1. Biochemistry
  2. Biophysics and Structural Biology
Download icon

Single-molecule studies contrast ordered DNA replication with stochastic translesion synthesis

  1. Gengjing Zhao
  2. Emma S Gleave
  3. Meindert Hugo Lamers Is a corresponding author
  1. MRC laboratory of Molecular Biology, United Kingdom
Research Article
Cited
0
Views
634
Comments
0
Cite as: eLife 2017;6:e32177 doi: 10.7554/eLife.32177

Abstract

High fidelity replicative DNA polymerases are unable to synthesize past DNA adducts that result from diverse chemicals, reactive oxygen species or UV light. To bypass these replication blocks, cells utilize specialized translesion DNA polymerases that are intrinsically error prone and associated with mutagenesis, drug resistance, and cancer. How untimely access of translesion polymerases to DNA is prevented is poorly understood. Here we use co-localization single-molecule spectroscopy (CoSMoS) to follow the exchange of the E. coli replicative DNA polymerase Pol IIIcore with the translesion polymerases Pol II and Pol IV. We find that in contrast to the toolbelt model, the replicative and translesion polymerases do not form a stable complex on one clamp but alternate their binding. Furthermore, while the loading of clamp and Pol IIIcore is highly organized, the exchange with the translesion polymerases is stochastic and is not determined by lesion-recognition but instead a concentration-dependent competition between the polymerases.

https://doi.org/10.7554/eLife.32177.001

Introduction

To ensure faithful replication of the genomic DNA, replicative DNA polymerases have a narrow active site that limits the incorporation of incorrect nucleotides. In addition, rare nucleotide mis-incorporations into the primer strand prevent further DNA synthesis and a 3'−5' exonuclease is required to remove the misincorporated nucleotides (Kunkel, 2004). In contrast, when the polymerase encounters a lesion on the template strand in the form of a modified base caused by diverse chemicals, reactive oxygen species, or UV light (Liu et al., 2016; Lindahl, 1996), the high-fidelity replicative DNA polymerases are stalled. To bypass these replication blocks, all cells harbor multiple specialized translesion DNA polymerases (Goodman and Woodgate, 2013) that have more open active sites and are therefore able to accommodate bulky DNA adducts and continue DNA synthesis. As a result of their more open active sites, the translesion polymerases are error prone and consequently associated with increased mutagenesis, drug resistance, and cancer (Fuchs and Fujii, 2013; Lange et al., 2011). Therefore, the access of the translesion polymerases to DNA needs to be tightly controlled, but how this is achieved has been the subject of debate.

The 'toolbelt' model (Indiani et al., 2005) predicts that in E. coli the replicative DNA polymerase Pol IIIα and the translesion DNA polymerase Pol IV bind simultaneously to the DNA sliding clamp β, a dimeric, ring-shaped protein that encircles the DNA and provides processivity to the replicative DNA polymerases (Johnson and O'Donnell, 2005). This way, the translesion polymerase functions in a manner analogous to the proofreading exonuclease: when the replicative polymerase inserts an incorrect nucleotide into the nascent strand, it will be removed by the proofreading exonuclease, whereas when the polymerase encounters a lesion on the template strand, the DNA is transferred to the translesion polymerase that can bypass the lesion. Thus both the exonuclease and translesion polymerase act as 'tools' that enable the replicative polymerase to overcome potential roadblocks to DNA replication. The toolbelt model, which was originally based on steady-state Förster Energy Resonance Transfer (FRET) experiments that showed the simultaneous binding of the replicative and translesion polymerases to the β-clamp, has found support in several subsequent studies (Indiani et al., 2009; Furukohri et al., 2008; Heltzel et al., 2012). However, all these studies used bulk studies that due to the asynchronous nature cannot separate out the sequential steps during a reaction (Robinson and van Oijen, 2013). More recently, single molecule approaches have also been used (Kath et al., 2014, 2016), but in these experiments the exchange of polymerases was inferred indirectly through the change in speed of DNA synthesis and therefore it cannot be determined whether the polymerases bind simultaneously. In addition, recent studies reveal that during DNA replication in E. coli the two binding pockets of the dimeric β-clamp are occupied by the replicative DNA polymerase Pol IIIα and the associated proofreading exonuclease ε (Toste Rêgo et al., 2013; Jergic et al., 2013). The cryo-EM structure of the trimeric Pol III-exonuclease-clamp (α, ε, β) complex (Fernandez-Leiro et al., 2015) also shows that most of the clamp is covered, leaving no space for a second polymerase. Therefore, it remains controversial whether the replicative and translesion polymerases can co-localize on a single clamp.

Consequently, an alternative view to the toolbelt model is that the translesion DNA polymerases compete for binding to clamp-DNA through 'mass action', as evidenced by the fact that the bypass of a N2-acetylaminofluorene guanine adduct by Pol V or Pol II depends on the relative concentrations of the two polymerases (Becherel and Fuchs, 2001; Fujii and Fuchs, 2004), and that the concentrations of Pol IV and Pol V are dramatically increased during the bacterial SOS DNA damage response (Sutton, 2010).

Regardless of the model, the DNA sliding clamp plays a pivotal role in controlling access of the translesion polymerases to the DNA. However, the control for access to the clamp-DNA is complicated by the fact that on the lagging strand, DNA synthesis is discontinuous and every ~1000 base pairs a new clamp is loaded, followed by the binding of the replicative DNA polymerase. Due to its closed circular shape, the clamp must be loaded and unloaded onto the DNA by the dedicated clamp loader complex (γ/τ-complex in bacteria, RFC in archaea and eukaryotes) (Hedglin et al., 2013; Kelch et al., 2012). Once the clamp is loaded onto primed DNA, the replicative polymerase associates and initiates DNA synthesis. How the repeated loading and unloading of the clamp and replicative polymerase Pol IIIα on the lagging strand is coordinated, while simultaneously preventing the untimely association of the translesion polymerases has not been studied so far.

Here, we use co-localization single molecule spectroscopy (CoSMoS) (Friedman et al., 2006) to directly visualize the loading of the E. coli clamp loader (γ/τ-complex), the DNA sliding clamp β, the replicative DNA polymerase Pol IIIα, the proofreading exonuclease ε, as well as the exchange with the translesion polymerases Pol II and Pol IV. The multi-color CoSMoS experiments enable us to follow the binding and dissociation of multiple proteins in real-time on a single DNA molecule, which makes it the most suitable method to discriminate between simultaneous or sequential binding of different molecules on a DNA substrate. Our work shows that the translesion polymerases Pol II and Pol IV do not form a stable complex with the replicative polymerase Pol IIIα on the clamp-DNA and therefore the clamp does not function as a molecular toolbelt. Furthermore, we find that the sequential activities of the replication proteins clamp loader, clamp, and Pol IIIα are highly organized while the exchange with the translesion polymerases Pol II and Pol IV is disordered and determined by mass action through concentration-dependent competition for the hydrophobic groove on the surface of the β-clamp. Hence, our results provide a unique insight into the temporal organization of the events in DNA replication and translesion synthesis, and contrast the highly organized replication events with stochastic polymerase exchange during translesion synthesis.

Results

Preparation of DNA substrates and fluorescently labeled proteins

The ring-shaped E. coli β-clamp is capable of threading and unthreading itself on free DNA ends and therefore we attached a primer-template DNA substrate to a glass surface and blocked its free end with monovalent streptavidin (Figure 1A). Subsequently, the binding of fluorescently labeled proteins to DNA was followed by two- or three-color total internal reflection fluorescence microscopy (see Materials and methods) with a frame rate of 0.44 and 0.66 s (s), respectively, on ~800 well separated DNA molecules per field of view (Figure 1B–C). Proteins were fluorescently labeled via maleimide-cysteine crosslinking or enzymatically via an N-terminal Ybbr tag (Yin et al., 2006), and the fluorescently labeled proteins retained wild-type activity as indicated by polymerase processivity assay (Figure 1—figure supplement 1A–C). For detection of Pol IIIcore (α, ε, θ), we fluorescently labeled the α subunit. For the clamp loader complexes (γ3δδ' and τ3δδ'), the fluorescent label was placed on the δ' subunit. The lifetime of the each of the fluorophores was measured individually on DNA-bound clamps (Atto 488 274.4 ± 16.5 s, Atto 565: 145.7 ± 7.5 s, Atto 647N: 93.0 ± 7.4 s) (Figure 1—figure supplement 1D–F). In most experiments, after initial detection of the DNA molecules, the fluorophore (Atto 488) on the DNA was bleached so that the same color could be re-used on one of the proteins. The bleaching of the DNA fluorophore has no effect on the lifetime of Pol IIIcore on clamp-DNA (τon without bleaching 18.1 ± 1.6, τon with bleaching 16.8 ± 1.8) (Figure 1—figure supplement 1G–H).

Figure 1 with 1 supplement see all
Experimental setup.

