conjugated dUTP (Cy3) in the case did blind evaluations of morphology and immunohistochemistry of the samples and with cyanine 5-conjugated dUTP (Cy5) in previous to entering into the study. Frozen reactive lymph node the case of the reference RNA. samples were used as controls. Hybridizations were done in the CNIO OncoChip (22). Hy- Forty-one cases were found to be suitable for study in accord- bridized slides were scanned with the Scanarray 5000 XL (GSI ance with the established criteria. Clinical data were obtained from Lumonics Kanata, Ontario, Canada) or with the Agilent G2565AA medical records. The series comprised 25 males and 16 females. Microarray Scanner System (Agilent, Palo Alto, CA), and images The mean age at diagnosis was 67 years, with only two patients under 50 (37 and 39) years of age. At the moment of diagnosis, all patients had lymph node involvement, in addition to peripheral blood lymphocytosis corresponding to a stage Ն1 on the Rai scoring system. According to the Binet classification, 10 cases Table 2 Correlation between markers analyzed in the TMA (24%) were stage A, 23 (56%) cases were stage B, five (12%) cases Syk Lyn Bcl2 were stage C, and three (7%) cases could not be assigned a Lyn category. The average follow-up after diagnosis was 52 months, Pearson 0.543 Ͻ with a maximum of 132 months. Criteria to consider progressive P 0.001 Bcl2 disease and to determine need for treatment were based on NCI Pearson 0.394 0.661 guidelines (20). First-line treatment was Chlorambucil Ϯ Predni- P Ͻ0.05 Ͻ0.001 sone in 19 patients, CHOP (cyclophosphamide, doxorubicin, vin- p-I␬B␣ cristine, and prednisone) or CHOP-like regimen in seven patients Pearson 0.627 0.409 0.54 Ͻ Ͻ Ͻ and nucleotide analogous-based therapy in four patients. Their P 0.001 0.05 0.001

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Clinical Cancer Research 6799

Fig. 2 Hierarchical clustering of the CLL series, showing the relative homogeneity of this se- ries. Red indicates high-level expression, and green indicates low-level expression.

were analyzed with the GenePix 4.0 program (Axon Instruments were averaged to the median. Only genes that had data present Inc., Union City, CA). in Ͼ70% of cases were considered in the subsequent analysis. To adjust for differences in the red and green labeling, To obtain the expression profile of each tumor we referred the Cy3/Cy5 ratios were normalized with the Diagnosis and Nor- ratios of the tumors to the mean of ratios of reactive lymph malization of Microarray Data (DNMAD) tool based on stand- nodes from four different patients. ard print-tip loess.7 Cluster analysis was done with the Cluster program To organize patterns on a symmetrical scale, ratios were (23). Student’s t test was used to identify genes differentially log-transformed (base 2), and duplicated spots on the OncoChip expressed in the classes defined by p-I␬B␣ expression.8

7 http://dnmad.bioinfo.cnio.es. 8 http://bioinfo.cnio.es/cgi-bin/tools/multest/multest.cgi.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. 6800 Differences in NF-␬B Activation in CLL

