Lesions IIa

In the last post, I introduced the cell's two main DNA lesion repair pathways--base excision repair (BER) and nucleotide excision repair (NER). Both operate beyond the replication fork during all of the phases of the cell cycle.

But what happens if a lesion escapes BER and NER and finds itself in front of a moving replication fork in S phase? Most lesions that arise in the cell are able to get past the CMG helicase but not the DNA polymerase. Some very large lesions can't get past the helicase. Our focus here, though, will be the majority of small- to medium-sized lesions that stall at the polymerase. We'll take a look at the cell's response to larger legions in a future post.

Replication Fork Stalling

Picture a replication fork humming along a double helix copying DNA at 50 nucleotides per second. Suddenly, there's a lesion in its path on the leading strand that it can't synthesize through because it doesn't fit in the polymerase's active site. Assume it's relatively modest: a chemical alteration to a base or a bond that's formed between neighboring bases.

When DNA polymerase epsilon does stall at a lesion like this on the leading template strand, the helicase keeps moving forward, unwinding DNA. This generates a stretch of single-stranded DNA (ssDNA) hundreds to thousands of nucleotides long between the advancing helicase and the stalled DNA polymerase epsilon. This stretch of ssDNA is extremely informative to the cell. But let's first pause to say something about lagging strand lesions.

Lagging strand lesions cause vastly fewer problems than leading strand lesions because the lagging-strand machinery naturally reprimes downstream. So, even if DNA polymerase delta stalls at a lesion on the lagging template strand, synthesis continues via repeated priming and synthesisis of Okazaki fragments past the lesion. Lagging strand lesions typically does cause a 20-200 nucleotide gap to be formed on the strand being synthesized. And the template strand lesion remains. But that's not a problem: the cell can fix both later. Lagging strand lesions generally don't slow down genome replication like leading strand lesions.

Communicating the Stall

Let's get back to our leading strand lesion. As soon as the ssDNA appears between the stalled DNA poylymerase epsilon and the CMG helicase--again, hundreds to thousands of nucleotides long--Replication Protein A (RPAs) arrive immediately to coat and protect the exposed single strand. Recall that the main job of RPAs is to protect ssDNA against nucleases that would otherwise destroy it, and secondary structures that could cause even more replication problems.

But the RPA coat on the ssDNA (RPA-ssDNA) serves another purpose. It is the main signal to the cell that there is a DNA replication problem. Specifically, it communicates that a lesion on a leading template strand has stalled at a DNA polymerase and must be dealt with. This distress signal--RPA-ssDNA--is powerfully agnostic with respect to type of lesion. Any polymerase-stalling lesion will produce RPA-ssDNA. But that begs the question: How is the signal--RPA-ssDNA--detected?

The detector molecule is one of the two subunits of a protein complex called ATR-ATRIP. In human cells, ATR-ATRIP is always present in the nucleus, searching for RPA-ssDNA. When it finds some (typically in S-phase, since that's when replication occurs), ATRIP subunits attach in clusters to RPA proteins along the ssDNA. Thus, the ATRIP subunit is the detector. But when ATRIPs attach to RPAs, they place very important--but not yet activated--ATR kinases near the lesion.

Now the ATR kinase must be activated. To accomplish this, a ring-shaped protein complex called 9-1-1 loads onto the lagging strand near the lesion with the help of a ring loader. 9-1-1 recruits TOPBP1, which binds to the 9-1-1 ring and also has an ATR-activation domain, or AAD. When TOPBP1's AAD is attached to an ATR, the ATR undergoes a shape change that activates its kinase. Each time the AAD attaches, it remains attached for milliseconds to seconds before becoming detatched and activating another ATR located nearby. In this way, a single TOPBP1 activates in succession, but transiently, many ATR molecules.

A signalling cascade

Now we've activated ATR near the lesion. There, the most important protein it will phosphorylate and thus activate will be another kinase called CHK1. Unlike ATR, which is anchored to the replication fork via its ATRIP subunit (and also dependent on local TOPBP1 proteins to function optimally), activated CHK1 kinases are diffusible. They can roam freely in the nucleus, themselves phosphorylating dozens of other proteins needed in the response to the RPA-ssDNA distress signal. These other activated proteins will then go on to do whatever it is they are charged with doing in the nucleus, including activating other proteins or turning on specific genes.

The scenario I just described illustrates a general strategy cells use to amplify messages called a signaling cascade. It's a bit like a pyramid scheme, but with no bad outcome! In this signaling cascade, ATR-ATRIP is the founder sitting atop the pyramid. Its ATR subunit phosphorylates and thus activates first-tier recruits: multiple CHK1 kinases. These now-phosphorylated CHK1s then go on to phosphorylate next level proteins to promote, inhibit or alter their activities. Some of these, in turn, will go on to activate other proteins. In this way, information gleaned by one protein is amplified by another protein and then further amplified by still other proteins below it in the pyramid.

