Polymerase Errors I

Life can't exist without the high fidelity transmission of genetic information from parent cell to daughter cell. DNA polymerases plays the central role in accomplishing this. DNA polymerases delta and epsilon only make an error (that is, attach a non-complementary nucleotide to a growing chain) once every 100,000 to 1 million nucleotide additions. But this is still too many. Those that arise must be fixed via a back-up system.

In addition to polymerase errors, the human genome--even in healthy cells--is under constant assault from endogenous (internal to the cell) and exogenous (external to the cell, or environmental) chemicals and other agents that alter its chemistry, break its backbone, and distort its helical structure. We'll refer to such alterations as "lesions" to distinguish them from polymerase "errors." Lesions must be fixed, too.

In this and the next few post, we'll learn how human cells handle errors and lesions. The umbrella term for the cell's response to these threats is DNA repair. DNA repair consists of an array of specialized pathways that, together, correct replication errors, remove lesions large and small, and rebuild damaged regions of the genome. This and the next post will cover DNA polymerase errors. Then several after that will address the topic of DNA lesions.

DNA Polymerase Errors

The most common DNA polymerase errors are substitutions. These occur when a DNA polymerase adds a non-complementary nucleotide to a growing chain. For example, given an "A" on the template strand, the DNA polymerase might accidentally pair it with a "C" instead of the correct "T." Think of a substitution as a misspelling.

In addition to substitutions, there are two other types of errors made by DNA polymerase. Sometimes the DNA polymerase slips, and either fails to add one or more nucleotides (creating a deletion) or adds one or more extra nucleotides (creating an insertion). Insertions and deletions are rarer than substitutions, but the cell must fix these, too. Scientists refer to insertions and deletions as "indels."

Finally, polymerases make another kind of mistake much more frequently than substitutions. They insert a ribonucleotide instead of a deoxyribonucleotide into the growing chain. Ribonucleotides are the monomers that make up RNA. So, a polymerase might insert an RNA version of an "A" nucleotide instead of a DNA version of an "A" nucleotide across from a DNA "T" on the template strand. Ribonucleotides can't remain in DNA because they cause. strand breaks.

So cells need ways to correct all of these kinds of errors--substitutions, indels, and ribonucleotide incorporations--all of which are caused by the replicative DNA polymerases delta and epsilon, themselves.

Why DNA Repair?

Now, let's pause for a moment. You might be wondering why we're focusing attention to DNA repair given that our main topic is DNA replication. The two, it turns out, are largely integrated with each other. First, as I've mentioned, replication is the primary generator of substitution errors, small insertions and deletions, and ribonucleotide incorporations. The repair of most of these kinds of errors takes place inside the replications fork immediately after nucleotide addition. So error repair is physically coordinated with the replication.

In addition, as we'll see in coming posts, lesions can cause DNA replication roadblocks. They interfere with DNA polymerases and, in some cases, CMG helicases as they translocate along the DNA. A lesion can halt a replisome in its tracks. Thus, DNA replication is dependent on DNA repair. So we can't ignore it. Nor do we want to. It's one of the most amazing activities of cells!

DNA Polymerase Proofreading

The first level of protection from DNA polymerase errors is the polymerase's own proofreading function thanks to its 3' to 5' exonuclease activity.* Here's a quick refresher.

In the rare event that DNA polymerase delta or epsilon attaches a non-complementary nucleotide to a nascent DNA chain, the enzyme "senses" this based on a change in its shape (a "conformation change") caused by the bulky geometry of the mispaired bases.

When it senses a mispair, it pauses. It then transfers the 3' end of the nascent strand to another site on the enzyme that holds its 3' to 5' exonuclease activity. The exonuclease removes 2-5 nucleotides, including the mismatched one. The 3' end is then returned to the polymerase active site where the few nucleotides removed are added back and the replisome continues on its way.

How effective is DNA polymerase proofreading? As I just mentioned, the baseline mismatch rate (that is, before proofreading) of a replicative polymerase is 1 error every 100,000-1 million nucleotides. Polymerase proofreading improves that 100- to 1000-fold. That equates to a mismatch, or error, rate after polymerase proofreading of 1 error every 10 million to 100 million nucleotides! But even this isn't good enough. Thus, cells have a backup error correction pathway called mismatch repair (MMR).

Mismatch repair (MMR)

Mismatch repair (MMR) corrects substitution and small insertion and deletion errors that DNA polymerase proofreading fails to fix. MMR occurs right within the replication fork, its initial detector protein being physically linked to the PCNA sliding clamp.

MMR and a few of the lesion-fixing pathways we'll discuss in the next post employ variations of a five step process referred to as "cut-and-patch" repair to fix errors and small- to medium-sized lesions. In cut-and-patch repair, the following generic steps occur in order:

1) an error or lesion is identified and marked by a detector protein

2) an enzyme recognizes the mark and makes a single strand nick close to the error or lesion

3) another enzyme starts at the nick and chews back the DNA, removing the error or lesion

4) a DNA polymerase uses the "good strand" as the template to fill-in the nucleotides that were chewed back

5) a DNA ligase seals the gap, creating a continuous strand

Now let's look at MMR specifically to make these steps more tangible. Step one of our generic cut-and-patch approach is identifying and marking the error or legion. In MMR three proteins combine in a modular manner to detect errors. MMR's core structural protein is MSH2. MSH2 pairs with either of two other proteins--MSH6 or MSH3--which fine-tune its error detection function in two different directions.

