GeneChip analyses point to novel pathogenetic mechanisms in mantle cell lymphoma

Authors


Dr rer. nat. Inga Vater, Institute of Human Genetics, Christian-Albrechts University Kiel, Schwanenweg 24, D-24105 Kiel, Germany.
E-mail: ivater@medgen.uni-kiel.de

Summary

The translocation t(11;14)(q13;q32) is the genetic hallmark of mantle cell lymphoma (MCL) but is not sufficient for inducing lymphomagenesis. Here we performed genome-wide 100K GeneChip Mapping in 26 t(11;14)-positive MCL and six MCL cell lines. Partial uniparental disomy (pUPD) was shown to be a recurrent chromosomal event not only in MCL cell lines but also in primary MCL. Remarkably, pUPD affected recurrent targets of deletion like 11q, 13q and 17p. Moreover, we identified 12 novel regions of recurrent gain and loss as well as 12 high-level amplifications and eight homozygously deleted regions hitherto undescribed in MCL. Interestingly, GeneChip analyses identified different genes, encoding proteins involved in microtubule dynamics, such as MAP2, MAP6 and TP53, as targets for chromosomal aberration in MCL. Further investigation, including mutation analyses, fluorescence in situ hybridisation as well as epigenetic and expression studies, revealed additional aberrations frequently affecting these genes. In total, 19 of 20 MCL cases, which were subjected to genetic and epigenetic analyses, and five of six MCL cell lines harboured at least one aberration in MAP2, MAP6 or TP53. These findings provide evidence that alterations of microtubule dynamics might be one of the critical events in MCL lymphomagenesis contributing to chromosomal instability.

Mantle cell lymphoma (MCL) is an aggressive B-cell non-Hodgkin lymphoma (B-NHL) genetically characterised by the translocation t(11;14)(q13;q32) that leads to overexpression of cyclin D1 (Williams et al, 1992). Based on studies with transgenic mice, it is well established that this chromosomal event alone is not sufficient to result in lymphomagenesis and that secondary genomic alterations are required for malignant transformation (Lovec et al, 1994; Gladden et al, 2006). In this multistep transformation process, tumour suppressor gene (TSG) inactivation has been shown to be a key mechanism. In MCL, frequently targeted TSGs encode for proteins that inhibit malignant transformation by protecting the genome from DNA damage (ATM) or from deregulated cell cycle progression (CDKN2A/P16, RB1) or by inducing apoptosis in cells with a disrupted cell cycle control (TP53)(Nielander et al, 2007).

MCL has been the subject of several array-based comparative genomic hybridisation (arrayCGH) studies, and genomic aberrations in addition to t(11;14)(q13;q32) have been extensively characterised (Kohlhammer et al, 2004; de Leeuw et al, 2004; Rubio-Moscardo et al, 2005a,b; Tagawa et al, 2005; Mestre-Escorihuela et al, 2007). Except one, all these studies used microarrays consisting of bacterial artificial chromosome (BAC) or P1-artificial chromosome (PAC) clones for which maximal resolution is restricted to 50–200 kb. Microarrays consisting of short synthetic oligonucleotides provide higher resolution. Some platforms allow genotyping simultaneously, so that loss of heterozygosity (LOH) without changes in DNA copy number is also detectable. This chromosomal event, so called partial uniparental disomy (pUPD), has been recently reported as a frequent mechanism of TSG inactivation in haematological neoplasms and solid tumours (Bignell et al, 2004; Bruce et al, 2005). In 2006, we could show that pUPD is a recurrent genetic mechanism alternative to chromosomal deletion in MCL cell lines (Nielaender et al, 2006). By loss of (part of) one parental chromosome and duplication of the remaining one from the other parent, pUPD results in LOH without chromosomal deletion. As the gene dosage is not altered, pUPD cannot be detected by conventional arrayCGH.

In the present study, we used 100 K GeneChip arrays to determine genomic imbalances and pUPD in a series of 26 MCL patients and six MCL cell lines. Taking advantage of the high resolution of this array, we focused on detection of small homozygous deletions. Moreover, this microarray enabled the identification of pUPD. The detection of these two genetic alterations has been shown to be a promising strategy to identify novel TSGs (Nielander et al, 2007). We identified minimally altered regions targeted by homozygous deletions harbouring novel candidate TSG loci, which might be associated with MCL pathogenesis. Moreover, we could show that pUPD is a recurrent chromosomal event not only in MCL cell lines, but also in MCL primary cases and leads to inactivation of TSGs. This further confirms that pUPD has to be considered as an alternative mechanism of TSG inactivation in MCL. Finally, our study identified genes encoding microtubule-associated proteins to be frequently targeted by chromosomal and epigenetic aberrations in MCL.

Methods and materials

Tissue samples

DNA-samples and fixed cells were selected from the files of the Institute of Human Genetics, the Second Medical Department and the Institute of Haematopathology, (University Hospital Schleswig-Holstein, Campus, Kiel, Germany). Genomic DNA was extracted from frozen dimethyl sulphoxide stocks derived from lymph node (LN), peripheral blood (PB), bone marrow (BM) and spleen (S) or from frozen tissue blocks of the affected LN using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). The MCL panel for the single nucleotide polymorphism (SNP) array comprised 26 patients with primary MCL. The median age of the patients at diagnosis was 66 years (range, 38–84 years). A subset of cases of this panel was also used for fluorescence in situ hybridisation (FISH) (= 13), MAP2 mutation analyses (n = 20), TP53 mutation analyses (n = 26) and epigenetic studies (n = 20). For the FISH screens and denaturing high performance liquid chromatography (DHPLC) analyses of MAP2, an additional 14 and 20 t(11;14)-positive MCL cases from files of the Institute of Human Genetics (University Hospital Schleswig-Holstein, Campus, Kiel, Germany) were included, respectively. Table SI shows all MCL samples included in this study. Tonsils and PB of healthy donors were used as non-tumours controls. For real time RT-PCR experiments, LN cDNA pooled from 12 Caucasians (aged 20–59 years) and purchased from the human immune system multiple tissue cDNA panel (Clontech Laboratories, Mountain View, CA, USA) and a tonsil freshly prepared from the Institute of Haematopathology (University Hospital Schleswig-Holstein, Campus, Kiel, Germany) were used as normal controls. The study was performed in the framework of the ‘European MCL Network’, for which central and local ethics approval was obtained.

