Real-time kinetic studies of Mycobacterium tuberculosis LexA–DNA interaction

Transcriptional repressor, LexA, regulates the ‘SOS’ response, an indispensable bacterial DNA damage repair machinery. Compared with its Escherichia coli ortholog, LexA from Mycobacterium tuberculosis (Mtb) possesses a unique N-terminal extension of additional 24 amino acids in its DNA-binding domain (DBD) and 18 amino acids insertion at its hinge region that connects the DBD to the C-terminal dimerization/autoproteolysis domain. Despite the importance of LexA in ‘SOS’ regulation, Mtb LexA remains poorly characterized and the functional importance of its additional amino acids remained elusive. In addition, the lack of data on kinetic parameters of Mtb LexA–DNA interaction prompted us to perform kinetic analyses of Mtb LexA and its deletion variants using Bio-layer Interferometry (BLI). Mtb LexA is seen to bind to different ‘SOS’ boxes, DNA sequences present in the operator regions of damage-inducible genes, with comparable nanomolar affinity. Deletion of 18 amino acids from the linker region is found to affect DNA binding unlike the deletion of the N-terminal stretch of extra 24 amino acids. The conserved RKG motif has been found to be critical for DNA binding. Overall, the present study provides insights into the kinetics of the interaction between Mtb LexA and its target ‘SOS’ boxes. The kinetic parameters obtained for DNA binding of Mtb LexA would be instrumental to clearly understand the mechanism of ‘SOS’ regulation and activation in Mtb.


Introduction
The expression of DNA damage and stress response genes, which serve to preserve genome integrity upon exposure to DNA damaging agents, is controlled by the 'SOS' response pathway. Activation of the 'SOS' response helps the bacteria to develop resistance to antibiotics, making it indispensable for survival and growth under adverse conditions [1][2][3]. This pathway is regulated by two key players, namely, RecA and LexA. LexA binds to a consensus sequence of DNA known as the 'SOS' box located in the operator region of several genes and transcriptionally represses them under normal physiological conditions. However, under stress conditions, LexA falls off from the operators leading to activation of these genes to facilitate DNA repair [4]. The 'SOS' regulons exhibit significant variations across the bacterial kingdom, reflecting their overall complexity. For example, while Bacillus subtilis harbors only 33 genes in its 'SOS' regulon, Escherichia coli contains over 45 genes [5]. Most of the 'SOS' regulons include genes that encode for error-prone DNA polymerases, LexA, RecA, and proteins involved in the nucleotide excision repair pathway, although exceptions are known to exist [6].
'SOS' activation occurs in the following sequence of events-(i) RecA interacts with single-stranded DNA to form activated nucleoprotein filament complex, (ii) activated RecA directly interacts with LexA leading to autoproteolytic cleavage of the latter, and finally (iii) LexA falls off from the operator regions causing transcriptional de-repression of the damage-inducible genes [7]. In E. coli, the LexA repressor binds to consensus 'SOS' box sequence present in different operators, with variable affinity [7]. Genes with lower operator-repressor affinity are activated early on in the 'SOS' pathway when compared with by the overlap PCR method. The primers used have been listed in Supplementary Table S2. All constructs generated have been confirmed by sequencing. The recombinant WT and its variant proteins were over-expressed in E. coli BL21 (DE3) cells. Cultures were induced with 0.5 mM IPTG at OD 600 0.6. The cells were pelleted after 4 h by centrifugation at 5000×g following which the pellet was resuspended in Lysis Buffer composed of 50 mM Tris-Cl (pH 8.0), 150 mM NaCl and 10 mM imidazole (pH 8.0), 5% glycerol, 1 mM phenylmethylsulfonyl fluoride (PMSF). After lysing the cells by sonication on ice (10" On, 30" Off cycles), the clarified lysate was centrifuged for 1 h at 20,000×g. The clarified supernatant was passed through a pre-equilibrated HisTrap HP column at 5 ml/min. Column was washed with 50 mM Tris-Cl, 500 mM NaCl, 30 mM imidazole (pH 8.0), and proteins were subsequently eluted under a gradient in 25 mM Tris-Cl (pH 8.0), 150 mM NaCl and 750 mM imidazole (pH 8.0). Pure fractions were pooled, diluted with a low salt buffer (25 mM Tris-Cl [pH 8], 100 mM NaCl, 1 mM EDTA, 5% glycerol), and loaded on to Q-Sepharose column for anion exchange chromatography. Proteins were eluted in gradient by passing a high salt buffer (25 mM Tris-Cl [pH 8.0], 1 M NaCl, and 1 mM EDTA). The pure fractions were concentrated using Gel Filtration Buffer (20 mM Tris-Cl [pH 8.0], 100 mM NaCl, and 5% glycerol) and separated using Superdex 75 10/300 GL for gel filtration. The purified proteins were run on 12% SDS-PAGE for analyzing their purity and concentrations were determined via spectrophotometric analysis.