(A) Schematic representation of the experimental setup. DNA molecules are attached to a PEGylated glass slide via a biotin-streptavidin layer, and end-blocked with monovalent streptavidin. Fluorescently labeled proteins will be detected when bound to the DNA molecules (B) Schematic representation of the three fluorescent channels from a single image (out of an 1000 image series) showing the presence of different molecules. (C) Schematic representation of kymographs from single position in the image series, revealing the binding and release of different proteins to the DNA at that position.

https://doi.org/10.7554/eLife.32177.002

Clamp loading and unloading are distinct processes

The isolated β-clamp shows no interaction with the end-blocked DNA (Figure 2—figure supplement 1A–B) When combined with the γ clamp loader complex (γ3δ1δ'1), frequent clamp loading events are observed where the loader and clamp arrive at the DNA simultaneously or in two adjacent frames (Figure 2A–B), due to the sequential data acquisition of the three laser channels (further explained in Figure 2—figure supplement 1C–E). Therefore, the loader and clamp bind DNA as a pre-formed complex. Shortly after DNA binding, the loader dissociates while the clamp remains bound to the DNA for the remainder of the data acquisition (Figure 2A). As the two lifetimes of the loader and clamp are vastly different it was not possible to accurately measure both in one experiment. Therefore, for the clamp lifetime, we first loaded the clamp, washed away the loader and then started data collection using 10 s intervals between measurements to avoid bleaching of the fluorophore. The clamps remain bound for more than 23 min (τon = 1429.7 ± 177.0 s; Figure 2C), excluding the time it takes to load the clamps, wash away the loader, and start data collection (~3 min). In contrast, the lifetime of the loader is very short lived. To accurately measure the loader lifetime, only one fluorescent channel was used to decrease the frame rate to 0.086 s. The lifetime of the isolated loader on DNA is 1.20 ± 0.05 s, which is shortened to 0.41 ± 0.01 s in the presence of the clamp (Figure 2D). The rapid release of the loader from clamp-DNA is dependent on ATP hydrolysis as evidenced by the fact that in the presence of the poorly hydrolysable analog ATPγS, the loader and clamp still bind to the DNA together but also release together (Figure 2E). The loader and clamp bind to the DNA briefly (τon = 2.7 ± 0.2 s; Figure 2F), which contrasts with the long lifetimes for the loaded clamps on DNA (Figure 2C). Taken together, our analysis is in agreement with the model that the clamp holds the loader in a conformation that suppresses ATPase activity and subsequent DNA binding triggers ATP hydrolysis and the release of the loader from clamp and DNA (Hedglin et al., 2013; Kelch et al., 2012).

Figure 2 with 1 supplement see all
Different mechanisms for clamp loading and unloading.

(A) Representative trace showing a clamp loading event on DNA in the presence of ATP. (B) Histogram showing the simultaneous arrival of loader and clamp on DNA (see also Figure 2—figure supplement 1C–E). (C) The distribution of lifetimes for the clamp on DNA after removal of the clamp loader (D) Lifetime of the clamp loader on DNA in the absence (blue bars) and presence (grey bars) of the clamp. (E) Representative trace showing the simultaneous arrival and release of loader and clamp on DNA in the presence of ATPγS. (F) The distribution of lifetimes for the loader and clamp on DNA in the presence of ATPγS. (G) Representative trace showing clamp loading and unloading by the loader in the presence of ATP. (H) The distribution of lifetimes for the loader on DNA during unloading. (I) The distribution of lag times between the arrival of the loader and the release of the clamp. (J) Representative trace showing unloading of a pre-loaded clamp in absence of ATP. (K) The distribution of lifetimes for the loader during clamp unloading in the absence of ATP. The first column (in dark grey) has been excluded from the fitting. The lower numbers in this column are possibly caused by the clamp that needs to be removed first before the loader can release. (L) The distribution of lag times between the arrival of the loader and the release of the clamp in the absence of ATP. All values represent mean lifetime/lag time ±s.e.m.

https://doi.org/10.7554/eLife.32177.004

During DNA replication, the clamps on the lagging strand need to be unloaded and recycled to allow for continuous DNA synthesis (Yao et al., 1996). We observe a different temporal organization for unloading events compared to the loading events: the loader arrives at the loaded clamp, releases the clamp within 4.1 ± 0.4 s of its arrival but remains bound for a total time of 10.8 ± 1.2 s (Figure 2G–I). Furthermore, unlike clamp loading, the unloading of a clamp does not require ATP hydrolysis: the nucleotide requirement of the unloading process was tested by first loading clamps onto DNA and then washing away the ATP and free proteins. Next, the loader was re-introduced in the absence of ATP, resulting in many unloading events with similar kinetics to those observed in the presence of ATP (Figure 2J–L). About half (53%) of the binding events of the loader to the pre-loaded clamp do not result in unloading of the clamp. These events last shorter (τon = 2.5 ± 0.1 s) than the unloading events (τon = 10.8 ± 1.2 s) (Figure 2—figure supplement 1F–G). Taken together, the data indicates that the loading and unloading of a clamp are not forward and backward reactions of the same mechanism but that they are separate processes, each with a distinct organization, possibly to prevent unwanted clamp unloading at the replication fork.

Pol IIIcore has an intrinsic lifetime on DNA that is independent of its activity

Once the clamp is loaded onto DNA, the replicative DNA polymerase Pol IIIcore will associate with the clamp and initiate DNA replication (Johnson and O'Donnell, 2005). Pol IIIcore is a stable trimeric complex containing the polymerase subunit α, the exonuclease ε and the accessory subunit θ (McHenry and Crow, 1979). Using a single fluorescent channel, we find that Pol IIIcore alone binds DNA very briefly for one image frame (0.086 s) or less, resembling more a collisions-like interaction rather than true binding events (Figure 3A–B). These short-lived collisions are in agreement with previous studies that show that Pol IIIα is a poor enzyme in isolation (Fay et al., 1981) that has a low affinity for DNA (Fernandez-Leiro et al., 2015; McCauley et al., 2008). The collisions of Pol IIIcore contrasts with the well-studied E. coli DNA polymerase I Klenow fragment, which is known to have a high affinity for DNA (Kuchta et al., 1987) and consequently shows a lifetime of 42.2 ± 1.8 s (Figure 3C–D).

Pol IIIcore binds transiently to clamp-DNA.

(A) Representative trace showing Pol IIIcore collisions with DNA in the absence of clamp. (B) The distribution of lifetimes for Pol IIIcore on DNA. (C) Representative trace showing Pol I binding on DNA. (D) The distribution of lifetimes for Pol I on DNA. (E) Representative trace showing Pol IIIcore binding to clamp-DNA shortly after the release of loader. (F) The distribution of lifetimes for Pol IIIcore on clamp-DNA in absence of dNTPs. (G) The distribution of lifetimes for Pol IIIcore binding events on clamp-DNA in the presence of dATP/dTTP. (H) Loading of clamp and polymerase by the τ clamp loader complex. (I) Lifetime of the τ clamp loader complex and Pol IIIcore on clamp-DNA (J) Cartoon of the binding sequence of the τ clamp loader - Pol IIIcore complex on clamp-DNA. All values represent mean lifetime ±s.e.m.

https://doi.org/10.7554/eLife.32177.006

The behavior of Pol IIIcore is dramatically altered in the presence of the clamp. Shortly after the clamp is loaded onto DNA, Pol IIIcore associates with the clamp-DNA, producing long lasting binding events (Figure 3E–G) with a lifetime of 15.7 ± 1.1 s (Figure 3E–F). During DNA replication, the E. coli replisome synthesizes DNA with up to 100,000 base pairs per binding event (Yao et al., 2009; Tanner et al., 2009), which contrasts with the relatively short binding times of Pol IIIcore we measure on clamp-DNA. Therefore, we tested the effect of nucleotides on the lifetime of Pol IIIcore on DNA, using only two of the four nucleotides to prevent the Pol IIIcore from extending the primer strand and dissociating at the end the DNA substrate. The omission of two nucleotides is often used in DNA replication assays to induce 'polymerase idling' at a primer terminus as a result of opposing polymerase and exonuclease activities. To our surprise however, addition of dATP and dTTP (0.5 mM each), has no significant effect on the lifetime of Pol IIIcore on clamp-DNA (τon = 16.1 ± 1.0 s, Figure 3G). This is also in agreement with the observation that an actively synthesizing Pol IIIcore has a lifetime of ~10 s in the presence of all four nucleotides on a 48.5 kb λ DNA substrate (Jergic et al., 2013; Tanner et al., 2008). Together, this shows that Pol IIIcore has a similar lifetime on clamp-DNA regardless of whether it is stationary, idling or actively synthesizing DNA.

During DNA replication, Pol IIIcore is tethered to the rest of the replisome via the clamp loader protein τ (Studwell-Vaughan and O'Donnell, 1991). Thus far, to separate clamp loading from polymerase loading, we have used a clamp loader complex comprising the subunits γ, δ, an δ' (γ3δ1δ'1). γ is a shorter product of τ that is fully active in clamp loading, but does not interact with the polymerase (Dallmann et al., 1995, 2000). To measure the effect of full length τ on clamp loading and polymerase loading, we also measured clamp and polymerase loading with the τ clamp loader complex (τ3δ1δ'1), in which each of the three τ proteins had a Pol IIIcore complex bound (see Material and Methods). Here, we find that clamp, loader, and Pol IIIcore arrive at the DNA together (Figure 3H). Unlike the γ clamp loader complex, the τ clamp loader complex does not dissociate immediately, as it remains bound for ~15 s and now leaves together with Pol IIIcore (τon = 14.8 ± 0.9 s, Figure 3H–I). We believe this is a result of the τ clamp loader complex and Pol IIIcore trading places on the clamp-DNA (Figure 3J). The presence of the τ, however, does not affect the lifetime of Pol IIIcore on clamp-DNA, as the lifetime of the τ clamp loader - Pol IIIcore complex is similar to that of Pol IIIcore alone.

However, during the simultaneous synthesis of leading and lagging strand, the polymerases on either strand are linked together by the multiple τ proteins of the clamp loader complex, which prevents the polymerase from diffusing away and enables it to quickly resume DNA synthesis. This may explain the higher processivity measured for the replisome (Yao et al., 2009; Tanner et al., 2009) compared to the bursts of activity from a single Pol IIIcore (Jergic et al., 2013; Tanner et al., 2008).

