Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2012 May 1.
Published in final edited form as: Mitochondrion. 2011 Feb 22;11(3):520–527. doi: 10.1016/j.mito.2011.01.011

Quality Assessment of Human Mitochondrial DNA Quantification: MITONAUTS, an International Multicentre Survey

Hélène CF Côté 1, Mariana Gerschenson 2, Ulrich A Walker 3, Oscar Miro 4, Gloria Garrabou 4, Emma Hammond 5, Joan Villarroya 6, Marta Giralt 6, Francesc Villarroya 6, Paola Cinque 7, Elena Garcia-Arumi 8, Antonio L Andreu Periz 8, Marcello Pinti 9, Andrea Cossarizza 9; The MITONAUTS group
PMCID: PMC3075360  NIHMSID: NIHMS283851  PMID: 21303702

Abstract

Mitochondrial DNA quantification by qPCR is used in the context of many diseases and toxicity studies but comparison of results between laboratories is challenging. Through two multigroup distributions of DNA samples from human cell lines, the MITONAUTS group anonymously compared mtDNA/nDNA quantification across nine laboratories involved in HIV research worldwide. Eight of the nine sites showed significant correlation between them (mean raw data R2=0.664; log10-transformed data R2=0.844). Although mtDNA/nDNA values were well correlated between sites, the inter-site variability on the absolute measurements remained high with a mean (range) coefficient of variation of 71 (37–212)%. Some variability appeared cell line-specific, probably due to chromosomal alterations or pseudogenes affecting the quantification of certain genes, while within cell line variability was likely due to differences in calibration of the standard curves. The use of two mtDNA and two single copy nDNA genes with highly specific primers to quantify each genome would help address copy number variants. Our results indicate that sample shipment must be done frozen and that absolute mtDNA/nDNA ratio values cannot readily be compared between laboratories, especially if assessing cultured cell mtDNA content. However, within laboratory and relative mtDNA/nDNA comparisons between laboratories should be reliable.

Keywords: Inter-laboratory variability, mtDNA content by qPCR

1. Introduction

Mammalian mitochondria contain their own genome, a circular 16.5 kb mitochondrial DNA (mtDNA) that encodes genes for 13 polypeptides, 22 tRNA, and 2 rRNA. Mitochondrial DNA is replicated by human polymerase γ, and the amount of mtDNA per cell can vary according to biogenesis and retrograde regulation. This regulation is affected by cell type and cellular energy demands, but can also be influenced by mitochondrial disease or dysfunction, acquired drug-related mitochondrial toxicity (Gerschenson and Brinkman 2004), and oxidative stress from various sources such as aging, cancer, and smoking (Cote 2005; Masayesva, Mambo et al. 2006; Higuchi 2007; Copeland 2008). Quantification of the relative ratio between mtDNA and nuclear DNA, the latter usually assumed to remain constant in human tissue, is therefore relevant to the study of many diseases and conditions, using either clinical, animal or cultured cell derived samples.

In 2005, representatives from 18 research groups around the world mostly involved in HIV drug toxicity research met for the first technical meeting of mtDNA researchers in Boston. During the meeting, methodologies were shared and the usefulness and standardization of mtDNA quantification between laboratories were discussed. Later that year, during a second meeting of the same group in Dublin, it was agreed that mtDNA quantity should be expressed as mtDNA/nDNA ratio as opposed to mtDNA copies per cell as few assays actually count cells but rather assume 2 copies of nDNA per cell, which is not true for all human tissues. The term MITONAUTS, standing for MITOchondria Network for Assay Utilization and Technique Standardization was coined and the present study designed, to compare mtDNA quantification between laboratories. The goal of this study was to assess the concordance between laboratories that quantify mtDNA using varied quantitative PCR assays and to assess how shipping affected the values.

2. Materials and methods

For ethical and international shipping regulation issues, we elected to use DNA extracted from human cell lines as opposed to human clinical samples. This presented definite advantages but also raised some comparison issues as discussed later.

2.1 DNA preparation

Table 1 summarizes the source of the human DNA samples. For the first shipment, total DNA was extracted from cultured human cells (see Table 1, left column) using QiaAmp DNA midi kit (Qiagen, Hilden, Germany). The DNA was resuspended in Tris-EDTA buffer and aliquoted (50 μL per tube). For sample #9, a larger volume (200 μL) of DNA was provided, to be used as internal control in future experiments. The first shipment samples' DNA concentration ranged from 57 to 150 ng/mL.

Table 1.

Description of the two shipments: the cell lines from which the DNA samples were extracted and the variability (CV) of the mtDNA/nDNA ratio values provided for each sample by the N participating sites.

Shipment # 1 (room temperature) Raw data Log-transformed data
sample Cell line Cell Type # sites mtDNA/nDNA Range CV mtDNA/nDNA Range CV
(N) (mean ± SD) (%) (mean ± SD) (%)
1 BKT-143 Osteosarcoma 8 337 ± 295 32–885 88 2.89 ± 0.23 2.69–3.39 8
2 A.301 CD4 T cell line 8 387 ± 348 47–1201 90 2.67 ± 0.43 2.04–3.37 16
3 CEM-1a Acute T lymphoblastic leukemia 8 418 ± 440 25–1310 105 2.69 ± 0.42 1.95–3.26 16
4 CEM-2 Acute T lymphoblastic leukemia 8 628 ± 655 90–2091 104 2.61 ± 0.47 1.80–3.26 18
5 CEM-3 Acute T lymphoblastic leukemia 8 644 ± 657 63–1835 102 2.64 ± 0.47 1.74–3.26 18
6 CEM-4 Acute T lymphoblastic leukemia 8 667 ± 623 55–1814 94 2.63 ± 0.41 1.95–3.32 16
7 HepG2 Hepatocellular carcinoma 8 711 ± 645 89–1806 91 3.07 ± 0.28 2.81–3.53 9
8 HL60 Promyelocytic leukemia 8 719 ± 550 158–1906 76 2.70 ± 0.46 1.94–3.39 17
9 HL60 Promyelocytic leukemia 8 740 ± 797 109–2328 108 2.67 ± 0.47 2.12–3.39 18
10 HUT78 T cell lymphoma 8 797 ± 689 309–2418 86 2.80 ± 0.29 2.49–3.38 10
11 HUT78 T cell lymphoma 8 801 ± 913 133–2447 114 2.75 ± 0.33 2.20–3.28 12
12 K562 Erythromyeloblastoid leukemia 8 803 ± 842 87–2435 105 2.85 ± 0.37 2.26–3.41 13
13 MCF7.2 Breast cancer 8 859 ± 478 396–1887 56 2.61 ± 0.62 1.64–3.58 24
14 MCF7.2 Breast cancer 8 878 ± 436 305–1821 50 2.67 ± 0.53 1.91–3.56 20
15 Molt-4 Acute T lymphoblastic leukemia 8 899 ± 658 494–2461 73 2.90 ± 0.22 2.48–3.26 8
16 Molt-4 Acute T lymphoblastic leukemia 8 940 ± 1257 80–3634 134 2.88 ± 0.22 2.60–3.28 7
17 PBMC Peripheral blood mononuclear cells 8 972 ± 1377 44–3796 142 2.40 ± 0.52 1.40–3.12 21
18 PBMC Peripheral blood mononuclear cells 8 983 ± 892 180–2560 91 2.36 ± 0.45 1.50–2.95 19
19 U937 Monocytic leukemia 8 1457 ± 1103 649–3421 76 2.45 ± 0.39 1.67–3.08 16
Shipment # 2 (dry ice) Raw data Log-transformed data
sample Cell line Cell Type mtDNA/nDNA Range CV mtDNA/nDNA Range CV
(N) (mean ± SD) (%) (mean ± SD) (%)
1 K562b Erythromyeloblastoid leukemia 8 832 ± 322 240–1245 39 2.88 ± 0.23 2.38–3.10 7.9
2 K562 Erythromyeloblastoid leukemia 8 1287 ± 830 359–3046 64 3.03 ± 0.28 2.56–3.48 9.2
3 CRL 2061c Fibroblast rhabdomyosarcoma 7 51 ± 26 21–101 51 1.66 ± 0.22 1.33–2.00 13
4 CRL 2061 Fibroblast rhabdomyosarcoma 8 268 ± 568 32–1671 212 1.96 ± 0.55 1.50–3.22 28
5 K562 Erythromyeloblastoid leukemia 8 805 ± 383 253–1273 48 2.85 ± 0.24 2.40–3.10 8.5
6 HEK 293 Embryonic Kidney 7 1659 ± 1280 315–3717 77 3.09 ± 0.39 2.50–3.57 13
7 K562 Erythromyeloblastoid leukemia 8 786 ± 357 304–1431 45 2.85 ± 0.22 2.48–3.16 7.6
8 K562 Erythromyeloblastoid leukemia 8 797 ± 318 237–1247 40 2.86 ± 0.23 2.38–3.10 7.9
9 K562 Erythromyeloblastoid leukemia 8 680 ± 298 296–1161 44 2.79 ± 0.22 2.47–3.06 7.8
10 K562 Erythromyeloblastoid leukemia 8 790 ± 453 240–1661 57 2.83 ± 0.27 2.38–3.22 9.4
11 CRL 2061 Fibroblast rhabdomyosarcoma 7 45 ± 17 20–66 37 1.62 ± 0.19 1.30–1.82 12
12 TF-1 Erythroleukemia 7 415 ± 196 151–658 47 2.56 ± 0.25 2.18–2.82 10
13 K562 Erythromyeloblastoid leukemia 8 1167 ± 614 375–2210 53 3.01 ± 0.25 2.57–3.34 8.4
14 K562 Erythromyeloblastoid leukemia 8 949 ± 649 275–2331 68 2.90 ± 0.27 2.44–3.37 9.4
15 K562 Erythromyeloblastoid leukemia 8 669 ± 321 211–1085 48 2.77 ± 0.25 2.32–3.04 9.2
16 Panc-1 Pancreatic carcinoma 7 1324 ± 590 340–1801 45 3.06 ± 0.28 2.53–3.26 9.1
17 CRL 2061 Fibroblast rhabdomyosarcoma 8 354 ± 723 47–2140 204 2.13 ± 0.52 1.68–3.33 24
18 K562 Erythromyeloblastoid leukemia 8 2873 ± 3293 720–10814 115 3.30 ± 0.36 2.86–4.03 11
19d HEK 293 Embryonic Kidney 7 1639 ± 1209 320–3588 74 3.10 ± 0.37 2.51–3.55 12
20e TF-1 Erythroleukemia 7 429 ± 199 151–621 46 2.58 ± 0.25 2.18–2.79 10
Open in a new tab
a

CEM 1–4 were derived from the same cell line exposed to NRTIs for 7 days (Galluzzi, Pinti et al. 2005)

b

the K562 samples were all derived from the same cell line exposed to various concentrations of zidovudine or stavudine for several weeks (Papp, Gadawski et al. 2008)

c

the three CRL 2061 samples were derived from the same primary cell line differentiated into muscle cells and exposed to 0.1 μM simvastatin (personal communication from Cote)

d

sample #19 is a duplicate of sample #6

e

sample #20 is a duplicate of sample #12

For the second shipment, DNA was also extracted from cultured human cells (see Table 1, right column) using the QiaAmp DNA mini kit (QIAGEN) and resuspended in Tris-EDTA buffer. The samples'DNA concentration ranged from 11 to 67 ng/μL.

In several instances, samples were prepared by treating a single cell line with drugs that modulate mtDNA content. For example, in the first shipment, four samples were derived from CEM cells exposed to nucleoside reverse transcriptase inhibitors (NRTI) for 7 days (Galluzzi, Pinti et al. 2005) while in the second shipment, 11 of the 20 samples were DNA extracted from K562 cells exposed to the NRTI zidovudine or stavudine (Papp, Gadawski et al. 2008).

2.2 Shipping

For the first shipment, two identical sets of 19 DNA samples were shipped by courier (DHL) to each participating laboratory from Modena, Italy, at room temperature. Each site was asked to ship one set back to the sender, also at room temperature to evaluate if shipping added to variability. For the second shipment, a single set of 20 DNA samples was shipped on dry ice from Vancouver, Canada, by FEDEX.

2.3 MtDNA quantification assays

Each site used its own mtDNA quantification assay methodology and reagents. Details on the methods used are presented in Table 2, in alphabetical order (unrelated to the order of the other result tables). One site (Barcelona I) used a different nuclear gene when assaying the second shipment as the gene typically used to quantify mtDNA depletion in human clinical samples yielded different results in cell line-derived samples. It was agreed that the data would remain anonymous. For this study, the free software Primer-BLAST (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) was used to blast the human genome with each set of primer against its intended target, under 55°C to 63°C PCR conditions. The size of the amplified fragment and the likelihood of amplifying unintended targets with each primer pair, based on the Primer-BLAST results, are reported in Table 2.

Table 2.

Characteristics of the various assays used by participating sites. The sites are listed in alphabetical order according to city name and the order does not correspond to the site # used throughout this report.

Site (alphabetical order) Mitochondrial gene Size (bp) Specific to single intended target? Nuclear gene Size (bp) Specific to single intended target? detection Instrument Reference
Barcelona I NADH dehydrogenase, subunit 2 (ND2) 235 Likely (mismatches position 2&7 from 3' end) 18S rRNA 531 Unlikely (several targets with single mismatch position 8 or more from 3' end) SYBR green LightCycler 1.5 (Lopez, Miro et al. 2004)
Barcelona I NADH dehydrogenase, subunit 2 (ND2) 235 Likely (mismatches position 2&7 from 3' end) RNA polymerase II 632 Likely (targets larger with mismatch position 6 or less from 3' end) SYBR green LightCycler 1.5 (Radonic, Thulke et al. 2004)
Barcelona II Cytochrome c oxidase subunit II (CCOII) 91 Yes CCAAT/enhancer binding protein-alpha or TFAM? (commercial ABI kit) N/A Commercial primer sequence not available Fluorescent probes ABI Prism 7700 (Vidal, Domingo etal. 2006)
Barcelona III 12S RNA 122 Likely (mismatch at 3' end) PDARs, RNAse P (commercial ABI kit) 86 Commercial primer sequence not available Fluorescent probes ABI Prism 7500 (Andreu, Martinez etal. 2009)
Freiburg ATPase subunit VI (ATP6) 79 Yes GAPDH exon 8 63 Likely (targets identified with mismatches position 9 or more from 3' end) Fluorescent probes ABI Prism 7700 (Setzer, Schlesier et al. 2005)
Honolulu NADH dehydrogenas, subunit 2 (ND2) 90 Yes Fas Ligand (FL) 95 Yes SYBR green LightCycler 480 (Gerschen son, Shiramizu etal. 2005)
Milan Cytochrome b 73 Likely (mismatch at 3' end) Chemokine (C-C motif) receptor 2 (CCR2) 66 Likely (targets identified with mismatches position 5 or more from 3' end) Fluorescent probes ABI 7900 Personal communication
Modena NADH dehydrogenase subunit 2 (ND2) 90 Yes Fas Ligand (FL) 95 Yes Fluorescent probes BioRadiCycler (Cossarizz a, Riva et al. 2003)
Perth mtDNA (1592–1675) (mostly tRNAVal) 84 Yes Human growth hormone (HGH) 100 Yes Fluorescent probes ABI Prism 7700 (Nolan, Hammond etal. 2003)
Vancouver Cytochrome c oxidase subunit I (CCOI) 197 Likely (mismatch position 4 from 3' end) Polymerase gamma accessory subunit (ASPG or POLG2) 186 Likely (mismatch(es) at 3' end) Fluorescent probes LightCycler 480 (Cote, Raboud et al. 2008)
Open in a new tab

2.4 Statistical considerations

For statistical analyses, mtDNA/nDNA values were compared using Pearson's correlations (XLstat 2009). For correlations, data from all sites were included. However, when analyzing variability between sites, data from site #2 were omitted since that site reported relative mtDNA/nDNA content and not the absolute ratio as for the other sites. Statistical analyses were performed on both raw and log10-transformed data due to the wide variability of the data.

3. Results

3.1 First shipment at room temperature

3.1.1 Sample mtDNA/nDNA stability

Globally, eighteen laboratories initially participated in this exercise and were sent two sets of 19 DNA samples extracted from 12 distinct human derived cell lines (Table 1) from Modena, at room temperature. Of those 18 sites, 11 shipped back one set of samples that were stored frozen until all shipments were received. The mtDNA content of each returned sample was then assayed by the Modena laboratory and compared (Pearson's correlation) with the values obtained for the set that never left Modena. As seen in Table 3, ten of the eleven returned sets of samples gave values that were generally higher than those of the reference set, with the traveling set showing an average change in mtDNA/nDNA ratio of +88% compared to the non-traveling set. Of note, the two sets of sample showing the lowest correlations between the reference mtDNA/nDNA ratio measured by Modena and the returned set of samples (Table 3) also happened to be those that traveled the longest distance.

Table 3.

Effect of shipping back and forth at room temperature on mtDNA/nDNA quantification: correlations between mtDNA/nDNA values measured in Modena for sample sets shipped from each site back to Modena and values determined by the Modena site for their set of samples.

Site Modena 1 2 3 4 5 6 7 8 9 10 11
Samplesa (N) 19 18 18 19 19 18 19 17 18 18 17 19
mtDNA/nDNA
  Range 25–775 53–966 95–1216 65–573 135–1689 157–2063 193–1731 93–743 53–395 76–699 68–644 88–2077
  % change N/A +40 +118 +0.6 +165 +244 +200 +68 −27 +26 +16 +118
Pearson's b
  R2 1.0 0.93 0.90 0.62 0.85 0.62 0.82 0.78 0.83 0.68 0.84 0.72
  Slope 1.0 1.07 1.45 0.42 1.71 1.56 1.66 0.83 0.36 0.55 0.72 1.69
  p value --- <0.0001 <0.0001 <0.0001 <0.0001 0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
Open in a new tab
a

In some instances, the assay did not meet the assay quality control and no value was generated.

b

The correlations were between the measurements made by the Modena laboratory on the set of samples assigned to their site and each of the sample sets shipped back from the participating sites.

3.1.2 MtDNA/nDNA concordance between sites

Eight sites submitted mtDNA/nDNA ratio data for the 19 samples. Data from one site (#2) were expressed as relative rather than absolute mtDNA/nDNA ratios with values approximately 300 times lower than all others. For that reason these data were only included in correlation analyses. Using both raw and log10-transformed data, results from each site were correlated to those of the other 8 sites individually (Table 4A). In this one on one comparison between each of the participating sites, five sites showed good correlations between them (#1, 3, 6, 8 and 9, all p<0.0001) while site #7 showed weaker correlation with those same five sites (raw data R2≥0.272, p≤0.022; log10 R2=0.250, p≤0.029). Three sites (#2, 4 and 5) showed poor correlation with the other sites with the exception that site 2 showed a strong correlation with site 4 (raw data R2=0.485, p=0.001; log10 R2=0.479, p=0.001) and a weak one with site 7 (raw data R2=0.266, p=0.024; log10 R2=0.204, p=0.052). This discordance was greatly ameliorated by excluding samples extracted from the Molt-4 cell line (samples 15 and 16), although site #5 remained poorly correlated to others (Table 4B).

Table 4A.

Pearson's correlations for the mtDNA/nDNA values obtained by each sites for the first shipment at room temperature for all samples (N=19). Correlations of raw data are shown against clear background while those of log10-transformed data are shown against a grey background. Bold indicates no significant correlation.

First shipment at room temperature, Pearson's correlation
Site 1 2 3 4 5 6 7 8 9
1 - R2=0.018
p=0.579 		R2=0.729
p<0.0001		 R2=0.022
p=0.543 		R2=0.069
p=0.276 		R2=0.746
p<0.0001		 R2=0.250
p=0.029		 R2=0.910
p<0.0001		 R2=0.835
p<0.0001
2 R2=0.175
p=0.074 		- R2=0.039
p=0.419 		R2=0.479
p=0.001		 R2=0.0004
p=0.936 		R2=0.007
p=0.732 		R2=0.204
p=0.052 		R2=0.014
p=0.626 		R2=0.004
p=0.806
3 R2=0.559
p<0.001		 R2=0.116
p=0.153 		- R2=0.004
p=0.793 		R2=0.029
p=0.487 		R2=0.637
p<0.0001		 R2=0.483
p=0.001		 R2=0.729
p<0.0001		 R2=0.667
p<0.0001
4 R2=0.135
p=0.122 		R2=0.485
p=0.001		 R2=0.029
p=0.484 		- R2=0.095
p=0.199 		R2=0.0001
p=0.969 		R2=0.002
p=0.857 		R2=0.022
p=0.548 		R2=0.014
p=0.633
5 R2=0.005
p=0.780 		R2=0.010
p=0.687 		R2=0.006
p=0.753 		R2=0.152
p=0.098 		- R2=0107
p=0.171 		R2=0.004
p=0.803 		R2=0.145
p=0.107 		R2=0.257
p=0.027
6 R2=0.515
p=0.001		 R2=0.025
p=0.522 		R2=0.638
p<0.0001		 R2=0.0003
p=0.983 		R2=0.008
p=0.766 		- R2=0.321
p=0.011		 R2=0.744
p<0.0001		 R2=0.864
p<0.0001
7 R2=0.272
p=0.022		 R2=0.266
p=0.024		 R2=0.652
p<0.0001		 R2=0.026
p=0.512 		R2=0.001
p=0.927 		R2=0.331
p=0.010		 - R2=0.288
p=0.018		 R2=0.290
p=0.017
8 R2=0.864
p<0.0001		 R2=0.135
p=0.122 		R2=0.744
p<0.0001		 R2=0.096
p=0.197 		R2=0.058
p=0.323 		R2=0.584
p=0.0001		 R2=0.412
p=0.003		 - R2=0.859
p<0.0001
9 R2=0.635
p<0.0001		 R2=0.039
p=0.421 		R2=0.764
p<0.0001		 R2=0.028
p=0.495 		R2=0.138
p=0.118 		R2=0.823
p<0.0001		 R2=0.419
p=0.003		 R2=0.794
p<0.0001		 -
Open in a new tab
Table 4B.

Pearson's correlations for the mtDNA/nDNA values obtained by each sites for the first shipment at room temperature for all samples except samples 15 and 16 (N=17). Correlations of raw data are shown against clear background while those of log10-transformed data are shown against a grey background. Bold indicates no significant correlation.

First shipment at room temperature, Pearson's correlation (minus samples 15 and 16)
Site 1 2 3 4 5 6 7 8 9
1 - R2=0.103
p=0.210 		R2=0.694
p<0.0001		 R2=0.169
p=0.101 		R2=0.063
p=0.329 		R2=0.808
p<0.0001		 R2=0.215
p=0.061 		R2=0.903
p<0.0001		 R2=0.830
p<0.0001
2 R2=0.288
p=0.026		 - R2=0.181
p=0.089 		R2=0.396
p=0.007		 R2=0.003
p=0.837 		R2=0.255
p=0.039		 R2=0.383
p=0.008		 R2=0.082
p=0.266 		R2=0.103
p=0.209
3 R2=0.571
p<0.001		 R2=0.348
p=0.013		 - R2=0.117
p=0.180 		R2=0.023
p=0.563 		R2=0.603
p<0.001		 R2=0.448
p=0.003		 R2=0.693
p<0.0001		 R2=0.593
P=0.0003
4 R2=0.289
p=0.026		 R2=0.430
p=0.004		 R2=0.222
p=0.056 		- R2=0.160
p=0.111 		R2=0.264
p=0.035		 R2=0.039
p=0.450 		R2=0.147
p=0.128 		R2=0.245
p=0.044
5 R2=0.005
p=0.784 		R2=0.010
p=0.698 		R2=0.009
p=0.721 		R2=0.184
p=0.086 		- R2=141
p=0.138 		R2=0.002
p=0.866 		R2=0.143
p=0.135 		R2=0.299
p=0.023
6 R2=0.795
p<0.0001		 R2=0.495
p=0.002		 R2=0.672
p<0.0001		 R2=0.397
p=0.007		 R2=0.096
p=0.226 		- R2=0.330
p=0.016		 R2=0.846
p<0.0001		 R2=0.859
p<0.0001
7 R2=0.270
p=0.032		 R2=0.478
p=0.002		 R2=0.635
p<0.0001		 R2=0.113
p=0.188 		R2=0.001
p=0.911 		R2=0.436
p=0.004		 - R2=0.246
p=0.043		 R2=0.242
p=0.045
8 R2=0.867
p<0.0001		 R2=0.255
p=0.039		 R2=0.743
p<0.0001		 R2=0.252
p=0.040		 R2=0.064
p=0.326 		R2=0.866
p=0.0001		 R2=0.383
p=0.008		 - R2=0.869
p<0.0001
9 R2=0.714
p<0.0001		 R2=0.249
p=0.042		 R2=0.697
p<0.0001		 R2=0.330
p=0.016		 R2=0.218
p=0.059 		R2=0.843
p<0.0001		 R2=0.383
p=0.008		 R2=0.867
p<0.0001		 -
Open in a new tab

3.1.3 MtDNA/nDNA measurement variability

The mean values and the inter-site coefficient of variation (%CV=mean*100%/standard deviation (SD) were calculated for each sample shipped (Table 1). For this calculation, site #2 was omitted since their data was on a relative scale. The average CV mean ± SD (range) for all samples was (raw data 94 ± 23 (50–142)%; log10 15.0 ± 4.8 (7.5–23.6)%). This decreased to (raw data 44 ± 8 (32–55)%, log10 10.7 ± 4.2 (4.1–18.1)%) if data from sites 4 and 5 were also omitted. There was no relationship between the samples' DNA concentration and their inter-site CV. This remained true with or without sites 2, 4 and 5.

3.2 Second shipment on dry ice

Sets of 20 DNA samples on dry ice were sent from Vancouver to eight laboratories, all were confirmed to have arrived still frozen except the shipment to Australia that was cold. Each site determined the mtDNA content of the samples, expressed as the mtDNA/nDNA ratio, and sent data back. As before, site 2 data were on a relative scale rather than an absolute one. While assaying samples from the second shipment, two sites noticed that, for some samples, their mtDNA/nDNA measurements showed gene-dependent variability and accordingly, sent back results that they considered reliable for 11/20 and 13/20 samples, respectively.

3.2.1 Concordance between duplicate samples

Within the 20 samples, two were present in duplicate (#6 was a duplicate of #19 and #12 of #20) (Table 1), something that was not known by the participants. Seven sites provided data for these samples. The absolute % difference between the duplicates (Δ between duplicates*100%/mean of duplicates) was calculated for each pair and averaged. Results (mean % difference ± SD (range) raw 11.7 ± 7.4% (0.8–26.3%); log10 1.8 ± 1.3% (0.2–4.2%) indicated generally good concordance between duplicates as six out of seven sites showed less than 15% (raw data) difference between duplicates.

3.2.2 MtDNA/nDNA concordance between sites

As before, for all 20 samples, results from each site were correlated to those of the other 8 sites individually, using both raw and log10-transformed data (Table 5A). This one on one comparison between the sites revealed that all sites showed good correlation between them except one site (#4) that showed generally poor correlation with most sites. However, site #4 was strongly correlated with site #2 (raw data R2=0.872, p<0.0001; log10 R2=0.736, p=0.001) and weakly so with site #1 (raw data R2=0.414, p=0.018; log10 R2=0.001, p=0.906). This poor correlation appeared driven in large part by two samples derived from the CRL 2061 cell line.

Table 5A.

Pearson's correlations for the mtDNA/nDNA values obtained by each sites for the second shipment on dry ice (N=20*). Correlations of raw data are shown against clear background while those of log10-transformed data are shown against a grey background. Bold indicates no significant correlation.

Second shipment on dry ice, Pearson's correlation
Site 1 2 3 4 5 6 7 8 9
1 - R2=0.646
p=0.003		 R2=0.917
p<0.0001		 R2=0.001
p=0.906 		R2=0.892
p<0.0001		 R2=0.885
p<0.0001		 R2=0.879
p<0.0001		 R2=0.901
p<0.0001		 R2=0.925
p<0.0001
2 R2=0.787
p<0.001		 - R2=0.780
p=0.0003		 R2=0.736
p=0.001		 R2=0.299
p=0.082 		R2=0.747
p=0.001		 R2=0.326
p=0.067 		R2=0.562
p=0.008		 R2=0.855
p<0.0001
3 R2=0.560
p=0.0001		 R2=0.873
p<0.0001		 - R2=0.007
p=0.782 		R2=0.936
p<0.0001		 R2=0.962
p<0.0001		 R2=0.968
p<0.0001		 R2=0.975
p<0.0001		 R2=0.982
p<0.0001
4 R2=0.414
p=0.018		 R2=0.872
p<0.0001		 R2=0.178
p=0.151 		- R2=0.096
p=0.302 		R2=0.054
p=0.445 		R2=0.045
p=0.485 		R2=0.024
p=0.612 		R2=0.009
p=0.759
5 R2=0.352
p=0.006		 R2=0.478
p=0.019		 R2=0.673
p<0.0001		 R2=0.001
p=0.941 		- R2=0.913
p<0.0001		 R2=0.926
p<0.0001		 R2=0.951
p<0.0001		 R2=0.927
p<0.0001
6 R2=0.650
p<0.0001		 R2=0.835
p<0.0001		 R2=0.709
p<0.0001		 R2=0.076
p=0.363 		R2=0.457
p=0.001		 - R2=0.865
p<0.0001		 R2=0.902
p<0.0001		 R2=0.973
p<0.0001
7 R2=0.351
p=0.006		 R2=0.362
p=0.050		 R2=0.840
p<0.0001		 R2=0.022
p=0.625 		R2=0.659
p<0.0001		 R2=0.521
p=0.0003		 - R2=0.916
p<0.0001		 R2=0.893
p<0.0001
8 R2=0.557
p=0.0002		 R2=0.709
p=0.001		 R2=0.817
p<0.0001		 R2=0.107
p=0.276 		R2=0.831
p<0.0001		 R2=0.622
p<0.0001		 R2=0.733
p<0.0001		 - R2=0.937
p<0.0001
9 R2=0.766
p<0.0001		 R2=0.917
p<0.0001		 R2=0.806
p<0.0001		 R2=0.198
p=0.128 		R2=0.505
p=0.0004		 R2=0.934
p<0.0001		 R2=0.559
p=0.0002		 R2=0.711
p<0.0001		 -
Open in a new tab
*

N=20 except for sites #2 for which N=11, and site 4, for which N=13.

Among the 20 samples, eleven were derived from the same cell line (K562) that had been cultured in the presence of thymidine analogues to alter the mtDNA content. The limited data sets provided by two laboratories both included values for all eleven K562 samples. If only the K562 samples were considered, the correlation between the 9 sites was generally more uniform (mean [range] R2 = raw data 0.69 [0.29–0.94]; log10 0.61 [0.19–0.88]), despite a tendency toward lower R2 values given the reduced sample size (N=11 instead of 20) (Table 5B). Notably, when all samples compared were derived from the same cell line, site #4 showed much improved correlations with the other sites.

Table 5B.

Pearson's correlation (R2) and p values between mtDNA/nDNA values obtained by individual sites for the second shipment on dry ice, considering only the samples derived from the K562 cell line (N=11). Correlations of raw data are shown against clear background while those of log10-transformed data are shown against a grey background. Bold indicates no significant correlation.

Second shipment on dry ice, K562 DNA only, Pearson's correlation
Site 1 2 3 4 5 6 7 8 9
1 - R2=0.646
p=0.003		 R2=0.796
p=0.0002		 R2=0.567
p=0.008		 R2=0.423
p=0.030		 R2=0.613
p=0.004		 R2=0.487
p=0.017		 R2=0.469
p=0.020		 R2=0.693
p=0.001
2 R2=0.787
p=0.0003		 - R2=0.780
p=0.0003		 R2=0.736
p=0.001		 R2=0.299
p=0.082 		R2=0.747
p=0.001		 R2=0.326
p=0.067 		R2=0.562
p=0.008		 R2=0.855
p<0.0001
3 R2=0.901
p<0.0001		 R2=0.873
p<0.0001		 - R2=0.726
p=0.001		 R2=0.561
p=0.008		 R2=0.799
p=0.0002		 R2=0.488
p=0.017		 R2=0.677
p=0.002		 R2=0.880
p<0.0001
4 R2=0.706
p=0.001		 R2=0.872
p<0.0001		 R2=0.854
p<0.0001		 - R2=0.249
p=0.119 		R2=0.479
p=0.018		 R2=0.185
p=0.187 		R2=0.450
p=0.024		 R2=0.656
p=0.002
5 R2=0.597
p=0.005		 R2=0.478
p=0.019		 R2=0.664
p=0.002		 R2=0.451
p=0.024		 - R2=0.431
p=0.028		 R2=0.368
p=0.048		 R2=0.431
p=0.028		 R2=0.567
p=0.007
6 R2=0.901
p<0.0001		 R2=0.835
p<0.0001		 R2=0.879
p<0.0001		 R2=0.669
p=0.002		 R2=0.541
p=0.010		 - R2=0.413
p=0.033		 R2=0.644
p=0.003		 R2=0.867
p<0.0001
7 R2=0.431
p=0.028		 R2=0.362
p=0.050		 R2=0.482
p=0.018		 R2=0.288
p=0.089 		R2=0.413
P=0.033		 R2=0.451
p=0.024		 - R2=0.407
p=0.035		 R2=0.314
p=0.073
8 R2=0.738
p=0.001		 R2=0.709
p=0.001		 R2=0.822
p=0.0001		 R2=0.641
p=0.003		 R2=0.539
p=0.010		 R2=0.771
p=0.0004		 R2=0.494
p=0.016		 - R2=0.662
p=0.002
9 R2=0.910
p<0.0001		 R2=0.917
p<0.0001		 R2=0.941
p<0.0001		 R2=0.805
p=0.0002		 R2=0.646
p=0.003		 R2=0.934
p<0.0001		 R2=0.379
p=0.044		 R2=0.795
p=0.0002		 -
Open in a new tab

3.2.3 MtDNA/nDNA measurement variability

The mean of the coefficient of variability between sites for all 20 samples (mean CV ± SD [range]) was (raw data 79 ± 48 [37–212]%; log10 24 ± 10 [9–35]%) (Table 1). Concordance improved if data from site #2 (relative scale) were omitted (raw data 71 ± 50 [37–212]%; log10 11 ± 5 [8–28]%), and further improved if site #4 was also omitted (raw data 56 ± 20 [37–125]%; log10 10 ± 2 [8–13]%). Interestingly, four of the five samples with the highest overall log10 data variability were extracted from the fibroblast rhabdomyosarcoma cell line CRL 2061. If only samples from a single cell line (K562 (N=11)) were considered for all sites, the mean CV was (raw data 71 ± 20 [55–126]%; log10 32 ± 2 [29–35]%), and this decreased to (raw data 56 ± 22 [39–115]%; log10 9 ± 1 [8–11]%) if the relative values from site #2 were omitted.

4. Discussion

The various assays used in this study were internally reliable. However, more work is needed before absolute quantification of mtDNA is sufficiently reproducible across laboratories to allow direct comparison between them, or development of clinically meaningful normal range values for use in clinical diagnosis and monitoring. Although mtDNA is a material of choice for forensic nucleic acid analyses and is known for its relative stability, travel at room temperature, though very affordable, did not favor mtDNA/nDNA measurement reproducibility. The apparent increase in mtDNA/nDNA content observed was likely caused by partial degradation of the nDNA during transport. This also implies that the standard sample that was distributed for future standardization between laboratories cannot be used for this purpose as it was part of the room temperature shipment. Alternatively, it is possible that partial degradation of the DNA linearized the mtDNA, rendering it more accessible to polymerases. From this exercise, it would clearly be recommended that DNA samples be kept frozen until analyzed. Because of this, the correlations between sites presented in Tables 4A and 4B should be interpreted with caution, as DNA degradation was likely a factor. Nevertheless, 6 of the 9 sites demonstrated good concordance between them.

For the second shipment on dry ice, in agreement with observations from a previous smaller study (Hammond, Sayer et al. 2003), good correlation was observed between 8 of the 9 sites. However, significant variability between sites remained with respect to the absolute mtDNA/nDNA values. This was illustrated by the inter-site CV which was above 200% for some samples derived from the CRL 2061 cell line, a high figure considering that intra-site variability (CV) for mtDNA/nDNA assay is typically ≤15%. Log10-transforming the data reduced the inter-site variability, as could be expected. A number of factors could influence the variability in mtDNA/nDNA values measured between sites. These include but are not restricted to the specificity of the assay primers and the specificity of the detection method used (SYBR green vs. fluorescent probes), the copy number of the nuclear gene amplified, potential polymorphisms and DNA rearrangements, the target gene's PCR efficiencies, and the methodology itself. A total of nine different assays were used among the participating sites, and the two sites using the same primer sets did not show higher than average correlation between them. This may be due to the fact that different detection systems were used.

Each assay uses unique sets of primers targeting a mitochondrial gene and a nuclear DNA gene. Some of the variability observed between the sites is intrinsic to the genes and primers they use to amplify the DNA as even within laboratories, some genes can yield more variable results than others. Insufficient specificity on the part of the nDNA primers would evidently impact this assay. Should any of the primers amplify unintentional targets such as pseudogenes or nuclear genes that are subject to chromosomal rearrangements, the value of the mtDNA/nDNA ratio would be affected. Nuclear DNA primers should ideally be targeted toward single-copy nuclear genes having low incidence of inter-individual polymorphisms and mutations. If a high copy number gene is chosen, the exact number of copies should be considered if the mtDNA/nDNA ratio is to be compared to that generated using single copy nuclear genes. Of course, the PCR efficiency of both the mtDNA and nDNA amplicons should be highly similar and the DNA concentration range yielding stable mtDNA/nDNA should be determined. Nuclear DNA non-coding pseudogenes, although less common than mitochondrial pseudogenes (Zhang and Gerstein 2004), are especially prevalent for ribosomal RNA genes (Griffiths-Jones 2007). Indeed, based on Primer-BLASTing, some assays used in this study may have unintentionally amplified other products including nuclear target pseudogenes with high or even complete homology. These homologous DNA amplicons may be undetectable by the Tm curve often used to evaluate PCR primer's specificity, yet they would significantly decrease mtDNA/nDNA ratio.

In addition, it is well recognized that chromosomal rearrangements, resulting in copy number variants, occur within the human genome. Although copy number variants have been associated with disease and malignancies (Conrad and Antonarakis 2007; Cooper, Nickerson et al. 2007), and are known to exist in several of the cell lines used in this study (Cottier, Tchirkov et al. 2004), they are also found in healthy individuals and are more common than initially expected (Scherer, Lee et al. 2007; Perry, Yang et al. 2008). As the cell lines used for this study are transformed and mostly derived from cancer patients, their DNA could bear important chromosomal rearrangements that may or may not affect amplification by primers targeting genes that vary from one assay to another. This could be at least partially responsible for the higher inter-site variability observed with some samples. Given that quantification can be nDNA gene-specific, consistent results with two independent nDNA genes can help rule out the possibility of unintended nuclear amplification.

Mitochondrial DNA pseudogenes are very common throughout the nuclear genome (Bensasson, Zhang et al. 2001; Yao, Kong et al. 2008) and pose many challenges to mtDNA research. As they can vary from one individual to another, from one cell line to another, some of the variability observed between the sites could be explained by mitochondrial pseudogenes. Testing two mtDNA genes rather than one or using Rho(o) cell (Hashiguchi and Zhang-Akiyama 2009) DNA as template would confirm the absence of mitochondrial pseudogene amplification. Using a single type of cultured cells when studying mtDNA quantification in would avoid many of the issues raised above.

Sequencing of the human genome and tools such as BLAST and Primer-BLAST can greatly assist in the design of assay primers. Of note, many primers used in this study were designed before the availability of these tools. Future studies such as this one should consider reporting not only the mtDNA/nDNA ratio but rather each gene copy number separately. This would allow the assessment of accuracy and concordance across sites and would give information on whether the source of discordance lies with the mitochondrial or the nuclear gene quantification.

The fact that all sites showed high concordance for the K562-only derived samples reinforces the likelihood that single vs. multicopy genes, cell line-specific DNA alterations and/or polymorphisms may have affected the performance or applicability of some assays. As several of the samples were extracted from cells exposed to drugs such as zidovudine, stavudine or simvastatin, there is a remote possibility that the drugs may affect the primer binding sites, hence the assay. From this data, it is difficult to ascertain how these factors may influence mtDNA measurements in human clinical samples from various genetic make-up and for the study of various diseases, however one can assume that clinical samples may harbor fewer chromosomal rearrangements than transformed cell lines. These results suggest that mtDNA quantification assays need to be designed carefully and several specific recommendations can be made based on this study to increase reproducibility and accuracy of mtDNA/nDNA determinations, in addition to the usual qPCR assay design steps.

5. Conclusions

Our results showed good correlation between laboratories, indicating that within lab comparisons or comparison of relative mtDNA/nDNA between labs should be reliable. However, absolute mtDNA/nDNA ratio values were highly variable across sites, something that is probably partially due to the fact that samples were derived mostly from transformed cultured cells. Furthermore, our results indicated that for such measurement as mtDNA/nDNA ratio, transportation of samples must take place under frozen conditions. Although human clinical samples may have yielded less variable results, further efforts in standardization and evaluation of proficiency in reporting mtDNA content are clearly needed if the goal is to standardize mtDNA content reporting and establish clinically relevant reference ranges for disease states, in order to assist clinical care and research.

Acknowledgements

The authors wish to thank Izabella Gadawski, Daniel E. Libutti, Dirk Lebrecht, Bernhard Setzer, and Eugenia Negredo for their contribution.

HCFC was supported by a MSFHR Scholar award (CI-SCH-50(02-1)) a CIHR New Investigator award (YSH-80511) and a CFI-New Opportunity (#10427). MG was supported by United States of America Department of Health and Human Services, National Institutes of Health grants # RO1 AI074554 and P20 MD000173. AC was partially supported by Istituto Superiore di Sanità., Progetto Nazionale AIDS 2007 (grant # 30G.62).EGA and ALAP work was supported by Grants from “Fondo de Investigación Sanitaria” (PI07/0347, PI08/90355, PS09/01602). OM and GG were supported by CIBERER (ISCIII), Grants from 'Fondo de Investigación Sanitaria' (PI08/0229) and 'Fundación para la Investigación y la Prevención del SIDA en España' (FIPSE 09/360745) and OM has been depositary of a Research Intensification grant from ISCIII (Spain) during 2009.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The University of British Columbia has a patent on one of the mtDNA/nDNA real-time PCR assays used in this study, on which HCFC is one of the inventors. AC has founded and has shares of GeneMore Italy srl, a company that has developed an assay for mtDNA quantification. Other authors have no other potential conflicts.

Preliminary results of this study have been reported at the meeting: “The Dark Side of the HAART – From Basic Science to Clinical Aspects”, 29–31 May 2008, Modena, Italy.

References

  1. Andreu AL, Martinez R, et al. Quantification of mitochondrial DNA copy number: pre-analytical factors. Mitochondrion. 2009;9(4):242–246. doi: 10.1016/j.mito.2009.02.006. [DOI] [PubMed] [Google Scholar]
  2. Bensasson D, Zhang D, et al. Mitochondrial pseudogenes: evolution's misplaced witnesses. Trends Ecol Evol. 2001;16(6):314–321. doi: 10.1016/s0169-5347(01)02151-6. [DOI] [PubMed] [Google Scholar]
  3. Conrad B, Antonarakis SE. Gene duplication: a drive for phenotypic diversity and cause of human disease. Annu Rev Genomics Hum Genet. 2007;8:17–35. doi: 10.1146/annurev.genom.8.021307.110233. [DOI] [PubMed] [Google Scholar]
  4. Cooper GM, Nickerson DA, et al. Mutational and selective effects on copy-number variants in the human genome. Nat Genet. 2007;39(7 Suppl):S22–29. doi: 10.1038/ng2054. [DOI] [PubMed] [Google Scholar]
  5. Copeland WC. Inherited mitochondrial diseases of DNA replication. Annu Rev Med. 2008;59:131–146. doi: 10.1146/annurev.med.59.053006.104646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cossarizza A, Riva A, et al. Increased mitochondrial DNA content in peripheral blood lymphocytes from HIV-infected patients with lipodystrophy. Antivir Ther. 2003;8(4):315–321. [PubMed] [Google Scholar]
  7. Cote HC. Possible ways nucleoside analogues can affect mitochondrial DNA content and gene expression during HIV therapy. Antivir Ther. 2005;10(Suppl 2):M3–11. [PubMed] [Google Scholar]
  8. Cote HC, Raboud J, et al. Perinatal exposure to antiretroviral therapy is associated with increased blood mitochondrial DNA levels and decreased mitochondrial gene expression in infants. J Infect Dis. 2008;198(6):851–859. doi: 10.1086/591253. [DOI] [PubMed] [Google Scholar]
  9. Cottier M, Tchirkov A, et al. Cytogenetic characterization of seven human cancer cell lines by combining G- and R-banding, M-FISH, CGH and chromosome- and locus-specific FISH. Int J Mol Med. 2004;14(4):483–495. [PubMed] [Google Scholar]
  10. Galluzzi L, Pinti M, et al. Changes in mitochondrial RNA production in cells treated with nucleoside analogues. Antivir Ther. 2005;10(1):191–195. [PubMed] [Google Scholar]
  11. Gerschenson M, Brinkman K. Mitochondrial dysfunction in AIDS and its treatment. Mitochondrion. 2004;4(5–6):763–777. doi: 10.1016/j.mito.2004.07.025. [DOI] [PubMed] [Google Scholar]
  12. Gerschenson M, Shiramizu B, et al. Mitochondrial DNA levels of peripheral blood mononuclear cells and subcutaneous adipose tissue from thigh, fat and abdomen of HIV-1 seropositive and negative individuals. Antivir Ther. 2005;10(Suppl 2):M83–89. [PubMed] [Google Scholar]
  13. Griffiths-Jones S. Annotating noncoding RNA genes. Annu Rev Genomics Hum Genet. 2007;8:279–298. doi: 10.1146/annurev.genom.8.080706.092419. [DOI] [PubMed] [Google Scholar]
  14. Hammond EL, Sayer D, et al. Assessment of precision and concordance of quantitative mitochondrial DNA assays: a collaborative international quality assurance study. J Clin Virol. 2003;27(1):97–110. doi: 10.1016/s1386-6532(02)00134-8. [DOI] [PubMed] [Google Scholar]
  15. Hashiguchi K, Zhang-Akiyama QM. Establishment of human cell lines lacking mitochondrial DNA. Methods Mol Biol. 2009;554:383–391. doi: 10.1007/978-1-59745-521-3_23. [DOI] [PubMed] [Google Scholar]
  16. Higuchi M. Regulation of mitochondrial DNA content and cancer. Mitochondrion. 2007;7(1–2):53–57. doi: 10.1016/j.mito.2006.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lopez S, Miro O, et al. Mitochondrial effects of antiretroviral therapies in asymptomatic patients. Antivir Ther. 2004;9(1):47–55. [PubMed] [Google Scholar]
  18. Masayesva BG, Mambo E, et al. Mitochondrial DNA content increase in response to cigarette smoking. Cancer Epidemiol Biomarkers Prev. 2006;15(1):19–24. doi: 10.1158/1055-9965.EPI-05-0210. [DOI] [PubMed] [Google Scholar]
  19. Nolan D, Hammond E, et al. Mitochondrial DNA depletion and morphologic changes in adipocytes associated with nucleoside reverse transcriptase inhibitor therapy. Aids. 2003;17(9):1329–1338. doi: 10.1097/00002030-200306130-00007. [DOI] [PubMed] [Google Scholar]
  20. Papp E, Gadawski I, et al. Longitudinal effects of thymidine analogues on mtDNA, mtRNA and multidrug resistance (MDR-1) induction in cultured cells. J Antimicrob Chemother. 2008;61(5):1048–1052. doi: 10.1093/jac/dkn067. [DOI] [PubMed] [Google Scholar]
  21. Perry GH, Yang F, et al. Copy number variation and evolution in humans and chimpanzees. Genome Res. 2008;18(11):1698–1710. doi: 10.1101/gr.082016.108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Radonic A, Thulke S, et al. Guideline to reference gene selection for quantitative real-time PCR. Biochem Biophys Res Commun. 2004;313(4):856–862. doi: 10.1016/j.bbrc.2003.11.177. [DOI] [PubMed] [Google Scholar]
  23. Scherer SW, Lee C, et al. Challenges and standards in integrating surveys of structural variation. Nat Genet. 2007;39(7 Suppl):S7–15. doi: 10.1038/ng2093. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Setzer B, Schlesier M, et al. Effects of of didanosine-related depletion of mtDNA in human T lymphocytes. J Infect Dis. 2005;191(6):848–855. doi: 10.1086/427655. [DOI] [PubMed] [Google Scholar]
  25. Vidal F, Domingo JC, et al. In vitro cytotoxicity and mitochondrial toxicity of tenofovir alone and in combination with other antiretrovirals in human renal proximal tubule cells. Antimicrob Agents Chemother. 2006;50(11):3824–3832. doi: 10.1128/AAC.00437-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Yao YG, Kong QP, et al. Pseudomitochondrial genome haunts disease studies. J Med Genet. 2008;45(12):769–772. doi: 10.1136/jmg.2008.059782. [DOI] [PubMed] [Google Scholar]
  27. Zhang Z, Gerstein M. Large-scale analysis of pseudogenes in the human genome. Curr Opin Genet Dev. 2004;14(4):328–335. doi: 10.1016/j.gde.2004.06.003. [DOI] [PubMed] [Google Scholar]

RESOURCES