Cross-linking reactions
The cross-linking reactions were performed by incubation of each of the proteins at 5 μM final concentration in presence of 0.01% v/v glutaraldehyde in 10 mM HEPES (pH 8.0), 50 mM NaCl for 30 min on ice. The reactions were stopped with 25 mM of DTT. Samples were separated on a 12% SDS-PAGE.

Circular dichroism
Circular dichroism (CD) spectra were recorded from 195 to 280 nm using a Jasco J-815 spectropolarimeter. A 1 mm pathlength quartz cuvette was used. Resolution up to 0.2 nm was maintained with a scan rate of 100 nm/min. The temperature of 25 • C was maintained for all experiments. About 5 μM of each protein in 10 mM Tris-Cl 50 mM NaCl (pH 7.5) was taken for analysis. The data presented are an average of three scans after correction for the buffer baseline. Recorded spectra were analyzed using Origin 8.1 software.

Extrinsic fluorescence
Extrinsic fluorescence spectra were obtained using a Jobin-Yvon Fluorometer FluoroMax3 at 25 • C. About 5 μM of each of the proteins in 10 mM Tris-Cl (pH 7.5), 50 mM NaCl was incubated with 44 mer ds dnaE2 'SOS' box containing DNA (sequence given in Table 1) at 1:2 ratio for 30 min at 37 • C. The samples were incubated with 40 μM of ANS in dark for 10 min. Samples were excited at 350 nm, and in the range of 400-600 nm, emission spectra were recorded. Measurements were corrected for fluorescence intensity of buffer, DNA, and ANS intrinsic fluorescence.

Electrophoretic mobility shift assay
WT and its variants at 128 nM were incubated in the presence of 3.5 nM of end-labeled (non-biotinylated) 44 mer ds dnaE2 'SOS' box containing DNA (sequence given in Table 1) in 10 mM HEPES (pH 7.5), 50 mM NaCl for 30 min on ice. The unbound DNA and DNA-protein complexes were resolved on 8% native PAGE at 100 V for 1 h in cold. Gels were dried and autoradiographed. The same procedure was followed for EMSA analysis in which increasing concentrations (0-128 nM) of WT Mtb LexA was incubated with 32 P end-labeled ds 44 mer ds dnaE2 'SOS' box containing DNA.

Bio-layer Interferometry (BLI)
The ForteBio Octet RED 96 (Forte Bio, U.S.A.) platform was used to conduct interaction studies between LexA and its variants with biotinylated ds 44mer of different 'SOS' boxes containing sequences (listed in Table 1). Streptavidin matrix-coated sensor chip (SA) was equilibrated in 10 mM HEPES (pH 7.5), 50 mM NaCl followed by immobilization of 100 nM of biotinylated ds DNA on it. Increasing concentrations of WT and mutant proteins were passed on to the chip and change in response units (RU) was analyzed. The program comprises 1 min stabilization of the baseline with the buffer followed by 10-min loading of sensors with biotinylated DNA, a 5-min association enabling interaction between the protein and DNA, a 5 min dissociation step finally followed by a 5 s regeneration step (unless mentioned otherwise). A reference sensor dipped in the buffer was used as a background control. All analyses were carried out at 25 • C. A 1:1 binding model was applied to globally fit the binding isotherms and kinetic parameters such as k on , k off , and K D were obtained. The experiments were performed in triplicates.

