Detection of hemizygous deletions in genomic DNA from leukaemia specimens for the diagnosis of patients

Leukemia Research 29 (2005) 165–171

Detection of hemizygous deletions in genomic DNA from leukaemia specimens for the diagnosis of patients Ursula R. Keesa,∗ , Philippa A. Terrya , Jette Forda , Janet Everettb , Ashleigh Murchb , Nick de Klerka , David L. Bakerc a

c

Telethon Institute for Child Health Research, and Centre for Child Health Research, The University of Western Australia, Perth, Australia b Pathology, Women’s and Children’s Health Service, Perth, Australia Department of Haematology–Oncology, Princess Margaret Hospital, Perth, Australia Received 5 February 2004; accepted 14 May 2004 Available online 11 September 2004

Abstract Hemizygous deletions in genomic DNA appear to play an important role in tumorigenesis. The loss or inactivation of tumour suppressor genes (TSGs) is of critical importance in most malignancies, and has been shown to affect response to therapy. Here, we report a quantitative real-time polymerase chain reaction (qPCR) designed to detect two TSGs at the CDKN2A locus, p16INK4A and p14ARF that allows the detection of hemizygous deletions. Testing by qPCR of 18 bone marrow specimens from paediatric acute lymphoblastic leukaemia (ALL) patients at diagnosis revealed nine to be GG, six to be GD and three to be DD for exon 2 of p14ARF /p16INK4A , concordant with Southern blotting analysis. A panel of 13 ALL cell lines was investigated for deletions at the CDKN2A locus and one of the lines, typed as GD for all exons, was further assessed by fluorescence in situ hybridisation, confirming the qPCR findings. The expression levels of p16INK4A and p14ARF were measured in all cell lines and these quantitative reverse transcriptase PCR results also agreed with the typing by qPCR. The qPCR method described is suitable for detection of hemizygous loss in primary patient material and the accuracy of the method was verified by three independent techniques. © 2004 Elsevier Ltd. All rights reserved. Keywords: Tumour suppressor genes; Detection of gene deletion; Hemizygous; Leukaemia specimens; Polymerase chain reaction

1. Introduction Inactivation of tumour suppressor genes (TSGs) can occur by gene mutation, deletion, rearrangement or epigenetic mechanisms. Knudson’s two hit model [1] states that TSGs are recessive and that both alleles must be inactivated for tumorigenesis. It appears that this may not apply to all TSGs. Several recently reported animal studies, reviewed by [2], demonstrated that loss of one allele could lead to predisposition for cancer. Haplo-insufficient TSGs include p27Kip1 , Dmp1, Ptch and others [3]. Moreover, a recently published

∗

Corresponding author. Tel.: +61 8 9489 7852; fax: +61 8 9489 7700. E-mail address: [email protected] (U.R. Kees).

0145-2126/$ – see front matter © 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.leukres.2004.05.021

study reported evidence for enhanced tumour formation in mice heterozygous for a Blm mutation (a Bloom syndromecausing mutant allele) [4]. Human carriers of a BLM mutation were similarly shown to have an increased risk for colorectal cancer [5]. These findings have important implications for cancer risk in humans. We have shown that deletion of one copy of the p16INK4A gene in paediatric leukaemia is associated with an increased risk of relapse [6]. In cases where gene deletion is the mechanism of inactivation of TSGs, a method frequently used to detect gene deletion is Southern blotting. Since primary patient specimens invariably contain a certain proportion of non-malignant cells, the sensitivity and reproducibility of this technique is not optimal for the detection of one allele. Here, we report that a quantitative real-time polymerase chain

166

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171

reaction (qPCR) is suitable for the detection of one gene copy in primary patient specimens.

