Expression profiling of B cell chronic lymphocytic leukemia suggests ...

Leukemia (2002) 16, 2429–2437  2002 Nature Publishing Group All rights reserved 0887-6924/02 $25.00 www.nature.com/leu

Expression profiling of B cell chronic lymphocytic leukemia suggests deficient CD1mediated immunity, polarized cytokine response, altered adhesion and increased intracellular protein transport and processing of leukemic cells Z Zheng1, S Venkatapathy2, G Rao2 and CA Harrington2,3 1

Division of Hematology, Stanford University School of Medicine, Stanford, CA, USA; and 2Affymetrix, Inc, Santa Clara, CA, USA

We used oligonucleotide microarrays to profile the expression of chronic lymphocytic leukemia (CLL) B cells from eight patients compared with CD5-expressing normal B cells from four donors and with pooled normal circulating B cells. Of 6790 genes examined, we identified 87 genes that were differentially expressed at least two-fold between CLL and the normal B cells. CLL cells significantly down-regulated transcripts from CD1c and CD1d genes, which encode proteins known to present lipid antigen and mediate innate and adaptive immunity. The expression pattern was also consistent with reduced signaling by interferon gamma but increased response to interleukin 4 in leukemic cells. CLL cells increased the expression of several collagen-associated extracellular matrix and adhesion molecules, up-regulated many genes involved in intracellular protein transport and processing, while downregulating genes involved in proliferation and metabolism. Based on the expression pattern, we propose that CLL-B cells prolong their survival through increased interaction with survival factors such as IL-4, and through various mechanisms of evading the immune response, such as turning off the expression of CD1c and CD1d, reducing immunogenic response to interferon gamma, inactivating T cell in B–T interaction and increasing the expression of immunoglobulin receptors which neutralize antibody-dependent cell-mediated cytotoxicity. Leukemia (2002) 16, 2429–2437. doi:10.1038/sj.leu.2402711 Keywords: CLL; microarray; expression profiling

Introduction B cell chronic leukemia is the most common leukemia in the western world, with approximately 10 000 new cases diagnosed each year in the United States.1 It is characterized by the relentless accumulation in the blood, marrow and lymphoid tissues of mature monoclonal B lymphocytes that express the CD5 surface molecule. The leukemic CLL B cell accumulation appears to be a result of defective apoptosis rather than uncontrolled proliferation.2 CLL patients suffer immune deficiencies due to impaired humoral immunity characterized by hypogammaglobulinemia. The leukemic B cells express very low levels of surface IgM or IgD, and upon antigen stimulation, are unable to undergo somatic hypermutation and isotype switching. In addition, CLL patients frequently develop autoimmune diseases, suggesting immune dysregulation.3 Unlike many forms of chronic and acute leukemia, CLL lacks recurrent reciprocal chromosomal translocations and distinctive patterns of oncogene or tumor suppressor gene expression. As a result, it has been difficult to identify the molecular events that lead to the disease phenotypes. CLL-B cells

Correspondence: Z Zheng at the present address: Amersham Biosciences, 928 E. Arques Ave, Sunnyvale, CA 94085, USA; Fax: (408) 773-8343; e.mail: [email protected] 3 Present address: Vaccine and Gene Therapy Institute, Oregon Health and Science University, Portland, OR 97006, USA Received 17 February 2002; accepted 26 June 2002

are thought to derive from CD5+ expressing B cells.2,4 CD5expressing normal B cells are a small population of B cells localized at the edge of the germinal center within the mantle zone of the secondary lymphoid follicles.5 They are considered a good candidate for the normal counterpart of CLL cells because they share several characteristics with the malignant CD5+ CLL cells. These include the propensity to produce polyreactive, low-affinity IgM autoantibodies and the expression of the cross-reactive idiotypes in the antibody produced.2,6–8 Recently, human memory B cells have been proposed as the normal counterpart of CLL-B cells.9 However, this alternative model has difficulty accounting for the above features of CLL-B, and the large percentage of CLL cases with leukemic cells having unmutated immunoglubulin gene would argue against an origin of post-germinal center stage cells such as the memory cells. The ideal normal counterpart of CLL will likely remain a subject of discussion. At present, the CD5+ B cell remains a reasonable candidate as the normal counterpart of CLL and has been used as control cells in many studies of CLL. One approach to understanding the biological basis of CLL pathophysiology would be to first correlate abnormal gene expression with the disease phenotypes. Recent advances in microarray technology allow for comprehensive profiling of human cancer cells. This technique has been applied successfully in cancer classifications and diagnosis,10,11 and in gaining insights into functional pathways.12–14 Although expression profiling establishes only a correlative rather than causative role of identified genes in the biological process of the disease studied, experience has shown that the unique and valuable global views of gene expression patterns, when coupled with available knowledge about gene functions, provide the basis for formulating testable hypotheses about the disease pathways.13,14 Microarray profiling of hematological malignancy for the purpose of exploring disease pathways instead of disease classification has been reported recently in a study of mantle cell lymphoma (MCL).15 However, a heterogeneous population of cells was profiled, complicating the interpretation of results and reducing the sensitivity of the study.15 CLL cells are especially appropriate for microarray analysis because of the relative ease in obtaining a purified, developmentally homogenous leukemic cell population in large quantities. Recently, oligonucleotide microarray has been used to compare the expression patterns of CLL-B cells with several normal B cell populations.9 That study, however, emphasized sample classification based on the overall expression pattern comparison. It did not focus on analyzing the potential roles of differentially expressed genes in the pathogenesis of CLL. Here, we report the use of oligonucleotide microarrays to compare gene expression profiles between purified CLL-B cells and CD5-expressing normal B cells, and between CLL and normal circulating B cells, as a way to identify potential molecular deregulations that may contribute to the pathophysiology of CLL. The results of our study revealed

Expression profiling of CLL Z Zheng et al

2430

that in CLL there was significant down-modulation of genes involved in both protein and lipid antigen presentation. We also observed reduced RNA levels for genes involved in interferon gamma response but increased transcript levels for IL-4 response genes. The profile also indicated reduced cell proliferation and metabolism but increased intracellular protein transport and processing, as well as increased production of extracellular matrix and adhesion molecules in CLL-B cells relative to normal CD5+ B cells. The significance of these abnormal gene expression patterns to CLL pathobiology is discussed. Materials and methods

netic beads coupled with goat anti-IgG (Dynal) were employed, exactly as in the protocol used for the isolation of CD5+ cells from human tonsils. Based on gene expression pattern comparisons, no obvious difference in gene expression can be attributed to the additional positive selection (see Results and discussion below). For additional comparison, we isolated human peripheral blood B cells directly from fresh buffy coats obtained from healthy donors (Stanford Blood Center, Stanford, CA, USA) using CD19-Dynabeads (Dynal). Purified B cells on beads were lysed directly in Trizol for total RNA extraction. Total RNA from seven individual buffy coats was pooled for subsequent Poly(A) selection. The use of all patient and normal samples above were according to an IRB approved protocol.