Cell lines

A panel of six MCL cell lines was studied: GRANTA-519, HBL-2, UPN-1, REC-1, MAVER-1 and JEKO-1. A compilation of studies characterising these cell lines is shown in Table SII.

Microarrays

The GeneChip Human Mapping 100 K array set (Affymetrix, Santa Clara, CA, USA) has been used according to the protocol provided by the manufacturer (Affymetrix) (http://www.affymetrix.com). Microarrays were washed and stained with the Fluidics Station 450 (Affymetrix) and scanned with the GeneChip Scanner 3000 (Affymetrix) using GeneChip Operating System (GCOS; Affymetrix) version 1.4. The BRLMM algorithm (http://www.affymetrix.com/support/technical/whitepapers/brlmm_whitepaper.pdf) was used with default parameters (score threshold = 0·5, prior size = 10 000 and DM threshold = 0·17) to genotype MCL tumour samples in combination with Hapmap reference arrays (http://www.affymetrix.com/support/technical/whitepapers/brlmm_whitepaper.pdf). The genotyping call rates of the hybridised SNP chips ranged from 96·06% to 99·88% (median = 98·59%) for the 50K XbaI array and from 94·54% to 99·77% (median = 98·92%) for the 50 K HindIII array. Ninety HapMap samples provided by Affymetrix (30 CEPH trios) were used as normal reference arrays (http://www.affymetrix.com/support/technical/sample_data/hapmap_trio_data.affx). A complete list of Affymetrix reference samples is shown in Table SIII.

Copy number analysis

Copy number analysis was performed using the Copy Number Analyser for GeneChip (CNAG) program v2.0 (Nannya et al, 2005). The optimised reference selection method implemented in CNAG was used (Nannya et al, 2005). Thus, CNAG selected a gender-specific reference set out of the 90 controls individually for each array. XbaI and HindIII arrays were combined for the analysis. Segmentation of raw copy number data was performed using the Hidden Markov Model (HMM) approach provided by CNAG. HMM parameters were adjusted individually for each array due to differences in hybridisation quality and tumour cell content. Starting with default parameters, the mean levels of HMM states were adjusted manually to increase smoothing for samples with higher noise while increasing sensitivity for low noise samples. With regard to outliers and technical artefacts, HMM segments with aberrant copy number were considered as a copy number aberration only if they consisted of at least 10 consecutive SNPs. High-level amplifications were defined as aberrations with HMM copy number ≥5. For male cases, the call of high-level amplifications on chromosome X was copy numbers ≥4.

Liberal screening for homozygous deletions.  Homozygous deletions were defined as aberrations with copy number = 0.

Sensitive screening for homozygous deletions.  In this more sensitive approach for detecting homozygous deletions, the Copy Number Analysis Tool (CNAT; Affymetrix) version 2.0 was used to calculate the copy number (CN), applying a 0·5 Mb genome smoothing filter. The data set was screened for regions with a copy number ≤0·6 in a minimum of two adjacent SNPs of which at least one showed a ‘NoCall’.

LOH analysis

A HMM-based method (Beroukhim et al, 2006) implemented in the dChip program (Lin et al, 2004; Zhao et al, 2004) (Build date: Apr 11 2007) was used to infer regions with LOH from tumour samples. The HMM considering haplotype (LD-HMM) (Beroukhim et al, 2006) method was selected for the LOH calculations to account for linkage disequilibrium (LD)-induced SNP dependencies. The LOH call threshold was set to the default value of 0·5. An empirical haplotype correction (Beroukhim et al, 2006) was applied. Thus, the genotypes of putative tumour-associated LOH regions were compared with the genotypes observed in euploid reference samples. If the genotypes in the respective region were highly concordant between the tumour sample and at least 5% of the normal reference samples the LOH region was rejected by dChip.

Partial uniparental disomy

Partial UPD regions represent genomic regions in which the LOH was not caused by altered copy number. A LOH region determined by dChip was called UPD if the copy number analysis using CNAG revealed no aberrations within that region. If a LOH region was partially affected by copy number aberrations, subregions with normal copy number were called UPD if they comprised at least 50 neighbouring SNPs.

Copy number polymorphisms

As an attempt to distinguish copy number aberrations from copy number variations (CNVs) present in healthy individuals, aberrant regions were compared to published data included in the ‘Database of Genomic Variants’ (http://projects.tcag.ca/variation/). Regions showing overlap ≥50% with known genomic variants were classified as CNVs.

Interphase FISH

The commercially available locus-specific identifier (LSI) IGH/CCND1 dual colour probe (Abbott/Vysis, Downers Grove, IL, USA) was applied to detect translocation t(11;14)(q13;q32). Differentially labeled bacterial artificial chromosome (BAC) clones, fosmid clones or commercial centromere probes (Abbott/Vysis) were applied to verify copy number results from 100K GeneChip analyses and to perform FISH screenings. A summary of the used FISH assays is shown in Table SIV. FISH was performed as published elsewhere (Martin-Subero et al, 2002). Hundred nuclei were evaluated per hybridisation whenever possible.

Delineation of homozygous deletions

Polymerase chain reaction (PCR)-based methods were used to confirm homozygous deletions detected in the SNP array data of the MCL cell lines. To identify novel candidate TSGs, regions of homozygous loss were delineated by PCR. Primer sequences, polymerase enzymes and PCR conditions are summarised in Table SV.

Mutation analyses

The coding exons and exon/intron-boundaries of MAP2 were amplified and analysed by DHPLC as previously described (Arnold et al, 1999). Sequences and annealing temperatures of primer pairs are shown in Table SV. PCR products that showed aberrant chromatograms were subjected to direct sequencing using the Big Dye Terminator v1.1 Cycle Sequencing Kit and the Genetic Analyser ABI PRISM 310 system (Applied Biosystems, Foster City, CA, USA). TP53 mutation analysis was performed as previously described (Gross et al, 2001). Sequence variations were checked and named according to the R12 release of the IARC TP53 Database (Petitjean et al, 2007) to determine whether they are rare polymorphisms or deleterious.

