Next-generation sequencing of the mitochondrial genome and association with IgA nephropathy in a renal transplant population.

Kidneys are highly aerobic organs that are critically dependent on the normal functioning of mitochondria. Genetic variations disrupting mitochondrial function are associated with multifactorial disorders including kidney disease. This study sequenced the entire mitochondrial genome in a renal transplant cohort of 64 individuals, using next-generation sequencing, to evaluate the association of genetic variants with IgA nephropathy and end-stage renal disease (ESRD, n = 100).

Mitochondrial DNA was amplified in two fragments using long-range PCR and sequenced on an Illumina Genome Analyzer II platform using TruSeq, with variants confirmed using bidirectional Sanger sequencing. A total of 427 differences from the mitochondrial DNA reference sequence were identified, of which 113 had a minor allele frequency >5% in this population. Five common SNPs revealed evidence of significant association (P ≤ 10−4) with ESRD where patients had a primary diagnosis of IgA nephropathy. Sixty-four percent of common SNPs were present in individuals with IgA nephropathy.

There is an excess burden of common mtDNA SNPs in renal transplant recipients with IgA nephropathy compared with individuals who have no evidence of renal disease. Genetic risk profiles may assist clinical stratification of renal transplant patients, and we have validated a cost-effective approach to genotype larger, carefully phenotyped cohorts.

Glomerulonephritis (GN) is the most common primary renal diagnosis in patients receiving renal replacement therapy (RRT) in the UK, accounting for 18.8% of the prevalent end-stage renal disease (ESRD) cohort in 20121. Kidneys are highly aerobic organs, and are critically dependent on mitochondrial activity. Mitochondria possess DNA that is discrete from the nuclear genome in the form of a circular, double-stranded molecule (mtDNA). The 16,569 base pair mitochondrial genome encodes 13 respiratory chain proteins, two rRNAs, and 22 tRNAs2. Genetic variations disrupting mitochondrial function are associated with increased molecular stress and multifactorial disorders including kidney disease34567891011.

Genetic variation in mtDNA may be inherited or acquired (termed somatic mutations) throughout life. MtDNA is particularly susceptible to acquiring somatic mutations as they both generate and respond to potentially mutagenic reactive oxygen species (ROS) and mtDNA lacks the protective and efficient repair systems associated with nuclear DNA. Mutations in mtDNA range from single base changes to large scale deletions. The 4977 bp “common deletion” has been associated with a number of disorders, including the renal diseases of focal segmental glomerulosclerosis (FSGS) and immunoglobin-A nephropathy (IgA nephropathy)12, the most common form of glomerulonephritis in the UK. Another frequently studied mutation is the 3243A>G single base change which is associated with MELAS (mitochondrial encephalomyopathy with lactic acidosis and stroke like episodes) and FSGS1314. MELAS can have renal involvement in the form of FSGS, but is uncommon in White European populations15.

This project sought to comprehensively investigate mitochondrial genetic variants for association with IgA nephropathy and ESRD, using a next-generation sequencing approach for the entire mitochondrial genome.

Results

There was an average of 1 million reads per DNA sample (range 36,492–3,327,485 reads) with 98% of 71 base reads aligned to the mitochondrial genome. There was >99.9% coverage of each mtDNA base with mapping quality score >40 and average coverage depth of 11,092x (Supplementary Table 1).

A total of 427 differences from the reference sequence were identified (Figure 1, Supplementary Figure 2, Supplementary Table 2), of which 55 had a minor allele frequency (MAF) of zero, leaving 372 SNPs in this population; 113 SNPs had a MAF greater than 5%. Five common SNPs revealed evidence of significant association with ESRD where patients had a primary diagnosis of IgA nephropathy (Table 1; 6419A>C, P = 3.67 × 10−6; 199T>C, P = 3.00 × 10−4; 150C>T, P = 4.00 × 10−4; 8251G>A, P = 4.00 × 10−4; 16565C>G, P = 9.00 × 10−4). The association of 6419A>C was maintained following permutation testing (n = 100,000 permutations) based on 427 variants with resultant P = 0.01. Ten SNPs with MAF less than 5% and 65 common SNPs were present only in the case group of individuals with IgA nephropathy. Strong linkage disequilibrium was not observed (Figure 2).

Figure 1

Vector NTI image of all SNPs on mitochondrial genome.

All genes are named in black text, with protein coding genes indicated by orange arrows and ribosomal RNA genes indicated by green arrows. Top five most significantly associated SNPs shown labelled in dark red. Other significant SNPs represented by dark red lines, common SNPs by blue lines, and rare variants by black lines.