Mtb LexA features distinct characteristics from its counterparts
LexA is present in most bacterial species and phyla [17]. Evolutionarily, the protein has retained its two distinct domains, the NTD, involved in DNA binding, and CTD, which is responsible for dimerization and autoproteolytic cleavage. Alpha helices involved in DNA binding and the residues critical for autoproteolysis have remained well conserved across different species, thereby preserving the overall functions of the protein.
From an evolutionary viewpoint, a comparison between LexA homologs from selected representatives belonging to major classes of Gram-positive, Gram-negative, Archaebacterial, and Actinobacterial phyla reveal the discrete clustering based on their classification, evident from the phylogenetic tree constructed ( Figure 1A). This tree has been deduced by comparing sequences from 24 representative bacterial species (shown in Supplementary Figure S1). LexA homologs from Actinobacteria closely resemble homologs from Gram-positive Firmicutes which is consistent with their relatedness at the species level. Interestingly, LexA from members of Actinobacteria that include the pathogenic tuberculous mycobacteria such as Mtb, M. canetti, and M. bovis shows significant similarity, suggesting a possible link between pathogenicity and sequence evolution of LexA.
Subsequent comparison of the LexA sequences among some of the well-known tuberculous and non-tuberculous mycobacteria revealed interesting results. Although the C-terminal regions remain almost identical, N-terminal regions exhibit sequence variations especially at the terminal end. Pathogenic mycobacteria have relatively smaller genome sizes [18] and are not expected to code for unwanted additional stretches of amino acids in their proteins unless they prove advantageous for their survival. Interestingly, tuberculous mycobacteria except for M. bovis were found to possess additional amino acids at the N-terminal end of LexA, unlike their E. coli counterpart, implying unexplored adaptive functions. Another region of less conservation spans the linker region that connects the NTD and CTD of the protein. While the latter half of the linker sequences (toward the CTD) exhibits more conservation, the initial half varies among different mycobacterial species. Although the linker is identical between Mtb and M. bovis harboring 25 amino acids, the number of residues and sequence conservation vary among the other mycobacteria ( Figure 1B). In M. leprae and M. haemophilum, the linker can extend up to 28 amino acids long. LexA possessing additional stretches of amino acids triggers the curiosity to explore their mechanistic roles in 'SOS' induction that could help mycobacterial survival and evolution.
Presumably, these stretches of amino acids in Mtb LexA that remain uncharacterized may confer additional functions unique to mycobacterial species when compared with its well-characterized homologs. LexA is a global repressor controlling the expression of DNA repair genes. Hence, assessing whether these additional stretches of amino acids may influence interactions with 'SOS' boxes could provide new information related to 'SOS' regulation. Considering both the additional stretches of amino acids to lie in proximity to the DNA binding regions of the protein, we deleted these stretches to explore their impact on DNA binding. Subsequently, the stretch of 24 amino acids (residues 1-24) was deleted to generate LexA 24aa, and LexA 18aa was generated by deleting 18 amino acids spanning the hinge Comparison of LexA sequences among tuberculous, non-tuberculous mycobacterial species, and model Gram-negative organism, E .coli. The first 24 amino acids extension, the N-terminal DNA binding domain, linker region, the stretch of 18 amino acids insert in the linker, and C-terminal dimerization domain are shown as a bar representation below the sequence in magenta, light blue, dark blue, yellow, and light green, respectively. DNA-binding residues chosen for mutation are shown with a brown bar representation below. Sequence alignment was done using Clustal Omega, and ESPript was used to generate the Figure. (C) Representation of constructs generated for this study is shown, and the gel picture shows purified Mtb LexA and its variants resolved and visualized on 12% SDS PAGE.
region (residues 94-111) of the protein. This long hinge region separates the NTD from its CTD in Mtb LexA. This is in sharp contrast with the much shorter hinge region of just four amino acids (Q70-E74) present in E. coli ( Figure  1B). The functional relevance of such a long hinge region in Mtb has not been explained. Possibly, the longer length of this inter-domain linker in Mtb LexA can enhance its flexibility to attain suitable conformations for binding to DNA [13]. Next, we have attempted to shed light on this aspect in the present study by characterizing and comparing the DNA-binding ability of the variants with that of WT LexA. RKG motif involved in DNA binding contains Arg 52, Lys 53, and Gly 54 in E. coli LexA lying in the third alpha helix of the protein [19], and its corresponding Arg 75, Lys 76, and Gly 77 residues of Mtb LexA have remained conserved. We have mutated these residues to assess whether they are essential for DNA binding in Mtb LexA.