Methylation studies

Sodium bisulfite conversion of genomic DNA was performed with EpiTect Bisulfite Kit (Qiagen). The methylation status of MAP2 was evaluated by methylation-specific PCR (MSP) (Herman et al, 1996) as well as bisulfite sequencing (Frommer et al, 1992). For bisulfite sequencing, the whole CpG island upstream of MAP2 was amplified after DNA bisulfite treatment, ligated into a pCR2.1-TOPO vector and transformed into competent Escherichia coli (TOP10) using the TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA, USA). Insertion of the DNA fragments was tested by PCR of the colonies. Cloned DNA was amplified from 10 positive colonies using vector-specific primers followed by direct sequencing using Big Dye Terminator v1.1 Cycle Sequencing Kit and the Genetic Analyser ABI PRISM 310 system (Applied Biosystems). Primer sequences for MSP and bisulfite sequencing, the used polymerase enzymes and PCR conditions used in the analyses are shown in Table SV. To control the result of bisulfite conversion and the specificity of the MSP reactions, we used CpGenome Universal Methylated and CpGenome Universal Unmethylated DNA (Millipore, Billerica, MA, USA).

Expression studies

MAP2 and MAP6 expression in MCL cell lines was analysed by SYBR Green-based real-time RT-PCR using the iCycler iQ multi-color-real-time PCR detection system (Biorad, Hercules, CA, USA). Total RNA was isolated from cultured cells with NucleoSpin RNA/Protein Kit (Macherey-Nagel, Düren, Germany), high quality was confirmed with the Experion automated electrophoresis system (Biorad) and cDNA synthesis was performed using the QuantiTect Reverse Transcription Kit (Qiagen). For the real-time PCR of MAP2 we used QuantiTect SYBR Green PCR Kit and the QuantiTect Primer Assays of MAP2 (Hs_ MAP2_1_SG) and MAP2 alternative transcripts (Hs_MAP2_va.1_SG) (Qiagen). These two primer assays are able to detect all known transcript variants of MAP2. For MAP6 real time PCR we used the QuantiTect Primer Assay Hs_ MAP6_1_SG. Normalisation for the quantity of cDNA was done by performing simultaneous real-time RT-PCR for hypoxanthine phosphoribosyltranferase 1 (HPRT1), β-glucuronidase (GUSB) and glucose-6-phosphat-dehydrogenase (G6PD) with appropriate QuantiTect Primer Assays (Hs_HPRT1_1_SG, Hs_GUSB_1_SG, and Hs_G6PD_1_SG) (Qiagen). The thermal profile for the SYBR Green-based PCRs consisted of 15 min Taq polymerase activation at 95°C followed by 45 cycles of PCR at 94°C for 15 s (denaturation), 55°C for 30 s (annealing), and 72°C for 30 s (extension). Following amplification, a melting curve analysis was performed to verify the correct product by its specific melting temperature (Tm). Melting curve analysis consisted of a denaturation step at 95°C for 1 min, lowered to 55°C for 30 s, and followed by 80 cycles of incubation in which the temperature is increased to 95°C at a rate of 0·5°C/10s/cycle with continuous reading of fluorescence. Results were analysed by iCycler iQ Optical Software Version 3.0a and ratios were calculated by ΔΔCT method.

Results

GeneChip array data

In this study, samples from 26 MCL patients (20 male and six female) were subjected to 100K GeneChip Mapping analyses. The characteristic translocation t(11;14)(q13;q32) was confirmed by cytogenetics and/or FISH. Based on FISH, the tumour cell content varied from 21% to 95% (mean 75%) (Table SI). In addition, a panel of six t(11;14)-positive MCL-derived cell lines was studied. Results of the LOH analysis based on this 100K GeneChip array data have already been published for five of these MCL cell lines (Nielaender et al, 2006).

Gains and losses.  Copy number analyses identified a pattern of genomic imbalances typical for MCL, including low-level gain and monoallelic loss of regions that are known to be recurrently affected in MCL. A genomic overview of gains and losses detected in the investigated MCL primary cases is given in Fig 1. Two hundred thirty-six and 204 genetic events leading to monoallelic loss were identified in the primary MCL and MCL cell lines, respectively (mean of 9 per MCL and 34 per MCL cell line). In addition to deletions, 206 low-level copy gains were detected in the primary MCLs (mean of 8 per MCL) and 181 in the MCL cell lines (mean of 30 per MCL cell line). By overlapping genomic segments showing copy number changes, we delineated the minimal altered regions (MARs) as well as minimal peak regions within these MARs, which were determined by imbalances of single cases (Fig SI). Chromosomal regions that were affected in at least four primary MCL or MCL cell lines are summarised in Table SVI. Frequently affected regions of chromosomal gain were identified in 1q23.3-q25.1, 2p25.1, 2q32.3, 3q27.3-3q28, 4q35.1-q35.2, 7p22.3-p15.3, 7p12.1, 7q21.11, 8p23.3, 8q22.1-q24.21, 10p15.3, 10p12.1-p11.22, 11q13.3-q21, 12q13.3-q14.1, 13q31.3, 14q11.2, 15q21.3, 17q23.2, 18q21.2-q22.3 and 19p12. Genomic loss was frequently detected in 1p33-p32.3, 1p31.1-p21.1, 2q13, 2q24.1-q31.2, 2q37.1, 3p14.2-p12.2, 6q23.3-q27, 8p23.2-p21.2, 9p24.3-p21.2, 9q13-q31.2, 10p14-p13, 11q22.3-q23.2, 12p13.2-p13.1, 13q12.3-q13.1, 13q14.2-q14.3, 13q22.2-q31.1, 13q34, 14q32.12, 15q11.2, 17p13.3-p12, 22q11.22, 22q13.2-q13.33, and Xp22.33. Some of these regions contain known CNVs, also present in healthy individuals, and are marked in Table SVI. In addition to common alterations previously reported in MCL, 100K GeneChip mapping identified five regions frequently targeted by chromosomal gain as well as seven regions of recurrent genomic loss, which have not been reported yet. Some of these novel regions contain candidate genes showing typical tumour suppressor or oncogene properties, such as FHIT in 3p13, CDKN1B in 12p13.1, MYCBP in 13q22.3, or CDKN2AIP and ING2 in 4q35.1 (Table SVI).

