Synchrotron Radiation Circular Dichroism, a New Tool to Probe Interactions between Nucleic Acids Involved in the Control of ColE1-Type Plasmid Replication

Featured Application: Our results indicate previously unknown macromolecular interplays in the regulation of ColE1-like plasmid DNA replication, which should be important in further studies on combating bacterial antibiotic resistance through controlling the propagation and spreading of plasmids as carriers of antibiotic resistance genes. New insights on this system are brought by synchrotron radiation-based circular dichroism (SRCD), which gives access to an extended spectral range down to the vacuum UV (vUV) range. Only the synchrotron light source provides intensities of the beam below 190 nm, which allows a precise determination of the nucleic acid conformation. Abstract: Hfq is a bacterial master regulator which promotes the pairing of nucleic acids. Due to the high molecular weight of the complexes formed between nucleic acids and the amyloid form of the protein, it is difﬁcult to analyze solely by a gel shift assay the complexes formed, as they all migrate at the same position in the gel. In addition, precise kinetics measurements are not possible using a gel shift assay. Here, we used a synchrotron-based biophysical approach, synchrotron radiation circular dichroism (SRCD), to probe the interaction of the Escherichia coli Hfq C-terminal amyloid region with nucleic acids involved in the control of ColE1-like plasmid replication. We observed that this C-terminal region of Hfq has an unexpected and signiﬁcant effect on the annealing of nucleic acids involved in this process and, more importantly, on their alignment. Functional consequences of this newly discovered property of the Hfq amyloid region are discussed in terms of the biological signiﬁcance of Hfq in the ColE1-type plasmid replication process and antibiotic resistance.


Introduction
Antibiotic resistance of bacteria has become one of the major medical problems in recent years [1]. The appearance of many bacterial strains resistant to one or many antibiotics makes this crucial therapy used to combat various infectious diseases less and less effective, resulting in alarming predictions on the possible lack of cures in such cases in the near future [2]. The spreading of antibiotic resistance is caused mainly by the overuse of these therapeutics and the selection of resistant strains. In fact, antibiotic-resistant bacteria occur naturally in the environment, predominantly because they bear genes coding for proteins  figure) is the pre-primer, synthesized by E. coli RNA polymerase (RNAP, oval), which hybridizes with the plasmid origin (Ori) during transcription (lower part of the figure). This hybridization, strengthened in the C-rich region (due to the formation of C-G pairs in the RNA-DNA hybrid), allows RNase H (fragmentary oval) to cleave the RNA strand hybridized with DNA, thus forming a primer that is used by DNA polymerase I (Pol I, thick arrow) to initiate plasmid DNA synthesis in the replication process (bottom part of the figure). However, ColE1-type plasmid replication initiation is negatively regulated by RNA I, another plasmid-encoded transcript, which interacts with RNA II through stem-loop (SL) structures called SL1, SL2, and SL3 (upper part of the figure). Effective RNA I-RNA II interactions prevent the formation of the RNA II-DNA hybrid despite the appearance of the fourth stem-loop structure in RNA II, thus inhibiting plasmid DNA replication initiation.

Hfq-CTR Peptide and Protein
Hfq-CTR peptide, corresponding to the amyloid CTR domain of Hfq (residues 64 to 102), was chemically synthesized (Proteogenix, France). The sequence of Hfq-CTR is SRPVSHHSNNAGGGTSSNYHHGSSAQNTSAQQDSEETE [25]. Before use, the Hfq-CTR peptide was reconstituted in water at 20 mg/mL. We determined previously that pH 5 was the most appropriate for the formation of the complex with nucleic acids [24]. We also chose to avoid the addition of salts (except those already present in DNA, protein and peptide solutions) to allow better investigation of the conditions in deep UV [34,35]. Previously, we ensured that the presence of salts (NaCl 50 mM) does not significantly change the activity of Hfq-CTR [24].

