Performance of mitochondrial DNA mutations detecting early stage cancer

Specimens were collected from 26 cancer patients from Johns Hopkins University Hospital after informed consent and IRB approval; and 24 non-cancer smoker controls provided blood and sputum from New York University after informed consent and IRB approval. All DNA samples were prepared from cryostat sectioned frozen cancer tissues, which had been freed of non-neoplastic cells by microdissection. Portions of the primary tumor tissue were routinely embedded in optimum cold temperature medium and the frozen specimen block was then cut with a cryostat. Initially, two 5 um sections were obtained for hematoxilyn staining and then examined by light microscopy. Sections from the tumor were reviewed carefully and the neoplastic cell content recorded as a percentage of the total cells on a section. Next several 10 um sections are cut and placed in TE9 containing 1% SDS/Protenase K overnight. Following overnight incubation, DNA was extracted and precipitated as described previously [49]. Primary tumor samples containing less than 50% neoplastic cells were further microdissected to enhance the neoplastic cell content. 80% ethanol was placed over the lightly stained tissue on the slide and non-neoplastic cell containing regions was dissected away from the tumor cells with fine tissue forceps or a 30-gauge needle using an inverted microscope. Free floating cells are then washed off the slide with 80% alcohol and the remaining regions of neoplastic cells scraped off and processed in buffer. Extracted DNA was stored at -20°C. DNA from primary tumors, body fluids and blood was evaluated. For the collection of the sputum, the study subjects inhale normal saline mist and then cough up a specimen. For the collection of the sputum, the study subjects inhale normal saline mist and then cough up a specimen. The sputum is from the airways and contains alveolar macrophages and bronchial epithelial cells. The sputum is processed for protein analysis and cells are separated by centrifugation and then placed in TRIzol^®. Bronchoalveolar lavage (BAL) is a saline wash introduced into a wedged bronchoscope and BAL cells retrieved in the saline. The cells are spun down and the BAL fluid stored (it contains 1% of epithelial lining fluid). The BAL cells are kept in TRIzol^®. Both the sputum and BAL specimens contain alveolar macrophages.

DNA from tumor sections was digested with 1% SDS/Proteinase K, extracted by phenol, chloroform and ethanol precipitated. Tumor samples were obtained from males and females. Samples were collected primarily from early stage (stage IA to IIB) tumors. We received DNA from 3 matched specimen samples from cancer patients. From 9 lung cancer patients, we received DNA from tumor, blood, and BALs (washings). From 4 bladder cancer and 14 kidney cancer patients, we received DNA from tumor, blood, and urine. Diagnosis and staging of patient tumors is provided in Table 1. In one patient sample set, DNA could not be amplified from the blood and the patient set was removed from analysis

Control samples (blood and matched sputum) were from 12 smokers who did not have cancer. Note that the control population were heavy smokers with a subset (6 of 12) having exposure to asbestos, (Table 2). Patients were considered cancer free based upon spiral CT analysis. DNA in TRIzol^® Reagent was recovered using the manufacturer's protocol (Invitrogen Corp., Carlsbad, CA).

All samples were genotyped using the PowerPlex^® 16 System (Promega Corp, Madison, WI) on a 3130xL genetic analyzer with a 36 cm capillary array and POP4 polymer and analyzed using GeneMapper^® ID v3.2 (Applied Biosystems, Foster City, CA (ABI)). Samples were diluted to 0.5 – 1.0 ng/μL and 1 μL of sample was added to a 24 μL reaction volume (18.2 μL H2O, 2.5 μL 10× buffer, 2.5 μL PowerPlex^® 16 10× primer pair mix, 0.8 μL (4 U) AmpliTaq Gold^® DNA Polymerase (ABI)), then PCR amplified using published conditions. 1 μL of ILS600 internal lane standard and 9 μL of HiDi™ Formamide (ABI) were added to 1 μL of reaction (or 1 μL Allelic Ladder Mix, one for each run) then the mix was briefly denatured and chilled to 95°C and placed on crushed ice for 3 minutes before each sequencing run. Based on STR typing, two tumor isolated mtDNA samples did not match the blood sample from the same patient, and were excluded from the data comparison.

Three primer sets resulting in amplicons of 5–6.1 kb in length were used to amplify the DNA for hybridization onto the Affymetrix Genechip^® Human Mitochondrial Resequencing Array 1.0 [50]. Together the primer sets allowed for full coverage of the mitochondrial genome. Due to low DNA quantity recovered from DNA isolation, all 12 matched controls (bloods and sputums from non cancer individials) and 8 BAL samples were pre-amplified by Whole Genome Amplification as described in our previous study [51] prior to long PCR. Amplification was conducted using the TaKaRa LA Taq™ Polymerase kit. Briefly, for the MitoChip, the mtDNA template (up to 10 μL), 0.75 μL each of forward and reverse primer (10 μM each), 0.5 μL polymerase (2.5 U), 5 μL LA PCR Buffer with MgCl₂, 8 μL dNTP mix (2.5 mM each dNTP) and 25 μL of H₂O were mixed for a total reaction volume of 50 μL. Thermal cycling conditions were as follows: pre-amplification denaturation: (1 cycle), 94°C for 2 min; amplification (30 cycles): 94°C for 15 sec; annealing and elongation, 68°C for 7 min; final elongation (1 cycle), 68°C for 12 min; 4°C hold. None of the bodily fluid samples could be amplified using the three primer pairs, presumably due to quality and/or quantity of mtDNA obtained from these samples. They were successfully amplified using the nine primer pairs previously validated for fluorescent DNA sequencing to provide full sequence coverage of the mitochondrial genome [42]. This required a slight modification in the pooling step of the MitoChip protocol to accommodate nine instead of three amplicons. Each PCR product was visualized on an agarose gel and analyzed for quality and quantity as previously described [17] or by spectrophotometric methods as described in The GeneChip^® CustomSeq^® Resquencing Array Protocol Version 2.0.