Pol IIIcore and Pol IV alternate binding to clamp-DNA

During DNA replication the replisome may encounter DNA adducts caused by diverse chemicals, reactive oxygen species or UV-light that form barriers to the high-fidelity replicative DNA polymerases. To overcome these replication blocks, the replicative DNA polymerases temporarily trade places with the error-prone translesion DNA polymerases (Goodman and Woodgate, 2013). To study the molecular mechanism of polymerase switching, we directly visualized the binding of Pol IIIcore and the translesion DNA polymerase Pol IV on clamp-DNA. At equal concentration of Pol IIIcore and Pol IV (30 nM each), we record a large majority of events (92%) that show alternating binding of the two polymerases on clamp-DNA, with Pol IIIcore binding first in 70%, and Pol IV binding first in 22%, of events (Figure 4A and C, Table 1). Similar to the lifetime of Pol IIIcore alone on clamp-DNA, this competition is unaffected by addition of nucleotides (dATP/dTTP, 0.5 mM each) (Pol IIIcore to Pol IV switch: 72%, Pol IV to Pol IIIcore switch 21%). In the polymerase switching events, there is a significant lag time between the release of Pol IIIcore and Pol IV arrival that decreases with increased protein concentrations (Table 1, Figure 4—figure supplement 1A–D). The lifetime of Pol IIIcore on the clamp-DNA remains unchanged at all protein concentrations (Table 1, Figure 4—figure supplement 1E–H), suggesting that Pol IV binding does not cause the release of Pol IIIcore and therefore the two polymerases bind independently.

Figure 4 with 3 supplements see all
The replicative and translesion polymerases compete for binding to clamp-DNA.

(A) Representative trace showing alternating binding of Pol IIIcore and Pol IV on clamp-DNA. (B) Lifetime of Pol IV on clamp-DNA. (C) Cartoon showing the frequency of different polymerase switching events. (D) Representative trace showing the independent arrival and release of Pol IIIcore and Pol IV on clamp-DNA during co-localization events. (E) Lifetime of the co-localization of Pol IIIcore on Pol IV on clamp-DNA. (F) Representative trace showing the simultaneous arrival and release of Pol IIIcore α subunit (polymerase) and Pol IIIcore ε subunit (exonuclease) on clamp-DNA. (G) Lifetime of the co-localization of the Pol IIIcore α subunit and Pol IIIcore ε subunit. (H) Representative trace showing alternating binding of Pol IIIcore and Pol II on clamp-DNA. (I) Lifetime of Pol II on clamp-DNA. All values represent mean lifetime ±s.e.m.

https://doi.org/10.7554/eLife.32177.007
Table 1
Competition of Pol IIIcore and Pol IV
https://doi.org/10.7554/eLife.32177.011
Concentration (nM)Polymerase exchange (%)*Lag time (s)Lifetime (s)
CompetitionPol IIIcorePol IVIII→IVIV →IIIIII + IVIII→IVPol IIIcore
Pol IIIcore - Pol IV3068112732.5 ± 4.616.3 ± 1.0
30307022920.3 ± 3.515.7 ± 1.1
301506320155.9 ± 0.516.6 ± 1.7
1501505124263.5 ± 0.216.0 ± 0.9
τ-complex§ - Pol IV30#30950511.3 ± 1.314.8 ± 0.9
  1. *Polymerase exchange observed on clamp-DNA showing the exchange from Pol IIIcore to Pol IV (III→IV), Pol IV to Pol IIIcore (IV→III), or co-localization of Pol IIIcore and Pol IV (III + IV).

    Time between Pol IIIcore release and Pol IV arrival.

  2. Lifetime on clamp-DNA.

    §τ-complex consists of τ clamp loader (τ3δ1δ'1) and three Pol IIIcore complexes (α, ε, θ).

  3. #Concentration of Pol IIIcore.

In addition to the switching events, we also observe a small number of co-localization events of the two polymerases on clamp-DNA (9%), which become more frequent at higher protein concentrations (Figure 4C, Table 1) but are unaffected by addition of nucleotides (7% with dATP/dTTP). In all events where Pol IIIcore and Pol IV co-localize on the clamp-DNA, the polymerases arrive and leave independently of one another (Figure 4D). This is different from the true binding partner exonuclease ε that arrives and leaves with Pol IIIα (Figure 4F). In addition, the co-localization time of Pol IIIcore and Pol IV (τcolocalize = 8.2 ± 0.6, Figure 4E) is shorter than that of Pol IIIcore alone (τon = 15.8 ± 0.9 s, Figure 4G). This therefore shows that Pol IIIcore and Pol IV do not form a stable complex on clamp-DNA and that the clamp does not function as a molecular toolbelt, but that the two polymerases compete for the binding of the clamp-DNA in a concentration-dependent manner. This competition is strongly favored towards Pol IIIcore in the presence of the τ clamp loader complex (Table 1) as it is directly tethered to Pol IIIcore and ensures that Pol IIIcore is immediately bound upon clamp loading (Figure 3H). The presence of the τ clamp loader complex does not alter the frequency of co-localization between Pol IIIcore and Pol IV (Table 1). This suggests that during replication, Pol IV may be able to frequently access the clamp-DNA, especially as the estimated cellular concentration of Pol IV is ~10 fold higher than that of Pol IIIcore (Sutton, 2010).

Finally, a second E. coli translesion polymerase, Pol II (Paz-Elizur et al., 1996) also competes with Pol IIIcore for access to the clamp but shows no co-localization with Pol IIIcore (Figure 4H–I, Table 2), possibly because of its large size (90 kDa vs. 40 kDa for Pol IV) that may prevent it from simultaneous binding to the clamp with Pol IIIcore.

Table 2
Lifetime of β-clamp binding mutants of Pol IIIcore, Pol IV and Pol II
https://doi.org/10.7554/eLife.32177.012
Polymerase exchange (%) *
PolymeraseMutationLifetime (s)III→IVIV →IIIIII + IV
Pol IIIcoreWT15.7 ± 1.170229
ε (β-)7.9 ± 1.270263
ε (β+)40.2 ± 8.771246
Pol IVWT14.2 ± 1.870229
β groove2.7 ± 0.240600
β rim14.9 ± 1.766295
PolymeraseMutationLifetime (s)III→IIII →IIIIII + II
Pol IIWT10.4 ± 1.371290
β groove4.4 ± 0.863370
  1. *Polymerase exchange observed on clamp-DNA showing the exchange from Pol IIIcore to Pol IV or Pol II, Pol IV or Pol II to Pol IIIcore, or co-localization of Pol IIIcore and Pol IV or Pol II.

    Lifetime on clamp-DNA.

  2. The Pol IV β cleft mutant was measured at high concentrations (90nM) in an attempt to catch co-localization events.

Polymerases compete for binding to the hydrophobic groove of the clamp

To further investigate the competition between the different polymerases and their access to the clamp, we created a series of polymerase mutants. Most clamp interacting proteins, including Pol IIIcore, Pol IV, and Pol II, bind to a hydrophobic groove on the surface of the β-clamp using the canonical sequence Qxx(L/M)xF (Dalrymple et al., 2001). The β-clamp is dimeric and thus has two binding grooves. Interestingly, Pol IIIcore contains two β-binding sequences: one in the polymerase subunit α (QADMF, residues 920–924) that is absolutely required for clamp binding (Dohrmann and McHenry, 2005), and a second sequence in the exonuclease subunit ε (QTSMAF, residues 182–187) that stabilizes the Pol IIIcore complex and stimulates processive DNA synthesis and exonuclease activity (Toste Rêgo et al., 2013; Jergic et al., 2013). To demonstrate that the ε β-binding motif also contributes to the lifetime of Pol IIIcore on the clamp, we made two variants of the exonuclease ε, one with a weakened the β-binding motif (QTSMAF to QTSAAA [Toste Rêgo et al., 2013]) and one with an enhanced β-binding motif (QTSMAF to QTSLPL [Fernandez-Leiro et al., 2015]). Indeed, the lifetime of Pol IIIcore is decreased ~2 fold by the weak β-binding motif, while it is increased ~2 fold by the enhanced β-binding sequence (Table 2, Figure 4—figure supplement 2A–B), showing that Pol III core occupies both binding grooves of the dimeric clamp.

In Pol IV, two β-clamp interacting motifs have been described: a canonical QLVLGL motif (residues 346–351) that binds the hydrophobic groove, and a rim binding sequence (VWP, residues 301–304) that interacts with the side of the clamp (Becherel et al., 2002; Bunting et al., 2003; Heltzel et al., 2009). Mutation of the groove binding motif in Pol IV was reported to inhibit clamp-dependent DNA synthesis, while mutation of the rim contacts resulted in loss of polymerase switching (Heltzel et al., 2009). In our experiments, mutation of the groove binding sequence (QLVLGL to QLVAGA, residues 346–351) results in a drastic reduction in the lifetime of binding to the clamp (Table 2, Figure 4—figure supplement 2C), and consequently no co-localization between Pol IV and Pol IIIcore are observed, even at elevated concentrations of Pol IV (Table 2). In contrast mutation of the rim binding motif (VWP to AGA, residues 303–305) shows little effect on the lifetime of Pol IV on the clamp, or on the co-localization frequency with Pol IIIcore (Table 2, Figure 4—figure supplement 2D). Finally, also mutation of the groove binding motif in Pol II (QLGLF to QLGAA, residues 779–783) leads to a reduced lifetime of binding to the β-clamp (Table 2, Figure 4—figure supplement 2E).