Figure 1.

 Genome-wide detection of copy number changes in MCL primary cases. Proportion of gains and losses analysing GeneChip data of 26 MCL primary cases are displayed from 1pter to Xqter. Light grey columns indicate chromosomal gain, whereas dark grey columns indicate loss of genetic material.

High-level amplifications.  Regions showing a copy number ≥5 were classified as high-level amplified. Amplicons included genes previously described as being amplified or overexpressed in MCL, such as BCL2 (18q21.33), GPC5/MIRN17 (13q31.3) and BMI1 (10p12.2) (Bea et al, 2001; de Leeuw et al, 2004; Rubio-Moscardo et al, 2005b). Novel detected regions of genomic amplification in MCL involved chromosomes 4p14, 6p21.2, 8p23.1, 8p23.1-p22, 8q24.21, 11q13.4-q13.5, 11q14.1, 11q22.3, 11q23.1-q23.2, 17p11.2, 18q12.2 and 18q12.2-q12.3 (Table SVII). Noteably, most of the detected high-level amplifications (∼90%) were observed in the MCL cell lines. There were only two primary MCL cases showing this kind of aberration. Each of these newly detected amplified regions was observed in only one sample with the exception of 11q13.4-q13.5, which was detected in three MCL cell lines. This region harbours nine genes including MAP6 (microtubule-associated protein 6), SERPINH1 (serine proteinase inhibitor 1), GDPD5 (glycerophosphodiester phosphodiesterase domain 5), RPS3 (ribosomal Protein S3), UVRAG (UV radiation resistance associated gene), WNT11 (wingless-type MMTV integration site family), MOGAT2 and DGAT2 (mono- and diacylglycerol O-acyltransferases), as well as the hypothetical protein LOC283212. In addition, this chromosomal region is coding for 2 snoRNAs (SNORD15A and SNORD15B).

Homozygous deletions.  Two different approaches were used for the identification of candidate homozygously deleted regions in the MCL GeneChip data. In the liberal screening of the MCL cell lines, 11 homozygous deletions were identified in eight different chromosomal regions and, except the CNV, all of them were confirmed by PCR and/or FISH (Table I). Applying the sensitive screening method, 19 additional candidate homozygous deletions, all affecting different loci, were identified in MCL cell lines; four were validated by PCR and/or FISH, two were consistent with known CNVs (Table I). The additional 13 regions suggestive for homozygous loss could not be confirmed. Noteably, this evaluation criteria detected every previously reported homozygous deletion in MCL cell lines, including TSG loci in 1p32.3 (CDKN2C/P18), 2q13 (BCL2L11), 9p21.3 (CDKN2A/P16), and 13q14.2 (RB1) (Nielander et al, 2007). Nevertheless, 43% of the regions suggestive for homozygous deletion represented false positive findings. However, four PCR-proven homozygously deleted regions and two CNVs detected by the sensitive screening were not detected by the liberal approach, most likely due to smoothing of the data. Table I displays the verified regions of homozygous loss in the MCL cell lines detected by both approaches. Moreover, the borders of five novel deleted regions were delineated by PCR to identify potential TSGs (Table I).

Table I.   Homozygous deletions in MCL cell lines.
Locus*Liberal screening [start-stop in bp*]Sensitive screening [start-stop in bp*]PCR-based confirmation and delineation [start-stop in bp*]FISH confirmationAffected cell linesTarget genes
  1. CNV, copy number variation.

  2. *NCBI Build 35.

  3. †According to database of genomic variants.

1p32.3Not detected50 837 181-57 453 350(Mestre-Escorihuela et al, 2007)Not doneUPN-1CDKN2C
2q13111 616 112-112 182 931111 616 112-112 155 057(Tagawa et al, 2005)Not doneJEKO-1BCL2L11
2q34210 048 358-210 472 004210 048 358-210 472 004210 079 601-210 462 202ConfirmedUPN-1MAP2
2q37.3Not detected237 396 643-237 486 004237 272 264-237 637 712ConfirmedUPN-1No genes
9p21.321 948 524-22 102 59921 948 524-22 090 176(Kohlhammer et al, 2004; de Leeuw et al, 2004; Tagawa et al, 2005)Not doneGRANTA-519, REC-1, MAVER-1CDKN2A, CDKN2B
9p21.227 316 780-27 716 91127 248 185-27 716 73127 287 093-27 940 630Not doneMAVER-1MOBKL2B, IFNK, C9orf72
9p21.128 761 537-30 316 11528 761 537-30 316 115Confirmed but not delineated by PCRNot doneMAVER-1No genes
12p13.1Not detected13 617 759-13 618 07813 616 272-13 652 667Not doneREC-1GRIN2B
13q14.2Not detected47 812 793-47 817 924(Pinyol et al, 2007)Not doneUPN-1RB1
13q33.1101 552 480-101 943 751101 552 480-101 943 751Confirmed by FISH not delineated by PCRConfirmedJEKO-1FGF14
18q22.162 853 023-63 491 35863 097 606-63 491 35862 390 461-64 346 179ConfirmedUPN-1DSEL
22q11.2220 994 635-21 479 13620 994 635-21 479 136CNV†Not doneGRANTA-519, HBL-2IGL@
Xp22.31Not detected6 492 343-6 543 186CNV†Not doneJEKO-1No genes
Xq28Not detected153 981 072CNV†Not doneGRANTA-519No genes

In the 26 primary MCL cases, 15 and six regions met the criteria for homozygous deletion in the sensitive and liberal approach, respectively. Confirmation by PCR-based methods was not performed due to contamination of the tumour samples by normal surrounding tissue. FISH was only applied for verification of deletions exceeding ∼100 kb (i.e. the insert size of the used BAC, PAC or fosmid clones) that were not classified as CNVs. Homozygous deletion could be confirmed in 9p21.3 and in Xp22.33. In Table II, regions of homozygous loss, which were detected by GeneChip data analysis of the MCL primary cases, are summarised with regard to size, CNVs, target genes and FISH verification.

