| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Molecular Diagnostics and Genetics |
1 Spectra Genetics, LLC, 4415 Fifth Ave., Suite 160, Pittsburgh, PA 15213.
aAddress correspondence to this author at: University of Pittsburgh, Hillman Cancer Center, 5117 Centre Ave., Pittsburgh, PA 15213-1863. Fax 412-623-7768; e-mail malehornde{at}msx.upmc.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We developed a new method, based on peptide mass signature genotyping (PMSG), for the detection of DNA mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. Exons of the gene were amplified, cloned, and expressed in Escherichia coli as peptide fusions, in natural as well as unnatural reading frames. Peptide analytes were purified by immobilized metal affinity chromatography and analyzed by matrix-assisted, laser desorption/ionization time-of-flight mass spectrometry. Synthetic and natural DNA samples with the 25 mutations recommended for CFTR carrier screening (Grody et al. Genet Med 2001;3:149–54) were assessed using the PMSG test for the CFTR gene.
Results: Peptide analytes ranged from 6278 to 17 454 Da and varied 30-fold in expression; highly expressing peptides were observed by electron microscopy to accumulate as inclusion bodies. Peptides were reliably recovered from whole-cell lysates by a simple purification method. CFTR mutations caused detectable changes in resulting mass spectrometric profiles, which were >95% reliably detected in blinded testing of replicate synthetic heterozygous DNA samples. Mutation detection was possible with both sample pooling and multiplexing. The PMSG CFTR test was used to determine compound heterozygous mutations in DNA samples from cystic fibrosis patients, which were confirmed by direct DNA sequencing.
Conclusions: The PMSG test of the CFTR gene demonstrates unique capabilities for determining the sequence status of a DNA target by sensitively monitoring the mass of peptides, natural or unnatural, generated from that target.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this report, we describe a PMSG test for mutation in the human CFTR gene. Our process differs from that of Garvin et al. (1), in several respects: (a), it uses in vivo peptide expression in Escherichia coli instead of cell-free expression; (b) the peptide analytes in our process receive their epitope tags from sequences in the plasmids from which they are expressed, instead of from 5' anchor sequences in the primers used to amplify the test sequences; and (c) our process generates and analyzes peptides from more than one reading frame of the test sequence, a feature that allows more sensitive detection and accurate specification of the nature of changes in the DNA in many cases.
PMSG affords several advantages over other mutation detection methods: (a) The method detects a wide variety of mutations throughout a target DNA sequence, unlike sequence-specific hybridization or amplification techniques that detect only those mutations that are covered by relatively short probes or primers. This confers on PMSG the ability to more comprehensively "scan" a target DNA sequence for a larger number of potential mutations and polymorphisms. Because of the colinearity of an entire target DNA sequence to the peptide analyte generated from it, PMSG has the ability to register many discrete individual variations, as well as the combined effect of multiple variations within a target DNA sequence, in a single analysis. (b) Because of the width of the useful mass spectrum and the resolution of MALDI-TOF MS, many analytes can multiplexed for simultaneous analysis. (c) Because of the sensitivity and dynamic range of MALDI-TOF MS, mutant species can be distinguished from a preponderance of wild-type signal, allowing for sample pooling. Together these attributes can be exploited to (among other things) detect novel mutations, establish haplotypes across regions with multiple variations, detect mutation in nonhomogeneous tissue samples, and correlate polymorphisms with phenotypes. Several research applications of PMSG in our laboratory have successfully detected previously known and unknown mutations in DNA samples from patients with clinically diagnosed genetic diseases, including mutations in the RDS/peripherin gene responsible for macular degeneration; in the CYP21 gene responsible for congenital adrenal hyperplasia, and in the TP53 gene associated with head and neck cancers (2).
