Lymphomas arising in mucosa-associated lymphoid tissue (MALT) are indolent B-cell tumors that have a predilection for epithelial sites and often develop in a setting of chronic inflammation or autoimmunity. As many as 50% of low-grade MALT lymphomas contain an (11;18)(q21;q21) chromosomal translocation. Using fluorescence in situ hybridization, we have analyzed the position of recombination within chromosome 18 DNA in three examples of MALT lymphoma bearing this translocation. In all three cases, the breakpoint maps to DNA in BAC b357H2, covering about 150 kb of sequence. A previously undescribed, ubiquitously expressed gene, which we refer to as MALT1, was identified within this sequence and was found to be broken in one case for which we have definitively located the position of recombination between chromosomes 18 and 11. The sequence of this gene indicates the presence of two immunoglobulin-like C2 domains and a region of partial homology to caspases, suggesting a possible role for MALT1 in the regulation of apoptosis.
Lymphoma arising in MALT4 is a recently recognized subtype of B-cell lymphoma with a number of unusual biological and clinical features (1). These features include the frequent origin in or close to mucosae or epithelia of a variety of organs at tissue sites that normally lack large amounts of lymphoid tissue. Additionally, rather than being composed of morphologically homogeneous cells like most other lymphomas, MALT lymphomas at early stages contain a diversity of lymphoid cell types, including small lymphocytes, centrocyte-like cells, monocytoid B cells, and plasma cells, all of which have been shown to be part of the same neoplastic clone (2, 3, 4). Also unlike other lymphomas, the neoplastic cells making up MALT lymphomas often remain confined to the primary site for relatively long periods of time–a characteristic that contributes to the indolent behavior of the tumor and presumably accounts for reported surgical cures in the older literature (5). However, even after the tumor disseminates to other sites, the predilection for epithelial involvement may be retained, at least for some time after the initial spread (3).
Another unusual feature of MALT lymphomas is that these neoplasms arise frequently at sites of preexisting inflammation, probably as an outgrowth of a neoplastic clone from among reactive lymphocytes. For example, MALT lymphomas develop in tissues affected by autoimmune processes, as in Hashimoto’s thyroiditis (6, 7) and Sjögren’s syndrome (8), and they are responsible for the high incidence of lymphoma among patients with these disorders. Furthermore, chronic bacterial infections have been implicated in the development of MALT lymphomas. Gastritis containing the bacterium Helicobacter pylori apparently predisposes to MALT lymphoma of the stomach (9), and chronic bacterial infections of the small intestine are believed to have an etiological role in Mediterranean lymphoma or α heavy chain disease–clinical presentations of lymphomas now regarded as examples of MALT lymphoma (10, 11). Evidence for a direct causal relationship between infection by H. pylori and MALT lymphoma comes from experiments in which cultured lymphoma cells proliferated in response to bacterial antigens only when the antigens were derived from the same strain of H. pylori as was present in the tissues of the particular tumor from which the culture was prepared (12). Consistent with these observations is that antibiotic therapy of patients with gastric MALT lymphomas in early phases of the disease has led to complete clinical remissions in the majority of patients treated (5).
Insights into the mechanisms underlying the development of MALT lymphoma could help to explain the multiple unusual features observed in this disease, as well as to improved diagnosis and treatment of the tumor. Although antigenic stimuli, such as those resulting from bacterial products or autoantigens, may promote growth of some MALT lymphomas, development of MALT lymphoma very likely requires specific alterations in gene structure and function that drive proliferation of the neoplastic cells. Sites of recombination in recurrent chromosomal rearrangements have often provided clues to the location of genes that have undergone structural or functional changes critical to the malignant phenotype. Several cytogenetic aberrations have been identified within MALT lymphomas, the most common of which appears to be the (11;18)(q21;q21) chromosomal translocation present in about 50% of low-grade MALT tumors (13, 14). To search for a gene or genes involved in the pathogenesis of MALT lymphomas, we have analyzed DNA surrounding the 18q21 breakpoint in cells from tumors with this translocation and found that recombination occurs in or around the DNA of a previously undescribed gene having partial homology to genes encoding members of the immunoglobulin gene superfamily and caspase enzymes.
MATERIALS AND METHODS
Tissues and Cells.
Tissues from three cases of MALT lymphoma confirmed by cytogenetic analysis to have the t(11;18)(q21;q21) were used in these studies. In case 1, the tissue was obtained from a pulmonary lobectomy; in case 2, it was obtained from a gastrectomy; and in case 3, it was obtained from a biopsy of an orbital mass. In each case, invasion of epithelial structures (lymphoepithelial lesions) was observed in histological sections of the tissues, and the neoplastic lymphocytes had a mature immunophenotype with expression of immunoglobulin restricted to a single light chain type and absence of CD5 and CD10.