Cell isolation and RNA extraction Normal CD5+ B cells were purified from four fresh human tonsils collected from routine tonsillectomy procedures. CD5+ B cells like these were routinely used in most laboratories as the normal controls in the studies of CLL-B cells. The tonsil tissues were disrupted by gentle mechanical means, and the resulting cell suspension was centrifuged through Ficoll– Hypaque to obtain mononuclear cells. T cells were removed by two rounds of rosetting with an excess amount of fresh, neuroaminidase-treated sheep red blood cells (SRBC). Less than 1% of the remaining cells were CD2+ after rosetting. The cells were then incubated with anti-CD5 MAb (Caltag, Burlingame, CA, USA) for 3–5 h at 4ºC, washed thoroughly in PBS and incubated for 12 h at 4ºC with 20-fold excess (relative to target cells) of the magnetic beads coated with goat antiIgG (Dynal, Lake Success, NY, USA). CD5+ B cells attached to the beads were then magnetically separated from the remaining cell suspension. After isolation, the CD5+ B cells on beads were directly lysed with Trizol (Gibco, Carlsbad, CA, USA) for total RNA extraction. Heparinized venous blood samples from eight CLL patients with Rai stages varying from 0 to 2 (Table 1) were obtained after written informed consent. None of the CLL patients received chemotherapy treatment during the 6 months prior to the blood drawn for this study. Mononuclear cells were obtained after centrifugation over Ficoll–Hypaque, and CLL cells from all but two samples were negatively selected by depleting T cells using two rounds of anti-CD2 coated magnetic beads (Dynal). More than 96% of the purified CLL cells were CD19+ and CD5+ in the CLL samples studies. Freshly isolated CLL cells were immediately lysed in Trizol for total RNA extraction. For two of the CLL samples (CM and C72), additional steps of anti-CD5 coating followed by positive selection using mag-

Leukemia

Table 1

Clinical features of enrolled CLL patients

Sample

Age

WBC (109/l)

Hb (g/dl)

Platelets (109/l)

Rai stage

sIg

C725 C1010 CH C72 CM C101 C117 C828

66 75 34 46 71 50 45 81

17.8 98.2 198 19.1 54 42.7 28.4 10.8

13.1 15 11.7 15.3 11.6 15.5 13 12.4

159 222 142 198 99 155 262 125

1 1 2 0 2 0 0 1

kappa kappa lambda kappa kappa NA kappa NA

GeneChip array expression analysis Poly(A)+ RNA was selected using Oligotex mRNA midi kits (Qiagen, Valencia, CA, USA). Double-stranded cDNA was synthesized from poly(A)+ RNA with the SuperScript Choice kit (Gibco) and oligo(dT) primers that contained a T7 RNA polymerase recognition sequence at the 5⬘ end. Approximately 1 ␮g of cDNA was subjected to in vitro transcription in the presence of biotinylated UTP and CTP (Enzo Diagnostic, Farmingdale, NY, USA) using the MegaScript T7 kit (Ambion, Austin, TX, USA). The labeled cRNA was fragmented and hybridized to a set of four oligonucleotide arrays (GeneChip Hu6000 array set; Affymetrix) overnight. The arrays were then washed, stained with phycoerythrin–streptavidin and scanned according to the manufacturer’s instructions. The array contains 6790 probe sets for 6416 genes (5223 known human genes and 1193 unnamed ESTs). Expression data were analyzed using GeneChip 3.0 software (Affymetrix, Santa Clara, CA, USA). Global scaling of intensity values was used. The average intensity over all genes on a chip was adjusted to 116 for all arrays to compensate for array-related variations in hybridization efficiency and signal acquisition. Fluorescence intensity levels of less than 20 units (after scaling) were converted to 20, since discrimination of expression below this level could not be performed with confidence. For each gene, the CLL/TB5+ expression ratio is calculated from the absolute expression value of one of the eight CLL samples and one of the four tonsilar CD5+ B cell samples. The average expression ratio (Avg(CLL/TB5+)) was calculated by averaging the 32 possible CLL/TB5+ expression ratios.

Northern blot analysis One microgram of polyA+ RNA from purified tonsillar, peripheral blood, and CLL B cells was separated on a 1% agarose gel containing 0.6 M formaldehyde, transferred on to a nylon membrane (Nytran; Schleicher and Schuell, Keene, NH, USA), and hybridized with DNA probes randomly primed with 32PdCTP. Hybridization signals were detected using a PhosphoImager (Molecular Dynamics, Sunnyvale, CA, USA). Results and discussion

Quality control of samples To investigate the molecular basis underlying the different phenotypes of normal and CLL-B cells, we profiled the


expression of 6500 independent genes in freshly isolated leukemic B cell samples from eight CLL patients (Table 1), as well as purified control B cell samples. The control B cell samples included four tonsillar CD5+ B cell samples and one pooled peripheral blood CD19+ B cell sample. Some genes with previously characterized expression in CLL were included in the array and were used as internal controls. The expression levels of these genes are summarized in Figure 1. As expected, tonsil CD5+ cells and CLL cells expressed high levels of the mRNA for the IgM heavy chain. Indicative of the monoclonal nature of the disease, there was predominance in CLL cells of one of the two Ig light chains (lambda or kappa) expression that was consistent with available clinical phenotyping results (data not shown). CD4 expression was not detectable (mean expression intensity (same below): 22 for CLL vs 24 for tonsil CD5+ cells), indicating little or no contamination with T cells in the preparations. Expressions of CD3 and CD14 mRNA were low with similar levels between normal and CLL samples (CD3: 63 for CLL vs 75 for tonsil CD5+ cells; CD14: 59 for CLL vs 55 for tonsil CD5+ cells). CD23, a molecule commonly associated with CLL,16 was highly expressed in CLL samples, while on average, CD23 expression in tonsillar CD5+ B cells was lower (2318 for CLL vs 1059 for tonsil CD5+ cells). As reported previously,2 there was an increase in the level of bcl-2 mRNA, but no abnormal expression of p53 or Rb in CLL relative to normal CD5+ cells (Figure 1). The results described above indicate little degradation or contamination of our mRNA samples and demonstrate the robustness of the sample purification procedure.

