Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A Novel Approach for Determining Cancer Genomic Breakpoints in the Presence of Normal DNA

  • Yu-Tsueng Liu ,

    To whom correspondence should be addressed. E-mail:

    Affiliation Moores UCSD Cancer Center, University of California San Diego, La Jolla, California, United States of America

  • Dennis A. Carson

    Affiliation Moores UCSD Cancer Center, University of California San Diego, La Jolla, California, United States of America


CDKN2A (encodes p16INK4A and p14ARF) deletion, which results in both Rb and p53 inactivation, is the most common chromosomal anomaly in human cancers. To precisely map the deletion breakpoints is important to understanding the molecular mechanism of genomic rearrangement and may also be useful for clinical applications. However, current methods for determining the breakpoint are either of low resolution or require the isolation of relatively pure cancer cells, which can be difficult for clinical samples that are typically contaminated with various amounts of normal host cells. To overcome this hurdle, we have developed a novel approach, designated Primer Approximation Multiplex PCR (PAMP), for enriching breakpoint sequences followed by genomic tiling array hybridization to locate the breakpoints. In a series of proof-of-concept experiments, we were able to identify cancer-derived CDKN2A genomic breakpoints when more than 99.9% of wild type genome was present in a model system. This design can be scaled up with bioinformatics support and can be applied to validate other candidate cancer-associated loci that are revealed by other more systemic but lower throughput assays.


Tumors evolve through the continuous accumulation and selection of randomly mutated genes. While sets of advantageous mutations are selected in tumors, neutral or even slightly detrimental mutations may also occur due to genomic instability and genetic drift. Recently, much effort has been expended to identify in primary human cancers point mutations in the exons of cancer-related genes. However, systemic mapping of genomic DNA rearrangements has lagged behind, due to technical difficulties in detecting smaller deletions, tumor heterogeneity, and the necessity to purify malignant from normal cells [1]. Historically, such work was done by time consuming and labor intensive genetics and molecular cloning on established cancer cell lines [2], [3], [4]. One of the most striking examples is the homozygous deletion of the CDKN2A (INK4A/ARF) tumor suppressor locus, which was discovered in this and other laboratories [3], [4], [5], [6], [7], [8]. The CDKN2A deletions occur early during tumor development [9], [10], [11]. The p16INK4a (one of the CDKN2A products [12]) protein constrains cell cycle progression by the Rb pathway and may be responsible for the decline in the replicative potential of stem cells during aging [13]. The p14ARF (the other alternative reading frame of CDKN2A [14]) gene product regulates the expression of MDM2, the turnover of p53, and thereby controls the cellular response to stress (reviewed in [6], [7], [8], [15], [16], [17]). Because the Rb and p53 pathways are central to cancer gate-keeping and caretaking [18], [19], strong selection pressures exist for the disruption of the entire CDKN2A gene segment on both chromosomes. Few other deletions are as well characterized, although it is expected that more will be found when more data from array based comparative genomic hybridization (array-CGH) are reported and also through The Cancer Genome Atlas (TCGA) project [20], [21], [22], [23], [24]. It will be important to validate the relevance of those genomic rearrangements to cancer development since many of the genomic structural changes may be simply due to genome instability in cancer. Large scale studies with clinical samples will be the most reliable confirmation.

While point mutations and very small insertions or deletions in genomic DNA can be detected by exon re-sequencing, it can be more difficult to detect gene dosage changes of larger genomic fragments, especially deletions [1]. Current established techniques for deletion mapping, including Southern blotting [25], fluorescent in situ hybridization (FISH) [26], quantitative PCR [26], [27], [28], [29], [30], and array-CGH [31] rely on the absence of a detectable wild type signal [1]. This is problematic when a significant number of normal cells are present in a tumor sample. Array-CGH has the potential to analyze alterations of DNA copy number on a genome-wide scale with relatively high resolution, depending on whether BACs, PCR products or oligonucleotides are used for the array elements. However, these techniques often fail where there is a heterogeneous cell population or samples of poor quality [31]. FISH is less vulnerable to the presence of heterogeneous cell populations, but has relatively low resolution and is difficult to scale up. Except for FISH, the other techniques mentioned are not practical for mapping genomic translocations and inversions. End-sequencing profiling was developed to address this issue but the approach was costly and hard to scale up [32]. Therefore, there is a need to develop a scalable approach for detecting such genomic structural changes in solid tumors where heterogeneous cell populations are present.

Here we report a novel approach, designated as Primer Approximation Multiplex PCR (PAMP), to enrich small amounts of deleted genomic DNA sequences in the presence of wild type DNA. The genomic locations of the enriched sequences are subsequently decoded by a genomic tiling array and confirmed by sequencing.


