Nucleotide sequence, transcription map, and mutation analysis of the 13q14 chromosomal region deleted in B-cell chronic lymphocytic leukemia

B-cell chronic lymphocytic leukemia (B-CLL) represents the most common leukemia in the Western countries with an estimated incidence of 1 per 100 000 per year. The disease is characterized by the monoclonal expansion of B lymphocytes expressing the CD5 marker and exhibiting a long life span, possibly because of a perturbed apoptotic program.1 

Current knowledge of the molecular pathogenesis of B-CLL is limited because no specific genetic alteration has yet been associated with this disease. In particular, B-CLL is not associated with reciprocal balanced chromosomal translocations.2 Accordingly, none of the proto-oncogenes involved in chromosomal translocations in other mature B cell malignancies, including cyclin D1,BCL-2, BCL-6, PAX-5, andc-MYC, are primarily altered in B-CLL.3,4Although in a fraction of B-CLL cases, inactivation of the tumor suppressor gene p53 (on chromosome 17p13) and deletions or mutations of the ATM gene (on chromosome 11q22-23) have been reported,5-9 such lesions were observed in late stages of the disease and may not represent primary tumorigenic events.

Chromosome 13q14 deletions (approximately 50% of the patients)2,10,11 are the commonest chromosomal alterations associated with B-CLL, followed by structural aberrations of chromosome 11q (19% of the cases) and trisomy of chromosome 12 (15%).2 Moreover, chromosome 13 deletions can be the only cytogenetically detectable abnormality, suggesting an early role in B-CLL pathogenesis. Together, these findings point to the presence in this region of a tumor suppressor gene whose loss or inactivation may be crucial for the leukemogenesis.

Deletions of chromosome 13q14 have also been reported in a variety of human tumors, including various other types of lymphoid tumors12-17 and myeloid leukemias,18 as well as prostate,19,20 head and neck,21 and non-small-cell lung cancers.22 Although the deleted region of chromosome 13q14 has not yet been defined precisely in most of these neoplasms, these observations suggest that a common tumor suppressor gene may reside on this chromosomal segment.

To identify the B-CLL-associated tumor suppressor gene, we previously constructed a high-density contiguous cosmid-based physical map encompassing the deleted interval.10 This map allowed us to demonstrate that somatic loss at 13q14 occurs in 54% of the patients10 (and unpublished results) and that the loss is monoallelic in 81% of such cases and biallelic in the remaining 19%. A minimal deleted region (MDR) in B-CLL, spanning less than 300 kilobase (kb) and representing the site of the B-CLL-associated putative tumor suppressor gene, was assigned on our physical map in the interval between markers 173a12-82 and 138G4/1.3R.10 This interval has also been confirmed by other studies.13 In fact, the MDR exhibits a partial overlap with the regions of deletion as assigned in additional reports,23-26 and it includes a locus of 10 kb that was previously defined as a minimal consensus for 13q14 deletion in B-CLL, based on the pattern of loss in 2 B-CLL cases and one cell line.27 

Here, we present the complete annotated sequence of approximately 347 kb encompassing the MDR and describe a complete characterization of the genes contained in this interval, an analysis of their expression status in B-CLL samples, and the screening of the entire MDR for mutations.

Sixteen overlapping cosmid clones covering the MDR were described previously.10 The nucleotide sequence of these cosmids was obtained by the shotgun DNA sequencing method.28 Briefly, cosmid DNA, purified as described,10 was sonicated or nebulized, and fragments between 1 and 4 kb were selected following agarose gel electrophoresis. The isolated DNA fragments were ligated into the dephosphorylatedHincII or SmaI sites of either M13mp18 or pUC19 and electroporated into DH5α (Life Technologies, Rockville, MD) or XL1Blue (Stratagene, La Jolla, CA) Escherichia coli cells. Plasmid DNA was obtained by using Qiagen Turbo kits on the BioRobot 9600 (Qiagen, Valencia, CA), and single-strand M13 DNA was isolated as described.29 Unidirectional sequencing with an M13 forward primer or bidirectional sequencing with both forward and reverse M13 primers was performed with either FS dye terminator or big dye terminator chemistry (PE Applied Biosystems, Foster City, CA). The products were analyzed, using ABI 377 sequencers. DNA sequences were filtered for vector and E coli sequences, assembled, and edited with the Phred, Cross_Match, Phrap, and Consed software package30-32 (available athttp://bozeman.mbt.washington.edu) on a Sun SPARC20 workstation. Sequence contigs were bridged by “walking,” using custom-generated primers (Life Technologies). Overall, 5.3-fold sequence coverage was achieved. A unique contig of 347 503 base pair (bp) encompassing the region of deletion in B-CLL was then compiled and used as the reference sequence. Homology searches on public databases were performed with the use of BLASTN and TBLASTX,33 after filtering out repeated and vector sequences with RepeatMasker, part of the Phred package.

