Tuberculosis, caused by Mycobacterium tuberculosis complex, is a global problem for modern healthcare. According to WHO expert estimates, in 2020 the number of deaths caused by the disease increased by 100,000 people compared to 2019 and amounted to 1.3 million. The negative trend is expected to continue in the coming years, reflecting a reduction in funding for tuberculosis diagnostics and treatment amid the overall strain on global healthcare due to the COVID-19 pandemic [1].

Despite the use of anti-tuberculosis therapy, the treatment success rate for drug-resistant cases does not exceed 60%. The ubiquitously emerging resistance of mycobacterial strains to new drugs such as linezolid, bedaquiline, clophosamine, etc. [2] accentuates the demand for new anti-tuberculosis drugs with a fundamentally different mechanism of action, as well as new targets for anti-tuberculosis therapy.

One plausible new target is represented by G-quadruplexes (G4) — non-canonical secondary structures formed by guanine-rich DNA and RNA sequences under physiological conditions. Each structural unit of G4 — G-quartet — consists of four guanine bases. The G-quartets are held together by π-π-stacking interactions and additionally stabilized by metal cations [3].

In eukaryotes, such non-canonical nucleic acid structures are fairly well studied and play an important role in genome stability regulation and maintenance [4]. In the 2000s, putative quadruplex sequences (PQS) were discovered in genomes of many bacteria and archaea, but their functional role requires further investigation [5]. G4 are possibly involved in various aspects of bacterial physiology, including survival in adverse environments, pathogenic bacteria interactions with the host, antigenic variation, etc. [6].

G4 ligands, typically compounds with low-molecular weight, bind the quadruplexes and affect their thermal stability, which may displace or reestablish certain protein factors and enzymes functionally associated with DNA or RNA and ultimately interfere with transcription and translation. The best known G4 ligands include acridine derivative BRACO-19 and cationic porphyrin TMPyP4 (fig. 1). Potential use of G4-ligands as antimicrobials has been demonstrated for Vibrio cholerae, Klebsiella pneumoniae, Streptococcus pneumoniae and Mycobacterium tuberculosis [6].

Tuberculosis mycobacteria, with their high GC content and considerable PQS density, represent a promising target for antimicrobial action of G4-stabilizing ligands. Inhibition of their growth by micromolar concentrations of known G4-ligands c-exNDI-2, BRACO-19 and TMPyP4 has been previously demonstrated. In these experiments, c-exNDI-2 and BRACO-19 exerted stabilizing effects on particular G-quadruplex elements located in promoter regions [7].

Similarly, TMPyP4 exerted a stabilizing effect on quadruplexes located in virulence-associated genes [8].

It should be emphasized that all the aforementioned studies were focused on quadruplex-forming sequences at particular genomic locations; notably, the presence of G4 was invariably associated with inhibited expression of the corresponding gene under the action of ligand. In our previous study, we analyzed the transcriptomic response to inhibitory concentrations of TMPyP4 (4 µM) and BRACO-19 (10 µM) in a model Mycobacterium smegmatis MC2155 strain. We found that 10% and 12% genes of the bacterium changed their expression under the action of TMPyP4 and BRACO-19, respectively. However, we found no statistically significant correlation between differential expression and the presence of PQS [9].

In this study, we aimed to assess the influence of subinhibitory concentrations of G4 ligands TMPyP4 (2 µM) and BRACO-19 (5 µM) on gene expression profiles in M. smegmatis MC²155 and interprete the results in connection with the previously obtained data.

METHODS

Bacterial strain and cultivation conditions

In this study Mycobacterium smegmatis MC²155 model strain was used in all experiments. The cultivation was carried out at 37 °С in a humid atmosphere containing 5% СО₂ using Middlebrook 7H9 broth (HiMedia; India) or Middlebrook 7H11 agar (HiMedia) with addition of 0.5% glycerol and 10% Middlebrook OADC growth supplement (HiMedia). Cryopreserved bacteria from a certified collection were seeded on agar plates and grown for 24 h before inoculation to a liquid medium.

