Article Text
PDF
Abstract
Aims Somatic insertions/deletions in exon 9 of the calreticulin gene have been identified in patients with essential thrombocythemia and primary myelofibrosis. Over 55 mutations have been discovered, 80% of which consist of either type 1 52-bp deletion or type 2 5-bp insertion. Other mutations (types 3–5) in conjunction with types 1 and 2 account for >87% of identified mutations. The aim of this study was development of a rapid PCR-based assay using LightCycler Hybridisation Probes for the detection of type 1–5 CALR mutations.
Method A real-time PCR assay using a novel HybProbe set was developed for use on the LightCycler 480 Instrument II. The acceptor probe was labelled with LC640 and Faststart DNA Master HybProbe kit was used for PCR reactions.
Results Assay limit of detection was determined to be seven target copies with a probability of 95%. The specificity of the assay was determined by using synthetic constructs of CALR wild-type and CALR mutation types 1–5 with no non-specific detection observed. Samples from 21 patients with essential thrombocythemia (ET) and 12 patients with primary myelofibrosis (PMF), together with 29 control samples from patients diagnosed with various conditions, were screened using the assay. Of these, 24 were found to have mutations in CALR exon 9, with the assay detecting 8 type 1 mutations, 12 type 2 mutations, 2 type 24 mutations, 1 type 20 mutation and 1 31-bp deletion.
Conclusions The novel assay described has potential for application as a rapid, sensitive, high-throughput screening method in the clinical diagnostics setting.
- MYELOPROLIFERATIVE DISEASE
- PCR
- MOLECULAR ONCOLOGY
Statistics from Altmetric.com
Introduction
Myeloproliferative neoplasms (MPNs) are a group of phenotypically heterogeneous, stem-cell-derived clonal haematopoietic disorders characterised by abnormal proliferation of terminally differentiated myeloid cells.1
Recent studies identified recurrent mutations in the gene encoding calreticulin (CALR) in the majority of patients with non-mutated JAK2 or MPL essential thrombocythemia and primary myelofibrosis, and provided evidence for their role in the development of MPNs. CALR mutations consist of somatic insertions and deletions (indels) in exon 9 of the CALR gene.2–6 These mutations lead to a +1 base pair (bp) frameshift in the open reading frame.
More than 55 individual CALR mutations have been described to date2–6 with two mutations being the most common variants, namely type 1 mutation, a 52 bp deletion (c.1092_1143del)(p.L367fs*46), and type 2 mutation, a 5 bp insertion (c.1154_1155insTTGTC)(p.K385fs*47). These two mutations collectively account for 84% of reported CALR mutations (50% and 34%, respectively). The next most frequent mutations in CALR exon 9 are a 46 bp deletion (c.1095_1140del), known as type 3; and a 34 bp deletion (c.1102_1135del), known as type 4, with each of these mutations representing 1.26% of the mutations recently published.2 ,5 ,6 Type 5 mutation, a 52 bp deletion (c.1091_1142del), brings the percentage of type 1 to type 5 mutations to 87% of published mutations.5 ,6
Indels affecting exon 9 of the CALR gene have been reported to be mutually exclusive with the common single-point mutation JAK2 V617F.5 However, recent studies from a small cohort of patients have shown that this is not always the case and there is a small subset of patients who harbour mutations in both JAK2 and CALR genes.7
The discovery of the JAK2 V617F mutation in 2005 led to progress in the diagnostic approach and therapeutic strategy for MPNs.8 ,9 Subsequently, JAK2 was incorporated into the WHO diagnostic criteria for polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF), which recommends testing for JAK2 V617F as a major diagnostic criterion for all suspected MPN. The growing application of molecular technologies for the identification of genetic alterations at the nucleotide level has revealed a number of genes other than JAK2, CALR and MPL, which are mutated in MPNs. These genes include TET2, ASXL1, IDH1, IDH2 and c-CBL, although, as of yet, their role as possible new diagnostic molecular markers is inconclusive.10 CALR has also recently been incorporated into the 2016 WHO classification of myeloid neoplasms.11 The high incidence of CALR mutations in ET and PMF, and its incorporation into the WHO diagnostic criteria, means there is a requirement for robust testing methodologies to be incorporated into clinical diagnostic settings for screening of patients with MPN.