samples), using the program GeneCluster as previously described.13 The SOM algorithm, like other clustering algorithms, serves to expose patterns in the gene expression data without using prior knowledge. The expression levels of the 6790 probe sets represented on the GeneChip arrays were used in the analysis. The default variation filter in the program was used to eliminate genes that did not change significantly across samples. SOM was used to organize the remaining genes into 18 clusters (Figure 2). Although these clusters were generated without presumptions, they corresponded to clear biological relevance. For example cluster c7 identified 113 genes that were consistently up-regulated in CLL-B cells and cluster c16 identified 148 genes consistently down-regulated in CLL. Most of the remaining clusters represent genes that were preferentially expressed in only one of the samples examined, thus capturing the unique signature pattern of the sample (eg cluster c1, c4 and c8). The lack of clusters representing genes differentially expressed in only those two positively selected CLL samples (C72 and CM) indicated that under our experimental conditions, additional positive selection based on CD5 ligation of CLL cells had little effect on the overall gene expression pattern in CLL.

2431

Global pattern of gene expression To identify patterns of gene expression in CLL that differed from those in normal control CD5+ B cells, we first employed a self-organizing map (SOM) algorithm to analyze the 12 samples (four tonsillar CD5+ normal control and eight CLL

Figure 1 Expression of control genes. Expression levels of selected genes in four tonsilar CD5+ B samples (triangles) and eight CLL-B samples (circles) studied. Expression levels of less than 20 are converted to 20.

Figure 2 SOM analysis of gene expression in normal and CLL-B cells. Using GeneCluster,13 the 1958 genes that passed the default variation filter were grouped into 18 clusters. Each cluster is represented by the average expression pattern for genes in the cluster. Normalized averaged expression levels are shown on the y-axis, with the two lines delineating 1 s.d. off the average. The 12 samples are represented on the x-axis, with the order being (from left to right) T1, T2, T3, T62, C101, C1010, C117, C72, C725, C828, CH and CM. The first four samples from the left are the tonsillar CD5+ B cells and the remaining eight CLL samples. The number of genes in each cluster is indicated at the top center of each cluster. Note that the cluster c7 represents 113 genes that showed up-regulation in all CLL samples and cluster c16 consists of 148 genes that were down-regulated in all CLL samples. Most other clusters, such as c8, consist of genes that were preferentially expressed in only one of the samples examined. Leukemia


2432

Differentially expressed genes between CLL and normal CD5+ B cells To represent quantitatively the gene expression differences between normal and CLL cells, for each gene we obtained an average expression ratio in each CLL sample, ie average of the four possible CLL/TB5+ ratio values from the CLL and four TB5+ expression values. We first selected for genes that showed two-fold or more differential expression between normal and CLL cells in at least six of the eight CLL samples. We then used a hierarchical clustering program (Cluster17) to cluster these differentially expressed genes. Among the 248 genes used in hierarchical clustering, 56 were excluded for further analysis due to low signal intensity (average of CLL and average of TB5+ both ⬍100) or inconsistent expression pattern (coefficient of variation for CLL/TB5+ ratios more than 100%), leaving 192 differentially expressed genes. Northern analysis of 10 randomly selected genes from the 192 genes all confirmed the results from the GeneChip hybridization (two shown in Figure 3). Among these 192 genes, about half (104 genes) were expressed in the normal peripheral blood B cells at a level not significantly different (less than two-fold) to the average expression level in the eight CLL samples. These 104 genes, although differentially expressed between tonsillar CD5+ B cells and circulating CLL cells, may contain genes related to the different anatomical locations of the two lymphocyte populations (circulating vs non-circulating) and were thus excluded for further consideration. Some of these 104 genes, for example a cluster of 19 down-regulated genes known to be involved in various aspects of cell proliferation, may account for some of the phenotypes of CLL-B (eg cell cycle arrest in CLL). They were nonetheless excluded, due to a similar expression pattern in the normal peripheral B cells, which were also non-cycling. The remaining 88 genes are likely to be an underestimate of genes differentially expressed between normal CD5+ B cells and CLL cells. The age distributions of our tonsil, normal blood, CLL blood samples are non-overlapping, with average age increasing in that order. There are 19 genes in the list of 88 with expression level in normal blood falling between average levels in tonsil and in CLL. Among them, one gene (interferon receptor ifnar21) showed significant correlation between expression level in CLL patient and patient age (absolute value of correlation coefficient ⬎0.7) and is excluded from the final list. The remaining 87 genes included 51 genes that were upregulated and 36 genes that were down-regulated in CLL cells

compared to normal CD5+ B cells. The complete list of these genes is shown in Table 2. The average expression levels together with their standard deviations in tonsil and CLL samples, as well as the average CLL/TB5+ ratios, are listed in Table 2. Their expressions in the pooled normal B cell sample are also included for reference. These genes could be grouped into different functional categories (Table 2). Overall, about one-third of differentially expressed genes appeared to be related to immunoregulation, suggesting that the immune deregulation was the most prominent abnormality of CLL. Some of the genes showing the most prominent differences are described below.

CD1 gene family CLL cells display the characteristics of anergic B cells that are defective in presenting antigens and mounting antibody responses to antigen.18,19 We observed down-regulation of B cell surface molecules that are believed to be critical in T-B cell interaction, such as CD40 and CD83.20 The costimulatory molecule CD27, whose expression in B cells is suppressed by CD40,21 was up-regulated as expected. Surprisingly, the two most down-regulated genes (25-fold and five-fold suppression, respectively) were CD1c and CD1d, two members of the CD1 family of MHC-like molecules responsible for presenting lipid rather than peptide antigen to T cells.22 In CLL-B cells these two CD1 mRNAs were virtually undetectable (Table 2). Unique among MHC and MHC-like molecules, CD1 is recognized by the antigen receptors of two prominent T cell subset with innate immune function: the CD 1d-restricted human Valpha24 natural killer T cell and the CD1c-specific human V␦1 gamma delta T cells.22 Thus, our data indicate for the first time that the innate immunity mediated by the activation of CD1-restricted T cells may be compromised in CLL. The activation of NK T cells results in engagement of NK T cell TCR and the lysis of microbial-infected antigen-presenting cells. Given that leukemic B cells are the major antigen presenting cells in CLL, lack of CD1 expression on CLL-B may explain the increased chance of microbial infection in CLL patients. However, more importantly, activation of NK T cells in vivo also induces a series of cellular activation events that leads to the activation of innate cells and adaptive cells such as B cells and T cells.22,23 Thus, defective CD1-restricted T cell activation conceivably can also lead to reduced adaptive response in CLL. Consistent with this hypothesis, CD69, which upon CD1-restricted T cell activation is induced on normal B cells,23 was expressed five-fold lower in CLL-B cells than in normal cells (Table 2). Given that CD1-restricted T cells play critical roles not only in microbial immunity, but also in autoimmune regulation24 and tumor prevention,25 it would be of great interest to study the status of T cells that recognize the CD1 molecules in CLL patients.

