Measurement of DNA Copy Number at Microsatellite Loci Using Quantitative PCR Analysis

Introduction

Analysis of allelic imbalance at specific genomic loci is an important step in the molecular genetic analysis of human cancers because recurrent regions of imbalance signal the locations of genes that are involved in disease progression. Often, analysis of allelic imbalance is accomplished by analysis of polymorphic microsatellite markers [e.g., carrying SSRs 2 (1)] . This approach is powerful because SSRs are dispersed throughout the genomes of most eukaryotes with an average spacing of ∼30 kb (2, 3, 4) . However, assays of allelic imbalance typically require allelic heterozygosity so that only a subset of the loci will be informative in any individual (5) . This is particularly limiting in experimental murine systems using inbred mouse strains that have lost heterozygosity. Analyses of these mice require cross-breeding to reintroduce heterozygosity. This is both time-consuming and experimentally undesirable because the cross-breeding may introduce DNA sequence variations that alter tumor phenotype.

Here we describe a technique called QuMA that overcomes some of these limitations by allowing analysis of the relative DNA copy number at microsatellite markers distributed throughout the genomes of higher eukaryotes. This real-time, quantitative PCR technique is an alternative to analysis of allelic imbalance because most (but not all) allelic imbalance also results in a change in relative copy number. We demonstrate that QuMA is sufficiently sensitive to detect gain or loss of a single copy of a test locus in mouse and human. We show the biological utility of QuMA in localization of a putative tumor suppressor gene associated with radiation-induced murine myeloid leukemia on chromosome 2 and in establishing an association between increased copy number of the putative oncogene ZNF217 and reduced survival duration in human ovarian cancer.

Materials and Methods

QuMA

During QuMA, test and reference loci containing (CA)n repeats were amplified using PCR in the presence of TaqMan probe homologous for a (CA)n repeat. The amount of FAM fluorescence in each reaction liberated by the exonuclease degradation of the TaqMan probe during PCR amplification was measured as a function of PCR cycle number using an ABI 7700 Prism (PE Biosystems, Foster City CA; Refs. 6, 7, 8, 9 ). The number of PCR cycles (Ct) required for the FAM intensities to exceed a threshold just above background was calculated for the test and reference reactions. Ct values were determined for three test and three reference reactions in each sample, averaged, and subtracted to obtain ΔCt [ΔCt = Ct (test locus) − Ct (pooled reference)]. ΔCt values were measured for each unknown sample [ΔCt (test DNA)] and for samples from several unrelated, normal (calibrator) individuals[ average = ΔCt (calibrator DNA)]. Relative copy number at each locus in the test sample was then calculated as: where ΔΔCt = ΔCt (test DNA) − ΔCt (calibrator DNA), and E = PCR efficiency. For simplicity, we used primers with PCR efficiencies of >85% and calculated the relative DNA copy number as 2^−ΔΔCt (× 2 when known to be a diploid test sample, such as normal male DNA, or mouse knockout DNA or with FISH analysis of mouse tumors).

To determine whether a QuMA measurement on a single sample was significantly different from the mean of measurements made on samples from a number of normal individuals, a TI was calculated using the mean and SD of ΔCt values from each marker and the pooled reference according to the following formula: where 2.28 was a two-sided tolerance limit factor assuming a normal distribution for a total of n = 140 mouse loci (10) . The measured relative copy number of a normal sample will be within 2 ± TI 95% of the time with 95% confidence. Thus, individual measurements outside this range were considered significantly different from normal.

PCR

PCR was conducted in triplicate with 50-μl reaction volumes of 1× PCR buffer A (PE Biosystems), 2.5 mm MgCl₂, 0.4 μm each primer, 200μ m each deoxynucleotide triphosphate, 100 nm probe, and 0.025 unit/μl Taq Gold (PE Biosystems) with 1–5 ng of genomic DNA. A large master mix of the above-mentioned components was made for each experiment and aliquoted into each optical reaction tube. Each primer set (5–10 μl volume) was then added, and PCR was conducted using the following cycle parameters: 1 cycle of 95°C for 12 min and 40 cycles of 95°C for 20 s, 55°C for 20 s, 72°C for 45 s. Analysis was carried out using the sequence detection software supplied with the ABI 7700 (PE Biosystems). The Ct values for each set of three reactions were averaged for all subsequent calculations. Pooled variance for all sets of PCR triplicates was 0.018 (n = 288), indicating sufficient statistical power with this level of intra-assay variation to detect a difference of 0.25 cycle between samples with greater than 95% confidence.