Molecular technologies used to date for the detection of CALR mutations include Sanger sequencing, high-resolution melting analysis, fragment analysis and targeted next-generation sequencing. Fragment analysis was used in one of the original studies carried out to identify somatic mutations in CALR exon 92 and has been shown to have a sensitivity of 5–10%.12 The most common mutations, types 1 and 2, can be identified by their distinct peaks, and Sanger sequencing can be used to determine the nature of other aberrant peaks. Fragment analysis has the capacity to determine mutation burden and, with evidence emerging to suggest clinical and prognostic differences between patients with type 1 and type 2 mutations,13 this is one of the key characteristics of the technology coupled with the ability to detect all known mutations. The ability to quantitate CALR mutant allele burden allows monitoring of molecular responses to treatment,14 ,15 similar to that used for monitoring targeted therapies for MPN.16
We report here the development and evaluation of a rapid, easy-to-use assay using LightCycler Hybridisation Probes (HybProbes) and melt curve analysis designed for the detection of type 1–5 CALR mutations, which account for 87% of total mutations discovered to date.6 The hybridisation probe (HybProbe) format used in this study consists of two oligonucleotides (donor and acceptor) that are specifically designed to hybridise to adjacent internal sequences of the amplified fragment during the annealing phase of the PCR reaction. Hybridisation probes were selected for use because the probes are not hydrolysed, and fluorescence is reversible, thus allowing for the generation of melt curves for analysis of polymorphisms within a DNA sequence. For evaluation purposes, the assay was used to screen known clinical samples obtained from patients diagnosed with MPNs and a control group consisting of patients diagnosed with a variety of both myeloid and lymphoid-derived haematological malignancies.
Materials and methods
Oligonucleotide primer and probe design
CALR exon 9 wild-type cDNA sequence (NCBI reference number: NM_004343.3) was aligned (Clustal W) against all documented CALR mutations.2 ,3 ,5 ,6 PCR primers and hybridisation probes were designed for application in a real-time PCR assay on the LightCycler 480 Instrument II. Primers and probes (Tib-MolBiol, Germany) were reconstituted at 50 and 20 µM respectively in nuclease-free water (Roche) and stored at −20°C. Table 1 lists the primers and probes used in the assay.
CALR assay primer and HybProbe sequences
Real-time PCR
Real-time PCR assays were performed on the LightCycler 480 Instrument II. The LightCycler Faststart DNA Master HybProbe kit (Roche) was used for PCR reactions in accordance with the manufacturer's instructions. The final reaction volume was 20 µL. A negative control (no template DNA in the reaction) was included in each PCR run. The combination of 0.1 and 0.7 µM final concentrations of forward and reverse primers, together with 0.2 µM of each HybProbe and 5 mM MgCl2, was used in this assay. Cycling conditions were 95°C denaturation for 10 min followed by 45 amplification cycles of 95°C for 10 s, 55°C for 20 s and 72°C for 10 s. The temperature transition rate was 4.4°C/s for all steps except for the annealing step of the amplification program, which had a temperature transition rate of 2.2°C/s. Melt curve analysis consisted of denaturation at 95°C for 30 s, annealing at 40°C for 30 s followed by a temperature increase of 40–80°C at a transition rate of 0.3°C/s with continuous monitoring in the Red 640 channel (498–660 nm).
Assay specificity
Assay specificity was determined by using synthetic constructs (Integrated DNA Technologies). Each construct contained an insertion corresponding to wild-type, or mutation types 1–5. Each construct was used as template in the assay.
Limit of detection of the assay
The limit of detection (analytical sensitivity) of the assay was determined by using a synthetic CALR wild-type sequence. DNA was quantified and serial dilutions were prepared from 1×106 to 1 DNA copies and used as target in the assay. Eight replicates of 10, 7.5, 5, 1, 0.5 and 0 DNA copies of target were tested. The lower limit of detection was established and a probit regression analysis (95% probability) performed (MiniTab, State College, Pennsylvania, USA).