This therefore shows that all three polymerases compete for the same binding groove on the clamp, and that the isolated polymerases compete with similar lifetimes on the clamp-DNA. This equilibrium is directly influenced by the concentration of the polymerases, or by physically tethering the polymerase to the clamp loader, as is observed for the τ clamp loader complex and Pol IIIcore (Table 1, Figure 3H–J)

DNA lesions do not affect the recruitment of translesion polymerases

The apparent lack of organization for the switching of the two polymerases raises the question of how Pol IV is recruited to the site of a lesion. We therefore wondered whether the lifetime of Pol IIIcore is affected by the nature of the DNA substrate. For this, we compared the lifetimes on three different DNA substrates: a matched, a mismatched, and a substrate containing a N2-furfuryl-dG lesion (Jarosz et al., 2006). These three DNA substrates should elicit very different outcomes, i.e. extension on a matched DNA substrate, mismatch removal by the exonuclease ε on a mismatched substrate, or polymerase switching on a DNA lesion. Surprisingly, the lifetime of Pol IIIcore on clamp-DNA is not altered by the presence of either a mismatch or the lesion (Table 3, Figure 4—figure supplement 3A–C). This indicates that Pol IIIcore does not 'discriminate' between the different DNA substrates. This is also observed in the presence of the two nucleotides dATP and dTTP (0.5 mM each), which give little change in the lifetime of Pol IIIcore on all three DNA substrates (Table 3, Figure 4—figure supplement 3D–F). Likewise, the exchange of Pol IIIcore to Pol IV is similar on all three substrates, (Table 3). Hence, Pol IIIcore dissociation is unaffected by mismatches or DNA lesions and the exchange between Pol IIIcore and Pol IV is not driven by the state of the DNA, but instead is a direct competition between the replicative and translesion polymerases.

Table 3
DNA lesion and mismatches do not affect the lifetime of Pol IIIcore on clamp-DNA or its competition with Pol IV
https://doi.org/10.7554/eLife.32177.013
Lifetime (s)Polymerase exchange (%)
No dNTPdATP/dTTPIII → IVIV → IIIIII + IV
Matched15.7 ± 1.116.1 ± 1.0731413
Lesion*17.6 ± 2.116.4 ± 1.4582418
Mismatched*19.0 ± 1.417.5 ± 0.5642313
  1. *Lesion DNA: N2-furfuryl-dG, mismatched DNA: G-T.

    Polymerase exchange on observed on clamp-DNA showing the exchange from Pol IIIcore to Pol IV (III→IV), Pol IV to Pol IIIcore (IV→III), or co-localization of Pol IIIcore and Pol IV (III + IV). Exchange rates measured in the absence of nucleotides

Discussion

In this work we use co-localization single molecule studies to show that the replication proteins of the clamp loader, clamp, and Pol IIIcore are a highly organized in their sequential actions, ensuring that DNA replication occurs in an efficient manner (Figure 5). In contrast, the translesion polymerases appear to gain access to the DNA in a more stochastic, concentration dependent manner. This implies that the cell needs a mechanism in order to tune the outcome of this competition towards the replicative polymerases during normal DNA synthesis and the translesion polymerases in the presence of DNA damage. Indeed, during DNA synthesis, two or three Pol IIIcores are tethered to the replisome (McInerney et al., 2007; Reyes-Lamothe et al., 2010), increasing the effective concentration of the replicative polymerase at the replication fork. Importantly, the Pol IIIcores are directly tethered to the clamp loader via the flexible linker of the clamp loader protein τ (Dallmann et al., 2000; Gao and McHenry, 2001), which ensures that as soon as a clamp is loaded, a Pol IIIcore will associate with the clamp (see Figure 3H–J). Secondly, during the SOS-response, a bacterial reaction to DNA damage, the cellular levels of the translesion DNA polymerases Pol II, Pol IV, and Pol V are increased (Bonner et al., 1988; Woodgate and Ennis, 1991; Kim et al., 2001), thus shifting the equilibrium of the polymerase competition in favor the translesion polymerases. Furthermore, it has been proposed that the recombinase RecA may favor the translesion polymerases by having opposing effects on the Pol III and translesion replisomes, albeit by a yet unknown mechanism (Indiani et al., 2013).

A model for DNA replication and translesion synthesis.

The DNA replication cycle consists of a sequence of carefully arranged steps of clamp loading, polymerase loading, DNA synthesis, polymerase release, and clamp unloading. In contrast, translesion DNA synthesis over DNA adducts shows no coordinated sequence of events, but is instead a direct competition between the replicative DNA polymerase Pol IIIcore, and the translesion DNA polymerases Pol II and Pol IV. DNA replication will resume once the lesion has been bypassed by one of the translesion DNA polymerases.

https://doi.org/10.7554/eLife.32177.014

Such a concentration-dependent exchange between the replicative and translesion polymerases is consistent with the discontinuous DNA synthesis by isolated Pol IIIcore (Jergic et al., 2013; Tanner et al., 2008), showing that individual Pol IIIcores can exchange. Recent studies (Lewis et al., 2017; Beattie et al., 2017) have also shown that in the context of the intact replisome, Pol IIIcore exchanges in a concentration-dependent manner during processive DNA replication. Taken together, these studies demonstrate that E. coli DNA synthesis is highly dynamic and utilizes a concentration-dependent mechanism to achieve a fine balance between stability and flexibility where the exchange of factors is determined by their availability.

Interestingly, in higher eukaryotes the access of the eukaryotic translesion polymerases to DNA shows a higher degree of coordination through ubiquitination of PCNA and the translesion polymerases (Chun and Jin, 2010), as well as formation of polymerase bridges through Rev1 (Pustovalova et al., 2016; Sale, 2013). However, given that there are up to 15 DNA polymerases to coordinate in eukaryotes (Plosky and Woodgate, 2004), the ubiquitination of PCNA and the interactions between the polymerases may not be sufficient to coordinate the specific recruitment of individual translesion polymerases, implying that in eukaryotes too, the translesion process may occur at least in part by a concentration-dependent mechanism, which is supported by studies showing that the intracellular levels of several human translesion polymerases (Pol η, Pol κ, and Pol ι) are increased upon DNA damage (Zhu et al., 2010; Zhu et al., 2012; Tomicic et al., 2014).

Materials and methods

Materials

All chemicals were purchased from Sigma-Aldrich (Gillingham, United Kingdom), unless stated otherwise. All chromatography columns were purchased from GE healthcare (Little Chalfont, United Kingdom) .

Cloning of protein expression vectors

Genes for E. coli β (dnaN), ε (dnaQ), θ (holE), γ (dnaX) and Bacillus subtilis Sfp phosphopantetheinyl transferase were cloned into pET28a vectors, and genes for δ (holA) and single cysteine δ' (holB) K83C/C217S/C294S(Goedken et al., 2004) were cloned into pET3d vectors. The sequence for Pol I (polA) Klenow fragment (residues 324–928) was cloned into a pETNKI-His-3C-LIC (Luna-Vargas et al., 2011) vector. For labeling purposes the Ybbr sequence DSLEFIASKLA (Yin et al., 2005) was added N-terminally to the following proteins during cloning into their respective vectors: Pol IIIα (dnaE) was cloned into a pETNKI-His-3C-LIC vector, Pol II (polB) and Pol IV (dinB) were cloned into pET11 vectors, and the gene for ε (dnaQ) was cloned into a pET28a vector. The plasmids for streptavidin 'alive' (biotin-binding) and streptavidin 'dead' (not biotin-binding) (Howarth et al., 2006) were generous gifts from M. Howarth (Univ. of Oxford).

Clamp binding mutants of ε, Pol IV, and Pol II were generated through site directed mutagenesis. The following sequences were changed: ε (β-) residues 182–187: QTSMAF to QTSAAA, ε (β+) residues 182–187: QTSMAF to QTSLPL, Pol IV (β groove) residues 346–351: QLVLGL to QLVAGA, Pol IV (β rim) residues 303–305 VWP to AGA, Pol II (β groove), residues 779–783: QLGLF to QLGAA.

Protein purification

Unless otherwise stated, protein purifications were performed with the following gradients: nickel affinity (25–500 mM Imidazole gradient in the presence of 500 mM NaCl), ion exchange (0–1 M NaCl gradient), hydrophobic interaction chromatography (2–0 M ammonium sulfate gradient) and gel filtration (150 mM NaCl). Pol IIIα-NYbbr was purified in 20 mM Hepes pH 7.5 and 2 mM DTT by nickel affinity, ion exchange and gel filtration. Pol I Klenow fragment was purified in 20 mM Tris pH 8.0 and 2 mM DTT by nickel affinity, anion exchange, hydrophobic interaction chromatography and and gel filtration. Pol IV-NYbbr was purified in 20 mM Hepes pH 7.5 and 2 mM DTT by ion exchange, hydrophobic chromatography and gel filtration. Pol II-NYbbr was purified in 20 mM Tris pH 8.0, 0.5 mM EDTA and 2 mM DTT by two steps of ion exchange separated by a step of hydrophobic chromatography, followed by gel filtration. ε-NYbbr was purified in 25 mM Hepes pH 8.2 and 2 mM DTT by nickel affinity under denaturing conditions (in the presence of 6 M Urea), followed by refolding overnight at 4°C in 25 mM Hepes pH 8.2 and 10 mM DTT and ion exchange. β, ε, θ and Sfp were purified in 20 mM Hepes pH 7.5 and 2 mM DTT by nickel affinity, ion exchange and gel filtration. δ and δ' K83C/C217S/C294S were purified in 50 mM Hepes pH 7.5, 0.1 mM EDTA and 2 mM DTT by hydrophobic chromatography and ion exchange, and γ/τ was purified in the same buffer by nickel affinity chromatography. The γ3δ1δ'1 and τ3δ1δ'1 complexes were assembled at a 1:2:1 ratio of γ/τ:δ:δ' and separated by ion exchange. To fully occupy all Pol IIIcore binding site, a 5-fold excess of Pol IIIcore was added to the τ3δ1δ'1 complex, and then purified by gel filtration. Streptavidin alive and dead were purified and prepared as described (Howarth et al., 2006).