CDKN2A locus

The CDKN2A is located on chromosome 9p21 (Figure 1). It encodes two proteins in different reading frames: p16INK4A and p14ARF, which both have 3 exons and share exons 2 and 3. CDKN2B (p15INK4B) and MTAP (methylthioadenosine phosphorylase) (not shown) are centromeric and telomeric neighboring genes respectively [3], [17], [33], [34]. BAC clone RP11-149I2 contains the whole CDKN2A genomic fragment and was used as template to generate probes (excluding repetitive regions) for printing on the minigenomic tiling array. The frequency of repetitive sequences predicted by RepeatMasker is shown at the bottom of the diagram.

Figure 1. The CDKN2A locus.

The genomic map covers about 55 kb around CDKN2A according to Ensemble [59]. CDKN2A/B is located at chromosome 9p21 and their RNA products are encoded by the reverse strand. CDKN2A encodes 2 proteins (p16INK4A and p14ARF) that share the same exons 2 and 3. The first exons of INK4A and ARF are about 20 kb apart. CDKN2B encodes p15INK4B that is homologous to p16INK4A. In addition to transcripts, the map also shows repetitive sequences (Repeat) and available BAC clone (Human tilepath clones), RP11-149I2.

Primer Approximation Multiplex PCR (PAMP)

It is difficult to detect a small fraction of deleted mutant genomic DNA in the presence of a vast excess of wild type DNA with array CGH or other popular molecular biology tools [26], [27], [35]. In typically contaminated tumor samples, genomic DNA is composed of various ratios of WT and CDKN2A deficient DNA. We aimed to take advantage of the fact that a shorter deleted genome sequence should be preferentially amplified compared to a much longer WT sequence using “approximated” flanking primers (Figure 2A) [36].

Figure 2. Primer Approximation Multiplex PCR (PAMP).

The efficiency of PCR amplification is inversely related to the distance of upstream and downstream primers. In this example, primers to amplify genomic sequences around the locus of interest (LOI) are divided into 20 groups: 10 each for forward (F1–F10) and reverse (R1-R10) groups (A). While all of the possible forward and reverse primer pairs are too far to each other for PCR amplification in wild type genome, certain pair of primers is brought closer (“approximated”) due to deletion (F3 and R3) in mutated genome. Multiplex PCR reactions are set and represented as a matrix to include one forward and one reverse primer group. The expected PCR results are shown as gray scale shadows in the matrix (B). This example shows that only group pairs close to breakpoint give PCR products (F3-R3, F3-R4, F4-R3).

The approach is illustrated in Figure 2. In this example, relatively even-spaced primers (average 1 kb apart) surrounding the locus of interest are divided into 20 groups for PCR. There are 10 groups each of forward primers, F1, F2 …, F10 and reverse primer R1, R2, …,R10, respectively (Figure 2A). Therefore, there are 100 pairs (F1-R1, F1-R2, …, F1-R10; F2-R1, F2-R2,…, F2-R10; …; F10-R1, F10-R2,…., F10-R10) of PCR reactions (Figure 2B). It is expected that only one or two pairs of PCR reactions will produce specific PCR products spanning the deletion boundary, since the other primer pairs should be too far from the breakpoint for efficient amplification. Then aliquots from each reaction can be mixed to hybridize on a single genomic tiling array. Unlike traditional array-CGH, it is expected that only spots representing genomic sequences near the breakpoints will be theoretically lit up, which was confirmed in the following experiments.

In order to increase the throughput and reduce the cost of reagents, every forward (F1-10) and reverse (R1-10) primer group can have multiple primers. Therefore, every PCR group (for example F1-R1) pair becomes multiplex PCR. Therefore, we designated this procedure as Primer Approximation Multiplex PCR (PAMP).

Deletion breakpoint cloning by PAMP and minigenomic tiling array

We reported previously that the Detroit 562 cell line has an approximate 20 kb (including INK4A exons 1 and 2) deletion on chromosome 9p21 [3]. We used this cell line to test our deletion scanning approach. Four groups (FA, FB, RY and RZ) of primers were used for four PAMP reactions (Figure 3A) using genomic DNA template either from Detroit 562 (CDKN2A deficient) or HEK293 (CDKN2A wild type) cell lines. Aliquots of all 4 PAMP reaction products were pooled and labeled for hybridization on an INK4A minigenomic tiling array that covers about 25 kb, including all of the exons of INK4A. As predicted in Figure 2, only spots with probes close to the breakpoints hybridized to the amplicons when Detroit 562 genomic DNA was used as a template (Figure 3B). Almost no signal was detected when HEK293 genomic DNA was used as a template. The control HEK293 sample had a significantly higher signal on Cot-1 DNA spots despite its general absolute signal intensity is low.

Figure 3. Breakpoint identification by PAMP with an INK4A minigenomic tiling array.