DNA extraction from whole blood
Anticoagulant treated whole blood (200 µL) collected from patients, following informed consent, at Galway University Hospital, was transferred to a sterile 1.8 mL microcentrifuge tube and digested with 20 µL proteinase K (Invitrogen) at 56°C for 10 min. The DNA was then extracted using the DNeasy Tissue Kit (QIAGEN, Valencia, California) and eluted with 100 μL (50 μL×2) of nuclease-free water. The DNA concentration was spectrophotometrically measured by light absorption at 260 nm of ultraviolet light using the Qubit Fluorometer (Life Sciences, http://www.invitrogen.com).
Clinical samples
In total, 62 clinical samples were screened using the CALR HybProbe assay. This sample set included 37 peripheral blood samples (samples 1–37) from patients at Galway University Hospital containing a variety of lymphoid and myeloid malignancies, and 25 genomic DNA samples (samples 38–62) obtained from patients diagnosed as having MPN from Belfast City Hospital (samples 38–42) and St. James Hospital Dublin (samples 43–62). Samples 1–37 had not previously been screened for CALR exon 9 mutations; however, samples 38–62 had been tested previously for CALR mutations.17 ,18
Results
Analytical specificity
The specificity of the CALR assay was evaluated using wild-type sequence and type 1–5 mutation-containing sequences as template in the PCR assay. When the wild-type melt peak at 72°C was compared with each individual mutated sequence type (figure 1A–E), a clear separation of melt peaks was observed. Compared with the wild-type sequence, type 1 mutation shows a Tm difference of 8°C at 64°C (figure 1A), type 2 mutation shows a Tm difference of 4°C at 68°C (figure 1B), types 3 and 5 have a TM difference of 6°C at 66°C (figure 1C, E) and type 4 has a Tm difference of 3°C at 69°C (figure 1D).
(A) Melt peaks obtained following melt curve analysis when wild-type and type 1 synthetic constructs were used as template in the PCR reaction. The wild-type melt peak is at 72°C while the type 1 mutation melt peak is at 64°C. (B) Melt peaks obtained following melt curve analysis when wild-type and type 2 synthetic constructs were used as template in the PCR reaction. The wild-type melt peak is at 72°C while the type 2 mutation melt peak is at 68°C. (C) Melt peaks obtained following melt curve analysis when wild-type and type 3 synthetic constructs were used as template in the PCR reaction. The wild-type melt peak is at 72°C while the type 3 mutation melt peak is at 66°C. (D) Melt peaks obtained following melt curve analysis when wild-type and type 4 synthetic constructs were used as template in the PCR reaction. The wild-type melt peak is at 72°C while the type 4 mutation melt peak is at 69°C. (E) Melt peaks obtained following melt curve analysis when wild-type and type 5 synthetic constructs were used as template in the PCR reaction. The wild-type melt peak is at 72°C while the type 5 mutation melt peak is at 66°C.
Analytical sensitivity
The limit of detection or analytical sensitivity of the assay was determined using 10, 7.5, 5, 1, 0.5 and 0 copies of wild-type CALR exon 9 sequence as target in the assay. Results showed that all eight replicates of 10 copy target input were detected while at 7.5 and 5 copy inputs seven replicates were detected. Four out of eight replicates were detected for 1 and 0.5 copy inputs. Hit rate analysis showed that the limit of detection was between 10 and 7.5 copies of target as detection begins to fall off after 10 copies (table 2). The detection limit of the assay was 7.3 genome equivalents per PCR reaction with a probability of 95% as determined (probit analysis).
