Distinct gene expression profiles in subsets of chronic lymphocytic leukemia expressing stereotyped IGHV4-34 B-cell receptors

Abstract

Background Numerous subsets of patients with chronic lymphocytic leukemia display similar immunoglobulin gene usage with almost identical complementarity determining region 3 sequences. Among IGHV4-34 cases, two such subsets with “stereotyped” B-cell receptors were recently identified, i.e. subset #4 (IGHV4-34/IGKV2-30) and subset #16 (IGHV4-34/IGKV3-20). Subset #4 patients appear to share biological and clinical features, e.g. young age at diagnosis and indolent disease, whereas little is known about subset #16 at a clinical level.Design and Methods We investigated the global gene expression pattern in sorted chronic lymphocytic leukemia cells from 25 subset/non-subset IGHV4-34 patients using Affymetrix gene expression arrays.Results Although generally few differences were found when comparing subset to non-subset 4/16 IGHV4-34 cases, distinct gene expression profiles were revealed for subset #4 versus subset #16. The differentially expressed genes, predominantly with lower expression in subset #4 patients, are involved in important cell regulatory pathways including cell-cycle control, proliferation and immune response, which may partly explain the low-proliferative disease observed in subset #4 patients.Conclusions Our novel data demonstrate distinct gene expression profiles among patients with stereotyped IGHV4-34 B-cell receptors, providing further evidence for biological differences in the pathogenesis of these subsets and underscoring the functional relevance of subset assignment based on B-cell receptor sequence features.

Introduction

The somatic hypermutation status of the immunoglobulin heavy variable (IGHV) genes is an independent prognostic marker in chronic lymphocytic leukemia (CLL) and subdivides patients into two subgroups with patients carrying unmutated IGHV genes having a poor prognosis.1,2 It is now well established that CLL displays a remarkably biased IGHV gene repertoire with over-representation of a limited number of genes such as IGHV1-69, IGHV4-34, IGHV3-23 and IGHV3-21.2–4 Furthermore, several groups have reported the existence of multiple CLL subsets with “stereotyped” B-cell receptors (BCR), with similar IG heavy and light-chain gene usage and almost identical complementarity determining region 3 (CDR3) sequences, in up to 30% of patients.5–9 These findings provide strong evidence for the involvement of antigens in the development of CLL. In addition, a number of studies have also indicated that stereotyped BCR may influence the clinical course in CLL.8,10

The IGHV4-34 gene is detected in approximately 10% of CLL cases and patients carrying this rearrangement generally have highly mutated IGHV sequences and a favorable prognosis.8 Recently, two relatively large subsets carrying stereotyped IGHV4-34 BCR were reported in CLL. The more common one, occurring at an overall frequency of approximately 1%, is designated as subset #4 and is characterized by IGHV4-34 rearrangements with highly homologous, 20 amino-acid long heavy variable CDR3 (VH CDR3), exclusive IGKV2-30 gene usage and uniform IgG-switching.8 Patients belonging to this subset also have an early age of onset of disease (median age at diagnosis, 43 years) and an indolent disease course, even when compared to cases expressing mutated IGHV4-34 rearrangements with heterogeneous VH CDR3.8

The second subset, occurring at an overall frequency of 0.3%, is known as subset #16 and is characterized by IGHV4-34 gene rearrangements with a distinct VH CDR3 of 24 amino acids in length, combined with IGKV3-20 gene usage. Little is known regarding the clinical outcome of this subset, probably because of the limited number of patients identified to date.8 Intriguingly, however, we recently found that this subset had a different spectrum of genomic aberrations compared to those of subset #4, even though both subsets consist of cases expressing mutated IGHV4-34 BCR.11

In contrast to the more favorable outcome of IGHV4-34-expressing cases, we previously showed that patients carrying IGHV3-21 BCR have a poor prognosis irrespective of the IGHV gene mutational status.4 In a study on global expression profiling, patients with IGHV3-21 had a different gene expression pattern from that of non-IGHV3-21 patients.12 Genes involved in DNA replication/cell cycle control, transcription and protein kinase activity were found to be differentially expressed, which may lead to a higher rate of proliferation and hence underlie the poor outcome of the patients with IGHV3-21.12

In the present study, we examined the gene expression profile of 25 IGHV4-34 patients belonging to subset #4, subset #16, or the heterogeneous non-subset 4/16 group with the aim of exploring whether the clustering of cases based on structural features of the antigen-binding site is also reflected in distinct “subset-specific” gene expression profiles.