Protein labeling

Pol IIIα, Pol IV, Pol II and ε exonuclease subunit were enzymatically labeled by Sfp, which conjugates CoA-linked Atto dyes to their N-terminal Ybbr tags (Yin et al., 2006). The labeled proteins were Pol IIIα-Atto 488, Pol IV-Atto 565, Pol II-Atto 647N, and ε-Atto 565. The β clamp and Pol I Klenow fragment were labeled on a single cysteine residue with maleimide-Atto647N. The γ-complex (δγ3δ') was labeled at the δ' subunit with the mutations K83C/C217S/C294S (Goedken et al., 2004) with maleimide-Atto 565 before complex assembly. All labeled proteins were purified away from the free dye by gel filtration. The labeling efficiencies of the proteins were determined using the protein and fluorophore absorption ratios, with the free fluorophore absorption at 280 nm subtracted from the protein absorption at 280 nm. The labeling efficiency was 85% for the γ- and τ-complexes, 68% for the β clamp monomer, 100% for Pol I Klenow fragment, 60% for Pol IIIα, 78% for ε, 62% for Pol IV and 66% for Pol II. In addition, the labeling efficiencies for Pol IIIα and ε were verified at the single molecule level by measuring the co-localization frequencies between Pol IIIα and ε on clamp-DNA, where each co-localization event was scored as having labeled Pol IIIα only, labeled ε only, or both labeled proteins. This way, the labeling efficiencies were 67% for Pol IIIα and 71% for ε, which are similar to the efficiencies measured by absorption.

φX174 primer extension assays

Protein activity was tested using single stranded ϕX174 phage DNA (New England Biolabs, Hitchin, United Kingdom), primed with a 5' fluorescein labeled primer (sequence: 5' FAM-ACCAACATAAACATTATTGCCCGGCGTACpG, where lowercase 'p' indicates the non-cleavable phosphothioate bond). Reactions were performed in 20 mM Tris pH 7.5, 2 mM DTT, 50 mM potassium glutamate, 8 mM magnesium acetate, and 0,05 mg/ml BSA. Each reaction contained 5 nM primed ϕX174 phage DNA, 50 nM β clamp, 10 nM γ clamp loader complex (γ3δ1δ'1), and 30 nM polymerase (Pol IIIcore, Pol II, or Pol IV). Reactions were quenched at 0, 0.5, 1, 2 and 5 min with 75 mM EDTA and 0.6% (W/V) SDS and stored on ice before separated on a alkaline agarose gel (0.8% agarose, 30 mM NaOH, 2 mM EDTA) for 15 hr at 14 V. Gels were scanned at 488 nM using a Amersham Typhoon (GE Healthcare)

DNA substrates

All DNA oligos were ordered from IDT (Leuven, Belgium), with the exception of the furfuryl-modified oligo that was purchased from Eurogentec (Seraing, Belgium). The following DNA substrates were used. A 33-nt template DNA: 5' Bio-CATAATATCCATGCTTCACC[amino-dT]TCATCCAAATCC for the matched and mismatched substrates or 5' Bio-CATAA[N2-furfuryl-dG]ATCCATGCTTCACC[amino-dT]TCATCCAAATCC for the lesion substrate. A 27-nt primer DNA: 5' Bio-GGATTTGGATGAAGGTGAAGCATGGApT for the matched and lesion substrates (where lowercase 'p' indicates the non-cleavable phosphothioate bond) or a 25-nt 5' Bio-GGATTTGGATGAAGGTGAAGCATGpT for the mismatched substrate. The template DNA was labeled on the internal amino-modified thymine with NHS-Atto 488 and purified away from the free dye by gel filtration. The labeled template DNA was subsequently bound to monovalent streptavidin and purified by gel filtration, before annealing to the primer DNA and binding to the glass cover slip for imaging. See Figure 1A for cartoon representation of final DNA substrate.

Preparation of slides

Glass slides and cover slips were washed in 3 M NaOH and Piranha solution (3:2 concentrated sulfuric acid: 30% hydrogen peroxide) and then silanized and pegylated essentially as described in (Ha et al., 2002). The imaging chamber (15 μL) was assembled by creating a sandwich between the cover slip and glass slide using double adhesive tape, and it was further passivated using 4 mg/ml PLL-PEG (SuSoS, Dübendorf, Switzerland), 1% (W/V) pluronic F127 (Sigma) and 10 mg/ml BSA (New England Biolabs). Streptavidin (1 mg/ml) (New England Biolabs) was added last to bind the biotin-DNA.

CoSMoS microscopy

All single-molecule measurements were performed at 23 (±1) °C in 20 mM Tris-HCl pH 7.5, 50 mM potassium glutamate, 8 mM MgCl2, 4% glycerol, 2 mM DTT, 0.1% Tween20 and 1 mM Trolox. The protein concentrations used were: 15 nM β2 clamp; 15 nM γ3δ1δ'1 complex; 30 nM or 150 nM Pol IIIcore; 6 nM, 30 nM, or 150 nM Pol IV; and 30 nM Pol II. Movies were acquired using Micromanager software on a Nikon (Kingston Upon Thames, United Kingdom) Eclipse Ti-E microscope with ApoTirf 100X/1.49 Oil, 0.13–0.20 WD 0.12 objective. The lasers used were 150 mW 488 nm, 150 mW 561 nm (both Coherent Sapphire Ely, United Kingdom) and 100 mW 638 nm (Coherent Cube) controlled by an acousto-optic tunable filter (Gooch and Housego, Ilminster, United Kingdom). Movies were acquired on an Andor (Belfast, United Kingdom) iXon (EM) + CCD camera at 20–40 mW laser power and an exposure of 50 millisecond (ms) per frame for a thousand frames with rapid alternation between the three laser channels (180 ms/change). Thus the frame rate is 660 ms for a 3-color experiment, 440 ms for a 2-color experiment, and 86 ms for a single color experiment (as there is no need to change the laser channels). The total duration of the movies for a 3-color experiment was 650–670 s, which gives an acquisition rate of 220 milliseconds per frame when the lag in the filter wheel was taken into consideration. Each field of view (54 μm × 54 μm, pixel size 105 × 105 nm) had an average density of 5900 DNA molecules of which 700–900 molecules were well separated and picked for analysis. The DNA-Atto 488 signal was bleached in order for this channel to be re-used for Pol IIIα-Atto 488.

Data analysis

The acquired movies were fed into the Imscroll analysis GUI developed by Jeff Gelles and Larry Friedman (Friedman and Gelles, 2015) to find the individual landing events and their dwell times. Histograms were plotted in Igor Pro and the data was fitted with an one-parameter exponential using a bin size ~1/2 the lifetime of the molecule. The decay constant τ represents the mean dwell time and the error in τ represents the error in the mean dwell time.

References

  1. 1
  2. 2
  3. 3
  4. 4
    Purification and characterization of an inducible Escherichia coli DNA polymerase capable of insertion and bypass at abasic lesions in DNA
    1. CA Bonner
    2. SK Randall
    3. C Rayssiguier
    4. M Radman
    5. R Eritja
    6. BE Kaplan
    7. K McEntee
    8. MF Goodman
    (1988)
    The Journal of Biological Chemistry 263:18946–18952.
  5. 5
  6. 6
  7. 7
  8. 8
    DnaX complex of Escherichia coli DNA polymerase III holoenzyme. Central role of tau in initiation complex assembly and in determining the functional asymmetry of holoenzyme
    1. HG Dallmann
    2. RL Thimmig
    3. CS McHenry
    (1995)
    The Journal of Biological Chemistry 270:29555–29562.
  9. 9
  10. 10
  11. 11
    Size classes of products synthesized processively by DNA polymerase III and DNA polymerase III holoenzyme of Escherichia coli
    1. PJ Fay
    2. KO Johanson
    3. CS McHenry
    4. RA Bambara
    (1981)
    The Journal of Biological Chemistry 256:976–983.
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
    Roles of chromosomal and episomal dinB genes encoding DNA pol IV in targeted and untargeted mutagenesis in Escherichia coli
    1. SR Kim
    2. K Matsui
    3. M Yamada
    4. P Gruz
    5. T Nohmi
    (2001)
    Molecular Genetics and Genomics : MGG 266:207–215.
  36. 36
  37. 37
    DNA replication fidelity
    1. TA Kunkel
    (2004)
    Journal of Biological Chemistry 279:16895–16898.
    https://doi.org/10.1074/jbc.R400006200
  38. 38
  39. 39
  40. 40
    The Croonian Lecture, 1996: endogenous damage to DNA
    1. T Lindahl
    (1996)
    Philosophical Transactions of the Royal Society B: Biological Sciences 351:1529–1538.
    https://doi.org/10.1098/rstb.1996.0139
  41. 41
  42. 42
  43. 43
  44. 44
    DNA polymerase III of Escherichia coli. Purification and identification of subunits
    1. CS McHenry
    2. W Crow
    (1979)
    The Journal of Biological Chemistry 254:1748–1753.
  45. 45
  46. 46
    Mechanism of translesion DNA synthesis by DNA polymerase II. Comparison to DNA polymerases I and III core
    1. T Paz-Elizur
    2. M Takeshita
    3. M Goodman
    4. M O'Donnell
    5. Z Livneh
    (1996)
    The Journal of Biological Chemistry 271:24662–24669.
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
    Constitution of the twin polymerase of DNA polymerase III holoenzyme
    1. PS Studwell-Vaughan
    2. M O'Donnell
    (1991)
    The Journal of Biological Chemistry 266:19833–19841.
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64

Decision letter

  1. Taekjip Ha
    Reviewing Editor; Johns Hopkins University School of Medicine, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Single-molecule studies contrast ordered DNA replication with chaotic translesion synthesis" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The reviewers agreed that your approach is novel and the messages touch about timely topics of broad interest. However, three major issues were raised. (1) Controls showing that the labeled components are functional in DNA synthesis are missing. (2) Bolstering their claims by adding much more data and analysis, instead of just showing one or few example time traces, is also a must. (3) Many more additional experiments are needed, for example including clamp loader that bridges pol III and clamp, and measurements that demonstrate the necessity of clamp interaction.

