Gre-family factors modulate DNA damage sensing by Deinococcus radiodurans RNA polymerase

ABSTRACT

Deinococcus radiodurans is a highly stress resistant bacterium that encodes universal as well as lineage-specific factors involved in DNA transcription and repair. However, the effects of DNA lesions on RNA synthesis by D. radiodurans RNA polymerase (RNAP) have never been studied. We investigated the ability of this RNAP to transcribe damaged DNA templates and demonstrated that various lesions significantly affect the efficiency and fidelity of RNA synthesis. DNA modifications that disrupt correct base-pairing can strongly inhibit transcription and increase nucleotide misincorporation by D. radiodurans RNAP. The universal transcription factor GreA and Deinococcus-specific factor Gfh1 stimulate RNAP stalling at various DNA lesions, depending on the type of the lesion and the presence of Mn²⁺ ions, abundant divalent cations in D. radiodurans. Furthermore, Gfh1 stimulates the action of the Mfd translocase, which removes transcription elongation complexes paused at the sites of DNA lesions. Thus, Gre-family factors in D. radiodurans might have evolved to increase the efficiency of DNA damage recognition by the transcription and repair machineries in this bacterium.

Introduction

Deinococcus radiodurans (Dra) is a highly stress resistant bacterium that survives high doses of ionizing and ultraviolet radiation, desiccation, and treatment with oxidizing agents. The factors proposed to contribute to the stress resistance in Dra include robust antioxidant and cell detoxication systems, a multicopy genome and highly effective DNA repair pathways (reviewed in [1–5]). Furthermore, Dra cells accumulate Mn²⁺ ions under stress conditions [6], and low molecular weight manganese complexes were shown to scavenge reactive oxygen species and protect the proteome from oxidation [7–10]. Mn²⁺ also modulates the activity of several enzymes involved in stress resistance including superoxide dismutase and DNA polymerases [3,5].

The common types of DNA lesions that are accumulated in cells under stress conditions include single- and double-stranded breaks, pyrimidine dimers, abasic sites, oxidized bases, such as 8-oxoguanine and thymine glycol, and alkylated bases, such as O6-methylguanine and 1,N6-ethenoadenine. Various DNA lesions can severely impair DNA replication and also interfere with the process of transcription, by directly affecting the activity of RNA polymerase (RNAP) [11–13]. Stalled transcription elongation complexes (TECs) may in turn present strong barriers for the DNA replication machinery, further compromising genome stability [14,15]. Consequently, cells have developed sophisticated mechanisms for DNA damage sensing and repair. In addition to the global repair mechanisms, RNAP serves as a major sensor of DNA lesions that recruits repair factors to transcribed regions, resulting in preferential repair of the template DNA strand in active genes in vivo (transcription-coupled repair, TCR) [16,17].

In bacteria, transcription complexes stalled at DNA lesions were proposed to be directly recognized by the Mfd translocase or UvrD helicase that bring other components of the nucleotide excision pathway (NER) to the site of damage. The Mfd protein was the first factor identified as a key participant of TCR in bacteria [18–21]. It is an ATP-dependent DNA translocase that interacts with RNAP and can push stalled or paused TECs in the forward direction, resulting in TEC dissociation and exposing the lesion to the repair factors [22–27]. On the opposite, UvrD was shown to induce backward TEC translocation and recruit the NER factors without TEC disassembly [28,29].

The effects of DNA damage on the activity of RNAP in vitro were studied for only a few of the known DNA lesions, including abasic sites, 8-oxoguanine, alkylated bases and thymine dimers, mostly with eukaryotic RNAP II from Saccharomyces cerevisiae [30–37] and bacterial RNAP from Escherichia coli [38–41]. Overall, published studies demonstrated that DNA lesions can have profound and diverse effects on transcription. In particular, nucleotide modifications that significantly disrupt DNA structure, such as thymine dimers and bulky adducts, can block RNAP progression [34,35,42], while others can be bypassed with transient transcriptional pausing [36,38–41]. Some lesions can also provoke transcriptional mutagenesis, by stimulating incorporation of incorrect nucleotides into the nascent RNA [36–39,41,43,44].

Despite extensive studies of the mechanisms of stress response in Dra, which revealed dramatic effects of various stress conditions on gene expression [45–49], very little has been known about the properties of its transcriptional machinery, and the activity of Dra RNAP on damaged DNA templates has never been studied. We previously characterized Dra RNAP and demonstrated that it has high intrinsic RNA cleavage activity, which might have a role in reactivation of backtracked TECs during stress response [50–53]. Furthermore, it was shown that Deinococcus/Thermus-specific Gfh (Gre factor homologue) proteins – paralogs of universal Gre factors that stimulate RNA cleavage by RNAP – strongly promote transcriptional pausing and termination by Dra RNAP in the presence of Mn²⁺, suggesting their role in stress response [54,55].