2. Materials and methods 2.1. Patient samples and cell lines Bone marrow specimens were obtained from paediatric leukaemia patients diagnosed and treated at the Princess Margaret Hospital for Children, Perth, Australia. Informed consent was obtained from parents, patients or both, as deemed appropriate. The cell lines studied (Table 1) were generated according to our previously published method [7] and the characterisation of the cell lines has been reported [8]. Two cell lines of T-cell phenotype (T-ALL), PER-427 and PER487 required the addition of 300 U/ml IL-2 to the growth medium. Immunophenotyping was performed by indirect immunofluorescence and flow cytometry, using a panel of monoclonal antibodies [7]. Each line was found to display the immunophenotypic features of the primary leukaemia cells of the respective patients. The RajiB, K562 and HeLa cell lines were used as controls. 2.2. Real-time genomic PCR analysis in multiplex format Genomic DNA was isolated from cryopreserved patient specimens and cell lines by standard methodology [9]. All primers and probes for the quantitative real-time PCR (qPCR) were designed using ABI PRISM Primer Express v1.5 software (PE Applied Biosystems). The primer sequences were as follows: p14 (E1␤)-forward (F) 5 -ccctcgtgctgatgctactg-

Table 1 Status of INK4A/ARF locus in childhood ALL cell lines Cell line

Exons∗

Genes∗

1␤

1␣

2

p14/ARF

p16

T-ALL PER-117 PER-255 PER-427 PER-487 PER-537 PER-550

DD DD GG DD GD GD

DD DD GG DD DD DD

DD DD GG DD DD DD

DD DD GG DD DD DD

DD DD GG DD DD DD

pre-B ALL PER-145 PER-278 PER-371

DD GD GD

DD GD GD

DD GD GG

DD GD GD

DD GD GD

B-ALL PER-377 PER-495

GD GG

GD GG

GG GG

GD GG

GD GG

Infant leukemia PER-485 DD PER-490 GG

DD GG

DD GG

DD GG

DD GG

∗

G, wild type; D, deleted.

3 , E1␤-reverse (R) 5 -gggcctttcctacctggtct-3 , p16 (E1␣)-F 5 -ccaacgcaccgaatagttacg-3 , E1␣-R 5 -gctacctgattccaattcccct-3 , p14/p16(E2)-F 5 -ggctctacacaagcttcctttcc-3 , E2-R 5 -tcatgacctgccagagagaaca-3 , ␤-actin(BA)-F 5 -agcgcggctacagcttca-3 and BA-R 5 -cgtagcacagcttctccttaatgtc-3 . The probe sequences were as follows: E1␤-probe (Pr) 5 -tctagggcagcagccgcttcctaga-3 , E1␣-Pr 5 -ccacctggatcggcctccga3 , E2-Pr 5 -cccccaccctggctctgacca-3 and BA-Pr 5 -atttcccgctcggccgtggt-3 . All primers were synthesised by Invitrogen Life Technologies (Sydney, Australia) and all probes by Applied Biosystems (Foster City, USA). For these experiments, the probes for E1␤, E1␣ and E2 were labelled with the FAM reporter and the endogenous control probe for βactin was labelled with VIC. All probes contained TAMRA as the quencher. The optimal concentrations for the primers and probes were determined to give the most efficient reaction, defined by the lowest value for cycle threshold (CT ) combined with the highest Rn (difference between the ratios of the reporter signal to passive reference dye in a positive and negative reaction). The conditions were then optimised for the exons of interest in a multiplex reaction with β-actin. In the case of E2, the reactions were carried out in a final volume of 50 ␮L. The final concentrations of the primers and probes were as follows: E2-F 130 nM, E2-R 140 nM, E2-Pr 200 nM, BA-F 168 nM, BA-R 484 nM, BA-Pr 200 nM and 50 ng of genomic DNA template was used in each reaction, as previously reported [6]. The multiplex reactions for E1␤ and E1␣ were further optimised to be performed in a 25 ␮L reaction. The primer and probe concentrations used for E1␤ were E1␤-F 300 nM, E1␤-R 500 nM, E1␤-Pr 125 nM, BAF 50 nM, BA-R 50 nM and BA-Pr 200 nM. The primer and probe concentrations used for E1␣ were E1␣-F 300 nM, E1␣R 100 nM, E1␣-Pr 125 nM, BA-F 100 nM, BA-R 50 nM and BA-Pr 200 nM. Both the E1␤ and the E1␣ reactions were performed using 25 ng of genomic DNA template. All reactions were set up using Taqman Universal Master Mix (Applied Biosystems) and run under standard real time PCR conditions on an ABI PRISM 7700 Sequence Detector. The thermal cycling conditions were Step 1: 2 min at 50 ◦ C, Step 2: 10 min at 95 ◦ C, Step 3: 15 s at 95 ◦ C, Step 4: 1 min at 60 ◦ C, repeating Steps 3 and 4 for a total of 40 cycles. A standard calibration curve was included in each experiment and they showed correlation coefficients of ≥0.97. The template for this was RajiB genomic DNA, in serial two-fold dilutions from 100 to 0.75 ng. The content of normal cells was determined morphologically as part of the routine diagnostic assessment of bone marrow specimens. After analysing the real time data, the ratios obtained were then adjusted to take into consideration the level of normal cells present in the marrow, as follows: (measured value − estimated percentage normal cells)/(1 − estimated percentage normal cells), to give final score. The deletion status of genes p14ARF and p16INK4A was determined based on findings for individual exons. In cases of GD status for E1 and E2, we made the assumption that the deletions had occurred on the same allele.

