On the Propagation of Mitochondrial Mutations
a new mutation get to appear in a mitochondrial
Address for correspondence:
How does a new mutation get to appear in a
Although the mechanisms of the production and propagation of chromosomal mutations can be explained, the behaviour of mitochondria and the propagation of mutations in the mtDNA are still very poorly understood. This paper, in no way, manages to resolve all the issues, but the steps involved in going from an initial mutational event to the final state of homoplasmy are discussed.
The study of mitochondrial
The Phylogenetic Tree
The phylogenetic tree for human mtDNA is now very extensive as it is found that only persons who have a maternal relationship within a genealogical time frame have identical mitochondrial sequences. And, as one compares the mtDNA sequences from persons who are more and more distantly related there is an increasing number of mutational differences between the sequences.
There are several papers that describe the phylogenetic tree; in particular, the paper from Kivisild, et al. (2006) has some very useful diagrams. This paper also gives estimates as to when the different branches of the phylogenetic tree were formed. As an example, the diagrams show that Haplogroup H and V have a common source from about 16,000 +/- 6,000 years ago.
The author’s own mtDNA sequence shows his maternal
origin is in Haplogroup V and has this set of mutational differences from the
T72C A263G 309.1C 315.1C A750G
A1438G A2706G A2880G G4580A A4769G
C7028T A8860G A15326G C15904T A16162G
The list contains 16 mutations of which 14 are base substitutions and 2 are insertions.
As Haplogroup V and Haplogroup H arose from a common source it is possible to split this list of mutations into two parts.
The first part contains the 6 mutations that have occurred after the split from the common source on the line leading to Haplogroup V:
T72C A2880G G4580A C15904T A16162G T16298C
the second part contains the other 10 mutations that occur on the line to the
A263G 309.1C 315.1C A750G A1438G
A2706G A4769G C7028T A8860G A15326G
However, if the list of mutations as given above is
viewed from the
G263A C309d C315d G750A G1438A
G2706A G4769A T7028C G8860A G15326A
Figure 1 illustrates the list of mutations divided into these two parts.
Figure 1. The mutations that have occurred along the lines leading to Haplogroups H and V from HV
So, over the last 16,000 years these 16 mutations have appeared in the two lines to Haplogroups H and V; but how might this have happened?
A point to be made at this stage is that in many places in this paper, and in the accompanying computer model, it has been found useful to personify a mutation as it passes through the different stages of its propagation; but it is important to appreciate that all the steps occur at random and a mutation does not have purpose, ambition, nor any form of intelligence.
Modelling the Propagation of Mitochondrial Mutations
A computer model that simulates the propagation of mitochondrial mutations has been written to accompany this paper. This model shows how a new mutation might go through the phases of propagation of a mutation from an initial mutational event to the states of heteroplasmy and homoplasmy.
The computer model is available in text form as supplementary file at:
And as a web page, accompanied by instructions for use and a discussion of the modeling process at:
Mitochondria are small organelles found in all the cells of the body. They are mainly involved in the production of “energy” within the cells and have been considered to act as small “power stations.” But, whereas most processes within a cell are controlled by the nucleus, mitochondria act independently and are thought to have arisen from bacteria that have formed a symbiotic relationship with the cell. For this reason mitochondria appear to behave more like bacteria than nucleated cells.
Most cells contain relatively few mitochondria, perhaps up to several hundred. But in a fully developed egg cell, an oocyte, there may as many as 200,000. Mitochondria appear to replicate by division and this process is independent of cell division. So, in a cell that does not divide as it matures, such as an oocyte, the number of mitochondria may become very high. However, in the cells of a tissue where there is a high rate of cell division the number of mitochondria per cell falls to a very low number.
The recent papers by Shoubridge and Wai (Shoubridge, 2007, Wai, 2008) deal particularly well with the subject of the behaviour of mitochondria in mammalian cells and discuss some of the problems concerning the emergence of mitochondrial mutations.
The Propagation of a Mitochondrial Mutation
mitochondrial mutation cannot just appear suddenly in every one of an individual’s
billions of mitochondrial
But because mitochondria act like bacteria it is possible to infer that a mitochondrial mutation propagates in the same manner as a mutated bacterial strain is known to arise. In bacteria for the emergence of a new bacterial strain there needs in the first instance to be a mutation occurring in the genetic code of just one organism, this mutated bacterium then goes through many generations until bacteria with this mutation form a new strain. In bacteria a mutation can develop because it offers an advantage to the bacteria, or at random with no advantage being present.
As described with bacteria, the spread of a mitochondrial mutation has to begin with an initial mutational event occurring in just one mtDNA molecule and this is followed by the mutation gradually coming to be present in more and more mtDNA molecules, until finally the state of homoplasmy is reached when it is considered that every mtDNA molecule in the body has the mutation.
