Abstract
Mantle cell lymphoma (MCL) is an aggressive non-Hodgkin's lymphoma with median patient survival times of ∼3 years. Although the characteristic t(11;14)(q13;q32) is found in virtually all cases, experimental evidence suggests that this event alone is insufficient to result in lymphoma and secondary genomic alterations are required. Using a newly developed DNA microarray of 32 433 overlapping genomic segments spanning the entire human genome, we can for the first time move beyond marker based analysis and comprehensively search for secondary genomic alterations concomitant with the t(11;14) in eight commonly used cell models of MCL (Granta-519, HBL-2, NCEB-1, Rec-1, SP49, UPN-1, Z138C and JVM-2). Examining these genomes at tiling resolution identified an unexpected average of 35 genetic alterations per cell line, with equal numbers of amplifications and deletions. Recurrent high-level amplifications were identified at 18q21 containing BCL2 , and at 13q31 containing GPC5. In addition, a recurrent homozygous deletion was identified at 9p21 containing p15 and p16. Alignment of these profiles revealed 14 recurrent losses and 21 recurrent gains as small as 130 kb. Remarkably, even the intra immunoglobulin gene deletions at 2p11 and 22q11 were detected, demonstrating the power of combining the detection sensitivity of array comparative genomic hybridization (CGH) with the resolution of an overlapping whole genome tiling-set. These alterations not only coincided with previously described aberrations in MCL, but also defined 13 novel regions. Further characterization of such minimally altered genomic regions identified using whole genome array CGH will define novel dominant oncogenes and tumor suppressor genes that play important roles in the pathogenesis of MCL.
INTRODUCTION
Mantle cell lymphoma (MCL) is an aggressive malignant lymphoid neoplasm with a poor prognosis. MCL comprises 6% of all non-Hodgkin's lymphomas (NHLs) and has a median survival of ∼3 years with few long-term survivors ( 1 , 2 ). MCL was originally identified as a morphologically distinct subtype of NHL, and subsequently the t(11;14)(q13;q32) was defined as a characteristic molecular feature of this subtype ( 3 ). This translocation places the CCND1 (cyclin D1) gene adjacent to the active immunoglobulin heavy chain ( IGH ) enhancer region. Normally, cyclin D1 expression is transient; however, under the control of the IgH enhancer region, cyclin D1 is constitutively expressed, causing cell cycle dysregulation ( 4 ). Importantly, cyclin D1 is not expressed in normal B-cells.
Although the t(11;14) is found in virtually all cases of MCL, experimental evidence in transgenic mice suggests that this event alone may not be sufficient to result in lymphoma ( 5 , 6 ). Moreover, secondary genomic alterations are frequently detected concomitantly with the t(11;14) ( 7 ). Au et al. ( 8 ) compiled a review of 214 cases of MCL and found that the most common secondary genomic alterations included +3/3q (12%), +12/12q (8%), −13/13q (19%), −6q (19%), −9/9p (13%), −1p (11%), −11q (11%), −Y (10%) and −17p (8%). In addition, leukemic and blastoid variants of MCL have been characterized to determine secondary genomic alterations that can act as diagnostic or prognostic markers ( 9 – 11 ). With few exceptions, these common secondary alterations encompass large regions encoding many candidate genes, complicating identification of markers and potential therapeutic targets.
Many of these regions of genomic alteration were identified using comparative genomic hybridization (CGH), which detects single copy number differences between two samples of DNA ( 12 ). This is accomplished by a competitive hybridization of differentially labeled sample and reference DNA to normal metaphase chromosomes, allowing the detection of regional copy number changes at a resolution of 5–10 Mb ( 13 ). A recent adaptation, array CGH, improves on the resolution of DNA copy number profiling by utilizing DNA segments representing discrete genomic loci spotted onto glass slides as the hybridization target. For example, arrays of 2460, 3500 and 4100 marker bacterial artificial chromosome (BAC) clones distributed at ∼1 Mb intervals have been constructed ( 14 – 16 ). Even though these arrays represent <10–15% of the genome, they have proven to be extremely useful for detecting chromosomal aberrations associated with congenital abnormalities and somatic malignancies ( 17 – 20 ). Recent studies have chosen to use overlapping coverage to determine alteration boundaries within regions of interest ( 21 – 24 ). A recent study assayed 53 MCL patient samples using an 812 segment matrix-CGH array designed for clinical classification of MCL ( 25 ). These studies show that the power of array CGH is maximized when the ability to detect single copy number changes is combined with the use of a tiling or overlapping set of BAC clones. With the development of our array of 32 433 overlapping BAC clone-derived segments spanning the genome, it is now possible to comprehensively determine segmental copy number alterations throughout the entire human genome in a single experiment while simultaneously fine mapping alteration breakpoints ( 26 ).
In this study, we use this new comprehensive segmental copy number profiling technology, called sub-megabase resolution tiling-set (SMRT) array CGH, to identify secondary genomic alterations concomittant with the t(11;14) by characterizing eight commonly used cell models of MCL (Granta-519, HBL-2, NCEB-1, Rec-1, SP49, UPN-1, Z138C and JVM-2). Characterization of these cell lines is necessary to determine if they adequately represent MCL. In addition, characterization of these cell lines may provide insights into MCL biology. Here, we describe the whole genome analysis and the identification of novel genetic alterations in these cell models. This constitutes the first disease-specific study using this new technology.