(A) Five groups of primers (FA, FB, Rx, RY and RZ, the small arrows and arrow heads) near the potential breakpoints were generated for PAMP based on our previous mapping [3]. The mapped CDKN2A breakpoints of the Detroit 562 cell line (Figure 5) are indicated for clarification. The “E1”, “E2” and “E3” designations (blue fonts) are the relative positions of INK4A exons. The first exon of ARF is further to the right of this diagram and is not covered by this array. The tiling probes for the array are indicated with two alternating colors (short black and orange lines) for ease of identification. (B) The first row of the INK4A minigenomic array was spotted with the tiling probes shown in panel A. Cot-1 DNA (repetitive sequence of genomic DNA) spots are indicated on this array. The rest of the spots are herring sperm DNA. Both Cot-1 and herring sperm DNA are used as nonspecific controls. This array was hybridized with labeled samples derived from two cell lines. The same sets of primers (FA, FB, RY and RZ) were used for PAMP reactions on Detroit 562 (mutant) and HEK293 (wild type) genomic DNA to map the potential CDKN2A breakpoints. The amplicons were labeled with different dyes, yielding a green signal (Cy-3) for the mutant sample and a red signal (Cy-5) for the wild type sample, to be simultaneously hybridized on the array (two-color array). The two green spots on the first row revealed the breakpoint location as been discussed in Figure 2.

In addition, four separate arrays were used to hybridize the individual PAMP products described above. A simple plot of signal intensity ratio of mutant/WT PCR products on the tiling array revealed the genomic location of the breakpoint (Figure 4). This analysis shows a very straightforward readout—the location of the deletion is bordered by two peaks. Only FB-RY (array 27) and all products pooling (array 29) produce the same result as shown in Figure 3. In contrast, the other three pairs yielded only faint background signals on the arrays. This result indicates that PAMP product pooling with a single array analysis gives the same breakpoint information as four individual arrays. The data support the original experimental predictions, and suggest that the procedure should be generally applicable for deletion and translocation scanning.

Figure 4. Pooling of individual PAMP reactions for single array hybridization.

Four groups (FA, FB, RY and RZ) of primers were used for four PAMP reactions each by pairing all of the possible forward and reverse primer groups using Detroit 562 (mutant) and HEK293 (control) as templates. The procedure has been briefly described in Figure 3. The products were labeled and used for array hybridization: FA-RY for array 25; FA-RZ for array 26; FB-RY for array 27 and FA-RZ for array 28. Aliquots of the individual PAMP samples were also pooled together and labeled for array hybridization (array 29, its array image is shown in Figure 3B). The results are presented with ratio signal intensity (Y-axis) of samples from Detroit 562 (mutant) and HEK293 (control) against the probe location (X-axis). The breakpoints can be identified through this plot by finding the two peaks that are analogous to the bright green spots in Figure 3B.

In order to pinpoint more precisely the area of deletion, nested PCR with pairs of specific primers was designed according to the earlier PAMP results. The PCR product was labeled for array hybridization, yielding a result very similar to that shown in Figure 4 and is shown in Figure 5A. Furthermore, the single major product of the PCR reactions was resolved by agarose gel electrophoresis, excised, extracted and sequenced (Figure 5B). The breakpoint cloned is in agreement with two other reports (Figure 5B) [37], [38].

Figure 5. Breakpoint mapping of Detroit 562 cell line.

To map the exact breakpoint, a nested set of PCR primers were designed for uniplex PCR based on the previous PAMP results (Figures 3 and 4). The PCR products were used for labeling and hybridized on the array and also for agarose gel electrophoresis. The array data is shown as the same plot in Figure 4. A single major band on the agarose gel was excised and purified for sequencing (B). The breakpoint is indicated (from #21975226 to #21960809 according to NCBI human genome sequence build 36).

To mimic the heterogeneous population of cancer and host cells typically found in solid tumors, various amounts of genomic DNA derived from Detroit 562 (mutant) and HEK293 (wild type) were mixed for PAMP and array hybridization. In order to test the sensitivity of our approach, we performed a titration experiment. The total genomic DNA for each assay was kept constant (100 ng). This is equivalent to about 28,000 copies of haploid genome (based on the estimate of 2.8×105 molecules/µg of haploid genome). The CDKN2A deleted cell line Detroit 562 was serially diluted with CDKN2A wild type HEK293 as shown in the Table 1. The assay was able to detect approximately 1 breakpoint sequence in the presence of an approximately 2000 fold excess of wild-type genome with sensitivity of 5–16 such molecules (Table 1). Thus, the PAMP approach provides a method for detecting genomic DNA deletions in the presence of more than 99.9% wild type DNA.

The PAMP breakpoint cloning strategy was confirmed in another cell line. Because our array only covers 25 kb of the genome, we scanned our previous mapping information on 100 cell lines to find one that might have breakpoints within this region [3], [33]. The Hs578T breast cancer cell line has a deletion in p16INK4A exons 1-3. With primers within and telomeric (FA and RX groups, Table 2) to the genomic fragment, we performed PAMP and array hybridization, and identified a single spot on the array (shown as a single peak in Figure 6A). The sequence of this probe is located from 69971 to 71219 in the RP11-149I2 BAC sequence (GenBank accession AL449423). Therefore, the centromeric end of the breakpoint should be located near this region. Uniplex PCR with primers from the two groups for PAMP was performed for sequencing. A PCR product about 2 kb was generated with a pair of primers and was subjected to direct sequencing (Figure 6B). The centromeric end of the breakpoint identified by PAMP is consistent with a previous report [37]. The telomeric end of the breakpoint was inferred from the primers used for PAMP and also confirmed by direct sequencing, although there was no genomic probe near the breakpoint that was included on the array.

