Saturday, December 29, 2007

DNA Denaturation and Renaturation and the Role of Hydrogen Bonds and Stacking Interactions

Several students have written to me with questions about the structure of DNA. The most troubling questions are from students who have read the article I wrote about a paper that measures the stacking interactions in polynucleotides [Measuring Stacking Interactions]. In that posting I wrote ...
The two strands of double- stranded DNA are held together by a number of weak interactions such as hydrogen bonds, stacking interactions, and hydrophobic effects [The Three-Dimensional Structure of DNA].

Of these, the stacking interactions between base pairs are the most significant. The strength of base stacking interactions depends on the bases. It is strongest for stacks of G/C base pairs and weakest for stacks of A/T base pairs and that's why it's easier to melt A/T rich DNA at high temperature. (It is often incorrectly assumed that this is due to having only two hydrogen bonds between A/T base pairs and three between G/C base pairs.)

Deoxyribonucleic Acid (DNA)
Many students have written to say that my statements contradict their Professors and their textbooks. I'm not surprised. The old-fashioned view of DNA denaturation (pre-1990) supposed that the differences between A/T rich DNA and G/C rich DNA were due to the extra hydrogen bond in G/C base pairs. Many of the Professors who teach introductory biochemistry aren't aware of the fact that this view is incorrect. Even more surprising, some of the current textbooks have not bothered to update their material on the structure of DNA.

Here's the story as we know it today. For those students who have written to me, I repeat the caution I mentioned in my reply to you—be sure to check with your Professor before you write any tests. Make sure he/she understands why you are contradicting what was said in class so you don't get marks taken off. It's always better to do this in advance instead of arguing your case after you have lost marks on the test.

Let's look first at what happens when DNA is denatured by raising the temperature.

As the temperature increases, you start to get local unwinding of the double-stranded DNA. This unwinding occurs preferentially in regions where the two strand are held together less strongly. In these regions the strands separate to form bubbles of single-stranded regions. The DNA sequence in these regions is enriched in A/T base pairs because the interactions between the two strands are weaker in A/T rich regions. In G/C rich regions strands are held together more strongly so they don't unwind until higher temperatures.

Incidentally, even at normal cell temperatures the DNA "breathes" and local regions become temporarily unwound. As you might expect, A/T rich regions are more likely to open up than G/C rich regions. This is one of the reasons why transcription initiation bubbles and DNA replication origins are often A/T rich. It's easier for the proteins (RNA polymerase, and origin binding proteins) to create the locally unwound regions.

When all of the base interactions are broken, the two strands separate. This is called denaturation. (Local unwinding is not denaturation.)

The base are now exposed to the aqueous environment. Single-stranded DNA is more stable than double-stranded DNA at higher temperature. Note that the edges of the bases will still form hydrogen bonds in this situation. They form hydrogen bonds with water molecules. In fact, they will form many more hydrogen bonds with water than they would form with complementary bases in double-stranded DNA.

As the temperature is lowered, the double-stranded form becomes more stable than the single strand in solution, and the DNA renatures. The first step is a nucleation event where two complementary regions come into contact. Nucleation is the rate-limiting step in renaturation. Once nucleation occurs, the rest of the molecule zips up pretty quickly.

It's easy to follow the denaturation of DNA because there's a difference in the absorbance of ultraviolet light between single- and double-stranded DNA. Single-stranded DNA absorbs more strongly.

In a typical melting curve, you measure the increase in UV absorbance as the temperature increases. This tracks the unwinding and denaturation of DNA. The melting point (Tm) is the temperature at which half the DNA is unwound.

DNA that consists entirely of AT base pairs melts at about 70° and DNA that has only G/C base pairs melts at over 100°. You can calculate the Tm of any DNA molecule if you know the base composition. The simplest formulas just take the overall composition into account and they are not very accurate. More accurate formula will use the stacking interactions of each base pair to predict the melting temperature [Wikipedia: DNA melting].

The question is why is there a relationship between the base composition of DNA and the stability of the double-stranded regions?

