|Home | About | Journals | Submit | Contact Us | Français|
Activating mutations in epidermal growth factor receptor-1 (EGFR) are found in 10–15% of Caucasian patients with non–small cell lung carcinoma (NSCLC). Approximately 90% of the mutations are deletions of several amino acids in exon 19 or point mutations in exon 21. Some studies suggest that these mutations identify patients that might benefit from targeted EGFR inhibitor therapy. DNA melting analysis of polymerase chain reaction products can screen for these mutations to identify this patient population. However, amplicon DNA melting analysis, although easily capable of detecting heterozygous mutations by heterodimer formation, becomes more difficult if mutations are homozygous or if the mutant allele is selectively amplified over wild type. Amplification of EGFR is common in NSCLC and this could compromise mutation detection by amplicon melting analysis. To overcome this potential limitation, we developed unlabeled, single-stranded DNA probes, complimentary to EGFR exon 19 and exon 21 where the common activating mutations occur. The unlabeled probes are incorporated into a standard polymerase chain reaction during the amplification of EGFR exons 19 and 21. The probe melting peak is easily distinguished from the amplicon melting peak, and probe melting is altered if mutations are present. This allows for easy identification of activating mutations even in homozygous or amplified states and is useful in the screening of NSCLC for the common EGFR activating mutations.
Tyrosine kinases are important cell-signaling molecules. Recently, activating mutations in tyrosine kinase genes have been discovered in several human malignancies.1 These mutations are somatic and the resultant proteins show constitutive activation. This results in uncontrolled cell-signaling and tumor growth. Because the mutations are somatic, the activated tyrosine kinases are unique to the tumor cell and represent novel tumor-specific drug targets.2 Examples include the activating mutations recently reported in the epidermal growth factor receptor-1 (EGFR) in a group of patients with non–small cell lung carcinoma (NSCLC).3 These mutations are in the active site of EGFR and result in a protein that is catalytically more active than the wild type.4 They are most common in tumors which show some bronchioloalveolar (BAC) histology. The patients tend to be women and nonsmokers. The frequency of these mutations in NSCLC patients in the United States is about 15% but is much higher in patients of Asian descent.5 Importantly, the activated EGFR proteins are more susceptible to small-molecule active site inhibitors, such as gefitinib and erlotinib, than to the wild-type EGFR. This suggests that the presence of these mutations in NSCLC might define a population of patients who could effectively be treated with EGFR active site inhibitors.6,7
Studies from early phase II clinical trials suggested that patients whose tumors harbored EGFR mutations were more likely to respond to EGFR inhibition than those patients whose tumors did not have mutations. Unfortunately these initially exciting results were not duplicated in larger phase III trials and it now appears that other factors, such as EGFR gene amplification, may also influence the response of NSCLC to EGFR inhibition.8 Remarkably, although EGFR mutations were reported in NSCLC over 3 years ago, there are still no guidelines available on how to use mutation testing in the clinic.9 To provide some guidance on the appropriate use of EGFR mutations testing, future prospective clinical trials are clearly needed in which a concerted effort is made to obtain tissue for both EGFR mutation testing and EGFR gene amplification.
EGFR gene amplification can be measured by fluorescence in situ hybridization (FISH).10 There are a variety of methods that can be used to test for EGFR mutations including direct DNA sequencing. However, since the majority of NSCLC cases will be wild-type with respect to EGFR, the ideal mutation testing method would be a scanning technology that will allow for the rapid elimination of all wild-type cases. We have recently reported that DNA melting analysis with high resolution techniques can provide this information. We used this technique to scan the entire coding sequence of exons 18, 19, 20, and 21 of EGFR and exons 19 and 20 of human epidermal growth factor receptor-2 (HER2) where mutations had been reported.11 However, it now appears that over 90% of the clinically important EGFR mutations are either a small deletion in exon 19 or a point mutation in exon 21.12,13 The restricted location of these mutations suggests that DNA melting analysis to detect them can be improved with the use of unlabeled probes spanning these areas. The use of unlabeled DNA probes improves the detection of mutations in which the tumor is homozygous for the mutant allele or in cases where the mutant gene is amplified with subsequent dilution of the normal.14 In this report we show that the use of unlabeled DNA probes to detect EGFR mutations is inexpensive, rapid, and suitable for large clinical trials where EGFR mutation scanning will be done.
These materials were obtained as described previously.11 In addition, DNA primers were obtained from Operon (Huntsville, Alabama). Unlabeled DNA probes with a 3′ amino modifier block were purchased from Integrated DNA Technologies (Coralville, IA). Primers and probes were desalted with no purification.
