Abstract Background Epidermal growth factor receptor (EGFR) variant III (vIII) is the most common oncogenic rearrangement in glioblastoma (GBM), generated by deletion of exons 2 to 7 of EGFR. The proximal breakpoints occur in variable positions within the 123-kb intron 1, presenting significant challenges in terms of polymerase chain reaction (PCR)–based mapping. Molecular mechanisms underlying these deletions remain unclear. Methods We determined the presence of EGFRvIII and its breakpoints for 29 GBM samples using quantitative PCR, arrayed PCR mapping, Sanger sequencing, and whole genome sequencing (WGS). Patient-specific breakpoint PCR was performed on tumors, plasma, and cerebrospinal fluid (CSF) samples. The breakpoint sequences and single nucleotide polymorphisms (SNPs) were analyzed to elucidate the underlying biogenic mechanism. Results PCR mapping and WGS independently unveiled 8 EGFRvIII breakpoints in 6 tumors. Patient-specific primers yielded EGFRvIII PCR amplicons in matched tumors and in cell-free DNA (cfDNA) from a CSF sample, but not in cfDNA or extracellular-vesicle DNA from plasma. The breakpoint analysis revealed nucleotide insertions in 4 samples, an insertion of a region outside of the EGFR locus in 1, microhomologies in 3, as well as a duplication or an inversion accompanied by microhomologies in 2, suggestive of distinct DNA repair mechanisms. In the GBM samples that harbored distinct breakpoints, the SNP compositions of EGFRvIII and amplified non-vIII EGFR were identical, suggesting that these rearrangements arose from amplified non-vIII EGFR. Conclusion Our approach efficiently “fingerprints” each sample’s EGFRvIII breakpoints. Breakpoint sequence analyses suggest that independent breakpoints arose from precursor amplified non-vIII EGFR through different DNA repair mechanisms. EGFR, EGFRvIII, genomic rearrangement, glioblastoma, liquid biopsy Importance of the study EGFRvIII is a truncation mutant lacking exons 2–7 and is associated with glioblastoma aggressiveness. The breakpoints of this large genomic deletion are generated by unknown mechanisms, are highly variable between patient samples, and are difficult to map. Here we show a sensitive and specific method for determining EGFRvIII-positive samples using discriminatory quantitative PCR followed by specific breakpoint mapping using arrayed primers spanning the long intron 1 of EGFR. Patient-specific PCR, designed based on the mapping results, is useful for the detection of EGFRvIII breakpoints unique to each tumor and we show a case where patient-specific breakpoint products were detected in the CSF using this method. Detailed breakpoint sequence analyses show traces of different types of genomic rearrangements and DNA repair mechanisms. Analyses of SNPs suggest that different deletions may evolve from precursor amplified non-vIII EGFR. Glioblastoma (GBM) is the most common primary malignant tumor of the central nervous system in adults.1 Epidermal growth factor receptor (EGFR) variant III (vIII) is the most common mutant EGFR found in GBM and is known to be associated with aggressive pathological features such as enhanced tumorigenicity, invasion, and therapeutic resistance.2–4 This constitutively active mutant of EGFR is generated through deletions encompassing exons 2–7, which result in a transcript producing a truncated receptor.5EGFRvIII in GBM resides primarily on extrachromosomal DNA (ecDNA) in an amplified manner.6,7 One study investigating EGFRvIII breakpoints suggests that Alu (Arthrobacter luteus) repeat elements may play a role in genomic rearrangements leading to the origin of the mutant.8 However, a link between underlying mechanisms of EGFR gene amplification and genomic rearrangements that result in a vIII deletion is not yet well understood. The proximal genomic breakpoints of EGFRvIII can occur anywhere within a 123-kb first intron, which presents a significant challenge for polymerase chain reaction (PCR)–based mapping.8,9 Next-generation sequencing is a method to map these deletions but has yet to be routinely applied in the clinic.10 Additionally, although reliable antibodies that can distinguish wild-type EGFR from EGFRvIII on tissue sections exist,11 the presence of amplified and rearranged genes does not always correlate with their expression due to epigenetic regulation.12 In an effort to provide a more efficient approach for analyzing EGFRvIII-positive GBMs, we modified previously reported mapping techniques8,9 by first screening samples for possible EGFRvIII deletions using genomic quantitative (q)PCR and then direct mapping of these deletions through arrayed PCR. The breakpoint sequence analyses suggest distinct DNA repair processes as biogenic mechanisms. Materials and Methods Samples Twenty-nine GBM tumor samples from 29 patients who underwent surgery at the University of California San Diego were analyzed. Paired plasma samples were obtained from all the patients. In one patient (BC010), who did not show any clear evidence of leptomeningeal disease, the cerebrospinal fluid (CSF) was collected through a puncture of the lateral ventricle before resection of the tumor. All samples were collected with informed consent according to the appropriate protocols approved by the institutional review board. Six GBM neurosphere cell lines and one GBM tissue that were analyzed in our previous studies7,9 and known to have EGFRvIII were included in the breakpoint sequence analysis. DNA Extraction Tumor DNA was extracted using the DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer’s instructions. Cell-free DNA (cfDNA) from plasma and CSF was extracted using the QIAamp Circulating Nucleic Acid Kit (Qiagen). The obtained cfDNA was unbiasedly amplified according to a previous report.13 Briefly, 5 ng of cfDNA was