Hit rate analysis of data generated from LOD testing of CALR assay
Clinical sample screening using the CALR real-time PCR assay
Of the peripheral blood samples tested (samples 1–37), three were found to have CALR exon 9 mutations (samples 13, 29 and 36). These 37 samples were also sequenced (Sequiserve Germany) to determine the amplicon sequence. This confirmed that there was one additional sample (sample 25) that had a CALR exon 9 mutation, which was not detected by the assay. This was due to the fact that the probe had only two mismatches against the target on the 5′ end, which did not produce a significant enough Tm shift to result in a distinct melt peak. The positive patient samples gave melt peaks at 63°C, 67°C and 68°C corresponding to CALR mutation types 1, 2 and 24 (figure 2A–C).
(A) Melt peaks obtained for samples 1–23 following melt curve analysis. A double peak with peaks at 68°C (type 2 mutation) and 72°C (wild-type) for sample 13. Results were confirmed by sequencing. (B) Melt peaks obtained for samples 24–33 following melt curve analysis. A double peak at approximately 67°C (type 2 mutation) and 72°C (wild-type) for sample 29. Results were confirmed by sequencing. (C) Melt peaks obtained for samples 34–37 following melt curve analysis. A double peak with peaks at approximately 63°C (type 1) and 72°C for sample 36. Results were confirmed by sequencing. (D) Melt peaks obtained for samples 38–42 following melt curve analysis. Double peaks indicating mutation in CALR exon 9 were obtained for samples 38, 39, 41 and 42. Results were confirmed by sequencing.
Genomic DNA samples 38–42 had previously been screened for CALR exon 9 mutations using fragment analysis.17 The CALR real-time PCR assay detected four out of five of these mutations. Peaks were seen at 62°C and 63°C, corresponding to a previously reported 31 bp deletion CALR mutation (p.D373fs*48),3 and type 1 CALR mutation, respectively, while the two other mutations detected by the assay share the same peak at 67°C corresponding to CALR mutation types 2 and 20 (figure 2D). The assay failed to detect a 4 bp deletion in sample 40 because the mutation (type 22) was outside the probe binding region. Results were confirmed by Sangar sequencing (data not shown).
The remaining genomic DNA samples (43–62) had also been previously screened by fragment analysis for CALR mutations.18 The CALR real-time assay detected mutations in 17 of these samples. Negative samples were sent for sequencing (Sequiserve Germany). The CALR assay results correlated 100% with the fragment analysis and the sequencing results. These samples showed three distinct melt profiles indicating three different mutations were present. Ten samples gave a peak at 68°C, corresponding to mutation type 2, six samples gave a peak at 64°C (mutation type 1) and one sample gave a peak at 67°C corresponding to mutation type 24 (figure 3A–C). A summary of all results is presented in table 3.
Summary of results obtained for all clinical samples tested
(A) Melt peaks obtained for six samples 47, 55, 56, 57, 58 and 61 at 64°C indicating a mutation in CALR exon 9. (B) Melt peaks obtained for 10 samples 43, 44, 45, 46, 48, 50, 51, 53, 54, 60 at 68°C indicating a mutation in CALR exon 9. (C) Melt peak obtained for sample number 52 at 67°C indicating a mutation in CALR exon 9.
Discussion
CALR mutations have recently been identified in a significant proportion of patients with ET/PMF without JAK2 and MPL mutations. The overall aim of this study was to design, develop, optimise and clinically validate a robust, rapid and user-friendly CALR exon 9 mutation screening assay using LightCycler hybridisation probes (HybProbes) and melt curve analysis. The assay described here was designed to detect the most common CALR mutations type 1–5, which account for >87% of mutations. However, while the assay was designed to detect mutation types 1–5, because of the mechanism of action of HybProbes and the ability to perform a melt curve analysis, the assay also detects other mutations. Based on the samples tested to date, additional mutations detected by the HybProbe assay are types 20, 24 and a 31 bp deletion. It is expected that as additional clinical samples are tested other mutations will also be detected by the assay.
This study demonstrated the feasibility of using melt curve analysis using HybProbes for the identification of CALR exon 9 mutations. The assay allows clear discrimination of mutated and wild-type sequences. Initial demonstration of the principle was carried out using synthetic CALR type 1–5 sequences as template in the PCR reaction. Following amplification, melt curve analysis was carried out and specific melt peaks were obtained for wild-type as well as for each mutated sequence type. All melt peaks obtained were in the range 72–64°C. The limit of detection of the assay was determined to be approximately seven copies of target DNA.