Cell lines NCI-H1975 (CRL-5908) and NCI-H1650 (CRL-5883) were obtained from American Type Culture Collection (Manassas, VA). Cell line NCI-H1650 has a deletion of amino acid residues 746–750 (del ELREA) in EGFR exon 19. To confirm this, DNA was isolated from the cells and subjected to direct DNA sequencing. The DNA sequencing electropherograms for the CRL5883 cell line indicated a del746–750 and showed that the deletion composed about 65% of the electropherogram peak, the rest being wild type (not shown). This observation suggests that the cells were not simply heterozygous for the mutant allele but perhaps contained an extra copy of the mutant gene. Fluorescence in situ hybridization to detect EGFR gene amplification showed low trisomy (not shown), suggesting an extra copy of the mutant gene was present in the CRL5883 cells. These results would explain the mutant to wild-type allele ratio of 2:1 observed in the DNA electropherogram.
Cell line NCI-H1975 has an EGFR mutation at amino acid residue 858 (L858R) in exon 21. This was confirmed by isolating DNA from the cells and performing DNA sequencing. The sequencing results indicated an L858R mutation and the mutant allele composed about 65% of the DNA electropherogram (not shown). Fluorescence in situ hybridization to detect EGFR amplification indicated low trisomy, so like the CRL5883 cells, this cell line appears to contain an extra copy of the mutant allele.
Human tissue containing EGFR mutation-positive lung cancer was obtained as described previously.11 One case, not previously described, was used in this study. This case was an EGFR deletion mutant from a primary lung adenocarcinoma in a 59-year-old woman. DNA sequencing of DNA isolated from the tumor showed a del747–752 P753S mutation in exon 19. Only the mutant allele was detected in the electropherogram, suggesting the tumor was homozygous for the mutation. Fluorescence in situ hybridization to detect EGFR revealed only low trisomy. A case described previously (case 1 in reference 11) had an exon 21 EGFR mutation and was from a primary lung carcinoma in a 73-year-old woman. DNA sequencing of DNA isolated from the tumor showed that the mutant allele made up about 50% of the electropherogram peak (not shown) suggesting the possibility of a heterozygous mutation.
To isolate DNA from the cell lines, the cells were pelleted using standard cytology methods. The cell pellet was placed in a cassette in 10% neutral buffered formalin and paraffin embedded. An unstained slide of the cells was made from the paraffin block. DNA was isolated from the cells on the unstained slide as previously described for the isolation of DNA from formalin fixed, paraffin embedded human tissue.11 Briefly, the cells from the unstained slide were scraped into in a Tris buffer containing Tween 20 and proteinase K. After boiling for 10 min to inactivate the proteinase K, the digest was used directly for polymerase chain reaction (PCR).
Unlabeled DNA probes spanning EGFR exons 19 and 21 where the majority of the EGFR activating mutations in NSCLC occur were designed with the use of Primer Designer Software (Scientific and Education Software, Durham, NC). The unlabeled probe for EGFR exon 19 is the reverse complement to the forward strand spanning nucleotide positions 155721 to 155751 in the EGFR gene (Genebank accession number NC-000007) and is 31 bps in length. The predicted Tm of the probe is 66.39ºC. The unlabeled probe for EGFR exon 21 is reverse complement to the forward strand, spanning nucleotide positions 172778 to 172803 in the EGFR gene and is 26 bps in length. The predicted Tm of the probe is 70.63ºC. The sequences of the unlabeled probes are shown in Table 1.
Asymmetric PCR is necessary to maximize the amount of the DNA strand complementary to the unlabeled probe. Asymmetric (10:1) PCR was carried out in a total volume of 10 μL in a LightCycler (Roche Diagnostics, Indianapolis, IN) capillary cuvette. Exons 19 and 21 were evaluated separately in independent reactions. The reaction mixture consisted of all provided components of LightCycler FastStart DNA Master Hybridization Probes kit (Roche Diagnostics, Indianapolis, IN) and reactions were run in accord with the manufacture’s instructions. In asymmetric PCR, the forward primer is in excess of the reverse primer. All reactions therefore were run with 0.5 μM of the forward primer and 0.05 μM of the reverse primer. The sequences of the primers have been described.11 They are included in Table 1 for reference. Unlabeled probes were included at the beginning of PCR and were at 0.5 μM. The unlabeled probes are blocked at the 3′ to prevent extension during PCR. All reactions contained 1 unit of uracil N-glycosylase (AmpErase) (Applied Biosystems, Foster City, CA) and the dye LC Green Plus (Idaho Technology, Salt Lake City, UT). Magnesium ion concentration was 4 mM. PCR was carried out on the LightCycler (Roche Diagnostics, Indianapolis, IN) with a preincubation step for 10 min at 95°C (to denature the uracil glycosylase and activate the FastStart Taq DNA polymerase). Cycling conditions consisted of 55 cycles of denaturing at 94°C for 10 sec, annealing at 60°C for 10 sec, and extension at 74°C for 10 sec. Transition rates were 20°C/sec from denaturation to annealing, 1°C/sec annealing to extension, and 20°C/sec extension to denaturation.