Local and global responses

CHK1 phosphorylations of various proteins initiate an immediate local response to the lesion and also contribute to a longer-term cell-wide global response. We'll focus on just the local response in this post.

The local response has two phases. The first, referred to as fork protection mode, physically stabilizes the replisome, buying it more time to bypass or fix the lesion. The fork protection phase lasts from minutes to tens of minutes. If the lesion remains, fork reversal mode will have time to unfold. In this mode, the cell literally reverses course, rewinding rather than unwinding the replication fork. This re-positions the lesion back in the context of double-stranded DNA where it will be easier to fix. Fork reversal (or "fork regression") mode takes from tens of minutes to hours.

Fork Protection Mode

The first phase of the local response is fork protection mode. It stabilizes and protects two replisome targets, creating a fork that can't move forward, but that won't fall apart. The first target is the physical structure of the replisome embodied by its core proteins, including the CMG helicase, the PCNA sliding clamp, the DNA polymerases, and other proteins like TIMELESS and AND-1. The second target is the ssDNA located between the polymerase and the helicase. This stretch of ssDNA will already be somewhat protected by RPA proteins. But with a stalled replication fork the ssDNA requires more protection.

Protecting the replisome structure

The local response involves CHK1 phosphorylation(s) of several replisome proteins we've already encountered: the helicase, AND-1, TIMELESS, and CLASPIN. In addition to protecting the replisome's structure, the CHK1 phosphorylations of TIMELESS and CLASPIN also create a positive feedback loop that generates even more phosphorylated CHK1, which speeds up fork protection mode. Let's first look at several phosphorylations that stabilize replisome proteins and the replisome's structure.

Perhaps most importantly, CHK1 phosphorylates the TIMELESS subunit of the TIMELESS-TIPIN complex. During normal leading strand replication, TIMELESS-TIPIN is a component of the fork protection complex (FPC). It travels with the replication fork, coupling CMG helicase unwinding with DNA polymerase epsilon synthesis.

During replication stress, TIMELESS is phosphorylated by CHK1 at two amino acids. These alter its shape with three effects, two of which pertain to the CMG helicase.

• they make TIMELESS bind more strongly to the helicase, stabilizing the helicase in place and preventing its premature disassembly.

• the TIMELESS phosphorylations make the helicase resistant to monoubiquitination at a specific amino acid, which would lead to its unloading off the DNA.

• they keep CLASPIN stably localized at the stalled fork. Without this anchoring, CLASPIN would dissociate frequently.

In fork protection mode, CHK1 also phosphorylates two of the MCM2-7 components of the CMG helicase itself. These slow down the helicase, preventing more polymerase-helicase uncoupling. These phosphorylations of MCM proteins plus those on TIMELESS keep the CMG helicase in position and intact.

Finally, CHK1 phosphorylates AND-1, the protein that tethers DNA polymerase alpha-primase. This stabilizes the AND-1 protein itself and it also helps maintain DNA polymerase alpha-primase's association with AND-1.

All of these phosphorylations--of TIMELESS, the CMG helicase, and AND-1--combine to create a reinforced, stabilized replisome that remains intact until the lesion is dealt with.

A positive feedback loop

A positive feedback loop is a situation in which a process (in our case, phosphorylations by CHK1) increases or alters the activity of one or more components of that process (in our case, TIMELESS and CLASPIN), that, in turn, enhance the original process (phosphorylations by CHK1). This positive feedback loop boosts all of the stabilization effects I just mentioned related to TIMELESS, the CMG helicase, and AND-1.

Our positive feedback loop has two parts. Recall that a CLASPIN scaffold is required for ATR phosphorylation of CHK1 and that CHK1 phosphorylations of TIMELESS keep CLASPIN stabilized at the fork. Stabilizing CLASPIN at the fork ensures that more CHK1 molecules will encounter it and will be phosphoylated by ATR. So, in this first part of the positive feedback loop, CHK1 phophorylation of another protein (TIMELESS) has, among other effects, that of increasing the number of activated CHK1s!

The second part of the positive feedback loop involves CHK1 phosphorylation of CLASPIN. A phosphorylated CLASPIN undergoes a shape change that makes it much more adept at facilitating ATR phosphorylations of CHK1 than an unphosphorylated CLASPIN. So by phosphorylating CLASPIN, CHK1 makes it a "super-CLASPIN"!

To summarize, then, the two aspects of the positive feedback loop are: (1) a more present CLASPIN scaffold (due to CHK1's action on TIMELESS) and, (2) a more effective CLASPIN scaffold (due to CHK1's action directly on CLASPIN).

Protecting ssDNA

To better protect the leading strand ssDNA between the DNA polymerase and the helicase, ATR--and to a lesser extent its primary target CHK1--first phosphorylate a protein called FANCD2, a component of the FANCD2–FANCI complex, at two specific amino acids. These change the complex's shape such that it clamps down on the DNA near the lesion. The shape change also permits a chemical addition called "monoubiquitination" of FANCD2. That is, it allows one ubiquitin molecule (chemical structure unimportant) to be attached to FANCD2 at a specific amino acid. This form is called “FANCD2-Ub”.