Abnormal expression of genes involved in cytokine response

Figure 3 Verification of array hybridization results. Representative Northern analysis of CCR7 and M-phase inducer phosphotase2 mRNA expressions in purified normal peripheral blood B cells (BB), tonsillar CD5+ B cells (TB) and CLL-B cells (CLL). Leukemia

In the category of cytokine signaling, the receptor for interleukin 4, IL-4R, is up-regulated by more than five-fold. IL-4 has been shown to protect CLL-B cells from spontaneous26–28 and Fas-mediated29 apoptosis in vitro. The potential increase in IL4 levels in T cells from CLL patients30–32 would further exacerbate the IL-4 induced response of CLL cells. Increased IL-4


Table 2

Accession

2433

Differentially expressed genes between CLL-B cells and normal CD5+ B cells

Name

Antigen presentation M63928 CD27 R44494 CD83 R49884 CD40 X60592 CD40 Z11697 CD83 X14974 CD1 R3 gene for MHC-related antigen (CD1d) Z22576 CD69 M28827 Thymocyte antigen CD1c

Avg TB5+ (s.d.)

392 116 506 243 774 198

(188) (75) (63) (108) (301) (48)

649 (338) 470 (69)

BB

Avg(CLL/TB5+)

(430) (18) (55) (55) (100) (37)

241 160 109 20 720 124

3.9 0.49 0.44 0.42 0.34 0.23

118 (104) 20 (0)

415 104

0.23 0.04

863 152 3681 501 807 1841

5.3 3.2 0.48 0.41 0.38 0.34

1789 839 154

5.7 2.8 2.4

AvgCLL (s.d.)

1340 37 219 86 232 43

Cytokine signaling X52425 IL-4 receptor X02875 (2⬘–5⬘) Oligo A synthetase E M26383 IL-8 J03143 Interferon-gamma receptor alpha chain R34698 Interferon-inducible protein 9–27 R73660 Gamma-interferon-inducible protein IP-30

503 130 119 267 522 1493

Immunoglobulin receptors U12255 IgG receptor FcRn M14766 Fc-Epsilon receptor CD23 M28696 Low-affinity IgG Fc receptor (FcGR2B)

163 (44) 1059 (650) 146 (30)

883 (179) 2318 (383) 343 (68)

Lymphocyte trafficking and adhesion/ECM/cytoskeleton U05291 Fibromodulin L31584 Chemokine receptor CCR7 U25956 P-selectin ligand T70046 Filamin B H14384 Agrin M95610 Human alpha 2 type IX collagen (COL9A2) L23823 Integrin beta 7 X69490 Titin T41204 92 kDa type V collagenase D44497 Coronin, actin-binding protein, 1A X59350 CD22 M15395 LFA-1

54 177 23 30 44 20 159 68 153 3207 1312 339

(17) (78) (6) (12) (13) (0) (42) (7) (67) (687) (196) (99)

1062 1342 151 170 240 115 667 280 58 984 272 37

(359) (371) (60) (157) (97) (44) (289) (144) (23) (479) (99) (38)

50 272 441 39 27 20 297 51 152 250 618 354

21.5 8.8 6.8 6.3 6.0 5.7 4.4 4.1 0.44 0.32 0.21 0.12

20 36 108 294 34

(0) (12) (114) (41) (21)

959 316 236 1743 146

(308) (125) (61) (402) (58)

313 40 106 195 20

47.9 9.9 6.6 6.0 5.5

45 95 364 184

(35) (61) (118) (92)

148 216 909 355

(72) (42) (451) (153)

64 71 181 154

4.9 4.1 2.7 2.4

410 138 199 184 282 189 238

(180) (65) (39) (27) (43) (117) (77)

134 40 63 59 77 43 40

(122) (21) (29) (32) (61) (20) (17)

42 113 287 117 21 686 293

0.41 0.39 0.33 0.32 0.28 0.27 0.18

Protein processing/metabolism X69910 P63 mRNA for transmembrane protein T92055 ABC-1 M55621 N-acetylglucosaminyltransferase I D16111 Phosphatidylethanolamine binding protein L41559 Pterin-4A-carbinolamine dehydratase (PCBD) X62822 Beta-galactoside alpha-2,6-sialyltransferase U26648 Syntaxin 5A T48904 Heat shock 27 kDa protein T96942 Mitochondrial short-chain enoyl-CoA hydratase R08183 10 kDa heat shock protein, mitochondrial X05309 C3B/C4B receptor (CR1) F allotype K03001 Aldehyde dehydrogenase 2 X69433 Mitochondrial isocitrate dehydrogenase M13792 Adenosine deaminase (ADA) R32804 Glucose transporter type 3 X76648 Glutaredoxin

(154) (22) (67) (51) (102) (268)

2474 405 43 107 191 489

(1153) (236) (8) (33) (85) (209)

(continued on next page)

signaling via increase in both IL-4R on the CLL-B cell and IL4 from T cell would likely offer a survival advantage for leukemic CLL-B cells. In contrast, molecules involved in responding to TH1 cytokine IFN-gamma (IFN-gamma receptor and two IFN-gamma regulated genes, IP-30 and 9-27 protein) are suppressed by two- to three-fold in CLL cells. The reduced response to IFNgamma in CLL has not been reported before. IFN-gamma

plays a central role in tumor surveillance and immunity,33 and this function of IFN-gamma is achieved primarily through direct action on the tumor cells to increase their immunogenicity.34,35 The major consequence of reduced IFN-gamma signaling in CLL may be decreased susceptibility of CLL cells to tumor immunity and thus increased CLL-B survival.