blunted and ligated using T4 DNA polymerase and T4 DNA ligase (New England Biolabs), respectively. Then, the elongated or circularized DNA was amplified via random primer-initiated multiple displacement amplification using the Illustra GenomiPhi HY DNA Amplification Kit (GE Healthcare Life Sciences). The amplified DNA was purified by ethanol precipitation and quantified using Qubit dsDNA HS (Thermo Fisher Scientific). Extracellular vesicles from plasma or CSF were purified using ExoQuick (System Biosciences) according to the manufacturer’s instruction. DNA was further extracted from extracellular vesicles using the DNeasy Blood and Tissue Kit (Qiagen) and then a portion of the DNA was treated with heparinase I (Sigma-Aldrich) according to a previous report.14 Briefly, 5 µL of DNA sample was combined with 5 µL of heparinase working solution (0.085 IU/mL of heparinase I, 10 mmol/L Tris HCl pH 7.5, 2 mmol/L CaCl2, 25 mmol/L NaCl) and incubated at 25°C for 3 h. Extracellular-vesicle DNA (evDNA) with and without heparinase I treatment was then blunted, ligated, amplified, and purified in the same manner as cfDNA. Quantitative PCR and Statistical Analyses Triplicate qPCR reactions containing 10 ng of DNA were run on a CFX96 Real Time System (Bio-Rad) with the following reaction conditions: 95°C for 5 min, 40 cycles of 95°C for 15 s and 58°C for 30 s. Three primer pairs between exons 2 and 7 were designed to quantify non-vIII EGFR and 2 pairs were set outside for the quantification of both vIII and non-vIII EGFR (Fig. 1A). The data were normalized to human hemoglobin beta (HBB) in each sample and again normalized to human genomic DNA (Promega), and the relative copy number of each region was determined using the 2-ΔΔCt formula. Primer sequences are listed in Supplementary Table S1. EGFR was regarded as amplified when the average of the relative copy numbers of the 2 regions outside the deletion compared with HBB was 2 or more. Datasets were analyzed by unpaired Student’s t-test using GraphPad Prism software. Fig. 1 View largeDownload slide Screening and mapping of EGFRvIII deletions in GBM. (A) Design of the qPCR primers specific for the EGFR gene. Blue bars indicate introns and orange numbered boxes show exons. Arrows on top indicate where the primers are designed. (B) Results of the copy number analysis of different regions on EGFR. A dotted line corresponds to a relative copy number of 2 compared with the control gene. Red squared samples are those with significantly greater copy numbers of regions outside of the EGFRvIII deletion compared with those inside in at least 1 of 6 different comparisons listed in the table below. (C) Design of mapping PCR primers corresponding to the EGFR gene. Blue bars indicate introns and orange numbered boxes show exons. Arrows on top indicate primers. (D) Agarose gel electrophoresis displaying mapping PCR amplicons. Red squares show serial samples specific to EGFRvIII deletions. Arrows indicate the smallest PCR amplicons among each series of amplicons and the primers that generated the smallest amplicons. Fig. 1 View largeDownload slide Screening and mapping of EGFRvIII deletions in GBM. (A) Design of the qPCR primers specific for the EGFR gene. Blue bars indicate introns and orange numbered boxes show exons. Arrows on top indicate where the primers are designed. (B) Results of the copy number analysis of different regions on EGFR. A dotted line corresponds to a relative copy number of 2 compared with the control gene. Red squared samples are those with significantly greater copy numbers of regions outside of the EGFRvIII deletion compared with those inside in at least 1 of 6 different comparisons listed in the table below. (C) Design of mapping PCR primers corresponding to the EGFR gene. Blue bars indicate introns and orange numbered boxes show exons. Arrows on top indicate primers. (D) Agarose gel electrophoresis displaying mapping PCR amplicons. Red squares show serial samples specific to EGFRvIII deletions. Arrows indicate the smallest PCR amplicons among each series of amplicons and the primers that generated the smallest amplicons. Mapping of EGFRvIII A forward primer within exon 1 (e1f) and 45 forward primers within intron 1 (i1.1–i1.45) were designed to be spaced approximately 2.7-kb apart from neighboring primers. Together with forward primers in exons 7 and 8 (e7f and e8f) as DNA quality controls, all 48 forward primers were combined with an anchor primer in exon 8 (e8r) and PCR was performed on 100 ng of genomic DNA from tumor samples (Fig. 1C). Primer sequences are listed in Supplementary Table S2. PCR was performed using Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific) in 10-μL reaction volume containing 0.2 μM of each primer with the following reaction conditions: 94°C for 2 min, 40 cycles of 94°C for 15 s, 59°C for 30 s, and 68°C for 15 min. PCR amplicons obtained from the mapping PCR were ligated into the pCR4-TOPO vector using the TOPO TA Cloning Kit (Life Technologies) and sequenced at Retrogen (San Diego, California) using M13 forward (−20) (5ʹ-GTAAAACGACGGCCAG-3ʹ) and M13 reverse (5ʹ-CAGGAAACAGCTATGAC-3ʹ) primers. Obtained sequences were aligned to the EGFR gene using the University of California Santa Cruz Genome Browser (GRCh38/hg38 assembly). Detection of Patient-Specific EGFRvIII in Different Samples Patient-specific primers flanking the EGFRvIII breakpoints mapped in each patient were designed. The primer sequences and the annealing temperature for each primer set are listed in Supplementary Table S3. PCR was performed on tumor DNA, cfDNA from plasma and CSF, and DNA from extracellular vesicles in plasma using Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific) in 10-μL reaction volume containing 0.2 μM of each primer with the following