Allele-Specific Polymerase Chain Reaction for the Imatinib-Resistant KIT D816V and D816F Mutations in Mastocytosis and Acute Myelogenous Leukemia

Tumor Samples and Cell Lines

Tumor samples were obtained from the Tissue Bank of the Oregon Health and Science University Cancer Institute, from the Oregon Health and Science University Departments of Pathology and Dermatology, the University of Chicago Section of Dermatology, and the Department of Pathology of Brigham and Women’s Hospital. The use of these tissues in the study was performed in accordance with the regulations of the institutional review boards of these institutions. The human HMC-1.2 mast cell line, which harbors both V560D and D816V mutations of the KIT gene, was kindly provided by Dr. C. Akin (Laboratory of Allergic Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD).28 LoVo cells (a colon cancer cell line that is wild type for KIT) were obtained from the American Type Culture Collection, Rockville, MD (http://www.atcc.org). In a previous study we generated stably transfected clones of Ba/F3 cells (murine interleukin 3-dependent hematopoietic pro-B cells) expressing wild-type, D816V, D816F, or D816Y mutant forms of human KIT cDNA.40 DNA extracted from these Ba/F3 subclones was used in validation experiments.

DNA Preparation

DNA was extracted from HMC-1.2, LoVo, and Ba/F3 cells with the QIAmp DNA mini kit (no. 51306; Qiagen, Valencia, CA). The same kit was used to extract genomic DNA from blood and bone marrow aspirates, as well as from formalin-fixed tissue (unembedded), and formalin-fixed, paraffin-embedded tissue samples, in accordance with the manufacturer’s recommendations.

Standard PCR

KIT exon 17 amplicons were generated from genomic DNA as previously described.41 For the standard (control) PCR reaction, the forward primer was 5′-TGTATTCACAGAGACTTGGC-3′, and the reverse primer was 5′-TAATGTTCAGCATACCATGCAA-3′. This reverse primer matches sequences in KIT intron 17 and was also used in the allele-specific PCR. To amplify KIT cDNA sequences from Ba/F3 subclones, the same forward primer was matched with the reverse primer 5′-GCTCCCAAAGAAAAATCCCATAGG-3′.

Allele-Specific PCR

Allele-specific PCR reactions were performed using 200 ng of DNA and a master mix based on the Expand High Fidelity Polymerase kit (no. 11759078001; Roche, Indianapolis, IN) with 1.4 U of polymerase, 160 μmol/L dNTP (Stratagene, Cedar Creek, TX), 400 nmol/L mutation-specific primer (Figure 1), 200 nmol/L blocking oligonucleotide (Figure 1), and 800 nmol/L reverse primer (5′-TAATGTTCAGCATACCATGCAA-3′). The cycling conditions were as follows: 95°C for 1 minute, followed by 45 cycles of 94°C for 1 minute, 55°C for 1 minute and 72°C for 1 minute, and a final 7-minute incubation at 73°C.

HPLC and DNA Sequencing

PCR products were routinely analyzed on a denaturing HPLC system (WAVE; Transgenomic, Inc., Omaha, NE), as previously described.4,38 This system allows the detection of DNA heteroduplexes that differ by as little as one nucleotide in 400 bp and is sensitive in the range of 15 to 20% mutant allele. DNA sequencing was performed using the BIGDye terminator kit (Applied Biosystems, Foster City, CA) and analyzed on an ABI 310 capillary sequencer.4

Topo Subcloning

To look for low-abundance KIT exon 17 mutations, amplicons were cloned into Topo plasmids and sequenced individually, as previously described.42

Assay Design

Our assay is modeled on the approach recently described by McKinzie and Parsons39 to study human KRAS and mouse Hras mutations. The mutation-specific primer (MSP) and the blocking oligonucleotide are shown in Figure 1. The MSP is designed to prime from the mutant A of the T>A substitution of D816V (A81402T based on GenBank U63834; KIT cDNA A2468T) but includes an intentional mismatch in the penultimate 3′ nucleotide. When aligned with wild-type (WT) sequence, the MSP is doubly mismatched at the 3′ end, limiting its ability to prime. The MSP can also prime from the second nucleotide of the two-nucleotide substitution that results in D816F (double substitution G81401T and A81402T in GenBank U63834; KIT cDNA G2467T and A2468T) (Figure 1), which is another mutation associated with SM and is resistant to imatinib in vitro.28 In theory, the MSP could also serve in amplifying D816I and D816M (both perfect matches), as well as D816L (penultimate mismatch), but to our knowledge these mutations have not been observed in SM. On the other hand, the known SM-associated mutations D816H (G81401C; KIT cDNA G2467C) and D816Y (G81401T; KIT cDNA G2467T) would not be expected to amplify, as both are doubly mismatched to the last two MSP nucleotides. This was confirmed for both of these mutations, as detailed below.