Design and Methods

Patients’ material

Tumor samples, derived from peripheral blood, were collected from 25 CLL patients known to express the IGHV4-34 gene for microarray expression analysis. The patients were untreated at the time of sample collection and came from collaborating institutes in France (n=3), Greece (n=10) and Sweden (n=12). Samples were classified according to recently revised criteria and displayed the typical CLL immunophenotype.13 Based on IG gene sequence features and following previously established criteria,8,9 cases were classified into three groups: (i) subset #4: 11 cases with IGHV4-34/IGKV2-30 usage, 20 amino acid long VH CDR3; (ii) subset #16: 5 cases with IGHV4-34/IGKV3-20 usage, 24 amino acid long VH CDR3; and, (iii) non-subset 4/16: 9 cases with IGHV4-34 BCR of varying VH CDR3 length and composition as well as heterogeneous light-chain usage. In keeping with previous reports, all subset #4 cases as well as three subset #16 cases with available information expressed IgG; this information was not available for non-subset 4/16 cases. Although the non-subset 4/16 group included two cases with unmutated BCR, the average IGHV4-34 gene mutational load in the three groups did not differ. The median age at diagnosis was 55 years (range, 37–73 years) for subset #4, 70 years (range, 55–81 years) for subset #16 and 60 years (range, 45–69 years) for non-subset 4/16 patients. A gender imbalance was observed for subset #16, in that it was primarily composed of women (male:female ratio: 1:4), in contrast to what is seen in CLL in general. Furthermore, two additional subset #16 cases (both IgG-switched) were collected for real-time quantitative polymerase chain reaction (RQ-PCR) validation. The clinical data and molecular characteristics of all patients are summarized in Table 1 and Online Supplementary Table S1. Informed consent was obtained according to the Helsinki declaration and the study was approved by the local Ethics Review Committees.

Isolation of chronic lymphocytic leukemia cells and RNA extraction

CLL cells were isolated through negative depletion of non-tumor cells using the Dynal B-Cell Negative Isolation Kit (Invitrogen, Carlsbad, CA, USA). The proportion of tumor cells following isolation was studied by FACS analysis of the cell surface markers CD5 and CD19 and was verified to be 93% or greater (median 98%, Table 1). RNA was extracted from isolated CLL cells using the RNeasy Mini kit (Qiagen, Hilden, Germany). The integrity of the RNA was evaluated using the Agilent 2100 Bioanalyzer system (Agilent Technologies Inc, Palo Alto, CA. USA) and its concentration measured with an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

Microarray expression analysis

A total of 100 ng RNA from each sample was used to prepare biotinylated fragmented complementary RNA (cRNA) with a two-cycle amplification step, according to the GeneChip Expression Analysis Technical Manual (Rev. 5, Affymetrix Inc., Santa Clara, CA, USA). The cRNA was hybridized to Affymetrix Human Genome U133 Plus 2.0 GeneChip expression arrays for 16 h in a 45°C incubator with rotation at 60 rpm. The arrays were then washed and stained using the Fluidics Station 450 and finally scanned using the GeneChip Scanner 3000 7G.

Microarray data analysis

The gene expression data were subsequently analyzed in R (http://www.r-project.org) using packages available from the Bioconductor project (www.bioconductor.org). The raw data were normalized using the robust multi-array average method.14,15 In order to search for differentially expressed genes between subset #4, subset #16 and non-subset 4/16 samples, an empirical Bayes moderated t-test was applied using the ‘limma’ package.16,17 Only probe sets with an average intensity at reliable levels (log₂ value of >5) in at least one subgroup of patients (subset #4, subset #16 and non-subset 4/16) in each pair wise comparison were used for subsequent analyses. To address potential problems with multiple testing, the P values were adjusted using the method of Benjamini and Hochberg.18 Genes with an adjusted P value less than 0.05 and an average fold change of at least 1.5 were regarded as differentially expressed.

Results were visualized by heat-maps and dendrograms using Genesis TreeView version 1.2.7 (www.genome.tugraz.at). Gene Ontology Tree Machine (GOTM) software was used to identify significant gene ontology categories in the data set (http://bioin-fo.vanderbilt.edu/gotm). The expression data were further analyzed using ingenuity pathway analysis in order to determine significantly deregulated genes and pathways (Ingenuity® Systems, Mountain View, CA, USA; www.ingenuity.com).