A cell line, MBCL-1, that was previously established after infection of lymphocytes from a MALT lymphoma of the stomach with EBV5 was also used in these studies. Both the tumor cells and the cell line displayed a karyotype with trisomy 18 but lacked the t(11;18)(q21;q21). This line had been carried continuously in culture for at least 6 months before these studies.
Hybrid cell lines derived from fusions of CHO cells with human cells carrying human chromosome 11 (CHO+11) or human chromosome 18 (CHO+18) as the only human chromosomal constituents (Coriell Cell Repositories, Camden, NJ) were grown in culture according to the conditions recommended by the repository. Total genomic DNA was extracted from CHO, CHO+11, and CHO+18 cell lines using standard procedures (15) and was used as a template for PCR.
YACs, BACs, and ESTs.
YACs were obtained from Centre d’Etude du Polymorphisme Humain megaYAC libraries purchased from Research Genetics (Huntsville, Alabama). YAC clones were expanded in SD media, and the DNA was isolated following standard protocols (16).
BACs were selected from three-tiered pools of BAC DNA contained within the California Institute of Technology BAC library (Research Genetics) by PCR using primers complementary to DNA in the STSs D18S831, D18S1129, and D18S1026 (17). Subsequent selections of BACs were performed by PCR amplification using primers based on the nucleotide sequence obtained at the ends of human genomic DNA inserts in previously identified BACS. Reaction conditions for these amplifications were performed as previously described, with 30 cycles of amplification (94°C denaturation, 55°C annealing, and 72°C extension) using minor variations of annealing temperature specific to each primer pair (18). BAC clones were expanded in Luria-Bertani media containing chloramphenicol (25mg/ml; Ref. 15), and the DNA was isolated using the Plasmid Midi-kit (Qiagen, Valencia, CA) following the manufacturer’s instructions.
ESTs were purchased from Genome Systems (St. Louis, MO). Plasmids were expanded in Luria-Bertani media supplemented with ampicillin (100mg/ml), and DNA was purified following standard plasmid preparation protocols (15).
Fluorescent in Situ Hybridization.
DNA probes for FISH were labeled with digoxigenin- (Boehringer Mannheim, Indianapolis, IN) or biotin-tagged nucleoside triphosphates (Life Technologies, Gaithersburg, MD) using the BioPrime labeling system (Life Technologies). Hybridization was performed on slides of intact nuclei isolated from tissue sections as previously described (19, 20). Slides were treated with 10 mg/ml RNase in 2× SSC for 60 min at 37°C followed by 50 mg/ml pepsin in 0.01 n HCl for 8 min at 37°C. Material on the slides was cross-linked in 10% buffered neutral formalin for 2 min at room temperature and dehydrated in graded alcohols. DNA on the slides was denatured in 70% formamide/2× SSC for 4 min at 72°C and dehydrated in graded alcohols at 4°C. A mixture of probe and COT-1 DNA (Life Technologies) was applied in Hybrisol VI (Oncor, Gaithersburg, MD) and denatured at 80°C for 10 min. Probe was allowed to anneal overnight at 37°C and washed in 0.5× SSC for 5 min at 70°C. Slides were rinsed in PN buffer (0.5 m Na2HPO4 to pH 8.0 with 0.5 m NaH2PO4 and 0.5% NP40 final concentration), and a signal was detected using rhodamine-conjugated antidigoxigenin and/or FITC-conjugated avidin and amplification reagents from Oncor following standard protocols. Results were observed and photographed with a Zeiss Axioskop II Fluorescent microscope (Carl Zeiss, Oberkochen, Germany).
Five-prime RACE was performed using the Marathon cDNA Amplification Kit (Clontech, Palo Alto, CA). Poly(A)+ RNA isolated from the MBCL-1 cell line was used as a template for first-strand cDNA synthesis carried out with gene-specific primers following the manufacturer’s directions. Nucleotide sequence of the RACE products was determined with standard M13F and M13R primers using the AmpliCycle Sequencing Kit (Perkin-Elmer, Branchburg, NJ) and the GeneAmp PCR 9600 Themocycler (Perkin-Elmer, Norwalk, CT).