To assess and develop the commercial capabilities of PMSG to screen for mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, we have undertaken the development of a PMSG test that identifies each of the 25 specific mutations that have been recommended for preconception carrier screening in the general population (3). We designed a panel of 18 peptide analytes, generated from 14 exons and 1 intron of the CFTR gene, that successfully report the presence of these 25 mutations, with a low incidence of false-positive and -negative results. Detection of the 25 mutations was demonstrated from synthetic DNAs and commercial reference DNAs as well as from DNA from suspected compound heterozygous patients, with the latter findings verified by DNA sequencing. All steps, including bacterial culture, expression, and purification, were performed in a 96-well format compatible with automation. Mutations were successfully detected when peptide analytes were generated and analyzed in multiplexed combinations, and the detection of mutations has been shown to tolerate moderate sample pooling. The latter properties (automation capability, multiplexing, and pooling) encourage the future development of a comprehensive low-cost test that scans the entire CFTR sequence for mutation.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
amplification of cftr
Primers used to generate first stage or "outside" PCR products of the CFTR gene are listed in supplemental file 1, which is available as part of the Data Supplement accompanying the online version of this article at Clinical Chemistry Online (http://www.clinchem.org/content/vol49/issue8/). Amplification reactions using SureStart Taq (Stratagene) included a total primer concentration of 2 µM in a 20-µL reaction volume; 1.5 mM MgCl2, 20 mL/L dimethyl sulfoxide, and 250 µM each of the deoxynucleotide triphosphates. Input genomic DNA extracted by automated means from dried blood spots (S. Dobrowolski; Neo Gen Screening, Bridgeville, PA) was included at <5% of reaction volume. The "reference patient" wild-type DNA had been sequenced previously in all relevant exons to ensure conformity with the published wild-type sequence. The hot-start thermal cycling conditions were as follows: 95 °C for 10 min; 35 repeats of three steps at 94 °C for 20 s, 56 °C for 60 s, and 72 °C for 40s; and 5 min at 72 °C. Ramping time was set at the minimum, and reactions were performed in an Eppendorf MasterCycler (Eppendorf).
Products of the first-stage PCR were reamplified in a second-stage or "nested" reaction with Eppendorf Taq, using the primers listed in supplemental file 2 (available in the online Data Supplement). Two different products each were generated from exons 4, 7, and 11. Amplification reactions included primer at a concentration of 2 µM in a 20-µL reaction volume, 1.5 mM MgCl2, 20 mL/L dimethyl sulfoxide, and 250 µM each of the deoxynucleotide triphosphates. We included a 20-fold diluted first-stage reaction product at <5% of reaction volume. Thermal cycling conditions were as follows: 95 °C for 2.5 min, followed 35 repeats of three steps at 94 °C for 10 s, 59 °C for 30 s, and 72 °C for 20 s, with a ramp rate of 2 °C/s, and a final step at 72 °C for 5 min.
molecular cloning
Plasmid pWZ4E DNA was purified from 250 mL of E. coli DH5
culture with use of the Qiagen plasmid maxiprep reagent set (Qiagen). The entire yield of one preparation was digested with 200 U each of DraIII and HindIII for 6 h at 37 °C, followed by the addition of 200 U of calf intestinal alkaline phosphatase for 2 h and additional digestion at 37 °C (all enzymes were from New England Biolabs). After heat inactivation for 20 min at 70 °C, nucleic acid was purified by two organic extractions each with phenol–chloroform–isoamyl alcohol (25:24:1 by volume) and chloroform–isoamyl alcohol (24:1 by volume), followed by ethanol precipitation with sodium acetate. Pelleted DNA was resuspended in sterile water at roughly 1 µg/µL, which was estimated by gel electrophoresis and ethidium bromide staining.
Second-stage PCR product was prepared for cloning by digesting it for 4 h at 50 °C with 10 U of SfiI (New England Biolabs), followed by purification with the GeneClean reagent set (Q-biogene). Roughly one-half of the eluate was ligated with 1 µL of vector DNA in a 6-µL reaction using 150 U of T4 DNA ligase (New England Biolabs) at room temperature for 1 h.
The E. coli host strain, NBDR, is NovaBlue DE3 (Novagen) bearing the pRARE plasmid (Novagen), which contains a tRNA cassette with genes for six rare tRNA species. The strain genotype is endA1 hsdR17(rK12-'K12+) supE44 thi-1 recA1 gyrA96 relA1 lac F'[proA+B+ lacIqZM15::Tn10(TcR)] (DE3) pRARE (CmR). The NBDR cells were made chemically competent by the method of Inoue et al. (5). Transformation and subsequent culture were conducted in sterile 96-well polypropylene plates (Costar no. 3960; 2-mL assay block; Corning Inc.). Transformations were performed by mixing 3 µL of ligations with 25 µL of cells in the bottom of each well, incubating the plate on ice for 15 min, heat shocking for 40 s at 42 °C, incubating on ice for 2 min, and finally adding 200 µL of warm SOC medium to each well. Cultures were incubated for 1 h at 37 °C by shaking the plates, covered by a storage mat (Costar no. 3080; Corning), at 200 rpm at an incline of 30 degrees, after which each well received 500 µL of warm Terrific Broth (Q-biogene) with antibiotics at 150% of the final concentration (final concentrations, 30 mg/L kanamycin, 10 mg/L tetracycline, and 10 mg/L chloramphenicol). Culture plates were shaken overnight at 37 °C at 300 rpm inclined at an angle of 30 degrees.
expression and purification
Cultures were restarted for induction by transferring 100 µL of overnight culture into 500 µL of warm Terrific broth with antibiotic selection in a fresh 96-well plate. Cultures were shaken at 37 °C for roughly 3–4 h, until A600 readings for the undiluted culture were >1.2. Isopropylthio-ß-galactoside (IPTG) was added to a final concentration of 1 mmol/L, and shaking was resumed for 3 h at 37 °C. After induction the plates were centrifuged at 800g (Eppendorf Centrifuge 5810R with an A-4-62 rotor; Eppendorf) for 5 min to pellet cells, and the medium was aspirated. Plates were routinely frozen at -80 °C at this point before purification. After freezing, plates were thawed quickly in warm water immediately before purification.