Definition of Functional Signatures. All genes printed Pearson correlation. The significance of associations between on the CNIO Oncochip array were analyzed with the Gene characteristics was determined with Fisher’s exact test. Ontology-based application FATIGO, which is part of the The mean expression for each value was compared to GEPAS suite (24).9 Genes that remained unclassified by identify genes discriminating between groups established on the FATIGO were sought in the GENECARDS database10 and basis of the presence or absence of a condition by Student’s t biological functions assigned. Genes were grouped in different test.13 The associated P value is presented as the adjusted P categories modified in accordance with published observations value derived from procedures that control the false discovery about pathogenesis of B-cell lymphoma as NF-␬B, B-cell re- rate. Genes with adjusted P values using False Discovery Rate ceptor (BCR) signaling and lymph node-homing categories. (FDR)_indep Ͻ0.2 are shown. Functional signatures for cell growth and maintenance (806 Cases were divided into two groups with values of p-I␬B␣ clones), apoptosis (146 clones), NF-␬B pathway (167 clones), greater than and less than the median of the series, respectively. BCR signaling (54 clones), cell communication (808 clones), Cox proportional hazard models were used to evaluate the lymph node-homing (13 clones) and metabolism (1409 clones) possible associations between OS and a number of variables. OS were included here. curves were derived with the Kaplan-Meier method. Statistical significance of associations between individual variables and VH Gene Analysis. Rearranged IgVH genes were ampli- fied with a semi-nested PCR method following standard pub- OS was determined with the log-rank test. Statistical analyses were done with the SPSS (SPSS Inc., lished procedures (25). A mixture of six framework 1 (FR1) VH family-specific primers and two consensus primers for the J Chicago, IL) program and with tools for random permutation H tests.14 gene in the first round of PCR and JH internal primers in the second round were used. The direct-sequencing procedure was done with an ABI RESULTS PRISM 310 or 3700 Genetic Analyzer (Applied Biosystems, Tissue Microarray Data. Captured images of expres- Weiterstadt, Germany) following the manufacturer’s procedure. sion of Syk, Lyn, Bcl2, p-I␬B␣, Ki67, c-FLIP, and IgM are Mutations were identified by comparison with the germline shown in Fig. 1. Cases were divided into two groups with sequence (immunoglobulin BLAST11 and V BASE sequence respect to the median value of protein detection in the whole ␬ ␣ directories12). series (Table 1). For p-I B , this threshold indeed separates two A sequence was considered mutated when there was Ͻ98% groups exhibiting distinct low and high levels of expression. Levels of expression for the markers showed that p-I␬B␣ homology with the closest germline VH genes. Fluorescence In situ Hybridization Analysis. Fluores- expression was closely correlated with the level of expression of cence in situ hybridization analysis was done with LSI p53/LSI Syk, Lyn, and Bcl2. Details of the results are given in Table 2. Other significant associations are also described in this table. ATM and LSI D13S319/LSI 13q34/CEP 12 Multi-color Probe cDNA Microarray Data. Hierarchical clustering (Stan- Sets, from Vysis (Downers Grove, IL). LSI p53 (17p13.1) ford University Cluster) of 41 CLL samples and four reactive covers the entire p53 gene, LSI D13S319, detects the 13q14.3 lymph nodes analyzed by cDNA microarrays revealed relative region, LSI 13q34 detects a region near the subtelomere of homogeneity of all of the CLL cases. Reactive lymph node cases 13q, and CEP 12 detects the ␣ satellite, centro- were grouped together (Fig. 2). The reproducibility of the tech- meric region of . LSI ATM detects the 11q22.3 nique was confirmed by repeating the hybridization of four region of chromosome 11, which includes the entire ataxia samples and finding that parallel hybridizations clustered to- telangiectasia-mutated gene (ATM). gether (supplementary information). Hybridizations on the tissue microarray were done as de- Two-fold over- or underexpressed clones with respect to scribed in Moreno-Bueno et al. (26). The cutoff value for reactive lymph nodes in at least 50% of CLL cases were con- deletions was taken as the mean plus three times the SD of the sidered to make up the CLL signature. This includes 213 genes, proportion of cells with one signal in the three normal non- comprising 114 down-regulated and 99 up-regulated genes. tumoral samples. This value corresponded to 25% of cells. Genes were grouped according to functional signatures, paying Tumor samples with Ͼ25% of cells with one signal were particular attention to the functional clusters of NF-␬B, BCR considered to be subject to genomic loss involving the targeted signaling and lymph-node homing. The results of the expression genomic region. The cutoff value for trisomy was established profiling of these gene sets are detailed in Tables 3 and 4. with the same three normal nontumoral samples. Gain or am- Relevant up-regulated genes and proteins related with plification was defined as the presence (in Ͼ5% of tumor cells) BCR- and NF-␬B-signaling included LYN, BTK, CCR7, BCL2, of three distinctive signals. BAFF and other genes. CCR7, a well characterized NF-␬B Statistical Analysis. Relationships between the levels of target (27), is a major receptor providing the triggering signals expression of the different markers were explored with the for B-cell entry into lymph nodes (28), which is consistent with the fact that the CLL cells analyzed here were derived from involved lymph nodes.

9 http://fatigo.bioinfo.cnio.es. 10 http://bioinformatics.weizmann.ac.il/cards. 11 http://www.ncbi.nlm.nih.gov/igblast. 13 http://bioinfo.cnio.es/cgi-bin/tools/multest/multest.cgi. 12 http://www.mrc-cpe.cam.ac.uk/vbase-ok. 14 http://bioinformatica.cnio.es/.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Clinical Cancer Research 6801