In this paper the gradual increase in the prevalence of a new mutation is described as the propagation of a mutation; and this process which can be split into two phases.
The first phase includes the steps that may be involved in going from an initial mutational event through to when many, or all, of the mtDNA molecules, firstly in a single mitochondrial nucleoid, and later a single mitochondrion, have the new mutation. This first phase can occur in any cell in the body, and not necessarily in a cell line that is going to be passed to a descendant.
However, the second phase deals with the special case where the new mutation has present in a germ cell line that goes on to produce the egg cells, the oocytes. In this phase a new mutation propagates and leads to heteroplasmy and homoplasmy in descendants.
Figure 2 illustrates the physical relationship between mtDNA molecules, a mitochondrial nucleoid, a mitochondrion and a cell.
Figure 2. (a) mtDNA molecules in a nucleoid, (b) Nucleoids in a mitochondrion, (c) Mitochondria in a cell.
Initial Mutational Events
Any mtDNA mutation must in the first instance come from an initial mutational event affecting a strand of mtDNA; and will usually involve either:
a substitution of one base for another,
a deletion of one or more neighbouring bases, or
the insertion of one of more extra bases,
It is also possible that an initial mutational event may, on occasions, be more complex, and lead to several different changes in a mtDNA molecule occurring at the same time. For example, a deletion of bases at one point and their insertion at another may occur.
Presumably, these initial mutational events occur as errors when a mtDNA strand is being replicated and are very common. But what is unusual is for an initial event to lead to the development of a new homoplasmic mutation.
In this paper it is assumed that initial events occur at random and do not exert any advantage, or disadvantage, to the mitochondrion, cell, or person, in which they are found. In this respect, the discussion of any possible positive or negative selection of individuals with certain mtDNA mutations is beyond the scope of this paper.
However, a very interesting paper on selection has recently been published by Doublet, et al.,(Doublet, 2008) and whilst the authors are discussing the mitochondria in a crustacean, some of their conclusions on the general behaviour of mitochondria are similar to those expressed in this paper.
Modelling initial mutational events
In the computer model the user is asked to select the number of initial events from the range 1 - 10,000,000; and this number then becomes the number of ‘runs’ that the model will use. In most instances an initial mutational event will not propagate, but each mutational event is considered to have the same chance of successful propagation as any other.
Propagation to a Mitochondrial Nucleoid
After its creation as an initial event, the next stage in the successful propagation of a mutation is when, by random, many, or all, of the mtDNA molecules in a particular mitochondrial nucleoid have the mutation. In this paper this successful propagation is described as the capture of a nucleoid by a mutation; and whereas initial events are likely to be very common, a capture of a nucleoid is presumably very uncommon.
Unfortunately, the biology of mitochondrial nucleoids is poorly understood - but a nucleoid appears to be a functionally separate area containing just a fraction of the mtDNA molecules found within a single mitochondrion. The recent paper by Gilkerson, et al. (2008) contains a detailed account of what is known about mitochondrial nucleoids, together with a very useful set of references.
An inference that can be drawn from the concept of the capture of a mitochondrial nucleoid is that different nucleoids in the same mitochondrion may become captured by different mutations. That is, the mtDNA molecules in one nucleoid may contain one particular mutation whilst other nucleoids may have other mutations
Phase A: Modelling the capture of a mitochondrial nucleoid
In the computer model the user is asked to determine the Nucleoid Size from the range 5 - 30, and this value represents the average number of mtDNA molecules in each nucleoid within a mitochondrion.
With the selections of initial events and Nucleoid Size made, the user can now click A:START to run the model.
The model will then for each initial event in turn show if a new mutation can capture a nucleoid.
In simple terms the model replicates a mtDNA molecule chosen at random and then deletes a mtDNA molecule, again chosen at random, over and over, until either the new mutation is lost or it captures the nucleoid.
Propagation to a Mitochondrion
Once a new mutation has become found in most, or all, of the mtDNA molecules in a nucleoid, the mutation may then become dominant within a mitochondrion by replication occurring at random, and without suggesting that there might be any selective advantage to the ‘mutated’ nucleoid.
It is not known how nucleoids are replicated - but they do act as functionally separate organelles. As such it is likely that a nucleoid increases in size as its mtDNA molecules are duplicated; and then the nucleoid divides into two forming daughter nucleoids which each contain roughly half the original number of mtDNA molecules.
This replication of nucleoids
appears to happen at random, and this may lead over a number of replication
cycles to the situation where all the nucleoids in a
mitochondrion are the descendants of just one nucleoid. So, if this nucleoid
does contain a new mutation in its
Phase A: Modelling the capture of a mitochondrion
In the computer model the user is asked to determine the Nucleoid Number from the range 5 - 20, and this value represents the average number of mtDNA molecules in a mitochondrion. The user also needs to specify ‘mitochondrion’ as opposed to ‘nucleoid’.