The entire consensus sequence of 347 503 bp was annotated with homologies to sequences in the public databases, using BLASTN and TBLASTX. In addition, computer-predicted exons were also identified, using the GENSCAN program34 (see the GENSCAN web server at Massachusetts Institute of Technology athttp://ccr-081.mit.edu/GENSCAN.html). A graphical output of these annotations is available at our B-CLL homepage (http//:genome2.cpmc.columbia.edu/∼bcll/).

Standard hybridization screenings using gene-specific probes were performed on complementary (cDNA) libraries generated from various RNAs derived from spleen (Clontech, Palo Alto, CA), placenta (ATCC, Rockville, MD), tonsil (ATCC), the B-cell lymphoma cell line BJAB (in the λZipLox vector [Life Technologies], using the SupersciptII lambda system for cDNA synthesis and cloning [Life Technologies]), and the B-CLL cell line MO104335 (in the λgt11 system [Promega, Madison, WI]). The cDNA clone inserts were amplified by polymerase chain reaction (PCR), using vector-derived primers, gel purified, and extracted with GFX columns (Amersham Pharmacia Biotech, Piscataway, NJ) and then fully sequenced with vector-derived or custom-generated oligonucleotides (Life Technologies).

Full-length cloning of the identified genes was completed, using the Rapid amplification of cDNA ends (RACE) Marathon-Ready cDNA kit (Clontech, Palo Alto, CA). Gene-specific primer pairs for both 3′ and 5′ extension starting from the most 3′ and 5′ cloned sequences were used. The cDNAs used were derived from spleen, lymph node, and testis.

The B-CLL specimens used in this study were diagnosed according to the criteria of the Fifth International Workshop on CLL and the National Cancer Institute-sponsored Working Group36,37 and were characterized by fractionation of peripheral blood from B-CLL patients on a Ficoll-Hypaque gradient (Amersham Pharmacia Biotech) and flow cytometric analysis. The majority of the cells (> 64%; average, 88%) expressed the CD19 and CD20 markers. These cells also expressed the CD5 and CD23 markers as well as low-density monoclonal surface immunoglobulin of the immunoglobulin M (IgM) or IgD (κ or λ) isotypes. The fraction of CD5-positive B cells was between 64% and 96%, with a median of 84%. Normal granulocytes in the postgradient pellet from the same patients were also collected. Various control B-cell lines were used in this study: the B-cell lymphoma-derived BJAB, the cord blood-derived CB33, the lymphoblastoid cell line EREB-E2,38 the B-CLL-derived cell lines MEC1 and MEC2,39 MO1043,35 Handman (G. Inghirami, unpublished), CLL183E95, and WaCaCD5⁺.40Cellular DNA from both tumor samples and cell lines was extracted by cell lysis, proteinase K digestion, “salting out” extraction, and ethanol precipitation, as described.41 

Somatic mutations were assessed on 20 B-CLL patient-derived DNAs displaying hemizygous deletion at 13q14,10 as follows: Exons and flanking intron sequences were amplified by standard PCR reactions in a final volume of 25 μL starting from 100 ng of tumor genomic DNA. An aliquot of the PCR product (2 μL) was examined on an agarose gel and re-amplified, if the total product yield was less than 20 ng. The PCR products were then subjected to direct sequencing (with the same primers used for the amplification reaction, except where indicated), after enzymatic degradation of primers and dNTPs with E coli exonuclease I and shrimp alkaline phosphatase (United States Biochemical, Cleveland, OH) as described.42The sequences of the primers used for the mutational analysis are available at the B-CLL homepage. Each sequence variant detected in a tumor sample was ruled out as a polymorphism by sequencing the same segment on normal case-matched DNA.