We resuspended cell pellets in 100 µL of 10 mmol/L Tris (pH 7.5) by vortex-mixing the plates and repeated pipetting, and then transferred the suspension to 96-well polypropylene plates (Costar no. 3958; 1-mL assay block; Corning). We then added 100 µL of 2x lysis buffer [20 g/L sodium dodecyl sulfate (SDS) in 40 mmol/L Tris, pH 7.5] to each well, sealed the plates with a storage mat, and inverted them to mix thoroughly. The storage mats were removed, and plates were covered with adhesive tape before being floated in a covered boiling water bath for 5 min. Plates were cooled by floating in a room-temperature water bath. Samples were diluted with 150 µL of a mixture of 0.5x Lysis buffer and 150 mL/L ethanol, and 50 µL of a slurry containing Ni-NTA beads and 500 mL/L ethanol (Qiagen) was added to each well. Plates were sealed with a storage mat and inverted to mix for 10 min. Samples were transferred to a filter plate (Unifilter800, 25 MBPP; Whatman Inc.) and centrifuged 1 min at 2000 rpm. The wells were rinsed three times with 400 µL of deionized water and once with 500 mL/L acetonitrile; all steps were followed by centrifugation for 1 min at 2000 rpm. After the final rinse, plates were centrifuged 5 min at 2500g to drain completely. Plates then were allowed to dry under laminar air flow for 15 min. Were eluted samples by adding 30 µL of 500 mL/L acetonitrile–3 g/L trifluoroacetic acid to each well, waiting 1 min, then centrifuging the plates at 4000 rpm for 2 min. Samples were spotted for MALDI immediately after elution; occasionally samples were concentrated by air drying under laminar flow for 30 min and resuspended in a minimal volume of elution buffer before spotting.
maldi-tof ms
Samples were spotted by mixing 2 µL of eluate with 2 µL of a saturated solution of sinapinic acid (Fluka no. 85429; Sigma-Aldrich) in 500 mL/L acetonitrile–3 g/L trifluoroacetic acid and immediately depositing the sample on one position of a Scout384 stainless target (Bruker Daltonics). Spots were air dried at room temperature before MS.
Samples were analyzed in a Bruker Autoflex instrument operated manually in linear mode with delayed extraction. Typical operating condition were as follows: ion source 1, 20 keV; ion source 2, 18.25 keV; lens, 8.8 keV; mass deflection at 3 kDa; extraction delay, 300 ns; laser power, 50–75% as needed; 35 transients collected at five locations per spot with manual positioning; Savitsky–Golay smoothing of the data sum; digitizer rate, 500 GS/s; collection in the 4- to 20-kDa range; no baseline subtraction. Some tuning was exerted on occasion for particular analytes, depending on their mass and the mass separation between mutant and wild-type species. External standardization was performed with purified calibrants (Protein Standards I; Bruker) with quadratic fit.
generation of synthetic mutants of cftr
Wild-type CFTR exons were amplified and molecularly cloned as described above, and then subjected to in situ mutagenesis with the QuikChange reagent set (Stratagene) to introduce the mutations of the ACMG25 panel. Mutated constructs were confirmed by DNA sequencing. Purified plasmids were used as templates for second-stage PCR, and the products were mixed with wild-type product generated from known wild-type reference patient DNA to give "reconstituted" heterozygous samples for test validation.
Natural heterozygous DNA sources for 22 of 25 mutations were ultimately collected from both an established repository (Coriell Institute for Medical Research, Camden NJ) and a collaborator (S. Dobrowolski, Neo Gen Screening Inc.) and used for multiplex and pooling demonstration tests.
test validation
The wild-type DNA and reconstituted heterozygous DNA for each mutation were used to set up a large number of test wells for amplification, cloning, expression, purification, and MS as described above. In general, roughly three-fourths of a 96-well plate was dedicated to heterozygous samples and one fourth to wild-type samples. The locations of the wild-type wells were blinded until after the completion of the data interpretation. In the case of two tests (exon 7 and exon 10), the test setup included two different mutations, as well as the wild type being introduced in similarly blinded fashion.