Your paper is being rejected because we do not feel that you would be able to perform all requested additional experiments (most critically point 3 above) that the reviewers found essential in two months, which is the time scale of additional work that eLife considers reasonable. However, if you are able to perform key experiments requested, I will be happy to consult the same reviewers with the new submission.

Reviewer #1:

Zhao et al. investigate the binding of E. coli DNA polymerases to the sliding clamp using single-molecule co-localization microscopy. The authors argue that replicative and translesion polymerase do not simultaneously interact with the clamp (toolbelt model), but exchange on the clamp in a concentration-dependent manner. This manuscript addresses a critical question in the field, albeit with a rather limited scope. Most importantly, clamp loader (with and without τ) is omitted from the polymerase co-localization experiments. Τ bridges clamp and polIII, and may further stabilize the "dual-polymerase" clamp complex in vivo. This will be a critical addition to the manuscript. Second, there are no biochemical controls indicating activity of fluorescently tagged enzymes. Statistical analysis is also lacking and in a few cases the key results are argued based on a single CosMoS trace. Overall, the manuscript feels more like the start of an interesting paper than a significant contribution to the field. I recommend that the authors make the following major edits to be considered for publication in eLife.

1) Include biochemical controls to confirm that clamp loader and polymerases are not perturbed by the fluorescent tag.

2) Include experiments with clamp loader, pol III and pol IV. Compare clamp loader with and without τ. Are more "dual polymerase" species seen with τ? There is growing evidence for multi-τ clamp loaders in vivo (from Sherratt lab and others).

3) Figure 1C: Clamp on-rates (~0.6 s) are limited by the experimental resolution (0.6 s / frame). The underlying rates may be significantly faster; the authors may be measuring slow events. Time resolution needs to be increased to make any claims here.

4) Once a clamp is loaded, clamp loader can catalyze unloading (e.g., Figure 1G,J). Does every clamp loader binding event result in clamp unloading (as suggested in Figures 1G,J)? Clamp unloading probabilities should be characterized and reported. If unloading is rare, more representative traces shown in Figure 1.

5) In the experiments detailing matched, mismatched, and lesion template, the authors investigate the lifetime without dNTPs and with dCTP/dGTP. If I understood the oligo table correctly, the next nucleotide to be incorporated is dATP or dTTP. Including these nucleotides may show changes in how the polymerase interacts with the templates.

6) Do changes in competing polymerase (Pol II or Pol IV) concentration affect the lifetime of Pol III core on the clamp? This has been suggested previously (PMCID: PMC4040570).

7) Does an increase in polymerase concentration increase the co-association of Pol III core and Pol IV?

8) What are the relevant concentrations of these polymerases in cells before and after SOS induction? These concentration ranges (or at least the relative polymerase ratios) should be investigated here.

9) A separation-of-function Pol III mutant that lacks interaction with the clamp will be useful to confirm that specific interactions are responsible for the increase in lifetime.

10) Controls with weakened clamp-pol IV interactions will also confirm the observed lifetimes and polymerase exchange dynamics (e.g., PMCID: PMC4040570).

11) Authors make a sweeping claim that all three polymerases "compete equally" for the clamp based on a single trace (Figure 3G). This is a shaky and unclear claim. Elaborate with more statistical analysis, and with wt and mutant proteins that weaken interactions.

Reviewer #2:

This paper uses co-localization microscopy with labeled DNA polymerases to determine the sequence of events of the binding of components of the replisome and different DNA polymerases to DNA. Clamp loading and unloading was observed, as well as association times of pol IIIcore and pol IV or pol II, which supports a model of stochastic binding of pol IV or pol II compared to a more ordered assembly of pol III components.

A major question is: are the tagged DNA polymerases active? The clamp loader and clamp clearly are competent for loading, but polymerization activity (and proofreading) by the tagged polymerases should be demonstrated.

"The shorter lifetimes recorded for the clamp are not caused by early release […] are instead due to the late loading of the clamps during the course of the experiment." Is this because this is observed near the end of the acquisition time? This should be explained.

Figure 2, were these experiments carried out with labeled α or labeled ε subunit? This question holds in every instance that specifies only "pol III core".

Subsection “Pol IIIcore has an intrinsic lifetime on DNA that is independent of its activity”, last paragraph and Table 1, what is the concentration of dGTP and dCTP used in these experiments?

"Pol IIIcore and Pol IV do not form a stable complex on clamp-DNA […] and there is a significant time lag between the release of Pol IIIcore and Pol IV arrival" However, it appears that sometimes Pol IIIcore and Pol IV do stably associate on the clamp (Figure 3C shows a significant amount of time in which the two pols are co-localized). And, top "rare co-localization events." How rare are these? Table 1 shows this to be 13-18% of total events, but what about the time of co-localization? It would seem prudent to include the co-localized times in the analysis as well and present those results here. In the representative trace shown, the co-localized time appears to exceed most of the association lifetimes of either polymerase alone on DNA. It would be of interest to conduct this experiment without β clamp. Also, does the percentage of co-localization change with nucleotides?

Subsection “Protein purification”, more information and/or references should be given describing the protein purification, including buffer compositions.

Subsection “CoSMoS microscopy”, the 488 label on DNA was photobleached before imaging pol IIIcore-α-488. What controls were done to ensure the photobleaching doesn't damage the DNA?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Single-molecule studies contrast ordered DNA replication with chaotic translesion synthesis" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Michael Marletta as the Senior Editor. The reviewers have opted to remain anonymous.

Your revised manuscript has been seen by the original reviewers. They generally agreed that the manuscript has improved significantly but made several important suggestions that should be helpful in improving rigor and clarity. Please revise your manuscript accordingly. No new experiments are required.

Reviewer #2:

This version of the manuscript is improved, but there are still some areas to address.

In the response, the authors state that the only report of polymerase concentrations in E. coli is Sutton 2009 BBA, which is a review article. This paper cites four other primary research articles, where the levels of polymerases are actually measured. I understand the desire to keep papers a reasonable length, and so cite a review when appropriate, but it is just not true that Sutton 2009 is the only report of polymerase levels. In addition, if one reads this paper and the cited references, it becomes clear that the levels of pol IV are well above those of pol III even without SOS induction, and therefore statements about "low intracellular levels" of pol IV under "normal" (non-SOS induced?) conditions (Discussion, first paragraph) need to be revisited.

In the first paragraph of the Introduction: "When the errors occur on the template DNA strand, in the form of DNA lesions […]" is misleading. DNA lesions that arise from chemicals, etc., are not replication errors. The wording should be revised.

In the Introduction, studies supporting the toolbelt model "all […] use bulk studies that due to the asynchronous nature cannot separate out the sequential steps." While this statement is true, it seems to ignore the fact that steady-state FRET experiments support the idea that two pols can bind β simultaneously (Indiani 2005 Mol Cell).

In the last paragraph of the Discussion, ordered vs. stochastic, the manuscript doesn't seem to be entirely convincing on this point. The exchange between two different pol III cores (two different labels) should be examined as a direct comparison to the other two-polymerase experiments. Either the experiments can be conducted to or it would be sufficient to revise the discussion, the general interpretation of experiments, and to change the title of manuscript.

Tables 1 and 2 are confusing. Why is pol II in Table 1? In Table 2, it is not clear what is being compared in each case. The last column gives co-localization, and the heading is "III+IV/II"; which is reported in the values for pol III (first 4 entries), pol II or pol IV?

https://doi.org/10.7554/eLife.32177.019

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

Summary:

Zhao et al. investigate the binding of E. coli DNA polymerases to the sliding clamp using single-molecule co-localization microscopy. The authors argue that replicative and translesion polymerase do not simultaneously interact with the clamp (toolbelt model), but exchange on the clamp in a concentration-dependent manner. This manuscript addresses a critical question in the field, albeit with a rather limited scope. Most importantly, clamp loader (with and without τ) is omitted from the polymerase co-localization experiments. Τ bridges clamp and polIII, and may further stabilize the "dual-polymerase" clamp complex in vivo. This will be a critical addition to the manuscript.

In the previous version of the manuscript, all experiments were performed in the presence of the γ clamp loader complex (i.e. γ3δδ', without τ), with the exception of the clamp unloading and the three-polymerase experiment, which were performed after washing away the clamp loader.

In the current version we have now also included the τ clamp loader complex (τ3δδ'), as shown in Figure 3. In summary, when using the τ clamp loader complex that is pre-loaded with Pol IIIcore, we find that the clamp loader, Pol IIIcore, and clamp arrive at the DNA together. Unlike the γ clamp loader that leaves immediately after clamp loading, the τ complex remains present as it is attached to Pol IIIcore that takes its place at the clamp.

The lifetime of Pol IIIcore is not altered in the presence of τ. As the τ clamp loader and Pol IIIcore arrive together, there is a dramatic shift in the competition between Pol IIIcore and Pol IV in favour of Pol III, as shown in the table below (Author response Table 1). The percentage of co-localization events between Pol IIIcore and Pol IV on the other hand is not much affected.