Figure 6. Breakpoint mapping of Hs578T cell line.

Two groups of primers: FA (FA1-FA4) and RX (RX1-RX5) were used for PAMP based on our previous mapping. The product was labeled for array hybridization (A). Only single peak is evident from the plot. It indicates the location of the other breakpoint is not covered by this minigenomic array. Two primers (RX3 and RX4) located near the genomic location of the probe (human chromosome 9, 21969229 to 21970477, NCBI build 36) that was hybridized and two primers (FA1 and FA2) located outside the array coverage were chosen for uniplex PCR. The FA2-RX3 pair is expected to have the shortest distance when a deletion occurs. A band of about 2 kb on agarose gel was excised from the gel, purified and sequenced (B). The breakpoint and location is indicated (from #21955827 to #21968338 according to NCBI human genome sequence build 36).


We have developed a general strategy that can be applied for pinpointing the genomic breakpoints in unpurified primary cancers. The amplification and tiling protocol described here allows for simple and precise CDKN2A breakpoint cloning, using contaminated DNA as a template. In contrast to the current available techniques for deletion mapping (including Southern blotting, fluorescent in situ hybridization, real time PCR, and array CGH) that rely on the absence of a detectable wild type signal, PAMP directly measures the deleted DNA. Therefore, this approach is much less vulnerable to problems associated with normal cell contamination. The experimental procedure is robust enough to detect deletions in the presence of at least 99.9% wild type sequence contamination, which could not be achieved by other procedures [3], [5], [26], [27], [28], [33], [35], [39].

Primer approximation PCR screening has been a useful tool for isolating deletion mutants in C. elegans [36]. The method relies on identifying a single band that is the product of a successful PCR reaction when a pair of specific primers is brought together by deletion, on an agarose gel. The procedure can only identify deletions that happen in a very small genomic fragment (3 kb) in a relatively low throughput fashion. It also suffers from relatively high false positive rate because the identity of the bands on the agarose gel is difficult to know. However, by applying multiplex PCR together with a genomic tiling array, one can simultaneously screen a wider range of genomic regions [40]. In addition, preferential amplification of the sequences near the breakpoints generates a relatively straightforward readout on the tiling array. The signal to noise ratio on the hybridized spots is obvious compared to the readout from array CGH (see Figure 4). The junction can be readily identified as long as one end of the nearby genomic location of the breakpoints is covered by the tiling array, as shown in the case of Hs578T breast cancer cell (Figure 6). Since high-density genomic tiling arrays are commercially available, this approach can be easily adopted. In addition, high-throughput genome sequencing technology may also pinpoint the exact breakpoint sequence after PAMP, bypassing the need for array hybridization [41], [42], [43].

We used multiplex PCR to reduce the workload and cost for PAMP. We were able to multiplex 28 primers easily in a single PCR reaction. Theoretically, one can cover over 90% of the 0.5 Mb of genomic fragment around CDKN2A locus with a total of 500 primers in one single PCR reaction through computational simulation, which will be described elsewhere (manuscript submitted). A recent paper reported a successful multiplex PCR with more than 1000 primer pairs through the aid of computational design [44]. The PAMP approach targets deletion sizes between 10 kb and 1 Mb. The smaller or larger deletions can be detected by resequencing and FISH respectively.

Like other PCR technologies, PAMP can be easily adopted to a robotic system for clinical and research purposes. One example of a potential clinical application is to use the unique breakpoint sequence as a personalized cancer-specific biomarker for disease monitoring after treatment, when the precise breakpoint has been mapped. For example, unlike many current tumor markers, such as CA19-9, CA125 and PSA, which are not truly cancer-specific, the CDKN2A breakpoints are specific and unique for each cancer with this locus deleted. A highly sensitive assay, such as real-time PCR, can be designed to monitor the status of cancer progression in the blood or other body fluids. The assay should be very specific because amplification is expected to occur only from deletion-containing DNA due to very long distance between the primers in the wild type genome (see Figure 2). This is analogous to the detection of a foreign virus sequence, which has been applied as a useful biomarker for Epstein-Barr virus associated nasopharyngeal carcinoma [45], [46].

Our approach can also ease traditional labor-intensive experiments that aim to understand how genomic breakpoints are generated during cancer development, particularly in primary tumors. Although illegitimate V(D)J recombination may be responsible for creating CDKN2A deletions in acute lymphoblastic leukemia, more breakpoint sequence data will be needed for other types of cancers to delineate the molecular mechanisms [37], [38], [47], [48]. Furthermore, the technique described in this paper can be used not only for deletion mapping, but it can also be applied to map other types of genomic rearrangement, such as translocations and inversions. Similar to the case of genomic deletion (see Figure 2), only “approximated” primers can generate amplicons when those primers are near the genomic fragments that are repositioned in translocations and inversions.

