Two Dimension Agarose Gel Electrophoresis (2D-AGE)

By Rubén Torregrosa-Muñumer

One limitation in science is that we usually study only what we can see. But thanks to the development of new technologies and strategies we can increase our scope. Two dimension agarose gel electrophoresis (2D-AGE) provided new insights to unravel the mechanism by which mtDNA is replicated; however it also gave rise a controversial discussion, since the scientific community do not agree on how mtDNA is replicated in mammals.

Using this method as the driving force of my PhD project, these are some questions I try to address in the lab:

  • How does mtDNA replication cope with a damaged DNA template?
  • How many different replication mechanisms do operate in mammalian mitochondria?
  • Can mammalian mtDNA recombine?

This is how a 2D-AGE (or 2D) looks:

Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Human 293T HEK cells untreated or treated with 2-3´dideoxycytidine (ddC). The treatment with this analogue nucleotide stalled mtDNA replication resulting in the increase of the double-stranded Y arc (indicated as “y”) that contains fully double-stranded DNA replication intermediates, one of the hallmarks of the strand couple replication model.

mtDNA genome

Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Mammalian mitochondrial genome. It encodes for 13 proteins and two rRNAs separated by 22 tRNAs (marked in orange). Modified from Pohjoismäki J and Goffart S, 2011.


Before going further we need to review what we know about the mtDNA replication initiation. Human mtDNA genome is around 16,5Kb double-stranded circular molecule which exists in hundreds of copies per cell. According to their different density in a CsCl gradient, we can distinguish the Heavy strand and the Light strand. Compared with the nuclear DNA, mtDNA is very compacted (it does not have introns) with only two non-codifying regions, the NCR (non-coding region, 1,1 Kb) containing the “OH” or origin of replication of the Heavy strand and other regulatory signals, and a small area (30 bp) containing the “OL” or origin of replication of the Light strand. These two replication origins are used in the asynchronous replication model (I discuss this below). On the other hand, bidirectional replication starts downstream the NCR in a wider area known as Ori-z.

How does the 2D-AGE technique work?

We exploit the fact that replicating DNA molecules progressively doubles in mass, which when are cut with a restriction enzyme adopt different bubbled or branched shapes depending on whether or not they contain a replication origin.

Basically, these are the steps of this technique:

  1. Digestion of mtDNA with a restriction enzyme (molecular scissors).
  2. Separation of the mtDNA fragments using agarose gel electrophoresis: DNA fragments are separated according to their size. This is the first dimension (low voltage, no EtBr).
  3. The gel slab containing the DNA of different sizes is cut, turned 90 degrees and a new agarose gel is cast. Then the second dimension (high voltage, EtBr) is run to separate the DNA fragments by shape.
  4. Transfer of DNA from the gel into a DNA-binding membrane (Southern blot).
  5. Hybridization with a labeled probe and capture of the signal with a photographic film.
  6. Interpretation of the different arcs such as the Y-arc or bubble arc (pss-b in the previows image).

If we are able to see the bubble arc in our 2D-AGE, it indicates that the fragment we are studying contains a replication origin. During DNA replication, replication fork advances resembling a growing “bubble” (Figure 1). Compared with linear DNA molecules, these “bubbles” have a reduced mobility when separated by gel electrophoresis due to their larger and bulky shape; and therefore, they can be easily distinguished. In contrast, if the restriction fragment does not contain a replication origin, we will see a single fork (Y-shaped molecule) advancing from one extreme of the fragment towards the other, generating the Y-arc (picture this: when a bubble is cut, two “Y” fragments are generated (Figure 2)). This Y-shaped molecules have also different mobility in an agarose gel electrophoresis.

Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Figure 1: Bidirectional replication: two replication forks advance in opposite directions until the full-length molecule is replicated. In unidirectional or asynchronous replication there is only one fork advancing.

Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Figure 2: Upper panel: replication bubble. Bubbles are only visible if our restriction fragment contains a replication origin (i.e. OH). Lower panel: Y-molecule. Y molecules resemble a cut bubble or a replication fork advancing from one edge of the fragment towards the other.[/caption]
Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Figure 3: Different restriction enzymes cut at different locations: in human mtDNA, HincII generates a 3.9 kb containing the NCR. In contrast, DraI generates a 3.7 kb fragment spanning the region downstream the NCR. Note that other fragments are generated too when using any of these enzymes, as in human mtDNA there are eleven restriction sites for HincII and four for DraI. However we only visualized the fragment of interested when using a labeled probe (indicated as “*” in the cartoon).[/caption]


We can use different restriction enzymes in order to study different regions of the mitochondrial genome (Figure 3). For example, we can use HincII to study only the control region or NCR (non-coding region) or DraI to study region downstream the NCR in human. Combining different restriction enzymes we can study different parts of the mitochondrial genome and map the exact location of replication initiation.

However, 2D-AGE do not only provides information about replication origins. We can see more than Y-arcs and bubble arcs in these gels. In addition, the shape and length of these arcs also provide more information. Using this method we can determine many other things such as:

  • Unidirectional or bidirectional replication
  • Location of pause sites
  • Replication termination
  • Replication fork direction
  • Nature of replication intermediates (single-stranded or double-stranded DNA)
  • Recombination intermediates
  • (…)

Putting together all these small pieces of data we can finally try to figure out some features of mtDNA replication, making this technique a very robust method.

2D-AGE and mtDNA replication

At the beginning of the eighties Bell and Byers (1983) at Washington University wondered how branched DNA molecules could be separated. Assuming that branched species would migrate more slowly than linear forms of the same mass when separated by electrophoresis, they successfully designed a method to easily separate them. In the next decade this method was optimized by another group from the same University who were interested in mapping replication origins and pause sites in Saccharomyces cerevisiae (Friedman and Fangman, 1987; Friedman and Brewer, 1995). Due to its versatility and robustness, several labs soon started to use this optimized method to analyze another small DNA molecule, the mitochondrial DNA, in simple organisms such as Sea urchin (Mayhook et al, 1992), Schizosaccharomyces pombe (Hand and Stachow, 1994) and the parasite Plasmodium (Preiser et al, 1996).

However, it would not be until the year 2000 when mammalian mtDNA was finally studied using this method by Holt and co-workers (2000) at University of Tampere (Finland). Previously, for almost 20 years it was assumed that mtDNA was exclusively replicated following an asynchronous replication mode described by Clayton (1982), the denominated Strand-Displacement Model, based on evidences obtained using electron microscopy and the mapping of 5´ends. However, two dimension gels revealed two different mechanisms of mtDNA replication co-existing in mammals: the Strand-Displacement Model (Clayton) which proposes a unidirectional replication that starts from discrete origins (OH and OL) and the Strand Coupled Replication (Holt) which proposes a classical bidirectional replication that starts from a wider region downstream the OH (known as Ori-Z). The Strand Coupled replication model is also known as COSCOFA (conventional, strand coupled, Okazaki fragment associating).

Six years later, using this method Yasukawa (2006) proposed that during mtDNA replication of vertebrates (birds and mammals) RNA is incorporated, forming the so-colled RITOLS (RNA introduced trough the lagging strand). Note that RITOLS basically follow the previously proposed strand-displacement model. In both models replication is asynchronous and likely starts at the same places (OH and OL), but the parental heavy strand in RITOLS is covered by RNA instead of proteins (this is a little bit controversial: we do not know for sure whether single-stranded binding proteins (mtSSB) or RNA covers the displaced parental heavy strand).

To summarize, proposed mtDNA replication mechanisms in mammals:

  • Asynchronous replication: Strand-Displacement Model (SDM) or RITOLS.
  • Bidirectional replication: Strand Coupled Replication (COSCOFA).