A genomic interval of 317 kb encompassing the B-CLL MDR was analyzed by a combination of fluorescence-assisted mismatch analysis (FAMA),43,44 and direct sequencing in 5 B-CLL patients (B-CLL Nos. 19, 22, 31, 33, and 85) carrying a hemizygous deletion at 13q14. The deletion status of B-CLL Nos. 19, 31, and 33 was described previously in detail10; B-CLL Nos. 22 and 85 showed a deletion throughout the 13q14 locus (data not shown). Because these B-CLL samples exhibited, on average, 23% contamination with normal lymphocytes (range, 25% to 19%), the allelic frequency (percentage of tumor-derived allele in the sample) was approximately 62%. The sensitivity of the FAMA method was tested with known mutations and detected a base change even when the frequency of the mutated allele was 15% (data not shown). FAMA was performed as described.43 Briefly, overlapping PCR amplicons of 900 to 1300 bp were generated with chromosome 13-specific primers containing a “universal” tail (see below), starting from 100 ng of tumor DNA under standard PCR conditions (final volume, 12-50 μL). These products were then labeled by PCR (final volume, 50 μL), using universal 5′-fluorescent primers (6-FAM-5′HUP: GGACCGTTAGTAGTCGAC and HEX-3′HUP: GGTCGGATAGCTAGTCGT) that anneal to the chromosome 13-specific primer “tail.” After chemical modification and cleavage, the products were electrophoresed on a 4% polyacrylamide denaturing gel, using an ABI377 sequencer together with standard size markers (TAMRA2500). The results were analyzed with GeneScan software (PE Applied Biosystems). For direct sequencing, the products were amplified under standard PCR conditions in a final volume of 50 μL and were purified by enzymatic degradation of primers and dNTPs as described above. All sequence variants detected by FAMA or by direct sequencing were confirmed on normal DNA from the same patient displaying such nucleotide substitutions. The 13q14 region analyzed for mutations in B-CLL patients was 317 kb (between position 1-316811 of the MDR sequence). The sequences of the primers used for FAMA and direct sequencing are available at the B-CLL homepage.

Total RNA was isolated from tumor samples and cell lines by guanidine isothiocyanate extraction.45Poly(A)⁺ isolation was then accomplished with the PolyATtract system (Promega). Northern analysis was performed by using 25 μg total RNA or 5 μg poly(A)⁺ RNA. Northern blot filters (including the “Immune System” and “Multiple Tissue”) obtained from Clontech (Palo Alto, CA) were also used for analysis. In addition, expression profiles were determined by reverse transcriptase (RT)-PCR assays, using gene- or exon-specific primers. Briefly, cDNAs were generated in a final volume of 50 μL, starting from 2 μg of total RNA that had been treated with RQ1 DNase (Promega, Madison, WI) using SuperscriptII RNaseH⁻ RT (Life Technologies). A control reaction lacking the RT enzyme was always performed in parallel to exclude the presence of contaminant DNA. For amplification, 2 μL of the cDNA were used per PCR reaction with gene-specific primers in a final volume of 20 μL (denaturation at 94°C for 10 seconds; annealing at 56°C to 60°C for 30 seconds; and extension at 72°C for 30 seconds for 30 cycles).

A total of 16 overlapping cosmids,10 spanning the interval bordered by markers WI-10171 and GCT16C05 covering the B-CLL MDR were sequenced to completion. A continuous sequence of 347 503 bp (available at the B-CLL homepage; GenBank accession No. AF272953) was generated. This was achieved by first generating 6 sequence contigs assembled by Phred/Phrap from cosmid shotgun sequences (ranging in size from 10 047-120 816 bp), which were then bridged with sequences derived from PCR products generated on genomic DNA (ranging in size from 150-800 bp). Overall, the calculated sequence coverage was 5.3-fold redundant. In addition, approximately 250 kb of sequence (available on request) was also generated, which covers the region between the MDR telomeric boundary and the marker D13S25 (Figure1).