Here, we analyse transcriptional properties of Dra RNAP on damaged DNA templates and demonstrate that various lesions can significantly affect the efficiency and fidelity of translesion synthesis by this RNAP. We show that factor Gfh1 can increase transcriptional pausing on damaged DNA templates and stimulate removal of stalled TECs by the Mfd translocase in the presence of Mn²⁺. Unexpectedly, GreA can also stimulate transcription stalling at the sites of lesions suggesting that various secondary channel factors may be involved in TCR in Dra cells under stress conditions.

Results

DNA lesions impair RNA extension by Dra RNAP

To reveal the effects of various types of DNA lesions on RNA synthesis by D. radiodurans RNAP, we analysed transcription of DNA templates containing several modifications commonly found in genomic DNA: 8-oxoguanine (8oxoG), O6-methylguanine (O6MeG), thymine glycol (TG), cyclobutane pyrimidine dimer (CPD, formed by two thymine residues), 1,N6-ethenoadenine (ϵA), and apurinic/apyrimidinic site (AP-site) (Supplementary Figure S1). We performed transcription in reconstituted TECs containing specific DNA lesions at a defined position downstream of the RNA 3′-end (Figure 1a, Supplementary Figure S2A). For most experiments, we used complete TECs containing upstream and downstream DNA duplexes and a transcription bubble with the 5′-labeled nascent RNA. The RNA 3′-end was located three nucleotides upstream of the lesion, which allowed to visualize RNAP stalling both before and after the lesion (Figure S2A). For some experiments, we also used minimal TECs containing downstream DNA duplex and a short RNA-DNA hybrid, in which the damaged template nucleotide was located immediately downstream of the RNA 3′-end (Figure S2B). Previous experiments demonstrated that minimal nucleic acid scaffolds of this structure form stable complexes with RNAP and mimic complete TECs [56].

Figure 1.

Translesion RNA synthesis by Dra RNAP and its modulation by GreA and Gfh1. (a) Scheme of the experiment. For TEC assembly, 5′-labeled 18/19 nt RNA (red) was annealed to the template DNA (tDNA) strand, followed by the addition of core RNAP and nontemplate DNA (ntDNA) strand. GreA or Gfh1 were added when indicated, and transcription was performed in the presence of all four NTPs. (b) Analysis of transcription products. The reactions were performed for 10”, 30”, 1ʹ, 20ʹ in the presence of 10 μM of NTPs at 30°C for the 8oxoG, 100 μM of NTPs at 30°C for the TG and O6meG templates, 100 μM of NTPs at 37°C for the CPD template. Positions of the starting RNA (18/19 nt), full-length run-off (RO) transcripts and the sites of lesions (asterisks) are indicated. (c) Quantification of the efficiencies of transcription stalling at the sites of DNA lesions. Means and standard deviations from 3 independent experiments are shown.

We first analysed the efficiency of translesion RNA synthesis on damaged DNA templates in reactions with all four NTPs. Transcription was performed in the presence of either Mg²⁺ or Mn²⁺ ions, because it was previously shown that Mn²⁺ has significant effects on RNAP activity and modulates the action of Dra-specific transcription factors [50,55]. The experiments were performed with complete TECs for the 8oxoG, O6meG, TG and CPD lesions, or with minimal TECs for the ϵA and AP lesions. Since various DNA lesions were expected to have different effects on transcription stalling, depending on the changes in DNA structure and complementary pairing, we used different NTP concentrations (from 10 μM to 1 mM) and different reaction temperatures (from 25 to 37°C) for various templates, as indicated in the figure legends. This corresponds to the temperature range of cell growth and RNAP activity in D. radiodurans [5,57].

No significant RNAP stalling was observed on control templates lacking damaged nucleotides (see Figures S3 and S4 for complete and minimal TECs). The presence of lesions in the template DNA strand significantly affected RNA extension by RNAP. In the presence of Mg²⁺, 8oxoG, O6meG and TG induced transient transcriptional pausing at a position preceding the damaged nucleotide, suggesting that these modifications inhibit nucleotide incorporation opposite the lesion (Figure 1b, upper panels). In the presence of Mn²⁺, additional stalling was observed directly at the site of the lesion, after nucleotide incorporation (Figure 1b, lower panels). Notably, Mn²⁺ decreased the overall efficiency of transcriptional stalling at 8oxoG, but significantly increased it for the TG lesion (Figure 1c, compare leftmost diagrams for the Mg²⁺ (grey) and Mn²⁺ (ochre) reactions), suggesting that Mn²⁺ may modulate translesion RNA synthesis by Dra RNAP under stress conditions.