The presence of mutation was determined qualitatively by manual interpretation of the mass spectrum of each test well relative to wild-type and mutant controls provided on each test plate. Results were scored by registering the presence and the experimental masses of wild-type and mutant peptides in each well.
screening of putative compound heterozygous patient samples
Deidentified DNA samples were obtained from patients previously determined to be positive for increased -trypsinogen and prescreened for the presence of the 
F508 mutation by fluorescence resonance energy transfer-based assay (S. Dobrowolski, personal communication). Amplification, cloning, expression, and MALDI were performed as described above.
electron microscopy
Suspended E. coli cells were pelleted and resuspended in a solution containing 20 mL/L glutaraldehyde in phosphate-buffered saline. The cells were fixed for 30 min at room temperature and stored at 4 °C for 2 days. The cells were pelleted and washed in three changes of phosphate-buffered saline (8 g/L NaCl, 0.2 g/L KCl, 1.2 g/L Na2HPO4 · 7H2O, 0.2 g/L KH2PO4). The pelleted cells were resuspended in a solution containing 10 g/L OsO4 in phosphate-buffered saline. After 30 min the cells were pelleted, washed in three changes of distilled H2O, and then dehydrated in an ascending ethanol series (500, 700, 800, and 900 mL/L, followed by three changes of absolute ethanol). Propylene oxide was used as a transitional solvent (two changes), and the sample was infiltrated overnight in equal parts of propylene oxide and epon-araldite (EA). The next day the sample was infiltrated with undiluted EA for 8 h. The sample was placed in the embedding capsules containing EA, and the EA was polymerized at 60 °C for 48 h. Thin sections (0.1 µm) were cut on a Reichert-Jung ultramicrotome. The sections were placed on 300 mesh copper grids and stained with uranyl acetate and lead citrate. Sections were viewed in a Hitachi 7100 TEM with an accelerating voltage of 50 kV, and 1 megabyte TIF images were recorded with an AMT Advantage 10 image acquisition system. A cross-grating replica grid (21 574 lines/cm) was used to determine image magnification.
multiplexing and pooling
Multiple species were typically combined for simultaneous expression analysis after separate purification of the individual second-stage PCR products, just before ligation into the expression vector. Ligation and subsequent steps were performed as described above.
Samples were pooled by amplifying exonic sequences from synthetic mutant and wild-type plasmids, after which the products were normalized for concentration by gel electrophoresis and staining of aliquots. Wild-type and mutant products were mixed at 1:1 and 10:1 ratios just before vector ligation. The mixed samples were then taken through cloning and expression as described above.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, which also shows the mass of peptide analytes generated from wild-type and mutant exonic sequences. The overall coverage of the CFTR gene in the final amplification strategy is graphically depicted in supplemental file 3 (available as part of the Data Supplement accompanying the online version of this article athttp://www.clinchem.org/content/vol49/issue8/). Of those exons expressed in forward orientation, exons 9, 11, 16, 19, 20, and 21, as well as the cryptic exon in intron 19, were expressed in the entirety of their natural forward reading frames; analyte design in these cases also included coding potential from intronic sequence flanking these exons. The inclusion of intronic flanking sequence was essential in many cases because eight of the targeted mutations affect mRNA splicing and as such are located outside exonic sequences. In the case of exon 11, a second analyte (exon 11.2) was designed in an alternative forward reading frame to detect the 1717-1 G>A mutation, but it also scanned the remainder of the exon and registered its own characteristic mass shifts resulting from G542X and R553X mutations.
|
Intentional mismatches in the annealing portion of the second-stage PCR primers were engineered to mutate natural stop codons, found in nonnatural reading frames or in intronic sequences, to create uninterrupted coding potential from an amplified fragment and permit fusion with the downstream 6xHis sequence provided by the expression vector. However, attachment to C-terminal 6xHis was not necessary for immobilized metal affinity chromatography (IMAC) capture because the histidine-rich N-terminal UE, common to all analytes in this report, was sufficient to permit nickel binding. In fact, five wild-type CFTR peptides lack C-terminal 6xHis by design, as do several mutant peptides that experience premature termination (Table 1
), and all were readily captured.
Despite the many demands and extraneous cloning information placed on the second-stage PCR primers, amplification was reproducibly robust, and cloning was straightforward. Enzymatic digestion by SfiI removed 22 bp in total, which was readily detected by gel electrophoresis of these small duplexes before and after digestion. Glass matrix purification of the small duplex products, ranging from 107 to 325 bp, yielded abundant material for ligation. The ligation with properly prepared plasmid vector DNA was efficient enough to produce thousands of transformants per test well and specific enough to permit en masse culture and expression of the transformants. Examination of the plasmid content of resulting cultures by restriction digestion or T7 amplification revealed a single species characteristic of proper addition of single exonic sequences into the expression vector. Self-ligation of the vector yielded very few transformants, and its subsequent culture and expression did not produce detectable peptide analytes. All of these criteria were routinely examined as quality assurance methods for the PMSG process.