PCR clean up was conducted using the QIAquick 96 well vacuum plate manifold and protocol [52]. DNAs were eluted in 65 μL of DNAse/RNAse free water.

The GeneChip^® CustomSeq^® Resquencing Array Protocol Version 2.0 was used with a few modifications. Briefly, either three or nine amplicons representing the patient and normal control mitochondrial genomes were separately pooled at equi-molar concentrations. Three amplicons were generated from primers provided by JHMI and nine from our previous publication [50]. The PCR amplification products were pooled, fragmented, labeled, hybridized, washed, and scanned. The total quantity of DNA applied to the array was 0.62 μg. Fragmentation of the pooled DNAs was conducted using 0.15 units of Fragmentation reagent (0.033 μL) per sample at 37°C for 15 minutes followed by 95°C for 15 minutes to inactivate. The fragments were labeled with 30 units of TdT at 37°C for 90 minutes followed by 95°C for 15 minutes. The hybridization cocktail, including the separately prepared control fragments, was hybridized for 16 to 18 hours at 45°C rotating at 60 rpm. Arrays were scanned on a GeneArray^® 2500 Scanner or a GeneChip^® Scanner 3000G7 Scanner, and analyzed with GeneChip^® DNA analysis (GDAS) and GSEQ softwares.

Final analysis of all data was conducted using Affymetrix software GCOS v1.4 and GSEQ v4.0. The probe intensities for each mutation reported by the software were examined on the forward and reverse stands for every instance of that base position on the chip. Heteroplamy was identified as 2 bases with probe intensities at least 1.75× the intensity of background for a sequence position. Mutations are defined as instances where the mitochondrial population in the blood for an individual differs from that of the matched sample at a sequence position, such that the base identified in the matched sample is unique and not found as part of the blood mitochondrial population. Mutations (both homoplasmic and heteroplasmic) were confirmed and only reported when the mutation was seen on both strands for locations tiled once on the MitoChip array, and 3 of 4 strands for locations that appear on the chip twice.

Twelve samples (3 matched specimens from 4 cancer patients) were sequenced by fluorescent sequencing. Amplification was performed in two steps using two different sets of primers (9 matched and 28 matched sets). The second set was needed to obtain amplicons suitable in length for automated fluorescent sequencing. Mitochondrial DNA amplification using the primary (nine primer sets) and the secondary (28 primer sets) nested PCR was previously described [42].

An enzymatic PCR cleanup protocol was optimized for high-throughput sequencing by enzymatic cleanup using Exonuclease I and Shrimp Alkaline Phosphotase (SAP) (GE Healthcare, Fairfield, CT). To each sample, 1.5 μL of Exonuclease I (1 U/uL) and 1.5 μL of SAP (10 U/uL) was added, then incubated at 37°C for 90 min followed by 15 min at 72°C with a 4°C hold. Samples were subsequently quantified using the Caliper AMS 90 protocol previously described [42].

The non-coding region (~1000 nt) for each sample was sequenced using CE DNA sequencing, as the MitoChip contains only the coding region. The PCR amplification conditions were as published [42]. PCR amplification primers selected were reported previously to specifically amplify mitochondrial encoded DNA sequences [42, 53, 54]. Primers contained M13 tags to facilitate DNA sequencing with M13 forward and reverse sequences. Briefly, the blood, tumor, urine, BAL, and sputum mtDNAs were sequenced using the Big Dye™ Terminator (BDT) version 3.1 cycle sequencing kit (ABI). A one eighth cycle sequencing reaction was used for all sequencing. Each reaction contained 1 μL of each of the following reagents: BDT reagent, DNA (3–6 ng/μL), M13 primer (forward or reverse; 5 pmol/μL), 5× Dilution Buffer (ABI), and dH₂O to a final volume of 5 μL. Cycling sequencing conditions for forward primers were as follows: (40 cycles): 96°C for 10 sec; annealing, 50°C for 5 sec; elongation, 60°C for 4 min; 4°C hold. Reverse primers were sequenced using the same protocol, but the annealing temperature was lowered to 37°C.