Concentration (nM) Polymerase exchange (%) Lifetime

Author response table 1
Competition between polymerases clamp-DNA.
https://doi.org/10.7554/eLife.32177.018
DNAConcentration (nM)Polymerase exchange (%)Lifetime
CompetitionPol IIIcorePol IV/IIIII→IVIV→IIIIII+IVPol IIIcore
γ complex
Pol IIIcore – Pol IV
30307022915.7 ± 1.1
τ complex
Pol IIIcore – Pol IV
3030950513.5 ± 1.7

Second, there are no biochemical controls indicating activity of fluorescently tagged enzymes.

We have now included bulk biochemical experiments, a 5.4 kb ϕX174 primer extension assay, and show that the labelled proteins (i.e. clamp, clamp loader, pol IIIcore, Pol IV, and Pol II) are unaltered in DNA synthesis when compared to the unlabeled proteins (see Figure 1—figure supplement 1A-C).

Statistical analysis is also lacking and in a few cases the key results are argued based on a single CosMoS trace.

We now have included the statistical analysis for all experiments throughout the figures and figure supplements.

Overall, the manuscript feels more like the start of an interesting paper than a significant contribution to the field. I recommend that the authors make the following major edits to be considered for publication in eLife.

1) Include biochemical controls to confirm that clamp loader and polymerases are not perturbed by the fluorescent tag.

As explained above, we have now included primer extension assays using a primed, 5.4 kb single stranded ϕX174 phage DNA that requires clamp loading for processive DNA synthesis. The results show that the labeled proteins retain wild-type activities (see Figure 1—figure supplement 1A-C).

2) Include experiments with clamp loader, pol III and pol IV. Compare clamp loader with and without τ. Are more "dual polymerase" species seen with τ? There is growing evidence for multi-τ clamp loaders in vivo (from Sherratt lab and others).

As discussed above (see "Summary: Most importantly"), these experiments have now been included in the revised manuscript. We have compared the clamp loader with and without τ. We are not sure what the reviewer means with "dual polymerase species". If it is referring to co-localization between Pol IIIcore and Pol IV, then no, we do not observe increased colocalization in the presence of τ Table 1 above). If the reviewers refer to multiple Pol IIIcores bound to one τ clamp loader complex, then yes, we make the τ-complex (τ3δ1δ'1) fully loaded with three Pol IIIcores.

3) Figure 1C: Clamp on-rates (~0.6 s) are limited by the experimental resolution (0.6 s / frame). The underlying rates may be significantly faster; the authors may be measuring slow events. Time resolution needs to be increased to make any claims here.

We have repeated the clamp loader experiments using a single fluorescent channel only, thus reducing the frame rate to 0.086 s/frame. At this frame rate, the lifetime of the clamp loader on DNA in the presence or absence of the clamp is 0.41 ± 0.01 s and 1.20 ± 0.05 s, respectively. This is now shown in Figure 2D and discussed in the first paragraph of the subsection “Clamp loading and unloading are distinct processes”.

4) Once a clamp is loaded, clamp loader can catalyze unloading (e.g., Figure 1G,J). Does every clamp loader binding event result in clamp unloading (as suggested in Figure 1G,J)? Clamp unloading probabilities should be characterized and reported. If unloading is rare, more representative traces shown in Figure 1.

We now measured the success rate of unloading, which is ~50%. This is now shown in Figure 2—figure supplement 1F-G discussed in the last paragraph of the subsection “Clamp loading and unloading are distinct processes”.

5) In the experiments detailing matched, mismatched, and lesion template, the authors investigate the lifetime without dNTPs and with dCTP/dGTP. If I understood the oligo table correctly, the next nucleotide to be incorporated is dATP or dTTP. Including these nucleotides may show changes in how the polymerase interacts with the templates.

In the initial manuscript, we chose not to use as dATP and dTTP as their incorporation would shorten the ssDNA overhang by 5 nucleotides and potentially change the interaction with the polymerase (in many crystal structures of polymerase-DNA complexes, the polymerase binds to 3-4 nucleotides of the single stranded overhang of the template strand. Our DNA substrate has a 6 nt overhang). Moreover, Pol IIIcore synthesizes DNA at speeds of 1000 nt/s, so the addition of 5 nucleotides could only add 5 milliseconds to the ~16 second lifetime of Pol IIIcore on clamp-DNA. Finally, the end product of incorporation of the five nucleotides would be the same as the starting DNA substrate with dCTP and dGTP, i.e. a primer that cannot be extended. Nonetheless, we have performed the idling experiment with dATP and dTTP and found that as for dCTP and dGTP, the lifetime of Pol IIIcore is comparable on all three DNA substrates (matched, mismatched, and lesion). This is now shown in Table 3 and discussed in the subsection “DNA lesions do not affect the recruitment of translesion polymerases”.

6) Do changes in competing polymerase (Pol II or Pol IV) concentration affect the lifetime of Pol III core on the clamp? This has been suggested previously (PMCID: PMC4040570).

We have now measured different concentrations of Pol IV (6-150 nM) and do not see an effect on the lifetime of Pol IIIcore. We do see a clear effect on the polymerase switching (III>IV vs IV->III), a decrease in the lag time between binding events, and an increase of colocalizations of Pol IIIcore and Pol IV with increased protein concentrations. This is now shown in Table 1 and discussed in the first paragraph of the subsection “Pol IIIcore and Pol IV alternate binding to clamp-DNA”.

7) Does an increase in polymerase concentration increase the co-association of Pol III core and Pol IV?

Yes it does. This is now shown in Table 1 and discussed in the second paragraph of the subsection “Pol IIIcore and Pol IV alternate binding to clamp-DNA”.

8) What are the relevant concentrations of these polymerases in cells before and after SOS induction? These concentration ranges (or at least the relative polymerase ratios) should be investigated here.

To our knowledge, there is currently only one report of the cellular concentration of the different polymerases: Sutton et al. 2009 (PMID 19540941). Here the cellular concentrations of E. coli Pol IIIcore and Pol IV are 20 and 330 nM respectively, and 20 and 3300 nM after SOS induction. In our experiments, we have measured Pol IIIcore and Pol IV at 6, 30, and 150 nM (Table 1). (Measuring higher protein concentrations results in increased background and non-specific binding to the glass slide.) At the measured concentrations it is clear that Pol IV can frequently switch places and/or co-localize with Pol IIIcore. Addition of τ favors the binding of Pol IIIcore to the DNA, although the frequency of co-localization with Pol IV is comparable to that of the isolated Pol IIIcore. In the cell, Pol IIIcore is connected to the rest of the replisome, implying that the local effective concentration will favor Pol IIIcore binding even more. This is now shown in Table 1 and discussed in the second paragraph of the subsection “Pol IIIcore and Pol IV alternate binding to clamp-DNA”.

9) A separation-of-function Pol III mutant that lacks interaction with the clamp will be useful to confirm that specific interactions are responsible for the increase in lifetime.

The clamp interaction motif of the polymerase subunit α of Pol IIIcore is essential for clamp binding and DNA synthesis (Dohrman 2005, PMID 15923012). Instead we have made two variants of the clamp binding motif in the exonuclease subunit ε of Pol IIIcore; one to decrease clamp binding and one to increase clamp binding. As expected, the reduced clamp binder of ε results in shorter lifetimes of Pol IIIcore on clamp-DNA, and the improved clamp binder of ε gives longer lifetimes of Pol IIIcore on clamp-DNA. This is now shown in Table 2 and discussed in the first paragraph of the subsection “Polymerases compete for binding to the hydrophobic groove of the clamp”.

10) Controls with weakened clamp-pol IV interactions will also confirm the observed lifetimes and polymerase exchange dynamics (e.g., PMCID: PMC4040570).

We have made two mutant version of Pol IV: one in the canonical groove binding sequence, and one in the rim binding sequence (see Heltzel 2009, PMID 19617571). We find that mutation of the groove binding sequence strongly reduces Pol IV binding to the clamp and prevents co-localization. Mutation of the rim binding sequence of Pol IV had no observable effect. This is shown in Table 2 and discussed in the second paragraph of the subsection “Polymerases compete for binding to the hydrophobic groove of the clamp”.

11) Authors make a sweeping claim that all three polymerases "compete equally" for the clamp based on a single trace (Figure 3G). This is a shaky and unclear claim. Elaborate with more statistical analysis, and with wt and mutant proteins that weaken interactions.

We have replaced this statement with: "This therefore shows that all three polymerases compete for the same binding groove on the clamp, and that the isolated polymerases compete with similar lifetimes on the clamp-DNA."

Reviewer #2:

[…] A major question is: are the tagged DNA polymerases active? The clamp loader and clamp clearly are competent for loading, but polymerization activity (and proofreading) by the tagged polymerases should be demonstrated.

We have now included primer extension assays using a primed, 5.4 kb single stranded ϕX174 phage DNA that requires clamp loading for processive DNA synthesis. The results show that the labeled proteins retain wild-type activities. This is shown Figure 1—figure supplement 1A-C, and discussed in the subsection “Preparation of DNA substrates and fluorescently labeled proteins”.

"The shorter lifetimes recorded for the clamp are not caused by early release […] are instead due to the late loading of the clamps during the course of the experiment." Is this because this is observed near the end of the acquisition time? This should be explained.

We have now measured the clamp lifetime in a different manner. We first loaded the clamp using the clamp loader, then washed away the loader and unbound clamps and then started data collection using 10 seconds intervals between measurements to avoid bleaching of the fluorophore. This resulted in a measured lifetime of ~23 minutes, not including the loading and washing time. This is now shown in Figure 2C and discussed in the first paragraph of the subsection “Clamp loading and unloading are distinct processes”.