The average GC content of the MDR region is 39.3%. Two CpG islands (at position 84091-85217 and 126795-129244) were identified, at least one of which is associated with the EST70/Leu1 and1B4/Leu2 genes. Repetitive elements comprised 40.2% of the complete sequence, mainly Alu and LINE1 repeats (12.6% and 14.3%, respectively; see the B-CLL homepage for a summary table and full description of repetitive elements).

A total of 200 nucleotide polymorphisms (SNPs) (average SNP frequency, 1/1400 bp) were also identified in the MDR region by comparing the sequences among 5 B-CLL patients. All of the identified SNPs are reported at the B-CLL homepage, and they have also been deposited in the GenBank SNP database at http://www.ncbi.nlm.nih.gov/SNP/.

Genes located within the B-CLL MDR were identified by homology searches of public databases, analysis of computer-predicted exons, and exon trapping reported previously.10 

The entire MDR consensus sequence of 347 503 bp and the sequence of 250 kb covering the interval between the MDR and the D13S25 marker were used as queries against the GenBank Expressed Sequence Tag database (dbEST). A total of 244 hits were detected, representing 74 unique cDNA clusters (28 within the MDR). These dbEST hits are summarized in a table at the B-CLL homepage and are also shown as part of a graphical map of the region. Of these clusters, only 7 yielded a positive signal in Northern analysis, whereas the remaining EST clusters gave positive signals only when analyzed by RT-PCR in a variety of B-CLL and lymphoid-cell line-derived cDNAs (data not shown), suggesting low levels of expression in the examined tissues. Our attempts to extend these cDNAs by library screening and RACE experiments were negative. Accordingly, these ESTs were considered as products of basal transcription and were not pursued further.

Strong homologies to 2 known genes were found in the region; the MDR segment from 102 255 to 102 851 exhibited 83% homology to the human ribosomal protein L18 gene (GenBank accession No. L11566), whereas the homology between the MDR segment between 174 971 and 176 557 and the progesterone-receptor associated protein p48/HSP70-interacting protein, also named putative tumor suppressor gene ST13 (U28 918 and U17 714), was 96% (Figure 1). In addition, a sequence with 99% homology with the human ribosomal protein L34 gene (L38 941) was observed in the region telomeric to the MDR. Several observations suggest that these sequences represent processed pseudogenes on chromosome 13 because they lack introns, harbor numerous stop codons interrupting the open reading frame of the homologous genes, and lack detectable expression (at least in B cells) when analyzed by RT-PCR with the use of primers specific for these loci.

GENSCAN34 predicts the presence of probable genes and/or exons, based on genomic sequence composition and gene structure features. A total of 22 exons predicted by GENSCAN with medium to very high probability (P > .5) was analyzed for expression. The results of this analysis, including exon location within the framework sequence, are summarized at the B-CLL homepage. Weak RT-PCR signals were detected for 12 exons, but in no case could these exons be extended further by cDNA library screening or RACE experiments. Only one predicted exon, overlapping with known EST sequences, was shown to recognize several transcripts by Northern analysis (see below).

A total of 9 transcripts, corresponding to the 3 genes described below, was identified within the approximate 347-kb interval (Figure 1B and Figure 2) and were fully cloned and characterized.

This gene (GenBank accession No. AF279659), named for “CLL-Associated RING finger” (and previously named Leu546), was identified both by homology searches as well as by GENSCAN prediction, and it includes the WI-10171 EST cluster that we described previously.10 The CAR gene includes at least 5 exons distributed over 21 kb of genomic DNA (bp 207-21543 of the approximate 347-kb MDR consensus sequence, Figure 1B). Different polyadenylation (see the EST hits summary table at the B-CLL homepage), and 3′ exon usage generate 4 transcripts of 1.6, 2.0, 2.3, and 7.5 kb sharing the same coding potential (Figure 1C). Expression of these messenger RNAs (mRNAs) was detected in all tissues tested. Exons 1 to 3, which recognize all 4 messages and correspond to the 1.6-kb messages, and possibly 2.0- and 2.3-kb messages, were found to be linked in dbEST- or library-derived cDNA clones. Exons 4 and 5 detected the 7.5-kb message only.