Table 1

SNPs demonstrating significant association with IgA nephropathy. HV2, Hypervariable segment 2; CO1, Cytochrome c oxidase subunit I; CO2, Cytochrome c oxidase subunit II; MAF, minor allele frequency

SNP base position	Location	MAF (%)	Alleles	Case frequency (%)	Control frequency (%)	P value for IgA nephropathy (n = 32)	P value for ESRD (n = 100)
150	MT-HV2	14	C:T	25.0	3.1	4.00 × 10−4	>0.05
199	MT-HV2	9	T:C	18.8	0	3.00 × 10−4	>0.05
6419	MT-CO1	22	A:C	40.7	3.7	3.67 × 10−6	>0.05
8251	MT-CO2	14	G:A	25.0	3.1	4.00 × 10−4	>0.05
16565	Control region	17	C:G	27.6	2.4	9.00 × 10−4	>0.05

Figure 2

Linkage disequilibrium plots for confidence intervals on D' and r2 respectively where all SNPs had a minimum minor allele frequency of 5%.

Discussion

This study of 32 cases and 32 controls provides ~60% power to identify a risk variant with an odds ratio of 2.0 at a 30% minor allele frequency. The relatively small number of persons assessed in this study means it is not sufficiently powered to identify significant association with rare mtDNA variants, nor risk variants with small effect sizes, however we have identified mitochondrial SNPs that are significantly associated with IgA nephropathy, confirmed significant association following correction for multiple testing, and shown there is an excess burden of common mtDNA variants in the case group with 75 SNPs present only in individuals with IgA nephropathy. Additionally, we have developed an efficient, cost-effective protocol for comprehensive analysis of the mtDNA sequence with extensive technical validation of the experimental approach. SNPs in the mitochondrial genome are often excluded from genome-wide association studies so this study provides complementary information to ongoing genetic epidemiology research of autosomes. We have provided frequency data, the exact location of all identified SNPs, and details of the relationship between common SNPs.

From the 113 SNPs observed with a minor allele frequency greater than 5%, five common SNPs showed significant association with IgA nephropathy where P ≤ 10−4. Association of 6419A>C with IgA nephropathy was maintained following permutation testing to adjust for multiple testing bias. This transversion SNP 6419A>C is relatively common with a minor allele frequency of 22% and is located in codon 172 of the mitochondrial DNA encoded cytochrome c oxidase subunit 1 (MT-CO1) gene, creating an amino acid change from lysine to asparagine. The functional result of this change in amino acid is unknown. Cytochrome c oxidase, a component of the respiratory chain, catalyses the reduction of oxygen to water; CO1 is the catalytic subunit of the enzyme. MT-CO1 expression has been shown to be increased in patients with chronic kidney disease and hemodialysis16. This increase may be part of a compensatory response to the increased ROS generation in the uremic state, as ROS can inhibit respiratory chain enzyme activity17.

Regions of the mitochondrial genome demonstrate homology with human autosomes in the nuclear genome. It is possible that ‘SNPs' identified in this study are derived from amplified regions not in the mitochondrial genome, but sequence alignment and SNP calling was performed (automated by the software and visual confirmation of results by two researchers) blinded to case versus control status so should have affected case and control samples equally. Off-target and duplicate reads were excluded from analysis; the coverage depth and quality of alignment to the mitochondrial genome was excellent, providing confidence in the results of this study. Additionally, we explicitly enriched for mtDNA based on long range PCR, called SNPs in duplicate for 32 samples that were independently amplified, and validated SNPs (>10% heteroplasmy) using Sanger sequencing with PCR primers designed to specifically amplify mtDNA and prevent co-amplification of nuclear DNA based on the human genome reference sequence GRCh37.

This study identified a number of novel variants significantly associated with IgA nephropathy using a case-control approach. Additionally, we have genotyped an independent cohort of 50 renal transplant recipients without IgA nephropathy, compared to 50 age and gender matched controls with no evidence of kidney disease, which supports genetic association with IgA nephropathy specifically. Ideally, future studies will be conducted using next generation sequencing to evaluate a larger cohort of patients with ESRD attributed to IgA nephropathy compared to healthy controls, utilising multi-centre collaborations.

Survival analysis was not performed as this study is not optimally designed to assess patient or graft survival. However, seventeen patients returned to dialysis following graft failure in this cohort; 48 SNPs, all with MAF >5%, were present only in those individuals whose kidney transplant failed.

Mitochondria also depend on a number of nuclear genome products for their function, and further study in this area might examine both mtDNA and nuclear genes associated with mitochondrial disorders. For example, a next generation sequencing, clinically targeted assay has been developed using RainDance and Illumina approaches to examine the mitochondrial genome and 108 selected nuclear genes implicated in mitochondrial disease18. Of note, mtDNA sequences can also be extracted from non-targeted mtDNA amplification, such as exome sequencing data, which provides substantial opportunities for further mtDNA research192021.