Table II.   Putative and confirmed homozygous deletions in MCL primary cases.
Locus*Sensitive screening [start-stop in bp*]Liberal screening [start-stop in bp*]Number of affected MCLFISH confirmationCandidate genes
  1. CNV, copy number variation; PAR1, pseudoautosomal region 1.

  2. *NCBI Build 35.

  3. †According to database of genomic variants.

1p21.2100 823 226-100 823 636Not detected1Not done (<100 kb)No genes
3p26.31 510 440-1 511 278Not detected1Not done (CNV†)No genes
4q22.190 356 219-90 356 586Not detected1Not done (CNV†)No genes
5q22.2111 710 742-111 711 216Not detected1Not done (CNV†)EPB41L4A
9p24.17 501 166-7 503 589Not detected1Not done (CNV†)No genes
9p2312 742 340-12 750 7188 916 956-12 974 7561Not done (CNV†)PTPRD, TYRP1
9p21.3Not detected21 762 317-22 801 3362Homozygous deletion verifiedCDKN2A, CDKN2B
9p21.227 692 974-27 693 855Not detected1Not done (<100 kb)No genes
9p21.1Not detected30 425 441-31 302 4601Not done (CNV†)No genes
11q22.2100 654 097-100 773 657100 546 098-102 615 3491No assay applicable due to repetitive sequencesBIRC2, BIRC3
11q22.3109 749 455-109 749 570Not detected1Not done (CNV†)No genes
12p13.113 617 759-13 618 078Not detected1Not done (<100 kb)GRIN2B
22q11.2221 099 653-21 337 56121 099 653-21 479 1361Not done (CNV†)IGL@
Xp22.33677 050-2 528 646677 050-2 561 0082Homozygous deletion verifiedPAR1 genes
Xp21.229 303 376-29 351 134Not detected1Not done (CNV†)IL1RAPL1
Xp21.133 396 094-33 415 950Not detected1Not done (CNV†)No genes
Xq21.3398 159 771-98 164 282Not detected2Not done (<100 kb)No genes

Loss of heterozygosity in regions with normal copy number.  The 100K GeneChip data of the primary MCL cases was also subjected to LOH analyses in order to identify regions of pUPD. A median number of 1.6 pUPDs (ranging from 0 to 5) was identified in the samples. As previously reported for MCL cell lines (Nielaender et al, 2006), regions frequently affected by pUPD in primary MCL are known to be commonly targeted by deletions in MCL, e.g. 11q and 13q (Fig 2A). Some of the detected regions showing pUPD are known to harbour common TSGs, like TP53 in 17p13.1.

Figure 2.

 Partial uniparental disomy (pUPD) in MCL. (A) Genome-wide distribution of pUPD in 26 primary MCL cases detected by 100K GeneChip Mapping (Affymetrix). Proportion of MCL cases showing pUPD is displayed in turquoise. Chromosomes are shown from 1pter (left) to Xqter (right). (B) Detection of pUPD in the short arm of chromosome 17 in MCL 8. The profile of genomic imbalances is given as black dots with a value of 2 indicating a balanced status. Chromosome 17 is shown from pter (left) to qter (right). Heterozygous calls are given as green dots and the estimated LOH region is overlaid in grey. A blue arrow indicates chromosomal location of the TP53 gene. (C) FISH analysis using a locus-specific probe consisting of BAC clones labelled in spectrum orange (TP53) and spectrum green (control locus in 17q21.2). Signal constellation indicates a normal gene dosage of two copies in the MCL 8. (D) Homozygous point mutation affecting exon 8 of TP53 in the same MCL showing pUPD in 17p13.1.

Mutation analyses of TP53 in one MCL sample displaying pUPD in the short arm of chromosome 17 (Fig 2B) revealed a homozygous missense mutation of a single nucleotide (g.14490 T>A) in exon 8, affecting a DNA binding domain (Fig 2C). This mutation has been reported to be deleterious for TP53-DNA interaction (http://www-p53.iarc.fr). FISH analysis using a TP53 locus-specific probe confirmed a normal gene dosage of two copies (Fig 2D).

Involvement of genes encoding microtubule-associated proteins in MCL

Remarkably, 100K GeneChip data and further investigation of candidate genes identified genes encoding different microtubule-associated proteins, such as MAP2, MAP6 and TP53, to be recurrently affected by chromosomal aberrations in MCL. Although it is widely known that MCL karyotypes show high complexity, alterations in microtubule organisation have not been reported so far.

MAP6 gene.  High-level amplification of MAP6 was confirmed in all three MCL cell lines MAVER-1, JEKO-1 and HBL-2 by FISH using a BAC clone spanning the MAP6 locus. A combined analysis of FISH and R-banding revealed the existence of two derivative chromosomes 6 harbouring the amplified 11q13.5 region in MAVER-1 (Fig 3). FISH screening of interphase nuclei from 27 primary MCL identified one case with five gene copies, although the control gene locus in 11q22.3 was also involved. To investigate whether gene dosage affected MAP6 expression, we performed real-time RT-PCR in the six MCL cell lines. Higher expression of MAP6 was detected in JEKO-1, GRANTA-519 and REC-1 compared to a freshly prepared tonsil, which was used as normal control (Fig S2).

Figure 3.