Five micrograms of tumor and control DNA were digested to completion with BglII and SpeI restriction enzymes (New England Biolabs, Beverly, MA) according to manufacturer’s instructions. Restriction enzymes were inactivated by standard extraction procedures (15) and ethanol precipitation. DNA was resuspended in 50 μl of water and quantified by spectrophotometric measurement at a 260 nm wavelength. Digested tumor DNA and control DNA (200 ng in 25 μl final volume) were incubated overnight with 800 units of T4 DNA ligase in 1× ligation buffer (NEB) at 15°C. IPCR (21) was performed using 8 ng of self-ligated DNA as a template in a total volume of 50 μl under the following reaction conditions: 1× buffer [50 mm Tris-HCl (pH 9.2), 16 mm (NH4)2SO4, 4 mm MgCl2, 0.05% Tween 20]; 200 μm dNTPs; 4 μm of each primer; 2.5U Taq DNA polymerase (Perkin-Elmer, Branchburg, NJ); and 0.05 units of Vent DNA polymerase (NEB). PCR cycling conditions were as follows: 94°C for 2 min; 15 cycles of 94°C for 10 s, 69°C for 210 s; 15 cycles of 94°C for 10 s, 69°C for 210 s plus 10 s per cycle; and 69°C for 8 min. Two rounds of amplification were performed using similar conditions but different pairs of primers nested with respect to one another and 1 μl from the initial round of PCR serving as a template for the second round.
Southern and Northern Blot Hybridization.
Genomic DNA was prepared following standard protocols (15) with the following modification: frozen tissue was thawed and minced with razor blades in extraction buffer rather than ground frozen with a Waring Blendor. Ten micrograms of genomic DNA were digested with 80–100 units of the indicated restriction enzyme, incubated at 37°C for no less than 6 h, and analyzed by Southern blot as described (15). Each digested sample was fractionated on a 0.8% agarose gel in 1× tris acetate EDTA buffer and subjected to electrophoresis for ∼25 h at 2.5 V/cm. After depurination, denaturation, and neutralization, DNA was transferred to a Biotrans nylon membrane (ICN, Irvine, CA) in 10× SSC. Blots were hybridized in ExpressHyb Hybridization Solution (Clontech) using probes radiolabeled by PCR with 32P-α-dCTP (NEN, Boston, MA). Blots were washed two times at room temperature for 20 min (2× SSC, 0.05% SDS) followed by two washes at 50°C for 20 min (0.1× SSC, 0.1% SDS) and exposed to BioMax MS film using a BioMax MS intensifying screen (Eastman Kodak, Rochester, NY) at −80°C.
Total RNA was prepared from the tumor samples and the MBCL-1 cell line using Trizol reagent (Life Technologies) as previously described (15). Poly(A)+ RNA for Northern blot analysis was selected from the total RNA using the Oligotex mRNA Mini Kit (Qiagen) according to the manufacturer’s instructions. Two micrograms of poly(A)+ RNA were size-fractionated by electrophoresis in a 1% agarose formaldehyde gel run in 3-(N-morpholino) propanesulfonic acid buffer and then transferred to a Biotrans nylon membrane. Both the total RNA tumor Northern blot as well as a Multiple Tissue Northern blot (MTN, Clontech) of poly(A)+ mRNA from various human tissues were hybridized in ExpressHyb Hybridization Solution (Clontech). Blots were washed according to the manufacturer’s instructions for the treatment of the multiple tissue blot.
Determination of Nucleotide Sequence and Computer-assisted Analysis of Derived Sequence.
Sequence data were obtained using the DyeDeoxy Terminator Cycle Sequencing Kit (Perkin-Elmer, Branchburg, NJ) and an ABI DNA Sequencer (Perkin-Elmer, Norwalk, CT). Sequence was analyzed using the National Center for Biotechnology Information BLAST Server and the GCG software package (Genetics Computer Group, Madison, WI). Peptides were analyzed using software available at the ExPASy world wide web server (Swiss Institute of Bioinformatics, Geneva).
Identification of YACs Flanking and Spanning the Chromosome 18 Breakpoint.
To identify YACs containing DNA sequences situated on either side of and closely linked to the breakpoint of chromosome 18 in the t(11;18)(q21;q21), two-color FISH was performed by hybridizing pairs of YACs to interphase nuclei of case 2. In these analyses, hybridization of the two YACs was distinguished by differentially labeling the DNA and detecting the probes with two different fluorochromes. Double signals, each consisting of two largely superimposed colors, would be expected for each of the two chromosome 18 homologues in normal nuclei observed by fluorescence microscopy. In contrast, two separate signals of one color (along with one double signal) would be expected in nuclei from tumors having a translocation breakpoint lying between the DNA represented in the two probes. YACs used for this purpose were selected from Whitehead CEPH megaYAC contig 18.4 (17), which covers DNA of the chromosome 18q21 region, with YAC DNA inserts spaced at roughly 15 cR (∼5 megabase) intervals. In an initial series of analyses, YAC y766F12 at 99 cM (485 cR), YAC y949C6 at 94 cM, and YAC y943B8 at 83 cM (445 cR) were tested in pairwise hybridizations. In each analysis, two pairs of red/green signals were observed in all nuclei, indicating that the breakpoint does not fall between these YACs. However, when the most centromeric YAC of this initial group, y943B8, was cohybridized to interphase nuclei together with the still more centromeric YAC y830B10 at 79 cM (420 cR), about 40% of the nuclei yielded one paired and one unpaired signal. These observations suggested that the breakpoint falls between y943B8 and y830B10, thereby localizing the breakpoint to the intervening 5 cM. That 60% of the nuclei yielded two paired signals probably reflects the fact that the tissues contain numerous nonlymphoma cells. Hybridization with pairs of successively closer YACs within the contig 18.14 and positioned between y943B8 and y830B10 (Fig. 1 A) established that the breakpoint lies between the DNA of y902A4 and y826H9.