Designing Oligonucleotides for Analyses
The detailed nucleotide sequence of the replication origin region of ColE1 plasmid DNA is shown in Figure 2, indicating the crucial regions and corresponding transcripts involved in the regulation of replication initiation. Oligonucleotides for detailed studies on interactions between RNA I and RNA II, RNA II and DNA at the origin region and DNA strands, in the presence and absence of CTR, were designed considering crucial fragments known to be involved in these processes. Thus, the synthetic fragments of RNA I and RNA II comprised the A stem regions, which contain the stem-loop 1 (SL1) sequence (see Figure 1). The minimal structure that involves RNA I-RNA II interactions was a single loop that can serve as the simplest model for interactions occurring between ColE1 regulatory RNAs. We have employed here such a model in the highly defined in vitro system, which allowed us to make the most unequivocal conclusions. RNA II and DNA fragments involved in the formation of the hybrid and the RNA/DNA switch during replication initiation contained sequences allowing us to monitor crucial interactions between strands of these nucleic acids, which were also analyzed. The sequences of the oligonucleotides used, corresponding to the ColE1 genome fragments, are indicated in Figure 2.

Synchrotron Radiation Circular Dichroism (SRCD)
SRCD measurements were carried out on the DISCO beamline at the SOLEIL synchrotron (proposal 20210003). The synchrotron light source provides intensities of the beam below 190 nm. Such analyses are not trivial due to the strong absorbance of light below 200 nm by air and water. Conventional spectrometers scan with noise below 200 nm and not at all below 190 nm, while references for NA have the most informative region in the 170-220 nm range. Samples (~4 µL) were loaded into a CaF2 circular cell of 12.5 µm pathlength [36]. Spectral acquisitions of 1 nm steps at 1.2 s integration time were recorded in triplicates between 320 and 180 nm. (+)-camphor-10-sulfonic acid (CSA) was used to calibrate amplitudes and wavelength positions of the experiment.
Note that the beam dimensions have been optimized and set to 4 mm × 4 mm at 1 nm bandwidth in order to reduce the eventual radiation damage of the samples [37,38]. We The precise RNA/DNA sequences used were as follows (with colors corresponding to those used in Figure 2 Before use, oligonucleotides in water were heated at 90 • C for 3 min and then slowly cooled down at 20 • C to allow proper folding. This procedure corresponds to the one used previously for the accumulation of the CD reference dataset (https://pcddb.cryst.bbk.ac. uk/, accessed on 6 February 2022). When the complex is analyzed in the presence of salts or far-UV absorbing buffers, the spectral bandwidth accessible is limited, reducing the spectral information obtained [35]. Annealing was checked on a gel before use, and we saw no effect of salts (Na + , Mg 2+ ) on the formation of the dsDNA duplexes, as the concentration used for its preparation was high (about 1 mM). Thus, we chose to use water in order to go further in far UV for SRCD analysis to get a broader spectral band (170-320 nm). Duplexes of oligonucleotides (i.e., RNA I:RNA II, etc.) were formed using the same protocol after stoichiometric addition of the two oligonucleotides in the same tube. Abbreviations ss and ds refer to single-and double-stranded NA, respectively.