reaction conditions: 94°C for 2 min, 40 cycles of 94°C for 15 s, each different annealing temperature for 30 s, and 68°C for 30 s. Whole Genome Sequencing Analyses Paired-end next-generation sequencing was performed for all the tumor samples on an Illumina Hi-Seq with 150 cycles at Novogene (Chula Vista, California). Sequence reads were then mapped to the GRCh38/hg38 genome using bwa-mem (https://arxiv.org/abs/1303.3997v2). Next, canonical alignment information was extracted from reads flagged with SA tags, which mark chimeric alignments (http://samtools.github.io/hts-specs/SAMtags.pdf). Genomic rearrangements were then predicted when multiple canonical alignments shared the same breakpoint coordinates. Based on alignments, breakpoints involving the EGFR locus were extracted and classified as EGFRvIII deletions if breaks spanned intron 1 and exon 7 or intron 7. The breakpoint sequences were obtained for 7 samples from our previous study7 in the same manner. The proportion of each EGFRvIII deletion among total EGFR copies was calculated based on the number of unique reads containing those junctions compared with the total number of unique reads mapped to the breakpoints, with or without chimeric alignments. Using the canonical alignment information extracted as above, interchromosomal and intrachromosomal rearrangements were analyzed. Breakpoints observed in more than 25 reads were plotted using Circos software (http://circos. ca/software/). Single Nucleotide Polymorphism Assays Total RNA was extracted from the tumor samples using the RNeasy Plus Mini Kit (Qiagen) and was reverse transcribed using RNA to cDNA EcoDry Premix (Clontech) according to the manufacturer’s instruction. RNA and cDNA were obtained from a glioma sphere cell line, GBM 39, which is known to have ecDNA harboring homogeneous amplicons of EGFRvIII, and was included in the analysis as a control. PCR was performed for the obtained cDNA using a forward primer within exon 1 and reverse primers within exons 15, 16, 20, and 25. The sequences of the primers were: exon 1 forward, 5ʹ-GGCTCTGGAGGAAAAGAAAG; exon 15 reverse, 5ʹ-GCAGTTTGGATGGCACAGGTG-3ʹ; exon 16 reverse, 5ʹ-TTCGTTGGACAGCCTTCAAG-3ʹ; exon 20 reverse, 5ʹ-GGCATGAGCTGCGTGATG-3ʹ; and exon 25 reverse, 5ʹ-TGCTGAAGAAGCCCTGCTGTG-3ʹ. Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific) was used and the reaction conditions were as follows: 94°C for 2 min, 35 cycles of 94°C for 15 s, 55°C for 30 s, and 68°C for 3 min. Single nucleotide polymorphism (SNP) sites rs2227983, rs17290169, s2227984, and rs1050171 in exons 13, 15, 16, and 20, respectively, of EGFR in the PCR amplicons derived from non-vIII EGFR and EGFRvIII were sequenced separately using 3 different primers: 5ʹ-TGTTTGGGACCTCCGGTCAG-3ʹ; 5ʹ-TGCCTCAGGCCATGAACATC-3ʹ; and 5ʹ-AGAAGCAACATC TCCGAAAG-3ʹ. These SNP sites were compared with sequence of DNA obtained from matched patients’ plasma. Results Screening and Mapping of EGFRvIII Breakpoints in Tumor Samples We analyzed 29 GBM specimens by genomic qPCR and found amplification (2.4- to 190-fold, average 50-fold) in 11 of these samples (38%; Fig. 1B). To identify EGFRvIII rearrangements, copy numbers of the 3 regions deleted in EGFRvIII were compared with those of the regions maintained in EGFRvIII. If any of the latter PCR amplicons yielded copy numbers significantly greater (P < 0.05) than any of the former ones, these samples were regarded as possibly containing EGFRvIII deletions. Eight of 29 samples displayed such differences (Fig. 1B). We then applied the arrayed PCR to the above 8 samples. Six of the 8 samples showed a stepwise pattern of PCR amplicons with approximately 2.7-kb incremental differences based on the spacing of neighboring primers (Fig. 1D). Two samples (BC010t and BC034t) harbored 2 different sets of these serial PCR amplicons. One set of the PCR amplicons was observed in 4 samples (BC039t, BC048t, BC051t, and BC083t), while amplicons were not obtained from 4 other samples (Fig. 1D). A forward primer that was designed to yield amplicons regardless of existence of EGFRvIII, e8f, yielded amplicons in all samples (Fig. 1D). In summary, 8 different EGFRvIII breakpoints in 6 samples were detected. All samples were subjected to whole genome sequencing (WGS) in parallel to validate the results derived from PCR-based mapping. All EGFRvIII breakpoints mapped by PCR were confirmed by WGS; no additional EGFRvIII breakpoints were identified by WGS (Supplementary Table S4). Thus, the sensitivity and specificity of our mapping procedure, as validated by WGS, were both 100% in detecting EGFRvIII, and by qPCR screening were 100% and 91%, respectively. The frequencies of the reads derived from each EGFRvIII breakpoint among all the reads from EGFR in each patient were 4.6%–76% (Supplementary Table S4). Additionally, WGS identified other EGFR rearrangements in these 29 samples, including a tandem duplication of chr7:55,173,011–55,173,064, which resulted in an in-frame 18 amino acid insertion in exon 17 (BC003t), and another case with EGFRvII(BC048t), a deletion encompassing exons 14 and 1515 (Supplementary Table S4). Detection of Patient-Specific EGFRvIII in Different Samples Primers designed for each specific sample generated amplicons of the expected size only in each corresponding tumor sample (Fig. 2A). To determine if such patient-specific EGFRvIII breakpoints could be detected in matched plasma and thus be used as a “liquid biopsy” platform,16EGFRvIII breakpoint PCR was performed using cfDNA17 or evDNA18,19 from plasma from 4 of the patients whose tumors were EGFRvIII positive (BC010, BC034, BC048, and BC083). Patient-specific EGFRvIII-derived PCR amplicons were not detected in 100 ng heparinase I treated or untreated plasma cfDNA or evDNA (Fig. 2B, C). However, patient-specific EGFRvIII