The expression cassette of the cloning vector is illustrated in supplemental file 4 in the online Data Supplement. The plasmid was based on the pET expression vectors (6). All coding information between the NcoI and XhoI sites of the parent plasmid, pET24d, was replaced through several separate additions. The amino terminus ultimately fused to all expressed exons consisted of the following sequence (after removal of n-formyl Met), GSGTPHHTTPHHTTPHHTTPHHTTRSHTTPHHGNSPLVA, with the sequence beyond this point dictated by the exonic information inserted into the DraIII sites of the vector. This sequence is collectively referred to here as the UE, which contains four tandem overlapping repeats of the peptide sequence HTTPHH. Those exonic sequences with continuous coding potential also acquired the following carboxy terminus from the expression vector: GQRCLEHHHHHH(stop).
At the DNA level, the sequence encoding the stop codon was UAA; this was created by directed mutagenesis of the UGA codon of the original pET vector to minimize translational read-through (TGA suppression). Furthermore, additional sequence was inserted immediately after the ochre codon to create proximal ochre codons in the two remaining forward reading frames; this served to more promptly terminate translation arising from exonic frameshifting mutations, but this attribute was not necessary for the detection of any of the analytes described in this report.
For the purposes of developing the CFTR test, all 25 mutations of the ACMG panel were constructed by in situ mutagenesis on plasmid clones of the wild-type exons in the expression vector. After confirmation by peptide expression and DNA sequencing, we used these mutated plasmids to reconstitute synthetic heterozygous samples by mixing them with equal concentrations of cloned wild-type plasmid DNA. Dilutions of these plasmid mixtures were used as DNA templates for amplification, cloning, and expression during the validation exercise (see below).
Reference patient samples for 22 of 25 mutations were also obtained and used for multiplex and pooling demonstration tests. Several of the mutations (I148T, 1078T, 1898 + 1 G>A, and 2789 + 5 G>A) could be represented in only heterozygous form using synthetic mixtures. The characteristic mass signatures generated from the natural heterozygous samples through the PMSG process, demonstrated in Fig. 1
, were seen to be indistinguishable from those observed for the reconstituted samples used for the validation exercise.
|
expression and purification
The E. coli host strain NBDR [NovaBlue (DE3) containing the pRARE plasmid] was used as the bacterial host for both molecular cloning of exons and their subsequent expression. The introduction of the pRARE plasmid, which encodes genes for six rare E. coli tRNA species, was necessary for efficient expression of the mutant peptide resulting from the 1078T mutation in exon 7.1. The mutant peptide experiences a frameshift that creates four consecutive rare arginine codons, and substantial truncated peptide accumulated that was consistent with translation stalling at these residues. Introduction of the pRARE plasmid restored efficient expression of the full-length peptide. Similar stalling at tandem rare codons was observed for other exons of CFTR, and the inclusion of the pRARE plasmid likewise restored efficient expression (data not shown).
Transformation and culture conditions were successfully adapted to the microplate format. Culture of 0.6-mL volumes in 2-mL assay wells, when plates were shaken vigorously on an incline, allowed adequate aeration of cultures and attainment of high cell density in rich medium (>3 x 108 colony-forming units/mL). Plates were covered with a storage mat during shaking, which permitted air exchange but prevented cross-contamination. Comparison of various minimal and rich media specially formulated for bacterial protein expression did not reveal any that substantially improved on the yield of purified peptide when cultured in Terrific broth (data not shown). NBDR cells grown in this context grew with a generation time under 50 min to a final corrected A600 well over 3, achieving saturation within the usual 16-h overnight culture after transformation. Cultures were brought back through log phase by regrowing dilutions of the overnight cultures until the onset of stationary phase, when expression was induced with IPTG.
Significant amounts of peptide were expressed for the variety of wild-type peptide analytes in the CFTR panel; a representative quality-assurance gel is presented in supplemental file 5 in the online Data Supplement. The typical yield of purified protein was 1–30 µg/test well, depending on the analyte species. Induction for 8 h instead of 3 h further increased the yield of all analytes, but this was a modest improvement only (data not shown). Detectable amounts of the analytes were observed in the absence of induction, and the analytes varied widely with regard to the amount that accumulated in these conditions, but this "leaky" expression did not significantly inhibit growth before IPTG induction. We did not adopt the use of pLys-expressing plasmids to more tightly suppress T7 leakiness (7) because this would be incompatible with the pRARE plasmid already in use in this system to alleviate some codon bias constraints on the expression of certain analytes.