Table 3 CLL signature genes divided into functional groups activated lymphocytes to lymph nodes and resulted in their Expression preferential migration to the spleen (35). Function % ratio Genes ZAP-70 Expression. ZAP-70 could be evaluated in 35 NF-␬B5Ͻ2-fold 6 cases of the complete series. Expression was positive in 20 Ͼ2-fold 6 (57%) cases and negative in the remaining 15 (43%) cases. In BCR signaling 3 Ͻ2-fold 2 the light of recent results suggesting that this expression is a Ͼ 2-fold 4 surrogate of IgV mutational status (8, 10, 36), we investigated Apoptosis 7 Ͻ2-fold 9 H Ͼ2-fold 6 this correlation in our series. We found that 15 of the 20 positive Cell growth and maintenance 14 Ͻ2-fold 19 cases, but only three of the15 negative cases, belonged to the Ͼ2-fold 11 unmutated group. This association is significant (P Ͻ 0.05). Ͻ Metabolism 28 2-fold 36 Mutational Status. DNA was obtained from 40 cases of Ͼ2-fold 24 Lymph-node homing 2 Ͻ2-fold 3 the complete series. IgVH gene rearrangement was identified in Ͼ2-fold 2 36 patients. Forty-seven sequences were amplified, with double Cell communication 27 Ͻ2-fold 37 or triple rearrangement in seven and two cases, respectively. Ͼ 2-fold 21 Only four of the nine multiple rearrangements showed differ- Other or unknown function 14 Ͻ2-fold 19 Ͼ2-fold 14 ences in mutational status with respect to VH family usage. The spectrum of V family usage was as follows: V 1 and V 3 NOTE. Functional groups are defined by and level H H H of deregulation (Ͼ2-fold up- or down-regulated in at least 50% of both in 16 cases, VH 4 in 11 cases, VH 2 in 2 cases, and VH 5 samples). Some genes were found in two or more groups. and VH 6 both in 1 case. Distribution of different subfamily usage was similar to the described, with a more frequent repre- Ͼ sentation of VH1.69 (10 sequences) (37, 38). Taking a 2% Lyn is a member of the Src-family protein tyrosine kinases variability with respect to the germline VH sequence, 18 IgVH that, along with Blk and Fyn, play an essential role in pre-B-cell sequences from 17 patients were considered as mutated, and 29 receptor-mediated NF-␬B activation and B-cell development sequences from 19 patients had a low mutational index. Per- (29). centages of mutated (48%) and unmutated (52%) cases were Btk has been shown to couple I␬B kinase, I␬B␣, and calculated based on 36 rearrangements. All cases with VH1.69 NF-␬B to the BCR, linking these key pathways (30); addition- subfamily usage were unmutated. ally, BTK directly regulates bcl-x expression by transcriptional FISH Analysis. To identify chromosomal abnormalities control in response to BCR engagement (31). and assess their involvement in the findings in this series, we did Increased Bcl2 expression was also observed in this series, FISH analysis on TMA cores, thereby investigating deletions in confirming previous observations (32) that linked Bcl2 expres- 11q22 (ATM), 17p13 (p53), 13q14 regions, and 12 trisomy. The sion to NF-␬B activation in CLL cells (33). distribution of chromosomal abnormalities was as follows: BAFF, a member of the tumor necrosis factor superfamily 11q22 deletion was present in five cases of 32 assessable cores is involved in normal and leukemic B-cell survival and differ- (15%); only one case featured deletion of 17p13 out of 30 entiation. BAFF binds to B cells and through the B-cell co- assessable cores (3%); 13q14 deletion was found in four of 28 receptor complex promotes NF-␬B activation (15, 34). assessable cores (14%); finally, only one of the 32 assessable Genes found to be up-regulated in the lymph node-homing cases was positive for trisomy of chromosome 12 (3%). Two group included CCR7 and SELL. SELL (L-selectin, CD62L) is a cases had double chromosomal abnormalities, both with 11q22 cell surface adhesion protein that mediates the adherence of and 13q14 deletions. All cases with 11q22 deletion were from lymphocytes to endothelial cells of high endothelial venules in the high-level p-I␬B␣ expression group. peripheral lymph nodes, the expression of which was found to Correlation with p-I␬B␣. CLL cases were divided into be increased in a subset of CLL patients (32). Interestingly, loss two groups according to the median value (2.63) of p-I␬B␣ of L-selectin from the cell surface inhibited the migration of protein detection in the whole series. Employing previously

Table 4 Expression-profiling genes upregulated in NF-␬B and BCR groups Gene symbol Gene description N-fold (average) NF-␬B genes BCL2 B-cell CLL/lymphoma 2 3.091 CCR7 chemokine (C-C motif) receptor 7 2.715 LYN v-yes-1 Yamaguchi sarcoma viral-related oncogene homolog 2.495 TLE1 transducin-like enhancer of split 1, homolog of Drosophila E(sp1) 2.269 IL1A interleukin 1, alpha 1.291 BAFF B-cell activating factor of the TNF family 0.9 BCR genes BCL2 B-cell CLL/lymphoma 2 3.091 BTK Bruton agammaglobulinemia tyrosine kinase 2.846 LYN v-yes-1 Yamaguchi sarcoma viral-related oncogene homolog 2.495 GAB1 GRB2-associated binding protein 1 1.967