The first people to think about this question didn't really understand the role of stacking interactions between base pairs in the middle of double-stranded DNA. They also didn't really appreciate hydrogen bonds. They naively assumed that the differences between G/C rich DNA and A/T rich DNA was due to the fact that G/C base pairs have three hydrogen bonds and A/T base pairs have only two [The Chemical Structure of Double-stranded DNA].

We now know that this explanation doesn't make sense. There is no net loss of hydrogen bonds when DNA is denatured, quite the reverse in fact. There are more hydrogen bonds formed between the bases in single-stranded DNA and water molecules than between base pairs in DNA. There's no reason why single-stranded DNA would renature if formation of double-stranded DNA was driven by the creation of hydrogen bonds between base pairs. For every hydrogen bond between bases you would have to break almost two hydrogen bonds to water molecules.

The most important interactions in double-stranded DNA are the stacking interactions between adjacent base pairs. You can think of this as the interactions of electrons on the upper and lower surfaces of the rings that form the bases.

There are ten possible interactions between adjacent base pairs. The energies of these interactions are shown in the table on the left. The arrows indicate the direction of the DNA stand from 3′→5′ [The Chemical Structure of Double-Stranded DNA].

Note first of all that the strength of these stacking interactions (about 30 kJ mol-1 on average) are greater than the strength of stability conferred by hydrogen bonds (about 3 4 kJ mol-1)1. Assuming there are on average six three hydrogen bonds per in two stacked G/C base pair, the total strength of the hydrogen bonds (18 12 kJ mol-1) is still much less than the stacking interactions.

Secondly, note that stacking interactions involving G/C base pairs are stronger (more negative) than those involving A/T base pairs. This is why the melting temperature of DNA depends on the base composition. It's not because G/C base pairs have one more hydrogen bond than A/T base pairs, it's because G/C base pairs form stronger stacking interactions.

This is why you can calculate a more accurate melting temperature for oligonucleotides if you use the stacking interactions. It's stacking interactions that determine the stability of double-stranded DNA and it's stacking interactions that are disrupted as the temperature increases and more thermal energy is added to the molecule.

Finally, the paper that I discussed in July [Measuring Stacking Interactions] measured the stacking interactions in single-stranded DNA (poly A). As it turns out, the stacking interactions between single bases are, in some cases, strong enough to force single-stranded DNA into a helical structure. This is further evidence of the importance of stacking interactions in conferring stability to the double helix.

1. The stability conferred by each hydrogen bond is the difference between the strength of the bond in double-stranded DNA and its strength in when bonded to water. Hydrogen bonds between bases and water molecules typically have strengths of about 25 kJ mol-1 and hydrogen bonds between base pairs are a bit higher.


  1. That is such a lucid explanation that it almost makes me wish that I were still teaching- I would make use of this post for sure.

  2. Why do the stacking interactions between adjacent bases depend on the DNA being base-paired into a double helix?

    Your table gives values for the interactions between adjacent base pairs. Am I right in thinking that the individual stacking interactions are between adjacent bases in the same strand? E.g. for a strand with sequence 5'CAGTG base-paired to a complementary strand, the A at position 2 in strand 1 makes stacking interactions with the C at position 1 and the G at position 3 in its own strand, and not with bases in strand 2.

    If so, why do these stacking interactions not also occur when the strand is not base-paired?

    I can see that, when DNA is single-stranded, the relative positions of adjacent bases might be less constrained (if any of the backbone bonds are free to rotate). But if the stacking interactions are more energetically favourable than hydrogen bonds formed with water, they should form and constrain the bases into stacked relationships even in the absence of a complementary strand.

    What am I missing?

  3. Rosie, the bases in single-stranded polynucleotides do stack but, as you noted, the backbone of single-stranded polynucleotides is not as constrained as it is in double-stranded DNA (or RNA).

    Furthermore, the van der Waals contacts between two adjacent base pairs is significantly greater than between single bases. (Think of a base pair as a single molecule.)

    In double-stranded DNA the base stacking interactions "lock in" the helix and the cumulative cooperative effect is enormous compared to the more flexible single strands. This is why DNA renaturation is enthalpically driven whereas protein folding is entropically driven.

    See the first figure in The Three-Dimensional Structure of DNA to get an impression of what the stacking interactions are doing.