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171

2.3. Real-time RT-PCR RNA was extracted from the cell lines using Trizol reagent (Invitrogen Life Technologies) according to the manufacturer’s protocol. The RNA was further purified on RNeasy mini columns (Qiagen) followed by sodium acetate/ethanol precipitation. Synthesis of cDNA was performed using Omniscript RT (Qiagen). The reaction volume was diluted to 40 ␮l. cDNA was then analysed by quantitative real time reverse transcriptase PCR (qRT-PCR) to determine the level of expression of p14ARF and p16INK4A in the cell lines. β-Actin was used as the endogenous control. The q-RT-PCR primers and probes were designed using Primer Express V1.5 (PE Applied Biosystems). In all cases, the probe was positioned to span exon–exon junctions. The primer sequences were as follows: p14cDNA-F 5 -gctactgaggagccagcgtcta-3 p14cDNA-R 5 -gggcgctgcccatca-3 , p16cDNA-F 5 ccaacgcaccgaatagttacg-3 , p16cDNA-R 5 -gggcgctgcccatca3 , ␤-actin cDNA-F 5 -ggcacccagcacaatgaag-3 , ␤-actin cDNA-R 5 -gccgatccacacggagtact-3. All primers were synthesised by Invitrogen Life Technologies (Sydney, Australia). The probe sequences were: p14 cDNAPr 5 -agccgattcctagaagaccaggtcatg-3 , p16 cDNA-Pr 5 -catgacctggatcggcctccga-3 and ␤-actin cDNA-Pr 5 tcaagatcattgctcctcctgagcgc-3 . Both probes for the target genes were labelled with the FAM reporter and the β-actin probe was labelled with the VIC reporter. All probes contained TAMRA as the quencher. Probes were synthesised by Applied Biosystems (Foster City, USA). The reactions for p14ARF and p16INK4A were optimised, as described above, to give the best CT and Rn in a single reporter reaction and then in a multiplex reaction with β-actin. The primer and probe concentrations used for analysis of p14ARF expression were p14cDNA-F 100 nM, p14cDNA-R 900 nM, p14cDNA-Pr 200 nM, ␤-actin cDNA-F 100 nM, ␤-actin cDNA-R 50nM and ␤-actin cDNA-Pr 100 nM. The primer and probe concentrations used for analysis of p16INK4A expression were p16cDNA-F 100 nM, p16cDNAR 900 nM, p16cDNA-Pr 150 nM, ␤-actin cDNA-F 50 nM, ␤-actin cDNA-R 25 nM and ␤-actin cDNA-Pr 100 nM. All qRT-PCR reactions were carried out in 50 ␮L volumes with 2 ␮L of cDNA template. The thermocycling conditions were the same as those described for qPCR above. Calibration curves were included in all tests and they showed correlation coefficients of ≥0.97. The cDNA generated from HeLa cells was used for the calibration curves for p14ARF and p16INK4A . 2.4. Southern blot analysis The presence of p16INK4A E2 was determined by Southern blotting using probes for p16E2 and for IL-2RB as control, as previously reported [9]. Genomic DNA was digested with EcoRI, separated on a 1.2% agarose gel and transferred onto Hybond-N+ (Amersham Pharmacia Biotech). The probes were labelled with ␣-32 P-dCTP using the Rediprime II la-