CPD presented a strong obstacle for nucleotide incorporation, as evident from the highly efficient RNAP stalling before nucleotide incorporation opposite the first dimer thymine, and the very slow kinetics of RNA extension (Figure 1, the CPD template). At the same time, some full-length RNA synthesis was observed at long incubation times, indicating that CPD does not completely block RNAP translocation. The AP-site induced transcriptional stalling at two positions, before and opposite the lesion, as follows from inefficient extension of the starting 12 nt RNA transcript and one nucleotide longer 13 nt RNA in comparison with control template (Figure S4, the AP template). ϵA strongly reduced nucleotide incorporation opposite the lesion and almost completely blocked further RNA extension (Figure S4, the ϵA template).

In summary, it can be concluded that the presence of damaged template nucleotides can significantly affect the efficiency and processivity of transcription by Dra RNAP, and the strength of this effect depends on the type of the lesion.

DNA lesions compromise the fidelity of nucleotide incorporation by Dra RNAP

We then tested how various DNA lesions can affect the fidelity of nucleotide incorporation by Dra RNAP by incubating the TECs with each of the four NTPs in the presence of either Mg²⁺ or Mn²⁺ ions (Figure 2a). The experiments were performed with minimal TECs, in which the RNA 3′-end was located immediately upstream of the lesions, thus allowing analysis of single-nucleotide incorporation. In control experiments with undamaged templates, RNAP incorporated predominantly complementary nucleotides (Figure 2b). Significant nucleotide misincorporation was observed in control reactions corresponding to the CPD and ϵA templates, likely because of higher NTP concentrations and higher reaction temperature used in these experiments (see next paragraph) (Figure 2b). Indeed, nucleotide misincorporation with these control templates was strongly decreased at lower NTP concentrations (10 μM instead of 1 mM, Figure S5).

Figure 2.

Fidelity of NTP incorporation opposite DNA lesions by Dra RNAP. (a) Scheme of the experiment. Minimal TECs were assembled from 5′-labeled 11/12 nt RNA, tDNA and ntDNA oligonucleotides, followed by RNAP addition. (b) Single-nucleotide incorporation opposite various DNA lesions and corresponding undamaged TECs. The reactions with 8oxoG, O6meG, TG and AP-site were performed for 1ʹ at 25°C with 10 μM of indicated NTPs; the reactions with CPD and ϵA were performed for 1ʹ at 37°C with 1 mM of indicated NTPs. Positions of the starting 11 nt or 12 nt RNAs, extended RNAs corresponding to the sites of lesions (indicated with asterisk) and the products of cleavage (c) are indicated. The percent of RNA extension for each reaction (including the cleavage products) is shown below the gels.

Various DNA lesions had distinct effects on the fidelity of RNA synthesis by Dra RNAP. While TG did not change the specificity of nucleotide incorporation, 8oxoG and O6meG significantly affected RNAP fidelity (Figure 2b). Opposite O6meG, Dra RNAP incorporated almost exclusively UTP instead of CTP, both in the presence of Mg²⁺ or Mn²⁺. In the case of 8oxoG, strong incorporation of ATP was observed, along with correct CTP (with only minor incorporation of UTP, which was less efficient than in the case of the control undamaged template). Opposite the AP-site, RNAP incorporated mostly ATP, and also some UTP and GTP in the presence of Mn²⁺. For the CPD and ϵA templates, the reactions were performed with higher substrate concentrations and at higher temperature to compensate for the severe base-pairing defects caused by these modifications. Preferable incorporation of ATP and GTP was observed for the CPD template; UTP and CTP were also incorporated to some extent in the presence of Mn²⁺. Nucleotide incorporation opposite ϵA was very inefficient for all four NTPs in the presence of Mg²⁺, while low levels of activity were detected in the presence of Mn²⁺ (Figure 2b). Essentially no nucleotide incorporation was observed for the CPD and ϵA templates when the same experiments were repeated at lower NTP concentrations, in agreement with strong effects of these lesions on transcription (Figure S5, compare with Figure 1).