FANCD2-Ub recruits—and creates a chromatin environment that favors the recruitment of—multiple chromatin remodelers and histone-modifying enzymes. These open chromatin just ahead of and just behind the stalled replication fork, allowing BRCA1 and PALB2 to assemble adjacent to the ssDNA. PALB2 then brings BRCA2 to the ssDNA–chromatin interface, where BRCA2 initiates RAD51 loading onto the RPA-coated ssDNA. BRCA2 enables RAD51 to displace RPA and form a stable, highly protective helical filament around the ssDNA between the polymerase and the helicase.

Also in support of better ssDNA protection, CHK1 phosphorylates RAD51 itself at a specific amino acid. This enhances RAD51’s ability to displace RPAs on the ssDNA and to assemble and form a RAD51 filament that better coats and protects that ssDNA. The CHK1 phosphorylations of FANCD2 (to initiate BRCA1, PALB2 and BRCA2 recruitment) and of RAD51 (to enhance ssDNA coating) are but two of the dozens that CHK1 performs as part of the replication stress response.

Let’s pause to say a few words about the RPA to RAD51 exchange, because I've been saying that RPAs protect ssDNA (and they do, to some extent!). An RPA molecule is a large heterotrimer that covers roughly 30 nucleotides of ssDNA. In contrast, a RAD51 monomer covers only about 3 nucleotides when incorporated into a filament. Unlike RPA, though, which more haphazardly attach to ssDNA, RAD51 monomers polymerize into a dense and continuous helical filament along the ssDNA. Thus, RAD51 is much smaller but achieves a denser and thus more protective coating of the ssDNA.

Fork reversal mode

Once the cell detects a lesion, if it isn't addressed (via pathways we'll discuss in the next post) in the first 20 or so minutes, then fork reversal mode begins. It typically starts 30-60 minutes after initial RPA-ssDNA detection by ATR-ATRIP, but it sometimes takes hours to complete.

In fork reversal, the cell's machinery is going to rearrange the replication fork into a different shape that allows the lesion or damaged site to be repaired more easily. Specifically, the cell is going to place the lesion back in the context of dsDNA where it will be easier to fix than it will be in the crowded environment of the replication fork.

A good analogy for fork reversal is the zipping (not the unzipping) of a zipper. Normally, in genome replication, the CMG polymerase rushes down dsDNA, separating it into single strands. This is like unzipping a zipper: when we unzip a zipper, the handled zipper itself is like the CMG helicase, separating one entity (the double helix) into two (two single strands).

In fork reversal, the single strands that were formerly separated come back together in a specific way (that doesn't really follow our zipper analogy). What was formerly a three-way junction at the replication fork (the dsDNA plus the two separated template strands being used for synthesis) will become a four-way junction that's often called a "chicken foot structure."

In essence, the two strands being synthesized--the one being synthesized off the leading strand template and the other one off the lagging strand template--will anneal to each other as the replication fork is pushed backward. Take a look at figure __.

To do this, the cell uses specialized motor proteins (proteins that can move along DNA and/or move DNA). I won't go into detail regarding all the proteins involved in this molecular gymnastics (like SMARCAL1, ZRANB3 and HLTF) in part because their roles are only partly understood. Fork reversal is an area of active research. But I'll clearly describe their effects.

In fork reversal,

Fork reversal is difficult to picture... without a picture!

The cell's response to RPA-ssDNA has a local component focused directly on addressing the lesion and a global component that communicates to the cell that there might be a larger, cell-wide problem with replication. This communication is used at the intra-S-phase cell cycle checkpoint to decide whether to continue genome replication or pause it. The local and global responses do overlap, but it's helpful to think in those terms.

At this point, the cell knows that a lesions is present (based on the presence of RPA-ssDNA) but it doesn't know the kind or severity of the lesion and therefore how involved the response will have to be. Given that, we'll see in the next post that the cell wisely takes a staged approach to coping with lesions. It starts with easier responses but then escalates its response in terms of energy consumption and risk to itself if initial attempts fail. But I'm getting ahead of myself. Back to our RPA-ssDNA distress signal.

In addition to the local response, which targets the lesion itself, the global response shuts down DNA replication cell-wide, giving the cell even more time to address the problem that's affecting replication globally. The problem might be that there are too many serious lesions, maybe due to UV exposure. Or it might be a lack of nucleotides--the raw materials for replication.

Regardless, the global response only occurs if phosphorylated CHK1 activity reaches a threshold level--a level only achievable if there are many sites of replication stress. One lesion won't generate enough phosphorylated CHK1 to initiate a global response. This is a good thing, because a cell-wide shutdown of replication in S-phase is a drastic measure.