amplicons were detected in cfDNA isolated from matched CSF (Fig. 2D). Fig. 2 View largeDownload slide Patient-specific EGFRvIII is detectable in CSF, but not in plasma. (A) Agarose gel electrophoresis of PCR products from tumor DNA generated with sets of patient-specific primers. Red arrowed amplicons indicate patient-specific PCR amplicons. (B) The results of patient-specific PCR on cfDNA from plasma. (C) Patient-specific PCR on tumor DNA (t) and evDNA from plasma (e). Red arrowed amplicons indicate patient-specific PCR amplicons. (D) The results of patient-specific PCR on CSF cfDNA (left). Red arrows indicate the patient-specific amplicons detected in CSF. The amplicon indicated by an asterisk is a nonspecific amplicon. The right panel shows the results of positive and negative control experiments. Fig. 2 View largeDownload slide Patient-specific EGFRvIII is detectable in CSF, but not in plasma. (A) Agarose gel electrophoresis of PCR products from tumor DNA generated with sets of patient-specific primers. Red arrowed amplicons indicate patient-specific PCR amplicons. (B) The results of patient-specific PCR on cfDNA from plasma. (C) Patient-specific PCR on tumor DNA (t) and evDNA from plasma (e). Red arrowed amplicons indicate patient-specific PCR amplicons. (D) The results of patient-specific PCR on CSF cfDNA (left). Red arrows indicate the patient-specific amplicons detected in CSF. The amplicon indicated by an asterisk is a nonspecific amplicon. The right panel shows the results of positive and negative control experiments. Detailed Breakpoint Analysis Reveals Different Genomic Rearrangements Among 15 EGFRvIII breakpoints in this analysis, 4 (BC010t-1, BC010t-2, BC034t-1, and GBM 6) had small insertions of up to 3 nucleotides (Fig. 3A, B). In one sample (HK359), an insertion of an approximately 42-kb sequence from outside of the EGFR locus was detected (Fig. 3A). Insertions of a duplicated 27-nucleotide sequence and a 72-nucleotide inverted sequence at breakpoint junctions accompanied by 6-nucleotide microhomologies were observed in BC010t-1 and BC039t, respectively (Fig. 3A, B). Nine breakpoint junctions (BC034t-2, BC048t, BC051t, BC083t, GBM 39, HK296, HK301, HK423, and UCSD002.26) were simple deletions without insertion (Fig. 3A, B). Among those 9 breakpoints, 3 (BC034t-2, BC083t, and HK296) showed microhomologies of 2 to 3 nucleotides at the junctions (Fig. 3B). Fig. 3 View largeDownload slide Detailed structures and sequences of mapped EGFRvIII breakpoints. (A) Blue arrows indicate intron 1 sequence for each breakpoint. Orange arrows are sequences downstream of EGFRvIII deletions. Black outlined boxes and green arrows show insertions at the junctions. Characters in the black outlined boxes are the inserted nucleotides, and the numbers in the arrows are the coordinates on chromosome 7 in the GRCh38/hg38 assembly. (B) Sequences at the junctions together with original sequences corresponding to upstream and the downstream sequence. Blue, green, and orange characters are the sequences corresponding to the same colored arrows in (A). The sequences highlighted in yellow show microhomology sequences of more than one nucleotide shared among breakpoint junctions and their original sequences. Suggestive DNA repair mechanisms involved in each breakpoint are listed in the right column. Abbreviations: FoSTeS, fork stalling and template switching; MMBIR, microhomology-mediated break-induced repair; MMEJ, microhomology-mediated end joining. Fig. 3 View largeDownload slide Detailed structures and sequences of mapped EGFRvIII breakpoints. (A) Blue arrows indicate intron 1 sequence for each breakpoint. Orange arrows are sequences downstream of EGFRvIII deletions. Black outlined boxes and green arrows show insertions at the junctions. Characters in the black outlined boxes are the inserted nucleotides, and the numbers in the arrows are the coordinates on chromosome 7 in the GRCh38/hg38 assembly. (B) Sequences at the junctions together with original sequences corresponding to upstream and the downstream sequence. Blue, green, and orange characters are the sequences corresponding to the same colored arrows in (A). The sequences highlighted in yellow show microhomology sequences of more than one nucleotide shared among breakpoint junctions and their original sequences. Suggestive DNA repair mechanisms involved in each breakpoint are listed in the right column. Abbreviations: FoSTeS, fork stalling and template switching; MMBIR, microhomology-mediated break-induced repair; MMEJ, microhomology-mediated end joining. Complex Chromosomal Rearrangements Observed in Tumor Samples Twelve samples had chromosomal rearrangements involving chromosome 7 (Fig. 4A). Ten out of 11 samples with EGFR amplification and all of the EGFRvIII-positive samples showed some rearrangements involving chromosome 7, while there was no EGFR amplification in 16 of 17 samples that did not show any chromosome 7 rearrangements (Fig. 4B). One sample with EGFR amplification without chromosome 7 rearrangements (BC068) had the lowest relative copy number of EGFR (2.4), suggesting either that there was no actual EGFR focal amplification or that chromosomal rearrangements were not detected due to a small fraction of the cells with EGFR amplification in this specimen. Fig. 4 View largeDownload slide EGFR amplification is associated with chromosome 7 rearrangements. (A) Circos plots of the 29 GBM samples. Samples of red characters are the samples with EGFR amplification and EGFRvIII. Samples of orange characters are the samples with EGFR amplification without EGFRvIII. Chromosome 7 and EGFR locus are zoomed in as shown in the plot of BC002. Red lines indicate intrachromosomal or interchromosomal connections involving chromosome 7. Other connections are drawn