  4. I believe Dr. Pearson used the fact that the stacking interaction energy of GpC is greater than CpG to a good pedagogical effect. Maybe you should mention it the next occasion you need to talk about DNA stability, because that factoid really stuck in my head due to its superficial counterintuitive-nature.

  5. Moran,
    It seems incredible to me, that you apparently feel that describing complex biological interactions and relationships somehow constitutes evidence for large scale evolutionary processes. To many, observations of living systems that take much study to describe, let alone explain lead them to believe that much more knowledge must be obtained in order to make any kind of legitimate statements regarding how these living systems "develop" through time given the environmental and chemical restraints whose impact must be fully and considerably addressed before making a "professional" claim that it has been shown to be a "fact" as so many "evolutionary proponents (and I would imagine you are one of them) assert.

    It gives one the impression that there is some sort of bias in the interpretation of data that could lead to claims regarding chemicals to life systems that may not have significant evidential matter to support the claims.

    mOran, Look, I think you are probably excited about what you do here. But i just wonder if you have some sort of personal philosophical agenda that drives you interest in you participation on a public forum.

    It is extremely agitating to many of us that someone takes a public forum "by the neck" to perpetuate what any honest scientist would call "assertions" in light of neutrally interpretated evidences regarding chemicals to life, and demand the data be vastly prematurely interpreted in a way that bolsters a personally favored interpretation. (YOu dum_ f__)

  6. Right on anonymous!!!!!!!

    Moran is a moron!!!!!!

    Sick of this unsubstantiated shit!!!!

  7. this is just like the central dogma issue. every course i've taken has gotten it wrong, including Biochemistry 370 (biochem lab). Maybe that was intentional b/c it's a course designed for non-biochemistry students?

  8. Back to the discussion of energetics. :) What is the reference for the table of base stacking energies? If CG/GC has a stacking energy of -61 kJ/mol, how do helicases that hydrolyze one ATP per bp unwound (like uvrD) work? Wouldn't the enzyme need to be nearly energetically perfectly efficient to pull this off? Could the enzyme even function?

    The value in the post of -3 kJ/mol for a single hydrogen bond... that should be -3 kcal/mol right? If most hydrogen bonds are somewhere from -3 to -8 kcal/mol then that would be -12 to -34 kJ/mol, putting base-pairing and stacking interaction energies in the same kind of ballpark.

  9. It seems incredible to me, that you apparently feel that describing complex biological interactions and relationships somehow constitutes evidence for large scale evolutionary processes.

    I assume you're asserting it doesn't. Present your alternate explanation. Cite your research.

    But i just wonder if you have some sort of personal philosophical agenda that drives you interest in you participation on a public forum.

    And I just wonder if you're a stereotypical chickenshit creationist troll who posts vapid, rambling comments on science blogs without demonstrating he has any real expertise in the subject he presumes to critique, and to which he's too cowardly to sign his name.

  10. On the energetics question, I never knew helicases expend 1 ATP per a base unwound; taking it provisionally, however, the average stacking interaction from the table is 34.3 kJ/mol. This value is close to the standard free energy change of ATP hydrolysis to ADP, which I believe is -32 kJ/mol (if I remember BCH242 correctly!).

  11. Dr. Moran,

    Thanks again for another clear and enlightening post. I was another one who'd always naively assumed the difference in melting temperatures of GC-rich and AT-rich DNA were simply due to counting Hydrogen bonds. Now I understand why nobody was willing to predict the optimal temperature for my PCR reactions with various primers.

  12. I still don't think Rosie's question has been adequately addressed.

    A single strand *could* adopt the same helical conformation as in a double strand. Yes, constraining the backbone configuration would be less favorable entropically, but it would still have the favorable stacking & van der waals interactions between adjacent bases. If the net is favorable in a double strand, why not in a single strand?

    Something still seems to be missing. Any evidence that stacking interactions between base pairs are more enthalpically favored than the sum of the potential stacking energies for the individual strands?

  13. Neal said in a prior post, yes hey Wagonass- I will give you my identity.

    "It seems incredible to me, that you apparently feel that describing complex biological interactions and relationships somehow constitutes evidence for large scale evolutionary processes."