E. coli induced for peptide expression were examined by transmission electron microscopy, and the results are shown in supplemental file 6 in the online Data Supplement. Cells expressing a "highly expressing" analyte (exon 13.2A) harbored one or more sizable darkly staining aggregates or inclusion bodies; higher magnification did not reveal a regular structure (data not shown). Inclusion bodies were not observed in induced cells bearing the nonrecombinant expression vector, which is predicted to generate a 4-kDa truncated peptide, but this species is not readily recovered and detected under these purification conditions. These "empty vector" cells had some irregularity on cytoplasmic staining, but whether this represents a less concentrated coalescence of the induced peptide or is a more general result of metabolic disturbance from IPTG induction of T7 expression is unknown. In contrast, the expression host itself, a DE3 lysogen induced with IPTG but harboring no expression plasmid construct, exhibited fairly uniform staining and morphology.
maldi-tof analysis
Purified peptides were eluted from the IMAC matrix and mixed promptly with the MALDI matrix (sinapinic acid) before spotting on the target plate. Spotting was performed routinely on a volumetric basis without normalizing concentration of peptides. Detection by MS was fairly robust despite the use of such variable and sometimes excessive amounts of analyte (frequently 10–100 pmol/3-mm spot). A single parameter file was generally used to monitor all analyte species, with laser power typically sufficient to register signal quickly and consistently across a sample spot (somewhat beyond threshold irradiance). Signal-to-noise ratios varied from <20 to nearly 1000, and resolution varied from 150 to nearly 1000, with the more weakly expressed and larger analyte species frequently giving smaller values. External standardization was used to maximize the mass spectrum available for analysis in a given spectrum and for convenience of sample preparation. Of the "splits", or mass shifts generated by the 25 mutations characterized, 16 were <130 Da and 6 were <30 Da. The narrowest splits of <15 Da were still readily resolved from the wild type, although the mass shifts values are less accurately calculated from centroid masses when peaks do not have baseline resolution. Fig. 1
illustrates all 25 signature mass shifts produced by the ACMG25 mutations, as interpreted and displayed with use of our SpecView software.
In addition to the expected sinapinic acid adducts and analyte dimers and doubly charged species, several types of artifact peaks were observed in certain spectra. Several analytes copurified with a species that is 42.4 Da larger (see, as an example, panels F and M in Fig. 1
). In addition, amber and opal stop codons were subject to suppression, the former because our standard E. coli expression strain contained the supE44 amber suppressor, and the latter because of low-level endogenous UGA suppression. For example, the internal amber codon within the exon 20 sequence experienced significant suppression, giving rise to a 12.9-kDa species that corresponds to the continuation of the peptide through the C-terminal 6xHis tag sequence (Fig. 1X
); this larger product can also be readily observed by SDS-polyacrylamide gel electrophoresis (PAGE; see supplemental file 5 in the online Data Supplement). Suppression of the opal codon created by mutation R553X caused the appearance of an 8495-Da analyte, 30 Da larger than the wild-type species, consistent with endogenous Trp-inserting suppression of the UGA mutation at the original Arg codon. The efficiency of opal suppression varied from case to case; e.g., the wild-type exon 11 sequence was not observed to experience detectable suppression at its innate opal codon.
Coexpressed wild-type and mutant peptides that differed minimally in sequence were observed to give very comparable signals by MALDI, i.e., heterozygotes generated nearly equal signals for mutant and wild-type analytes when they showed a "close split", which can be seen for 16 of the spectra in Fig. 1
. This corresponded to comparable protein expression when these peptides were expressed separately and analyzed by SDS-PAGE (data not shown). Coexpression of substantially different wild-type and mutant peptides, because of nonsense or frameshift mutations, produced more variable ratios of expression judged by either criterion. Most notable was the intron 19 mutant peptide, which gave markedly higher expression by both MALDI and SDS-PAGE (Fig. 1W
)
blinded validation
To test the reliability of mutation detection, we used reconstituted mixtures of wild-type and synthetic mutant DNAs to represent heterozygous mutant DNA samples in a large number of test wells in the microplate format, interspersed with a smaller number of wells with wild type only. These plates were then taken through the entire PMSG process. The layout of the sample distribution was blinded after performance of PCR and was unknown to the purification/MALDI operator. After cloning, expression, and purification, the presence of a mutation was determined qualitatively by manual interpretation of the mass spectrum for each test well after calibration and display by SpecView software, relative to wild-type and mutant controls provided on each test plate. Spectra were scored by registering the presence and masses of wild-type and/or mutant peptides in each well. In two cases (exons 7 and 10.2), two different mutations were validated simultaneously on the same plate. The results are presented in supplemental file 7 (available the online Data Supplement).
