Hi Tias,

Really great question! Actually DNA synthesis does proceed in both directions, just by a bit of a different mechanism! When the DNA strands are separated to form a single-stranded bubble at an origin of replication, a short RNA molecule is synthesized complementary to each strand (Okazaki fragment) to serve as a primer for DNA synthesis. DNAP cannot initiate DNA synthesis de novo like RNAP does, so it uses the RNA primer as a double stranded starting point. This occurs in both directions. But to actually extend the a single DNA molecule in both directions as you asked, several people pointed out 2 different substrates (dNTPs) would be required for the 2 distinct enzymatic reactions (although tdT (terminal deoxynucleotide transferase) does 3’ extensions just fine with plain, ordinary dNTPs), in addition, another enzyme would be needed to remove the Okazaki fragment, replace it with DNA, and then ligate that to the 2 flanking, newly replicated DNAs. Faster maybe, but probably not easier. That might possibly allow for more rapid genome replication, which could potentially provide selective advantage to fast replicating organisms such as bacteria and viruses. Probably not as significant an issue for multicellular organisms.

Also, importantly, evolution does not operate to generate the best long-term solution. Its selection process is always current and local. Given the PRESENT choices and environment, which option best promotes survival and subsequent reproduction? Actually, I think the choice needs to be viewed in retrospective – which option best promoted survival and reproduction because concurrent options are always in competition. Most new features evolve gradually from preexisting features (legs from fins, hands from feet, brains from ganglia) and utilize/co-opt preexisting developmental signaling pathways. A new feature that simultaneously requires too many novel functions or adaptations (say ESP or self-teleportation :)  ) is less likely to evolve, even if it was physically possible and  would present a tremendous competitive advantage. I think the jump from unidirectional DNA replication to bi-directional might fall in this category (although the existence of the tdT enzyme certainly argues otherwise). Instead, what did evolve, was the acquisition of error correction – utilizing the reverse direction for polymerase as the eraser function to rid mistakenly incorporated bases (recognized due to the antisense strand nature of DNA!) The resulting importance of that can be seen with a comparison of PCR products utilizing plain, old-fashioned Taq and error-correcting enzymes such as Phusion and Q5 (mass vs quality of product).

Regarding antiparallel vs parallel DNA strands, parallel might simplify genome organization because all genes would have to be oriented in the same direction and on the same strand. But in actuality, genomic DNA is not a static linear (2D) molecule. It is often in cruciform (cross-shaped) or more complicated structures due to proteins binding to local inverted repeat sequences, with major roles in both local gene regulation and global genome structural organization.

Hello Tias,

In my opinion, the evolutionary basis for DNA polymerase would be the type of Genome present in primates. They all had single stranded genome and hence only require one direction of synthesis. Also during that period, there was no need of two DNA polymerases which have different direction as the genome was very simple. Another angle would be requirement of cell growth. In early days, there was not much competition for survival & so low rate of growth and hence no complexity was introduced at genome level.

I hope this would helpful.

Thanks RAJESHKUMAR!

But, I feel, when DNA was selected as genetic material, then after that, DNAP could have evolved to be capable of synthesizing in both the directions. Why is this DNAP selected over any other?

Hello,

One reason for that could be efficiency of DNAP. As it is very accurate and cell had decided to go with this, there might be no need of second DNAP.   

Not a second. This could have evolved to have better functions (duality in direction). 

More complex systems are not always better. Sometimes, nature keep things simple. DNA is highly regulated, adding another polymerase (or bidirectional pol) would probably just make the process of replication even more messy. You would need several more layers of regulation as the pol would bind any 5' and 3' DNA strand. This would probably lead to higher genome instability, especially in recombination.

You could ask the same question on the chirality of sugar in nature, but the answer is probably the same. The ratio effectiveness/cost was not good enough to drive evolution toward a wide development of a new set of enzyme.

Isn,t it related to the dNTP? Since dNTP have triphosphate at the 5' end only and that is the one available for 3' OH to make a phospho-diester bond. I am not aware of dNTP having OH at the 5' end and triphosphate at the 3' end.