Synchrotron Radiation Circular Dichroism (SRCD)
SRCD measurements were carried out on the DISCO beamline at the SOLEIL synchrotron (proposal 20210003). The synchrotron light source provides intensities of the beam below 190 nm. Such analyses are not trivial due to the strong absorbance of light below 200 nm by air and water. Conventional spectrometers scan with noise below 200 nm and not at all below 190 nm, while references for NA have the most informative region in the 170-220 nm range. Samples (~4 µL) were loaded into a CaF 2 circular cell of 12.5 µm pathlength [36]. Spectral acquisitions of 1 nm steps at 1.2 s integration time were recorded in triplicates between 320 and 180 nm. (+)-camphor-10-sulfonic acid (CSA) was used to calibrate amplitudes and wavelength positions of the experiment.
Note that the beam dimensions have been optimized and set to 4 mm × 4 mm at 1 nm bandwidth in order to reduce the eventual radiation damage of the samples [37,38]. We have also shown that using this setup, no spectral decrease owing to UV radiation could be observed after 50 consecutive scans (~4 h) on Hfq-CTR (Supplementary Figure S1). Therefore, no radiation damage can be assumed. Radiation damage may also induce degradation of the nucleic acid. In a parallel study using primer extension on polynucleotides subsequent to a full SRCD spectral data collection, no cyclobutane dimers had been formed between adjacent pyrimidines on the same DNA strand. This suggests that little if any photochemical modifications due to UV exposure had occurred on the DNA during the accumulation of the spectra.
Data analyses (averaging, baseline subtraction, smoothing, scaling and spectral summations) were carried out with CDtoolX [39]. Spectra are presented in units of mdeg versus nm, maintaining the same molar ratios for all presented samples. Due to the origin of absorption, spectra of mixed samples (polynucleotides + peptide) could not be standardized to ∆ε.
Two types of experiments have been performed: the effect of Hfq or CTR were tested either on individual RNAs/DNAs or on duplexes such as RNA I:RNA II. As they are not negligible, the contribution of the added Hfq or CTR peptide was subtracted from the SRCD spectra of the complex (the SRCD spectra of the free peptide and protein are presented in Supplementary Figure S2). For all experiments, RNA concentrations were in excess compared to Hfq-CTR. The Hfq-CTR (monomeric) concentration was 0.4 mM. RNAI_Astem and RNAII_Astem concentrations were 13 mM (in bp), the RNAII_C-rich_stretch concentration was 11 mM, the DNA_C-stretch_ssColE1 concentration was 15 mM, and the DNA_ColE1orifor concentration was 14 mM (RNA and DNA concentrations were normalized, taking into account the respective size of fragments).
The effect of Hfq-CTR was analyzed at 15 • C to minimize spontaneous annealing between nucleic acids that could occur [24]. For kinetic measurements, an apparent catalytic kinetic constant k cat app was determined during initial rate conditions. This constant was For melting curves, triplet SRCD spectra were acquired every 3 • C between 15 • C and 96 • C. Averaged SRCD values of the maximum of the peaks around 180 and 270 nm are presented as a function of temperature to measure the melting point (Tm). A Boltzmann sigmoid equation, which assumes a two-state model, has been used for fitting of melting curves: y = Bottom + (top − bottom)/(1 + e((Tm − x)/slope)).

Synchrotron Radiation Linear Dichroism (SRLD)
For SRLD measurements, the settings of the modulator (0.608xλ) and acquisition (2xf) were adapted as described previously [40]. A total of 4 µL of samples were loaded into the same CaF 2 circular cell that was used for SRCD. Doublet spectra were recorded and averaged every 90 • over 360 • , which resulted in 9 spectra including a repetition of the 0/360 • angle for reproducibility verification. LD results in strong spectral amplitudes, which change depending on the angle of the cell. The magnitude of LD is typically higher than the CD of the same molecule or complex. This allows the alignment of DNA/RNA or peptide:DNA/RNA complexes to be observed [41].

Hfq-CTR Does Not Melt ColE1 Origin
The effect of Hfq-CTR on ColE1 dsDNA origin was first analyzed. Melting of dsDNA is mandatory for the replication process, as it is not achieved by the DNA polymerase. SRCD makes it possible to follow the transition from double-to single-strand DNA, as well as the reverse transition [24,42]. Indeed, the spectral region between 320 and 170 nm contains several electronic transitions of interest. Annealing or conversely melting activities of a protein can be observed by the increase or decrease in amplitudes at~185 and 260 nm, respectively. Around 270 nm, the positive CD signal shows base-pairing and base-stacking [42][43][44]. On the other hand, the positive band at 185 nm is indicative of the formation of double-stranded right-handed NA molecules (here in B-form, see below). A decrease in the amplitudes of these peaks thus signifies dsDNA melting of base pairs and the right-handed dsDNA structure.
As seen in Figure 3, no effect of Hfq-CTR was observed on melting the dsDNA ColE1 origin. The same result was observed for full-length Hfq, even for melting kinetics recorded for hours or at 37 • C. This indicates that Hfq-CTR does not melt ColE1 origin dsDNA.

Hfq-CTR Stabilizes ColE1 Origin dsDNA
To investigate the effect of CTR on ColE1 origin dsDNA stability, we analyzed melting curves of the ColE1 origin duplex. As shown in Figure 4, we observe that the T m in the presence of CTR was 8 • C higher than that measured for dsDNA alone. We thus show that Hfq-CTR does not melt ColE1 origin, but rather, it significantly stabilizes the dsDNA duplex in this region of replication priming.