167

belling kit (Amersham Pharmacia Biotech). Hybridisation was performed overnight at 65 ◦ C and signal was detected on an FLA-3000 using Image Reader v1.8E software (Fuji Photo Film Co. Ltd., Japan). The ratio of signals for p16E2 to IL-2RB was determined from the scanned image using Image Gauge v3.4 software (Fuji Photo Film Co. Ltd., Japan). 2.5. Cytogenetic analysis The cytogenetic analysis was performed according to standard protocols [10]. Fluorescence in situ hybridisation (FISH) for p16INK4A was carried out using the Vysis LSI p16/DEP 9 Dual Color Probe (Vysis Inc., Downers Grove, IL) according to the directions supplied by the manufacturer. 2.6. Assessment of optimal cut-off ratios The qPCR method was previously shown to be suitable for detection of homozygous gene loss (DD), hemizygous loss (GD) and wild type status (GG) [6]. Our original method used cut-off ratios that were based on the triphasic distribution of specimens, 0.4 (between DD and GD) and 0.8 (between GD and GG) [6]. We re-examined these ratios using a statistical simulation method and a larger number of specimens. The method consisted of randomly sampling from all feasible distributions of the genotype frequency and it took the documented clonal heterogeneity of lymphoid tumours into account, using estimates based on receptor rearrangements in lymphoid leukaemias, and the independently measured level of blast cells, hence the presence of normal cells. This process was repeated 10,000 times to give 10,000 estimated values. A discriminant function analysis was then carried out on these simulated values so as to choose the cut-off ratios that gave minimum incorrect allocation, which was about 2%. Full details of the statistical method applied are available from the authors on request. The cut-off ratios were determined to be 0.26 (between DD and GD) and 0.71 (between GD and GG). The specimens tested in our previous studies were reanalysed using these ratios and the classification according to p16INK4A status did not change.

3. Results In order to determine whether qPCR can detect deletion of p16INK4A , we initially examined mixed cell populations using DNA from two cell lines, RajiB cells which are wild type for p16INK4A (GG) and K562 cells which show homozygous deletion of p16INK4A (DD) [6]. Mixtures ranging from 5 to 100% of Raji B cells relative to concentrations of K562 cells were made. The ratio of p16INK4A E2/β-actin for duplicate specimens was determined experimentally in a multiplex qPCR (for details see Section 2) and these values were expressed as a function of the input ratios of RajiB/K562 cells. Reproducibly, this test yielded a linear graph with a correlation coefficient of 0.9687–0.9742, indicating clearly that the