Evaluation of Mtb LexA and its variants for dimerization and DNA-binding property
All proteins were purified to ∼98% purity ( Figure 1C). Near and far UV spectra of Mtb LexA and its mutants were used to assess changes in their secondary structures due to mutations ( Figure 2A). Interestingly, both far (195-250 nm) and near (250-280 nm) UV spectra from CD studies revealed the comparable secondary structures of WT Mtb LexA and its mutants.
All the variants predominantly exist as dimers in solution as analyzed from profiles of gel filtration chromatography (Supplementary Figure S2). To confirm this further, the purified proteins were subjected to cross-linking using the chemical cross-linker glutaraldehyde. Upon cross-linking, the predominant form appeared to be dimeric in all cases as evident from the top band running between 48 and 63 kDa ( Figure 2B). Structural analysis of the C-terminal segment of Mtb LexA revealed residues 229-236 along with residues 139-153 from its NTD to be involved in dimer formation between two LexA monomers [13]. The mutants generated in the present study have no overlapping sequences with the aforementioned residues and therefore, all of them retained the ability to form dimers. Therefore, a comparable profile of results in CD spectroscopy and protein dimerization experiments revealed that the selected regions had minimal or no influence on the secondary structure of Mtb LexA.
Next, we used fluorescence spectroscopy to assess the structural changes of Mtb LexA and its variants upon interaction with DNA. Hydrophobic extrinsic fluorescent dye 8-anilino-1-naphthalenesulfonic acid (ANS) has been widely used for screening the alterations in the tertiary structure of proteins and to monitor their ligand-binding events such as protein-nucleic acid interactions [20]. The ligand displaces the fluorescent dye upon binding to the protein, resulting in a quench in fluorescence. Changes in fluorescence intensity are a direct readout of protein-DNA binding. The maximum quench in fluorescence intensity was noted when WT LexA formed a complex with DNA ( Figure 2C). A similar quench in fluorescence intensity was observed for LexA 24aa and the WT protein upon DNA binding ( Figure 2F), suggesting that the 24 amino acids extension is not crucial for DNA binding. However, we cannot overlook the possibility that it may have a regulatory role in DNA binding that could be dependent on the 'SOS' box sequences [21]. In striking contrast with LexA 24aa, deleting 18 amino acids from the linker region has significantly reduced the fluorescence quenching to nearly half compared with that for the WT, suggesting that this linker region may play a prominent role in DNA binding ( Figure 2E). The deletion of 18 amino acids had weak or no effect on the secondary structures; however, the observed reduction in nucleic acid binding affinity may be influenced by the Van der Waals interactions offered by either glutamic acid or aspartic acid residues (4 out of 18 amino acid residues) within the linker region. Lastly, RKG/AAA mutant showed no significant fluorescence quench upon interaction with DNA ( Figure 2D), thereby establishing that RKG residues play a critical role in DNA binding in Mtb LexA similar to the other orthologs. Further, we have carried out electrophoretic mobility shift assays (EMSA) to see the DNA binding of WT LexA to the dnaE2 'SOS' box. The shift in the 32 P-labeled dnaE2 'SOS' box DNA suggests that Mtb LexA can bind to 'SOS' box with nM affinity ( Figure 3A). We have also performed EMSA with Mtb LexA variants. EMSA analysis revealed that LexA 24aa and LexA 18aa showed mobility shift, whereas RKG/AAA mutant did not show any DNA binding ( Figure 3B). This further confirms the importance of the RKG motif in DNA binding. Based on these observations, we further decided to quantitate the real-time kinetic parameters of LexA-DNA interactions with WT LexA and its variants.