Hfq-CTR Aligns ColE1 Origin dsDNA
Next, as the formation of NA bound to CTR could result in the alignment of the complex, we analysed this possible alignment using synchrotron radiation. Indeed, we previously observed such activity for the Hfq-CTR amyloid region with a model of DNA (dA 59 :dT 59 ) [35]. Synchrotron radiation linear dichroism (SRLD) provides information on the orientation of chromophores in an orientable molecule such as DNA [46]. Here, molecular alignment may be induced by the protein, and the axes orthogonal to the direction of the alignment will not be equivalent. This would result in different SRCD and SRLD spectra when rotating the cell every 90 • . As shown in Figure

Hfq-CTR Stabilizes ColE1 Origin dsDNA
To investigate the effect of CTR on ColE1 origin dsDNA stability, we analyzed melting curves of the ColE1 origin duplex. As shown in Figure 4, we observe that the Tm in the presence of CTR was 8 °C higher than that measured for dsDNA alone. We thus show that Hfq-CTR does not melt ColE1 origin, but rather, it significantly stabilizes the dsDNA duplex in this region of replication priming. The effect of Hfq-CTR on RNA annealing was analyzed. Indeed, the best-known Hfq function is to promote RNA annealing [47], and Hfq was strongly suspected to play a role in RNA I-RNA II complex formation during ColE1-type plasmid DNA replication [17]. However, due to the high molecular weight of the complex with the amyloid form of CTR that stays on the top of the gel, it is difficult to analyse this process solely with EMSA and even more difficult to make precise kinetics measurements. As explained previously, an increase of the amplitudes of the SRCD peaks at 262 and 183 nm would signify the formation of base pairs and of right-handed double-stranded RNA, respectively. Thus, SRCD analysis of Hfq-CTR allows measurement at 262 nm of an apparent catalytic kinetic constant k cat app of 0.015 mdeg.M −1 .min −1 and, at 183 nm, an apparent catalytic kinetic constant k cat app of 0.088 mdeg.M −1 .min −1 ( Figure 6). This activity is similar to that reported previously for DsrA small RNA annealing to rpoS mRNA [24]. Appl. Sci. 2022, 12, x FOR PEER REVIEW 8 of 19

Hfq-CTR Aligns ColE1 Origin dsDNA
Next, as the formation of NA bound to CTR could result in the alignment of the complex, we analysed this possible alignment using synchrotron radiation. Indeed, we previously observed such activity for the Hfq-CTR amyloid region with a model of DNA (dA59:dT59) [35]. Synchrotron radiation linear dichroism (SRLD) provides information on the orientation of chromophores in an orientable molecule such as DNA [46]. Here, molecular alignment may be induced by the protein, and the axes orthogonal to the direction of the alignment will not be equivalent. This would result in different SRCD and SRLD spectra when rotating the cell every 90°. As shown in Figure 5, ColE1 dsDNA becomes aligned with the Hfq-CTR, while dsDNA alone do not align in the absence of the protein [35].

Hfq-CTR Strongly Aligns RNA I-RNA II Duplex
The effect of Hfq-CTR on RNA I-RNA II alignment was also analyzed. Indeed, Hfq has been shown to promote DNA alignment [35], but not that of RNA, especially short RNA pieces such as for RNA I-RNA II stem-loops. Here, we found that Hfq-CTR strongly aligns RNA I and RNA II, while these RNAs alone only slightly align (Figure 7; 4  Next, the effect of Hfq-CTR on the stability of the RNA I-RNA II complex formed was evaluated using meting curves. SRCD analysis was performed on the RNA I-RNA II complex in the absence and presence of Hfq-CTR. As described under Methods, SRCD spectra were acquired every 3 • C between 15 • C and 96 • C. As seen in Figure 8A, from the increase in CD signal at 190 nm, the RNA I-RNA II complex was considerably more stable in the presence of Hfq-CTR, especially at high temperatures (>80 • C). This result was unexpected as usually high temperatures (>80 • C) melt NA structures and do not stabilize it, as is the case for the RNA I-RNA II complex without CTR ( Figure 8B). Three independent repetitions of this experiment clearly demonstrate that the protein stabilizes the complex between tested RNAs at high temperatures. Without Hfq-CTR, this effect was not observed. Indeed, this observation is unique and needs further analysis leading to the understanding of the thermodynamics involved. The effect of Hfq-CTR on RNA annealing was analyzed. Indeed, the best-known Hfq function is to promote RNA annealing [47], and Hfq was strongly suspected to play a role in RNA I-RNA II complex formation during ColE1-type plasmid DNA replication [17]. However, due to the high molecular weight of the complex with the amyloid form of CTR that stays on the top of the gel, it is difficult to analyse this process solely with EMSA and even more difficult to make precise kinetics measurements. As explained previously, an increase of the amplitudes of the SRCD peaks at 262 and 183 nm would signify the formation of base pairs and of right-handed double-stranded RNA, respectively. Thus, SRCD analysis of Hfq-CTR allows measurement at 262 nm of an apparent catalytic kinetic constant kcat app of 0.015 mdeg.M −1 .min −1 and, at 183 nm, an apparent catalytic kinetic constant kcat app of 0.088 mdeg.M −1 .min −1 (Figure 6). This activity is similar to that reported previously for DsrA small RNA annealing to rpoS mRNA [24].