Leukemia


2434

Table 2

Differentially expressed genes between CLL-B cells and normal CD5+ B cells (continued)

Accession

Name

Transcriptional and growth regulation M24069 DNA-binding protein A (dbpA) T40507 dbpA M74718 TCF4 M81934 Human CDC25B X66924 Heir-1 mRNA for helix–loop–helix protein (ID3) H46136 Homeobox protein HOX-A10 X51345 JUN-B U00115 BCL-6 M22638 LYL-1 L25941 Integral nuclear envelope inner membrane protein (LBR) H24033 C-MYC R37053 P55-c-FOS Proto-oncogene U22431 Hypoxia-inducible factor 1 alpha R12588 G2/Mitotic-specific cyclin B2 Antiapoptosis M14745 D15057 U27467 Signaling M77349 M80899 U20982 H24596 R70806 M97675 M93425 H50438 L25081 U02570 X07109 R61874 H67367 H24635 R67358

Miscellaneous X79204 T52999 R56038 M16405 H82820 R40717 T55709 R81175 R43911 H40677 R45299 X04011

BCL-2-alpha Defender against apoptosis (DAD-1) BCL-2 related mRNA (A1) TGF-beta induced gene product (BIGH3) Human novel protein AHNAK Insulin-like growth factor binding Protein-4 (IGFBP4) RAF kinase inhibitor protein Diacylglycerol kinase Human transmembrane receptor (ROR1) Protein-tyrosine phosphatase G1 M-phase inducer phosphatase 2 Transforming protein RHOC Human CDC42 GTPase-activating protein PKC type beta II PAK-interacting exchange factor alpha RAN-specific GTPase-activating protein Protein-tyrosine phosphatase PAC-1 MAP kinase phosphatase-1 Ataxin 1 Unknown Unknown M4 Muscarinic acetylcholine receptor Glycophorin B Unknown Unknown Glutamate receptor 7 Unknown Unknown Unknown X-CGD gene involved in chronic granulomatous disease

Avg TB5+ (s.d.)

37 106 120 22 163

(19) (26) (20) (4) (37)

AvgCLL (s.d.)

BB

Avg(CLL/TB5+)

18.0 11.3 4.9 4.7 4.0

533 1143 574 100 630

(177) (404) (214) (70) (402)

164 409 46 20 52

123 2073 295 498 321

(28) (1185) (90) (128) (149)

53 689 92 136 75

(46) (423) (27) (93) (26)

20 4467 303 57 347

0.4 0.4 0.33 0.29 0.27

245 282 169 254

(98) (239) (76) (82)

44 21 25 37

(17) (1) (9) (21)

178 385 218 91

0.20 0.17 0.17 0.16

39 (16) 85 (34) 451 (100)

230 (71) 216 (49) 160 (69)

98 96 1081

6.8 3.0 0.4

37 (33) 20 (0) 34 (18)

643 (169) 495 (377) 686 (311)

192 52 30

26.0 24.8 24.6

48 175 20 40 62 68 35 196 324 162 156 764

(56) (107) (0) (5) (33) (34) (20) (39) (49) (56) (21) (425)

494 782 182 276 302 256 101 592 148 59 50 182

(281) (307) (119) (134) (90) (120) (56) (163) (61) (35) (37) (176)

20 128 20 123 106 100 35 222 311 20 173 1516

19.5 12.4 9.1 7.0 6.0 4.7 3.6 3.1 0.5 0.4 0.3 0.3

24 48 30 74 50 35 51 47 1131 156 316 241

(7) (8) (12) (77) (42) (17) (36) (28) (104) (44) (212) (50)

179 225 124 163 129 109 120 104 2277 37 36 34

(74) (84) (66) (84) (70) (48) (58) (26) (575) (37) (18) (21)

53 90 39 36 283 47 20 28 669 110 79 496

8.0 4.8 4.6 4.6 4.0 3.8 3.7 3.0 2.0 0.3 0.2 0.1

The genes are listed in putative functional categories and within each category are sorted by fold changes (Avd(CLL/TB5+)). BB, normal peripheral CD19+ cells pooled from seven individuals.

Increased expression of immunoglobulin receptors We also observed induction of several immunoglobulin receptor messages in CLL-B cells. These include the neonatal IgG receptor FcRn (5.7-fold increase), the Fc-epsilon receptor CD23 (2.8-fold increase) and the low affinity IgG Fc receptor FcGR2B (2.4-fold increase). The overexpression of CD23 is consistent with increased response to IL-4, as CD23 can be induced by IL-4 in vitro in both normal B cells and CLL-B.26 Leukemia

FcGR2B was recently shown to be the target for deregulation by chromosomal translocation in B cell follicular lymphoma, and overexpression of FcGR2B may play a role in tumor progression in follicular lymphoma.36 Increased FcGR2B engagement may reduce antibody-dependent cellmediated cytotoxicity of antibodies against tumor cells in vivo,37 therefore it is likely that increased FcGR2B also contributes to the survival of leukemic cells. The observation of increased FcGR2B expression in CLL should be relevant to the


treatment of CLL as several antibody therapeutics, including rituximab (anti-CD20) and CAMPATH (anti-CD52), are currently in clinical trials for treatment of CLL.

Increased expression of the extracellular matrix proteins It is surprising that CLL cells strongly up-regulate several extracellular matrix proteins. The most highly induced gene in this category is fibromodulin, with 21.5-fold induction in CLL. Fibromodulin is a proteoglycan that interacts with the type I and II collagen fibrils in the extracellular matrix and has a role in the assembly of collagen fibrils in connective tissues.38 Also overexpressed in CLL-B cells is the type IX collagen, a fibrilassociated collagen with similar function to fibromodulin.39 Corresponding to increased expression of these matrix proteins, there is decreased expression of collagenase V involved in the degradation of extracellular matrix. These together with the increased expression of P-selectin ligand, agrin, and integrin beta7 suggest increased adhesiveness and reduced mobility of CLL cells relative to normal counterparts. This is particularly noteworthy given that the CLL-B cells are circulating whereas the CD5+ normal counterparts for this study are cells in tissue. Most of these molecules were not overexpressed in normal circulating B cells (Table 2). The contribution of these phenotypic changes to the characteristics of CLL cell trafficking remains to be determined. There was no correlation, however, between the expression of these molecules and the degree of leukemic infiltration of lymphoid organs in our patients. The interactions between cancer cells and tissue microenvironment are critical for cancer survival and progression.40 For example, interaction with fibronectin prevents CLL-B from apoptosis.41 It is not clear with what cell types the CLL-B cells may have increased contact. One possible candidate could be the bone marrow stromal cells, which adhere to CLL cells but not normal B cells and provide survival factors for CLL.42,43 The increased expression of chemokine receptor CCR7 (Figure 3) is noteworthy, as the role of this receptor on the B cells is not primarily for B cell homing to lymph organs,44,45 but rather for bringing T cells and activated B cells within lymph organs together for interaction.45 CLL-B cells express CCR7 at a level similar to activated normal B cells (unpublished observation). Being the major accessory cells in CLL, the leukemic cells may compete with competent, activated B cells for T cell interaction.46 Moreover, once in contact, CLL-B cells quickly inactivate T cells by downregulating CD40L and costimulatory molecules on the T cell surface,47,48 rendering T cells anergic. Further study is needed to elucidate the role of CCR7 in CLL pathogenesis. The expression of agrin in CLL-B cells, but not normal B cells, is also unusual. When secreted from T cells agrin can promote the formation of the contact site (immunological synapse) between activated T cells and antigen-presenting cells.49,50 It will be of interest to test whether agrin from CLL-B cells has a similar function in promoting immunological synapse formation, or in clustering molecules on CLL-B cells to form contact site with other cell types.