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. 6802 Differences in NF-␬B Activation in CLL

defined functional signatures, Student’s t test was used to iden- IGF1 and its receptor, IGF1-R, are key regulators of cell tify genes that were differentially expressed in these two groups. growth and development. After activation, they induce a cas- In total, 894 genes with the best adjusted P value (see Material cade of tyrosine phosphorylations, leading to the eventual acti- and Methods) were differentially expressed (Fig. 3A and B). The vation of Akt and up-regulation of the antiapoptotic protein most characteristic genes of the high-level p-I␬B␣ expression Bcl-xL (40). group are listed in Tables 5 and 6. FAF1 is a member of the death-inducing signaling complex Among the genes to be deregulated in the cases with in Fas-mediated apoptosis and has also been shown to be able to NF-␬B activation, some of the most noticeable were PIK3CD, play a role as an NF-␬B activity suppressor through cytoplasmic LYN, IRAK1, IGF1, FAF1, BAFF, RBBP2, HDAC6, CDC16, retention of p65, via physical interaction (41). PMSCL2, PPAP2B, SELL, and MALT1. BAFF expression was greater in cases with NF-␬B activa- PIK3CD codes for the p110␦ catalytic subunit of the phos- tion, thereby confirming the proposed role of BAFF as a pro- phoinositide 3-kinase (PI3K) family, preferentially expressed by moter of NF-␬B activation in leukemic B-cells (15, 34). leukocytes, which plays a role, after antigen-receptor triggering It is of particular note that the increased expression of RBBP2 and SYK activation, in protein kinase C-␤- and AKT/PKB- and HDAC6, a binding partner for NF-␬B p50 and p65 (42), dependent B-cell survival (39). associates NF-␬B activation in CLL with histone acetylation.

Fig. 3 Student’s t test analysis. A, differentially expressed genes identified in cases with high ver- sus low levels of expression of phosphorylated I␬B␣. B, detail showing the most closely asso- ciated genes with highest P value within each group. Red indicates high-level expression, green indicates low-level ex- pression, and missing values are in gray.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Clinical Cancer Research 6803

In contrast, cases with a low level of NF-␬B activation to estimate a number of variables of potential prognostic inter- ␬ ␣ express CUL3, PTH, CARD12, TRAF3, MADH4, GAD2, est, such as IgVH mutational status, ZAP-70, and p-I B ex- DNAI1, UBE2D3, GTF2B, HEAB, POL2RG, G3BP, and ARG1. pression. This multivariate analysis showed ZAP-70 expression These findings suggest abnormalities in the ubiquitination and and p-I␬B␣ to be independent prognostic factors of survival protein degradation in this group of cases. (P Ͻ 0.01; Table 7). ␬ ␣ Associations of p-I B expression with VH mutational status and chromosomal abnormalities were sought using Fish- er’s exact test. There was no substantial relationship between the DISCUSSION ␬ ␣ ␬ level of p-I B expression and IgVH mutational index, although Results from our study show that activation of NF- B, as we observed that high levels of p-I␬B␣ were associated with determined by the expression of p-I␬B␣, is a relatively frequent 11q22 (ATM) deletion (P Ͻ 0.05). phenomenon in CLL with lymph node involvement. The inten- Survival Analysis. Using the Kaplan-Meier method to sity of this expression parallels NF-␬B activation and oscillates assess the prognostic importance of p-I␬B␣ in the OS of CLL, along the series. Additionally, the expression of p-I␬B␣ is we found that cases with a high level of expression had better linked with the expression of a large set of key genes involved survival probability than cases with a low level of expression in the control of B-cell signaling, signal transduction and apo- (P Ͻ 0.05) (Fig. 4). The five-year OS probability was 40 and ptosis. 60% in the low- and high-level p-I␬B␣ expression groups, CLL cells are capable of responding to external cell pro- respectively. liferation signals, such as microenvironmental stimuli fostering

Kaplan-Meier analysis of IgVH mutational status revealed CD40 stimulation (43), and anti-apoptotic signals after antigen- a trend toward longer survival in the mutated group. No relation receptor triggering. BCR complex is formed by surface IgM that was found between chromosomal abnormalities and OS in this is non-covalently associated with CD79a/CD79b signal trans- series. duction units. Following sIg cross-linking, it activates the BCR A Cox proportional hazards regression model was derived signaling pathway downstream. In normal B-lymphocytes, this

Fig. 3 Continued

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. 6804 Differences in NF-␬B Activation in CLL