RESULTS
Copy number profiles of the eight MCL cell lines were created by co-hybridizing differentially labeled sample DNA with reference male DNA on the SMRT array that contains 32 433 BAC-derived amplified fragment pools spotted in triplicate ( 26 ). The analysis of 97 299 data points for each of the eight samples (778 392 in total) facilitated the localization of chromosomal breakpoints to within single BAC clones and the subsequent identification of genomic imbalances between breakpoints. For example, alterations in MCL cell line HBL-2 included numerous gross chromosomal losses (−3p21.31–p21.1, −6q22.32–qter, −9p21.2–pter, −13q14.13–q21.32, −17p11.2–pter, −18q11.2–q21.2, −18q22.2–qter) and gains (+1q32.2–q42.2, +2q36.1–qter, +3q13.32–q22.2, +7, +8q21.12–q22.3, +8q23.1–q24.21, +9q22.33–qter, +11p11.2–pter, +11q13.3–q14.2, +13q31.2–q34, +16q, +18q21.2–q22.2). Additionally, using this comprehensive genomic assay we were able to detect localized chromosomal alterations affecting small numbers of overlapping clones (−1p36.22–p36.21, −2p11.2, −3p22.2, −3q13.31, −4p15.1–p14, −6p25.2, −6p24.3–p24.2, −6p22.3, −13q21, −15q15.1–q15.2, −15q23, −16p11, −22q11.22, +1q21.1, +1q42.3, +2p11.1, +2q33.1, +2q33.2–q33.3, +2q34, +2q35, +4p14, +4p11, +10p11.1, +10q11.21, +10qter, +11q22.3, +12q14.2, +21p11) (Fig. 1 ; aCGH smooth data in Supplementary Material).
Comparable numbers of chromosomal alterations were observed in all MCL lines assayed excluding JVM-2, resulting in 250 genomic imbalances (mean per tumor line, 35.6; range, 21–57) including 125 genomic gains (mean per tumor line, 17.9) and 125 genomic losses (mean per tumor line, 17.7). JVM-2 contained only seven genomic imbalances including five gains and two losses. The high number of genomic imbalances observed in seven of the eight MCL cell lines is likely due to the comprehensive genomic coverage of array CGH using the SMRT array. Identification of novel minute alterations combined with the subdivision of previously defined large alterations into smaller regions may account for the increased numbers of genomic imbalances detected. Because highly recurrent genetic imbalances may be indicative of important regions involved in pathogenesis, copy number profiles were aligned to reveal patterns of recurrent alteration (Fig. 2 ). Only alterations recurring in at least three of the eight lines were used to define minimally altered regions (MARs) (Table 1 ). Using this definition we can confidently define the boundaries of our regions because the clones near boundaries have shown that they can respond to copy number changes in at least two alternate lines. Of the resulting 35 MARs, we observed a range of sizes from 130 kb to 40 Mb with nine (26%) at <1 Mb in size. These regions could not have been defined by conventional chromosomal or array-based CGH. Of the 35 MARs observed in this study, 22 (63%) recurred in three samples, 10 (29%) recurred in four samples and three (9%) recurred in five or more samples.
We first investigated whether the identified MARs included the chromosomal alterations commonly reported in MCL. The most common MAR was the loss of chromosome arm 9p. HBL-2, Rec-1 and UPN-1 demonstrated loss from 9p21.2–pter, whereas Granta-519, SP49 and Z138C showed multiple smaller regions of deletion and loss. After alignment of the profiles, a 1.24 Mb MAR of loss was identified in six of the eight samples. Notably, in SP49 and Z138C the localized deletion and loss (1.67 and 1.24 Mb, respectively) encompassed CDKN2A (p16 INK4A ) and CDKN2B (p15 INK4B ), which are inhibitors of cyclin dependent kinase 4. Both CDKN2A and CDKN2B are reported to be both deleted and under-expressed in MCL ( 3 , 27 , 28 ). Three additional MARs of loss were identified at 9p23, 9p23–24.1 and 9p24.3–pter with sizes of 640 kb, 3.31 Mb and 480 kb, respectively (Fig. 3 ).
Trisomy 12 is a common cytogenetic event in MCL, although almost never seen as a sole cytogenetic alteration. Surprisingly, examination of 1488 DNA segments representing chromosome 12 revealed multiple distinct regions of gain on both the long and the short arm as opposed to trisomy 12. The multiple regions observed in Granta-519, HBL-2, NCEB-1, Rec-1, Z138C and JVM-2 included six MARs of gain at 12p11.21–p12.3, 12q13.13–q13.2, 12q24.21–qter, 12q14.2, 12q13.2–14.1, 12q15–21.2 (Fig. 4 ). Two genes known to be over-expressed in MCL, CD63 (melanoma 1 antigen) and CDK4 (cyclin dependent kinase 4), reside within the 3.41 Mb MAR of gain at 12q13.2–14.1 ( 29 , 30 ). Curiously, RARG (retinoic acid receptor gamma), presently reported to be under-expressed in MCL, resides within a gained 2.97 Mb MAR at 12q13.13–q13.2 ( 30 ).