Two Dimension Agarose Gel Electrophoresis (2D-AGE)

Two Dimension Agarose Gel Electrophoresis (2D-AGE)
Pohjoismäki J and Goffart S, 2011.

Why do mitochondria have two different replication modes? we support the idea that the former (asynchronous replication: SDM or RITOLS) is involved in regular mtDNA maintenance while the latter (bidirectional replication: COSCOFA) is used only under replicative stress, as we demonstrated in Torregrosa-Muñumer (2015) (see below). That is, two co-existing replication mechanism differently utilized depending on the reigning conditions. But, why something too small is too complicated? Considering the importance of mitochondria in energy production -and other cellular functions-, it is plausible that mitochondria have developed such adaptive mechanism.

Human 293T HEK cells treated with UVB light. DNA digested with DraI which cuts downstream the NCR (this mtDNA fragment does not contain the OH; and therefore, we cannot see the bubble arc created in the asynchronous replication. But this fragment contains the Ori-Z where replication starts in COSCOFA). Untreated cells followed the classical asynchronous replication model (most replication intermediates appear in the sllow-moving Y arc or “smy” and the bubble arc is missing). Four hours after the UV exposure, mitochondria switched to COSCOFA (increase of fully double-stranded DNA intermediates (labeled as “y”) and apparition of the bubble arc (“b”) in this fragment which does not contain the OH but includes the Ori-Z, the origin of replication in COSCOFA). Note that the increase of the “Y” arc and bubble arc “b” in the DraI digestion are two hallmarks of COSCOFA in 2D-AGE. Torregrosa-Muñumer, 2015.

However, it is important to highlight that mtDNA replication in mammals is still under a debate. Although there is a number of evidences supporting two co-existing replication modes (asynchronous and bidirectional replication), we do not fully understand neither how they operate nor are regulated. Interestingly, some people consider only the SDM as real, arguing he lack of evidences supporting COSCOFA or claiming that RITOLS are artifacts. And at the same time other people actually think that the SDM is an artifact. On top of this, and to make this topic a little bit more complicated, some evidences support the existence of a third replication mechanism involving recombination (Recombination-Dependent Replication or RDR) observed in adult human heart (Pohjöismaki et al, 2013). The only thing that is clear is that we still have a lot of work to do in order to uncover the mechanism/s by which mtDNA is replicated in mammals.

Controversial topics:

  • Strand Displacement Model vs RITOLS: mtSSB or RNA covering the parental heavy strand.
  • Asynchronous (Strand Displacement Model or RITOLS) vs bidirectional (COSCOFA) replication: according to some people there are not enough evidences supporting COSCOFA (which is basically based on 2D-AGE and LM-PCR experiments) and only the SDM is real.
  • Is there recombination in mammalian mitochondria? although there is a number of evidences supporting it, the mitochondrial recombination machinery is still mostly missing.


  • Bell L, Byers B (1983). Separation of branched from linear DNA by two-dimensional gel electrophoresis. Anal Biochem. 1983 Apr 15;130(2):527-35.
  • Brewer and Fangman (1987). The localization of replication origins on ARS plasmids in S. cerevisiae. Cell. Volume 51, Issue 3, 6 November 1987, Pages 463-471.
  • Clayton (1982). Replication of animal mitochondrial DNA. Cell. Apr;28(4):693-705.
  • Friedman and Brewer (1995). Analysis of replication intermediates by two-dimensional agarose gel electrophoresis. Methods in Enzymology. Volume 262, 1995, Pages 613-627. Katherine L. Friedman.
  • Holt, Lorimer and Jacobs (2000). Coupled leading- and lagging-strand synthesis of mammalian mitochondrial DNA. Cell. 2000 Mar 3;100(5):515-24.
  • Pohjöismaki et al (2013). Overexpression of Twinkle-helicase protects cardiomyocytes from genotoxic stress caused by reactive oxygen species. Proc Natl Acad Sci U S A. 2013 Nov 26;110(48):19408-13. doi: 10.1073/pnas.1303046110. Epub 2013 Nov 11.
  • Torregrosa-Muñumer, R (2015). Low doses of ultraviolet radiation and oxidative damage induce dramatic accumulation of mitochondrial DNA replication intermediates, fork regression, and replication initiation shift. Mol Biol Cell. 2015 Nov 15;26(23):4197-208. doi: 10.1091/mbc.E15-06-0390. Epub 2015 Sep 23.
  • Yasukawa et al (2006). Replication of vertebrate mitochondrial DNA entails transient ribonucleotide incorporation throughout the lagging strand. EMBO J. 2006 Nov 15;25(22):5358-71. Epub 2006 Oct 26.