But I think, making replication process unidirectional, makes the entire process slower (as lagging strand would not allow the polymerisation as fast as leading strand, due to multiple re-priming). If the process is slow then there is more caution and lesser chances of mis-incorporation.  It is hence a way to regulate the speed of replication making it more efficient. 

@Rajesh Kumar

Do you really mean PRIMATES had single stranded genomes?

Parallel arrangement of DNA strands in dsDNA is thermodynamically impossible.

Sometimes going slower is better than making an error.

Replication in only one direction i.e. 5' to 3' is the only way which allows error free correction. The proof reading activity of  the DNA  polymerase hydrolyses the wrongly incorporated nucleotide for incorporation of the correct nucleotide.  

Please  look at the  attached  figure  for DNA synthesis to occur either in 5' to 3' direction or 3'-5' direction..

There you would see that the energy for condensation of the phosphodiester bond in the 3' to 5' case is on the 5' end of the growing chain.  So in case i.e. a  wrong base is incorporated and exonucleolytically removed, there will be no energy on the resulting monophosphate end for the condensation to occur. On the other hand, in case of synthesis from 5' to 3', the energy for condensation is always the triphosphate of the incoming base. This way, when editing occurs, the energy to make the bond is still present. Therefore, it appears that replication in the 5' to 3' direction is dictated by the need for the polymerases to edit their work.

You can also please refer  to  Molecular Biology of the Cell by Bruce Alberts

(http://www.ncbi.nlm.nih.gov/books/NBK26850/) and search for

"Only DNA Replication in the 5′-to-3′ Direction Allows Efficient Error Correction" Fig. 5-11 on this link. 

Tias: What would be the advantage of synthesizing DNA in both directions?

---

It could not be done with one DPOL; there would have to be specific DPOL for each direction (5'->3' vs 3'->5'); they would compete for the same substrate and would have to be about equally fast/accurate: No net benefit plus the organism would have to keep two DPOLs for a job that can be done by one of them alone. Not to mention the "traffic accidents" of two DPOLs driving in opposite directions "in the same lane" . . .  

Besides, there would have to be two sets of dNTPs: 2'-deoxyribonucleotide-5'-triphosphates (for DPOL acting in 5'->3' direction) and 2'-deoxyribonucleotide-3'-triphosphates (for DPOL acting in 3'->5' direction). Do the latter dNTPs actually exist as stable chemical entities?

Dear Klvana,

Excellent explanation!

But I would like to add few things.

Advantage: faster replication, no hazard of trombone model! (may be)

What if DNAP had two subunit, each dedicated for each direction?

Why don't the other type of dNTPs (3' triphosphates) exist? I think, they'll be stable as the 3'OH of sugar can easily have a nucleophilic attack on phosphate as it happens during polymerization reaction. Why is 5' triphosphate synthesis favoured over the other? Is there no enzyme which can catalyse this reaction? If not, then why are the 5' triphosphate synthesizing enzymes evolutionary selected? Is it because nature wants to maintain this bidirectional and anti parallel nature of DNA?

Hi Tias,

Really great question! Actually DNA synthesis does proceed in both directions, just by a bit of a different mechanism! When the DNA strands are separated to form a single-stranded bubble at an origin of replication, a short RNA molecule is synthesized complementary to each strand (Okazaki fragment) to serve as a primer for DNA synthesis. DNAP cannot initiate DNA synthesis de novo like RNAP does, so it uses the RNA primer as a double stranded starting point. This occurs in both directions. But to actually extend the a single DNA molecule in both directions as you asked, several people pointed out 2 different substrates (dNTPs) would be required for the 2 distinct enzymatic reactions (although tdT (terminal deoxynucleotide transferase) does 3’ extensions just fine with plain, ordinary dNTPs), in addition, another enzyme would be needed to remove the Okazaki fragment, replace it with DNA, and then ligate that to the 2 flanking, newly replicated DNAs. Faster maybe, but probably not easier. That might possibly allow for more rapid genome replication, which could potentially provide selective advantage to fast replicating organisms such as bacteria and viruses. Probably not as significant an issue for multicellular organisms.