Changes in non-immune-related gene expression The comprehensive nature of microarray gene expression profiling allows one to gain valuable insights beyond immune

deregulation. For example, several genes involved in cellular proliferation pathways were also identified in the differentially regulated set. Most of the identified proliferative protooncogenes (c-MYC, Jun-B, c-Fos, Hox-A10, BCL-6, G2-specific Cyclin B2) were down-regulated, while inhibitor for proliferation (Id3) was over expressed. This pattern is expected given the cell cycle arrest phenotype of CLL-B cells. Consistent with the quiescent state of CLL-B cells, many genes involved in cellular metabolism (such as aldehyde dehydrogenase 2) were down-regulated (Table 2). It should be noted that although c-Myc is the primary transcriptional factor that regulates proliferation in many tumor cell types, it is not clear whether its down-regulation is the cause for the growth arrest in CLL. Many c-Myc target genes have been identified51 and are represented in our array. However, they were not significantly down-regulated in CLL (data not shown). In addition, the c-Myc partner Max expression was not significantly decreased relative to CD5+ B cells (average fold change = 0.70). The transcriptional factor dbpA was highly induced in CLL (average 15-fold) relative to CD5+ B cells. Moreover, the high expressions were specific to CLL as they were not significantly up-regulated in purified CD5+ B cells from a MCL patient (unpublished observation). A closely related protein dbpB (YB1) was highly but not differentially expressed in CLL (average fold change = 0.76), indicating the specificity of dbpA overexpression. As a DNA binding protein containing the cold-shock domain, dbpA is known to inhibit the induction of MHC molecules by IFN-gamma,52 but it has never been described in human B cells. It will be of interest to investigate the potential role of this transcriptional regulator in pathogenesis of CLL. Although there are at least 25 genes on our GeneChip with known roles in apoptosis, we found only three anti-apoptotic molecules differentially expressed in CLL. In addition to Bcl2, the anti-apopototic molecule DAD-1 was overexpressed in CLL cells by about three-fold. DAD-1 does not belong to the Bcl2 family and it protects cells from apoptosis by mechanisms distinct from those used by bcl-2.53 Paradoxically, the Bcl2 family antiapoptotic molecule A1 was decreased in CLL. While the Bcl-2 family of proteins plays a major role in protecting CLL-B cells from apoptosis induced by therapeutic agents,54 it is not clear whether they are critical in establishing the pro-survival phenotype of CLL in vivo, since prior to therapy CLL cells do not encounter those therapeutic agents. In fact, when cultured in vitro, CLL-B cells, even with overexpression of Bcl2, are more prone to spontaneous apoptosis than normal peripheral blood B cells or CD5+ normal B cells.42 This observation suggests the increased steady-state level of antiapoptotic molecules, while important in determining response to therapy, is not sufficient to explain the CLL pathogenesis. Finally, we found in CLL cells significant up-regulation of genes for protein processing (N-acetylglucosaminyltransferase 1, PCBD, beta-galactoside alpha-2,6-sialyltransferase) and protein trafficking/transport (p63, ABC-1, phosphatidylethanolamine binding protein, syntaxin 5A). Among them, the p63 mRNA, for a transmembrane protein localized to the ER-Golgi intermediate compartment,55 shows a 48-fold increase, the highest in the entire list. Few studies have explored the role of intracellular protein processing and transport in the disease phenotype of CLL. Our results correlate the deregulation of this process with CLL. Some proteins encoded by genes identified here may modulate functions of immunoregulatory proteins. For example, beta-galactoside alpha 2,6-sialyltransfer-

2435

Leukemia


2436

ase sialylates CD22 and can abrogate CD22-mediated B lymphocyte adhesion in vitro.56 Identifying specific protein processing and trafficking pathways that are affected, and the target molecules that they regulate, will help understand the pathogenesis of the disease. CLL is a complex human disease characterized by abnormal cell death rather than increased proliferation. Our comprehensive gene expression profiling provides important information on the global pattern of gene expression in CLL and identifies many candidate genes potentially involved in CLL pathology. The physiological phenotype of CLL predicts deregulation of a number of genes that were identified in our study, for example, the down-regulation of genes involved in T–B cell interaction and in cell proliferation. More importantly, however, our study revealed several unexpected patterns of gene expression in CLL that may shed new light on the disease mechanisms. These include the suppression of CD1c and CD1d involved in innate immunity, the reduced IFN-gamma signaling, increased adhesiveness, as well as increased intracellular protein trafficking and processing. The prolonged survival of CLL-B cells may not only be due to increased expression of anti-apoptotic molecules, but also through increased response to survival factors, characterized by increased response to IL-4 and adhesion to extracellular matrix, and through various mechanisms of evading the immune response, for example, by turning off the expression of CD1c and CD1d, reducing the immunogenic response to interferon gamma, inactivating T cell in B–T interaction and increasing the expression of immunoglobulin receptors which neutralize antibody-dependent cell-mediated cytotoxicity. The changes in these cellular phenotypes likely result from abnormal transcriptional regulation, for which we identified a highly abnormally expressed transcriptional factor dbpA, and from abnormal post-transcriptional regulation, where significantly altered intracellular protein processing and trafficking seems apparent. Our results provide a valuable basis for additional, more directed and focused studies on CLL pathobiology.

8 9

10

11

12

13

14

15

16 17

Acknowledgements We are grateful to Lenn Flechter for help with clinical samples. The GeneChip array hybridizations were performed in the Affymetrix Academic User Center. The authors acknowledge the support of NIH grants 5R01CA056764-08 and P01HG01323 for this work.