The predicted protein product of the CAR gene is encoded entirely by exon 3 and consists of 407 amino acids (aa). It belongs to the RING-finger (aa 10-57) B-box (aa 89-131) coiled-coil domain-containing subfamily of RING-finger proteins.47 It is notable that a region included in the 3′ untranslated sequence potentially encodes a 213-aa-long protein containing a BTB/POZ domain (aa 5-106) found in a variety of potassium channel proteins.

The full-length cDNA cloning of this gene (GenBank accession No.AF279658) revealed that it is composed of at least 5 exons spanning approximately 40 kb of DNA (bp 46 433-84 921 of the approximate 347-kb MDR consensus sequence; Figure 1). Two transcripts of 1.3 and 1.9 kb (previously described partially by our group and also by others10 27) are generated by differential polyadenylation of exon 5. Expression was detected in all of the lymphoid tissues examined as well as in fetal liver, bone marrow, and thymus (Figure 2and data not shown). Various alternatively spliced RNA species were also identified in spleen or lymph node (Figure 1C). A minor transcript of 1.0 kb was also detected occasionally, deriving from polyadenylation after the exon 4 sequences (Figure 1C and Figure 2). We note that several additional exons downstream of exon 5 (ie, extending the gene size) were found by 3′ RACE experiments. However, such sequences failed to generate a signal in Northern blot experiments. Therefore, they may belong to very rare transcripts (ie, detectable only by PCR) with questionable physiologic relevance, at least for B cells). Nevertheless, such exons were also included in the mutational analysis of the 1B4/Leu2 gene.

No homology between the 1B4/Leu2 and other genes was detected. This gene may not encode a protein product because only a short (55 aa) open reading frame is detectable that is not conserved in the mouse homolog, despite a high degree of conservation at the nucleotide sequence level (85% identity through exons 1-4; data not shown). This suggests evolutionary conservation of the primary nucleotide sequence independent of coding potential.

The third gene (GenBank accession No. AF279660) located within the B-CLL MDR includes at least 3 exons, spanning approximately 25 kb of genomic DNA (bp 85 201-110 160 of the approximate 347-kb MDR consensus sequence, Figure 1). A transcript of approximately 1.2 kb (described previously10 27) and a less abundant 3.1-kb transcript were detected in all the lymphoid tissues tested (Figure 2and data not shown). In some of the transcripts, exon 2 was excluded by alternative splicing (Figure 1C). Additional exons following exon 3 and further extending the gene were found by 3′ RACE experiments (see “Materials and methods” section; data not shown) but did not detect any transcript when tested by Northern analysis (nonetheless, they were included in the gene mutation analysis).

No homologies with previously identified genes or proteins were identified by BLAST searches with EST70/Leu1 sequences. The longest open reading frame accounts for a putative product of 78 aa (or 68 aa, if exon 2 is included). Southern analysis of mouse genomic DNA with human-specific probes under low stringency conditions (allowing approximately 30% mismatches), as well as mouse database searches using the EST70/Leu1 sequences, suggested that this gene may not be conserved in the mouse genome (data not shown). This gene may also represent a noncoding RNA.