by black lines. (B) Association between chromosome 7 rearrangements and EGFR amplification or EGFRvIII positivity. P-value is based on Fisher’s exact test. Fig. 4 View largeDownload slide EGFR amplification is associated with chromosome 7 rearrangements. (A) Circos plots of the 29 GBM samples. Samples of red characters are the samples with EGFR amplification and EGFRvIII. Samples of orange characters are the samples with EGFR amplification without EGFRvIII. Chromosome 7 and EGFR locus are zoomed in as shown in the plot of BC002. Red lines indicate intrachromosomal or interchromosomal connections involving chromosome 7. Other connections are drawn by black lines. (B) Association between chromosome 7 rearrangements and EGFR amplification or EGFRvIII positivity. P-value is based on Fisher’s exact test. SNP Analysis Indicates EGFRvIII Is Derived from Amplified Non-vIII EGFR RNA from 2 patient samples (BC010t and BC034t) that had 2 different breakpoints (Fig. 1D) was further analyzed to identify which allele the EGFRvIII was derived from. Two different-sized amplicons from non-vIII EGFR and EGFRvIII were obtained by reverse transcription PCR with a forward primer in exon 1 and a reverse primer in exon 20 in the tumors (Fig. 5A). Plasma DNA from these patients showed nucleotide differences between 2 alleles at 2 SNP sites (rs2227983 and rs2227984) (Fig. 5B, C); however, these sites in cDNA derived from non-vIII EGFR, and EGFRvIII showed homogeneous nucleotide compositions in both patients (Fig. 5B, C). Assuming the cDNA from non-amplified or the normal allele of EGFR is undetectable by sequencing due to being overwhelmed by amplified non-vIII EGFR in these samples (Fig. 1B), these results suggest that EGFRvIII, with 2 different breakpoints in each patient, was derived from amplified full-length EGFR, as both species of EGFR shared the same allele-determining SNP. Fig. 5 View largeDownload slide EGFRvIII is derived from the same allele as amplified non-vIII EGFR. (A) Agarose gel electrophoresis of reverse transcription PCR amplicons of EGFR transcripts from different samples. Red arrowed samples indicate specific samples of either EGFRvIII or non-vIII EGFR transcripts. (B) The nucleotide compositions at each SNP site in each sample and (C) the original chromatograms. Fig. 5 View largeDownload slide EGFRvIII is derived from the same allele as amplified non-vIII EGFR. (A) Agarose gel electrophoresis of reverse transcription PCR amplicons of EGFR transcripts from different samples. Red arrowed samples indicate specific samples of either EGFRvIII or non-vIII EGFR transcripts. (B) The nucleotide compositions at each SNP site in each sample and (C) the original chromatograms. Discussion Here we present simplified methods to screen tumor samples for the presence of EGFR amplification and EGFRvIII. The screening by qPCR targeting several loci in EGFR was sensitive and specific to detect EGFRvIII. Though the specificity was 91% in our analysis of 29 tumor samples, if the analysis was limited to only samples with EGFR amplification, with an assumption that EGFRvIII is exclusively associated with amplification of the gene,20 our specificity reached 100%. In our analysis, 11 of 29 samples (38%) had EGFR amplification, and 6 (55%) of those with the amplification had EGFRvIII, which is consistent with previous reports.20,21 Although EGFRvIII-targeted therapeutics have not yet proven to be efficacious through phase III clinical trials for GBM patients,22 and therefore clinical relevance of detecting EGFRvIII is limited for now, our method would be a rapid and sensitive way to detect EGFRvIII in tumor DNA and would have potential for diagnostic application if therapeutic challenges, such as drug delivery and intrinsic resistance of tumor cells, are overcome in the future.23 Mapping of EGFRvIII has been performed with similar approaches as described here, although the sensitivity and specificity of these methods have not been evaluated.8,9 In our experience from a previous study,9 it is often difficult to distinguish specific EGFRvIII breakpoint amplicons from nonspecific PCR products. Thus, using a large number of intron 1 primers spaced equidistant from each other, our goal was to reduce the risk of a false negative result and to distinguish specific and nonspecific amplicons easily by determining if a stepwise pattern of PCR amplicons derived from the placement of neighboring primers is present. Indeed, our mapping method combined with the qPCR screening in the current study had 100% sensitivity and specificity for detecting EGFRvIII breakpoints that were found by WGS analysis. The sensitivity for detection of EGFRvIII breakpoints by this PCR approach was high enough to detect a rare population of EGFRvIII, as in the case of BC010t-2, which accounted for only 4.6% of total EGFR reads in WGS. As for the locations of the breakpoints, a previous report suggesting involvement of Alu repeat sequences in EGFRvIII rearrangement shows that all the intron 7 breakpoints are located in or downstream of the intron 7 Alu sequence.8 On the other hand, some of the samples analyzed in the current study and in our previous work9 had intron 7 breakpoints upstream of this region, suggesting involvement of other mechanisms. In one tumor (BC010t-1), we detected the 3ʹ EGFRvIII breakpoint in the middle of exon 7. In this case, it was assumed that the deletion of the splice acceptor site in intron 6 probably resulted in skipping of exon 7 during splicing, thus generating EGFRvIII. In terms of intratumoral heterogeneity, there were 2 cases with 2 different EGFRvIII breakpoints, which is in agreement with previous studies showing breakpoint heterogeneity within a single tumor.24,25 Assuming that these 2 different breakpoints were derived from separate