Using breakpoint sequences as cancer-specific biomarkers to monitor minimal residual diseases has been explored [49], [50],[51],[52],[53]. Disease monitoring based on personalized genomic DNA breakpoint is considered to be highly attractive approach for several reasons[49]. First, many genomic DNA rearrangements are directly related to oncogenic process, therefore, are truly cancer-specific and stable over time. This is in contrast to more convenient Ig/TCR rearrangement based assay. Indeed, we found exactly the same CDKN2A breakpoints of the two cell lines used in this study as reported by others. Second, the DNA is more stable than RNA although it is easier to map fusion transcript if it exists, such as BCR-ABL. Third, the genomic breakpoints are very likely to be different from each patient and become personalized biomarkers, thereby, reducing the risk of false positive results due to cross contamination. However, this is also the biggest hurdle to overcome. For example, many efforts to improve the PCR amplification range for detecting C-MYC/immunoglobulin translocations have had limited success because the breakpoints are scattered across a more than 300 kb region [54], [55], [56]. Our strategy may be useful for such application.

Our approach aims to identify breakpoints within a 1 Mb genomic fragment as FISH or other cytogenetic techniques are available for larger genomic rearrangements and the cost and labor significantly increase when the target region expands. We are able to inexpensively produce a tiling array covering a genomic fragment of 0.5 Mb around the CDKN2A with 1 kb resolution. We are currently working on methods to increase multiplexing and reduce the volume of each PAMP reaction for broader applications.

Materials and Methods

Cell culture and sample preparation

The cell lines described in the paper were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and cultured as recommended. The genomic DNA was extracted with DNAzol (Molecular Research Center, Inc., Cincinnati, OH) following the instructions from the manufacturer.

Minigenomic tiling array

We created an INK4A minigenomic tiling array covering a 25 kb fragment in the CDKN2A locus for proving the concept of our approach. DNA probes were generated by PCR with BAC clone RP11-149I2 (obtained from BACPAC Resources Center at Children's Hospital Oakland Research Institute, Oakland, CA) as template and avoiding the repetitive genomic sequences that were predicted by RepeatMasker. The PCR products were purified with DNA Clean-up and Concentrator-5 (Zymo Research, Orange, CA), resuspended in 3×SSC and printed on poly-L-lysine slides at 0.1 mg/ml along with Human Cot-1 DNA (Invitrogen, Carlsbad, CA), which is enriched for repetitive sequences, and herring sperm DNA (Promega, Madison, WI), which was used as nonspecific control. The printing procedure has been described and essentially followed the manual of the DeRisi arrayer with silicon microcontact printing pins (Parallel Synthesis Technologies, Inc. Santa Clara, CA) [57], [58]. Arrays were post-processed with succinic anhydride-based method for blocking before hybridization as previously described [57]. The protocols related to array printing and hybridization in this paper generally can be found in (

Primer-Approximation Multiplex PCR (PAMP) and array hybridization

A simplified PAMP scheme is shown in Figure 2. A series of primers (Table 2) toward INK4A exons 1-2 along the CDKN2A locus were synthesized by Integrated DNA Technologies (Coralville, IA). Groups of forward and reverse primers (250 nM each in the final reaction) were used to generate amplicons from 0.1 µg of genomic DNA templates in a total of 10 µl of solution mixing with 10 µl of Taq 2×Master Mix (New England Biolabs, Ipswich, MA). The reaction was assembled at 4°C in a PCR workstation and transferred to a thermocycler with the block preheated to 94°C. The cycling conditions were a 3-minute denaturation step at 94°C followed by 35 cycles at 92°C for 30 sec, 55°C for 30 sec and 68°C for 2.5 minutes with a final extension step at 68°C for 5 minutes. One µl of unpurified product was subsequently used as templates for another round of amplification to label the amplicons with the same PCR protocol except that dTTP was replaced by a 4∶1 mixture of aminoallyl dUTP (Ambion, Austin, TX) and dTTP for probe labeling. The labeled amplicons were purified with DNA Clean-up and Concentrator-5 columns, eluted in 9 µl of sodium bicarbonate (pH 9.0) and coupled with 1 µl of DMSO dissolved Cy3 or Cy5 NHS esters (GE Healthcare, Piscataway, NJ) for 30 to 60 minutes. The Cy3 and Cy5 labeled amplicons were purified with DNA Clean-up and Concentrator-5 columns and eluted with 10 µl of 10 mM Tris-HCl (pH 8.0). Paired Cy3 and Cy5 labeled amplicons were combined with 3.6 µl of 20×SSC, 0.5 µl of Hepes (pH 7.0) and finally 0.5 µl of 10% SDS. The mixed solution was heated for 2 minutes at 95°C, cooled to room temperature and hybridized to the minigenomic tiling arrays at 63°C overnight essentially as previously described [57], [58]. The hybridized arrays were washed and scanned with GenePix 4000B scanner (Molecular Device, Sunnyvale, CA) and analyzed by GenePix Pro 6.0 software.