 High level amplification in 11q13.5. (A) Genomic profiles of part of chromosome 11 in three MCL cell lines. Copy number is given as black dots with a value of 2 indicating a balanced status. Chromosomal region in the long arm is shown from centromeric to telomeric. The minimal altered region is indicated by blue vertical bars. (B–D) Interphase FISH of the MCL cell lines using a locus-specific probe for MAP6 (spectrum green) and a control locus in 11q22.3 (spectrum orange). (E–F) R banding followed by FISH on metaphases of MAVER-1 using the described MAP6 probe. High level amplification was shown to affect two derivative chromosomes 6 [a:der(6)(6pter?6q22::11q13?::11q22?11q23::11q22->11q23::?),b:der(6)(?::6p21?6q22::11q13?11q14::11q23? amp)].

MAP2 gene.  Homozygous deletion of MAP2 detected in the MCL cell line UPN-1 was confirmed by PCR and FISH. In addition, the REC-1 cell line showed a heterozygous deletion of part of the long arm of chromosome 2, including MAP2. FISH screening of the additional MCL cell lines and 27 primary cases failed to detect further cases with chromosomal loss affecting MAP2.

Mutation analysis of MAP2 was performed in 40 primary MCL and five MCL cell lines by DHPLC of the coding exons and the exon/intron boundaries followed by sequencing aberrant fragments. DHPLC screening of MAP2 identified a G>A transition in one MCL case (MCL 16). This mutation affected the first position of a coding triplet in exon 5 causing an amino acid change from glutamic acid (E) to lysine (K) (Fig 4). There was no evidence for a SNP at this DNA sequence position in NCBI Build database and DHPLC analyses failed to detect this alteration in 50 DNA samples (100 alleles) derived from peripheral blood of healthy individuals (data not shown).

Figure 4.

 Homozygous deletion in 2q34. (A) Genomic profile of part of the chromosome 2 in the MCL cell line UPN-1. Copy number is given as black dots with a value of 2 indicating a balanced status. Chromosomal region in the long arm is shown from centromeric to telomeric. A red arrow indicates the homozygously deleted region. A Genome Browser extract (NCBI Build 35) displays gene content and the chromosomal position of primers (red bars) for delineation of the minimal affected region. (B) Multiplex-PCR of the target gene and a control locus confirmed the homozygous deletion of MAP2 but not of the adjacent coding sequences of RNAz s59616 and C2orf21. (C) DHPLC chromatograms of the primary MCL 16 (red) and a healthy control (blue). (D) Sequencing of the aberrant fragment revealed a heterozygous mutation in exon 5 of MAP2 changing amino acid composition of the protein (E163K). Here, the reverse strand is displayed showing a C>T transition.

To investigate the DNA methylation status of the CpG island within the MAP2 promotor (Fig 5A), bisulfite-treated DNA probes of five MCL cell lines and 20 MCL cases were subjected to MSP. In 18 of the 20 MCL cases the pattern of PCR products indicated partial DNA hypermethylation, including the MCL 16 with the MAP2 mutation. The DNAs of the MCL cell lines JEKO-1 and HBL-2 were also partially hypermethylated and the MSP pattern of the REC-1 cell line indicated virtually complete DNA methylation of the MAP2 CpG island (Fig 5B). Noteably, REC-1 had only one MAP2 allele due to a heterozygous deletion of part of chromosome 2. The MCL cell line GRANTA-519 showed only a weak PCR product of methylated DNA and the additional two MCL cases and the MCL cell line MAVER-1 seemed to be completely unmethylated. In contrast to the distribution in MCL, eight non-tumours controls (four tonsils and four PB) showed a MSP pattern indicating an unmethylated status of MAP2 promotor. These findings could be confirmed by bisulfite sequencing (Fig 5C and D, Fig S3).

Figure 5.

 Epigenetic and expression studies of MAP2. (A) Genome Browser extract (NCBI Build 35) displaying location of MAP2 CPG island. (B) Methylation specific PCR (MSP) of REC-1, two primary MCL and controls. Methylation specific amplification was verified using universal methylated (MC) and universal unmethylated (UC) DNA as controls. Genomic DNA (gC) was tested to exclude unspecific amplification. (C) Part of the analysed MAP2 CpG island sequences investigating bisulfite-treated DNA of REC-1 and a tonsil. The included CpG dinucleotides 5–11 are marked by black bars. (D) Bisulfite genomic sequencing of the MAP2 CpG island in the MCL cell line REC-1 and two primary MCL. Tonsil and peripheral blood (PB) samples from healthy donors were used as controls. Each column represents a CpG dinucleotide and each row represents one of 10 sequenced clones. Black and white spots indicate methylated and unmethylated CpGs, respectively. Grey spots indicate CpG dinucleotides that failed to be analysed. (E) Real-time Reverse Transcription (RT-) PCR. Two primer assays were applied to investigate all known MAP2 transcripts. Primer position is indicated by black arrows. Expression of MAP2 transcripts was examined in six MCL cell lines and compared to MAP2 expression in lymph node tissue of healthy individuals. Each value has been normalised to the average expression of three housekeeping genes. In UPN-1 and REC-1 no MAP2 expression was detected, thus log2 ratio was not calculable (N/C).

To investigate how DNA methylation status of the MAP2 CpG island correlates with MAP2 gene expression in MCL cell lines we performed real-time RT-PCR. Two different primer assays were used to detect all known MAP2 transcript variants (Fig 5E). In UPN-1 and REC-1, no expression of MAP2 transcripts could be detected. Real-time RT-PCR results of HBL-2 indicated reduced expression of all MAP2 transcripts compared to control LN tissue. In GRANTA-519 and MAVER-1 the expression of MAP2 transcripts was similar or minimally reduced whereas JEKO-1 showed a considerable higher expression of all MAP2 transcripts compared to the LN tissue (Fig 5E). Similar results were obtained using a freshly prepared tonsil as reference (data not shown).