After PCR, samples were momentarily heated to 94°C and cooled to 40°C. Samples were subjected to melting analysis with high resolution in the HR-1 (Idaho Technology, Salt Lake City, UT) as described previously.11 Melting curves were acquired by heating the sample from 55°C to 95°C at a rate of 0.3°C/sec. The melting data were directly converted to a derivative plot (−dF/dT vs. temperature) with the HR-1 software.
DNA sequencing was performed by the University of Utah Scientific Core Facilities using automated fluorescence sequencing (ABI3700 PRISM BigDye Terminator v3.1 cycle sequencing kit and sequencher version 4.0 (Applied Biosystems). Sequencing reactions used 3.2 pM of forward and reverse primer and approximately 60 ng amplified sample.
Cell line CRL5883 contains an EGFR exon 19 deletion (del 746–750) and cell line CRL5908 contains an EGFR exon 21 mutation (L858R). Two patients with NSCLC were also studied. One NSCLC was from a 59-year-old woman and contained the EGFR exon 19 deletion of amino acid residues 747–752 and a change of proline to serine at amino acid residue position 753 (del 747–752; P753S). The other NSCLC was from a 72-year-old woman and contained the common EGFR exon 21 mutation of leucine to arginine at amino acid residue 858 (L858R). The histologies of the two NSCLCs are shown in Figure 1.
Our research efforts in characterizing adenocarcinomas of the lung with respect to EGFR mutations involve DNA melting analysis with high-resolution techniques.11 The high-resolution DNA melting curve obtained after amplification of tumor DNA with exon 19 specific primers from the patient described above with the EGFR exon 19 mutations is shown in Figure 2. The melting curve obtained after amplification with the exon 21 primers was normal (not shown). As indicated in Figure 2 the EGFR exon 19 melting curve showed only a slight perturbation with tumor DNA compared with normal. This prompted us to evaluate the exon 19 amplicon by direct DNA sequencing. Surprisingly, direct DNA sequencing easily revealed an EGFR exon 19 deletion and also indicated the apparent lack of the wild-type allele, suggesting the mutation was homozygous (not shown). The lack of sufficient quantities of the wild-type allele makes it difficult to detect mutations by DNA melting because of the lack of heteroduplex formation. This suggests that homozygous mutants or cases in which the mutant allele is amplified with dilution of the wild-type could be challenging to detect simply by melting curve analysis. This potential problem, especially in NSCLC where EGFR amplification is a common event,10,15 led us to develop an alternative mutation detection method using unlabeled DNA probes.
Unlabeled probes are designed to span the area of the exon where activating mutations are known to occur. Probes are single-stranded and are present during the PCR. Asymmetric PCR is performed to increase the amount of the strand complimentary to the probe. Melting analysis detects a separate probe melting peak in addition to the amplicon. If a mutation is present in the area spanned by the unlabeled probe, due to the mismatch between probe and its complement, the unlabeled probe melting temperature will be lowered. Unlabeled probes should detect mutations that are homozygous or when the mutant allele is in abundance over the wild type. To test this, we used a 31 base pair probe, designed to detect the common EGFR deletion exon 19 mutation (see Table 1). As shown in Figure 3, a unique probe melting peak is easily detected with DNA from the NSCLC case that contained an exon 19 deletion even though the mutation was barely detected by amplicon melting (Figure 2). The absence of a wild-type probe melting peak agrees with the sequencing data indicating the lack of detectable wild type. The probe also easily detects the EGFR exon 19 mutation in the CRL5883 cell line. Both mutant and wild-type alleles were detected, in agreement with the DNA sequencing data.