Tiling array data analysis

The human Cot-1 DNA and herring sperm DNA were designed to be positive and negative controls respectively and spotted multiple times on the array (see Figure 3). To normalize for day-to-day and sample-to-sample variation, the median intensity of all features representing herring sperm DNA (I50%-HS) were used to divide the intensity (IG) of each feature representing genomic probes. Each (IG):(I50%-HS) ratio, the normalized genomic probe signal, was plotted at the Y-axis against the corresponding probe's genomic location at the X-axis to ease data interpretation (see Figure 4).

CDKN2A breakpoint cloning

To confirm CDKN2A breakpoint mapping by PAMP approach, the genomic fragments flanking the breakpoints were cloned by traditional PCR approaches guided by the results from PAMP. Two examples are given in this paper.

Detroit 562 cell line: The nested approach was used to clone the CDKN2A breakpoint in Detroit 562 human epithelial carcinoma cells. The primers were designed with clues from the PAMP experiment (see Figures 3 and 4). External primers (AGGTTTGGTTAAGAGTCGTTC and AAGATCTATATGGTGGCCTTTAG) were used for 35 cycles of PCR (92°C, 30 seconds; 55°C, 30 seconds; 68°C, 2 minutes). The two sets of internal primers (ATGCTAGCTGTAACTGGAGC and CTTAAGGCTAAATTGACTTG; GGCTTAGAGCTAACTCTTCACCC and TATGTGTGTGTGTGTCTGTGTGATG) were used for the second PCR reaction under the same conditions. A single band about 1 kb in size was excised and extracted with Qiaquick gel extraction kit (Qiagen, Valencia, CA). The purified product was directly sequenced with internal primers.

Hs578T cell line: Four sets of uniplex PCR reactions were performed by pairing 2 single forward (FA1 and FA2) and 2 single reverse (RX3 and RX4) primers. The PCR program was the same as that for PAMP. The products were analyzed by agarose gel electrophoresis. A single band from the shortest distance pair (FA2-RX3) was excised and sequenced with an internal primer.


We thank Yandong Zhao and Qing Lu for technical assistance in array hybridization, Vineet Bafna and Ali Bashir for discussions and April Livengood and Richard Schwab for critical reading of the manuscript.

Author Contributions

Conceived and designed the experiments: DC YL. Performed the experiments: YL. Analyzed the data: DC YL. Contributed reagents/materials/analysis tools: DC YL. Wrote the paper: DC YL.