Table 5 Differentially expressed genes in cases with high This study also underlines the relevance of the PI3K path- p-I␬B␣ expression way in the molecular pathogenesis of CLL cells. Although the Expression molecular details of the recruitment, activation, and signaling of Function Total genes ratio (median) Genes PI3Ks by the BCR are not fully understood, it is clear that they NF-␬B33Ͻ2-fold 1 are at least partially regulated by BCR- and Syk-dependent Ͼ2-fold 3 phosphorylation of CD19 and other B-cell PI3K adaptor mole- BCR signaling 12 Ͻ2-fold 0 cules (39). Here, the results highlight a simultaneous expression Ͼ 2-fold 3 of a large set of the key molecules in this process in NF-␬B Apoptosis 44 Ͻ2-fold 2 Ͼ2-fold 2 activated CLL cases, including SYK, IRAK1, IL1, PIK3CD, and Cell growth and maintenance 179 Ͻ2-fold 6 PIK3CB. Studies at the protein level have confirmed these Ͼ2-fold 5 findings; these observations also extend to the expression of Ͻ Metabolism 415 2-fold 13 BclII and Lyn. Ͼ2-fold 14 Lymph node-homing 4 Ͻ2-fold 1 In this study, Bcl2 protein expression level was closely Ͼ2-fold 1 related with p-I␬B␣, Syk, and other molecules involved in the Cell communication 177 Ͻ2-fold 7 BCR signaling that leads to NF-␬B activation. Indeed, the high Ͼ2-fold 9 level of BclII protein expression is routinely observed in the Ͻ Other or unknown function 288 2-fold 13 diagnosis of small lymphocytic neoplasms and may be seen in Ͼ2-fold 12 bone marrow and lymph node examinations. Thus, increased NOTE. Genes are divided into functional groups, as defined by Bcl2 protein expression is a common and substantial finding in Gene Ontology. Genes are considered up- or down-regulated if the median was Ͼ2-fold or Ͻ2-fold, respectively. the major types of small lymphocytic neoplasms, including mantle cell lymphoma, follicular lymphoma, and CLL. The routes leading to Bcl2 expression, nevertheless, differ in distinct neoplasms (21). Table 6 NF-␬B genes associated with high level of ␬ p-I␬B␣ expression Variation of NF- B activation in CLL is not an entirely novel observation, because electrophoretic mobility shift anal- Gene N-fold ysis data showing variability in tumoral cells and cell lines have symbol Gene description (average) been already reported (16). Phosphorylated I␬B␣ has been used IRAK1 Interleukin-1 receptor-associated kinase 1 1.519 as a surrogate for the activation of NF-␬B, because it constitutes PAWR PRKC, apoptosis, WT1, regulator 1.295 ␬ IGF1 Insulin-like growth factor 1 (somatomedin C) 1.125 a key step in the release of NF- B subunits and eventual nuclear BAFF B-cell activating factor of the TNF family 1.250 translocation (17, 18). Moreover, a variable response to BCR FAF1 Fas (TNFRSF6)-associated factor 1 0.998 signaling in CLL cells has been reported previously (33, 50, 51) and linked to the presence of decreased levels of functionally relevant adhesion molecules, cell-signaling receptors, and 11q22-q23 deletion (16). Gene expression profiling studies (52,

leads to proliferation, differentiation, apoptosis, survival, and 53) have also showed the presence of an IgVH mutational status immune tolerance, depending on the B-cell developmental stage and cell environment. Molecules downstream of BCR include members of the Src kinase family (Lyn, Fyn, and Blk; ref. 29), the Tec kinase family (Btk; ref. 44), Syk tyrosine kinase (45, 46), members of the protein kinase C family of serine-threonine kinases (29), PI3K; ref. 39), the mitogen-activated protein ki- nase family (45), the antiapoptotic NF-␬B complex (29), and ZAP-70 (47) in the case of CLL cells. Activation of B-cell co-receptor complex (CD21, CD19 and CD81) by the BAFF gene (34) and tumor necrosis factor receptor CD40-CD154 interactions amplify BCR signaling (48). This study, including the mRNA and protein expression level of a majority of these genes, suggests the possible signif- icance of BCR signaling in the activation of NF-␬B and high- lights the expression of a selected number of genes playing a critical role in this process, including SYK, IRAK1, and PI3K. Interestingly, this phenomenon seems to involve preferentially a group of CLL cases, thus further increasing the biological het- erogeneity of CLL. IRAK1 is a key TRAF6-interacting protein that mediates NF-␬B activation downstream of tumor necrosis factor and Toll/IL-1R-containing receptors (49). In fact, in this CLL series, the expression of both IRAK1 and IL1 appears to be closely Fig. 4 Kaplan-Meier curve showing association between p-I␬B␣ ex- related to that of phosphorylated I␬B␣. pression and OS.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Clinical Cancer Research 6805

