Reader Comments

Post a new comment on this article

Does a pseudo-mitochondrial genome haunt this disease study ?

Posted by IanLogan on 12 Jul 2010 at 10:48 GMT

In 2008 Yong-Gang Yao, et al. [1] published a paper that discussed the manner in which nuclear mitochondrial pseudogenes (NUMTs) can pose a serious problem for mitochondrial disease studies. Their paper dealt with errors that occur when the results of mitochondrial DNA sequencing are confounded by the presence of mutations coming from a NUMT.

The present study by Rollins, et. al. did not use formal DNA sequencing, but instead used an Affymetrix mitochondrial array which is said to have 'assigned high quality base pair sequencing calls to 98.2% of the mtDNA genome'.

However, in the paper there are over 17 instances of a very uncommon mutation, T10652C out of 41 subjects and 36 controls; and the authors do not appear to have seriously considered the possibility of NUMT contamination as the underlying reason for their unusual findings. But to us it would seem that the involvement of a NUMT is the cause of the unexplained results.

It is all well and good to suspect the involvement of a NUMT, but it is another matter to try to prove it. However, here, we suggest a mechanism by which we can explain the results and furthermore show which particular NUMT is likely to be responsible.

Although we do not know the details of the primer used by Affymetrix to determine the nature of the base at 10652; it must by necessity match the Cambridge Reference Sequence (CRS) for a number of bases either before or after the base at 10652. It is probable that the primer used has a length of between 10 bases and 20 bases.

We explain our mechanism by starting with the CRS where the sequence of bases from 10652 is:
So a suitable primer to detect a base change at 10652 could match against this sequence if it was, for example:
the reciprocal of 'GCCATACTAGTCTTTGC'
where the next base will correspond to 10652, and can be determined.

Unfortunately for the present researchers there is a NUMT which has a 'C' at the position corresponding to 10652, and hence if a primer of this structure were to have been used it could quite possibly lead to the conclusion that the mutation T10652C was present.

The NUMT is question was detailed in the paper by Tourmen, et al. [2] as being on Chromosome 5 at locus 5q31.1. This NUMT is said to be 5,219 bases in length and corresponds to bases 10,269 - 15,487 of the CRS.

The more recent Build 37.1 gives the chromosomal coordinates of this NUMT as chr5:134258999..134264217

The way in which this NUMT matches the CRS can be shown by following these steps.
Open the Mapviewer from NCBI using http://www.ncbi.nlm.nih.g...
Find the line: Homosapiens, Human, Build 37.1
Go to the tools section and click on the 'B' icon to BLAST Human Sequences.
In the main query box enter the string:
(i.e. the CRS string above, but with T10652C)
In the database box select the option:
genome(reference only)
In the Program box select the option:
BLASTN: Compare nucleotide sequences
In the Expect box select the option:
Next click on BEGIN SEARCH,
and on the next page click on VIEW REPORT.

The report obtained shows that, as expected, the CRS, termed here 'Homo sapiens mitochondrion, complete genome' matches the query string from base 2 to base 42.
(Note: the corresponding part of the reference sequence is said to be 10654-10694, which is 1-base different to the CRS, and this is explained by the reference sequence having the base 3107N.)

The report also shows that the query string matches well against the sequence on contig NT_034772.6 at 42577706-42577665; a segment that does come in the NUMT identified by Tourmen, et al [2].
Indeed there is a perfect match for the 17 bases that correspond to the CRS bases 10653-10669 which would mean that a primer containing the sequence as described here can be expected to 'lock' on to both real mitochondrial DNA and the DNA from this NUMT.

It is therefore our opinion that the mutations observed at T10652C are the result of a NUMT 'haunting' the results.

There is just one further point we would like to make about the paper and that is in relationship to haplogroup N1b1.
This haplogroup is very rare in the general population with an incidence of around 1:500 people. However, the researchers note that 5 persons (out of 77) have mental illness and come from this small haplogroup.
The reason for this over representation of haplogroup N1b1 is not clear; and probably merits further consideration.

Ian Logan, M.B., Ch.B.
Exmouth, England
Ann Turner, M.D.
Menlo Park, CA, USA

1. Yao YG, Kong QP, Salas A, Bandelt HJ. Pseudomitochondrial genome haunts disease studies.J Med Genet. 2008 Dec;45(12):769-72.
2. Tourmen Y, Baris O, Dessen P, Jacques C, Malthièry Y, Reynier P. Structure and chromosomal distribution of human mitochondrial pseudogenes. Genomics. 2002 Jul;80(1):71-7.

No competing interests declared.