regions from the same tumor, this raises the possibility that further varieties of EGFRvIII break structures exist in other portions of a large tumor mass. Furthermore, discrepancy between the results of tumor DNA and RNA in those patients whose RNA data were available from our previous study26 (Supplementary Table S5) might have resulted from sampling bias of the tumors where EGFRvIII-positive regions are scattered within each tumor as described previously.25 Considering such heterogeneity within each tumor, multiple sampling of tumor regions coupled with comprehensive immunohistochemistry and DNA and RNA analyses for EGFRvIII would be necessary for informative diagnostic information. Our mapping results enabled patient-specific PCR that yielded EGFRvIII-derived PCR amplicons specific to each patient’s tumor, and for a single case where a CSF sample was available, this approach detected the same patient-specific amplicon from cfDNA. Diagnosis using cfDNA from liquid biopsies, as a surrogate for tissue biopsy, has been used for cancer patients to monitor treatment responses and tumor relapses.16 In the setting of a brain tumor, cfDNA is difficult to detect in plasma DNA due to the blood–brain barrier,27 which is consistent with our results where EGFRvIII-derived PCR amplicons were not obtained from either cfDNA or evDNA derived from plasma, even with modifications to enhance sensitivity, such as preamplification13 and heparinase I treatment.14 In contrast, a previous study analyzing patients with brain tumors, including 8 GBMs, shows that tumor DNA was detectable in CSF,28 as in one of the cases in the current study, suggesting that this source of cfDNA provides a means to detect EGFRvIII mutations based on a tumor-defined, patient-specific breakpoint fingerprint. Several approaches for liquid biopsy of GBM have been attempted.29 These include detection of EGFRvIII in evRNA in plasma and CSF, with detection sensitivity ranging from 36% to 61%.26,30,31 The benefit of a DNA assay would be that it could track EGFRvIII breakpoints specific to each tumor, as each breakpoint presents a unique junctional fingerprint. In contrast, an RNA analysis cannot discriminate different populations of EGFRvIII with different genomic breakpoints, since the junction sequence between exons 1 and 8 in EGFRvIII RNA is constant among different samples. On the other hand, the commonality in RNA breakpoints might enable the detection of EGFRvIII in evRNA from plasma.31 Although there was only one patient for whom both results of plasma cfDNA in the current study and evRNA from our previous study26 were available (Supplementary Table S5) and further studies to compare the sensitivity of DNA and RNA detection in biofluid are necessary, we expect that the approach of screening and mapping genomic EGFRvIII breakpoints presented here could be integrated with other diagnostic tools in the future. Furthermore, with the advent of CRISPR/Cas9 gene-editing technology, mapped EGFRvIII junctions could serve as loci for insertion of a suicide gene, such as herpes simplex virus thymidine kinase.32 Such an approach could be a highly specific method to target EGFRvIII-positive tumor cells for which the breakpoints were accurately mapped. By sequencing EGFRvIII breakpoints, we found different types of rearrangements and some with sequences inserted between intron 1 and exon 7 or intron 7 sequences. Such insertions could be large, as in HK 359, where there was an insertion of approximately 42 kb. This insertion scenario would be impossible to detect using our mapping PCR strategy due to length of the inserted sequence. Our mapping PCR would be able to generate amplicons at the breakpoints with relatively short insertions, presumably up to a few kilobases. Thus, the size of an insertion is one of the limitations encountered with our mapping method. Twelve of the 15 breakpoints (80%) analyzed had either short insertions up to 3 nucleotides or simple deletions of intron 1 to exon 7 sequences, suggestive of non-homologous end joining (NHEJ) as a mechanism for repair.33 Three (20%) showed microhomology of 2 to 3 nucleotides between their original upstream and downstream sequences, suggestive of NHEJ or microhomology-mediated end joining based on the length of the homologies.34,35 A duplication or an inversion found in another 2 cases (13%), accompanied by microhomologies, may suggest involvement of replication mechanisms such as fork stalling and template switching or microhomology-mediated break-induced replication36,37 (Fig. 3B). Together, these varieties of repair mechanisms play a role in the biogenesis of EGFRvIII rearrangement. One of the EGFR-amplified samples was accompanied by extensive rearrangement of chromosome 7 (BC020), suggesting potential involvement of chromothripsis,38,39 which is associated with generation of double minutes or ecDNA,39–41 where amplified oncogenes in cancer cells, including EGFR in GBM, mostly reside.7,42 All 6 cases with EGFRvIII in our analysis had amplification of both EGFRvIII and non-vIII EGFR and these amplicons are likely to be localized to ecDNA.7 It has been previously reported that ecDNA is prone to acquire additional mutations, probably due to its localization in micronuclei.43,44 The commonality in the alleles of origin of all EGFRvIII and amplified non-vIII EGFR may suggest that EGFRvIII is derived through further rearrangements of amplified full-length EGFR on ecDNA, which is also supported by previous studies.24,25 In conclusion, detailed analyses of EGFRvIII deletions efficiently detected by our PCR approach gives us implications as to how this aggressive mutant arises in GBM. We expect such findings will lead to a better understanding of how clonal evolution and the heterogeneity observed in GBM25,45,46 