Furthermore, RNA cleavage was observed for most templates, especially in the presence of Mn²⁺, which likely occurred after single-nucleotide incorporation, resulting in the appearance of RNA products that were one-nucleotide shorter than the starting RNA (due to the 3′-terminal dinucleotide cleavage) (Figure 2b; the cleavage products are indicated with ‘C’). For all damaged templates, RNA cleavage could also be detected after nucleotide (mis)incorporation, and it was also more efficient in the presence of Mn²⁺.

Therefore, various DNA lesions impair correct nucleotide incorporation by Dra RNAP and decrease transcription fidelity. Mn²⁺ increases nucleotide misincorporation opposite lesions and also stimulates RNA cleavage by Dra RNAP in stalled transcription complexes.

GreA and Gfh1 inhibit translesion RNA synthesis by Dra RNAP

D. radiodurans encodes three members of the Gre-family transcription regulators, including the universal transcript cleavage factor GreA and lineage-specific GreA homologues, Gfh1 and Gfh2 (see Introduction). Gre factors are well-known to stimulate RNA cleavage by RNAP in backtracked transcription complexes, and to suppress transcriptional pauses associated with backtracking (e.g. [58,59]). We therefore expected that Dra GreA might stimulate translesion RNA synthesis by Dra RNAP, especially if RNAP stalling at the sites of lesions could be accompanied by backtracking.

The effects of GreA on Dra RNAP activity were analysed in complete TECs containing 8oxoG, O6meG, TG or CPD in the presence of all four NTPs (Figure 1). Unexpectedly, GreA promoted RNAP pausing at all four tested lesions and slowed down accumulation of full-length RNA products in the presence of either Mg²⁺ or Mn²⁺ (Figures 1b and c). Under most conditions, transcriptional pauses stimulated by GreA exactly corresponded to the sites of lesions, and no RNA cleavage products of shorter lengths were detected. In the case of the TG template, however, a shorter RNA product was observed in the presence of GreA, suggesting that it may appear after cleavage of the paused RNA during transcription of this template (Figure 1b, the TG template).

Both Gfh1 and Gfh2 were previously shown to stimulate transcriptional pausing and termination in the presence of Mn²⁺ ions, with Gfh1 showing stronger inhibitory effects on RNAP activity [55]. Since it was shown that Gfh1 could increase RNAP stalling at pause sites of various nature [54,55], we tested whether it could also inhibit translesion RNA synthesis by Dra RNAP. In agreement with previous reports, in the presence of Mg²⁺ Gfh1 had little effect on transcription (Figure 1b and c, Gfh1 reactions with Mg²⁺). In contrast, in the presence of Mn²⁺ Gfh1 enhanced RNAP stalling at the sites of DNA lesions. This effect was especially pronounced in the case of 8oxoG, and to a lesser extent, O6meG and TG templates (Figures 1b and c). The CPD lesion already blocked transcription almost completely even in the absence of any factors; however, both Gfh1 and GreA inhibited the weak readthrough RNA synthesis observed on the CPD template in their absence (Figure 1b).

In agreement with published data, in the presence of Mn²⁺ we also detected stimulation of transcriptional pausing by Gfh1 on control templates lacking damaged nucleotides, confirming that Gfh factors universally stimulate various pause types (Figure S3) [55]. GreA could also induce some pausing, especially at lower NTP concentrations. However, these pauses did not coincide with positions of TEC stalling observed on damaged DNA templates, and had significantly shorter half-lives (compare Figure 1 and Figure S3). Thus, both GreA and Gfh1 can specifically enhance Dra RNAP stalling at the sites of lesions, which might facilitate DNA damage recognition by the DNA repair factors in Dra (see below).

In contrast to Dra, E. coli encodes two Gre factors, GreA and GreB, but does not contain Gfh factors. To reveal whether E. coli GreA could stimulate RNAP stalling on damaged DNA templates similarly to Dra GreA, we performed experiments with the same DNA templates and purified RNAP and GreA from E. coli. E. coli GreA only weakly affected translesion RNA synthesis by E. coli RNAP (Figure 3A and B). In particular, no increase in RNAP stalling was observed for the 8oxoG template, whose transcription was most strongly inhibited by Dra GreA and Gfh1 (compare Figures 3 and Figure 1c). Therefore, the ability to stimulate RNAP stalling at DNA lesions may be a specific characteristic of Gre-family factors in Dra.

Figure 3.