Figure 2, were these experiments carried out with labeled α or labeled ε subunit? This question holds in every instance that specifies only "pol III core".

All experiments with Pol IIIcore were performed with a labeled α subunit. This is now stated in the subsection "Preparation of DNA substrates and fluorescently labeled proteins". Only for the co-localization of α and ε we also used a labeled ε subunit (Figure 4F-G).

Subsection “Pol IIIcore has an intrinsic lifetime on DNA that is independent of its activity”, last paragraph and Table 1, what is the concentration of dGTP and dCTP used in these experiments?

0.5 mM for each nucleotide. This is now stated in the text (subsection “Pol IIIcore has an intrinsic lifetime on DNA that is independent of its activity”, second paragraph; subsection “Pol IIIcore and Pol IV alternate binding to clamp-DNA”, first paragraph; subsection “DNA lesions do not affect the recruitment of translesion polymerases”)

"Pol IIIcore and Pol IV do not form a stable complex on clamp-DNA […] and there is a significant time lag between the release of Pol IIIcore and Pol IV arrival" However, it appears that sometimes Pol IIIcore and Pol IV do stably associate on the clamp (Figure 3C shows a significant amount of time in which the two pols are co-localized). And, top "rare co-localization events." How rare are these? Table 1 shows this to be 13-18% of total events, but what about the time of co-localization? It would seem prudent to include the co-localized times in the analysis as well and present those results here.

We have extended our analysis of the co-localization events by testing different concentrations of Pol IIIcore and Pol IV (see Table 1). In the table we report the percentage of co-localization events, that increases from 9% at low protein concentrations to 26% at high protein concentrations. The measured lifetime is ~8 seconds. This is now discussed shown in Table 1 and discussed in the second paragraph of the subsection 2Pol IIIcore and Pol IV alternate binding to clamp-DNA”.

In the representative trace shown, the co-localized time appears to exceed most of the association lifetimes of either polymerase alone on DNA. It would be of interest to conduct this experiment without β clamp.

In the absence of the clamp, Pol IIIcore binds very transiently to the DNA (τon=<0.086 s). This therefore makes it highly unlikely for any co-localization between Pol IIIcore and Pol IV to occur. For example, a β-clamp mutant of pol IV that has a reduced lifetime of 2.7 s (compared to 15.4 s for wild type Pol IV) shows no co-localization with Pol IIIcore on clampDNA even at elevated concentrations (see Table 2).

Also, does the percentage of co-localization change with nucleotides?

No it does not. This is now mentioned in thein the second paragraph of the subsection 2Pol IIIcore and Pol IV alternate binding to clamp-DNA”.

Subsection “Protein purification”, more information and/or references should be given describing the protein purification, including buffer compositions.

We have now expanded the information regarding the protein purification (see Materials and methods, subsection “Protein purification”).

Subsection “CoSMoS microscopy”, the 488 label on DNA was photobleached before imaging pol IIIcore-α-488. What controls were done to ensure the photobleaching doesn't damage the DNA?

We have measured the lifetime of Pol IIIcore on clamp-DNA before and after photobleaching of DNA and find there is no difference. This is now shown in Figure 1—figure supplement 1 and mentioned in the subsection “Preparation of DNA substrates and fluorescently labeled proteins”.

Furthermore, we bleach the Atto 488 dye at its absorption peak at 488 nm. The absorption of DNA drops sharply to zero a >300 nm, so the amount of energy absorbed by the DNA at 488 nm is extremely low. Consequently, UV induced damage is negligible at > 300 nm (see Besaratinia et al., 2011, PMID 21613571).

[Editors' note: the author responses to the re-review follow.]

Reviewer #2:

[…] In the response, the authors state that the only report of polymerase concentrations in E. coli is Sutton 2009 BBA, which is a review article. This paper cites four other primary research articles, where the levels of polymerases are actually measured. I understand the desire to keep papers a reasonable length, and so cite a review when appropriate, but it is just not true that Sutton 2009 is the only report of polymerase levels. In addition, if one reads this paper and the cited references, it becomes clear that the levels of pol IV are well above those of pol III even without SOS induction, and therefore statements about "low intracellular levels" of pol IV under "normal" (non-SOS induced?) conditions (Discussion, first paragraph) need to be revisited.

We now cite the original references (Bonner et al., 1988, Woodgate and Ennis, 1991, Kim et al., 2001) that report the increased expression of Pol II, Pol IV, and Pol V during the SOS response (see: Discussion, first paragraph). Three other references listed in the Sutton 2009 paper (#36 Qiu '97, #37 Bonner '90, #146 McHenry '03) actually do not report protein levels and are therefore not included in our revised manuscript. In addition, we adjusted our wording regarding 'normal' and 'low' levels of pol IV and changed to "Secondly, during the SOS-response, a bacterial reaction to DNA damage, the cellular levels of the translesion DNA polymerases Pol II, Pol IV, and Pol V are increased [Bonner et al., 1988; Woodgate and Ennis, 1991; Kim et al., 2001], thus shifting the equilibrium of the polymerase competition in favor the translesion polymerases."

In the first paragraph of the Introduction: "When the errors occur on the template DNA strand, in the form of DNA lesions […]" is misleading. DNA lesions that arise from chemicals, etc., are not replication errors. The wording should be revised.

We have modified the text to be more clear: "In addition, rare nucleotide mis-incorporations into the primer strand prevent further DNA synthesis and a 3'-5' exonuclease is required to remove the misincorporated nucleotides [Kunkel, 2004]. In contrast, when the polymerase encounters a lesion on the template strand in the form of modified base caused by diverse chemicals, reactive oxygen species, or UV light [Liu et al., 2016; Lindahl, 1996], the high-fidelity replicative DNA polymerases are stalled."

In the Introduction, studies supporting the toolbelt model "all […] use bulk studies that due to the asynchronous nature cannot separate out the sequential steps." While this statement is true, it seems to ignore the fact that steady-state FRET experiments support the idea that two pols can bind β simultaneously (Indiani 2005 Mol Cell).

We have now included a statement about the FRET experiments of the Indiani 2005 paper: "The toolbelt model, which was originally based on steady-state Förster Energy Resonance Transfer (FRET) experiments that showed the simultaneous binding of the replicative and translesion polymerases to the β-clamp, has found support.…"

In the last paragraph of the Discussion, ordered vs. stochastic, the manuscript doesn't seem to be entirely convincing on this point. The exchange between two different pol III cores (two different labels) should be examined as a direct comparison to the other two-polymerase experiments. Either the experiments can be conducted to or it would be sufficient to revise the discussion, the general interpretation of experiments, and to change the title of manuscript.

As we show in Figure 2 and 3, the loading of the clamp and pol IIIcore occur in a well-defined and ordered sequence. In contrast, as shown in Figure 4, the exchange between Pol IIIcore and the translesion polymerase Pol II and Pol IV does not follow any particular sequence, and any possible exchange between the polymerase are observed. Hence, the polymerase exchange follows a random probability, which is the definition of stochastic. We therefore stand by our distinction between ordered replication and stochastic translesion synthesis. However, we have replaced "chaotic" with "stochastic" in the title, and have toned down the Discussion.

Tables 1 and 2 are confusing. Why is pol II in Table 1? In Table 2, it is not clear what is being compared in each case. The last column gives co-localization, and the heading is "III+IV/II"; which is reported in the values for pol III (first 4 entries), pol II or pol IV?

We have adjusted Table 1 and 2 to remove the confusion. Pol II has been removed from the column heading in Table 1, and we have added a footnote to explain the polymerase exchange ("Polymerase exchange observed on clamp-DNA showing the exchange from Pol IIIcore to Pol IV (III→IV), Pol IV to Pol IIIcore (IV→III), or co-localization (III+IV)"). In Table 2 we have inserted a division between the first six lines that record the exchange between Pol IIIcore and Pol IV, and the last two lines that record the exchange between Pol IIIcore and Pol II.

https://doi.org/10.7554/eLife.32177.020

Article and author information

Author details

  1. Gengjing Zhao

    MRC laboratory of Molecular Biology, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Emma S Gleave

    MRC laboratory of Molecular Biology, Cambridge, United Kingdom
    Contribution
    Data curation, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Meindert Hugo Lamers

    MRC laboratory of Molecular Biology, Cambridge, United Kingdom
    Present address
    Leiden University Medical Center, Leiden, Netherlands
    Contribution
    Formal analysis, Supervision, Validation, Writing—original draft, Writing—review and editing
    For correspondence
    m.h.lamers@lumc.nl
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-4205-1338

Funding

Medical Research Council (U105197143)

  • Gengjing Zhao
  • Emma S Gleave
  • Meindert Hugo Lamers

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank Andrew Carter and Nick Barry for help with TIRF microscopy, Jeff Gelles and Larry Friedman for help with CoSMoS data analysis, and David Rueda for helpful suggestions. This work was supported by the UK Medical Research Council through grant U105197143 to MHL.

Reviewing Editor

  1. Taekjip Ha, Reviewing Editor, Johns Hopkins University School of Medicine, United States

Publication history

  1. Received: September 21, 2017
  2. Accepted: December 5, 2017
  3. Accepted Manuscript published: December 6, 2017 (version 1)
  4. Accepted Manuscript updated: December 7, 2017 (version 2)
  5. Accepted Manuscript updated: December 11, 2017 (version 3)
  6. Version of Record published: December 15, 2017 (version 4)

Copyright

© 2017, Zhao et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 634
    Page views
  • 183
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Scopus, Crossref.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)