With the selections of initial events, Nucleoid Size, Nucleoid Number and ‘mitochondrion’ made, the user can now click A:START to run the model.
The model will then for each initial event in turn show if a new mutation can capture a mitochondrion.
In simple terms the model will as before first try to capture a nucleoid, and when successful, proceeds with this nucleoid to see if, at random, it captures the mitochondrion.
These results suggest that with a Nucleoid Number of 10 mtDNA molecules per mitochondrion capture of a mitochondrion may occur, but as the Nucleoid Number is increased then mitochondrial capture becomes very much less likely.
The Persistence of a Mutated Mitochondrion
Once a mitochondrion has most, or all, of its mtDNA molecules containing a new mutation, it is possible that a fairly stable situation has been reached. Mitochondria are organelles that appear to have a long life within cells so it is likely that a mutated mitochondrion once it has arisen may persist in tissues.
It is interesting to speculate that the phenomenon whereby many new mutations appear in mtDNA taken from cancerous tissue that the mutations have arisen because of the proliferation of persistent mutated mitochondria in the tissue; and not as a result of the actual cancer causing mutations in the mtDNA of its cells.
The Importance of the Germ Cell Line
The previous discussion has concerned itself with how a new mutation might appear in any line of cells in an organism, but in genealogy we are concerned with how mutations are transferred from a person in one generation to their offspring in the next generation. And, for this to happen it is necessary to consider that mutated mitochondria are present in the germ cell line—meaning that for a new mutation to appear, mutated mitochondria need to be present in the egg cells, the oocytes.
After fertilisation, a human oocyte enlarges and divides repeatedly to form the first stage of embryonic growth, the blastula. This has the form of a hollow ball of cells; and each of the different cells of the blastula goes on to divide and make a different cell line; and these cell lines in turn form the different tissues of the body.
One of these lines in a female embryo is the cell line that form the primordial germ cells which over time form the next set of oocytes; and mitochondria containing mtDNA molecules with a new mutation must be present in this cell line if the new mutation is to be taken forward to the next generation.
Another important cell line to be considered is the
one which goes on to produce buccal cells (cells forming the inner
surface of the cheeks); as sequencing of the mitochondrial
But, as before whenever there is replication of cells and their organelles, there is a random allocation of the organelles into the cells. So when different cell lines are produced from the blastula it is possible that some cell lines may have a below average number of mutated mitochondria in their cells and other cell lines an above average number. If the new primordial germ cell line has a higher proportion of mutated mitochondria in its cells than were present in the germ cell line of the parent then the new mutation has successfully achieved a higher level of propagation.
Figure 3 illustrates these stages in embryonic growth and shows how two cell lines may vary in the number of mutated mitochondria that they contain in their cells.
Figure 3. New cell lines containing different proportions of mutated mitochondria
Heteroplasmy and Homoplasmy
The above discussion has described how it is possible for the cells in a particular cell line of the body to acquire a new mutation; and how if the cell line is the Germ Cell Line that produces the egg cells, the oocytes, for the next generation it is possible for the proportion of mtDNA molecules with the new mutation to be higher in the next, and subsequent generations.
At present it is not feasible to test cells from the Germ Cell Line for Genealogical purposes, so it is usual to test buccal cells which are very easy to collect using a mouth swab. A buccal cell sample is generally assumed to represent the cells of the Germ Cell Line - but as the above discussion has pointed out, there may be differences in the proportion of mutated mtDNA molecules in different cell lines.
The mtDNA molecules obtained in a cell sample can theoretically be considered as having heteroplasmy at very low levels, but it is really only possible when sequencing for Genealogical purposes for heteroplasmy to be detected when a new mutation is present in maybe a tenth of the mtDNA molecules, i.e. at a level of, maybe, 10%. This also means that homoplasmy is assumed on testing when the new mutation might actually only be present in about 90% of mtDNA molecules.
Just how a new mutation present at a low level of heteroplasmy propagates to become a homoplasmic mutation is not properly understood. But it is likely that in many cases where a new mutation is being propagated successfully that heteroplasmy levels rise slowly over a number of generations. This, however, does not mean that the level of heteroplasmy rises with every generation, and indeed the level can be expected to fall in some generations.
But whilst the gradual rise in levels of heteroplasmy over a number of generations is possibly the usual way in which a new mutation is propagated, this does not explain the observation that on occasions new mutations can be seen to appear with a high level of heteroplasmy, or even homoplasmy, in an individual, but the mutation is not found in the individual’s mother.
However, one explanation of this phenomenon may be to consider that at a particular stage in the development of Germ Cells in an embryo there is such rapid growth that each cell contains very few mitochondria. And, if one of these mitochondria should contain a new mutation it is possible that the mutation may appear to propagate suddenly over a single generation. But, as discussed above, this does not mean the initial mutational event is recent, as the first phase of propagation leading to the presence of at least a single mutated mitochondria will have needed to have occurred beforehand.