With external standardization, the mass accuracy was generally within 5–10 Da. SD for mass determination varied with the analyte; however, the determination of the splits, or difference between wild and mutant peaks, was much more consistent. Even across the widest separations of wild-type and mutant peptides (>10 kDa), the experimentally determined splits (given as values) were accurate to within a few Daltons of the prediction. Ten of the 25 mutation tests succeeded with no false-positive or -negative results. Of the remaining tests, one to three false-positive and -negative wells were observed per plate. Aggregate error rates were 4.0% false positive (20 of 499 wild type) and 1.2% false negative (17 of 1425 mutants), but these averages include all sources of error, including trivial errors of manual sample handling, which were suspected in several cases.
screening of putative compound heterozygous patient samples
DNA samples were obtained from deidentified patient samples judged to be likely compound heterozygous on the basis of immunoassay for -trypsinogen and fluorescence resonance energy transfer-based screening for the very common F508 mutation. PMSG was performed on these samples, using the entire panel of CFTR analytes. Mass spectra were processed using SpecView software and then individually inspected for the presence of wild-type and/or mutant analytes, in comparison with known positive control samples. Results of the test for 10 patients are presented in Table 2
. Nine patients were compound heterozygotes and one was homozygous for the F508 mutation. Seven of the nine had the F508 mutation and one of six different additional mutations in exon 11, 13, or 20. The remaining two compound heterozygotes (P175 and P176) had both mutations in exon 11; the detection of apparently wild-type analyte in these samples is attributable to translational suppression of the G542X mutation. The G542X and R553X mutations registered as mass shifts in both the exon 11 and exon 11.2 analytes, as predicted. Mutations predicted on the basis of their peptide mass signatures were confirmed by DNA sequencing, with one exception. The defect detected in exon 13 for sample P166 was not the simple deletion of A at base 2184 (the ACMG mutation at this position), but rather the replacement of AA at 2183–2184 with G (8), which gives an indistinguishable mass with the present analyte design.
|
pooling and multiplexing
Several exons of the current CFTR panel were examined for their abilities to be coexpressed in multiplex combinations and to withstand a moderate amount of pooling, while not compromising the ability to detect the presence of mutant species.
We tested pooling individually for each mutation by combining amplified wild-type and mutant exonic DNAs at estimated 1:1 and 10:1 ratios (wild type:mutant) before vector ligation and carrying the mixture through the entire expression, purification, and analysis process. Results for two representative samples are shown in Fig. 2
. In both cases shown, mutant analyte was detected as a distinct peak, even in the 1:10 dilution. Results for the entire panel of CFTR mutant analytes showed various tolerances for pooling; in general those mutations giving the smallest mass shifts (<20 Da) were less tolerant because their signals more quickly merged into the larger, nearby wild-type peaks. Differing expression between wild-type and mutant peptides also affected pooling ability; for example, the 3849 + 10kB C>T mutation in the intron 19 peptide is more highly expressed than the wild type and would likely therefore tolerate a significantly higher pooling factor.
|
Shown in Fig. 3
is the multiplexed detection of 13 different CFTR analytes, which were cloned and expressed en masse in one culture, in a single spectrum. To obtain this spectrum, nine different reference patient samples, bearing heterozygous mutations in exon 3, 4.2, 10.2, or 11, were amplified for their mutated exon regions. Products were combined just before vector ligation and taken through the rest of PMSG as a single culture. As a control, the four wild-type exons were similarly amplified and taken through PMSG in the same multiplexed combination. Despite the variation in expression observed with such a multiplex setup, there was still sufficient resolution and sensitivity to detect the presence of each mutation within its expected mass range. It should be noted that combination by volume of multiple heterozygous alleles within a single exon actually constitutes twofold pooling for exons 4.2 and 10.2 and fourfold pooling of exon 11. This pooling is more noticeable in the diminution of the signal for the mutations of exon 11 relative to the wild-type analyte (e.g., mutants G551D and R560T in Fig. 3
). Of course, no single patient would be expected to carry more than two different CFTR alleles. Therefore, a spectrum of this complexity would not be encountered naturally, but the experiment nevertheless provides a demonstration of the dynamic range and sensitivity of the PMSG method.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
During test development, some analytes were encountered that accumulated in unacceptably low amounts, as indicated by their being detectable on gels only by silver staining. This did not present a major problem, however, because we found more highly expressed peptides that covered the DNA regions of interest by exploiting alternative reading frames and/or by changing the boundaries of the amplicons. The ability to dictate the extent, reading frame, and direction of expressed sequences and to suppress particular innate stop codons, all through primer design, is a fundamental strength of the PMSG system. Some exons (e.g., exons 4 and 7) were subdivided into separate analytes to adequately express and detect specific mutant peptides, and some larger exons (e.g., exons 10 and 13) were covered only in part to optimize analyte expression and size. A considerable intron sequence (e.g., exons 4 and 12) was occasionally included beyond that necessary to detect mutations at splice junctions to attain suitable expression or ideally sized analytes.