The Montage™ SEQ₉₆ plate (Millipore Corp., Billerica, MA) was used for cleanup following cycle sequencing. Thirty microliters of Wash Solution (Millipore) was added to each well of the cycle sequencing plate. The samples were transferred to the clean-up plate and placed on the vacuum manifold for 5 to 20 minutes or until the wells were dry. A second wash of 30 μL Wash Solution was added and vacuumed dry for an additional 5 to 30 minutes. Once dry, 20 μL of Injection Solution (Millipore) was added to each well and the plate was mixed vigorously on a plate shaker for 10 minutes.

Resuspended samples were transferred to a 3100 Optical Plate and diluted with 15 μl of HI-DI Formamide (ABI). All separations were performed using the ABI 3130xL Genetic Analyzer using an 80 cm capillary and POP7 polymer system. Samples were electrokinetically injected (30 seconds, 1 kV) and separated at 14.6 kV. Sequences were aligned using the DNA Star SeqMan II (5.05) program and scanned for polymorphisms and sequence variants.

To determine the statistical significance of mutation and heteroplasmy rates, probability estimates were compared. To test for equivalence the confidence intervals were evaluated for equality at 95% confidence. Distribution of mutations with respect to gene length was tested by Chi square.

MitoChip analysis of the coding region combined with CE analysis of D-loop regions revealed mutations in tumor and body fluid specimens. Tumor samples revealed 18 of 24 speciemens with mutations (6/8 lung [79%]; 3/3 bladder [100%] 9/13 kidney [69%]), for a preliminary sensitivity of 75%. This is significantly higher than observed in prior studies [7]. These data also identified mutations in patient bodily fluids that differed from peripheral blood in 8 of 22 cancer patients; 3/7 BALs, 1/3 bladder urine and 4/12 kidney urine specimens (Table 1). Samples contained one or more mutation. Two of the twelve (17%) sputum specimens from heavy smokers without cancer contained one or more mitochondrial mutations (Table 2); a preliminary specificity of 83%. One specimen contained 3 mutations and the other contained one mutation. Three of these mutations are silent.

A total of 34 mutations were detected in the tumor specimens. Three of the mutations occurred in the D-loop, while the majority of the mutations were spread throughout the coding region. To evaluate whether statistically significant clustering of mutations was found, Chi square values were calculated to compare the location of mutations with the gene targets represented as a proportion of the entire mtDNA genome. For tumor mutations in all genes a nonrandom clustering was suggested for tRNAs and the ND complex genes. The percentages of mutations found within a gene in relation to the percentage of the genome occupied by that gene are; ND complex: 33.4% of the mtgenome and 47.1% of the mutations, tRNA genes: 9.1% of the mtgenome and 17.6% of the mutations, rRNA genes: 15.2% of the mtgenome and 20.6% of the mutations, D-loop: 6.8% of the mtgenome and 8.8% of the mutations, CO: 20.8% of the mtgenome and 2.9% of the mutations, and CytB: 6.9% of the mtgenome and 2.9% of the mutations. There were no mutations detected in the ATPase genes which account for 5.1% of the mtgenome. Thus, mutations more frequently occurred in specific genes, and were statistically rare in the CO genes. Eleven mutations were detected in the body fluids, and the majority of these were heteroplasmic with all mutations occurring in the coding region. The entire list of mutations detected in this study is shown in Table 3, with only purine and pyrimidine heteroplasmies found.

The reported limit of detecting mixed bases or heteroplasmies in solid tumors and body fluids using the MitoChip approaches 2% [50]. We examined the frequency of heteroplasmy in specimens from cancer patients and a control cohort. A total of 202 sequence variants (with respect to the reference sequence) were identified in the mitochondria from the blood of 12 heavy smokers without cancer (Table 4); only 7 of these (3.5%) were heteroplasmic. All seven of these were detected in both blood and sputum and are not haplotype markers. Overall, the rate of heteroplasmy in the control population was quite low; 7 nucleotides in 185,412 nucleotides sequenced, or 0.0038%. The rate of heteroplasmic calls was highest in the tumor DNAs, where 10.5% of all sequence variants detected (43 in 411) were heteroplasmic. In the cancer population, mtDNA heteroplasmies were detected in the blood of 18 of the 26 samples (69.2%) a statistically significantly higher heteroplamy rate than observed in, control blood samples where only 4 of the 12 (33.3%) contained one or more heteroplasmies (Table 4).

The MitoChip sequencing results were compared to bi-directional fluorescent CE DNA sequencing for the 3 matched specimens from 4 patients, a total of 12 samples (Table 5). The MitoChip and CE DNA sequencing results were 99.996% identical (8 differences in 198,816 nt). The percent sequence coverage for the MitoChip was 95.8% (+/-1.4), and the percent sequencing coverage for fluorescent CE sequencing was 97.5% (+/-1.2). Due to MitoChip's increased sensitivity, more mutations were identified by this method than by CE DNA sequencing. All of the 8 differences were heteroplasmic, and 7 of these were detected by the MitoChip and not CE sequencing. The 7 heteroplasmies detected by the MitoChip were in the following gene complexes: 1 CO I, 1 ATPase, 1 ND (all 3: Thr→Ala), 2 tRNA, and 2 silent mutations. The one heteroplasmy detected only by CE DNA sequencing was a silent mutation.