Real-time quantitative reverse transcriptase polymerase chain reaction analysis

In order to validate gene expression data obtained from the microarray analysis, we selected three genes (IL15, HOXA1 and ZHX1) for RQ-PCR analysis. Primers were designed using the Primer3 software (Broad Institute, Boston, USA) (sequences are available upon request). RQ-PCR analysis was performed using Maxima SYBR Green Master Mix according to the manufacturer’s protocol (Fermentas, Burlington, Canada). Beta-actin expression was used as an internal reference. Differences in expression between subsets were evaluated using student’s T-test and presented in box plots using the Statistica 8.0 software (Stat Soft, Tulsa, OK, USA).

Table 1.Clinical and molecular data for patients included in the study.

Results

Expression profiling of IGHV4-34-expressing chronic lymphocytic leukemia

We performed high-resolution Affymetrix gene expression arrays on 25 sorted IGHV4-34-expressing samples with the aim of identifying biological features capable of distinguishing the different IGHV4-34 subsets. After robust multi-array average normalization and log transformation, we detected few significantly differentially expressed genes when studying subset #4 and non-subset 4/16 cases (3 genes) and similarly when comparing subset #16 patients and non-subset 4/16 patients (10 genes) (data not shown). In contrast, we identified 111 significantly differentially expressed genes, when comparing subsets #4 and #16 (fold change differences are indicated in Online Supplementary Table S2). We then clustered patients according to their gene expression profiles for the 111 genes that displayed differential expression and could successfully distinguish subset #4 from subset #16 patients (Figure 1), although two subset #4 patients had an intermediate gene expression profile. The results revealed an overall lower gene expression profile for subset #4 compared to subset #16 (Figure 1 and Online Supplementary Table S2).

Next, we evaluated differences in gene expression between subset #4 cases and all other cases, i.e. subset #16 and non-subset 4/16 cases, identifying 14 differentially expressed genes of which 9 were also present in the initial list of 111 genes. These included genes involved in cell cycle regulation, such as TLK1, as well as genes implicated in apoptosis, e.g. RPS27L.19,20 Figure 2 illustrates the gene expression profiles for these two groups, where notably, subset #4 cases again appear to have a lower expression of these genes.

Biological annotation of identified genes

In order to determine the biological relevance of the differentially expressed genes, we performed gene ontology enrichment analysis using Gene Ontology Tree Machine software. Here, we focused on the 111 genes that differed in expression between subset #4 and subset #16. Categories that showed a significant enrichment for these are shown in Table 2. Briefly, we found significant over-representation of genes affecting proliferation, cell cycle control and regulation of transcription activity including STOML2, PPP2CA and HOXA1.21–23

Table 2.Gene ontology categories enriched for differentially expressed genes in subset #4 versus subset #16.

To further analyze our data, a second search using ingenuity pathway analysis was performed, again focusing on genes that differed in expression between subset #4 and subset #16, and identified a number of gene networks which were significantly enriched. These were classified as follows and are listed in Table 3: (i) cellular growth and proliferation (16 genes); (ii) humoral immune response (12 genes); (iii) cell-mediated immune response (13 genes); and (iv) hematopoiesis (7 genes).

Figure 1.Visualization of differentially expressed genes using hierarchical clustering of genes in patients in subset #4 versus subset #16.

Several of the differentially expressed genes studied, displaying lower expression in subset #4, have previously been implicated in cancer and cancer-related pathways where tumorigenesis is most often a consequence of over-expression. However, none of these has previously been associated with CLL. These genes are summarized in Online Supplementary Table S3 and categorized as follows: (i) tumor proliferation; (ii) PI3K/AKT/NF-κB pathways; (iii) the p53 pathway; (iv) viral adhesion and replication in host cell; (v) apoptosis; and (vi) cancer therapy.

Figure 2.Visualization of differentially expressed genes using hierarchical clustering of genes in subset #4 versus all other cases.

Confirmation of array data using real-time quantitative polymerase chain reaction analysis

To confirm the array data, we selected three differentially expressed genes in functionally relevant pathways, IL15, a cytokine involved in immune response,24–26 HOXA1, a DNA binding transcription factor,23 and ZHX1, a regulator of transcription,27 for RQ-PCR analysis in seven subset #16 cases, of which two had not been analyzed using gene expression microarrays, and seven subset #4 cases. All genes analyzed validated the results from microarray analysis, having significantly higher expression in subset #16 cases (Figure 3).