Loss of chromosome 1 material was also observed at p36.11 and p21.1–p31.1. The 1p36.11 MAR was detected in Granta-519, SP49 and Z138C with Z138C defining the boundaries. This 1.18 Mb MAR of loss contains the FGR (Gardner-Rasheed feline sarcoma viral oncogene homolog) gene. The expression level of FGR in Granta-519 and Z138C according to the Lymphochip was 4-fold lower than that of NCEB-1. This is consistent with the fact that NCEB-1 showed genomic gain in this region as opposed to the reduced copy number in Granta-519 and Z138C. The 1p21.1–p31.1 MAR of loss was detected in Granta-519, Rec-1 and SP49 with Rec-1 defining the boundaries. This 35.25 Mb region contains 111 Refseq annotated genes, of which, four genes are reportedly differentially expressed in MCL, BCL10 (B-cell CLL/lymphoma 10), CNN3 (calponin 3, acidic), GBP1 (guanylate-binding protein 1) and TGFBR3 (transforming growth factor-beta type III), with under-expression of BCL10 , GBP1 and TGFBR3 correlating well with our observation of the loss of this region ( 29 – 31 ).
The short arm of chromosome 17 is commonly deleted in many tumor types including MCL. In our panel, we observed loss of the entire p-arm in Granta-519, HBL-2 and UPN-1. In addition, we observed the loss of a 1.14 Mb telomeric region in Z138C outside the TP53 region (Table 1 ).
Alignment of Granta-519, HBL-2, NCEB-1 and Rec-1 defined a 760 kb MAR of loss at 13q14.3. Loss of chromosome 13q material is commonly reported in MCL. Consistent with previous reports, this MAR did not include the tumor suppressor gene RB1 (retinoblastoma), but was 1.2 Mb distal to RB1 ( 3 , 7 ). Granta-519 and NCEB-1 showed 1.26 and 2.55 Mb regions of loss, respectively. The 1.26 Mb deletion in Granta-519 was subsequently confirmed by locus-specific fluorescence in situ hybridization (FISH) analysis (Fig. 5 A). This MAR at 13q14.3 coincided with the deleted regions described in chronic lymphocytic leukemia (CLL) and recently in MCL ( 25 , 32 ). Several candidate tumor suppressor genes including DLEU1 (deleted in lymphocytic leukemia 1), DLEU2 (deleted in lymphocytic leukemia 2), RFP2 (ret finger protein 2) and C13orf1 (CLL deletion region gene 6) reside within the 760 kb gene-rich region. The expression level of DLEU2 revealed that Granta-519, HBL-2 and NCEB-1 had an average 3-fold lower expression than Z138C according to Lymphochip. This correlates with the reduced copy number of this region in the three cell lines as compared with Z138C.
Duplication of 7p has been reported in MCL, particularly in blastoid variants ( 9 ). Detailed examination of 7p using the SMRT array revealed two distinct MARs of gain. The most recurrent of these was a 5.85 Mb region between 7p22.1 and 7pter. This region contains 37 Refseq annotated genes, of which MAD1L1 (mitotic arrest deficient-like 1) is reported to be differentially expressed in MCL ( 30 ). HBL-2, Rec-1, SP49 and Z138C demonstrated gain of this region with HBL-2 defining the MAR. The other gained MAR on 7p was large, spanning 40.4 Mb, containing 171 Refseq annotated genes, including NMB (neuromedin B) reported to be over-expressed in MCL ( 29 ).
A recent study by Bentz and coworkers delineated a 2.4 Mb consensus region of deletion at 8p21 ( 25 ). We observed a 730 kb MAR of loss at 8p21.2–p21.3 containing three putative tumor suppressor genes, RHOBTB2 (Rho-related BTB domain containing 2), TNFRSF10B (tumor necrosis factor receptor superfamily, member 10B) and DBC-1 (deleted in breast cancer 1). RHOBTB2 and TNFRSF10B overlap with the previously described consensus region; however, unlike their consensus region TNFRSF10C and TNFRSF10D were excluded in our MAR. A similar deleted region was previously described in a breast cancer study ( 33 ).
Next, we investigated the genomic loci of genes known to be differentially expressed in MCL. Reported over-expression of the anti-apoptotic dominant oncogene BCL2 in MCL coincides with a 7.26 Mb amplified MAR between 18q21.33 and 18q22.1 ( 30 ). Lymphochip data showed a similar result. The average BCL2 expression in Granta-519, HBL-2, NCEB-1 and Z138C is 4-fold higher than that of the Rec-1 sample. This correlates with increased copy numbers of BCL2 in Granta-519, HBL-2 and Z138C as compared with normal copy number in Rec-1. However, NCEB-1 also showed high-level BCL2 expression. As there is no increase in copy number at 18q21.33 in NCEB-1 this eludes to a different mechanism of over-expression. The well characterized MYC oncogene that is reported to be over-expressed in MCL ( 30 ) coincided with a 5.56 Mb MAR of gain in HBL-2, NCEB-1, Rec-1 and UPN-1. The boundaries of this MAR were defined by the overlapping regions of gain in HBL-2 and Rec-1. Another MAR of gain is 2.50 Mb in size located at 11p15.5–pter. It contains genes IFITM2 (interferon-induced transmembrane protein 2), H-RAS (Harvey rat sarcoma viral oncogene homolog) and IFITM1 (interferon-induced transmembrane protein 1), known to be differentially expressed in MCL ( 29 , 30 ). Curiously, all three of these genes were reported to be under-expressed in the region we describe as gained. However, this region is densely packed with Refseq annotated genes (59 genes) and the importance of other genes in this region remains to be determined. REL (avian reticuloendotheliosis viral oncogene homolog) is another over-expressed gene in MCL ( 29 ) and lies within a 1.07 Mb gained MAR between 2p15 and 2p16.1 recurrent in Rec-1, SP49, UPN-1 and JVM-2. The MAR was defined by the amplified region in Rec-1.