  1. 1. Armour JA, Barton DE, Cockburn DJ, Taylor GR (2002) The detection of large deletions or duplications in genomic DNA. Hum Mutat 20: 325–337.
  2. 2. Heisterkamp N, Groffen J (2002) Philadelphia-positive leukemia: a personal perspective. Oncogene 21: 8536–8540.
  3. 3. Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K, et al. (1994) Deletions of the cyclin-dependent kinase-4 inhibitor gene in multiple human cancers. Nature 368: 753–756.
  4. 4. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, et al. (1994) A cell cycle regulator potentially involved in genesis of many tumor types. Science 264: 436–440.
  5. 5. Cairns P, Polascik TJ, Eby Y, Tokino K, Califano J, et al. (1995) Frequency of homozygous deletion at p16/CDKN2 in primary human tumours. Nat Genet 11: 210–212.
  6. 6. Rocco JW, Sidransky D (2001) p16(MTS-1/CDKN2/INK4a) in cancer progression. Exp Cell Res 264: 42–55.
  7. 7. Ruas M, Peters G (1998) The p16INK4a/CDKN2A tumor suppressor and its relatives. Biochim Biophys Acta 1378: F115–177.
  8. 8. Sharpless NE (2005) INK4a/ARF: a multifunctional tumor suppressor locus. Mutat Res 576: 22–38.
  9. 9. Bardeesy N, DePinho RA (2002) Pancreatic cancer biology and genetics. Nat Rev Cancer 2: 897–909.
  10. 10. Mao L, Hong WK, Papadimitrakopoulou VA (2004) Focus on head and neck cancer. Cancer Cell 5: 311–316.
  11. 11. Maley CC, Reid BJ (2005) Natural selection in neoplastic progression of Barrett's esophagus. Semin Cancer Biol 15: 474–483.
  12. 12. Serrano M, Hannon GJ, Beach D (1993) A new regulatory motif in cell-cycle control causing specific inhibition of cyclin D/CDK4. Nature 366: 704–707.
  13. 13. Kim WY, Sharpless NE (2006) The regulation of INK4/ARF in cancer and aging. Cell 127: 265–275.
  14. 14. Quelle DE, Zindy F, Ashmun RA, Sherr CJ (1995) Alternative reading frames of the INK4a tumor suppressor gene encode two unrelated proteins capable of inducing cell cycle arrest. Cell 83: 993–1000.
  15. 15. Sherr CJ (2004) Principles of tumor suppression. Cell 116: 235–246.
  16. 16. Campisi J (2005) Senescent cells, tumor suppression, and organismal aging: good citizens, bad neighbors. Cell 120: 513–522.
  17. 17. Gil J, Peters G (2006) Regulation of the INK4b-ARF-INK4a tumour suppressor locus: all for one or one for all. Nat Rev Mol Cell Biol 7: 667–677.
  18. 18. Kinzler KW, Vogelstein B (1997) Cancer-susceptibility genes. Gatekeepers and caretakers. Nature 386: 761, 763.
  19. 19. Vogelstein B, Kinzler KW (2002) The genetic basis of human cancer. New York: McGraw-Hill Professional.
  20. 20. Vastag B (2006) NIH Institutes launch joint venture to map cancer genome. J Natl Cancer Inst 98: 162.
  21. 21. Zender L, Spector MS, Xue W, Flemming P, Cordon-Cardo C, et al. (2006) Identification and validation of oncogenes in liver cancer using an integrative oncogenomic approach. Cell 125: 1253–1267.
  22. 22. Hicks J, Krasnitz A, Lakshmi B, Navin NE, Riggs M, et al. (2006) Novel patterns of genome rearrangement and their association with survival in breast cancer. Genome Res 16: 1465–1479.
  23. 23. Aguirre AJ, Brennan C, Bailey G, Sinha R, Feng B, et al. (2004) High-resolution characterization of the pancreatic adenocarcinoma genome. Proc Natl Acad Sci U S A 101: 9067–9072.
  24. 24. Rivera MN, Kim WJ, Wells J, Driscoll DR, Brannigan BW, et al. (2007) An X chromosome gene, WTX, is commonly inactivated in Wilms tumor. Science 315: 642–645.
  25. 25. Petrij-Bosch A, Peelen T, van Vliet M, van Eijk R, Olmer R, et al. (1997) BRCA1 genomic deletions are major founder mutations in Dutch breast cancer patients. Nat Genet 17: 341–345.
  26. 26. Perry A, Nobori T, Ru N, Anderl K, Borell TJ, et al. (1997) Detection of p16 gene deletions in gliomas: a comparison of fluorescence in situ hybridization (FISH) versus quantitative PCR. J Neuropathol Exp Neurol 56: 999–1008.
  27. 27. Kees UR, Terry PA, Ford J, Everett J, Murch A, et al. (2005) Detection of hemizygous deletions in genomic DNA from leukaemia specimens for the diagnosis of patients. Leuk Res 29: 165–171.
  28. 28. Carter TL, Watt PM, Kumar R, Burton PR, Reaman GH, et al. (2001) Hemizygous p16(INK4A) deletion in pediatric acute lymphoblastic leukemia predicts independent risk of relapse. Blood 97: 572–574.
  29. 29. M'Soka TJ, Nishioka J, Taga A, Kato K, Kawasaki H, et al. (2000) Detection of methylthioadenosine phosphorylase (MTAP) and p16 gene deletion in T cell acute lymphoblastic leukemia by real-time quantitative PCR assay. Leukemia 14: 935–940.
  30. 30. Batova A, Diccianni MB, Nobori T, Vu T, Yu J, et al. (1996) Frequent deletion in the methylthioadenosine phosphorylase gene in T-cell acute lymphoblastic leukemia: strategies for enzyme-targeted therapy. Blood 88: 3083–3090.
  31. 31. Pinkel D, Albertson DG (2005) Array comparative genomic hybridization and its applications in cancer. Nat Genet 37 SupplS11–17.
  32. 32. Volik S, Zhao S, Chin K, Brebner JH, Herndon DR, et al. (2003) End-sequence profiling: sequence-based analysis of aberrant genomes. Proc Natl Acad Sci U S A 100: 7696–7701.