TP53 gene.  Due to its influence on microtubule dynamics, TP53 was subjected to mutation analyses in all samples of the MCL GeneChip panel. One case showing a homozygous TP53 mutation had already been analysed as part of the pUPD investigation. In six of the additional 25 primary cases mutations within the coding sequence of TP53 were detected using DHPLC followed by sequencing aberrant fragments. Detected mutations include six different point mutations and one microdeletion of one single base pair (Table III). In all these cases copy number and LOH analyses of the 100K GeneChip data indicated genomic loss of the second allele (Table SVI). Three point mutations and one microdeletion affecting the coding sequence of TP53 were detected in MCL cell lines UPN-1, HBL-2, MAVER-1 and JEKO-1 (Table III). In all these cell lines the second TP53 allele was deleted. Both REC-1 and GRANTA-519 exhibited the wild type allele but the latter showed a heterozygous deletion of the TP53 locus. Some of the MCL cell line data has been already reported elsewhere (Amin et al, 2003; M’Kacher et al, 2003; Camps et al, 2006; Zamo et al, 2006).

Table III. TP53 mutation analyses in MCL.
MCL primary casesTP53 mutationSecond abberation in 17p13.1
MCL 4g.12178 delC - p.122XDeletion
MCL 6g.14487 G>T - p.R273LDeletion
MCL 8g.14490 T>A - p.V274DpUPD
MCL 15g.13203 G>A - p.R175HDeletion
MCL 17g.14070 C>T - p.R248LDeletion
MCL 19g.12108 G>T - p.E62XDeletion
MCL 21g.13091 G>C - p.A138PDeletion
MCL cell linesTP53 mutationSecond abberation in 17p13.1
  1. g., genomic sequence position; p., protein sequence position (http://www-p53.iarc.fr).

UPN-1g.14525 G>A - p.E286KDeletion
MAVER-1g.14511 A>G - p.D281GDeletion
JEKO-1g.12096 delC - p.58XDeletion
HBL-2g.14511 A>G - p.D281GDeletion

Discussion

100K GeneChip microarray data

In the present study, we performed 100K GeneChip microarray analyses of 26 primary MCL and six MCL cell lines. The use of short synthetic oligonucleotides as arrayed elements enabled the detection of genomic imbalances with high resolution (approximately 24 kb) and genotyping was performed simultaneously to identify regions of pUPD. Overall, the identified genomic imbalance pattern was consistent with those previously described by cytogenetics or arrayCGH studies in MCL (Kohlhammer et al, 2004; de Leeuw et al, 2004; Rubio-Moscardo et al, 2005b; Tagawa et al, 2005; Mestre-Escorihuela et al, 2007; Pinyol et al, 2007).

Using the GeneChip mapping technique, we not only delineated previously reported alterations but also identified novel regions of recurrent genomic imbalance in MCL. Taking advantage of the high resolution of the used SNP arrays, we focused on detecting high-level amplifications and small regions of homozygous loss. In this way, novel regions harbouring potential oncogenes or candidate TSG were identified that might be involved in tumourigenesis. Moreover, we introduced two useful bioinformatic approaches for identifying homozygous losses in 100K GeneChip data and point out advantages and disadvantages.

In line with a recent study of five MCL cell lines (Nielaender et al, 2006), LOH analysis of primary MCL tumor tissue demonstrated that pUPD is a recurrent genetic mechanism in MCL tumourigenesis. Genomic distribution of detected pUPD in the analysed primary MCL showed that recurrently affected regions, such as 11q and 13q, are commonly targeted by deletions in MCL. Furthermore, we explicitly demonstrated TSG inactivation by pUPD targeting the TP53 locus in 17p13.1. A homozygous missense mutation affecting a DNA binding domain was detected. This mutation is frequent in lymphomas (28%) and has been reported to be deleterious for TP53-DNA interaction (http://www-p53.iarc.fr). TP53 inactivation by chromosomal deletion is a common chromosomal event in MCL and is associated with poor prognosis (Rubio-Moscardo et al, 2005b). The present study showed that pUPD is an alternative mechanism to chromosomal deletion leading to homozygosity of a TSG inactivating mutation. Thus, pUPD seems to be a critical genetic event in MCL pathogenesis.

Genes encoding microtubule-associated proteins as targets of chromosomal aberrations

Interestingly, analysis of the 100K GeneChip data identified different genes encoding microtubule-associated proteins (MAPs) to be involved in chromosomal alterations in MCL. MAPs are cellular proteins that are associated with microtubules and alter their dynamics. Microtubule dynamic property is crucial for the assembly of the mitotic spindle and the attachment and movement of chromosomes along the spindle (Zhai et al, 1996). Microtubule-targeting drugs suppressing microtubule dynamics are widely used as cancer chemotherapeutic agents (Jordan & Wilson, 2004). In addition to their direct involvement in the physical process of mitosis, microtubules also serve as scaffolds for signalling molecules (Mollinedo & Gajate, 2003). The family of MAPs includes products of oncogenes, tumour suppressors and apoptosis regulators, suggesting that alteration of microtubule dynamics and changes in the scaffolding properties of microtubules may be critical events in tumourigenesis and tumour progression (Bhat & Setaluri, 2007). Until now, alterations in microtubule organisation have not been reported in MCL (Jares et al, 2007).