The design of the blocking oligonucleotide (B-oligo) is the reverse of the MSP, such that it is doubly mismatched to the mutant D816V and D816F sequences, but only singly mismatched to WT. In addition, the 3′ terminal nucleotide of the B-oligo was linked in a chemically reversed configuration (3′ to 5′) to eliminate any possibility of priming.

Pilot studies were performed using DNA purified from HMC-1.2 cells, which are heterozygous for the D816V mutation. The expected sequence of the amplified product was confirmed by bidirectional sequencing. Temperatures examined for primer annealing included 58°C, 57°C, and 55°C, with the latter providing the best results. We tested a reduced number of PCR cycles35 but found that 45 cycles provided more consistent results. The addition of the B-oligo increased the sensitivity of the assay by 10- to 100-fold (data not shown). Molar ratios of MSP to B-oligo ranging from 2:1 to 1:2 were tested, and the most consistent results were observed at a 2:1 ratio, which was used in all subsequent assays. In addition, it was found that doubling the concentration of the reverse primer relative to the MSP yielded more consistent amplifications.

Parson and colleagues43 reported that substitution of a modified DNA polymerase that lacks 5′ to 3′ exonuclease activity (Stoffel fragment) in their allele-specific mouse Hras assay increased allele discrimination. However, in our assay a Stoffel fragment polymerase (no. N8080038; Applied Biosystems) generated spurious products; therefore, we used the Expand High Fidelity polymerase (no. 11759078001; Roche).

Assay Sensitivity for KIT D816V

The sensitivity of the AS-PCR assay for KIT D816V was influenced by the quality of the template DNA. Positive signals were routinely detectable at a level of 0.1% mutant allele using DNA extracted from HMC-1.2 cells diluted into wild-type DNA from LoVo cells (not shown). The detection threshold was higher when paraffin-derived DNA from a seminoma heterozygous for D816V was diluted with the same wild-type LoVo DNA (not shown) or with genomic DNA from a formalin-fixed, paraffin-embedded GIST that was wild-type for KIT exon 17 by denaturing HPLC (Figure 2). The D816V mutation in paraffin-derived DNA was reproducibly detected at the level of 1% but not below this. The positive control used for all AS-PCR reactions was a parallel tube containing the same reverse primer matched to a generic exon 17 forward primer, which was 5′ to the allele-specific primer and yielded a larger amplicon.

Validation Studies

The AS-PCR assay was validated through the analysis of five types of samples. First, 64 samples of genomic DNA wild type for KIT exon 17 were tested in the allele-specific reaction. No amplifications were observed with the MSP/B-oligo mixture, whereas control primers showed strong signals. The wild-type status of these 64 samples had been determined by denaturing HPLC screening, which in our experience is more sensitive than DNA sequencing. Thirty-one of the samples were also wild type by direct sequencing; the remaining cases were not sequenced. Because 20 of the samples were derived from cell lines or fresh-frozen tumor tissue (six AML, three GIST, seven myxoid chondrosarcoma, and one lymphoma), the quality of the DNA was probably not a factor in these AS-PCR reactions. The remaining 44 samples were from paraffin-embedded tumors (GIST, non-GIST sarcoma, hemangioma, thymic carcinoma, renal cell carcinoma, renal oncocytoma, and desmoid fibromatosis). Genomic DNA from a cell line wild type for KIT exon 17 (LoVo cells) was included as a negative control in all subsequent reactions.

Second, a group of Ba/F3 subclones expressing wild-type or mutant forms of human KIT cDNA were tested. As illustrated in Figure 3, positive AS-PCR signals were obtained from cells stably transfected with D816V or D816F cDNA, but not from cells containing WT or D816Y forms of KIT. Control PCR reactions for human KIT exon 17 sequences were positive in all cases. Because the amplification target in these cell lines was cDNA, a different reverse primer was used in the AS-PCR reactions. Nevertheless, the results support the specificity of the MSP/B-oligo combination for the D816V and D816F mutations. As we did not have any paraffin-derived samples with KIT D816F, which is a rare mutation in mast cell disease and has not been reported in seminoma or GIST, we could not confirm that KIT D816F can be amplified from genomic DNA. However, the positive result obtained from the D816F cDNA-expressing Ba/F3 clone suggests that any signal generated by MSP/B-oligo combination could represent either D816V or D816F.