Cultivation of M. smegmatis for the transcriptomic analysis was carried out in accordance with the previous publication [9]. Bacterial cells were grown to 0.47 optical density measured at 570 nm, corresponding to the mid-exponential phase, and transferred to 5 mL tubes (NUOVA APTACA; Italy). G4-stabilizing agents were added to the cultures in final concentrations corresponding to a half of minimal inhibitory concentration (5 µM for BRACO-19 and 2 µM for TMPyP4). The cells were incubated for 4 h (cell division time described previously [10]) at 37 °C in a humid atmosphere containing 5% CO₂ in a thermal shaker (250 rpm). All experiments were carried out in three biological replicates.

RNA extraction and transcriptomic analysis

RNA extraction and transcriptomic assay were carried out using previously described protocols [9]. Bacterial cells were collected by centrifugation (8000 g, 10 min, 4 °C); the pellet was washed with phosphate buffered saline and mixed with RNAprotect Bacteria Reagent (Qiagen; USA) to stabilize RNA. The cells were lyzed in Lysing Matrix B 2 mL tubes placed in a MagNA Lyzer instrument (Roche; Switzerland) for 30 s. The lyzates were processed in a KingFisher automated station (Thermo Fisher Scientific; USA) using MagMAX mirVana Total RNA Isolation Kit (Thermo Fisher Scientific) in accordance with the manufacturer’s protocol. The isolated total RNA was treated with TURBO DNA-free kit (Thermo Fisher Scientific) in a 50 µL volume. Additional RNA purification was performed with Agencourt RNA Clean XP kit (Beckman Coulter; USA).

The libraries were prepared starting from 300 ng of total RNA. The ribosomal RNA was removed by Ribo-Zero Plus rRNA Depletion Kit (Illumina; USA) in accordance with the manufacturer’s protocol. The transcriptomic libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit (NEB; USA). The libraries were pooled in equimolar amounts, diluted to a 12 pM final concentration and high-throughput sequenced on HiSeq 2500 Illumina platform using HiSeq Rapid SBS Kit v2 (50 cycles) and HiSeq SR Rapid Cluster Kit v2 with 1% Phix (Illumina) added as control. The sequencing data are deposited in NCBI (Accession: PRJNA765512).

Bioinformatics analysis

The sequencing reads were mapped to M. smegmatis MC²155 genome (CP000480.1) using HISAT2 [11]. The sorting of SAM files, their conversion to BAM files and subsequent indexing were carried out in SAMtools software [12]. The mapping quality and gene coverage were assessed using QualiMap [13]; the reports were aggregated by MultiQC [14]. The mapped reads were assigned to specific genes using featureCounts [15]. Differential gene expression analysis was performed with edgeR package [16] for R. Genes with false discovery rate

For intersample comparisons the counts were normalized using the gene length corrected trimmed mean of M-values [17]. Subsequent analysis of functional enrichment for GO terms (categories) and KEGG pathways involving differentially expressed genes was performed with clusterProfiler package [18]; the categories were considered enriched at FDR ≤ 0.05.

RESULTS

Transcriptomic influence of G4-stabilizing ligands BRACO-19 and TMPyP4 in M. smegmatis MC²155

To assess the influence of G4 ligands on transcriptomic profiles of M. smegmatis, the bacteria were exposed to subinhibitory concentrations of BRACO-19 and TMPyP4 (5 µM and 2 µM, respectively) for 4 h. The results of RNA sequencing for the ligand-treated bacterial cultures were compared mutually as well as against previously obtained data for higher, inhibitory concentrations of the ligands (10 µM and 4 µM, respectively).

The principal components analysis revealed clusterization of the replicates in every experiment as well as pronounced effects of both ligands compared with the control (fig. 2A). At the same time, the analysis revealed close similarity of the effects for different concentrations of the same ligand.