Because the YAC y817B11 overlaps both y902A4 and y826H9, y817B11 was predicted to span the chromosome 18 breakpoint. To test this prediction, FISH with y817B11 as a probe was performed on metaphase chromosome spreads from case 2. This analysis yielded three major signals: one to the normal chromosome 18, one to the derivative 18 chromosome, and one to the derivative 11, confirming that y817B11 spans the breakpoint in chromosome 18 (Fig. 2,A)−. The identities of derivative 18 and derivative 11 chromosomes in these analyses were confirmed by rehybridization of the same metaphase with probes specific for the pericentromeric regions of chromosomes 11 and 18 (Fig. 2, B and C).
To further define the position of the breakpoint, two-color interphase FISH was performed using y789F3. This YAC partially overlaps both the spanning YAC y8l7B11 and the more centromeric flanking YAC y9O2A4. YAC y789F3 was hybridized to interphase nuclei of case 1 along with YAC y826H9, which lies on the telomeric side of the breakpoint and, as such, contains a DNA sequence that should lie within the derivative 11 chromosome. Examination of the interphase fluorescence pattern revealed that part of the y789F3 separated from the y826H9 label. Therefore, y789F3 also spans the breakpoint. This localized the breakpoint between the marker Dl8Sll29, the first STS telomeric to y789F3, and the Whitehead single copy probe WI5827 (D18S1055), the most centromeric STS common to both 817B11 and y9O2A4.
Construction of a BAC Contig across the Chromosome 18 Breakpoint.
To identify BACs in the region of the chromosome 18 breakpoint, pools of DNA from a BAC library with 6-fold redundancy (Research Genetics) were screened by PCR using primers from STSs compiled by the Human Genome Project for the region of the YACs described above. D18S831, an STS that lies within DNA of y826H9 and maps adjacent to D18S1129, identified BAC b245Il4; D18S1129 identified BAC b357N13; and D18SlO26, an STS within the DNA of y789F3, identified BAC b54Lll (Fig. 1, A and B). To determine the position of the breakpoint with respect to these BACs, labeled DNA from pairs of BACs was used as a probe in two-color FISH on interphase nuclei of case 2. Splitting of the two signals was not observed when b357N13 was tested together with the telomeric flanking BAC b245I14, indicating that the breakpoint does not lie telomeric to b357N13. In contrast, splitting was observed when either of these two BACs was individually tested together with the centromeric flanking BAC b54L11 (Fig. 2,D), localizing the breakpoint to the interval between b357N13 and b54L11. PCR primers constructed to be complementary to the nucleotide sequence determined for the ends of the BACS were used in PCR amplifications to select additional sets of BACs from the library. Probes produced from DNA of this second generation of BACs (b230N13 and b24P22) centromeric to the breakpoint were tested in two-color FISH with probes produced from second generation telomeric BACs (b32I18 and b79O17). Splitting was again observed, further limiting the breakpoint to a region bounded by b230N13 and b32I18. A third iteration of selection by PCR for BACs containing sequence at the centromeric end of b32I18 insert yielded the BACs b128K1, b246H13, b357H2, and b441H2, with each of the latter three overlapping the centromeric end of b128K1. PCR with primers prepared from the telomeric end of b230N13 identified BAC b205O5, the telomeric end of which was contained within BACs b357H2 and b441H21, indicating that a contiguous overlapping array of specific BACs across the breakpoint could be assembled with BACs selected from the library (Fig. 1 B).
The map deduced for the various BACs in the region of the chromosome 18 breakpoint was further validated by performing PCR amplifications with primers complementary to the ends of the BAC inserts on the DNA of YACs comprising a contig of this region constructed by the Whitehead Genome Project. Additionally, restriction analyses were performed on BAC DNAs, and the results from digests of the individual BACs were compared to define the regions of overlap. The sizes of human DNA inserts in the BACs were calculated based on the total sizes of the restriction fragments produced by digestion of the BAC DNAs.