The absence of strong effects of GreA on the efficiency of translesion transcription by E. coli RNAP. (a) Transcription was analysed in complete TECs under the same conditions as in Figure 1. The reactions were performed for 10”, 30”, 3ʹ, 20ʹ with 10 μM NTPs at 30°C for 8oxoG, with 100 μM NTPs at 30°C for O6meG and TG, and with 100 μM NTPs at 37°C for CPD templates. GreA was added to 5 μM where indicated. Positions of the starting 18 or 19 nt RNA, damaged nucleotide in the template DNA (asterisk) and full-length RNA (RO) are indicated. (b) The efficiencies of transcription stalling are shown for various DNA lesions (means and standard deviations from 3 independent experiments). Transcription was performed in complete TECs in the presence of Mg²⁺ or Mn²⁺ ions under the same conditions as in Figure 1.

Gfh1 stimulates disassembly of transcription complexes by the Mfd translocase

The Mfd translocase was mainly characterized in E. coli as a protein involved in DNA repair, which preferentially acts on stalled TECs, causes their dissociation by pushing RNAP forward, and recruits NER factors (see Introduction). The properties of the Mfd protein in Dra have remained unknown, despite extensive studies of the DNA repair mechanisms in this bacterium. We therefore purified Dra Mfd and characterized its in vitro activities.

Analysis of the effects of Dra Mfd on translesion transcription was performed in complete TECs containing upstream DNA duplex to allow Mfd binding behind RNAP. The reactions were performed in the presence of either Mg²⁺ or Mn²⁺ with addition of 1 mM dATP, which can be used by Mfd instead of ATP [22], to prevent ATP misincorporation by RNAP. To reveal Mfd-dependent dissociation of TECs during translesion transcription, the TECs were immobilized on magnetic streptavidin beads via 5′-biothynilated nontemplate DNA strand, which allowed to separate RNA molecules bound within the TEC (remaining in the pellet fraction, P) and dissociated into the solution (present in the supernatant fraction, S) (Figure 4a). To calculate the efficiency of RNA release for each lesion, the amount of RNA products of corresponding lengths found in the supernatant fraction was divided by the total amount of all RNAs (starting from the lesion site and longer) in the sum of supernatant and pellet fractions. Analysis of RNA products for the 8oxoG template is shown in Figure 4b. In this experiment we observed pausing at two sites, prior to and immediately after nucleotide incorporation opposite 8oxoG, and the second pause was stronger in the presence of Mn²⁺; we therefore used both paused RNA products for calculation. Quantification of TEC dissociation for all analysed TECs is presented in Figures 4c and d.

Figure 4.

Effects of Mfd and Gfh1 on TEC disassembly during translesion transcription. (a) Scheme of the experiment. Complete TECs containing 5′-biotinylated ntDNA strand were assembled in the same way as in Figure 1a and immobilized on streptavidin magnetic beads in transcription buffers containing 10 mM MgCl₂ or MnCl₂, followed by the addition of Gfh1 and/or Mfd, 1 mM dATP and 10 μM NTP substrates. Transcription was performed for 20 minutes at 30°C, and RNA products from the pellet (P) and supernatant (S) fractions were analysed by PAGE. (b) Analysis of RNA products synthesized in the 8oxoG TEC. Positions of the starting RNA (18 nt), damaged nucleotide (asterisk) and full-length product (RO) are indicated. (c and d) Quantification of RNA release at the sites of lesions in the presence of Mg²⁺ (c) and Mn²⁺ (d).

It was shown that in the absence of factors, only full-length RNA was released into the solution with either Mg²⁺ or Mn²⁺, while RNA products at the sites of lesions remained bound to the beads and therefore corresponded to stalled TECs. Mfd induced almost complete dissociation of the stalled TECs in the presence of Mg²⁺, with most RNA products released into the solution (Figure 4b). Since only a fraction of all released RNAs corresponded to the site of lesion (20–21 nt RNAs), with longer read-through products, the calculated efficiency of RNA release at the site of lesion varied between ~20% for 8oxoG and TG and 55–60% for O6meG and CPD (Figure 4c, the Mfd reactions). Interestingly, the efficiency of TEC dissociation was significantly decreased in the presence of Mn²⁺ (10–25% RNA release at the site of lesion for the four TECs, Figures 4b and d), suggesting that accumulation of Mn²⁺ ions in Dra cells may compromise the action of Mfd under stress conditions.