168

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171

gene status can be accurately measured by this technique and over the range from 0 to 100%, simulating DD to GG. In order to use the method for patient specimens that invariably contain normal cells, the experimentally determined ratios were adjusted for the independently measured percentage of normal cells. Optimal cut-off ratios between DD/GD/GG were determined by statistical simulation models (see Section 2). Based on these studies, the ratios determined for bone marrow specimens were set as follows. Ratio for test gene/β-actin <0.26: gene deletion (DD); ratio 0.26–0.71: one gene copy, hemizygous (GD); ratio >0.71: two gene copies, germline (GG). Eighteen patient specimens were tested for presence of p16E2 by qPCR and nine were found to be GG, six were GD and three were DD. In order to determine whether an independent method would produce the same results, all specimens were analysed by the Southern blotting analysis, using a probe for p14/p16E2 and IL-2RB [9] as the control hybridisation. Fig. 1 shows the results for five patient specimens determined to be DD by qPCR (two cases), GD (two cases) and GG (one case). The data indicated concordance between the two methods. The ratios for p16E2/IL-2RB obtained from Southern blot analysis were then determined for all 18 patient specimens (nine typed as GG, six as GD and three as DD), see Fig. 2, and this provided further confirmation for the qPCR method, since specimens typed as GG by qPCR yielded signals by Southern blot analysis of 1.17 ± 0.38 and for GD specimens of 0.33 ± 0.14, while the three DD specimens had no discernible signal for the test gene. When specimens were divided into genotype groups (DD, GD, GG), there was perfect agreement between qPCR and Southern blot data (κ = 1.0, lower confidence limit 0.62). When measured on a continuous scale, the Spearman correlation coefficient was 0.84, indicating good agreement. To further verify the data from the qPCR technique by independent methods, we employed fluorescence in situ hybridisation (FISH) and qRT-PCR. We intended to examine the patient specimens by these additional techniques, however, due to lack of sufficient material, this was not feasible. Hence, we made use of a panel of ALL cell lines

Fig. 1. Southern blot analysis of p16E2 deletion in leukaemia specimens from five patients. DNA samples were digested with EcoRI and hybridised with a probe for p16E2 (Panel A) revealing a band of 4.3 kb. The same blot was hybridized with a probe for the IL-2RB gene (Panel B), demonstrating the presence of DNA in all lanes. The ratio for the signal intensity for p16E2/IL2RB gene and the typing of specimens by qPCR are indicated at the bottom of each lane.

Fig. 2. Gene status of p16E2 determined for each of 18 patient specimens, using Southern blot and qPCR methods. The qPCR method typed 3 as DD, 6 as GD and 9 as GG. All specimens were analysed to determine gene status by the independent Southern blot analysis (as shown in Fig. 1) yielding ratios of signal intensities of 1.17 ± 0.38 for GG specimens, 0.33 ± 0.14 for GD specimens and no detectable signal for DD specimens.

(Table 1). The panel comprised six T-ALL lines representing various stages of differentiation and three lines of pre-B phenotype. Based on their expression of IgM on the cell surface, two lines were classified as mature B-cell ALL. Two lines were established from infants and both expressed Blineage markers. The status of the CDKN2A locus was determined by qPCR and it comprised the assessment of three exons, E1␣ (p16INK4A gene), E1␤ (p14ARF gene) and E2 (p16INK4A and p14ARF ) and the results are summarised in Table 1. Five of six T-ALL cell lines scored as DD across the locus, while only two of seven B-lineage ALL cells were found to be DD. Three of the lines were typed as GD and one of them, PER-278, was found to be GD for all exons. This cell line is known to contain an aberration of chromosome 9 and we previously reported the karyotype as 46,XY,der(9)t(1;9)(q23;p13),der(19)t(1;19)(q23;p13) [11]. The typing of PER-278 cells as GD provided the opportunity to apply FISH as another independent technique to verify the gene deletion findings based on qPCR. We made use of the LSIp16 probe which spans the CDKN2A locus, comprising the p14ARF and p16INK4A genes, in conjunction with the reference CEP9 probe which hybridises to alpha satellite sequences specific to chromosome 9. The FISH study on 20 metaphases showed deletion of the CDKN2A sequence on the derivative chromosome 9 in all of them (Fig. 3), and this finding was further confirmed by results from 100 interphase nuclei examined, all of which showed a single signal for CDKN2A. These results demonstrated that a second independent technique confirmed the typing by the qPCR method. In order to confirm the presence or the absence of the genes at the functional level, we examined the expression of p14ARF and p16INK4A in the panel of cell lines. A qRTPCR method was established to measure expression of the two genes in a multiplex reaction with β-actin as reference gene. The expression level of β-actin was found to be similar in all cell lines under test. The data summarised in Fig. 4 clearly showed that gene expression was present in cell lines