For BRACO-19, we identified 589 (316↑; 273↓) and 865 (555↑; 310↓) differentially expressed genes at 5 µМ and 10 µМ, respectively (fig. 2B). For TMPyP4, we observed the opposite trend with decreased number of differentially expressed genes at higher concentration of the ligand: 754 (337↑; 417↓) and 702 (359↑; 343↓) for 2 µМ and 4 µМ, respectively.

Ligand-induced changes in expression of genes associated with putative quadruplex-forming sequences

Correlations of the ligand sensitivity with the presence of putative quadruplex-forming sequences were assessed for the following groups of genes identified previously: PQS in both DNA strands (615 genes of n = 665), PQS in coding strand (255 genes of n = 267), PQS in antisense (template) strand (360 genes of n = 398), PQS in intergenic regions (two flanking genes for 14 elements of n = 53) and PQS in promoter regions (17 genes of n = 17) [9]. Both ligands revealed no statistically significant differences with control samples for all groups of genes regardless of concentration (Games-Howell post-hoc test ^p_adj. >> 0.05; fig. 3).

Interestingly, the number of differentially expressed quadruplex-associated genes increased as 58 → 75 with increasing concentration of BRACO-19 (5 µM → 10 µM), but decreased as 72 → 59 with increasing concentration of TMPyP4 (2 µM → 4 µM) (fig. 4). 

Ligand-induced changes in expression of genes of the replication and repair system

Comparative analysis of RNA sequencing data for BRACO-19 and TMPyP4 revealed their differential influence on the replication and repair system of M. smegmatis at transcriptional level (table). The effects of BRACO-19 included elevated expression levels for DNA gyrases, translesion synthesis DNA polymerases, DNA and RNA helicases and DNA repair system proteins. Translesion synthesis DNA polymerases are SOS response proteins involved in active recovery of the bacterial chromosome integrity [19]. DNA gyrases (type II topoisomerases) participate in DNA synthesis by introducing negative supercoils [20], whereas methyltransferase Ogt and DNA glycosylase AlkA prevent DNA damage during alkylation [21]. It should be noted that the use of BRACO-19 in subinhibitory concentration (5 µM) affected expression of fewer genes compared with the inhibitory concentration (10 µM). A > 2-fold increase in expression upon switching the dose of BRACO-19 from subinhibitory to inhibitory, observed for ogt and alkA genes only, may indicate a specific dose-dependent effect of the given ligand on these genes.

For TMPyP4, the only gene with an increased expression level of the repair system was ssbb. This gene encodes a protein which participates in replication and recombination repair processes through binding to single-stranded DNA to keep it from rehybridization. Expression level of the DNA helicase gene RecQ was increased by both ligands.

Ligand-induced changes in metabolic pathways of M. smegmatis MC²155

The analysis of GO terms and KEGG pathways revealed significant enrichments (FDR < 0.05; fig. 5) for the transcription (GO:0006351), arginine biosynthesis (GO:0006526), transport (GO:0006810), sulfur metabolism (msm00920), siderophore group nonribosomal peptide biosynthesis (msm01053) and ABC transporter (msm02010) functionalities.

Similarly, with the use of BRACO-19 in inhibitory concentrations, the exposure to subinhibitory levels of the same compound affected the only metabolic pathway associated with iron metabolism. The analysis identified elevated expression of genes involved in siderophore biosynthesis, transport, and reuptake. By contrast, genes involved in iron storage showed reduced expression. It is important to note that, judging by the level of enrichment, the response of the bacterial cell to subinhibitory concentrations of the ligand was stronger.

The use of TMPyP4 stimulated expression of transcription factor-encoding genes (GO:0006351) and genes involved in sulfur metabolism (msm00920) associated with the enrichment of cellular transport systems observed at different concentrations of the ligand: the majority of GO:0006810, msm02010 and msm00920 genes were common between these groups. By contrast, genes involved in arginine biosynthesis (GO:0006526) showed reduced expression at both concentrations of the ligand.

DISCUSSION

This study continues our previous research on the transcriptomic response of M. smegmatis to inhibitory concentrations of G4-stabilizing compounds BRACO-19 and TMPyP4 [9]. In this work, we used subinhibitory concentrations of the ligands for more refined specification of the links between PQS and gene expression, as well as identification of primary responders among metabolic pathways.