Table 7 Multivariate Cox hazard ratios of Zap-70 and p-I␬B␣ 8. Wiestner A, Rosenwald A, Barry TS, et al. ZAP-70 expression expression identifies a chronic lymphocytic leukemia subtype with unmutated immunoglobulin genes, inferior clinical outcome, and distinct gene Hazard ratio 95% CI P expression profile. Blood 2003;101:4944–51. Ͻ Zap-70 6.45 1.84–22.64 0.01 9. Krober A, Seiler T, Benner A, et al. V(H) mutation status, CD38 ␬ ␣ Ͻ p-I B 10.3 2.56–41.5 0.01 expression level, genomic aberrations, and survival in chronic lympho- cytic leukemia. Blood 2002;100:1410–6. 10. Crespo M, Bosch F, Villamor N, et al. ZAP-70 expression as a gene signature, composed of genes that have various expres- surrogate for immunoglobulin-variable-region mutations in chronic lymphocytic leukemia. N Engl J Med 2003;348:1764–75. sions among mutated and unmutated cases, thus underlining the 11. Durig J, Nuckel H, Cremer M, et al. ZAP-70 expression is a relevance of BCR signaling to the generation of the cellular prognostic factor in chronic lymphocytic leukemia. Leukemia 2003;17: expression profile. 2426–34. The data from this study reveal the variable degree of 12. Furman RR, Asgary Z, Mascarenhas JO, Liou HC, Schattner EJ. expression of p-IkB␣, a surrogate of NF-␬B activation, at least Modulation of NF-kappa B activity and apoptosis in chronic lympho- in CLL cases with nodal involvement, linking this heterogeneity cytic leukemia B cells. J Immunol 2000;164:2200–6. with the expression of genes involved in the BCR signaling and 13. Viatour P, Bentires-Alj M, Chariot A, et al. NF- kappa B2/p100 NF-␬B activation. Whether this finding is also relevant in CLL induces Bcl-2 expression. Leukemia 2003;17:1349–56. cases with early clinical stages remains to be clarified. 14. Zaninoni A, Imperiali FG, Pasquini C, Zanella A, Barcellini W. If this finding is confirmed in larger CLL series with early Cytokine modulation of nuclear factor-kappaB activity in B-chronic lymphocytic leukemia. Exp Hematol 2003;31:185–90. or advanced stages, p-I␬B␣ expression could be added to the list 15. Kern C, Cornuel JF, Billard C, et al. Involvement of BAFF and of biological variables predicting clinical behavior that already APRIL in the resistance to apoptosis of B-CLL through an autocrine includes IgVH somatic mutation, ZAP-70 expression, and other pathway. Blood 2004;103:679–88. cytogenetic alterations. Data here obtained do not provide an 16. Sembries S, Pahl H, Stilgenbauer S, Dohner H, Schriever F. Re- explanation for the more adverse clinical course in cases lacking duced expression of adhesion molecules and cell signaling receptors by NF-␬B activation; a reason for this phenomenon would be given chronic lymphocytic leukemia cells with 11q deletion. Blood 1999;93: by the presence of alternative cytogenetic or molecular alter- 624–31. ations. This contrasts with the situation observed in multiple 17. Sun Q, Matta H, Chaudhary PM. The human herpes virus 8-en- myeloma, where an increased NF-␬B activity is associated with coded viral FLICE inhibitory protein protects against growth factor withdrawal-induced apoptosis via NF-kappa B activation. Blood 2003; enhanced tumor cell survival (54). 