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171

169

Fig. 3. FISH analysis confirming deletion of one p16INK4A allele in ALL cell line PER-278. Metaphase spreads showing two signals for the CEP9 probe (green) which hybridises to alpha satellite sequences specific to chromosome 9, and the absence of signal from the LSIp16 probe (orange) on the derivative chromosome 9, a probe that hybridizes to the CDKN2A locus.

typed as GG or GD and absent in DD cell lines. The three GG cell lines expressed p14ARF ranging from 0.59 to 3.43, while p16INK4A expression levels ranged from 0.12 to 0.29, in both instances determined relative to β-actin. The expression levels observed for GD cell lines were in the same range as for GG lines (Fig. 4). Thus, in all 13 cases the expected presence or absence of mRNA from the two genes was confirmed by qRT-PCR. These findings provided functional validation for the findings from qPCR.

4. Discussion Recent insights gained from molecular genetic studies show that diagnostic assessment of human cancers requires methods that accurately detect gene deletions, particularly TSGs. Gene expression profiles allow the molecular classification of cancers as evidenced by the seminal work on lymphoid tumours [12,13]. Recent studies have demonstrated that DNA microarrays provide a strategy for possible discovery of new subtypes of cancer, independent of previous knowledge of the biology of the cancers. Certain genetic lesions occur at high frequencies in many cancers, regardless of patient age or tumour type. A well documented example is the aberration of the RB1 tumour suppressor pathway which includes mutations of the p16INK4A , cyclin D, cyclin-dependant kinase 4 (CDK4) and RB1 genes. The p16INK4A TSG can be inactivated by deletions, mutations or epigenetic mechanisms, depending on tumour type [14] and in ALL the dominant mechanism is deletion [9,15]. Moreover, evidence is accumulating that the presence or the absence of particular TSGs in cancer

cells is correlated with the response to chemotherapy, as reported by Schmitt and colleagues who studied animals with primary lymphomas [16]. They showed that the in vivo response to chemotherapy was dependent on mutations of p53 and genes at the CDKN2A locus. Taken together, these findings make it mandatory to establish suitable techniques for the reliable detection of mutations in TSGs in human cancers. The qPCR method designed here for the detection of deletion of two TSGs at the CDKN2A locus was verified by three independent techniques—Southern blotting, FISH and qRTPCR. Eighteen primary patient specimens were analysed by Southern blotting and in all cases the results agreed with assessment by qPCR. Metaphase and interphase FISH studies on a leukaemia cell line unequivocally confirmed the findings by qPCR. The incidence of homozygous deletion in T-ALLs was five out of six, in contrast to two out of seven in B-lineage ALL, reflecting data from primary patients [9,15,17]. Moreover, RNA studies were conducted on a panel of cell lines and both the genes, p16INK4A and p14ARF , were found to be expressed in all cases typed as GG and GD, while absent in DD lines. The transcriptional control of these genes is complex and a growing number of proteins are known to regulate their expression, reviewed in [18]. The cell lines under test are genetically very heterogenous and this appears to be the most likely explanation for similar expression levels in GG and GD lines. We are currently analysing primary patient specimens to further examine the finding. Bertin and et al. [17] recently reported a real-time PCR assay for genes at region 9p21, modelled on our original method [6]. Importantly, their method uses SYBR Green detection which does not achieve the degree of specificity attained by gene-specific probes, as