The known genes and previously reported regions of alteration in MCL accounted for only 22 of the 34 (63%) identified MARs. We next examined the remaining 13 MARs, in order of increasing size, for genes described in the context of cancer but not previously implicated in MCL (indicated in bold in Table 1 ). Two sub-megabase recurrent losses were observed in seven and six of the cell lines, respectively. As these cell lines are derived from mature B-cells, one would expect the presence of deletions resulting from the rearrangement of immunoglobulin (Ig) genes during B-cell maturation ( 34 ). However, conventional chromosomal or array CGH has not detected these localized alterations. Strikingly, examination of both the kappa and the lambda Ig light chain loci revealed recurrent deletions. A 130 kb MAR at 2p11.2 coinciding precisely within the kappa Ig light chain locus was observed in Granta-519, HBL-2, NCEB-1, Rec-1, SP49, UPN-1 and JVM-2. Similarly, a 390 kb MAR at 22q11.22 coincided with the lambda Ig light chain locus and was observed in Granta-519, HBL-2, NCEB-1, SP49 and Z138C. The deletion at the lambda Ig light chain region was verified by locus-specific FISH in cell line Granta-519 (Fig. 5 B). Probe hybridization was confirmed in normal metaphases (data not shown).
A 430 kb region of recurrent loss on chromosome 8 was observed at p23.3 containing only three genes: ZNF596 (zinc finger protein 596), FBXO25 (F-box only protein 25) and INM01 (hypothetical protein). However, none of these genes have been described in the context of cancer.
The fifth, seventh and eighth smallest novel MARs are all gains on chromosome 2 at q37.2 (870 kb), q37.1 (1.66 Mb) and q37.3 (2.42 Mb) and contain 2, 14 and 23 genes, respectively. The gained region at 2q37.3 contains two genes that have been described in the context of ovarian and pancreatic cancer, BOK (BCL2-related ovarian killer) and GPC1 (glipican 1), respectively. BOK has been described as a pro-apoptotic BCL2 protein with restricted expression in reproductive tissue ( 35 ). As such, it is unlikely to be involved in MCL pathogenesis. However, the glypican family of cell surface proteins are expressed primarily during development with differential expression in a tissue and stage-specific manner, affecting both growth and morphogenic signals ( 36 ). As such, both GPC1 and GPC3 have been implicated in pancreatic and ovarian cancers as putative oncogenes, respectively ( 37 , 38 ).
In addition, the sixth smallest novel MAR (1.16 Mb) at 13q31.3 is one of only two regions of recurrent amplification. The only gene affected by this amplification is GPC5 (glypican 5). GPC5 has recently been identified as amplified in various types of lymphoma ( 39 , 40 ). This MAR was present in HBL-2, Rec-1 and Z138C and defined by the overlap between amplifications in HBL-2 and Z138C.
The 3.10 Mb MAR of gain at 1p36.32–pter contains 37 Refseq annotated genes. Among these are several tumor necrosis factor receptor super family genes, a cell division cycle control gene ( CDC2L2 ) and matrix metaloproteases. Unfortunately, none of these genes are represented on the Lymphochip, and a candidate oncogene in this MAR remains to be determined.
The 9q34.2–q34.3 (4.31 Mb) gene dense MAR of gain encompasses 62 Refseq annotated genes. Among these, several putative oncogenes have been identified, VAV2 (vav2 oncogene) NOTCH1 (notch homolog 1, translocation associated), and TRAF2 (tumor necrosis factor receptor-associated factor 2 isoform 2) ( 41 – 43 ). VAV2 is a guanine nucleotide exchange factor involved in Rac1 activation by Src that induces morphological changes when over-expressed in NIH-3T3 cells ( 41 , 44 ). Over-expression of NOTCH1 has been shown to promote tumor cell proliferation and survival in both T- and B-cell lymphomas as well as epithelial neoplasms such as small cell lung cancer ( 45 ). TRAF2 is involved in cytokine mediated activation of NF-kappaB, which is commonly activated in lymphomas ( 43 ).
An equal size (4.31 Mb) MAR of gain at 7q11.23 contains 49 Refseq annotated genes, including BCL7B.BCL7B shares 90% identity with BCL7A , which is recurrently translocated in high-grade B-cell NHL ( 46 ).
The three largest novel MARs are gains of 1q32.2–q32.3 (4.69 Mb) and 8q22.1–q22.3 (4.79 Mb), and loss of 18q11.2–q21.2 (17.07 Mb). These MARs encompass 31, 22 and 51 genes, respectively. Neither candidate oncogenes nor tumor suppressor genes have been described within these regions, and further characterization is necessary to determine the selective advantage of having these genomic alterations.