arise. Supplementary material Supplementary material is available at Neuro-Oncology online. Funding This research was supported by National Institutes of Health grants (R01 NS080939 to F.B.F., 2P01CA069246-20 and UH3TR000931 to B.S.C., and 1RO1NS097649-01 to C.C.C.). B.R. received funding support from Ludwig Cancer Research. C.C.C. also received support from the Doris Duke Charitable Foundation Clinical Scientist Development Award, The Sontag Foundation Distinguished Scientist Award, the Kimmel Scholar Award, and BWF 1006774.01. Acknowledgments F.B.F. and B.S.C. conceived the study. B.S.C. and J.M.F. collected the samples. T.K. and F.B.F. designed and T.K. conducted the experiments. B.L. and B.R. assisted with the bioinformatic analyses. F.B.F., B.S.C., C.C.C., J.M.F., and T.K. interpreted the data. T.K., F.B.F., and C.C.C. wrote the manuscript. Conflict of interest statement. The authors declare no conflicts of interests. References 1. Cloughesy TF , Cavenee WK , Mischel PS . Glioblastoma: from molecular pathology to targeted treatment . Annu Rev Pathol . 2014 ; 9 : 1 – 25 . Google Scholar CrossRef Search ADS PubMed 2. Huang HS , Nagane M , Klingbeil CK et al. The enhanced tumorigenic activity of a mutant epidermal growth factor receptor common in human cancers is mediated by threshold levels of constitutive tyrosine phosphorylation and unattenuated signaling . J Biol Chem . 1997 ; 272 ( 5 ): 2927 – 2935 . Google Scholar CrossRef Search ADS PubMed 3. Nagane M , Levitzki A , Gazit A , Cavenee WK , Huang HJ . Drug resistance of human glioblastoma cells conferred by a tumor-specific mutant epidermal growth factor receptor through modulation of Bcl-XL and caspase-3-like proteases . Proc Natl Acad Sci U S A . 1998 ; 95 ( 10 ): 5724 – 5729 . Google Scholar CrossRef Search ADS PubMed 4. Nishikawa R , Ji XD , Harmon RC et al. A mutant epidermal growth factor receptor common in human glioma confers enhanced tumorigenicity . Proc Natl Acad Sci U S A . 1994 ; 91 ( 16 ): 7727 – 7731 . Google Scholar CrossRef Search ADS PubMed 5. Wong AJ , Ruppert JM , Bigner SH et al. Structural alterations of the epidermal growth factor receptor gene in human gliomas . Proc Natl Acad Sci U S A . 1992 ; 89 ( 7 ): 2965 – 2969 . Google Scholar CrossRef Search ADS PubMed 6. Vogt N , Lefèvre SH , Apiou F et al. Molecular structure of double-minute chromosomes bearing amplified copies of the epidermal growth factor receptor gene in gliomas . Proc Natl Acad Sci U S A . 2004 ; 101 ( 31 ): 11368 – 11373 . Google Scholar CrossRef Search ADS PubMed 7. Turner KM , Deshpande V , Beyter D et al. Extrachromosomal oncogene amplification drives tumour evolution and genetic heterogeneity . Nature . 2017 ; 543 ( 7643 ): 122 – 125 . Google Scholar CrossRef Search ADS PubMed 8. Frederick L , Eley G , Wang XY , James CD . Analysis of genomic rearrangements associated with EGRFvIII expression suggests involvement of Alu repeat elements . Neuro Oncol . 2000 ; 2 ( 3 ): 159 – 163 . Google Scholar CrossRef Search ADS PubMed 9. Nathanson DA , Gini B , Mottahedeh J et al. Targeted therapy resistance mediated by dynamic regulation of extrachromosomal mutant EGFR DNA . Science . 2014 ; 343 ( 6166 ): 72 – 76 . Google Scholar CrossRef Search ADS PubMed 10. Kamps R , Brandao RD , Bosch BJ et al. Next-generation sequencing in oncology: genetic diagnosis, risk prediction and cancer classification . Int J Mol Sci . 2017 ; 18 ( 2 ):doi: 10.3390/ijms18020308 . 11. Gupta P , Han SY , Holgado-Madruga M et al. Development of an EGFRvIII specific recombinant antibody . BMC Biotechnol . 2010 ; 10 : 72 . Google Scholar CrossRef Search ADS PubMed 12. Li J , Taich ZJ , Goyal A et al. Epigenetic suppression of EGFR signaling in G-CIMP+ glioblastomas . Oncotarget . 2014 ; 5 ( 17 ): 7342 – 7356 . Google Scholar PubMed 13. Li J , Harris L , Mamon H et al. Whole genome amplification of plasma-circulating DNA enables expanded screening for allelic imbalance in plasma . J Mol Diagn . 2006 ; 8 ( 1 ): 22 – 30 . Google Scholar CrossRef Search ADS PubMed 14. Kondratov K , Kurapeev D , Popov M et al. Heparinase treatment of heparin-contaminated plasma from coronary artery bypass grafting patients enables reliable quantification of microRNAs . Biomol Detect Quantif . 2016 ; 8 : 9 – 14 . Google Scholar CrossRef Search ADS PubMed 15. Frederick L , Wang XY , Eley G , James CD . Diversity and frequency of epidermal growth factor receptor mutations in human glioblastomas . Cancer Res . 2000 ; 60 ( 5 ): 1383 – 1387 . Google Scholar PubMed 16. Crowley E , Di Nicolantonio F , Loupakis F , Bardelli A . Liquid biopsy: monitoring cancer-genetics in the blood . Nat Rev Clin Oncol . 2013 ; 10 ( 8 ): 472 – 484 . Google Scholar CrossRef Search ADS PubMed 17. Gormally E , Caboux E , Vineis P , Hainaut P . Circulating free DNA in plasma or serum as biomarker of carcinogenesis: practical aspects and biological significance . Mutat Res . 2007 ; 635 ( 2–3 ): 105 – 117 . Google Scholar CrossRef Search ADS PubMed 18. Akers JC , Gonda D , Kim R , Carter BS , Chen CC . Biogenesis of extracellular vesicles (EV): exosomes, microvesicles, retrovirus-like vesicles, and apoptotic bodies . J Neurooncol . 2013 ; 113 ( 1 ): 1 – 11 . Google Scholar CrossRef Search ADS PubMed 19. Akers JC , Hua W , Li H et al. A cerebrospinal fluid microRNA signature as biomarker for glioblastoma . Oncotarget . 2017 ; 8 ( 40 ): 68769 – 68779 . Google Scholar CrossRef Search ADS PubMed 20. Aldape KD , Ballman K , Furth A et al. Immunohistochemical detection of EGFRvIII in high malignancy grade astrocytomas and evaluation of prognostic significance . J Neuropathol Exp Neurol . 2004 ; 63 ( 7 ): 700 – 707 . Google Scholar CrossRef Search ADS PubMed 21. Ohgaki H , Dessen P , Jourde B et al. Genetic pathways to glioblastoma: a population-based study . Cancer Res . 