Also, importantly, evolution does not operate to generate the best long-term solution. Its selection process is always current and local. Given the PRESENT choices and environment, which option best promotes survival and subsequent reproduction? Actually, I think the choice needs to be viewed in retrospective – which option best promoted survival and reproduction because concurrent options are always in competition. Most new features evolve gradually from preexisting features (legs from fins, hands from feet, brains from ganglia) and utilize/co-opt preexisting developmental signaling pathways. A new feature that simultaneously requires too many novel functions or adaptations (say ESP or self-teleportation :)  ) is less likely to evolve, even if it was physically possible and  would present a tremendous competitive advantage. I think the jump from unidirectional DNA replication to bi-directional might fall in this category (although the existence of the tdT enzyme certainly argues otherwise). Instead, what did evolve, was the acquisition of error correction – utilizing the reverse direction for polymerase as the eraser function to rid mistakenly incorporated bases (recognized due to the antisense strand nature of DNA!) The resulting importance of that can be seen with a comparison of PCR products utilizing plain, old-fashioned Taq and error-correcting enzymes such as Phusion and Q5 (mass vs quality of product).

Regarding antiparallel vs parallel DNA strands, parallel might simplify genome organization because all genes would have to be oriented in the same direction and on the same strand. But in actuality, genomic DNA is not a static linear (2D) molecule. It is often in cruciform (cross-shaped) or more complicated structures due to proteins binding to local inverted repeat sequences, with major roles in both local gene regulation and global genome structural organization.

Tias Saha wrote: "Advantage: faster replication."

Why do you think that running the nucleotidyl-transfer reaction in both 5'->3' and 3'->5' directions will be faster than 5'->3' alone?

Tias Saha wrote: "What if DNAP had two subunit, each dedicated for each direction?"

Two modules. Completely different active sites. Different reaction mechanisms. Extremely unlikely to evolve by duplication and adaptation. Difficult to imagine how the two modules will catalyze the reactions simultaneously. But more importantly, the two modules would come from two separate DNA polymerases (competing and colliding) . . . so the living thing will have to resolve the "pick one or die" dilemma or ---- glue the two modules together. Extremely, extremely fanciful. 

Tias Saha wrote: "Why don't the other type of dNTPs (3' triphosphates) exist? I think, they'll be stable . . ."

If they are stable, they can exist. If they do not exist it is because they are not stable entities. Evolution can act on an array of possibilities. What is possible depends on the thermodynamics and (as Craig M Neville noted) on the context. As the system becomes more complex, even chemically possible reactions might not be incorporated into the living thing because of the constraints of the complex system. But as for the nucleoside-3'-triphosphates, I think that the problem is in the chemistry. but I am not organic (synthetic) chemist to make the claim. So, let's ask organic chemist . . . 

Tias Saha wrote: ". . . as the 3'OH of sugar can easily have a nucleophilic attack on phosphate as it happens during polymerization reaction."

Note that the formation of O3'-PA bond is accompanied by PA-O3A bond cleavage, which yields the pyrophosphate byproduct; the extended DNA has a free 3'OH.

Most likely we started with RNA molecules. The problem is that they have what are called vicinal hydroxyls (on the 2' and 3' carbons), which very readily attack phosphodiester bonds. A 3' triphosphate would then be readily destroyed by a simple base-catalyzed reaction, which could easily happen in aqueous solution. Therefore, evolving to use only the 5' triphosphate system is to the benefit of long-term information storage, first in RNA, then in DNA. BTW, one of the reasons RNA is unstable is because of this room temperature catalysis; RNA is quite easily chewed up by itself, without the help of any exogenous RNAses.

If the polymerase ever did evolve to utilize 3' triphosphate molecules, an entire set of new enzymes would also have to evolve to add each phosphate, to ligate molecules, etc. The handedness and directionality have all evolved in concert, and adding one enzyme wouldn't really help unless every other step in the replication process were co-evolving.