In this study, a homozygous deletion of the MAP2 locus was identified in the MCL cell line UPN-1. Real time RT-PCR revealed absence of MAP2 expression in UPN-1 and REC-1. REC-1 harbours a heterozygous deletion of part of the long arm of chromosome 2 and epigenetic studies showed complete DNA methylation of the CpG island of the remaining MAP2 allele. Moreover, the DNAs of the MCL cell lines JEKO-1 and HBL-2 were also partially hypermethylated. MSP analysis demonstrated partial hypermethylation in 90% of 20 investigated primary MCL. In one of these cases showing partial hypermethylation, a point mutation affecting the coding sequence of MAP2 was identified. These findings suggest that DNA hypermethylation is a frequent mechanism leading to MAP2 gene inactivation in MCL. MAP2 protein participates in the stabilisation of microtubules and is predominantly expressed in neurons, where it is essential for the regulation of organelle transport within axon and dendrites (Sanchez et al, 2000). Moreover, MAP2 was the first protein shown to copurify and interact directly with the regulatory subunit of the protein kinase A (PKA), also known as cAMP-dependent protein kinase (cAPK) (Vallee et al, 1981; Theurkauf & Vallee, 1982). Among other cellular effects, PKA-catalysed phosphorylation modulates cell growth, cell division and actin cytoskeleton rearrangements. MAP2 protein operates as an A-kinase anchoring protein (AKAP) and targets the PKA to microtubules. Attachment to microtubules occurs through its tubulin-binding domain (Serrano et al, 1984; Hirokawa, 1994). MAP2 also harbours a conserved binding site for phosphatase PP2A, although direct binding of the phosphatase has yet to be reported. PP2A represents a family of heterotrimeric serine/threonine phosphatases implicated in the regulation of a plethora of cellular processes such as apoptosis, transcription, translation, DNA replication, signal transduction, protection against tumourigenesis and cell division (Janssens et al, 2005). Soltani et al (2005) reported that MAP2 expression is associated with prognosis in melanoma. A five-year clinical follow-up study showed longer disease-free survival of patients whose primary tumors express abundant MAP2 as compared with patients with weak or no MAP2 expression. Moreover, exogenous expression by adenovirus leads to cell cycle arrest, growth inhibition and apoptosis in metastatic melanoma cells (Fang et al, 2001; Soltani et al, 2005). Thus, lack of MAP2 expression might be also associated with MCL pathogenesis.

The p53 protein is also associated with microtubules in vitro and in vivo (Giannakakou et al, 2000) and has been reported to regulate other microtubule-associated proteins (Murphy et al, 1999; Johnsen et al, 2000; Mirza et al, 2002). TP53 inactivation by chromosomal deletions or by mutations is a common genetic alteration in MCL. Galmarini et al (2003) demonstrated that microtubule protein composition was altered in TP53 mutants (mut-p53) and dynamic instability of microtubules was significantly increased. Mutation analyses in this report identified seven primary MCL cases to harbour homozygous mutations in the coding sequence of TP53. In all these cases, the second allele got lost by chromosomal deletion or pUPD. Similarly, TP53 was shown to be homozygously mutated in four of six investigated MCL cell lines (Amin et al, 2003; M’Kacher et al, 2003; Camps et al, 2006; Zamo et al, 2006). Galmarini et al (2003) also reported that the MAP6 (also called STOP) protein and its corresponding mRNA-expression were increased in the mut-p53 cells, than in the wt-p53 cells suggesting negative transcriptional regulation of MAP6 by p53 protein. Interestingly, our MCL GeneChip data showed high-level amplification of MAP6 in the three MCL cell lines MAVER-1, JEKO-1 and HBL-2. Real-time RT-PCR detected higher expression of MAP6 in JEKO-1, GRANTA-519 and REC-1 compared with non-tumours tonsil tissue. The apparent lack of correlation between MAP6 dosage and expression in MAVER-1 and HBL-2 might be caused by putative non-neuronal transcripts variants that cannot be detected with the primers used. According to this, the murine homologue shows non-neuronal transcript variants, which lacks whole exons (Bosc et al, 2003). So far, no human non-neuronal transcript variant has been identified. High-level amplification of MAP6 was frequently detected in MCL cell lines but in none of the investigated primary MCL. As cell lines are frequently derived from cases with advanced disease, this finding might indicate that MAP6 amplification is associated with disease progression. In line with this hypothesis, a chromosomal rearrangement of another MAP gene, i.e. MAP4, was recently identified as secondary alteration in a large B-cell lymphoma (DLBCL), which was present at relapse but not at initial diagnosis (Murga Penas et al, 2006). According to our GeneChip data, it is widely assumed that cell lines generally harbour an increased number of chromosomal aberrations compared to the primary tumour cells.

Our study provides evidence that alterations of microtubule dynamics might be critical in MCL tumourigenesis and tumour progression. Nineteen of the 26 primary MCL from the SNP array panel and five of the six MCL cell lines harboured a genetic or epigenetic defect in at least one of the three microtubule-associated genes MAP2, MAP6 and TP53. Six primary cases were not analysed by methylation-specific methods. Only one case and one MCL cell line, in which all three gene loci were analysed, did not show any of the investigated alterations.

The complexity of the mitotic spindle requires fine-tuning of the dynamics of all microtubules for proper function. MAPs can either stabilise or destabilise microtubules. Changes in levels of expression have been reported to correlate with aggressiveness of cancer cells or their sensitivity to microtubule-targeting agents. Although plenty of studies exist regarding microtubule-associated proteins in neurons, where they play a critical role in neurite outgrowth and dendrite development, their mechanisms of operation in mitotic spindle regulation are rather unclear. Mitotic spindle organisation is a fine-tuning process and investigation of the involved proteins is difficult due to low dosage. In contrast to abundant gene expression in brain tissue, expression of MAP2 and MAP6 is hardly detectable in other kinds of tissues. In our study, genomic alterations, such as MAP6 amplification or MAP2 and TP53 inactivation, provide evidence for the involvement of microtubule-associated genes in MCL tumourigenesis. Alterations in microtubule dynamics and mitotic spindle organisation might contribute to the karyotype complexity and chromosomal instability that are characteristic features of MCL. Supporting this hypothesis, a recent study underlined the high expression level of centrosome-associated gene products in blastoid MCL (Neben et al, 2007). MCL has one of the worst prognoses amongst all lymphomas. There is no therapy that can be considered as standard. Resistance against microtubule-targeting chemotherapeutic agents may be the consequence of changes in microtubule dynamics.

In conclusion, our study demonstrated that 100K GeneChip microarray analyses is a useful strategy to analyse MCL genomes with regard to genomic imbalances, particularly homozygous deletions, as well as pUPD. Moreover, we identified novel candidate TSG and oncogene loci that might harbour genes involved in MCL pathogenesis. Interestingly, different genes encoding microtubule-associated proteins could be identified as targets of chromosomal aberrations in MCL. Our findings suggest that alteration of microtubule dynamics is a critical genetic event in MCL.

Acknowledgements

This study was supported by the Lymphoma Research Foundation (New York) and the EU (LSHC-CT 2004-503351) in the framework of the ‘European MCL Network’.

Ancillary