Global Profiling of DNA Replication Timing and Efficiency Reveals that Efficient Replication/Firing Occurs Late during S-Phase in S. pombe
Majid Eshaghi, R. Krishna M. Karuturi, Juntao Li, Zhaoqing Chu, Edison T. Liu, Jianhua Liu
PLoS ONE 2(8): e722.
1 comment:
This paper by Eshaghi et al. provides interesting, important new data on the rates of replication of each of the ORFs in the S. pombe genome when wild-type and cds1-delta cells are released from a 3-hour treatment with 8 mM hydroxyurea (HU). Previous microarray analyses (Feng et al., 2006; Heichinger et al., 2006; Hayashi et al., 2007) examined the extent of replication at those origins capable of firing during HU treatment in wild-type or checkpoint-mutant cells. This new paper, by Eshaghi et al., is the first to examine events when wild-type and checkpoint-mutant cells resume more rapid DNA replication after release from HU treatment.
Unfortunately, in my view, the value of the primary data obtained by Eshaghi et al. is diminished by some serious problems with their interpretation of the data.
The first problem is the authors' assumption that insignificant DNA synthesis occurred during the 3-hour HU treatment. As recently emphasized by the Brewer/Raghuraman laboratory (Alvino GM, Collingwood D, Murphy JM, Delrow J, Brewer BJ, Raghuraman MK. Replication in Hydroxyurea: It's a matter of time. Mol Cell Biol. 2007 Jul 16; [Epub ahead of print]), HU does not block DNA synthesis in wild-type cells. It simply slows the rate of synthesis, but the overall timing program remains intact. Those origins that are going to fire at a specific time in the program (relative to other origins) do so. If the HU treatment is maintained sufficiently long, late origins will fire.
The extent of slowing of the replication program is dependent on the concentration of HU. Eshaghi et al. employed 8 mM HU, which is insufficient to prevent significant DNA synthesis at early origins during a 3-hour treatment. If one examines Figure 9 in the EMBO J paper of Kim & Huberman (2001), one can see that a somewhat higher concentration of HU (12 mM) was insufficient to prevent run-off of most replication forks from the early-firing origins in the "K-repeats" of centromeric DNA. Like most replication origins in S. pombe, the K-repeats are replicated by a combination of active origin firing and passive replication due to forks from neighboring origins. The 2D gel results in Fig. 9 of Kim & Huberman (2001) show that both active origin firing (indicated by bubble arcs) and passive replication (indicated by Y arcs) of the K-repeats are abundant after 1.5 hours of 12 mM HU treatment and are greatly reduced by 3 hours of HU treatment. That means replication was complete at most K-repeats after 3 hours of HU treatment. Not much additional synthesis could possibly take place at the K-repeats after removal of HU.
Eshaghi et al. used the DNA from their "0-minute" time point (the time at which HU was removed from cells, after 3-hour incubation with HU) as a reference with which to compare their other DNA samples (from subsequent time points). As I've already indicated, in the 0-minute time point DNA replication would have been nearly complete in the vicinity of the earliest-replicated regions, and it would have been partially complete in other early-replicated regions (for example, in the vicinity of ars2-1; see Fig. 9 of Kim & Huberman [2001]). Therefore, these early-replicated regions could not have achieved a 2-fold increase in copy number during subsequent time points. In fact, in the case of wild-type cells, if the authors had plotted copy number increase as a function of distance along the chromosome at the 60-minute time point (when replication was complete), the resulting graph should have displayed valleys centered on early-firing origins. The graph should have been a perfect inverse of the graph the authors would have obtained if they had plotted the results obtained when they measured copy numbers along the chromosomes in 3-hour-HU-treated cells compared to G1 or G2 cells. However, Eshaghi et al. do not mention this anticipated phenomenon of reduced copy number at early origins. Instead, they treat the data points at each ORF along the chromosome as if the ratio of signals at the 0-minute and 60-minute time points represents a full 2-fold increase in DNA content. I haven't yet thought through what effects this distortion might have on the final results and interpretations, but I suspect that this distortion would alter the apparent relative contributions to post-HU chromosome replication of early- and late-replicating regions.
The second problem with the authors' interpretations is their unusual definition of the term "efficiency". In the past the term "efficiency", when used in connection with origin firing, has meant the frequency with which the origin fires during S phase. Thus an origin with 30% efficiency (typical for S. pombe) would fire in 30% of S phases and not fire in 70% of S phases. When an origin doesn't fire, it is passively replicated by forks from neighboring origins. So all origins (and all other chromosomal segments) _replicate_ with 100% efficiency, but origins _fire_ with efficiencies that, in S. pombe, always seem to be less than 100%. In contrast to this standard definition of "efficiency", Eshaghi et al. use the term to denote the _rate_ at which chromosomal segments, including origins, are replicated during S phase. By this definition, an origin that replicated during most of S phase would have a lower efficiency than an origin that replicated only during the third quarter of S phase. In both cases, though, each origin (and all other chromosomal segments) would be 100% replicated by the end of S phase.
Eshaghi et al. observed that, in wild-type cells, the regions replicating late had higher "efficiencies" (replicated faster) than regions replicating early. This is consistent with the prediction ([9], [29]) that the replication process would have to be faster in late S phase in order to ensure timely replication of all chromosome regions. This may well be true. But before this conclusion can be considered solid, one would need to test the possibility that replication kinetics might have been altered by chilling the cells to 4 degrees when removing the HU. Indeed, slower replication during early post-HU S phase and faster replication later would be the predicted consequence of chilling the cells during HU removal.
It is important to note that the authors' definition of "efficiency" has nothing to do with the classic efficiency of origin firing. The authors' definition is concerned with overall replication of an origin, whether actively or passively, and the time scale over which that replication takes place. One can easily imagine that an origin with high firing efficiency (by the classic definition) might fire asynchronously in S phase and thus score as inefficient by the definition of Eshaghi et al.
The danger of employing this non-classic definition is that readers will become confused. For example, when I first read the abstract of this paper, I thought the authors were referring to late origins that fired efficiently instead of to late origins that replicated rapidly. Confusion is even more likely when the authors use both definitions in the same sentence, as in this example: "To ensure completion of DNA duplication by the end of S-phase in an inefficient replication system such as S. pombe [9], [29], it is necessary to increase the efficiency of origins that fire late in S-phase." In this sentence, the first reference to efficiency (the word "inefficient") refers to the classic definition (origin firing frequency), while the second reference refers to the speed of replication. Note that it is not necessary that individual origins fire more efficiently in late S phase--just that the overall process of replication be faster. This can be accomplished by greater origin firing efficiency, greater synchrony of origin firing, and/or by more rapid replication fork movement. This microarray study does not discriminate between these possibilities.
The final problem that I would like to comment on is the use of a ±12-kb window to determine whether origins identified in one study are located at the same positions as origins identified in a different study. A ±12-kb window is, in fact, a 24-kb window. If a study should identify 500 origins, and if these origins should be evenly distributed throughout the genome, and if each origin should be considered "coincident" with anything ±12 kb from it, then the portion of the genome that would be considered coincident with one of the origins would be 12 Mb (24 kb X 500). Since the S. pombe genome is only 14 Mb, it's not surprising that there appears to be a high degree of overlap between the origins identified in this study and those identified in other studies.
I hope that the authors and other investigators will respond to these comments.
Post a Comment