  33. 33. Nobori T, Takabayashi K, Tran P, Orvis L, Batova A, et al. (1996) Genomic cloning of methylthioadenosine phosphorylase: a purine metabolic enzyme deficient in multiple different cancers. Proc Natl Acad Sci U S A 93: 6203–6208.
  34. 34. Olopade OI, Pomykala HM, Hagos F, Sveen LW, Espinosa R 3rd, et al. (1995) Construction of a 2.8-megabase yeast artificial chromosome contig and cloning of the human methylthioadenosine phosphorylase gene from the tumor suppressor region on 9p21. Proc Natl Acad Sci U S A 92: 6489–6493.
  35. 35. Dijkman R, Tensen CP, Jordanova ES, Knijnenburg J, Hoefnagel JJ, et al. (2006) Array-based comparative genomic hybridization analysis reveals recurrent chromosomal alterations and prognostic parameters in primary cutaneous large B-cell lymphoma. J Clin Oncol 24: 296–305.
  36. 36. Jansen G, Hazendonk E, Thijssen KL, Plasterk RH (1997) Reverse genetics by chemical mutagenesis in Caenorhabditis elegans. Nat Genet 17: 119–121.
  37. 37. Sasaki S, Kitagawa Y, Sekido Y, Minna JD, Kuwano H, et al. (2003) Molecular processes of chromosome 9p21 deletions in human cancers. Oncogene 22: 3792–3798.
  38. 38. Raschke S, Balz V, Efferth T, Schulz WA, Florl AR (2005) Homozygous deletions of CDKN2A caused by alternative mechanisms in various human cancer cell lines. Genes Chromosomes Cancer 42: 58–67.
  39. 39. Caldas C, Hahn SA, da Costa LT, Redston MS, Schutte M, et al. (1994) Frequent somatic mutations and homozygous deletions of the p16 (MTS1) gene in pancreatic adenocarcinoma. Nat Genet 8: 27–32.
  40. 40. Fan JB, Chee MS, Gunderson KL (2006) Highly parallel genomic assays. Nat Rev Genet 7: 632–644.
  41. 41. Shendure J, Porreca GJ, Reppas NB, Lin X, McCutcheon JP, et al. (2005) Accurate multiplex polony sequencing of an evolved bacterial genome. Science 309: 1728–1732.
  42. 42. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, et al. (2005) Genome sequencing in microfabricated high-density picolitre reactors. Nature 437: 376–380.
  43. 43. Service RF (2006) Gene sequencing. The race for the $1000 genome. Science 311: 1544–1546.
  44. 44. Wang HY, Luo M, Tereshchenko IV, Frikker DM, Cui X, et al. (2005) A genotyping system capable of simultaneously analyzing>1000 single nucleotide polymorphisms in a haploid genome. Genome Res 15: 276–283.
  45. 45. Lin JC, Wang WY, Chen KY, Wei YH, Liang WM, et al. (2004) Quantification of plasma Epstein-Barr virus DNA in patients with advanced nasopharyngeal carcinoma. N Engl J Med 350: 2461–2470.
  46. 46. To EW, Chan KC, Leung SF, Chan LY, To KF, et al. (2003) Rapid clearance of plasma Epstein-Barr virus DNA after surgical treatment of nasopharyngeal carcinoma. Clin Cancer Res 9: 3254–3259.
  47. 47. Cayuela JM, Gardie B, Sigaux F (1997) Disruption of the multiple tumor suppressor gene MTS1/p16(INK4a)/CDKN2 by illegitimate V(D)J recombinase activity in T-cell acute lymphoblastic leukemias. Blood 90: 3720–3726.
  48. 48. Kitagawa Y, Inoue K, Sasaki S, Hayashi Y, Matsuo Y, et al. (2002) Prevalent involvement of illegitimate V(D)J recombination in chromosome 9p21 deletions in lymphoid leukemia. J Biol Chem 277: 46289–46297.
  49. 49. van der Velden VH, Hochhaus A, Cazzaniga G, Szczepanski T, Gabert J, et al. (2003) Detection of minimal residual disease in hematologic malignancies by real-time quantitative PCR: principles, approaches, and laboratory aspects. Leukemia 17: 1013–1034.
  50. 50. Goulden N, Virgo P, Grimwade D (2006) Minimal residual disease directed therapy for childhood acute myeloid leukaemia: the time is now. Br J Haematol 134: 273–282.
  51. 51. Hughes T, Deininger M, Hochhaus A, Branford S, Radich J, et al. (2006) Monitoring CML patients responding to treatment with tyrosine kinase inhibitors: review and recommendations for harmonizing current methodology for detecting BCR-ABL transcripts and kinase domain mutations and for expressing results. Blood 108: 28–37.
  52. 52. Szczepanski T (2007) Why and how to quantify minimal residual disease in acute lymphoblastic leukemia? Leukemia..
  53. 53. Cazzaniga G, Gaipa G, Rossi V, Biondi A (2006) Monitoring of minimal residual disease in leukemia, advantages and pitfalls. Ann Med 38: 512–521.
  54. 54. Janz S (2006) Myc translocations in B cell and plasma cell neoplasms. DNA Repair (Amst) 5: 1213–1224.
  55. 55. Joos S, Falk MH, Lichter P, Haluska FG, Henglein B, et al. (1992) Variable breakpoints in Burkitt lymphoma cells with chromosomal t(8;14) translocation separate c-myc and the IgH locus up to several hundred kb. Hum Mol Genet 1: 625–632.
  56. 56. Joos S, Haluska FG, Falk MH, Henglein B, Hameister H, et al. (1992) Mapping chromosomal breakpoints of Burkitt's t(8;14) translocations far upstream of c-myc. Cancer Res 52: 6547–6552.
  57. 57. Eisen MB, Brown PO (1999) DNA arrays for analysis of gene expression. Methods Enzymol 303: 179–205.
  58. 58. Wang D, Urisman A, Liu YT, Springer M, Ksiazek TG, et al. (2003) Viral discovery and sequence recovery using DNA microarrays. PLoS Biol 1: E2.
  59. 59. Hubbard T, Andrews D, Caccamo M, Cameron G, Chen Y, et al. (2005) Ensembl 2005. Nucleic Acids Res 33: D447–453.