Identification of a BAC Spanning the Chromosome 18 Breakpoint.
Because b230N13 and b32I18 contain DNA on either side of the breakpoint, it was clear that one or more BACs within the contig between these flanking BACs must span the breakpoint. To determine which of these BACs spans the breakpoint, DNA from each BAC mapping between b230N13 and b32I18 was used as a single color probe on interphase nuclei of the MALT lymphoma designated case 2. Both b128K1 and b246H13 gave only two signals on the majority of interphase nuclei, suggesting that the breakpoint was either not located within the BAC sequence or that the breakpoint was too eccentrically located within the BAC sequence to yield a third signal. BAC b357H2 alone yielded three distinct signals in about 40% of the interphase nuclei. Control FISH with b357H2 to interphase nuclei with normal karyotypes yielded one or two signals in nearly all nuclei examined. To rule out the possibility that the third signal represented spurious cross-hybridization to an irrelevant region of the genome rather than chromosome 18 DNA separated by the translocation, a two-color FISH assay was performed in which b357H2 DNA was labeled with one fluorochrome while the flanking BACs were together labeled with a second fluorochrome. Each of the three signals generated from b357H2 DNA colocalized with signals generated by the flanking BACs, demonstrating the reliability of the three b357H2 signals.
Two-color FISH interphase assays with DNA of b357H2, b32I18, and b205O5, as described above, were repeated on interphase nuclei from cases 1 and 3. In both cases, b357H2 yielded three signals in close proximity to the signal produced by the flanking BACs b32I18 and b20505. These results indicate that in all three of our cases confirmed to have the t(11;18)(q21;q21) by cytogenetic analysis, the chromosome 18 breakpoint lies within DNA of the BAC b357H2.
Localization of a Chromosome 18 Breakpoint in the t(11;18)(q21;q21).
To facilitate ultimately a search for genes in the region of the chromosome 18 breakpoint, DNA of b128K, the smallest of the BACs that mapped to the region, was randomly sheared and subcloned in M13 viral vectors, the inserts of which were then subjected to sequence analysis designed to cover the sequence of the BAC with 6-fold redundancy. This analysis permitted the construction of a 58-kb contig representing the centromeric portion of b128K1, believed to flank the breakpoint region described above. To investigate whether sequence eccentrically located in BAC b128K1 (hence, undetectable by interphase FISH analysis) was rearranged by the breakpoint, probes for Southern blot analysis were generated to test all SpeI and BamHI restriction fragments in the most centromeric 30 kb. Additionally, to extend the analysis in the centromeric direction into a sequence unique to b357H2 (the BAC demonstrated by FISH to contain DNA divided by the translocation), BamHI fragments of 24 kb, 5 kb, and 8 kb were mapped to the portion of b357H2 centromeric to b128K1 using the combined results of restriction digests and Southern blot analysis performed with probes produced from the sequence at the ends of the insert. Nucleotide sequence was obtained from the ends of these fragments, and complementary PCR primers were designed. A probe generated from the centromeric end of the 8-kb BamHI fragment, termed B6, detected a DNA rearrangement in Southern blot analyses (Fig. 3) with three of five restriction enzyme digests of genomic DNA isolated from case 1. DNA extracted from normal peripheral blood of the patient in case 1 and analyzed in parallel with the tumor DNA showed only unrearranged, germ-line bands in Southern blot autoradiograms. Restriction enzyme digests that demonstrated rearrangements included those with SpeI, BglII, and HindIII. Mapping of restriction sites for these enzymes within B6 further localized the position of recombination in the t(11;18)(q21;q21) to the most centromeric 3 kb of this fragment.
Southern blot hybridizations with probes made from the B6 fragment failed to show rearrangement in the DNA of tissues from each of the other two cases of MALT lymphoma or in DNA from the MBCL-1 cell line established from a MALT lymphoma containing trisomy 18 but not the t(11;18)(q21;q21). Further analyses covering genomic DNA about 60 kb telomeric and 10 kb centromeric to the breakpoint in case 1 also failed to disclose rearrangements in cases 2 or 3.
As a further step toward precise localization of the chromosome 18 breakpoint in case 1, the nucleotide sequence of B6 was determined. DNA fragments containing the fusion of the chromosome 18 nucleotide with the chromosome 11 sequence were then amplified by IPCR using tumor DNA from case 1. “Head-to-head” primers in these reactions were designed to complement the nucleotide sequence at the centromeric end of B6. BglII and SpeI digestion of the IPCR products yielded fragments corresponding in size to those expected from the Southern blot analyses. IPCR of both digests preferentially amplified the smaller rearranged fragments compared to the larger germ-line fragments. IPCR products were ligated into the vector PCR 2.1 (Topo-TA Cloning Kit, Invitrogen, Carlsbad, CA) and subjected to sequence analysis, which produced 39 bp of novel sequence from the BglII template and 190 bp from the SpeI template that diverged from that contained within the B6 of normal DNA (Fig. 4, A and B). Oligonucleotide primers designed from these new sequences were used for standard PCR with the following DNA templates: normal human genomic DNA, DNAs from two CHO/human hybrid cell lines containing either chromosome 11 or chromosome 18 as the only human chromosomal constituents (CHO+11 and CHO+18), and the chromosome 18 BAC b357H2. PCR of normal human genomic DNA and CHO+11 DNA yielded identical products of 116 bp, whereas PCR of the other templates showed no amplification products (Fig. 4 C). Using DNA templates from a panel of chromosome 11q21 YACs, PCR generated the same 116-bp product from YAC y947E2 (data not shown). Additionally, amplification across the translocation breakpoint, using one primer complementary to the putative chromosome 11 sequence and one to the chromosome 18 sequence, yielded a product of predicted size with tumor DNA from case 1 serving as a template but not with normal genomic DNA as a template.
Disruption of an Intronic Sequence in the MALT1 Gene by the t(11;18)(q21;q21).
To identify a gene or genes in the vicinity of the breakpoint, the 58-kb sequence contig was analyzed with the software programs GRAIL, ORF finder, and BLASTN against the genebank databases. Although nine clusters of ESTs comprised of sequence found within this region were identified by this analysis, only a single EST, AA214173, contained more than a single putative exon and was free of highly repetitive genomic sequence elements. The 5′ 79 bases of this EST encode U6 RNA. Because 5′ attachment of a U6 sequence is a known artifact of cDNA cloning and such a sequence does not appear within the sequence determined for b128K1, we regarded it as irrelevant. Following the U6 sequence in the 3′ direction, sequence analysis revealed an open reading frame of 1336 bp. Therefore, to obtain an additional 5′ coding sequence for the gene represented by EST AA214173, RACE amplification was performed using primers at the 5′ end of the known coding sequence in the EST and RNA prepared from the MBCL-1 cell line. Analysis of the resulting amplification products yielded an additional 696 bp of sequence. Comparison of this sequence with that of b128K1 revealed two additional exons of length 101 bp and 73 bp in the centromeric most 2 kb of the b128K1 insert. Analysis of the 696-bp sequence with the BLASTN program against the NCBI ALU database detected Alu and LINE elements in the most 5′ 250 bp. Analysis of the roughly 696 bp of remaining sequence of the RACE product with BLASTN against the human EST database found only one human EST, AA321297, which had been obtained from a human bone marrow library. The 345 nucleotides available for this EST are 97% identical to the RACE product. The match included 250 bp of Alu sequence. That an independently prepared cDNA duplicated the Alu in the RACE product suggests that the repetitive element could be a real exonic component of this gene.
To further map the unique sequence of the RACE product within the genome, the nonrepetitive portion of the RACE product was generated by PCR, and the resulting fragment was used to probe Southern blots of human genomic DNA and restriction digests of BACs b128K1, b246H13, and b357H2. Similar fragments were detected in the restriction digests of genomic DNA and BAC DNA, indicating that the portion of the RACE product used as a probe consists of a single copy sequence. Analysis of the blots prepared from BAC DNA revealed hybridization to the 24-kb BamHI fragment (B1) of b357H2 centromeric to b128K1. However, there was no hybridization to 5-kb (B10) or 8-kb (B6) BamHI fragments derived from DNA between that of B10 and the end of b128K1. Because B6 contains DNA broken by the t(11;18)(q21;q21) in case 1, recombination in this tumor occurred in the fifteenth intron from the 3′ end of a gene, for which we propose the name MALT1.
To determine exon/intron boundaries in the RACE sequence not present in b128K1, cycle sequencing was performed using internal primers from the RACE product and b357H2 DNA as a template. This analysis revealed four exon/intron boundaries demarcated by appropriate consensus splice donor/acceptor sequences. The first exon contains the Alu and LINE iterated sequence elements, with the recombination breakpoint in chromosome 18 of case 1 falling in the third intron 3′ of this exon. The entire exon/intron structure deduced for the segment of MALT1 identified thus far is shown in Fig. 1 C.
Structural and Expression Analyses of the MALT1 Gene.