101:1956–61. The study has also identified a large set of molecules that 18. Liu L, Eby MT, Rathore N, Sinha SK, Kumar A, Chaudhary PM. could be functionally assayed in the future with the aim of The human herpes virus 8-encoded viral FLICE inhibitory protein identifying therapeutic targets. physically associates with and persistently activates the Ikappa B kinase complex. J Biol Chem 2002;277:13745–51. ACKNOWLEDGMENTS 19. Jaffe ES Harris NL, Stein H, Vardiman JW. Pathology and genetics of tumours of haematopoietic and lymphoid tissues. World We thank the CNIO Tumor Bank for providing most of the samples health organization classification of tumours. Lyon, Fance: IARC analyzed. The authors acknowledge the technical assistance of the Press; 2001. Genomics and Histology Units and the Molecular Pathology Program. 20. Cheson BD, Bennett JM, Grever M, et al. National Cancer Institute- We also thank Juan C. Cigudosa for carrying out FISH analysis and sponsored Working Group guidelines for chronic lymphocytic leuke- Ramon Diaz for statistical advice. mia: revised guidelines for diagnosis and treatment. Blood 1996;87: 4990–7. REFERENCES 21. Martinez N, Camacho FI, Algara P, et al. The molecular signature 1. Binet JL, Lepoprier M, Dighiero G, et al. A clinical staging system of mantle cell lymphoma reveals multiple signals favoring cell survival. for chronic lymphocytic leukemia: prognostic significance. Cancer Cancer Res 2003;63:8226–32. (Phila) 1977;40:855–64. 22. Tracey L, Villuendas R, Ortiz P, et al. Identification of genes 2. Rai KR, Montserrat E. Prognostic factors in chronic lymphocytic involved in resistance to interferon-alpha in cutaneous T-cell lym- leukemia. Semin Hematol 1987;24:252–6. phoma. Am J Pathol 2002;161:1825–37. 3. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and 23. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis survival in chronic lymphocytic leukemia. N Engl J Med 2000;343: and display of genome-wide expression patterns. Proc Natl Acad Sci 1910–6. USA 1998;95:14863–8. 4. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Un- 24. Herrero J, Al-Shahrour F, Diaz-Uriarte R, et al. GEPAS: a web- mutated Ig V(H) genes are associated with a more aggressive form of based resource for microarray gene expression data analysis. Nucleic chronic lymphocytic leukemia. Blood 1999;94:1848–54. Acids Res 2003;31:3461–7. 5. Hamblin TJ, Orchard JA, Ibbotson RE, et al. CD38 expression and 25. Algara P, Mateo MS, Sanchez-Beato M, et al. Analysis of the immunoglobulin variable region mutations are independent prognostic IgV(H) somatic mutations in splenic marginal zone lymphoma defines variables in chronic lymphocytic leukemia, but CD38 expression may a group of unmutated cases with frequent 7q deletion and adverse vary during the course of the disease. Blood 2002;99:1023–9. clinical course. Blood 2002;99:1299–304. 6. Damle RN, Wasil T, Fais F, et al. Ig V gene mutation status and 26. Moreno-Bueno G, Sanchez-Estevez C, Cassia R, et al. Differential CD38 expression as novel prognostic indicators in chronic lymphocytic gene expression profile in endometrioid and nonendometrioid endome- leukemia. Blood 1999;94:1840–7. trial carcinoma: STK15 is frequently overexpressed and amplified in 7. Ibrahim S, Keating M, Do KA, et al. CD38 expression as an impor- nonendometrioid carcinomas. Cancer Res 2003;63:5697–702. tant prognostic factor in B-cell chronic lymphocytic leukemia. Blood 27. Mathas S, Lietz A, Janz M, et al. Inhibition of NF-kappaB essen- 2001;98:181–6. tially contributes to arsenic-induced apoptosis. Blood 2003;102:1028–34.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. 6806 Differences in NF-␬B Activation in CLL