The recent papers by Shoubridge and Wai (Shoubridge, 2007, Wai, 2008) discuss this phenomenon in some detail and describe it as being a genetic bottleneck.
The Computer Model accompanying this paper attempts to illustrate how a new mutation might propagate successfully from a very low level of heteroplasmy to homoplasmy - or, as happens in most cases, fails in its attempt to propagate. The model allows for the number of mitochondria per cell to be reduced to very low levels, thereby simulating the effect of a genetic bottleneck.
Phase B: Modelling the steps from Mitochondrial Capture to the appearance of Heteroplasmy and Homoplasmy
This part of the model shows how the level of heteroplasmy might vary from one generation to the next; and whether homoplasmy is achieved.
The starting condition can be considered as representing the presence of a single mitochondrion with a new mutation in all of its mtDNA molecules; and running the model shows over a series of Runs whether the states of heteroplasmy and homoplasmy are achieved.
In the computer model the user is asked to determine:
- the number of Runs from the range 10 - 1000,
- the Mitochondrial Number from the range 10 - 800.
In effect this number fixes the initial level of heteroplasmy. Choosing ‘200’ mitochondria will give an initial level of 1/200, or 0.5%.
- the number of Replications from the range 5 - 20.
This number can be considered to represent the number of cell divisions at the ‘blastula’ stage. The higher the number selected the more randomisation will occur between generations.
- the number of Generations form the range 10 - 300.
This number alters the number of generations that are to be followed.
- there is also the option of suppressing the display of the Full Results, which can be useful when the output from the model is large.
The model is run by choosing B:START. Results for three sets of example inputs are shown below:
The above discussion and accompanying Computer Model offer some ideas as to how mitochondrial mutations might arise and be propagated.
First, a mutation appears as an initial mutational
event, before becoming the dominant form in the mitochondrial
But if a particular mutation cannot be detected on sequencing as being heteroplasmic, does this mean the mutation is totally absent, or is it present in some of the mtDNA molecules of an individual? The ideas discussed above do suggest that the mitochondria of an individual may contain many new mutations, each of which has a very small chance of propagating over future generations to be a heteroplasmic or homoplasmic mutation.
However, other problems still remain to be solved:
For example, there are many examples in the phylogenetic tree where a new mutation appears in branches near to each other, but is absent in many other branches. The usual explanation of this phenomenon has in the past involved the concepts of parallel mutations and back mutations.
A parallel mutation being said to occur when the same mutation has appeared in different places in the tree, i.e. the mutation is said to have emerged by coincidence in different places. This phenomenon is generally considered simply to be the result of the randomness of nature as there is only a limited number of possible mutations that can occur and so it is likely the same mutation will occur by chance in different places in the phylogenetic tree.
And, a back mutation is said to occur when a new mutation is found higher up the tree, but is absent in lower branches where it would be expected. This phenomenon is difficult to explain as it would appear to depend on a mutation having a memory of its previous state--something that is unlikely to occur.
However, this paper describes how a new mutation must in the first instance appear as an initial mutational event and then may propagate to be a heteroplasmic or homoplasmic mutation. And, whereas the ideas described here do not explain fully the finding of back mutations in the phylogenetic tree, they do suggest a mechanism by which a new mutation may not appear in all the descendants of an individual.
There is clearly still a great deal to be learnt about the behaviour of mitochondrial mutations and better models will be produced, but the present paper together with its Computer Model appears able to explain many of the observed phenomena.
A listing of the code for the computer model is available both in text form and as a web page at:
Anderson S, Bankier AT, Barrell BG, de Bruijn MHL, Coulson AR, Drouin J, Eperon IC, Nierlich DP, Roe BA, Sanger F, Schreier PH, Smith AJH, Staden R, Young IG (1981) Sequence and organization of the human mitochondrial genome. Nature, 290:457-465.
RM, Hubacka I, Chinnery PF,
Lightowlers RN, Turnbull DM, Howell N (1999) Reanalysis and
revision of the
Cann RL, Stoneking M,
Doublet V, Souty-Grosset
C, Bouchon D, Cordaux R, Marcadé I. A thirty million year-old inherited
heteroplasmy (2008) PLoS
RW, Schon EA, Hernandez E,
Kivisild T, Shen P, Wall DP, Do B, Sung R, Davis K, Passarino G, Underhill PA, Scharfe C, Torroni A, Scozzari R, Modiano D, Coppa A, de Knijff P, Feldman M, Cavalli-Sforza LL, Oefner PJ (2006) The role of selection in the evolution of human mitochondrial genomes. Genetics, 172):373-87.
EA, Wai T (2007) Mitochondrial
Wai T, Teoli D, Shoubridge EA
(2008) The mitochondrial