This study utilised a cost-efficient, high throughput approach to identify novel genetic variants in mtDNA associated with IgA nephropathy. Additionally, we revealed an excess burden of mtDNA variations in renal transplant recipients with ESRD, secondary to IgA nephropathy, compared with individuals who have no evidence of renal disease. Building up genetic risk profiles may assist risk stratification of renal transplant patients, and aid the development of treatment strategies. These results are encouraging, confirming that further research is warranted in larger, carefully phenotyped cohorts.

Methods

Participants

The Regional Nephrology Unit at Belfast City Hospital is the only centre providing kidney transplant procedures in Northern Ireland. This single-centre site facilitates comprehensive analyses since DNA is available for all transplant recipient-kidney donor pairs in first deceased donor kidney transplants since 1986 with extensive prospective clinical follow-up2223. DNA was extracted from blood samples obtained prior to kidney transplantation using the salting out method23. All experiments were performed in accordance with relevant guidelines and regulations. This study had full ethical approval from the Office of Research Ethics Committees, Northern Ireland (ORECNI reference: 08/NIR03/79). Sixty-four individuals were selected for mtDNA next-generation sequencing (NGS) of the entire mitochondrial genome. This consisted of transplant recipients (cases, n = 32) with age and gender matched donors (controls, n = 32). All transplant recipients had ESRD secondary to primary IgA nephropathy proven by immunofluorescence staining of renal biopsies. A further 50 renal transplant recipients with ESRD, but without IgA nephropathy (cases), were compared to 50 age and gender matched kidney donor (control) individuals with no evidence of kidney disease from Belfast.

Next-generation sequencing

The mtDNA of all DNA samples were individually amplified using long-range PCR (Qiagen, Crawley, UK LongRange PCR Kit), in two overlapping fragments (fragment 1 = 9,289 bp; fragment 2 = 7,626 bp) [15]. Agarose gel electrophoresis (0.7%) was used to confirm the presence of PCR product. PicoGreen quantitation (Life Technologies, Warrington, UK) was employed to quantitate PCR products for fragments 1 & 2, which were then pooled in equimolar amounts for each individual and barcoded for downstream analysis. Barcoding each sample facilitates inexpensive pooling of DNA to generate sequence data, which can be later extracted to analyse each individual separately using bioinformatic tools. Samples were prepared using the Illumina TruSeq NGS protocol according to manufacturer's instructions, enriched DNA was quantitated using Qubit (Life Technologies) and sequencing was performed using an Illumina Genome Analyzer II platform. DNA from the 32 individuals in the case group was separately prepared and sequenced in duplicate for optimisation and validation of the technique; once in a 32 sample pool, and again using two pools comprising 16 samples each. Targeted next generation sequencing data was analysed for SNPs associated with IgA nephropathy in an independent case-control cohort for ESRD.

Validation by Sanger sequencing

Four samples, two renal transplant recipients and two kidney donors, underwent comprehensive Sanger sequencing of the entire mitochondrial genome to validate the NGS results derived from Illumina. Sixty-two internal sequencing primers were employed alongside nine overlapping primer pairs to amplify adequately sized fragments for Sanger sequencing2425. Bidirectional sequencing was performed on an ABI 3730 Genetic Analyser (Life Technologies)

Analysis

Initial alignment and quality control was performed using Partek Flow version 2.2.3, 64 bit architecture for pair-end FASTQ files (Partek Inc, Missouri, USA). Post-alignment quality control and analysis (QC/QA) was performed in Partek Genomics Suite version 6.6 (Partek Inc), in order to visualise the quality of aligned reads. Coverage across the mitochondrial genome was visualised using Partek Genomics Suite (Supplementary Figure 1). Partek Genomics Suite was used to call SNPs against the revised Cambridge Reference Sequence for mtDNA (rCRS, NC_012920), and an allele percentage test against the rCRS was also undertaken to evaluate heteroplasmy. Sequence data was analysed for quality control and the presence of SNPs blinded to case and control status. Concordance was analysed between duplicate samples. Case-control analysis was performed using the open source software Haploview26. Linkage disequilibrium plots for the IgA nephropathy cohort were generated using D' and r2 measures in Haploview.

Author Contributions

D.R.V., E.M.K., D.W.M. performed Illumina next generation sequencing. A.P.M. provided phenotyping expertise. A.P.D. & A.J.M. conducted sequencing and all analysis. A.P.M., A.P.D. & A.J.M. co-wrote the manuscript. All authors critically reviewed the manuscript.