Determination of DNA binding kinetics of Mtb LexA and its variants
We used BLI for determining the DNA binding affinity for Mtb LexA and its mutants. First, we characterized the interaction between WT Mtb LexA and its variants to the perfectly palindromic dnaE2 'SOS' box ('SOS' box sequence given in Table 1 Table 1) indicates the conformational changes of the proteins upon DNA binding. Fluorescence intensity is shown in arbitrary units.  Table 1), and EMSA was carried out according to standardized conditions mentioned in Materials and Methods.  Figure S3). LexA18aa variant shows significantly reduced affinity as seen from the sensograms depicted in Figure 4. Deleting 18 amino acids from the linker connecting NTD and CTD resulted in a more than 15 times reduction in affinity when compared with the full-length protein, with an obtained K D of 34.4 + − 0.19 nM ( Table 2). The rate of association (k on ) was highest in the order of 10 6 for full-length Mtb LexA and decreased to the order of 10 4 in the case of the 18aa variant. While the association rate constants varied, the dissociation rate constants (k off ) did not change significantly. We speculate that deleting the 18 amino acids from the linker affected the conformation to bind DNA suitably, thereby reducing its association rate ( Figure 4B and Table 2). We did not observe any detectable binding for RKG/AAA, even after doubling the time of interaction (from 300 to 600 s of association) ( Figure 4G), indicating the RKG motif in DBDs is critical for the DNA binding. Altogether, Mtb LexA exhibits similarity to E. coli counterpart in terms of binding to its cognate 'SOS' box with nanomolar affinity. Additionally, the presence of the longer linker in Mtb LexA is found to positively affect its DNA-binding ability. The kinetic parameters obtained by performing experiments at physiological pH intrigued us to investigate how they would vary when subjected to acidic conditions. The rationale behind choosing a highly acidic pH condition to monitor changes in DNA binding affinity of Mtb LexA arises from the fact that Mtb is known to face a hostile environment of acidic pH inside host macrophages and is challenged to maintain internal pH homeostasis for survival [22]. Although internal pH lower than 6 is noted to be lethal for mycobacteria, it has to endure external pH as low as 4 [22]. We wanted to assess whether Mtb LexA and its variants could exhibit DNA binding even at an extreme pH such as pH 4.

R K G / A A A 8 a a a a W
We found that Mtb LexA exhibited maximal and optimal binding with the perfectly palindromic dnaE2 'SOS' box near physiological pH (pH 7.5). Surprisingly, Mtb LexA retained the ability to bind DNA even at pH 4, although with reduced affinity (  nearly six times for LexA 24aa, and nearly 3.2 times for LexA 18aa when compared with physiological conditions ( Table 2). From the kinetic parameters observed, we notice a significant reduction in association rate constants (k on ) at low pH (pH 4) compared with physiological conditions (pH 7.5) for wild-type protein and its 24aa variant while the change is not so pronounced in the case of the 18aa variant. Dissociation rate constants (k off ) in the case of WT Mtb LexA remained comparable in both the pH conditions tested but reduced by a power of 10 for the mutant proteins at low pH. Mutants seem to both associate and dissociate faster at physiological pH as compared with low  Table 2. The corresponding response versus concentration curves has been plotted in (D-F) from the results of three independent experiments.
pH and as for the wild-type protein, there has been a drastic reduction only in its rate of association to bind DNA at low pH conditions. Our observations results corroborate with those made from in vitro studies that have assessed the effect of variations in pH in regulating the 'SOS' response for E. coli LexA [23][24][25]. Relan and colleagues reported that E. coli LexA bound to its operator maximally near physiological pH displaying about 10-fold better binding compared with that at pH 4 [23]. Similarly, we noticed a reduction in DNA binding by Mtb LexA as well, at acidic pH. It will be interesting to decipher the molecular events leading to this.