An unlabeled probe to detect the common EGFR NSCLC exon 21 mutation (L858R) is shown in Table 1. The use of the probe in an asymmetric PCR amplifying exon 21 from either an EGFR exon 21 mutation positive cell line or an NSCLC tissue sample in shown in Figure 4. The species corresponding to the mutant and wild type are not clearly resolved by amplicon melting because the change in melting temperature is only about 3°C. Nonetheless, a clear probe melting abnormality is observed in the presence of DNA containing the L858R mutation.
To determine the sensitivity of the unlabeled probes to detect their respective EGFR mutations, DNA from the NSCLC containing the exon 19 mutation, del 747–752 P753S, and the NSCLC with the exon 21 L858R mutation were each separately mixed with varying amounts of normal genomic DNA and subjected to asymmetric PCR with the respective unlabeled probes. As shown in Figure 5, abnormalities in the unlabeled probe melting curve can still be detected when the mutant allele makes up as little as 12.5% of the DNA.
DNA melting analysis with high resolution technology is being applied frequently to detect EGFR and KRAS activating mutations in NSCLC.11,16–18 Mutation detection is important because it may help identify groups of neoplasms most susceptible to EGFR inhibition. We have routinely been using DNA melting analysis to molecularly characterize gastrointestinal stromal tumors (GIST).19 Because most GISTs are heterozygous, amplicon DNA obtained after PCR of an exon of interest will contain both normal and mutant DNA strands. After heating and reannealing the DNA strands, heteroduplexes form between the normal and mutant strands and these lower the melting curve of the amplified DNA. The lowered melting curve is easily detected by high-resolution DNA melting analysis and indicates a mutation. Homozygous mutations are harder to detect in standard melting analysis because heteroduplexes are not formed from tumor DNA and therefore the perturbation of the melting curve from normal is more difficult to detect. The unavoidable normal cell contamination in tumor samples contributes wild DNA which can drive heteroduplex formation and can help uncover homozygous mutations. However, in tumors in which the mutant allele is amplified so that the percent normal DNA becomes minimal, heteroduplexes may not be detected by melting curve analysis at all and without sequencing, a false negative result may occur.
Recently, DNA melting analysis incorporating unlabeled probes has been shown to be useful in genotyping.14 This technology does not require sequence analysis. The system is attractive because it is rapid, inexpensive, and is all performed in a single cuvette (a closed tube system) which greatly diminishes contamination. The unlabeled probes are designed to be complementary to an area of the gene where a mutation(s) of interest is known to occur. Melting analysis performed in the presence of an unlabeled probe will yield two melting peaks. The species with the higher Tm is due to amplicon melting and the species with the lower Tm is due to the probe melting from its complement. Asymmetric PCR is used to provide an excess of the strand complimentary to the unlabeled probe. If a mutation in the area of the unlabeled probe is present, DNA containing it will be produced in excess in asymmetric PCR and form a mismatch with the probe leading to an altered probe melting peak. Therefore, even if the mutation is homozygous or amplified in the tumor, it will form a heteroduplex with the probe (which has a wild-type sequence) and an abnormal DNA melting profile will result. Using unlabeled DNA probes in melting analysis provides a means to circumvent problems that could occur in detecting homozygous mutations or in detecting mutant alleles which are increased in copy number.
In NSCLC, 90% of the EGFR activating mutations are either deletions in exon 19 or point mutations in exon 21.12,13 Because of the limited spectrum of EGFR mutations and because of the known EGFR amplification that occurs in NSCLC,10,15 it appeared that this was an ideal system to develop an EGFR mutation detection assay.
In this report we designed unlabeled DNA probes which span the known activating mutations in EGFR exon 19 and 21. By using these probes in PCR and melting analysis, we were able to easily detect the mutations in both cell lines known to be EGFR mutation positive and from paraffin embedded tumor tissue from known EGFR mutation positive NSCLC. Dilution experiments indicated that the mutant alleles can be detected if they make up at least 12.5% of the total DNA. This is slightly more sensitive than standard high resolution melting11 but in addition, unlabeled probes are able to detect the mutations even in homozygous cases or in cases with an abundance of the mutant allele.
There are no absolute clinical guidelines on how to use EGFR mutation analysis in the therapy of NSCLC.9 Such guidelines will only come from well-planned clinical trials in which a concerted effort is made to obtain tumor tissue for molecular testing. The use of unlabeled DNA probes to detect the EGFR mutations provides a rapid and inexpensive molecular test that could be incorporated into future clinical trials which seek to determine the importance of EGFR activating mutations in the therapy of NSCLC.
We thank the Associated and Regional University Pathologists Institute for Clinical and Experimental Pathology for supporting this work.