DISCUSSION
Unlike solid tumors, cell lines are readily created from lymphomas. High-throughput techniques allow for the rapid characterization of cell lines, with the hope of identifying new molecular aberrations that can act as biomarkers or therapeutic targets. However, the transcriptome and proteome are volatile and too readily influenced by culture conditions. Genomic analysis has the advantage of being more stable, making it theoretically easier to identify key molecular abnormalities. The eight cell lines used in this study had not been thoroughly characterized at the genomic level. The recent development of the SMRT array allowed the first detailed whole genome examination of copy number changes at tiling resolution in these cell lines. Assessment of copy number alterations with 32 433 DNA segments identified an unprecedented average of 35 genetic alterations per cell line with equal numbers of amplifications and deletions. Because recurrent alterations are more likely to be indicative of critical events in pathogenesis, we considered only those alterations recurrent in at least three cell lines. Of the recurrent MARs identified, only 18 (51%) were within previously characterized genetic alterations. Surveying regions encompassing genes known to be differentially expressed in MCL accounted for an additional four (11%) of the identified MARs. The remaining 13 (37%) identified MARs include novel regions of alteration in MCL. These regions may require further verification in paired normal and tumor samples to eliminate the possibility of polymorphisms within the human population.
In addition, nine (26%) of the defined recurrent regions were <1 Mb in size. Using previously available assays with 1 Mb or greater resolution, these regions would have been impossible to define. However, owing to the comprehensive overlapping nature of the SMRT array, we were able to define these regions. The utility of this comprehensive assay is made evident by the detection and localization of genomic rearrangements within the Ig loci, which previously went undetected in whole genome copy number assays.
Trisomy 12 has been observed in many types of lymphoma including MCL. This suggests that there are multiple oncogenes spanning chromosome 12. Our observation of six distinct MARs of gain throughout chromosome 12 supports this hypothesis and provides regions for further characterization. Similar to trisomy 12, chromosome arm 9p deletions are common in MCL. This is traditionally associated with the deletion of tumor suppressor genes p15 and p16. However, our data would suggest additional tumor suppressor genes at three separate MARs. Further characterization of these regions in clinical specimens is required to identify the candidate tumor suppressor genes. One of the MARs containing a single gene at 13q31.3 is associated with GPC5.GPC5 amplification has been observed in multiple B-cell lymphomas. Interestingly, we also identified a 2.42 Mb MAR at 2q37.3 containing GPC1. This finding raises the possibility of GPC family involvement in MCL pathogenesis. Over-expression of GPC1 and GPC3 has been found in ovarian and pancreatic cancer, respectively.
For the first time we can comprehensively describe MCL cell model genomes at tiling resolution. The extension of high-resolution whole genome analysis to tumor genomes will prove useful in identifying novel genetic alterations in MCL. Inclusion of these alterations will potentially enhance focussed MCL-specific arrays. (The complete panel of ‘SeeGH Karyograms’ output from SeeGH are given in Supplementary Material and the raw image data are available at http://www.bccrc.ca/cg/ArrayCGH_Group.html ).
MATERIALS AND METHODS
Cell lines, culture conditions and DNA extraction
A panel of eight cell lines, seven MCL-derived (Granta-519, HBL-2, NCEB-1, Rec-1, SP49, UPN-1 and Z138C) ( 10 , 11 , 47 ) and one prolymphocytic leukemia-derived (JVM-2) ( 47 ) was analyzed in this study. All cell lines studied contained CCND1-IGH gene rearrangements indicative of MCL. Granta-519, HBL-2, JVM-2, NCEB-1 and Z138C were cultured in RPMI 1640 supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin and 1% L -glutamine to a density of ∼10 6 cells/ml and genomic DNA was extracted via standard proteinase K/RNAse treatment and phenol–chloroform extraction. DNA isolation for the remaining three cell lines (Rec-1, SP49 and UPN-1) was previously described ( 10 , 11 ).
Array CGH
Copy number alterations were assayed in the eight MCL model cell lines by performing array CGH with microarrays containing 97 299 elements, representing 32 433 BAC-derived amplified fragment pools spotted in triplicate. The construction of this SMRT array was previously described ( 26 , 48 ). The overlapping clone coverage of this array allows the assessment of genomic integrity across the entire genome in a single experiment.
Six hundred nanograms of sample or pooled normal reference male genomic DNA (Novagen, Mississauga, Ontario) was denatured at 100°C for 10 min in a 33.5 µl volume containing 1× Klenow buffer, and 28 µ M random octamers (Alpha DNA, Montreal, Quebec) then cooled on ice. Fifteen nanomoles each of dATP dTTP and dGTP, 9 nmol dCTP, 45 U Klenow (Promega, Madison, WI, USA) and 4 nmol cyanine-3 dCTP or cyanine-5 dCTP, respectively, was added to the reaction for a final volume of 50 µl, mixed and incubated at 37°C overnight.
Labeled sample and reference DNA were combined and purified using ProbeQuant Sephadex G-50 Columns (Amersham, Baie d'Urfe, Quebec) followed by addition of 200 µg human Cot-1 DNA (Invitrogen, Burlington, Ontario) and precipitation with 0.1× volume 3 M sodium acetate and 2.5× volume 100% ethanol. The DNA pellet was resuspended in 100 µl hybridization solution containing 80% DIG Easy hybridization solution (Roche, Laval, Quebec), 200 µg sheared herring sperm DNA and 100 µg yeast tRNA. Resuspended probe was denatured at 85°C for 10 min and repetitive sequences blocked at 45°C for 1 h prior to hybridization. Pre-blocked probe was added to the array and placed into a hybridization chamber (Telechem, Sunnyvale, CA, USA) and hybridized at 45°C for ∼40 h.
After hybridization, arrays were washed five times for 5 min each in 0.1×SSC, 0.1% SDS at room temperature in the dark with agitation followed by five rinses in 0.1×SSC and dried using an oil free air stream.
Imaging and analysis
Images of the hybridized array were captured through cyanine-3 and cyanine-5 channels using a charge-coupled device (CCD) camera system (Applied Precision, Issaquah, WA, USA). Images were then analyzed using SoftWoRx Tracker Spot Analysis software (Applied Precision). Custom software called ‘SeeGH’ was used to visualize all data as log 2 ratio plots and this software is freely available ( http://www.flintbox.ca/technology.asp?tech=FB312FB ) ( 49 ).
To determine the experimental variation within copy number profiles using SMRT array CGH, two self versus self and five repeats of pooled male versus normal female hybridizations were performed (data available at http://www.bccrc.ca/cg/ArrayCGH_Group.html ). The standard deviation of the autosomal log 2 ratios for all control hybridizations was determined to be 0.086. Similar to Veltman et al. ( 17 ) we chose a value slightly more conservative than two times the standard deviation, yielding thresholds of +/−0.2 log 2 ratios. Therefore, genomic alterations were classified as follows: normal copy number between log 2 ratios of −0.2 and 0.2, genomic loss between log 2 ratios of −0.2 and −0.7, homozygous deletion with log 2 ratio of <−0.7, genomic gain between log 2 ratios of 0.2 and 0.5, and amplification with log 2 ratio of >0.5. Genomic imbalances and their associated breakpoints were identified using genetic local search algorithms within the software package aCGHsmooth developed by Jong et al. ( 50 ) using recommended program parameters, and subsequently confirmed visually within the primary raw normalized data (available at http://www.few.vu.nl/~vumarray/ ). Briefly, aCGHsmooth determines breakpoints within chromosomes by performing a maximum likelihood estimation for each clone by calculating the probability that the clone of interest lies within the set of previous clones. These putative breakpoints are then shifted randomly in both directions and an overall fitness is calculated. This is repeated until either no improvement in fitness can be achieved or the maximum number of iterations have been completed. The mean values of breakpoint segments are calculated and closely smoothed levels are joined to reflect the assumption that there are few copy number values present in chromosomes.
Expression microarray procedure
Lymphochip DNA microarrays ( 51 ) containing 12 196 cDNA elements were used to quantitate mRNA expression for five MCL cell lines (Granta-519, HBL-2, NCEB-1, Rec-1 and Z138C) as previously described ( 52 ).
Fluorescent in situ hybridization
FISH probes were created by labeling purified BAC DNA in random priming reactions containing either spectrum-red or spectrum-green labeled nucleotides. Metaphase spreads of Granta-519 were created from cells, at 8×10 5 cells/ml, treated with colchazine for 3 h. Hybridization and imaging was performed as previously described ( 53 ).
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG Online.
ACKNOWLEDGEMENTS
We thank Bradley P. Coe for assistance with aCGH data analysis, Baljit Kamoh and Spencer K. Watson for assistance with FISH analysis, and Ali Turhan and Catherine Tucker for providing cell lines. This work was supported by funds from Genome BC/Genome Canada and the Lymphoma Research Foundation.
Figure 1. Whole genome SMRT aCGH SeeGH karyogram of MCL cell line HBL-2 versus pooled normal male genomic DNA. Each dot represents data from one BAC-derived segment on the array. Data points to left and right of center line represent genetic losses and gains, respectively. Lines at log 2 ratios of −0.5 and +0.5 are scale bars only.
Figure 1. Whole genome SMRT aCGH SeeGH karyogram of MCL cell line HBL-2 versus pooled normal male genomic DNA. Each dot represents data from one BAC-derived segment on the array. Data points to left and right of center line represent genetic losses and gains, respectively. Lines at log 2 ratios of −0.5 and +0.5 are scale bars only.
Figure 2. Summary of chromosomal imbalances detected by SMRT aCGH in eight MCL cell line models. Lines on left side of the ideogram indicate loss of chromosomal material with thick lines representing homozygous deletion; lines on right side indicate gain of chromosomal material with thick lines indicating regions of high-level amplification. Each numbered line represents a loss or gain in a cell line: (1) Granta-519, (2) HBL-2, (3) NCEB-1, (4) Rec-1, (5) SP49, (6) UPN-1, (7) Z138C and (8) JVM-2.
Figure 2. Summary of chromosomal imbalances detected by SMRT aCGH in eight MCL cell line models. Lines on left side of the ideogram indicate loss of chromosomal material with thick lines representing homozygous deletion; lines on right side indicate gain of chromosomal material with thick lines indicating regions of high-level amplification. Each numbered line represents a loss or gain in a cell line: (1) Granta-519, (2) HBL-2, (3) NCEB-1, (4) Rec-1, (5) SP49, (6) UPN-1, (7) Z138C and (8) JVM-2.
Figure 3. Representation of genomic copy number alterations on chromosome arm 9p. Each bar represents data from one BAC-derived segment on the array with each column corresponding to the indicated cell line. Green and red bars indicate genomic copy number losses and gains, respectively.
Figure 3. Representation of genomic copy number alterations on chromosome arm 9p. Each bar represents data from one BAC-derived segment on the array with each column corresponding to the indicated cell line. Green and red bars indicate genomic copy number losses and gains, respectively.
Figure 4. Representation of genomic copy number alterations on chromosome 12. Each bar represents data from one BAC-derived segment on the array with each column corresponding to the indicated cell line. Green and red bars indicate genomic copy number losses and gains, respectively.
Figure 4. Representation of genomic copy number alterations on chromosome 12. Each bar represents data from one BAC-derived segment on the array with each column corresponding to the indicated cell line. Green and red bars indicate genomic copy number losses and gains, respectively.
Figure 5. Locus-specific FISH validation of genetic copy number alterations. ( A ) Locus-specific FISH of 13q14.3 deletion in cell line Granta-519. ( B ) Locus-specific FISH 22q11.22 homozygous deletion in cell line Granta-519. Each bar represents data from one BAC-derived segment on the array at the chromosomal location indicated. Data points to left and right of center purple line represent genetic losses and gains, respectively. Blue lines are scale bars at log 2 ratios of −0.5 and +0.5. Red and green arrows indicate loci of spectrum red and spectrum green labeled FISH probes, respectively.
Figure 5. Locus-specific FISH validation of genetic copy number alterations. ( A ) Locus-specific FISH of 13q14.3 deletion in cell line Granta-519. ( B ) Locus-specific FISH 22q11.22 homozygous deletion in cell line Granta-519. Each bar represents data from one BAC-derived segment on the array at the chromosomal location indicated. Data points to left and right of center purple line represent genetic losses and gains, respectively. Blue lines are scale bars at log 2 ratios of −0.5 and +0.5. Red and green arrows indicate loci of spectrum red and spectrum green labeled FISH probes, respectively.
Summary of genomic alterations recurring in a minimum of three MCL cell line models
| Cell line | Cytoband | Size (Mb) | Minimal region boundary clones | Number of refseq genes in region | Genes known to be differentially expressed in MCL | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Granta-519 | HBL-2 | NCEB-1 | Rec-1 | SP49 | UPN-1 | Z138 | JVM-2 | |||||
| 2p11.2 | 0.13 | 685C7–685C7 | 0 | – | ||||||||
| 9p21.3 | 1.24 | 328C2–275H17 | 4 | P15 (↓)( 2 ), P16 (↓)( 2 ) | ||||||||
| 22q11.22 | 0.39 | 757F24–50L23 | 7 | – | ||||||||
| 9p23 | 0.64 | 771D23–577M1 | 1 | – | ||||||||
| 13q14.2–13q14.3 | 0.76 | 195L15–M2217H7 | 7 | – | ||||||||
| 17p13.3–17pter | 1.14 | 818O24–M2348K1 | 11 | – | ||||||||
| 9p24.3–9pter | 0.48 | 165F24–143M1 | 2 | – | ||||||||
| p23–9p24.1 | 3.31 | 702P22–645K2 | 2 | – | ||||||||
| 7p22.1–7pter | 5.85 | 568L2–379K15 | 37 | MAD1L1 (↓)( 30 ) | ||||||||
| 8q24.13–8q24.21 | 5.56 | 497O14–279F20 | 24 | MYC (↑)( 30 ) | ||||||||
| 11p15.5–11pter | 2.50 | 583A17–401C19 | 59 | 1-8D (↓)( 29 ), H-RAS (↓)( 30 ), IFI17 (↓) ( 29 , 30 ) | ||||||||
| 2p15–2p16.1 | 1.07 | 607B19–119H15 | 6 | REL (↑) ( 30 ) | ||||||||
| 12q13.13–12q13.2 | 2.97 | 260F8–M2265L24 | 68 | RARG (↓)( 30 ) | ||||||||
| 17p11.2–17pter | 15.60 | 304M17–M2348K1 | 208 | P53 (↓)( 2 ) | ||||||||
| 18q11.2–18q21.2 | 17.07 | 1076F2–373G16 | 51 | – | ||||||||
| 1p36.11 | 1.18 | 443P17–553K16 | 23 | SRC2 (↓)( 29 ) | ||||||||
| 1p21.1–1p31.1 | 35.25 | 478L17–437B7 | 111 | BCL-10 (↓)( 29 ), CNN3 (↑)( 29 ), GBP1 (↓)( 29 ), TGFBR3 (↓)( 30 ) | ||||||||
| 2q37.1 | 1.66 | 707E3–599F11 | 14 | – | ||||||||
| 2q37.2 | 0.87 | 367B19–473L20 | 2 | – | ||||||||
| 18q21.33–18q22.1 | 7.26 | 155A6–177K16 | 23 | BCL2 (↑)( 29 , 30 ) | ||||||||
| 1p36.32–1pter | 3.10 | 151F10–34P13 | 37 | – | ||||||||
| 12q24.21–qter | 19.10 | 749J2–M2140B24 | 115 | – | ||||||||
| 1q32.2–1q32.3 | 4.69 | 818N18–757D10 | 31 | – | ||||||||
| 2q37.3 | 2.42 | 546M8–811O7 | 23 | – | ||||||||