2004 ; 64 ( 19 ): 6892 – 6899 . Google Scholar CrossRef Search ADS PubMed 22. Taylor TE , Furnari FB , Cavenee WK . Targeting EGFR for treatment of glioblastoma: molecular basis to overcome resistance . Curr Cancer Drug Targets . 2012 ; 12 ( 3 ): 197 – 209 . Google Scholar CrossRef Search ADS PubMed 23. Thorne AH , Zanca C , Furnari F . Epidermal growth factor receptor targeting and challenges in glioblastoma . Neuro Oncol . 2016 ; 18 ( 7 ): 914 – 918 . Google Scholar CrossRef Search ADS PubMed 24. Francis JM , Zhang CZ , Maire CL et al. EGFR variant heterogeneity in glioblastoma resolved through single-nucleus sequencing . Cancer Discov . 2014 ; 4 ( 8 ): 956 – 971 . Google Scholar CrossRef Search ADS PubMed 25. Eskilsson E , Rosland GV , Talasila KM et al. EGFRvIII mutations can emerge as late and heterogenous events in glioblastoma development and promote angiogenesis through Src activation . Neuro Oncol . 2016 ; 18 ( 12 ): 1644 – 1655 . Google Scholar CrossRef Search ADS PubMed 26. Figueroa JM , Skog J , Akers J et al. Detection of wild-type EGFR amplification and EGFRvIII mutation in CSF-derived extracellular vesicles of glioblastoma patients . Neuro Oncol . 2017 ; 19 ( 11 ): 1494 – 1502 . Google Scholar CrossRef Search ADS PubMed 27. De Mattos-Arruda L , Mayor R , Ng CK et al. Cerebrospinal fluid-derived circulating tumour DNA better represents the genomic alterations of brain tumours than plasma . Nat Commun . 2015 ; 6 : 8839 . Google Scholar CrossRef Search ADS PubMed 28. Wang Y , Springer S , Zhang M et al. Detection of tumor-derived DNA in cerebrospinal fluid of patients with primary tumors of the brain and spinal cord . Proc Natl Acad Sci U S A . 2015 ; 112 ( 31 ): 9704 – 9709 . Google Scholar CrossRef Search ADS PubMed 29. Santiago-Dieppa DR , Steinberg J , Gonda D , Cheung VJ , Carter BS , Chen CC . Extracellular vesicles as a platform for ‘liquid biopsy’ in glioblastoma patients . Expert Rev Mol Diagn . 2014 ; 14 ( 7 ): 819 – 825 . Google Scholar CrossRef Search ADS PubMed 30. Kros JM , Mustafa DM , Dekker LJ , Sillevis Smitt PA , Luider TM , Zheng PP . Circulating glioma biomarkers . Neuro Oncol . 2015 ; 17 ( 3 ): 343 – 360 . Google Scholar PubMed 31. Skog J , Würdinger T , van Rijn S et al. Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide diagnostic biomarkers . Nat Cell Biol . 2008 ; 10 ( 12 ): 1470 – 1476 . Google Scholar CrossRef Search ADS PubMed 32. Chen ZH , Yu YP , Zuo ZH et al. Targeting genomic rearrangements in tumor cells through Cas9-mediated insertion of a suicide gene . Nat Biotechnol . 2017 ; 35 ( 6 ): 543 – 550 . Google Scholar CrossRef Search ADS PubMed 33. Moore JK , Haber JE . Cell cycle and genetic requirements of two pathways of nonhomologous end-joining repair of double-strand breaks in Saccharomyces cerevisiae . Mol Cell Biol . 1996 ; 16 ( 5 ): 2164 – 2173 . Google Scholar CrossRef Search ADS PubMed 34. Chiruvella KK , Liang Z , Wilson TE . Repair of double-strand breaks by end joining . Cold Spring Harb Perspect Biol . 2013 ; 5 ( 5 ): a012757 . Google Scholar CrossRef Search ADS PubMed 35. Sfeir A , Symington LS . Microhomology-mediated end joining: a back-up survival mechanism or dedicated pathway ? Trends Biochem Sci . 2015 ; 40 ( 11 ): 701 – 714 . Google Scholar CrossRef Search ADS PubMed 36. Liu P , Erez A , Nagamani SC et al. Chromosome catastrophes involve replication mechanisms generating complex genomic rearrangements . Cell . 2011 ; 146 ( 6 ): 889 – 903 . Google Scholar CrossRef Search ADS PubMed 37. Zhang F , Khajavi M , Connolly AM , Towne CF , Batish SD , Lupski JR . The DNA replication FoSTeS/MMBIR mechanism can generate genomic, genic and exonic complex rearrangements in humans . Nat Genet . 2009 ; 41 ( 7 ): 849 – 853 . Google Scholar CrossRef Search ADS PubMed 38. Maher CA , Wilson RK . Chromothripsis and human disease: piecing together the shattering process . Cell . 2012 ; 148 ( 1–2 ): 29 – 32 . Google Scholar CrossRef Search ADS PubMed 39. Stephens PJ , Greenman CD , Fu B et al. Massive genomic rearrangement acquired in a single catastrophic event during cancer development . Cell . 2011 ; 144 ( 1 ): 27 – 40 . Google Scholar CrossRef Search ADS PubMed 40. Zhang CZ , Spektor A , Cornils H et al. Chromothripsis from DNA damage in micronuclei . Nature . 2015 ; 522 ( 7555 ): 179 – 184 . Google Scholar CrossRef Search ADS PubMed 41. Holland AJ , Cleveland DW . Chromoanagenesis and cancer: mechanisms and consequences of localized, complex chromosomal rearrangements . Nat Med . 2012 ; 18 ( 11 ): 1630 – 1638 . Google Scholar CrossRef Search ADS PubMed 42. Collins VP . Gene amplification in human gliomas . Glia . 1995 ; 15 ( 3 ): 289 – 296 . Google Scholar CrossRef Search ADS PubMed 43. Hatch EM , Hetzer MW . Linking micronuclei to chromosome fragmentation . Cell . 2015 ; 161 ( 7 ): 1502 – 1504 . Google Scholar CrossRef Search ADS PubMed 44. Shimizu N , Kanda T , Wahl GM . Selective capture of acentric fragments by micronuclei provides a rapid method for purifying extrachromosomally amplified DNA . Nat Genet . 1996 ; 12 ( 1 ): 65 – 71 . Google Scholar CrossRef Search ADS PubMed 45. Abou-El-Ardat K , Seifert M , Becker K et al. Comprehensive molecular characterization of multifocal glioblastoma proves its monoclonal origin and reveals novel insights into clonal evolution and heterogeneity of glioblastomas . Neuro Oncol . 2017 ; 19 ( 4 ): 546 – 557 . Google Scholar CrossRef Search ADS PubMed 46. Zanca C , Villa GR , Benitez JA et al. Glioblastoma cellular cross-talk converges on NF-kappaB to attenuate EGFR inhibitor sensitivity . Genes Dev . 2017 ;doi: 10.1101/gad.300079.117 . © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Neuro-Oncology – Oxford University Press
Published: Apr 12, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera