Correction to: Nature https://doi.org/10.1038/nature20592, published online 30 November 2016.
Several errors were discovered in this Letter regarding the assignments of mitochondrial (mt)DNA haplotypes for a subset of egg donors from our study. We used whole mtDNA sequencing and then determined the haplotypes by running the sequences through the publicly available MITOMAP database (http://www.mitomap.org/MITOMAP). During this process, the haplotype names were accidentally swapped for egg donors cED4 and hED4, owing to the notational similarity. Specifically, the mtDNA haplotype name H49 was assigned to the carrier egg donor cED4 rather than to the healthy egg donor hED4, which was given the haplotype B2k. Thus, the correct haplotype for cED4 is B2k and for hED4 is H49. In addition, owing to typographic errors in our database, several errors occurred: the haplotype for healthy egg donor hED6 was inaccurately assigned as V3, whereas its correct haplotype is H1g; and the correct donor mtDNA haplotype for NT-ES5 is D4a. The corrected version of Table 1 in the original Letter is shown in Table 1 of this Amendment; the corrected versions of Extended Data Fig. 3a, 5a and 7b in the original Letter are shown in Figs. 1–3 of this Amendment; and the corrected versions of Supplementary Tables 1, 3, 4 and 5 in the original Letter are provided as Supplementary Tables 1–4 with this Amendment; all changes are shown in red. In addition, in Fig. 3 of the original Letter, in all cases ‘H49’ (on the y axes and in the legend of panel f) should be ‘B2k’. The changes are summarized in Table 2 of this Amendment.
We also identified sequence errors as follows. In Supplementary Table 1 in the original Letter, the nucleotide at mt11253 for H6a haplotype of carrier 1 should be C and the nucleotide at mt11812 for T2b haplotype of carrier 3 should be G. Because the mismatched haplotypes T2 and T2b have the same G at mt11812, the correct distance between these samples is 21 single nucleotide polymorphisms (SNPs), not 22 SNPs as reported in the original Letter. The correct sequences are shown in Supplementary Table 1 of this Amendment, with changes highlighted in red. In Supplementary Table 1 of this Amendment, we have also removed the mt3107 SNP position for U5a haplotype because this nucleotide is probably a hotspot. Therefore, the SNP distance between U5a and H1g is reduced from 33 to 32 in Table 1 and Fig. 1 of this Amendment. The correct B2k haplotype at mt11177 is T and is correct in the ‘Summary’ tab of Supplementary Table 4 of this Amendment.
We also found typographic errors in Supplementary Tables 1, 3, 4 and 5 of the original Letter in the assignment of several hED numbers: the correct number for original hED7 is hED10; hED10 is hED7; hED6 is hED9; hED9 is hED6; hED8 is hED11; hED11 is hED5; and hED5 is hED8. The correct hED numbers appear in Supplementary Tables 1–4 of this Amendment, with differences compared to the originals highlighted in red.
We initially assigned egg donor IDs on the basis of mitochondrial phylogenic tree, and included the assignments in Supplementary Tables 1 and 5 in the original Letter. During the formatting of Extended Data Fig. 3a of the original Letter, the order of egg donor IDs was rearranged; however, the corresponding Supplementary Tables were not corrected accordingly.
We found discrepancies due to errors in designating SNPs in the D-loop and other non-coding regions between maternal and donor haplotypes for H49–B2k and D4a–A2g combinations in Supplementary Table 1 of the original Letter. We incorrectly included nine SNPs in the non-coding NC7 locus as D-loop SNPs. Therefore, the number of SNP differences for H49–B2k should be 5 instead of 14 in D-loop, and the number for D4a–A2g should be 19 instead of 18 in coding genes; Supplementary Table 1 of this Amendment shows the correct values, highlighted in red.
We also wish to clarify that the MiSeq-generated whole mtDNA sequencing dataset for the 3243ST-ES1 cell line at passage 3 showed the presence of the mutant mt3243A>G variant at 32% heteroplasmy, which possibly indicates the reversion back to the mutant maternal mtDNA haplotype (SRA project ID SRP091883). We re-analysed the same mtDNA sample that we prepared for MiSeq and a newly prepared mtDNA sample from 3243ST-ES1 cells collected at the same passage 3 by using Sanger and ARMS–qPCR assays; we determined that the presence of the mt3243A>G variant was amplification bias at mt3243.
This cell line was then re-tested 11 times: between passages 4 and 10, after differentiation in teratomas into ectoderm, mesoderm and endoderm representatives, and after in vitro differentiation into neural progenitor cells (NPCs) and cardiomyocytes (SRA project ID SRP091883). Repeated whole mtDNA sequencing did not detect any mt3243A>G variant, verifying the reversion to the non-mutant (mt3243A) maternal haplotype. In addition to whole mtDNA sequencing, we also corroborated the reversal by Sanger sequencing. The PCR primers amplified a fragment that included mt3243, mt4820 and mt4977 SNPs, differentiating maternal and donor mtDNA. Sanger chromatograms showed gradual reversal to the maternal mtDNA at mt4820 and mt4977, but not mt3243 (see Fig. 4).
We also note that our healthy oocyte donor hED5 haplotype was homoplasmic for mt14484T>C. This variant has been reported in some patients with Leber hereditary optic neuropathy (LHON). However, our oocyte donor was asymptomatic, with no evidence of LHON in her family. It is possible that the mt14484T>C variant is implicated in LHON only when presented in certain background haplotypes or in combination with specific nuclear genes. Our IRB inclusion and exclusion criteria clearly state that to be considered a pathogenic mtDNA mutation the oocyte donor family must be diagnosed with the maternally inherited mitochondrial disease. Therefore, this carrier was listed in the healthy oocyte donor cohort.
In conclusion, several errors and discrepancies were discovered in our database in the designation of mtDNA haplotypes, egg donor IDs and counting SNP differences. However, the actual mtDNA sequences for all samples and combinations of mitochondrial replacement therapies or techniques (MRT) were correct. Therefore, these errors do not alter the results or conclusions of the original Letter. The original Letter has not been corrected. We thank C. Bravi for bringing some of the errors to our attention.
Supplementary information is available in the online version of this Amendment.