References 1 Parker SL, Tong T, Bolden S, Wingo PA. Cancer statistics, 1997. CA Cancer J Clin 1997; 47: 5–27. 2 Caligaris-Cappio F. B-chronic lymphocytic leukemia: a malignancy of anti-self B cells. Blood 1996; 87: 2615–2620. 3 Kipps TJ. Chronic lymphocytic leukemia. Curr Opin Hematol 1997; 4: 268–276. 4 Wierda WG, Kipps TJ. Chronic lymphocytic leukemia. Curr Opin Hematol 1999; 6: 253–261. 5 Lydyard PM, Lamour A, MacKenzie LE, Jamin C, Mageed RA, Youinou P. CD5+ B cells and the immune system. Immunol Lett 1993; 38: 159–166. 6 Kipps TJ, Robbins BA, Tefferi A, Meisenholder G, Banks PM, Carson DA. CD5-positive B-cell malignancies frequently express cross-reactive idiotypes associated with IgM autoantibodies. Am J Pathol 1990; 136: 809–816. 7 Kipps TJ, Rassenti LZ, Duffy S, Johnson T, Kobayashi R, Carson Leukemia

18 19

20 21 22 23

24

25

DA. Immunoglobulin V gene expression in CD5 B-cell malignancies. Ann NY Acad Sci 1992; 651: 373–383. Kipps TJ, Duffy SF. Relationship of the CD5 B cell to human tonsillar lymphocytes that express autoantibody-associated crossreactive idiotypes. J Clin Invest 1991; 87: 2087–2096. Klein U, Tu Y, Stolovitzky GA, Mattioli M, Cattoretti G, Husson H, Freedman A, Inghirami G, Cro L, Baldini L, Neri A, Califano A, Dalla-Favera R. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med 2001; 194: 1625–1638. Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD, Lander ES. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 1999; 286: 531–537. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell JI, Yang L, Marti GE, Moore T, Hudson J, Jr., Lu L, Lewis DB, Tibshirani R, Sherlock G, Chan WC, Greiner TC, Weisenburger DD, Armitage JO, Warnke R, Staudt LM. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000; 403: 503–511. Iyer VR, Eisen MB, Ross DT, Schuler G, Moore T, Lee JC, Trent JM, Staudt LM, Hudson J, Jr., Boguski MS, Lashkari D, Shalon D, Botstein D, Brown PO. The transcriptional program in the response of human fibroblasts to serum. Science 1999; 283: 83–87. Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E, Lander ES, Golub TR. Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation. Proc Natl Acad Sci USA 1999; 96: 2907– 2912. Karpf AR, Peterson PW, Rawlins JT, Dalley BK, Yang Q, Albertsen H, Jones DA. Inhibition of DNA methyltransferase stimulates the expression of signal transducer and activator of transcription 1, 2, and 3 genes in colon tumor cells. Proc Natl Acad Sci USA 1999; 96: 14007–14012. Hofmann WK, de Vos S, Tsukasaki K, Wachsman W, Pinkus GS, Said JW, Koeffler HP. Altered apoptosis pathways in mantle cell lymphoma detected by oligonucleotide microarray. Blood 2001; 98: 787–794. Dorfman DM, Pinkus GS. Distinction between small lymphocytic and mantle cell lymphoma by immunoreactivity for CD23. Mod Pathol 1994; 7: 326–331. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998; 95: 14863–14868. Ranheim EA, Kipps TJ. Activated T cells induce expression of B7/BB1 on normal or leukemic B cells through a CD40-dependent signal. J Exp Med 1993; 177: 925–935. Dazzi F, D’Andrea E, Biasi G, De Silvestro G, Gaidano G, Schena M, Tison T, Vianello F, Girolami A, Caligaris-Cappio F. Failure of B cells of chronic lymphocytic leukemia in presenting soluble and alloantigens. Clin Immunol Immunopathol 1995; 75: 26–32. Zhou LJ, Tedder TF. Human blood dendritic cells selectively express CD83, a member of the immunoglobulin superfamily. J Immunol 1995; 154: 3821–3835. Ranheim EA, Cantwell MJ, Kipps TJ. Expression of CD27 and its ligand, CD70, on chronic lymphocytic leukemia B cells. Blood 1995; 85: 3556–3565. Park SH, Bendelac A. CD1-restricted T-cell responses and microbial infection. Nature 2000; 406: 788–792. Nishimura T, Kitamura H, Iwakabe K, Yahata T, Ohta A, Sato M, Takeda K, Okumura K, Van Kaer L, Kawano T, Taniguchi M, Nakui M, Sekimoto M, Koda T. The interface between innate and acquired immunity: glycolipid antigen presentation by CD1dexpressing dendritic cells to NKT cells induces the differentiation of antigen-specific cytotoxic T lymphocytes. Int Immunol 2000; 12: 987–994. Lehuen A, Lantz O, Beaudoin L, Laloux V, Carnaud C, Bendelac A, Bach JF, Monteiro RC. Overexpression of natural killer T cells protects Valpha14- Jalpha281 transgenic nonobese diabetic mice against diabetes. J Exp Med 1998; 188: 1831–1839. Smyth MJ, Thia KY, Street SE, Cretney E, Trapani JA, Taniguchi M, Kawano T, Pelikan SB, Crowe NY, Godfrey DI. Differential tumor