The peptide affinity purification process was developed without any special steps to promote protein stability or prevent degradation. The host strain has neither lon nor ompT mutations (9), and no protease inhibitors or nucleases were used in the purification. The purification process was straightforward: cell pellets were simply resuspended in Tris buffer, adjusted to 10 g/L SDS, and boiled, and the peptide analytes were captured by IMAC directly from diluted SDS lysates. With the entire process conducted in microplate format using multichannel and repeater pipettes, several 96-well microplates could be processed in parallel, from cell harvest to peptide elution, in roughly 3–4 h. The inherent robustness of the process should therefore lend itself well to automation in the future.
The biology of peptide expression in our system has yet to be thoroughly characterized. Because expressed peptides were purified for mass determination only and because the purification method used whole cells and exploited denaturing conditions, it was not relevant to monitor biological activity or peptide solubility. Indeed, because many of the CFTR exons are expressed from unnatural reading frames, there is no expectation per se of "proper" folding or native biological activity. Transmission electron microscopy of peptide-expressing E. coli revealed the presence of darkly staining aggregates or inclusion bodies, but this was not monitored across the entire analyte panel. Sequestration of peptides in inclusion bodies might conceivably provide some protection from proteolytic degradation, but this was not examined in the PMSG system and may not be the case in general (10)(11). Differences in CFTR peptide aggregation/accumulation might be attributable to intermolecular interaction of exon-specific peptide sequences, which do vary widely in their hydrophobicity and predicted secondary structure. Alternatively, differences in accumulation may arise from exon-specific determinants for instability, e.g., protease cleavage or other signals promoting turnover. The fact that all of our peptides share a substantial amount (>4 kDa) of a common N-terminal, highly charged UE peptide sequence, but ultimately vary in their expression, does not necessarily rule out intramolecular interactions involving UE that may control or modulate stable aggregation/accumulation.
During test design and development, smaller analytes were generally favored because smaller peptides provide better signals in MALDI MS. However, there may be a biological constraint at the expression level, because analytes <6–7 kDa were more often poorly expressed. The current design of the expression vector contributes 4.2 kDa at the amino terminus, which is common to all peptides, and 1.5 kDa at the carboxy terminus (for those sequences that do not terminate internally), which may aid in satisfying any postulated "minimum size" requirement for protein accumulation. The cloning vector itself is predicted to generate a peptide of only 4.1 kDa consisting of the N-terminal UE sequence only; the fact that this is not regularly observed to accumulate on induction of nonrecombinant control samples may be attributable to its small size or to the fact that this sequence ends in a rather hydrophobic sequence (SLPGG), which may contribute to instability (12). Replacing the N-terminal UE with a smaller 6xHis sequence generally served to decrease expression of CFTR test peptides, although certain species were less affected (data not shown).
The PMSG test described here detected all 25 CFTR mutations in a panel of synthetic targets with reasonably low rates of false results. Natural samples performed equally well, and mutations were reliably detected in reference patient samples and uncharacterized patient samples, with the latter results confirmed by DNA sequencing.
The test described here was configured to assure detection of the 25 mutations specified in the ACMG guidelines. Commercially available allele-specific oligonucleotide-based CFTR tests can register up to 87 specified mutations; such expanded mutation panels have been shown to improve the rate of detection of CFTR mutation carriers, especially in historically underrepresented ethnic populations (13). It is important to note that the PMSG test for CFTR is capable of detecting many other mutations that may lie in the regions it covers. In fact, the amplified regions (inside the primer annealing sites) cover a total of 398 known CFTR mutations (14). Of these, 358 are predicted to register changes in analyte mass of more than 9 Da and thus be detectable by the present methodology. Thus, even in its current configuration, as a research tool of modest throughput with which several dozen patients can be screened manually in 48 h, the test described provides considerable power for comprehensive monitoring of genetic variation.
Finally, because PMSG uses MALDI-TOF MS as its analytical method, samples can be both pooled and multiplexed. Although multiplexing for CFTR is provided here as a preliminary demonstration, two- to fourfold multiplexing is routine in analysis of other genes in this laboratory. Sample pooling for CFTR is likewise presented here as a representative example only. We anticipate that the consistency in sample handling that can be provided by automation will allow for standardized multiplexing and pooling in future generations of PMSG tests that can scan all exons and splice sites to detect almost all known and unknown mutations in the gene under analysis.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
G. Hum Mutat 1994;3:330-332.[CrossRef][Web of Science][Medline]
[Order article via Infotrieve]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