To investigate whether any of the 3 candidate genes was affected by mutations in B-CLL specimens in which one allele was retained, exon and exon/intron junction sequences were analyzed in 20 cases (the complete description of the sequences included in the mutation analysis is available at the B-CLL homepage). No sequence variant was detected, except for SNPs that were also found in normal samples (see SNPs summary at the B-CLL homepage). Absence of mutations affecting some of the EST70/Leu1 and 1B4/Leu2 transcripts in hemizygously deleted patients had also been reported previously.27 

To exclude the possibility that inactivating mutations may target unidentified gene regulatory elements, we searched for somatic mutations in the entire MDR area in B-CLL samples. If additional genes were present within the MDR and had escaped our attention, a complete mutation analysis would permit their recognition. Somatic sequence alterations were sought by FAMA and direct sequencing in 317 kb of the approximate 347-kb region in 5 hemizygously deleted B-CLL patients that displayed unmutated gene sequences (see above). A total of 221 kb were analyzed by FAMA, and 60 kb were analyzed by direct sequencing. A total of 36 kb consisting of low complexity repeated sequences were excluded from the analysis. Several sequence variants (200 in total) distributed along the entire sequence were detected. These variants were classified as SNPs, because they were also found in matching normal control DNA derived from the same patients (see SNPs summary and Bentley48).

The absence of somatic mutations suggests that inactivation of these genes may occur through epigenetic mechanisms, such as DNA methylation, leading to loss of their expression in a tumor. To investigate this possibility, expression of the 3 B-CLL candidate gene transcripts was tested in a panel of normal tissues, tumor cell lines, and B-CLL-derived poly(A)⁺ RNAs (Figure 2). Tumor cases, as well as cell lines representative of the germline or with deleted 13q14 configurations, were included in the analysis. When the levels of transcription of the CAR and EST70/Leu1 genes were examined in nondeleted versus deleted patients, they were comparable to those observed in the positive normal tissues (data not shown and Figure 2), indicating that the retained allele of these genes is apparently normally expressed. Our analysis, however, would not detect 2-fold differences resulting from loss of expression of one allele, because of contaminating normal cells in the samples.

The 1B4/Leu2 gene showed no detectable expression in B-CLL samples, regardless of their 13q14 status, whereas expression was detected in other normal (CB33) or tumor-derived B cells (Figure 2). Low levels of 1B4/Leu2 transcription, comparable to the level detected in CB33 cells, were detected in Epstein-Barr virus-immortalized B-CLL cell lines, including one hemizygously deleted sample (MEC1; Figure 2). This finding is in contrast to what was observed in the primary tumor samples and may reflect peculiar features of the exceedingly small fraction of CLL cases that can grow in vitro, or gene expression changes associated with the immortalization process.

In an effort to identify the B-CLL-associated tumor suppressor gene located on chromosome 13q14, we analyzed approximately 347 kb of completely annotated sequence encompassing the B-CLL MDR and characterized the genes located in this region. We have also examined their mutational and expression status in B-CLL. Moreover, we investigated the mutational status of the entire MDR interval in B-CLL patients.

The 3 genes described here may represent all the active transcriptional units present in the MDR in B cells. Although the transcriptional characterization of the MDR through a combination of database homology searches, exon trapping, and computer predictions revealed a total of 74 unique EST clusters, the majority (67) of these sequences were not bona fide transcripts either because of lack of expression (at least in B cells) or because they represented “noise” dbEST hits (ie, clones derived from unprocessed nuclear RNA fragments or from contaminating genomic DNA primed at poly(A) stretches within low complexity regions, such as Alu elements, etc). The remaining 7 EST clusters recognized RNA species on Northern blots. These corresponded to a RING finger-encoding gene (CAR) and 2 apparently noncoding genes (1B4/Leu2 and EST70/Leu1). Characterization of these genes revealed several novel transcripts described here for the first time (detailed in Figure 1C, rows c-f).

On the basis of its structural features and the well-proven link between RING-finger containing proteins and tumorigenesis,47 the CAR gene was a credible candidate for the B-CLL-associated tumor suppressor. However, no alterations were discovered in the gene exon or exon-intron junction sequences in 20 tumor cases, indicating that this gene is not affected by a typical mechanism involving loss of one allele and mutation of the retained allele. In addition, when transcription of CAR was tested in B-CLL samples with loss of 13q14, we found that the wild-type nondeleted allele was expressed at levels comparable to normal cell line controls, although a 50% reduction would not be detectable by our analysis. Finally, we observed that some B-CLL patients with 13q14 deletion retained the CAR gene (Figure 1B). On the basis of these results, the CAR gene is unlikely to represent the B-CLL-associated tumor suppressor gene.