    Then the "Band Wagoner" Said"
    "I assume you're asserting it doesn't. Present your alternate explanation. Cite your research."

    Neal says:

    Since when the f______ does "real science" start out with propositions that it holds to be true and demand alternative, non-sympathetic observers who legitimately point out the short falls of the assertions (who base their arguments on the real currently scientifically verifiable evidence available) move into a legitimate position of commanding authority? God, when this happens, and I think it is, there are big problems, after taking into consideration the important role that REAL science has in the world today!!!!! Wagoner, if you are a real scientist, and there are significant numbers of others like you is the "profession", this country and the rest are probably in some pretty big trouble, (you dum_ a__s)

    I will say my name again! NEAL, NEAL, NEAL , NEAL , NEAL , Is that good enough for you you idiot?????

  14. Wagoner said:
    And I just wonder if you're a stereotypical chickenshit creationist troll who posts vapid, rambling comments on science blogs without demonstrating he has any real expertise in the subject he presumes to critique, and to which he's too cowardly to sign his name.

    NEAL said:

    I think i have answered the "cowardly" question in that last post.
    The rest of your diversionary assertions, you should reconsider and, if you were honest, would encourage anybody else on this forum to do the same!!!!!

    The "expertise" comment has many facets which, if you were really interested in understanding, would require you to essentially "open your mind", set aside your prejudices, philosophical preferences and whatever else, and LOOK AT THE EVIDENCES FROM NON-BIASED ACTUALLY SCIENTIFIC POINT OF VIEW!!!!(S) No one can command that you do that and have that kind of phenomena occur. You are responsible if anybody is, for your responses to informational stimulus, you have the (some folks might say) privilege of having the equipment to comprehend, that (guess the f--k what) has yet to be anywhere near described let alone explained "chemicals to ecosystems" even taking into consideration the "mythical" multiverse concept that those who want to continue to suck off of society continue to "promote" to those (the chickens in the henhouse will continue to swallow regardless of the infinite lack of evidences)But you and yours will continue to be confronted with legitimate questions regarding your assertions you have had the unmitigated (until now) privileges
    of "asserting down the throats" of the unsuspecting public in the name of "Science"!!!!!!!
    You pissz______t!!!!


  15. Bleh, I messed up. Earlier, I meant to say CpG has a greater stacking interaction energy than GpC. With respect to the energetics question posed earlier by anon., the bond energy of hydrogen bonds at 3 kJ/mol sounds right; anything more would be a "real" bond (e.g. covalent bond). I believe if the bond energy of hydrogen bonds was more, we wouldn't have very watery water, and conversely if the bond energy was less, water would be a gas.

    Qetzel, I believe Larry may have already addressed your questions. In the actual post, he mentions that single strand poly-A DNA can form a helical conformation solely by the strength of stacking interactions. So in at least one instance, the stacking interactions of adjacent adenine bases enthalpically overcomes the entropic barrier posed by constraining the sugar-phosphate backbone of single-stranded DNA. There is no reason to expect that there aren't other ssDNA sequences that can form regular structures due to stacking interactions, but I think the reason why poly-A ssDNA works is because adenine is a purine (bigger base); perhaps A/G-rich ssDNA can form regular structures as well. Also, he mentions that double-stranded DNA has more stacking interactions than ssDNA because one can think of base-pairs as single molecules (i.e. more interacting surface area than two adjacent bases). Even if one base in dsDNA swivels out (which it does, because DNA is a dynamic structure), the collective stability of dsDNA due to stacking interactions/hydrogen bonds/entropic considerations ensures that the base has time to swivel back in and preserve the structure. Finally, I believe ssDNA of a known sequence can form predictable globular structures when the temperature is lowered, and stacking interactions (although not necessarily of adjacent bases) probably enthalpically drive to a degree the formation of such a globular structure.

  16. Opps, scotch the sentence about A/G-rich ssDNA. G-rich ssDNA would probably form guanine quartets, not a regular structure.

  17. The strength of the hydrogen bond in water is (optimally) 23.3 kJ/mol.

    [Hydrogen bond thermodynamic properties of water from dielectric constant data
    S. J. Suresh and V. M. Naik
    J. Chem. Phys. 113 (2000) 9727-9732.]