170

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171

be applied, e.g. a PCR or flow cytometry method to detect specific markers on the malignant or the normal cell population. For specimens from solid tumours higher accuracy may be achieved by selecting malignant cells by laser capture microdissection for subsequent assessment by qPCR. We believe that the most appropriate way of assessing deletion of p16INK4A in the clinical diagnosis of cancer is to use techniques capable of routinely detecting loss of only one allele of the gene, such as the qPCR described here. Such an approach will be particularly valuable for TSGs shown to be semidominant or haploinsufficient [2,20]. Furthermore, it is applicable to the increasingly recognised group of cancers where deletion of a TSG is of prognostic significance, e.g. PTEN deletion in prostate and brain cancer [21,22]. Our recent study on paediatric leukaemia patients suggested that hemizygous p16INK4A deletion may confer susceptibility to leukaemia [23]. As molecular profiling of subtypes gains increased usage in clinical diagnosis, such methods as described here will find broad application, exemplified by a recent study on patients with mantle cell lymphoma which identified subsets that showed distinct gene expression signatures. Importantly, differences in cyclin D1 mRNA abundance synergised with gene deletion at the CDKN2A locus [24]. The accurate identification of hemizygous gene deletion in cancer patients will have profound impact on prognosis, risk stratification and clinical therapy, as well as diagnosis in future cancer trials. Fig. 4. Expression of p14ARF and p16INK4A in childhood ALL cell lines. Expression for both genes was determined by qRT-PCR and expressed as ratio of test gene: β-actin. Values are means and S.D. of four independent measurements. Annotated is the designation and genotype of each cell line as determined by qPCR.

is the case in our method. Bertin et al. were unable to assess gene dosage in 13 cases, whereas our study overcame this drawback with appropriate statistical analysis. The knowledge of a particular genetic lesion in the cancer cells of a patient is particularly useful if therapy tailored to such cancer cells is available. Efforts to design novel anticancer agents specifically targeting the cell cycle are being made and a range of small molecular inhibitors are under investigation. Flavopiridol is one such agent. It is an inhibitor of several CDKs and displays unique anticancer properties [19]. If the current clinical trials prove successful then flavopiridol, or similar agents, may in future be included in therapies for patients whose cancer cells display relevant genetic lesions, e.g. gene deletions in the RB1 pathway, including the CDKN2A locus. The qPCR method reported here was found to be very robust. In order to use it for detection of gene deletion in primary patient specimens, the content of non-malignant cells needs to be established by an independent method. In this study and in our previously reported work which showed deletion of p16INK4A in paediatric leukaemia to be associated with an increased risk of relapse [6], this was done morphologically since it is part of the routine diagnostic assessment of bone marrow specimens. However, other techniques could

Acknowledgements This work was supported by the Children’s Leukaemia and Cancer Research Foundation, Western Australia and NIH Grant CA83088.

References [1] Knudson Jr AG. Hereditary cancer, oncogenes, and antioncogenes. Cancer Res 1985;45:1437–43. [2] Cook WD, McCaw BJ. Accommodating haploinsufficient tumor suppressor genes in Knudson’s model. Oncogene 2000;19:3434–8. [3] Fodde R, Smits R. A matter of dosage. Science 2002;298:761–3. [4] Goss KH, Risinger MA, Kordich JJ, Sanz MM, Straughen JE, Slovek LE, et al. Enhanced tumor formation in mice heterozygous for Blm mutation. Science 2002;297:2051–3. [5] Gruber SB, Ellis NA, Scott KK, Almog R, Kolachana P, Bonner JD, et al. BLM heterozygosity and the risk of colorectal cancer. Science 2002;297:2013. [6] Carter TL, Watt PM, Kumar R, Burton PR, Reaman GH, Sather HN, et al. Hemizygous p16(INK4A) deletion in pediatric acute lymphoblastic leukemia predicts independent risk of relapse. Blood 2001;97:572–4. [7] Kees UR, Ford J, Price PJ, Meyer BF, Herrmann RP. PER-117: a new human ALL cell line with an immature thymic phenotype. Leuk Res 1987;11:489–98. [8] Kees UR, Ford J, Watson M, Murch A, Ringner M, Walker RL, et al. Gene expression profiles in a panel of childhood leukemia cell lines mirror critical features of the disease. Mol Cancer Ther 2003;2:671–7.