When MSH2 pairs with MSH6 the heterodimer is called MutS-alpha. It identifies single base mismatches and small loop structures caused by insertions and deletions of 1-2 nucleotides. These represent about 85-90% of all errors. But when MSH2 pairs with MSH3 it is called MutS-beta and it detects larger insertions and deletions (2-13 nucleotides). These represent about 10-15% of all DNA polymerase errors.

We'll focus here on MutS-alpha. It handles the vast majority of mismatch errors and does so at the replication fork tethered to the back face of PCNA via a PIP box on either MSH6 or MSH3. There, the MSH6 or MSH3 detection subunits constantly scan the DNA as it's synthesized, looking for DNA polymerase errors. Any mismatch or small loop distorts the DNA, creating a small kink. The kink, in turn, creates a small gap between the strands of the double helix. Once the MutS protein complex detects the kink, it's MSH6 or MSH3 subunit inserts a specific amino acid into the gap to stabilize the kink so that it clearly flags the error.

When this amino acid is inserted into the double helix gap, it also causes an ATP-driven conformation changes in the core MSH2 (structural) subunit in MutS-alpha that closes it around the DNA, turning it into a sliding clamp (unrelated to PCNA) right at the mismatch or loop. This marks the mismatch and provides a docking station for the next actor in the pathway, MutL-alpha.

Mismatch Repair (MMR) pathway. MutS-alpha and -beta identify and mark the error and then recruit MutL-alpha. MutL-alpha makes an incision in the error strand using its endonuclease activity and then recruits other proteins needed to finish the job. These include EXO1 exonuclease to remove the mismatched nucleotide(s), RPAs to protect single-stranded DNA, DNA polymerase delta or epsilon to fill in the nucleotides removed by EXO1, and DNA ligase I to connect the strands.

Like MutS-alpha, MutL-alpha is a heterodimer that performs two roles. First, it makes a single-stranded incision in the DNA strand with the mismatch. This is the second step of the "cut-and-patch" repair model. It then recruits the next set of required proteins including EXO1 exonuclease, RPAs, DNA polymerase delta and DNA ligase I. Thus, MutL-alpha performs step 2 (it makes the incision) then it orchestrates the transition to steps 3 and 4 (excision and fill-in).

MutL-alpha makes an incision in the "bad" strand somewhere 5' of the mismatch. Let's pause. MutL-alpha's job is to cut the strand containing the error. But how does MutL-alpha "known" which of the two strands contains the error? One strand contains the correct nucleotide and the other the incorrect one. Which is which?

MutL-alpha uses two kinds of clues to determine which strand to cut. On the lagging strand, any gaps between Okazaki fragments that haven't yet been filled by DNA ligase identify the new strand, which is the strand with the error. Cut the strand with the gaps!

Because the leading strand contains no nicks in its newly synthesized strand, MutL-alpha needs a different clue. It uses the orientation of the PCNA sliding clamp. The PCNA disk has a front and a back face. Since DNA polymerase only synthesizes 5' to 3', the strand exiting the PCNA's back face with its 5' end is the nascent strand--the strand to nick!

The incision made, we're now at the third step of "cut-and-patch" repair: chewing back the DNA containing the error. In MMR, EXO1 exonuclease (a 5' to 3' exonuclease) chews the strand with the error starting at the incision.

As EXO1 chews back the "bad" strand, how will it "know" when it has removed the error? Given that the MutS-alpha sliding clamp is flagging the error, by the time EXO1 physically runs into it, the error will have been removed! The MutS sliding clamp serves as a molecular stop signal, or roadblock, for EXO1.

Of course, as EXO1 is chewing back the strand containing the error, the complementary strand is being exposed to destructive nucleases. By now we know what that means. RPAs (single-stranded binding proteins) will be needed. They arrive and coat the complementary single strand--the strand without the error--to protect it.

DNA polymerase delta arrives next to perform step four of our MMR "cut-and-stitch" repair. Starting at the nick, and with the mismatch "bad" strand now absent due to the action of EXO1, DNA polymerase delta re-synthesizes a new strand. It continues until it reaches the 5' end of the strand right in front of it. At that point, it stops. DNA polymerases can't connect two abutting strands. That job goes to DNA ligase 1, which performs the fifth and last step in "cut-and-stitch" repair: it ligates the two strands.

The error--be it a substitution, an insertion, or a deletion--has now been corrected. It takes human cells between several minutes and 30 minutes to complete the MMR pathway and repair an error. It does this simultaneous with replication fork movement. The first seconds are critical, however, because only then are there clues--in the form of lagging strand Okazaki fragment gaps and leading strand PCNA orientation--as to which strand to fix.

Next we'll discuss a type of DNA polymerase error that's actually much more common than the one's we've been discussing thus far: ribonucleotide insertion.

• We refer to the different functions that an enzyme can perform as that enzyme's "activities." So DNA polymerase epsilon, for example, has two activities: a DNA polymerization activity and a 3' to 5' exonuclease activity.