The bioinformatics analysis revealed extensive transcriptomic alterations induced by subinhibitory concentrations of G4 ligands (fig. 2). The influence of ligand concentration on the extent of transcriptomic effect depended on the ligand: increasing the ligand concentration to inhibitory led to either increase or decrease in the number of differentially expressed genes (for BRACO-19 and TMPyP4, respectively). Further analysis of the data revealed no significant regulation of PQS-associated genes by G4 ligands or concentrationdependent ligand-mediated G4 stabilization (fig. 3). We also analyzed common effects of the two ligands by focusing on particular genes reacting similarly to both ligands by either up- or downregulation. However, even for genes selected on this basis, we failed to find specific evidence of the ligand-mediated G4 stabilization or dose-dependent effects. Stabilization of the quadruplex structures previously described in M. tuberculosis could not be confirmed, given significant genomic divergence between M. tuberculosis and M. smegmatis.

The analysis of KEGG pathways and GO categories also revealed no signs of dose dependency. All transcriptomic effects exerted by subinhibitory concentrations of the ligands were consistent with the previously published results obtained by using inhibitory concentrations of these ligands [9] (fig. 5). The only interesting exception is transcriptomic representation of the iron metabolism system, which reacted to BRACO-1 with higher enrichment at subinhibitory concentrations of the ligand. This cellular system becomes activated under conditions of iron deficiency, with oxidative or nitrosative stress, as well as under increased demand for metalloproteins involved in DNA replication and repair [22]. The use of BRACO-19 in subinhibitory concentrations led to overrepresentation of the siderophore nonribosomal peptide biosynthesis pathway (msm01053) including the mycobactin synthesis genes. Other systems of the pathway revealed a gradual increase in gene expression levels at higher concentrations of the ligand.

Among the replication and repair system genes of M. smegmatis, the only gene activated by subinhibitory concentrations of BRACO-19 and TMPyP4 was recQ. The ability of RecQ helicase to resolve the quadruplex structures has been demonstrated for E. coli; however, RecQ of M. smegmatis is not homologous to the eponymous protein of E. coli; moreover, RecQ is absent in M. tuberculosis [23]. This point questions the ability of RecQ protein of M. smegmatis to resolve G4 structures and subjects it to further scrutiny.

It should be recognized that transcriptional response in bacteria is a complex phenomenon modulated by multiple factors, including subtle genetic differences, environmental exposures, etc. Recent findings indicate that quadruplexes, apart from their inhibitory effect, can act as transcriptional activators depending on the conditions. In eukaryotes, quadruplexes may provide structural cores for transcription factor binding and promote functional changes in chromatin architecture, R-loop stabilization, etc. [24]. In bacteria, the quadruplex stabilization by naphthalene diimide derivatives has been shown to inhibit gene expression in Gram-negative strains while enhancing it in Grampositive strains, which supports the bidirectional nature of the quadruplex-mediated regulation [25]. The regulatory role of RNA quadruplexes in bacteria should not be ignored, as their functional presence in prokaryotes has been proved experimentally [26]. The use of advanced techniques for quadruplex detection and verification, including proteomic experiments, will ensure further progress in our understanding of the role of particular G4 elements in genomic regulation.

CONCLUSIONS

Guanine quadruplexes, described in the majority of microorganisms, represent an important element of functional genomic regulation. The current study, however, secures no direct relationship between stabilization of the putative quadruplex structures and expression of genes associated with these structures. Both quadruplex-stabilizing ligands tested in this study exerted diverse transcriptomic effects, encompassing numerous genes and several metabolic pathways. The changes observed for the transcription factor-encoding genes and the replication and repair system genes under the action of these ligands indicate their potential involvement at the RNA/DNA level, consistently with the current concept of quadruplex-mediated genomic regulation. The enhanced transcription of RecQ DNA helicase gene observed in the studied systems positions its product as a plausible candidate enzyme participating in the resolution of quadruplex structures.