Oligonucleotides

PCR primer sequences for microsatellite loci were obtained from either the Whitehead Institute Massachusetts Institute of Technology Center for Genome Research 3 or Washington University (St. Louis, MO). 4 Primers were synthesized by Life Technologies, Inc. (Gaithersburg, MD) or Integrated DNA Technologies (Coralville, IA). The reference pools contained primer pairs for six or seven different loci. For human analyses, loci were chosen in regions of the genome that usually did not show alterations in human breast tumors (human pool 1, D4S1605, D5S478, D11S913, D12S1699, D14S988, D21S1904, and D22S922) or ovarian tumors (human pool 2, D1S2868, D2S385, D4S1605, D5S643, D10S586, and D11S1315) when analyzed by CGH (11) . The murine reference loci were D1MIT64, D2MIT175, D3MIT12, D12MIT10, D13MIT250, and D14MIT5. Primer sequences for the test and reference loci analyzed in this study are listed in Table 1 ⇓ . The TaqMan CA-repeat fluorogenic probe consisted of the following sequence: 5′-FAM-TGTGTGTGTGTGTGTGTGTGT-6-carboxy tetramethyl rhodamine-3′. ZNF217 primers and probe were as reported previously (12) . Both probes were purchased from PE Biosystems.

FISH

Murinespleen cells on microscope slides fixed using methanol-acetic acid were analyzed using dual-color FISH with BAC probes as described previously (13) . Hybridization domains of each color were enumerated in at least 100 metaphase or interphase nuclei for each analysis. BAC clones containing SSRs D2MIT356 (BAC356), D2MIT62 (BAC62), and D2MIT415 (BAC415) were selected from a BAC library as described previously (14) . DNA from each BAC clone was extracted and purified using a midi-prep column according to manufacturer’s recommendations (Qiagen, Santa Clarita, CA) and labeled by nick translation to incorporate dCTP FITC or Texas Red (13) .

Specimens

Murine.

Two inbred strains of mice were evaluated: (a) a FVB knockout mouse (designated M11) that lacked a 700-kb segment of chromosome 11 (kindly provided by Dr. E. Rubin; Lawrence Berkeley National Laboratory, Berkeley, CA); tail DNA samples were obtained from M11^+/+, M11^+/−, and M11^−/− FVB mice; and (b) 3-month-old SJL mice (Jackson Laboratories, Bar Harbor, ME) exposed to 3 Gy of ionizing radiation from a Cs¹³⁷ irradiator. Spleen and marrow samples were collected for DNA isolation after 7–9 months.

Human.

All specimens were collected and used with the approval of the Committee on Human Research at the University of California, San Francisco. These specimens included: (a) blood samples collected from normal, healthy, male and female donors for analysis of X chromosome copy number; (b) ovarian tumor samples collected as described by Suzuki et al. (15) ; and (c) human breast cancer cell lines HBL100, MDA-MB-361, HS578T, SKBR3, and MCF-7 obtained from American Type Culture Collection (Manassas, VA) and cultured as recommended by the supplier.

DNA Isolation

DNA was extracted from either tail clippings or spleens of mice and processed with a Puregene DNA extraction kit (Gentra Systems, Minneapolis, MN) used according to the manufacturer’s purification recommendations. DNA was extracted from human blood specimens and cell lines using a Wizard Genomic DNA purification kit from Promega (Madison, WI). Ovarian tumor DNA was isolated as described previously (15) .

CGH

CGH was performed on DNA from three SJL mice as described by Kallioniemi et al. (16) . Mouse metaphase chromosome spreads were prepared from phytohemagglutinin-stimulated murine peripheral blood lymphocytes of a normal C57BL mouse. Image analysis was conducted as described by Piper et al. (17) .

Results and Discussion

Validation

Amplification.

We used QuMA to measure the levels of amplification of SSR loci mapped to 20q13.2 and the putative oncogene ZNF217 in five breast cancer cell lines that had been assessed previously using FISH (11 , 18) . Fig. 1 ⇓ shows good concordance between the two methods. Cell lines HBL100, MDA-MB-361, and HS578T show only modest increases (less than five copies) in DNA copy number, whereas SKBR3 and MCF-7 show high-level amplifications (up to 40 copies). However, the levels of amplification measured using FISH and QuMA differ by as much as 2× at some loci. We speculate that this is because the extent of the genome interrogated differs between the two methods. QuMA assesses copy number at a specific ∼200-bp wide locus, whereas FISH measures copy number using an ∼100-kb probe. The number of hybridization signals generated using FISH will vary in an uncertain manner if the level of amplification changes across the extent of the region covered by the probe. In addition, copy number is difficult to measure in interphase nuclei using FISH when the level of amplification is high. On the other hand, FISH is less sensitive to the presence of admixed normal cells. The level of aneusomy in these cell lines may also contribute to the difference because the reference loci may be elevated in copy number, resulting in underrepresentation of the absolute DNA copy number. Thus, analysis using both methods may be appropriate in cases of high-level amplification or aneusomy.

Single Copy Number Changes.