The full sequence for MALT1 assembled from genomic sequence, ESTs, and RACE products consists of 3910 bp, including an open reading frame of 2325 bp encoding a polypeptide of 775 amino acids (Fig. 5). An in-frame stop codon is located immediately upstream of the putative first ATG, which is found at nucleotide 266 in the sequence. It has homology to members of the immunoglobulin superfamily, with two immunoglobulin-like C2 domains from amino acids 69 to 243 resembling sequence found in both cell adhesion molecules and receptor tyrosine phosphatases. MALT1 is homologous to a conceptually translated open reading frame, F22D3.6, in Caenorhabditis elegans, with 27% identity and 44% similarity between regions extending from amino acid 197 to 569 of MALT1 and from amino acid 92 to 506 of F22D3.6. A ProfileScan of the PROSITE profile database reveals an interleukin-1β-converting enzyme (ICE, or caspase-1) p20 motif from amino acid 292 to 370. These polypeptide sequences share 52% similarity and 27% identity with caspase-1 p20 from amino acid 160 to amino acid 218. However, the alignment terminates abruptly 19 amino acids N-terminal of the active site amino acid (237 in caspase-1 p20). Following submission of this manuscript, much of the sequence of MALT1 was described by Dierlamm et al. (22), which they refer to as MLT. However, our sequence differs from that presented by those investigators by including 5′ and further 3′ untranslated regions, as well as 46 additional N-terminal codons within the open reading frame.
Northern blots containing polyadenylated RNA from 16 adult human tissues (heart, brain, placenta, lung, liver, skeletal muscle, kidney, pancreas, spleen, thymus, prostate, testis, ovary, small intestine, colon, and peripheral blood lymphocytes) were hybridized with a 1228-bp EcoRI fragment of MALT1 cDNA containing nucleotides 446 to 1674 of the open reading frame. Four bands appeared in the autoradiograms with RNA from each tissue. These bands corresponded to species with approximate lengths of 3.2, 4.6, 5.4, and 9.6 kb. The presence of multiple bands in this autoradiogram could be due to cross-hybridization by a portion of the probe to RNA other than that transcribed from MALT1 or to various forms of RNA derived from MALT1 possibly by alternative splicing of a precursor transcript. To evaluate the accuracy of these two explanations, one blot including RNA from eight tissues was sequentially rehybridized with probes containing sequences from 452 to 1310 of the MALT1 open reading frame and from 1476 to 1880. Both hybridizations yielded the same pattern of bands, suggesting that all four RNA species are transcribed from MALT1 and may be alternatively spliced products from this one gene.
Northern blot hybridization analysis of polyadenylated RNA from B-cell tumors also indicated the same four MALT1 species, although the relative abundance of the bands varied somewhat among specimens (Fig. 6). Several specimens showed an additional faint band corresponding to an RNA of about 7.8 kb. However, a unique band migrating slightly below the 7.8 kb band was detected by MALT1 probes only in the MALT lymphoma from case 1.
Our results indicate that the chromosome 18 breakpoint of t(11;18)(q21;q21) in all three cases of MALT lymphoma having this translocation and available to us for study lie within a region of less than about 140 kb (the size of the DNA insert in the BAC b357H2). This region corresponds to the same general area reported by others to contain this breakpoint (23). However, within this region, the breakpoints might be widely scattered given that Southern blot analysis of DNA for 20 kb in both directions from a breakpoint in case 1 failed to find breakpoints in two other cases having the translocation. Further testing of additional MALT lymphomas will be necessary to determine the distribution of breakpoints, as well as to establish the incidence of the t(11;18)(q21;q21) and other chromosomal aberrations within chromosome 18 band q21 that may evade conventional cytogenetic analysis. That abnormalities within this region of the genome may be particularly important in MALT lymphoma is suggested by the frequent occurrence of trisomy 18 in these tumors and the prior finding of structural changes in trisomic chromosomes at the same loci as involved in chromosomal translocations found in other examples of that particular tumor type (24). Indeed, this precedent shaped our rationale for initially pursuing the breakpoint in chromosome 18 rather than in chromosome 11 to understand the molecular consequences of the t(11;18)(q21;q21).