The EST70/Leu1 locus probably is a new member of a growing family of noncoding RNA genes. Such genes exert their function through their RNA product and play important roles in various cellular processes, including transcriptional (and posttranscriptional) regulation or regulation of enzymatic activities.49-51 A link between sterile RNA transcripts and tumorigenesis has also been proposed in the case of the H19 and tropomyosin genes.52,53 Despite the fact that B-CLL samples with monoallelic deletion at 13q14 invariably lose one allele of theEST70/Leu1 gene, the retained allele lacks somatic mutations, indicating that the EST70/Leu1 gene cannot be implicated in B-CLL pathogenesis through a classic 2-hit inactivation mechanism. Because transcription of the retained unmutated alleles is maintained in the tumors, it remains possible that haplo-insufficiency may contribute to B-CLL development, in a way analogous to other tumor suppressor genes in the mouse.54 55 

It is likely that the 1B4/Leu2 gene also produces a sterile transcript. 1B4/Leu2 is highly conserved in the mouse genome, strongly supporting its functional significance.1B4/Leu2 sequences are not mutated in B-CLL, but, in contrast to CAR and EST70/Leu1, no expression was detected in the B-CLL cases tested, regardless of the presence or absence of chromosome 13q14 deletion. It is possible that the lack of expression of 1B4/Leu2 in B-CLL may reflect the lack of transcription of this gene in the normal cellular counterpart of B-CLL. This possibility is difficult to address because various human lymphocyte subpopulations are heterogeneous for 1B4/Leu2expression and the normal counterpart of B-CLL is currently unknown.1 However, epigenetic silencing may be tumor associated and, in alternative to or together with gene loss, contribute to the lack of 1B4/Leu2 expression in all B-CLL cases. Thus, this gene remains a candidate for the B-CLL associated tumor suppressor.

Two groups23,26 reported exclusion of theEST70/Leu1 and 1B4/Leu2 genes from the deleted interval in 3 B-CLL cases when analyzing a total of 39 tumor samples with deletion at 13q14. However, this result was never observed by us and by 2 other groups10,13,25 when mapping the deleted interval in more than 300 B-CLL samples with 13q14 deletion. This discrepancy could either reflect different technical approaches used in mapping the deleted area (PCR23,26 versus Southern blotting hybridization and/or fluorescence in situ hybridization10,13 25) or point to the involvement of a second tumor suppressor locus in a small minority of B-CLL cases with deletion at 13q14.

In conclusion, the results presented here leave open several possibilities on the underlying pathogenetic mechanisms in B-CLL-associated deletions. First, it remains formally possible that 13q14 deletions are not of pathogenetic significance in CLL, but rather they represent epiphenomena in the disease. However, the presence of these lesions as sole cytogenetic abnormality, their high frequency, and their clonal nature strongly argue in favor of their selection during tumorigenesis. Second, the B-CLL-associated gene is within the MDR but has escaped detection because of its atypical features, including unusually large introns potentially spanning the entire MDR, lack of homologous sequences in the public domain, or lack of polyadenylation, as in the case of small nucleolar RNA genes (not represented in the dbEST collection of cDNAs). Alternatively, reduced dosage of the product of one or more of the candidate genes may be sufficient to contribute to B-CLL development. Reduced dosage can occur as a result of monoallelic loss in the case of EST70/Leu1, or as a consequence of allelic loss or transcriptional silencing in the case of 1B4/Leu2. The complete MDR sequence presented here will be useful in the continuing search for transcriptional units within this region and for the design of gene targeting experiments that will explore the role of haplo-insufficiency in mice.

We thank Andrey Rzhetsky for discussions during various phases of the project; Jing Chen, Hye Park, Juyan Zhou, Gifty Asamani, Kendra Hill, and Vladan Miljkovic for expert technical assistance.

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under Accession Nos. AF272953 (MDR sequence), AF279658 (1B4/Leu2), AF279659 (CAR), and AF279660 (EST70/Leu1). The B-CLL homepage can be found at http://genome2.cpmc.columbia.edu/∼bcll.