We further tested the effects of Gfh1 on TEC dissociation. In the presence of Mg²⁺, Gfh1 had no effect on the efficiency of RNAP pausing or RNA release either in the absence or in the presence of Mfd (Figures 4b and c, the Gfh1 reactions). On the contrary, in Mn²⁺-containing buffer Gfh1 significantly increased the observed efficiency of RNAP pausing (Figure 4b, the Gfh1 reaction with Mn²⁺) and also enhanced TEC disassembly by Mfd (Figures 4b and 4c, the Mfd/Gfh1 reaction with Mn²⁺). In particular, for the 8oxoG template the calculated percentage of RNA release at the site of lesion was increased more than 2-fold, from ~10% to >20% under these conditions (Figure 4d, blue bars). Similarly, the efficiency of RNA release was significantly increased in comparison with the Mfd-only reactions for the O6meG and TG templates (Figure 4d, orange and gray). In the case of the CPD template, the efficiency of RNAP stalling and RNA release was already high in the absence of Gfh1 and was not further increased in its presence (Figure 4d, ochre). Therefore, Gfh1 can increase the efficiency of TEC disassembly by Mfd in the presence of Mn²⁺ ions, thus compensating for the decrease in the Mfd activity under these conditions.

Discussion

Dra is an extremotolerant bacterium highly resistant to various types of DNA damaging agents, suggesting that its DNA replication, repair and transcription machineries have evolved in a unique way to deal with damaged DNA [1,3–5]. While many previous studies have addressed the mechanisms of DNA repair and gene regulation during stress response in Dra [45–49], none of them has studied the properties of Dra RNAP during transcription of damaged DNA. Here, we analysed the activity of Dra RNAP on damaged DNA templates and showed that the outcome of transcription on such templates greatly depends on the nature of a lesion and the action of regulatory factors.

Contrary to expectations, our analysis did not reveal any features of Dra RNAP that would enable it to transcribe damaged DNA more effectively or accurately in comparison with previously studied RNAPs. In particular, the Dra RNAP activity was greatly inhibited by the bulky CPD lesion and by ϵA, which disrupted base-pairing [31–34,41,42]. The AP-site also strongly impaired the activity of Dra RNAP, in agreement with previous reports on its effects on transcription by RNAPs from all three domains of life [40,41,60–63]. Similarly to other RNAPs [36,38,39,43,44,61,64,65], 8oxoG and O6meG induced transient TEC pausing and changed the specificity of nucleotide incorporation by Dra RNAP, resulting in the insertion of AMP opposite 8oxoG and UMP opposite O6meG.

At the same time, Dra RNAP activity could be significantly modulated by Mn²⁺ ions and by Mn²⁺-dependent Gfh factors, which might have important physiological consequences. Mn²⁺ stimulated nucleotide misincorporation on various damaged templates and also on control templates suggesting that it may compromise the fidelity of RNA synthesis by Dra RNAP under stress conditions. On the other hand, Mn²⁺ also enhanced intrinsic RNA cleavage by Dra RNAP on undamaged templates and opposite DNA lesions (ref [50]. and this study), possibly promoting transcriptional proofreading. Mn²⁺ also increased the efficiency of RNAP stalling at the TG lesion suggesting that its recognition is stimulated under stress conditions. Importantly, regulation by Mn²⁺ ions is characteristic for many stress-related Dra proteins, including all three Dra DNA polymerases, PolA [66], PolX [67], and the α subunit of Pol III [68], likely facilitating replication of damaged DNA.

We observed that Dra-specific factor Gfh1 can stimulate TEC stalling on damaged DNA templates (Figure 1 and Figure S2). Similarly to previously reported effects of Gfh factors on transcriptional pausing and termination by Dra RNAP [54,55], this effect strictly depended on the presence of Mn²⁺ ions, whose coordination in the RNAP active site might be changed by Gfh [55,69]. Surprisingly, Dra GreA also enhanced transcriptional pausing at the sites of DNA lesions, both in the presence of Mg²⁺ or Mn²⁺ (Figure 1). To our knowledge, a possible role of GreA in translesion RNA synthesis by bacterial RNAP has not been previously studied. At the same time, eukaryotic RNA cleavage factor TFIIS was tested on several damaged DNA templates and facilitated translesion transcription for 8oxoG but not for other lesions [37,70]. Since TECs stalled at the sites of lesions can perform RNA cleavage (Figure 2), GreA might increase TEC stalling by stimulating RNAP backtracking and reiterative RNA cleavage by these complexes, which would compete with efficient RNA extension. Indeed, a recent study of yeast RNAP I demonstrated that it cannot bypass the CPD lesion because of high rates of intrinsic RNA cleavage, a competing reaction with nucleotide addition [71]. At the same time, the absence of Gre-induced RNA cleavage products for most TECs, except for the TG template (Figure 1), suggests that both GreA and Gfh1 might instead directly prevent RNA extension. In support of this hypothesis, structural studies of GreA and Gfh1 in complex with Thermus thermophilus RNAP demonstrated that they stabilize RNAP in an inactive ratcheted conformation incompatible with nucleotide addition [72,73]. Both Gfh1 and GreA could also block access of NTPs to the active site of RNAP by their N-terminal coiled-coil domain directly interfering with nucleotide binding [69,72,73]. Since the concentration of both factors in our experiments (5 μM) was comparable to or higher than their reported affinities to RNAP [53,69,74], they likely formed stoichiometric complexes with the TEC resulting in strong inhibition of transcription.

In comparison, E. coli GreA had much weaker effects on translesion RNA synthesis by E. coli RNAP (Figure 3). Similarly, E. coli GreA had no strong effects on the rate of RNA elongation by E. coli RNAP (e.g. [69]). Recent single-molecule analysis of GreB interactions with the E. coli TEC suggested that it cannot stably bind actively transcribing RNAP, and thus avoids blocking transcription [75]. Thus, the observed stimulation of RNAP pausing by Gre-family factors may be a specific property of the transcriptional machinery in Dra.

GreA/Gfh1-dependent stalling of the TECs may make them a target for more efficient recognition by the TCR factors, including the Mfd translocase. Previous studies demonstrated that E. coli Mfd can recognize paused or stalled TECs and displace them from the DNA template, further recruiting NER factors and translocating with them in search for DNA lesions [22,24–26,76]. Mfd can also likely prevent RNAP backtracking and reactivate backtracked complexes [27]. However, Mfd by itself has no strong effects the efficiency of RNAP stalling at the sites of DNA lesions (ref [24]. and our unpublished observations). In our experimental system, Mfd successfully displaced TECs formed by Dra RNAP from the sites of lesions in the presence of Mg²⁺, but the efficiency of its action was decreased in reactions containing Mn²⁺.

Strikingly, Gfh1 promoted TEC dissociation by Mfd in the presence of Mn²⁺, thus compensating for the decrease in Mfd activity in these conditions (Figure 4). We propose that the Gfh1-dependent increase in TEC dissociation is primarily explained by an increase in RNAP stalling at the sites of lesions, thus giving Mfd more time to act on such complexes. Indeed, Gfh1 also stimulated dissociation of minor TEC fractions paused at several other template positions (Figure 4). Alternatively, Gfh1 binding might also stabilize a specific RNAP conformation more sensitive to the Mfd action, a hypothesis that deserves further scrutiny.

Taken together, these results suggest a potential role of Gre-family factors in transcription regulation and its coordination with DNA repair under stress conditions. Gfh factors may decrease general DNA occupancy by RNAP by inhibiting the initiation and stimulating the termination steps of transcription [54,55]. During translesion synthesis, Gfh1 stimulates RNAP stalling and facilitates Mfd action, likely promoting DNA repair. GreA can also increase TEC stalling, possibly making transcription complexes more accessible to TCR. Importantly, expression of both GreA (gene DR1162) and Gfh1 (gene DR1970) in Dra cells is increased after irradiation suggesting their role in stress resistance [47,77]. It remains to be tested whether other Dra-specific or universal factors, including the UvrD helicase [28,29], can also act on transcription complexes of Dra RNAP depending on the presence of lesions. Further experiments are required to establish whether GreA and Gfh proteins may have any effect of the sensitivity of stalled TECs to these factors, and to reveal their possible functions in DNA damage recognition and repair in Dra cells in vivo.

Materials and methods

Proteins and nucleic acids

Core Dra RNAP was purified from E. coli BL21(DE3) cells expressing all four core RNAP subunits from the pET-28-rpoACBZ-Dra plasmid using Polymin P precipitation, Heparin-Sepharose, Ni²⁺-affinity and MonoQ anionic-exchange chromatography steps as described [78]. Gfh1 and GreA were expressed from corresponding pET-28-based plasmids and purified by Polymin P precipitation, Ni²⁺-affinity and Q anionic-exchange chromatography [50,55]. The Dra Mfd gene was amplified from genomic DNA and cloned into pET-28. The Mfd protein was expressed in E. coli BL21(DE3) at 16 °C with the addition of 0.2 mM IPTG. The cells were collected by centrifugation and disrupted in a French press in the lysis buffer (40 mM Tris-HCl pH 8.0, 500 mM NaCl, 0.1 mM DTT, 5% glycerol and 1 mM PMSF). After centrifugation, the cleared lysate was loaded onto a HiTrap column charged with Ni²⁺ in the presence of 10 mM imidazole, washed with buffer A (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 5% glycerol) containing 40 mM imidazole and eluted with the same buffer containing 100 mM imidazole. The fractions with the Mfd protein were dialysed against buffer B (40 mM Tris-HCl pH 8.0, 0.5 mM EDTA, 100 mM NaCl, 0.1 mM DTT, 5% glycerol) and loaded onto a Mono Q column equilibrated with buffer B. The proteins were eluted by a NaCl gradient in the same buffer. Mfd was eluted at 280 mM NaCl. Glycerol was added to 50% and the protein was stored at −20 °C or −80 °C (for long-term storage). DNA and RNA oligonucleotides used in the assays were purchased from DNA Synthesis, Syntol, Evrogen (Moscow, Russia) and TriLink BioTechnologies (San Diego, CA).

In vitro transcription

Transcription assays were performed in two types of reconstituted TECs. Minimal TECs were assembled from short synthetic DNA and RNA oligonucleotides (see Figure S1B for oligonucleotide sequences) and core RNAP. The 5′-P³²-labeled RNA transcript (0.5 μM final concentration) was mixed with the template (1 μM) and nontemplate (5 μM) DNA oligonucleotides in transcription buffer containing 40 mM Tris-HCl, pH 7.9, 40 mM NaCl and 10 mM MgCl₂ or MnCl₂, heated to 65°C, and cooled down to 20°C at 1°/min. For the AP-site, the template was assembled for 10 minutes at 37°C. Core RNAP (25 nM) was mixed with the annealed oligonucleotides (10 nM final RNA concentration) and incubated for 10 minutes at 30°C. The samples were transferred to the required temperature and NTPs were added as described in the figure legends.

Complete TECs were assembled from 70(71) nt DNA and 18(19) nt RNA oligonucleotides (Figure S1A). The 5′-P³²-labeled RNA oligonucleotide (0.5 μM final concentration) was mixed with the template DNA oligonucleotide (1 μM) in buffer containing 40 mM Tris-HCl, pH 7.9 and 40 mM NaCl, heated to 65°C, and cooled down to 20°C at 1°/min. Core RNAP (60 nM) was mixed with the annealed oligonucleotides (6 nM final RNA concentration) and incubated for 20 minutes at 30°C. The nontemplate DNA oligonucleotide was added (120 nM), the samples were incubated at 30°C for another 20ʹ and then transferred to the required temperature. GreA or Gfh1 were added to 5 μM. NTPs were added as described in the figure legends along with 10 mM of MgCl₂ or MnCl₂ and heparin (15 μg/ml). The reactions were stopped after various time intervals, as indicated in the figures, and the RNA products were resolved by 18% denaturing PAGE, followed by phosphorimaging. The efficiencies of RNAP stalling were calculated as the ratio of RNA products corresponding to the site of DNA lesion to the sum of these products and read-through RNA transcripts.

Experiments with immobilized TECs were performed with 5′-biotinylated nontemplate DNA oligonucleotides. After assembly, the TECs were incubated with magnetic streptavidin beads in 20 μl of the transcription buffer at room temperature for 10ʹ with shaking, followed by the addition of Gfh 1 (5 μM) and/or Mfd (2 μM) and NTPs at 30°C (no heparin was added in this case). dATP was added at 1 mM final concentration to all samples, including reactions with Gfh1 and control reactions without the addition of any factors. After 20ʹ, a 10 μl aliquot of the supernatant fraction was taken, and the remaining samples were diluted with 1 ml of a buffer solution lacking catalytic ions. After washing, the solution was completely removed, the pellet fraction was resuspended in 20 μl of the same buffer, and 10 μl were used for electrophoretic analysis. The percentage of RNA dissociated from the TECs at the site of a lesion was calculated as the ratio of RNA products of corresponding lengths in the supernatant fraction to the sum of all RNA products in both the pellet and supernatant fractions. The Mann-Whitney U-test was used for statistical analysis.

Funding Statement

This work was supported by the Russian Science Foundation (grant 17-14-01393; analysis of translesion transcription by Dra RNAP) and Russian Foundation for Basic Research (grant 18-34-00905; analysis of DNA repair factors from Dra).

Acknowledgments

We thank Irina Artsimovitch for the gift of CPD templates, Artem Ignatov for help in figure preparation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

suppelmental data for this article can be accessed here.