We assessed the ability of QuMA to detect monosomy by comparing copy number measurements on the X chromosome in human males with copy number at three autosomal loci and by analyzing copy number in a region of chromosome 11 deleted in a mouse knockout experiment. QuMA analyses in the human showed a relative copy number of 1.08 ± 0.035 for three autosomal loci (D8S258, D15S126, and D17S849) in male DNA and 0.46 ± 0.06 for the three nonpseudoautosomal X chromosome loci (DXS453, DXS1052, and DXS1196). QuMA measurements of relative copy number at D11MIT23 in a 700-kb segment of chromosome 11 deleted by homologous recombination 6 and at six autosomal loci (D1MIT64, D2MIT175, D3MIT12, D12MIT10, D13MIT250, and D14MIT5) were 0.59 ± 0.06 and 1.03 ± 0.13, respectively. Both studies clearly show that QuMA is sufficiently sensitive to distinguish one copy from two copies.

To further define the sensitivity of QuMA, we mixed DNA from a chromosome 11 knockout mouse with increasing amounts of DNA from normal FVB mice. QuMA was performed on these DNA mixtures at D11MIT23, D2MIT175, and D13MIT250. Fig. 2A ⇓ shows that QuMA was able to distinguish the reduced copy number at D11MIT23 even in the presence of 30% of normal DNA. Specifically, QuMA showed a 90% chance of detecting a difference between one copy and two copies, with 95% confidence under these conditions. Thus, QuMA should be able to detect the loss of one allele in tumor samples contaminated by this amount of normal DNA.

Demonstration Applications

Deletion Analysis in Tumors in Inbred Mice.

The SJL mouse strain has been shown to develop acute leukemias at high frequency when irradiated. Recurrent allelic imbalance of an ∼30-cM region of chromosome 2E has been reported in tumors in F₁ crosses of these mice (19 , 20) . We used QuMA to assess deletions at 27 loci along chromosome 2 using spleen DNA from eight irradiated SJL mice. Six of the eight spleens were comprised mostly of leukemic cells. Fig. 2B ⇓ shows that these analyses define an ∼23-cM-wide region of common deletion between D2MIT9 and D2MIT337. The QuMA results are consistent with those obtained using FISH and CGH. CGH analysis of DNA from spleens from two irradiated mice with leukemia (mice 2.4B and 11.2) and one irradiated mouse without leukemia (mouse 2.1B) showed chromosome 2 copy number losses in the two leukemic mice and no loss in the nonleukemic mouse. FISH analyses of spleens from two other irradiated mice with leukemia (mice 2.5 and 12.2) and one mouse without leukemia (mouse 8.5) were performed using BAC clones mapped to chromosome regions 2A(BAC356), 2E(BAC62), and 2H(BAC415). For the nonleukemic mouse (mouse 8.5), the 2A:2E copy number ratio was 0.92, whereas in leukemic mice (mice 2.5 and 12.2), the ratios were 1.67 and 1.89, respectively. The 2A:2H ratios for the same mice were both 1.02, indicating that these mice have an interstitial deletion of in the region of 2E. These results demonstrate the utility of QuMA for high-resolution mapping of the extent of physical deletions in inbred mouse strains, obviating the need for expensive and time-consuming backcrossing.

Reproducibility.

To assess the reproducibility of QuMA, 13 markers were measured three times on four irradiated mice. The average relative copy number for the disomic loci was 2.00 ± 0.26 (n = 123), with 5 of 123 measurements outside the TI (three above the cutoff values and two below the cutoff values). The average relative copy for monosomic loci was 0.94 ± 016 (n = 30), with all values outside the TI.

Assessment of Prognostic Markers in Human Tumors.

Amplifications of several regions of chromosome 20q occur frequently in a broad range of human cancers including ovarian cancer (11) . Several putative oncogenes have been identified in these regions including the putative zinc-finger transcription factor ZNF217 (12) . Amplification of this gene has been associated with reduced survival duration in human breast cancer (21) . We used QuMA to test the association between ZNF217 copy number increase and reduced survival for 60 ovarian cancer patients. The Kaplan-Meier survival curves for patients with or without increased copy number of ZNF217 in Fig. 3 ⇓ show that increased copy number of this gene was significantly associated with reduced survival duration in late-stage tumors and all tumors. This study demonstrates the utility of QuMA for rapid detection of prognostic copy number changes in human cancers.

Conclusion

We have demonstrated the utility of QuMA for rapid measurement of relative DNA copy number in the genomes of mouse and human tumors. QuMA distinguishes between increases and decreases in copy number and takes advantage of the large number of SSRs that have already been mapped and for which PCR primers are conveniently available. Of course, it does not detect events leading to loss of heterozygosity or other abnormalities that do not alter genome copy number. The use of a single TaqMan probe for all test and reference loci makes the method cost effective. The use of a “pooled reference” reduces the possibility that loss/gain of a single reference marker might lead to erroneous assignment of copy number at a test locus. However, the pooled reference should be designed to include loci that are not frequently altered in copy number in the system under study and may not provide a useful reference for determining absolute copy number in highly aneusomic tumors. Published CGH and LOH analyses may be used to guide the selection of reference loci. QuMA is advantageous because it does not require that the SSRs be polymorphic. This is particularly helpful in high-resolution analyses of the extent of regions of genome copy number abnormality in humans and inbred mouse tumors and for assessment of the presence of prognostic copy number abnormalities in individual tumors.