The CALR mutation status of 62 clinical samples from patients was evaluated using the new assay. DNA extraction was carried out directly from whole blood. Of these, 24 were found to be mutation positive. In two patient samples, two CALR mutations were not detected using the CALR HybProbe screening assay, but were found to be mutation positive when samples were sequenced. The mutations were a type 12 34-bp deletion (c.1098_1131del) and a type 22 4-bp deletion (c.1120_1123del). These mutations are located just outside the probe binding region, and therefore, the assay was unable to detect them. The assay was, however, found to detect three additional mutations, even though they were not included for detection at the time of the design of the HybProbe assay.
Of the 26 mutation-positives in the clinical sample set tested, 24 were detected. The eight different types of mutations detected by the CALR screening assay (types 1–5, 20 and 24, as well as a 31 bp deletion) account for around 15% of the 55 published mutations, but given the frequency of type 1 and 2 mutations, the assay can detect >90% of known CALR exon 9 mutations.
Other techniques are available for detecting CALR mutations, including HRM19 and Sanger sequencing.20 Both methods have been reported to give false positive and negative results. In a cohort of 32 patients analysed in a previous study,21 high resolution melting (HRM) identified CALR mutations in 22 (69%) of 32 ET patients compared with 16 (50%) patients by Sanger sequencing. Further investigation revealed that 21 patients were positive for CALR mutations, giving a false positive rate using HRM of 3%, with no false negative results. The authors of the study state that the inability of Sanger sequencing to detect mutations in a number of samples may be due to the fact that the mutation burden may have been low in these particular samples rather than a limitation with the technique. The HRM technique has been used in many studies and a range of mutation types emerged from these studies. For example, three different mutation types were detected in one study19 and six different mutations detected in a second study.20 HRM technology is similar to the HybProbe technology described in this study, but without the inclusion of a specific set of detection probes. From an ease-of-use perspective, HybProbes and HRM are closed-tube methods, meaning they are rapid and melt analysis can be conducted immediately after PCR amplification. The use of HybProbes or HRM for the detection of CALR mutations provides a medium-throughput screening option for the clinical diagnostic laboratory setting. They also enable CALR mutant allele burden to be estimated, based on measuring the relative area under the curve of mutant and wild-type alleles, thus enabling monitoring of treatment. Both methods give distinct melting curves for positive samples, but if exact genotyping of the mutation is required, Sanger sequencing is necessary as frequently, when using HRM or HybProbes, more than one specific CALR mutational type will have the same melt profile. For either method to be absolutely robust, melt curves would have to be generated for all CALR mutations and grouped into similar melting profiles to give an indication of what CALR mutation is present. Even if this was carried out, many mutation types would have the same melt profile due to the melting dynamics of the target and probe.
Fragment analysis is another technique used for detection of CALR mutations.2 ,5 This method detects only indels and does not detect point mutations. It is a two-step method resulting in it being a more cumbersome technique to carry out in the routine clinical setting and it also has the potential for contamination. Additionally, specific training to produce and interpret results is necessary.12 Despite these limitations, it has the advantage of having a relatively high sensitivity for CALR mutation detection.21 It also enables mutant allele burden to be estimated.15 Targeted next-generation sequencing has also been used to identify CALR mutations.12 While it is more complex than the other mutation detection techniques, the authors describe the development of a novel analysis pipeline and a limit of detection that they report to be superior to other methods.12 Implementation of next generation sequencing (NGS) into the clinical diagnostic laboratory setting is complex with specific technical training of staff and informatics expertise for result interpretation required, although advances have been made in an effort to make the technique more user-friendly. Digital PCR (dPCR) has recently been described for accurate and highly sensitive quantification of type 1 and/or type 2 CALR mutant allele burden.22 ,23 However, the quantitative detection of all identified CALR mutations would require the development of a large number of individual or multiplex assays using dPCR.