    Theoretical and physical measures of hydrogen bonding strength in nucleotide pairs shows:

    G:C -87.9 kJ/mol
    A:T -54.5 kJ/mol

    [Hydrogen bonding of DNA base pairs by consistent
    charge equilibration method combined with
    universal force field
    Tetsuji Ogawa, Noriyuki Kurita, Hideo Sekino,
    Osamu Kitao, Shigenori Tanaka
    Chemical Physics Letters 374 (2003) 271–278]

  18. Dunbar,

    No, I don't think that answers the question.

    I agree that some stacking still occurs in ssDNA. (That's partly why even ssDNA shows a hyperchromic effect upon heating, IIRC.)

    The point is, if stacking is the primary force in forming dsDNA, that implies that the stacking energy is greater in dsDNA than in ssDNA. If so, then why?

    Thinking of base pairs as single molecules doesn't obviously answer the question either, unless the stacking energy of two successive base pairs is greater than the sum of the stacking energy of the two successive bases in each single strand.

    I vaguely remember that some cross-strand stacking may occur in dsDNA, depending on the sequence (e.g. between a purine in one strand and an adjacent purine in the opposite strand). However, I'm not sure if I'm remembering that right. Alternatively, maybe the pi-orbitals hybridize to some extent across the entire base pair, and that leads to greater stacking than would otherwise occur?

  19. This comment has been removed by the author.

  20. Sorry professor, I'm too dumb. Does the table of stacking energies tell us that even two DNA strands have the same numbers of A-T and G-C pairs, their Tm can still be different?

    Also, what exactly is stacking effect o_O ?

    1. That's actually a very smart experiment to do. One should be able to design such oligos and measure the heat released during binding process with ITC. If the heat released differs then the order of nucleotides plays a role in renaturation; otherwise, the model would be inaccurate. The stacking energies posted on this webpage is from Larry's textbook "Biochemistry". The version I have is the second edition, page 24.14. The table was adapted from Ornstein, R.L. et al, (1978), Biopolymers 17:2341-2360. I wonder how the values were measured. Given Larry's lab expertise in molecular evolution, some data must have been collected to support his assertion, I hope. Robert Cedergren
      also mentioned base stacking in a chapter that he wrote in Elementary Mol. Bio., and he argued that it could provide more rigidity than expected to a single-stranded polynucleotide sequence. While William Galley, who teaches phys chem at McGill, is a strong proposer of the entropy model of renaturation.

      However, it doesn't matter how Larry was known among the young undergrads as the "snob" and rejected hundreds of students who wish to study biochem at UofT each year, or how 70-year-old William made his golf swing on top of a desk in the front of the class room and brags about how he beat a black kid with initials of T.W., personally I don't see why the stacking energy contribution and the entropy contribution from H-bonding with water should be exclusive. Looks like it could be both entropy- and enthalpy-driven, if the numbers of kj/mol differ a lot, then we may just call it a "mostly-winner"-driven process.

  21. anonymous says,

    Theoretical and physical measures of hydrogen bonding strength in nucleotide pairs shows:

    G:C -87.9 kJ/mol
    A:T -54.5 kJ/mol

    [Hydrogen bonding of DNA base pairs by consistent
    charge equilibration method combined with
    universal force field
    Tetsuji Ogawa, Noriyuki Kurita, Hideo Sekino,
    Osamu Kitao, Shigenori Tanaka
    Chemical Physics Letters 374 (2003) 271–278]

    I'm sorry. I wasn't clear in my posting so I've made a correction.

    I was referring to the contribution to DNA stability conferred by a hydrogen bond. This is the difference between the strength of the bond in a base pair (e.g. 28 kJ mole^-1) and in water (e.g. 24 kJ mole^-1.