| 7p11.2–7p21.2 | 40.42 | 760D02–160E4 | 171 | NMB (↑)( 29 ) | ||||||||
| 7q11.23 | 4.31 | N1328G23–85K14 | 49 | – | ||||||||
| 8q22.1–8q22.3 | 4.79 | 655F3–486H6 | 22 | – | ||||||||
| 9q34.2–9q34.3 | 4.31 | 746P3–424E7 | 62 | – | ||||||||
| 12q14.2 | 0.37 | 749L20–415I12 | 2 | |||||||||
| 13q31.3 | 1.16 | 511F12–487A2 | 1 | |||||||||
| 12q13.2–12q14.1 | 3.41 | 644F5–1P10 | 82 | CD63 (↑)( 29 ), CDK4 (↑)( 30 ) | ||||||||
| 12p11.21–12p12.3 | 12.31 | 48N6–607H18 | 51 | – | ||||||||
| 12q15–12q21.2 | 7.27 | 428C23–401H20 | 26 | – | ||||||||
| 8p23.3 | 0.43 | 91J19–130K11 | 3 | – | ||||||||
| 8p21.2–p21.3 | 0.73 | 784K2–806N20 | 12 | – | ||||||||
| Cell line | Cytoband | Size (Mb) | Minimal region boundary clones | Number of refseq genes in region | Genes known to be differentially expressed in MCL | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Granta-519 | HBL-2 | NCEB-1 | Rec-1 | SP49 | UPN-1 | Z138 | JVM-2 | |||||
| 2p11.2 | 0.13 | 685C7–685C7 | 0 | – | ||||||||
| 9p21.3 | 1.24 | 328C2–275H17 | 4 | P15 (↓)( 2 ), P16 (↓)( 2 ) | ||||||||
| 22q11.22 | 0.39 | 757F24–50L23 | 7 | – | ||||||||
| 9p23 | 0.64 | 771D23–577M1 | 1 | – | ||||||||
| 13q14.2–13q14.3 | 0.76 | 195L15–M2217H7 | 7 | – | ||||||||
| 17p13.3–17pter | 1.14 | 818O24–M2348K1 | 11 | – | ||||||||
| 9p24.3–9pter | 0.48 | 165F24–143M1 | 2 | – | ||||||||
| p23–9p24.1 | 3.31 | 702P22–645K2 | 2 | – | ||||||||
| 7p22.1–7pter | 5.85 | 568L2–379K15 | 37 | MAD1L1 (↓)( 30 ) | ||||||||
| 8q24.13–8q24.21 | 5.56 | 497O14–279F20 | 24 | MYC (↑)( 30 ) | ||||||||
| 11p15.5–11pter | 2.50 | 583A17–401C19 | 59 | 1-8D (↓)( 29 ), H-RAS (↓)( 30 ), IFI17 (↓) ( 29 , 30 ) | ||||||||
| 2p15–2p16.1 | 1.07 | 607B19–119H15 | 6 | REL (↑) ( 30 ) | ||||||||
| 12q13.13–12q13.2 | 2.97 | 260F8–M2265L24 | 68 | RARG (↓)( 30 ) | ||||||||
| 17p11.2–17pter | 15.60 | 304M17–M2348K1 | 208 | P53 (↓)( 2 ) | ||||||||
| 18q11.2–18q21.2 | 17.07 | 1076F2–373G16 | 51 | – | ||||||||
| 1p36.11 | 1.18 | 443P17–553K16 | 23 | SRC2 (↓)( 29 ) | ||||||||
| 1p21.1–1p31.1 | 35.25 | 478L17–437B7 | 111 | BCL-10 (↓)( 29 ), CNN3 (↑)( 29 ), GBP1 (↓)( 29 ), TGFBR3 (↓)( 30 ) | ||||||||
| 2q37.1 | 1.66 | 707E3–599F11 | 14 | – | ||||||||
| 2q37.2 | 0.87 | 367B19–473L20 | 2 | – | ||||||||
| 18q21.33–18q22.1 | 7.26 | 155A6–177K16 | 23 | BCL2 (↑)( 29 , 30 ) | ||||||||
| 1p36.32–1pter | 3.10 | 151F10–34P13 | 37 | – | ||||||||
| 12q24.21–qter | 19.10 | 749J2–M2140B24 | 115 | – | ||||||||
| 1q32.2–1q32.3 | 4.69 | 818N18–757D10 | 31 | – | ||||||||
| 2q37.3 | 2.42 | 546M8–811O7 | 23 | – | ||||||||
| 7p11.2–7p21.2 | 40.42 | 760D02–160E4 | 171 | NMB (↑)( 29 ) | ||||||||
| 7q11.23 | 4.31 | N1328G23–85K14 | 49 | – | ||||||||
| 8q22.1–8q22.3 | 4.79 | 655F3–486H6 | 22 | – | ||||||||
| 9q34.2–9q34.3 | 4.31 | 746P3–424E7 | 62 | – | ||||||||
| 12q14.2 | 0.37 | 749L20–415I12 | 2 | |||||||||
| 13q31.3 | 1.16 | 511F12–487A2 | 1 | |||||||||
| 12q13.2–12q14.1 | 3.41 | 644F5–1P10 | 82 | CD63 (↑)( 29 ), CDK4 (↑)( 30 ) | ||||||||
| 12p11.21–12p12.3 | 12.31 | 48N6–607H18 | 51 | – | ||||||||
| 12q15–12q21.2 | 7.27 | 428C23–401H20 | 26 | – | ||||||||
| 8p23.3 | 0.43 | 91J19–130K11 | 3 | – | ||||||||
| 8p21.2–p21.3 | 0.73 | 784K2–806N20 | 12 | – | ||||||||
Aligned with data from University of California, Santa Cruz Human Genome Browser April 2003 Freeze. Novel regions are in bold. Gray shading indicates loss; dark shading indicates gain. (↑) indicates over-expression and (↓) indicates under-expression in the reference mentioned.