Discussion

In the present study, we examined the gene expression profile of 25 IGHV4-34 patients including subset #4, #16 and non-subset 4/16 cases. Initially, we compared the gene expression profiles between subset #4 and non-subset 4/16 patients and between subset #16 patients and non-subset 4/16 patients, and detected only few significant differences. This is probably because, overall, non-subset 4/16 IGHV4-34 cases exhibited a more heterogeneous gene expression profile, likely reflecting the structural heterogeneity of their BCR, which would be expected to be responsive to a far wider range of antigens than that recognized by stereotyped subsets. Interestingly, however, we detected distinct differences in gene expression patterns when comparing subset #4 and #16 cases, both of which can be reliably defined at the molecular level based on subset-specific VH CDR3 and subset-biased features of somatic hypermutation.8 This finding is supported by the recent observation that stereotyped subset cases have consistent antigen reactivity profiles.28 In addition, it may be considered as further evidence that the clustering of cases based on IG primary sequences is biologically and, very likely, clinically relevant.

Intriguingly, subset #4 patients consistently exhibited a lower expression of the genes identified. Detailed characterization of the differently expressed genes in biological processes, using alternative approaches such as gene ontology tree machine and ingenuity pathway analysis, revealed that these genes are involved in cell cycle control and proliferation, such as STOML2 and PPP2CA,21,22 suggesting that these tumors have a low proliferative capacity compared to other CLL subsets, including subset #16.29 This is supported by the identification of a subset of genes involved in hematopoiesis as well as humoral and cell-mediated immune responses (Table 3). IL15 is one such example and acts as a pleiotropic cytokine shown to induce the proliferation of natural killer (NK) cells, B cells, and interferon-producing killer dendritic cells.24–26

Several of the differentially expressed genes detected to have consistently lower expression in subset #4 have previously been implicated in cancer and cancer-related pathways (Online Supplementary Table S3). In almost all cases, over-expression of these genes was associated with tumorigenesis. For instance, over-expression of SSFA2 has been found to promote tumorigenicity in multiple myeloma,30 KIF11 is over-expressed in several cancers and its inhibition has been shown to induce cell cycle block and cell death via the mitochondrial pathway in acute myeloid leukemia cells,31 ARF6 is over-expressed in breast cancer and has been shown to be involved in metastatic cancer32 and PPP2CA is over-expressed in breast tumors and implicated in the negative control of cell growth and division.22 The finding that subset #4 cases display a distinctly lower gene expression profile may partly explain the favorable prognosis of these cases although the impact of individual genes or pathways in CLL is difficult to discern and needs to be studied in more detail.

Furthermore, when comparing expression profiles for subset #4 with those of other IGHV4-34-expressing cases (including subset #16), we detected differences in expression levels for only a few genes, again showing low expression in subset #4 (Figure 2). Several of these genes are important in tumorigenesis, including RPS27L and INHBC, which play important roles in p53 regulation and oncogenic transformation.20,33 Nevertheless, when subset #16 cases were pooled with non-subset 4/16 cases, the observed gene expression differences between subset #4 and #16 were “masked”, probably because of the diverse BCR expressed by the latter cases, as discussed above. This result may justifiably be taken to imply that the comparison between distinct CLL subsets defined by BCR structural features is a pre-requisite for ensuring meaningful biological conclusions on the mechanisms underlying the selection and clonal expansion of CLL progenitors with distinctive BCR. Admittedly, this task will not be easy, given that, individually, each subset accounts for only a small fraction of a CLL cohort. Eventually, however, it may elucidate the pathogenesis of a sizable proportion of CLL cases, since, collectively, the stereotyped subsets account for up to 30% of a cohort.

Table 3.Enriched ingenuity pathway analysis (IPA) categories including differentially expressed genes between subset #4 and subset #16.

Although the experimental platforms differed significantly, the findings of the present study are consistent with those of our recent work, which showed that subset #4 cases carry remarkably few genomic aberrations, again indicating that the CLL cells are less proliferative. In contrast, subset #16 cases displayed several known recurrent aberrations including the poor-prognostic deletion of 11q and trisomy 12.11 We consider these findings as different pieces of the biological puzzle underlying the indolent clinical behavior of subset #4 cases. This is also reflected by the fact that only 18% (2/11) subset #4 patients required subsequent treatment while 60% (3/5) subset #16 patients were later treated.