26 27

28 29

30

31

32 33

34

35 36

37 38

39 40 41

surveillance by natural killer (NK) and NKT cells. J Exp Med 2000; 191: 661–668. Fournier S, Yang LP, Delespesse G, Rubio M, Biron G, Sarfati M. The two CD23 isoforms display differential regulation in chronic lymphocytic leukaemia. Br J Haematol 1995; 89: 373–379. Dancescu M, Rubio-Trujillo M, Biron G, Bron D, Delespesse G, Sarfati M. Interleukin 4 protects chronic lymphocytic leukemic B cells from death by apoptosis and upregulates Bcl-2 expression. J Exp Med 1992; 176: 1319–1326. Pu QQ, Bezwoda WR. Interleukin-4 prevents spontaneous in-vitro apoptosis in chronic lymphatic leukaemia but sensitizes B-CLL cells to melphalan cytotoxicity. Br J Haematol 1997; 98: 413–417. Williams JF, Petrus MJ, Wright JA, Husebekk A, Fellowes V, Read EJ, Gress RE, Fowler DH. fas-mediated lysis of chronic lymphocytic leukaemia cells: role of type I versus type II cytokines and autologous fasL-expressing T cells. Br J Haematol 1999; 107: 99–105. Mu X, Kay NE, Gosland MP, Jennings CD. Analysis of blood Tcell cytokine expression in B-chronic lymphocytic leukaemia: evidence for increased levels of cytoplasmic IL-4 in resting and activated CD8 T cells. Br J Haematol 1997; 96: 733–735. de Totero D, Reato G, Mauro F, Cignetti A, Ferrini S, Guarini A, Gobbi M, Grossi CE, Foa R. IL4 production and increased CD30 expression by a unique CD8+ T-cell subset in B-cell chronic lymphocytic leukaemia. Br J Haematol 1999; 104: 589–599. Mainou-Fowler T, Miller S, Proctor SJ, Dickinson AM. The levels of TNF alpha, IL4 and IL10 production by T-cells in B-cell chronic lymphocytic leukaemia (B-CLL). Leuk Res 2001; 25: 157–163. Shankaran V, Ikeda H, Bruce AT, White JM, Swanson PE, Old LJ, Schreiber RD. IFNgamma and lymphocytes prevent primary tumour development and shape tumour immunogenicity. Nature 2001; 410: 1107–1111. Kaplan DH, Shankaran V, Dighe AS, Stockert E, Aguet M, Old LJ, Schreiber RD. Demonstration of an interferon gamma-dependent tumor surveillance system in immunocompetent mice. Proc Natl Acad Sci USA 1998; 95: 7556–7561. Colgan SP, Morales VM, Madara JL, Polischuk JE, Balk SP, Blumberg RS. IFN-gamma modulates CD1d surface expression on intestinal epithelia. Am J Physiol 1996; 271: C276–C283. Callanan MB, Le Baccon P, Mossuz P, Duley S, Bastard C, Hamoudi R, Dyer MJ, Klobeck G, Rimokh R, Sotto JJ, Leroux D. The IgG Fc receptor, FcgammaRIIB, is a target for deregulation by chromosomal translocation in malignant lymphoma. Proc Natl Acad Sci USA 2000; 97: 309–314. Clynes RA, Towers TL, Presta LG, Ravetch JV. Inhibitory Fc receptors modulate in vivo cytoxicity against tumor targets. Nat Med 2000; 6: 443–446. Svensson L, Aszodi A, Reinholt FP, Fassler R, Heinegard D, Oldberg A. Fibromodulin-null mice have abnormal collagen fibrils, tissue organization, and altered lumican deposition in tendon. J Biol Chem 1999; 274: 9636–9647. Olsen BR. Collagen IX. Int J Biochem Cell Biol 1997; 29: 555–558. Liotta LA, Kohn EC. The microenvironment of the tumour–host interface. Nature 2001; 411: 375–379. de la Fuente MT, Casanova B, Garcia-Gila M, Silva A, GarciaPardo A. Fibronectin interaction with alpha4beta1 integrin prevents apoptosis in B cell chronic lymphocytic leukemia: correlation with Bcl-2 and Bax. Leukemia 1999; 13: 266–274.

42 Lagneaux L, Delforge A, Bron D, De Bruyn C, Stryckmans P. Chronic lymphocytic leukemic B cells but not normal B cells are rescued from apoptosis by contact with normal bone marrow stromal cells. Blood 1998; 91: 2387–2396. 43 Lagneaux L, Delforge A, De Bruyn C, Bernier M, Bron D. Adhesion to bone marrow stroma inhibits apoptosis of chronic lymphocytic leukemia cells. Leuk Lymphoma 1999; 35: 445–453. 44 Warnock RA, Campbell JJ, Dorf ME, Matsuzawa A, McEvoy LM, Butcher EC. The role of chemokines in the microenvironmental control of T versus B cell arrest in Peyer’s patch high endothelial venules. J Exp Med 2000; 191: 77–88. 45 Forster R, Schubel A, Breitfeld D, Kremmer E, Renner-Muller I, Wolf E, Lipp M. CCR7 coordinates the primary immune response by establishing functional microenvironments in secondary lymphoid organs. Cell 1999; 99: 23–33. 46 Kneitz C, Goller M, Wilhelm M, Mehringer C, Wohlleben G, Schimpl A, Tony HP. Inhibition of T cell/B cell interaction by BCLL cells. Leukemia 1999; 13: 98–104. 47 Yellin MJ, Sinning J, Covey LR, Sherman W, Lee JJ, Glickman-Nir E, Sippel KC, Rogers J, Cleary AM, Parker M. T lymphocyte T cellB cell-activating molecule/CD40-L molecules induce normal B cells or chronic lymphocytic leukemia B cells to express CD80 (B7/BB-1) and enhance their costimulatory activity. J Immunol 1994; 153: 666–674. 48 Cantwell M, Hua T, Pappas J, Kipps TJ. Acquired CD40-ligand deficiency in chronic lymphocytic leukemia. Nat Med 1997; 3: 984–989. 49 Khan AA, Bose C, Yam LS, Soloski MJ, Rupp F. Physiological regulation of the immunological synapse by agrin. Science 2001; 292: 1681–1686. 50 Bromley SK, Burack WR, Johnson KG, Somersalo K, Sims TN, Sumen C, Davis MM, Shaw AS, Allen PM, Dustin ML. The immunological synapse. Annu Rev Immunol 2001; 19: 375–396. 51 O’Hagan RC, Schreiber-Agus N, Chen K, David G, Engelman JA, Schwab R, Alland L, Thomson C, Ronning DR, Sacchettini JC, Meltzer P, DePinho RA. Gene-target recognition among members of the myc superfamily and implications for oncogenesis. Nat Genet 2000; 24: 113–119. 52 Lloberas J, Maki RA, Celada A. Repression of major histocompatibility complex I–A beta gene expression by dbpA and dbpB (mYB-1) proteins. Mol Cell Biol 1995; 15: 5092–5099. 53 Hong NA, Flannery M, Hsieh SN, Cado D, Pedersen R, Winoto A. Mice lacking Dad1, the defender against apoptotic death-1, express abnormal N-linked glycoproteins and undergo increased embryonic apoptosis. Dev Biol 2000; 220: 76–84. 54 Kitada S, Andersen J, Akar S, Zapata JM, Takayama S, Krajewski S, Wang HG, Zhang X, Bullrich F, Croce CM, Rai K, Hines J, Reed JC. Expression of apoptosis-regulating proteins in chronic lymphocytic leukemia: correlations with in vitro and in vivo chemoresponses. Blood 1998; 91: 3379–3389. 55 Schweizer A, Rohrer J, Jeno P, DeMaio A, Buchman TG, Hauri HP. A reversibly palmitoylated resident protein (p63) of an ERGolgi intermediate compartment is related to a circulatory shock resuscitation protein. J Cell Sci 1993; 104: 685–694. 56 Braesch-Andersen S, Stamenkovic I. Sialylation of the B lymphocyte molecule CD22 by alpha 2,6-sialyltransferase is implicated in the regulation of CD22-mediated adhesion. J Biol Chem 1994; 269: 11783–11786.

2437

Leukemia