    See Kool et al. (2000)
    Are hydrogen bonds necessary for stabilization of the DNA helix? Since Watson and Crick's elucidation of the duplex structure, it has been generally accepted that the hydrogen bonds between the bases are critical to the stability of DNA. A Watson-Crick pair has two or three such bonds, and these add approximately 0.5 - 1.8 kcal mol^-1 of stabilization per base pair of DNA. The uncertainties in this value arise from the fact that the experimentally measured value varies greatly depending on the molecular context. Of course, this relatively low energetic stabilization reflects the competition with water, because breaking each hydrogen bond in a base pair (by separating the bases) also means the simultaneous formation of two hydrogen bonds with water molecules. In the gas phase, where such competition does not occur, experiments show that individual hydrogen bonds add stabilization of 6 - 7 kcal mol^-1 with the kinds of functional groups found in DNA. Although DNA has been studied for decades, the role of hydrogen bonds in stabilization of the helix and selective base-pair formation is still unclear.

    The stacking interactions are for formation of stacked bases from unstacked bases in water so they represent a real gain in stabilization energy.

  22. If strand separation is hard for GC-rich sequences, why CpG islands are transcirptionally strong and why not the repeats with similar GC-richness ?

  23. Peter asks,

    If strand separation is hard for GC-rich sequences, why CpG islands are transcirptionally strong and why not the repeats with similar GC-richness ?

    CpG islands are site of methylation (the "C" is methylated). Methylation is thought to regulate gene expression and that's why you find regions that are enriched in CpG near genes.

    The CpG islands are far away from the promoter and the beginning of the gene so the strands are never separated during transcription.

  24. I have been perusing some of your pages here, they are pretty well done I must say. Thanks for put an obviously significant effort into them.

    However, I do have a comment on this particular page. I agree with much of your discussion on the hydrogen bonding of the DNA to both itself and to water. However, I feel you have neglected to fully describe one important detail in regard to DNA-DNA and DNA-water hydrogen bonding. You do correctly say that DNA will form h-bonds with water (more than with itself in dsDNA form); however, you have neglected the entropic contribution to this problem. Whilst DNA does form h-bonds with water it comes at a heavy entropic cost, this cost is in fact why the DNA preferentially binds to itself and not the water. In other words DNA does not h-bond with water because it is entropically unfavourable and ultimately DNA - DNA binding is an entropy driven effect. A point that is both not trivial and generally not obvious.
    It's a fairly straightforward calculation to illustrate this, for example one might take a look at the book: "Intermolecular Surfaces and Forces" to more fully appreciate this effect.

  25. anonymous says,

    ... ultimately DNA - DNA binding is an entropy driven effect.

    No, it is not. Renaturation is an enthalpy driven reaction, unlike protein folding. That's the point.

    It's why scientists realized that the hydrogen bonding isn't very important. The enthalpic change comes from base stacking. We use DNA renaturaion and denaturaion to illustrate the difference between double-stranded DNA stability and the stability of a folded protein—which is entropically driven.

    The interior of double-stranded DNA is hydrophobic but not enough to make a difference in stability.

  26. I should clarify, I am not saying that the DNA bases being slightly hydrophobic is the driving force in binding, I am saying that the base binding is a result of the fact that breaking reducing hydrogen bonds in water is very entropically unfavourable. I am not referring to only to the entropy of the DNA, I am referring to the water + DNA, breaking the hydrogen bonds in water is indeed part of the driving force in renaturation and that is entropic in nature.

  27. it's easily to note when you extract DNA, the last step is to resuspend it in water, because of the h-bonds that for the DNA with water, these can be done.

    Moran, i had been doing the DNA-DNA hybridization technique, your proposal can be development more accurately if you denatures and renatures genomes with different GC content organisms. Also, the absorbance in this technique can be a parameter for your fact.


  29. Sir,
    I'm a student of Microbiology (UG) 1st year. And one of my professors already taught in class that the H-bond is the reason of melting and renaturation. If you will kindly make me understand what is staking interaction, then I will be more helpful in my studies. I know my question is very naïve, but so am I - new and naïve in this field.
    Thank you.

  30. I know this is an old post, but thanks a ton for the explanation! It's really clear and informative.

  31. its very interesting to read morans explanation which i should say has been well documented. DNA yesterday, today and tomorrow is the same and i dont think any alternative explanation will outrightly replace the basic concepts (works done and published by erudite and eminent scientists) that has been taught and learned over the years.