DNA binding kinetics of Mtb LexA with different 'SOS' boxes
The differential gene expression profile following DNA damage led to the identification of genes that fall under direct regulation of LexA in Mtb [21]. However, DNA binding kinetics of LexA to 'SOS' boxes of these DNA damage-inducible genes remained uncharacterized. We, therefore, determined the DNA binding affinity for LexA and its mutants to different mycobacterial 'SOS' boxes (the kinetic parameters determined for the interaction of mutants to different 'SOS' boxes are provided in Supplementary Figure S4 and Table S3). The 'SOS' boxes chosen have unique characteristics (Table 1). While the dnaE2 'SOS' box is a perfect palindrome throughout (as mentioned in the previous section), lexA and recA 'SOS' boxes have one mismatch toward their 3 ends (on the flank). rv3074 'SOS' box is unique in displaying a perfect palindrome of sequences on either side repeat flanks but showing mismatches in sequences between the flanks. All the genes whose 'SOS' boxes have been chosen for the present study are highly induced following DNA damage in Mtb [21]. Increasing concentrations of WT LexA and its variants (analytes) were allowed to interact with biotinylated 'SOS' boxes till saturation in binding was achieved. WT LexA was found to bind to different 'SOS' boxes with close affinities ( Figure 6 and Table 3). K D values ranged from 0.98 + − 0.01 nM for the lexA 'SOS' box to 3.86 + − 0.03 nM as noted for the rv3074 'SOS' box. The association rate was relatively higher for dnaE2 and lexA 'SOS' boxes as compared with the other two 'SOS' boxes. The perfectly palindromic nature of the dnaE2 'SOS' box facilitates faster association with LexA. dnaE2 encodes an error-prone DNA polymerase; hence, its regulation must be strictly controlled. It is known that genes that are involved in mutagenesis such as these error-prone DNA polymerases are expressed later in the 'SOS' response cascade and are tightly controlled [26]. Our observation confirms the same in the case of Mtb.
Although the association rate constants (k on ) varied for different 'SOS' boxes, the corresponding dissociation rate constants (k off ) are also seen to change proportionately; hence, the overall K D is not widely altered for different 'SOS'

Conclusion
Mtb LexA controls gene expression patterns of the crucial 'SOS' response pathway that facilitates mycobacterial adaptation to stress [12]. However, lack of thorough understanding at the molecular level, taking into account the unique regions of Mtb LexA that could potentially influence its interaction with DNA, prompted us to execute the present study by analyzing the impact of such truncations/ mutations on Mtb LexA-DNA interaction. Together, we present our detailed analysis of Mtb LexA and the role of its additional stretches of amino acids in regulating the 'SOS' response.
To begin with, the deletion and mutated variants displayed comparable secondary structure as that of wild-type Mtb LexA protein, inferred from the ellipticity measurements carried out using circular dichroism. Moreover, they retained the ability to form dimers as observed from size exclusion chromatography and crosslinking studies.
Qualitative estimation and comparative analysis of protein-nucleic acid interaction of the variants compared with the WT revealed that while the 24 amino acids extension at the N-terminal is not critical for Mtb LexA-DNA association, deletion of the 18 amino acids linker connecting the NTD and CTD of the protein resulted in a marked reduction in DNA binding compared with the full-length protein. The 18 amino acids present in the linker most likely accounts for the conformational flexibility of Mtb LexA to suitably bind DNA. Moreover, mutating the RKG motif in the DNA binding helix abolished LexA-DNA binding, highlighting the significance of strong evolutionary conservation of this motif across different organisms.
The quantitation of DNA binding in real-time has been carried out using BLI and kinetic parameters of Mtb LexA-DNA interaction have been determined. The binding affinity of WT Mtb LexA (K D 2.16 + − 0.01 nM) and LexA 24aa (K D 4.94 + − 0.03 nM) was within 2-fold range, while a 17-fold reduction with LexA 18aa (K D 34.4 + − 0.19) and no observable binding with RKG/AAA mutant was observed. Mtb LexA was found to bind different 'SOS' boxes under mycobacterial 'SOS' regulation with comparable affinity. Association with perfectly palindromic sequence was found to be stronger. However, since association and dissociation rates changed proportionately for all the 'SOS' boxes, the overall affinities were found to fall in a close range. Although Mtb LexA binds to different 'SOS' boxes with comparable affinity in vitro, in the cellular context, the time and spatial regulation of 'SOS' genes might be altered by other transcription factors, intracellular pH, specific cations, and anions. In vivo studies may uncover their actual regulation under DNA damaging and normal conditions. DNA binding assays under both physiological and extreme acidic pH conditions in vitro revealed that mycobacterial LexA retains DNA binding even at pH as low as 4, albeit with reduced affinity as compared with its optimum binding at physiological pH. Taken together, our study provides a better understanding of the real-time kinetics of mycobacterial 'SOS' regulation. Extensive characterization of Mtb LexA in controlling one of the key stress-responsive pathways of the bacteria is imperative and will facilitate designing unconventional and yet more effective therapeutic strategies in counteracting TB infection.

Data Availability
All supporting data and sequence information are included within the main article and its supplementary material.