Euromit 2017

These are the projects we are planning to present in our next conference in Cologne (Germany): Euromit 2017.

Mitochondrial Topoisomerases (Anu hangas)

Topoisomerases are essential enzymes for replication, transcription and repair of DNA in any known organism as they regulate the supercoiling and winding of DNA molecules. While their function is rather well characterized in nuclear DNA metabolism, surprisingly little is known about topoisomerases in mitochondria, where the constant expression and replication of mitochondrial DNA equally requires their function. Information about those enzymes is urgently needed as topoisomerase inhibitors are broadly used, e.g. in cancer therapy or as antibiotics often having severe side effects.

To address the function of Top2β in mitochondria, we studied mitochondrial DNA alterations upon knockdown of Top2β. In addition, we investigated the effects of a suspected topoisomerase 2 inhibitor, the commonly used antibiotic ciprofloxacin, on mitochondrial DNA maintenance in proliferating HeLa cells and in differentiated C2C12 cells.

Role of PrimPol in mtDNA replication (Rubén Torregrosa-Muñumer)

During the mitochondrial DNA (mtDNA) replication, replication fork can stall due to damaged DNA template or deleterious mutations in enzymes required for mtDNA maintenance. The stalled replication forks can collapse, resulting in double-strand breaks and the loss of partially replicated genomes, being a potential cause of the pathological mtDNA deletions seen in human diseases. Replication stalling occurs also in the nucleus, where multiple pathways are involved in replication fork resolution, rescue and re-initiation. However, the fate of stalled replication forks in mitochondria is unclear. To make even more complicated the understanding of mtDNA maintenance, there are indications suggesting that at least two mechanistically different mtDNA replication mechanisms operate in mammals, the so-called RITOLS and COSCOFA. Interestingly, oxidative and UV damage seems to specifically stall RITOLS intermediates and induce COSCOFA replication. It might be that RITOLS represents a high fidelity, housekeeping mode, whereas COSCOFA is a more robust replication mechanism operating under stress.

Figure 1 – mtDNA replication fork stalling: during mtDNA replication, the mtDNA replication fork can stall due to damaged DNA template, mutated mtDNA maintenance proteins or inhibitors of the mitochondrial DNA polymerase, among others. Prolonged mtDNA stalling can lead to fork collapse and result in rearrangements and deletions, with dramatic consequences for the mitochondrial function. However, nothing is known about the fate of mtDNA stalled replication forks. In the nucleus, fork collapse can be resolved by double strand break formation and repair or fork regression to restart of replication. In another scenario, PrimPol has risen as a potential re-priming enzyme, which seems to allow replication restart downstream the lesion.

For the last few years our research group has been trying to understand the interplay between replicative stress and replication mode switching in mitochondria. Our investigations have concentrated on recently discovered primase-polymerase enzyme PrimPol, which is located both in the nucleus and in the mitochondria. Although PrimPol was firstly identified as a translesion polymerase, able to bypass damage on the DNA template, together with some other recent studies our results give more importance to the priming activity of PrimPol, including the priming of lagging-strand replication at unconventional origins on mtDNA. We show for the first time that PrimPol, although it is not an essential enzyme for mtDNA maintenance, it plays a significant role in maintaining replication under genotoxic stress and possibly explaining some of key features of the different replication mechanisms.