U.R. Kees et al. / Leukemia Research 29 (2005) 165–171 [9] Kees UR, Burton PR, Lu C, Baker DL. Homozygous deletion of the p16/MTS1 gene in pediatric acute lymphoblastic leukemia is associated with unfavorable clinical outcome. Blood 1997;89:4161–6. [10] Webber LM, Garson OM. Fluorodeoxyuridine synchronization of bone marrow cultures. Cancer Genet Cytogenet 1983;8:123–32. [11] Kees UR, Lukeis R, Ford J, Willoughby ML, Garson OM. Establishment and characterization of a childhood pre-B acute lymphoblastic leukemia cell line, PER-278, with chromosome translocations t(1;19) and t(1;9). Cancer Genet Cytogenet 1990;46:201–8. [12] Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 1999;286:531–7. [13] Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, et al. Distinct types of diffiuse large B-cell lymphoma identified by gene expression profiling. Nature 2000;403:503–11. [14] Hall M, Peters G. Genetic alterations of cyclins, cyclin-dependent kinases, and Cdk inhibitors in human cancer. Cancer Res 1996;68:67–108. [15] Drexler HG. Review of alterations of the cyclin-dependent kinase inhibitor INK4 family genes p15, p16, p18 and p19 in human leukemia-lymphoma cells. Leukemia 1998;12:845–59. [16] Schmitt CA, Fridman JS, Yang M, Lee S, Baranov E, Hoffman RM, et al. A senescence program controlled by p53 and p16INK4A contributes to the outcome of cancer therapy. Cell 2002;109:335–46. [17] Bertin R, Acquaviva C, Mirebeau D, Guidal-Giroux C, Vilmer E, Cave H, et al. CDKN2A, CDKN2B, and MTAP gene dosage permits precise characterization of mono- and bi-allelic 9p21 deletions in

[18]

[19]

[20] [21]

[22]

[23]

[24]

171

childhood acute lymphoblastic leukemia. Genes Chromosomes Cancer 2003;37:44–57. Lund AH, Turner G, Trubetskoy A, Verhoeven E, Wientjens E, Hulsman D, et al. Genome-wide retroviral insertional tagging of genes involved in cancer in Cdkn2a-deficient mice. Nat Genet 2002;32:160–5. Zhai S, Senderowicz AM, Sausville EA, Figg WD. Flavopiridol, a novel cyclin-dependent kinase inhibitor, in clinical development. Ann Pharmacother 2002;36:905–11. Quon KC, Berns A. Haplo-insufficiency? Let me count the ways. Genes Dev 2001;15:2917–21. Sasaki H, Zlatescu MC, Betensky RA, Ino Y, Cairncross JG, Louis DN. PTEN is a target of chromosome 10q loss in anaplastic oligodendrogliomas and PTEN alterations are associated with poor prognosis. Am J Pathol 2001;159:359–67. Kwabbi-Addo B, Giri D, Schmidt K, Podsypaninia K, Parsons R, Greenberg N, et al. Haploinsufficiency of the PTEN tumor suppressor gene promotes prostate cancer progression. Proc Natl Acad Sci USA 2001;98:11563–8. Carter TL, Terry PA, Gottardo N, Baker DL, Kees UR, Watt PM. Deletion of one copy of the p16INK4A tumor suppressor gene is implicated as a predisposing factor in pediatric leukemia. Biochem Biophys Res Commun 2004;318:852–5. Rosenwald A, Wright G, Wiestner A, Chan WC, Connors JM, Campo E, et al. The proliferation gene expression signature is a quantitative integrator of oncogenic events that predicts survival in mantle cell lymphoma. Cancer Cell 2003;3:185–97.