To our knowledge, this is the first report of the use of HybProbes for the detection of CALR mutations using DNA extracted directly from whole blood. The assay and associated melt curve analysis described in this study offers advantages as a screening tool, such as ease of use, and its ability to detect >90% of identified exon 9 CALR mutations. It also enables the mutant allele burden to be determined. Its inability to detect two mutation types in the clinical samples analysed, type 12, a 34 bp deletion (c.1098_1131del), and type 22, a 4 bp deletion (c.1120_1123del), is a limitation, and further improvements could be made to increase the number of mutations detected by the HybProbe assay. Bioinformatics analysis indicated that these two mutations, and six additional mutations (type 23, a 5 bp insertion (c.1120_1131delinsTGCGT), type 26, a 1 bp deletion (c.1122del), type 27, a 5 bp insertion (c.1123_1125delinsTGTTT), type 37, a 34 bp deletion (c.1091_1124del), a 34 bp deletion (c.1096_1129del) and a 5 bp insertion (c.1127_1129delinsTTTGC)) were located outside the probe binding region and therefore would not be detected. There are multiplexing options available that could provide the necessary configuration for inclusion of additional probes allowing detection of more mutated sequences.
The HybProbe assay described here detects >90% of all identified CALR mutations as well as having the advantage of being a simple and straightforward assay to set up, requiring limited hands-on time. The key advantage of the HybProbe assay format is that because of the presence of specific probes, primer-dimers that may result from the reaction do not interfere with the fluorescence signal. This is in contrast to intercalating dyes such as those used in HRM, which potentially integrate into any double-stranded nucleic acid including any primer-dimers present in the reaction. Another advantage of HybProbes is that a single-sensor probe can detect multiple mutations. With the addition of more than one sensor probe in a reaction, many more mutations can be detected. Additional probes can be labelled with the same dye label or using a different label allowing groups of mutations to be detected required. With further work, the assay described here could be made even more specific by the inclusion of additional mutation probes, and quantitative, allowing monitoring of the mutant allele burden in patients. Given that the assay detects >90% of identified CALR exon 9 mutations, we propose that the HybProbe assay described here has potential as a CALR mutation screening assay for use in the clinical laboratory diagnostic setting.
Take home messages
The work described here involved the design, development and optimisation of a robust, rapid CALR exon 9 mutation detection screening assay, using HybProbes, which can be performed on widely available laboratory instrumentation. The assay was developed with the clinical laboratory workflow in mind with minimal hands-on time and result interpretation.
The work described demonstrates the potential of the use of HybProbes for CALR exon 9 mutation detection and demonstrates the usefulness of the technique as a first-line screening assay in a clinical diagnostic setting.
While HybProbes are generally used for single nucleotide polymorphism detection, in the novel assay described here, multiple indels were detected. The current assay facilitates genotyping of type 1 and 2 mutations, which is important given that these mutations are the most common and can lead to different clinical phenotypes. Additional discriminating capabilities could be added to the assay with further development.
References
Footnotes
Handling editor Mary Frances McMullin
Contributors TK and LOC contributed equally. TK performed the experimental work described in the manuscript. LOC is a member of the project management team. Contributed intellectually to the design of the study and prepared the manuscript. JK is a member of the project management team responsible for the study concept and design. Assisted with collection of clinical samples. AHH and MM assisted with collection of clinical samples and contributed to study concept and design. MOD is a member of the project management team responsible for the study concept and design. MP, SL and KH provided access to clinical samples and provided critical review of the manuscript. BG , CM , EK and SL assisted with technical aspects of assay development. TS is a member of the project management team responsible for the study concept and design, critical review of the experimental work. Responsible for manuscript preparation and critical review.
Funding Irish Research Council (EBPPG/2013/62).
Competing interests None declared.
Patient consent Obtained.
Ethics approval National University of Ireland Galway Research Ethics Committee and GUH REC.
Provenance and peer review Not commissioned; externally peer reviewed.