No matter how diffusely scattered translocation breakpoints may be within DNA about a chromosomal locus, they generally exert their effect in tumors containing them by altering the expression of genes closely linked to the breakpoint. However, in some instances, the distance between a gene and a breakpoint can be relatively great, such as the 1000 kb or more of DNA that may separate translocation breakpoints in chromosome 8 from the MYC gene of some endemic Burkitt’s lymphomas (25). This fact complicates the determination of the gene directly affected by any translocation that has not been previously characterized at the molecular level. In the present situation, the MALT1 gene covers roughly the middle 65 kb of the BAC b357H2 found by us to span the breakpoint in three examples of the t(11;18)(q21;q21). Although t(11;18)(q21;q21) could act through other genes either contained in b357H2 or in DNA adjacent to the region within this BAC, recombination within MALT1 in at least one case would suggest that this gene plays a role in the pathogenesis of MALT lymphomas. Additionally, FISH using b357H2 as a probe on interphase nuclei from all three tumors with the t(11;18)(q21;q21) yielded signals of roughly equal intensity from the two derivative translocation products, thereby favoring the presence of breakpoints in the mid-portion of the DNA contained in the BAC; that is, the region of DNA covered by MALT1. On the other hand, Northern blot autoradiograms of MALT1 RNA in normal tissues revealed a multiplicity of bands, some of which correspond to rather large RNA species. The size of the gene may therefore be far bigger than the 65 kb we have assigned to it and could constitute a larger target for recombination than we have assumed based on data from ESTs, sequencing of genomic chromosome 18 DNA, and RACE amplification.
Although case 2 was shown by FISH to have the same translocation as case 1, Northern blot analysis of case 2 failed to detect a band similar to the abnormal band seen in case 1. There are several possible explanations for this finding. For instance, an abnormal band may be present at very low intensity in case 2 RNA, either due to variation in the level of transcript among tumors or to a paucity of tumor cells within the tissue sections from which RNA was extracted. It could also be that the breakpoint in the second tumor tested lies within the MALT1 gene but does not affect the RNA sequences recognized by the probes used in this assay. Another possibility is that the breakpoint in the second tumor falls outside the MALT1 gene entirely. Assuming that MALT1 is directly involved in the pathogenesis of MALT lymphoma, the latter explanation would imply that alterations in MALT1 could occur by recombination both within or outside the gene, yet somehow disrupt or alter function in either circumstance. A similar situation has been reported with respect to the MYC gene in translocations of Burkitt’s lymphoma (24). Finally, the breakpoints within the MALT1 gene could be variable in position and lead to transcripts of size different from the abnormal band produced from RNA of case 1. Some of these transcripts may be unstable or generate bands obscured by the multiple normal bands detected in the Northern blot autoradiograms in normal and tumor tissues. Ultimately, these issues can only be resolved by further structural analysis of the MALT1 gene in MALT lymphomas and by functional studies directed at the ability of this gene in either normal or mutant form to affect the proliferative behavior of lymphocytes in culture and in vivo.
Structural analysis of the coding sequence contained within the MALT1 cDNA identified several regions of homology to known genes. The presence of two immunoglobulin-like C2 domains suggests a transmembrane or extracellular location for the normal MALT1 protein; however, no amino acid sequences compatible with signal peptides or transmembrane regions could be found by computer scans of the cDNA sequence. On the other hand, the protein TITIN is a well-characterized component of striated muscle that also contains immunoglobulin-like C2 domains (25), so that an intracellular role for MALT1 is a possibility. A second homology of note is the partial caspase motif within MALT1. The exclusion from MALT1 of the critical amino acid in the active caspase protease site (26) would seem to imply that the protein lacks caspase activity but may bind substrate for caspase and conceivably act as a competitive inhibitor of caspase enzymes. According to this line of reasoning, MALT1 would have antiapoptotic function reminiscent of BCL-2, a gene implicated in the malignant transformation of B lymphocytes in follicular lymphoma (27), another form of indolent B-cell neoplasia. At any rate, an effect of the t(11;18)(q21;q21) on apoptosis is suggested by the recent report of Dierlamm et al. (22), who describe fusion of MALT-1 (MLT) in this translocation to API2, the product of which is an inhibitor of apoptosis. Investigation of a role for MALT1 in regulating apoptosis or in some other cellular function could possibly gain from genetic studies of the homologous open reading frame F22D3.6 in the Caenorhabditis elegans genome, as well as from analysis of mice carrying transgenes and knock-out mutations for MALT1.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by Grants CA75354, CA72744, and AI013575 from the NIH.
The abbreviations used are: MALT, mucosa-associated lymphoid tissue; CHO, Chinese hamster ovary; YAC, yeast artificial chromosome; BAC, bacterial artificial chromosome; STS, sequence-tagged site; EST, expressed sequence tag clone; FISH, fluorescence in situ hybridization; RACE, rapid amplification of cDNA ends; IPCR, inverse PCR; cR, centiRay.
L. Soreng, E. Robinson, E. Kieff, and J. Sklar, unpublished data.
We thank Marcia Lux for advice and assistance with FISH and Dr. Arthur Skarin for help in obtaining biopsy specimens.