28. Okada T, Ngo VN, Ekland EH, et al. Chemokine requirements for 42. Zhang W, Kone BC. NF-kappaB inhibits transcription of the H(ϩ)- B cell entry to lymph nodes and Peyer’s patches. J Exp Med 2002;196: K(ϩ)-ATPase alpha(2)-subunit gene: role of histone deacetylases. Am J 65–75. Physiol Renal Physiol 2002;283:F904–11. 29. Saijo K, Schmedt C, Su IH, et al. Essential role of Src-family 43. Granziero L, Ghia P, Circosta P, et al. Survivin is expressed on protein tyrosine kinases in NF-kappaB activation during B cell devel- CD40 stimulation and interfaces proliferation and apoptosis in B-cell opment. Nat Immunol 2003;4:274–9. chronic lymphocytic leukemia. Blood 2001;97:2777–83. 30. Petro JB, Khan WN. Phospholipase C-gamma 2 couples Bruton’s 44. Khan WN. Regulation of B lymphocyte development and activation tyrosine kinase to the NF-kappaB signaling pathway in B lymphocytes. by Bruton’s tyrosine kinase. Immunol Res 2001;23:147–56. J Biol Chem 2001;276:1715–9. 45. Kawauchi K, Ogasawara T, Yasuyama M. Activation of extracel- 31. Petro JB, Castro I, Lowe J, Khan WN. Bruton’s tyrosine kinase lular signal-regulated kinase through B-cell antigen receptor in B-cell targets NF-kappaB to the bcl-x promoter via a mechanism involving chronic lymphocytic leukemia. Int J Hematol 2002;75: 508–13. phospholipase C-gamma2 following B cell antigen receptor engage- 46. Lanham S, Hamblin T, Oscier D, Ibbotson R, Stevenson F, Pack- ment. FEBS Lett 2002;532:57–60. ham G. Differential signaling via surface IgM is associated with VH 32. Korz C, Pscherer A, Benner A, et al. Evidence for distinct patho- gene mutational status and CD38 expression in chronic lymphocytic mechanisms in B-cell chronic lymphocytic leukemia and mantle cell leukemia. Blood 2003;101:1087–93. lymphoma by quantitative expression analysis of cell cycle and apopto- 47. Chen L, Widhopf G, Huynh L, et al. Expression of ZAP-70 is sis-associated genes. Blood 2002;99:4554–61. associated with increased B-cell receptor signaling in chronic lympho- 33. Bernal A. Survival of leukemic B cells promoted by engagement of cytic leukemia. Blood 2002;100:4609–14. the antigen receptor. Blood 2001;98:3050–7. 48. Haxhinasto SA, Bishop GA. Synergistic B cell activation by CD40 34. Hase H, Kanno Y, Kojima M, et al. BAFF/BLyS can potentiate and the B cell antigen receptor: Role of BCR-mediated kinase activation B-cell selection with the B-cell co-receptor complex. Blood 2003. and TRAF regulation. J Biol Chem 2003. 35. Venturi GM, Tu L, Kadono T, et al. Leukocyte migration is regu- 49. Suzuki N, Suzuki S, Duncan GS, et al. Severe impairment of lated by L-selectin endoproteolytic release. Immunity 2003;19:713–24. interleukin-1 and Toll-like receptor signalling in mice lacking IRAK-4. 36. Orchard JA, Ibbotson RE, Davis Z, et al. ZAP-70 expression and Nature (Lond) 2002;416:750–6. prognosis in chronic lymphocytic leukaemia. Lancet 2004;363:105–11. 50. Zupo S, Cutrona G, Mangiola M, Ferrarini M. Role of surface IgM 37. Fais F, Ghiotto F, Hashimoto S, et al. Chronic lymphocytic leuke- and IgD on survival of the cells from B-cell chronic lymphocytic mia B cells express restricted sets of mutated and unmutated antigen leukemia. Blood 2002;99:2277–8. receptors. J Clin Investig 1998;102:1515–25. 51. Hivroz C, Geny B, Brouet JC, Grillot-Courvalin C. Altered 38. Kipps TJ, Tomhave E, Pratt LF, Duffy S, Chen PP, Carson DA. signal transduction secondary to surface IgM cross-linking on B- Developmentally restricted immunoglobulin heavy chain variable re- chronic lymphocytic leukemia cells. Differential activation of the gion gene expressed at high frequency in chronic lymphocytic leukemia. phosphatidylinositol-specific phospholipase C. J Immunol 1990;144: Proc Natl Acad Sci USA 1989;86:5913–7. 2351–8. 39. Okkenhaug K, Vanhaesebroeck B. PI3K in lymphocyte develop- 52. Rosenwald A, Alizadeh AA, Widhopf G, et al. Relation of gene ment, differentiation and activation. Nat Rev Immunol 2003;3:317–30. expression phenotype to immunoglobulin mutation genotype in B cell 40. Sekharam M, Zhao H, Sun M, et al. Insulin-like growth factor 1 chronic lymphocytic leukemia. J Exp Med 2001;194:1639–47. receptor enhances invasion and induces resistance to apoptosis of colon 53. Klein U, Tu Y, Stolovitzky GA, et al. Gene expression profiling of cancer cells through the Akt/Bcl-x(L) pathway. Cancer Res 2003;63: B cell chronic lymphocytic leukemia reveals a homogeneous phenotype 7708–16. related to memory B cells. J Exp Med 2001;194:1625–38. 41. Park MY, Jang HD, Lee SY, Lee KJ, Kim E. Fas associated factor 54. Berenson JR, Ma HM, Vescio R. The role of nuclear factor-kappaB 1 inhibits NF-kB activity by interfering with nuclear translocation of in the biology and treatment of multiple myeloma. Semin Oncol 2001; RelA (p65) subunit of NF-kB. J Biol Chem 2003. 28:626–33.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research. Variability in the Degree of Expression of Phosphorylated Iκ Bα in Chronic Lymphocytic Leukemia Cases With Nodal Involvement

Antonia Rodríguez, Nerea Martínez, Francisca I. Camacho, et al.

Clin Cancer Res 2004;10:6796-6806.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/20/6796

Cited articles This article cites 49 articles, 30 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/20/6796.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/20/6796.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/20/6796. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on October 2, 2021. © 2004 American Association for Cancer Research.