Contrary to the findings for subset #4, we earlier reported a distinct gene expression profile in the group of patients with aggressive CLL carrying IGHV3-21 BCR. In particular, we observed up-regulation of several genes involved in DNA replication/cell cycle control, transcription and protein kinase activity in IGHV3-21 CLL cases compared to in non-IGHV3-21 CLL cases, indicating a proliferative phenotype, which might account for the more aggressive clinical course.12 This is in turn supported by the high number of copy number alterations detected in this subset.11 That notwithstanding, both the gene expression and genomic profiles in poor-prognostic IGHV3-21 patients and good-prognostic subset #4 IGHV4-34 patients allude to the involvement of different immune-mediated pathways in the development of subgroups of CLL carrying distinct IG molecular features and, perhaps, distinct antigen reactivity and BCR signaling activity.

The IGHV4-34 gene encodes antibodies which are intrinsically autoreactive due to recognition of the N-acetyl-lactosamine epitopes that are present on both various self-antigens (I/i blood group antigen, B-cell isoform of CD45) and microbial pathogens (Epstein-Barr virus, cytomegalovirus, Mycoplasma pneumoniae).34,35 The IGHV4-34 gene is used at a high frequency in healthy individuals; however, IGHV4-34 antibodies are virtually undetectable in healthy sera.36 In contrast, IGHV4-34 antibodies are secreted at a high level in patients with systemic lupus erythematosus and in response to acute infections with herpes viruses and M. pneumoniae.37–41 It has, therefore, been proposed that IGHV4-34 B cells may normally be maintained in a state of diminished responsiveness, sparing the host from potential (auto)immune-mediated pathologies. In line with this, most CLL clonotypic IGHV4-34 rearrangements are mutated, indicating that these CLL cells do not behave in an inherently different way from non-CLL IGHV4-34 B cells. However, we previously noted that subset #4 may be distinguished from other IGHV4-34 CLL by a distinctive somatic hypermutation profile, indicative of selective (super)antigenic processes.9,42 Perhaps relevant to this distinctive somatic hypermutation profile, we obtained evidence of a possible link between persistent infections with Epstein-Barr virus and cytomegalovirus and subset #4 CLL.43 Additionally, we recently reported that subset #4 cases display extensive intraclonal diversification through precisely targeted somatic hypermutation, possibly as a result of ongoing antigenic stimulation, providing support for these findings.44,45 Altogether, these observations could be taken to imply that though still actively recognizing antigen(s), subset #4 CLL cells are maintained in a state of diminished responsiveness, perhaps due to as yet unidentified mechanisms affecting signal transduction and eventual biological effects. The elucidation of these mechanisms will not be easy; however, it could be argued that the gene expression results reported here have offered useful suggestive evidence that might assist in designing further research in subset #4.

Figure 3.RQ-PCR validation of array data. Relative expression data presented as box plots for the following genes: IL15, HOXA1 and ZHX1. Seven subset #16 and seven subset #4 cases were investigated.

In conclusion, we show here that the gene expression profiles for subset #4 and subset #16 IGHV4-34-expressing CLL cases are distinctly different. Although the number of cases studied was relatively small because of the low overall frequency of patients in the two subsets, our findings provide evidence for differences in the underlying biological mechanisms in these subsets, particularly those involved in tumor maintenance and possibly even malignant transformation. RQ-PCR analysis also confirmed the microarray data for selected genes, validating the use of microarray analysis for gene expression profiling. The differentially expressed genes were found to be involved in important cell regulatory pathways including cell cycle control and proliferation and were commonly found to have lower expression in subset #4 cases which may reflect the lack of genomic complexity and indolent phenotype in this group of patients. Finally, we showed that non-subset 4/16 cases have a heterogeneous gene expression profile, which highlights the need for further sub-characterization of CLL in order to identify differences between biologically distinct subgroups of CLL.

Footnotes

• ↵* These authors contributed equally.
• Funding: this work was supported by the Swedish Cancer Society, the Swedish Research Council, the Medical Faculty of Uppsala University, Uppsala University Hospital, and Lion’s Cancer Research Foundation in Uppsala, Sweden.
• The online version of this article has a Supplementary Appendix.
• Authorship and Disclosures The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
• Financial and other disclosures provided by the authors using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are also available at www.haematologica.org.

• Received June 4, 2010.
• Revision received August 17, 2010.
• Accepted August 25, 2010.