Hfq-CTR Stabilizes RNA II Secondary Structure, but Not That of RNA I
As Hfq appears to have a stabilizing effect on some RNAs [49], we tested the effect of Hfq-CTR on RNA I and RNA II secondary structures. We observed different effects on the stabilization of RNA I and RNA II SL1 structures. Specifically, we observed that SL1 of both RNAs did not form alone and that Hfq-CTR significantly stabilized the RNA II SL1 structure ( Figure 9A) but did not stabilize the RNA I SL1 structure ( Figure 9B). This could be observed by a significant increase in peaks at 180 and 270 nm. This effect is nearly instantaneous and occurs in less than 2 min.
Note that the SRCD increase at 183 nm (and to a minor extent at 262 nm) in Figure 6 could be due to CTR-induced RNA II folding. In this case, the increase would only bẽ 5 mdeg as in Figure 9A; nevertheless, it is more than 10 mdeg for RNAI and RNAII annealing by the CTR. Therefore, we can conclude that the effect observed in Figure 6 is indicative for RNAI:RNAII duplex formation and not only from RNAII folding by the CTR. Here an initial increase in A-form helix formation (183 nm) is followed by a slower gradual formation of base pairing (262 nm). Kinetic constants kcat app were determined during initial rate conditions. Note that spontaneous RNA annealing also occurs without the protein but to a lesser extent, and it was subtracted.

Hfq-CTR Strongly Aligns RNA I-RNA II Duplex
The effect of Hfq-CTR on RNA I-RNA II alignment was also analyzed. Indeed, Hfq has been shown to promote DNA alignment [35], but not that of RNA, especially short RNA pieces such as for RNA I-RNA II stem-loops. Here, we found that Hfq-CTR strongly aligns RNA I and RNA II, while these RNAs alone only slightly align (Figure 7; 4 vs. 30 mdeg) [48].

Hfq-CTR Strongly Stabilizes the RNA I-RNA II Duplex, in Particular at High Temperature
Next, the effect of Hfq-CTR on the stability of the RNA I-RNA II complex formed was evaluated using meting curves. SRCD analysis was performed on the RNA I-RNA II complex in the absence and presence of Hfq-CTR. As described under Methods, SRCD spectra were acquired every 3 °C between 15 °C and 96 °C. As seen in Figure 8A, from the increase in CD signal at 190 nm, the RNA I-RNA II complex was considerably more stable in the presence of Hfq-CTR, especially at high temperatures (>80 °C). This result was unexpected as usually high temperatures (>80 °C) melt NA structures and do not stabilize it, as is the case for the RNA I-RNA II complex without CTR ( Figure 8B). Three independent repetitions of this experiment clearly demonstrate that the protein stabilizes the complex between tested RNAs at high temperatures. Without Hfq-CTR, this effect was not observed. Indeed, this observation is unique and needs further analysis leading to the understanding of the thermodynamics involved. Note that spontaneous RNA annealing also occurs without the protein but to a lesser extent, and it was subtracted.

Hfq-CTR Stabilizes the C-Stretch Region of ColE1 Plasmid DNA
Finally, we also analyzed the effects of CTR on the dsDNA C-stretch region of the ColE1 plasmid. First, we checked, as we did for the ColE1 origin sequence, if CTR was able to melt dsDNA ( Figure 10). We observed that CTR stabilized the structure of the dsDNA region as the signal at~185 nm increased slightly. SRCD analysis allows measurement at 185 nm of an apparent catalytic kinetic constant k cat app of 0.069 mdeg.M −1 .min −1 . To confirm this effect, we analyzed the melting curves of the duplex ColE1 C-stretch. As shown in Figure 11, we observed that the T m in the presence of CTR is 10 • C higher than that measured for dsDNA alone. We thus confirm that Hfq-CTR significantly stabilizes the dsDNA duplex in this region of regulation of the plasmid. Note that in the case of the dsDNA C-stretch region oppositely to dsDNA origin region, we can observe a slight effect of the CTR on the kinetics of annealing. This indicates that Hfq-CTR slightly changes this C-stretch region, reinforcing the helical parameters of the dsDNA. could be observed by a significant increase in peaks at 180 and 270 nm. This effect is nearly instantaneous and occurs in less than 2 min.
Note that the SRCD increase at 183 nm (and to a minor extent at 262 nm) in Figure 6 could be due to CTR-induced RNA II folding. In this case, the increase would only be ~5 mdeg as in Figure 9A; nevertheless, it is more than 10 mdeg for RNAI and RNAII annealing by the CTR. Therefore, we can conclude that the effect observed in Figure 6 is indicative for RNAI:RNAII duplex formation and not only from RNAII folding by the CTR.  [50]. Secondary structures of RNAs corresponding to relevant spectra are also shown.

Hfq-CTR Stabilizes the C-Stretch Region of ColE1 Plasmid DNA
Finally, we also analyzed the effects of CTR on the dsDNA C-stretch region of the ColE1 plasmid. First, we checked, as we did for the ColE1 origin sequence, if CTR was able to melt dsDNA ( Figure 10). We observed that CTR stabilized the structure of the dsDNA region as the signal at ~185 nm increased slightly. SRCD analysis allows measurement at 185 nm of an apparent catalytic kinetic constant kcat app of 0.069 mdeg.M −1 .min −1 . To confirm this effect, we analyzed the melting curves of the duplex ColE1 C-stretch. As shown in Figure 11, we observed that the Tm in the presence of CTR is 10 °C higher than that measured for dsDNA alone. We thus confirm that Hfq-CTR significantly stabilizes the dsDNA duplex in this region of regulation of the plasmid. Note that in the case of the dsDNA C-stretch region oppositely to dsDNA origin region, we can observe a slight effect of the CTR on the kinetics of annealing. This indicates that Hfq-CTR slightly changes this C-stretch region, reinforcing the helical parameters of the dsDNA. Figure 10. Hfq-CTR does not melt the C-stretch region of ColE1 plasmid DNA. Rather, Hfq-CTR appears to stabilize the dsDNA structure. Spectra of the kinetics are from blue (taken 2 min after the addition of the CTR) to red (80 min after the addition of the CTR). This can be observed by the evolution of the spectra, mainly at 185 nm. Inset: kinetics of annealing at 185 nm. Kinetic constants kcat app is 0.069 mdeg.M −1 .min −1 . This indicates that Hfq-CTR stabilizes the dsDNA, as confirmed below (see Figure 11). Note that we suspect that the effect could be quasi-instantaneous after binding but cannot be analyzed during the first seconds of the kinetics due to deadtime to load the cell. Figure 10. Hfq-CTR does not melt the C-stretch region of ColE1 plasmid DNA. Rather, Hfq-CTR appears to stabilize the dsDNA structure. Spectra of the kinetics are from blue (taken 2 min after the addition of the CTR) to red (80 min after the addition of the CTR). This can be observed by the evolution of the spectra, mainly at 185 nm. Inset: kinetics of annealing at 185 nm. Kinetic constants k cat app is 0.069 mdeg.M −1 .min −1 . This indicates that Hfq-CTR stabilizes the dsDNA, as confirmed below (see Figure 11). Note that we suspect that the effect could be quasi-instantaneous after binding but cannot be analyzed during the first seconds of the kinetics due to deadtime to load the cell.

Discussion
Significant effects of the E. coli Hfq protein on the regulation of ColE1-type plasmid DNA replication have been proposed based on genetic studies [17,18]. Namely, the efficiency of transformation of E. coli hfq mutant cells with a ColE1-like plasmid was decreased significantly relative to isogenic wild-type cells [17,18]. Such results suggest a positive regulation of plasmid DNA replication by Hfq. However, a higher efficiency of plasmid DNA synthesis occurred in the absence of the active Hfq protein in host cells [17], which is rather compatible with an opposite effect, i.e., inhibition of ColE1 DNA replication by Hfq. Recent studies on E. coli transformation by ColE1-like plasmids suggested that Hfq might act at the stage of establishment of plasmid DNA replication after the plasmid has entered the cell [18]. Clearly, the effects of Hfq on the regulation of ColE1 plasmid replication are complex and possibly involve different stages and reactions which are not yet determined.
In light of the above-described uncertainty regarding the specific role(s) of Hfq in ColE1-type plasmid replication and considering the recently identified effects of CTR of this protein on plasmid biology [18], it is clear that detailed studies on interactions of different domains of Hfq with NA involved in ColE1 DNA replication are crucial. Such stud-

Discussion
Significant effects of the E. coli Hfq protein on the regulation of ColE1-type plasmid DNA replication have been proposed based on genetic studies [17,18]. Namely, the efficiency of transformation of E. coli hfq mutant cells with a ColE1-like plasmid was decreased significantly relative to isogenic wild-type cells [17,18]. Such results suggest a positive regulation of plasmid DNA replication by Hfq. However, a higher efficiency of plasmid DNA synthesis occurred in the absence of the active Hfq protein in host cells [17], which is rather compatible with an opposite effect, i.e., inhibition of ColE1 DNA replication by Hfq. Recent studies on E. coli transformation by ColE1-like plasmids suggested that Hfq might act at the stage of establishment of plasmid DNA replication after the plasmid has entered the cell [18]. Clearly, the effects of Hfq on the regulation of ColE1 plasmid replication are complex and possibly involve different stages and reactions which are not yet determined.
In light of the above-described uncertainty regarding the specific role(s) of Hfq in ColE1-type plasmid replication and considering the recently identified effects of CTR of this protein on plasmid biology [18], it is clear that detailed studies on interactions of different domains of Hfq with NA involved in ColE1 DNA replication are crucial. Such studies are also important because plasmid-borne antibiotic resistance is a serious medical problem; thus, understanding details of molecular mechanisms of plasmid DNA replication regulation may indicate ways to limit this very adverse phenomenon [3,4]. On the other hand, methods for precise studies on interactions between small proteins and NA are restricted. Results of experiments are often difficult to interpret due to the limited resolutions of available laboratory techniques. Therefore, in this work, we have developed SRCD-based methods of investigation of interactions of Hfq, and especially its CTR, with NA. This allowed us to assess the effects of the tested protein on interactions between oligonucleotides representing crucial fragments of NA involved in the regulation of ColE1 plasmid replication. SRCD was employed in these studies, as the synchrotron light source could provide intensities of the beam below 190 nm, allowing the determination of NA conformation.
The crucial results of our studies on the interactions of CTR of Hfq with NA involved in the regulation of ColE1 plasmid DNA replication can be summarized as follows. First, CTR does not melt dsDNA at the ColE1 origin; conversely, it significantly stabilizes the dsDNA ColE1 origin structure. Second, CTR aligns the ColE1 origin dsDNA. Third, CTR accelerates RNA I-RNA II annealing. Fourth, CTR aligns the RNA I-RNA II duplex. Fifth, CTR stabilizes the RNA I-RNA II duplex, especially at high temperatures. Sixth, CTR stabilizes RNA II secondary structure but not that of RNA I. Seventh, CTR does not melt the C-stretch region of ColE1 plasmid DNA; conversely, it stabilizes it.
How can the above-summarized results, and demonstrated properties of CTR, be interpreted in the light of in vivo results which might lead to ostensibly contradictory conclusions about the positive vs. negative regulation of ColE1 replication by Hfq? In fact, CTR revealed properties that might both potentially stimulate DNA replication by aligning the dsDNA origin region (thus making the DNA region more susceptible to further specific interactions with pre-primer RNA and subsequent replication initiation) and inhibit the initiation of this process by the stimulation and stabilization of RNA I-RNA II interactions. On the other hand, no effects clearly suggesting activation of the replication process by CTR could be observed, e.g., melting of the C-stretch DNA region or annealing of RNA II to this region. In fact, CTR even stabilized the dsDNA duplex in two regions of the ColE1 origin, which suggests an inhibitory effect on DNA replication initiation. Therefore, one might propose that Hfq, or more precisely, its CTR, is probably a negative rather than positive regulator of ColE1-type plasmid DNA replication. Such a proposal is compatible with the finding that in the hfq mutant, the efficiency of synthesis of ColE1-like plasmid DNA was higher than in the wild-type control cells [17]. However, a decreased efficiency of transformation of mutants devoid of either the whole hfq gene or its part coding for CTR by ColE1-like plasmids [18] might suggest a seemingly opposite conclusion, i.e., that Hfq positively affects ColE1 plasmid replication. Nevertheless, one should note that such a suggestion is based on the most frequently accepted interpretation of results of the efficiency of transformation measurement, assuming that a lower number of host cells can be transformed if plasmid replication is impaired. However, it was demonstrated previously that if negative control of plasmid replication is abolished, the plasmid DNA synthesis process proceeds extremely efficiently. This so-called runaway replication causes non-controlled overproduction of plasmid DNA in cells leading to adverse effects, including cell death [51]. Such a phenomenon was also observed in E. coli cells bearing ColE1-type plasmids, though the final effect was a decreased viability of bacteria, from about 2-fold [52] to about 10-fold [53], rather than the complete elimination of living cells. Therefore, we suggest that less efficient negative control of plasmid replication initiation by Hfq, observed in E. coli hfq mutants [17,18], might result in runaway replication of ColE1-like plasmid and impaired viability of transformed bacteria. If this hypothesis is true, the decreased number of hfq mutant transformants appearing in experiments with these plasmids might be compatible with the role of Hfq in the negative regulation of plasmid replication initiation, as the over-replication (confirmed by the higher efficiency of incorporation of labeled precursors (nucleotides) into plasmid DNA) would cause lower viability of bacteria that acquired plasmids.
An independent, intriguing observation demonstrated in this study was that CTR of Hfq efficiently stabilized RNA I-RNA II interactions under extreme temperature conditions (above 80 • C). This effect might be related to the thermostability of Hfq. However, the annealing of oligonucleotides is highly sensitive to temperature; thus, such a phenomenon is surprising. Obviously, 80 • C cannot be applied to in vivo conditions, as E. coli dies at temperatures over 54 • C [54]. Nevertheless, in E. coli cells, effects of Hfq could be more pronounced at temperatures like 42-45 • C, while at very high temperatures, we observed enhanced effects which cannot occur in living cells, but in vitro, they were clearly represented. One might speculate that the stabilization effect at the high temperature arises from the formation of high-molecular-weight, amyloid-like complexes, formed by Hfq and its CTR. Such structures were reported previously in vitro and in vivo [25,55], and it is possible that interactions of this part of the Hfq protein with RNA strands result in the building of macromolecular complexes due to the trapping of nucleic acid strands by amyloid bodies. Such structures, if formed indeed, would be of extremely high stability. Interestingly, some reports suggested that the Hfq protein could form aggregates or higherorder complexes in E. coli cells, and it was possible to observe concentrated signals inside cells that represented Hfq [56,57]. Moreover, foci-like structures formed by this protein were observed in bacterial cells. These complexes, composed of Hfq molecules, were especially abundant under conditions of nitrogen starvation of E. coli [58]. Therefore, it is possible that our in vitro observation also reflects processes occurring in bacterial cells, especially under some environmental stresses such as high temperature. Nevertheless, this is unlikely, as we observed that temperatures above 70 • C disrupt the amyloid structure of Hfq-CTR [59].

Conclusions
The main and unexpected result of this study is that an amyloid-like region (CTR) of the E. coli Hfq protein promotes RNA annealing and induces a strong alignment of RNAs and DNA involved in the replication control of ColE1-type plasmid. Moreover, CTR stabilizes the dsDNA duplex in two regions of the ColE1 origin region. These results open perspectives to understand ColE1 plasmid DNA replication in more detail. On the basis of experiments utilizing SRCD, we propose that Hfq is mainly a negative regulator of ColE1 plasmid replication and suggest that such a proposal is compatible with otherwise ostensibly contradictory results of in vivo experiments indicating lower efficiency of transformation of E. coli hfq mutants by ColE1-like plasmids and higher efficiency of synthesis of plasmid DNA. The hypothesis that alleviated inhibition of plasmid replication initiation in hfq mutants might lead to runaway replication and resultant decreased viability of transformants is compatible with all results published to date and provides a possible explanation of the role of the Hfq protein in ColE1 plasmid biology.