Third, a series of tumors with sequence-confirmed KIT D816V was analyzed with the AS-PCR assay. These included eight samples of AML (DNA from fresh cells) and three samples of paraffin-embedded seminoma (Table 1, cases 1 to 11). All of these samples yielded positive amplicons by AS-PCR. As noted above, we could not further validate the AS-PCR reaction for the D816F mutation because none of our tumor samples had this substitution.

We next examined a group of tumors with alternate mutations of codon 816 using the AS-PCR assay. Two seminomas with sequence-confirmed D816H did not amplify by AS-PCR, as expected (Table 1, samples 12 and 13). D816H is also a mutation found in GISTs from patients with acquired resistance to imatinib, and two such tumors were negative by AS-PCR despite robust amplifications with the control primers (samples 14 and 15). Imatinib-resistant GISTs with D816G or D816A were AS-PCR-negative as well (samples 16 and 17, respectively).

Interestingly, two AML samples that had KIT D816Y mutation by direct sequencing were positive by AS-PCR (Table 1, samples 18 and 19). Topo subclones were prepared and sequenced from these amplification products, and D816V-positive clones were detected at low levels (1 of 20 from case 18, 1 of 10 from case 19). These results are consistent with the estimated sensitivity of the AS-PCR assay and provide evidence of heterogeneity among KIT mutations that occur in AML. We suspect that mutation heterogeneity also accounts for the small positive AS-PCR signal observed in one of the imatinib-resistant GISTs that appeared to be D816H by direct sequencing (Table 1, sample 20). Although the AS-PCR signal in this case could represent a false positive, it is perhaps more likely that it reflects low-abundance D816V that accumulated under the selective pressure imatinib therapy. This view is supported by the fact that there were no positive signals among 25 untreated GISTs that were wild type for KIT exon 17 (see above) or among the four other D816H-mutant tumors examined (two seminoma, two GIST).

Finally, a group of biopsies from adult patients with suspected mast cell disease was screened for KIT exon 17 mutation using our standard PCR reaction and HPLC, with confirmatory sequencing as necessary, and by AS-PCR. Among 21 bone marrow samples (aspirates or cores), six were positive by both assays, seven were positive by AS-PCR only, and the remaining eight were negative by both assays (Table 2). The tested marrows were predominantly referral samples from patients with suspected SM; unfortunately, clinical details for most of these patients could not be obtained. Marrow samples from a patient with mast cell leukemia (Table 2, sample 29) and a patient with SM-AML (sample 30) were positive by both the standard and AS-PCR assays. In contrast, marrow samples from three patients with a clinical diagnosis of SM-MPD (samples 56 to 58) were all negative by both assays. Interestingly, one of these was found to be positive for a JAK2 V617F mutation when analyzed in the Molecular Diagnostics Center of Oregon Health and Science University (data not shown).

Among 13 skin biopsies with mast cell infiltrates, three were positive by both the standard assay and AS-PCR, another was AS-PCR-positive only, and the remaining nine were completely negative (Table 2). Because no attempt was made to microdissect the areas involved by mast cells, it is possible that the negative cases had too few mast cells to allow detection, even by AS-PCR. Alternatively, the infiltrates may not have been related to a mutant KIT-related neoplasia. Other examples of tissues that yielded positive AS-PCR results included one colon biopsy, one lymph node, and one of two blood samples (Table 2).

The results summarized in Table 2 demonstrate an excellent correlation between the standard PCR/HPLC assay and AS-PCR results, but with a clear advantage for the latter. Sequential biopsies from one particular patient are of interest in this regard. A colon biopsy from this 61-year-old male (sample no. 32) was morphologically consistent with the clinical diagnosis of SM but was negative for an exon 17 mutation by the standard assay. The AS-PCR assay on this sample was clearly positive. The patient subsequently developed AML, and a blood sample (no. 31) was positive by both the standard assay (D816V on sequencing) and the AS-PCR assay.