RNase HI Is Essential for Survival of Mycobacterium smegmatis

RNases H are involved in the removal of RNA from RNA/DNA hybrids. Type I RNases H are thought to recognize and cleave the RNA/DNA duplex when at least four ribonucleotides are present. Here we investigated the importance of RNase H type I encoding genes for model organism Mycobacterium smegmatis. By performing gene replacement through homologous recombination, we demonstrate that each of the two presumable RNase H type I encoding genes, rnhA and MSMEG4305, can be removed from M. smegmatis genome without affecting the growth rate of the mutant. Further, we demonstrate that deletion of both RNases H type I encoding genes in M. smegmatis leads to synthetic lethality. Finally, we question the possibility of existence of RNase HI related alternative mode of initiation of DNA replication in M. smegmatis, the process initially discovered in Escherichia coli. We suspect that synthetic lethality of double mutant lacking RNases H type I is caused by formation of R-loops leading to collapse of replication forks. We report Mycobacterium smegmatis as the first bacterial species, where function of RNase H type I has been found essential.


Introduction
RNases H are ubiquitous enzymes present is eukaryotes, prokaryotes, archaeons and viruses. They are involved in the removal of RNA from RNA/DNA hybrids during DNA replication, repair and transcription. Based on the differences in their amino acid sequences, RNases H have been divided into two types and three classes. Prokaryotic RNase HI or eukaryotic H1 represent RNases H type 1, while prokaryotic RNases HII and HIII or eukaryotic RNases H2 represent RNases H type 2.
Type 1 RNases H require at least four ribonucleotides as a substrate for a cleavage to occur [1,2] and they do not prefer any consensus sequence to perform this reaction [1]. In vitro studies showed that introducing various modifications that decrease DNA flexibility in RNA/DNA duplex abrogates cleavage by RNase H1 [3]. Thus, the enzyme must recognize both RNA and

Materials and Methods
In silico analysis Genes presumably encoding RNases H in M. smegmatis mc 2 155, E. coli K12_MG1655 and M. tuberculosis H37Rv were identified and obtained from National Center for Biotechnology Information (NCBI) database. The span of the domains was defined using Simple Modular Architecture Research Tool (SMART) [37]. The homology between the domains was estimated using Basic Local Alignment Search Tool (BLAST) of NCBI. The alignments were visualized using MultiAlin [38] and ESPript 3.0 [39].
For determination of growth rates bacterial cells were transferred to fresh NB medium and cultured until the cultures reached OD 600 between 0.6 and 0.9. Aliquots of these seed cultures were inoculated in fresh 7H9 broth supplemented with OADC at starting OD 600 = 0.05. The cultures were incubated at 37°C with vigorous shaking for 48 h. At desired intervals of time samples of cultures were harvested and analyzed using spectrophotometer (Pharmacia Biotech

M. smegmatis mutants
The mutants were obtained by using gene replacement protocol as previously described [40][41][42]. The procedure required construction of gene replacement plasmids and complementation plasmids. Primers used for obtaining the mutants are listed in Table 3. Briefly, for gene replacement purpose, the sequences flanking desired deletion were amplified by PCR and consecutively introduced into p2NIL plasmid, followed by introduction of PacI suicidal cassette excised from pGOAL17 vector. For complementation under an acetamide promoter, a native copy of the gene of interest was amplified by PCR and introduced into pJAM2. The gene was next excised from the pJAM2 together with an acetamide promoter and introduced into pMV306Hyg. All cloning was performed in E. coli Top10 cells (Invitrogen). A plasmid allowing deletion in dnaA, previously used in [43], was modified by addition of gentamycin resistance cassette within BstBI restriction site. Plasmids were introduced into mycobacterial cells, which were further subjected to multistep selection process allowing detection of the mutants. Presence of intact copy of the dnaA gene within the ΔrnhA/ΔdnaAattB::dnaA strain at the attB site was confirmed by sequencing.

Exchange of complementation vectors with genetic cassettes for an empty vector
Conditional mutants carrying a copy of a presumably essential gene were electroporated with an empty pMV306Km plasmid. For exchange of complementation vectors with genetic cassettes for an empty vector (ExEV) on rich medium, bacteria after transformation were inoculated in NB broth and cultivated overnight at 37°C with vigorous shaking. Afterwards, bacteria were plated on NB selective medium supplemented with Km and cultivated up to one week at 37°C. For ExEV on minimal medium, bacteria after transformation were inoculated in 7H9 broth without the addition of OADC and cultivated overnight at 28°C with vigorous shaking. Afterwards, bacteria were plated on 7H10 medium without the addition of OADC supplemented with Km and cultivated for up to three weeks at 28°C. In case of ExEV on cassettes containing MSMEG4305, an additional variant of the experiment was performed which included supplementation of both rich and poor media with B12. Clones growing on Km containing media were screened in search of an intact version of the gene of interest. Encountering a clone devoid of an intact version of the gene would mean that the gene is non-essential. On the other hand, inability to find such clones would confirm the essentiality of the gene.

Southern blotting
Primers used for production of hybridization probes and restriction enzymes used for digestion of genomic DNA are listed in Table 3.

Recombinant RnhA expression and purification
Sequence encoding RnhA, identified in NCBI database, was amplified by PCR (primers listed in Table 3) and introduced into pHIS2 expression plasmid. Cloning was performed in E. coli Top10 cells (Invitrogen). The plasmid was further introduced into E. coli BL21. Expression of the protein was performed at 37°C overnight in the presence of 0.4 mM IPTG. The pellet was resuspended in 10 ml of Binding Buffer (50 mM Tris-HCl pH 8 (Sigma); 6-8M urea (Sigma)) and sonicated in short bursts (Bioblock Scientific Vibracell). The addition of urea was necessary to denature and solubilize otherwise insoluble protein. Next, the sample was incubated for 2 h with mild shaking. The sample was centrifuged at 17000 x g at 12°C for 30 min. The supernatant was transferred through filtered syringe (0.45 μm, Millex) and placed on affinity column containing Ni-NTA resin (ThermoScientific). Following binding of the protein to the Ni-NTA resin, the column was washed with Binding Buffer and Wash Buffer (60 mM imidazole (Sigma), 0,4 M NaCl (Sigma), 20 mM Tris-HCl pH 8 (Sigma)). Next, the recombinant protein was washed out with Elution Buffer (1M imidazole (Sigma), 0.5 M NaCl (Sigma), 20 mM Tris-HCl pH 8 (Sigma)). The sample containing recombinant protein was concentrated on a Novagen concentrator until achieving the final concentration of 1 mg/ml. The protein was used to immunize rabbits and for production of polyclonal antibodies.

Production of anti-RnhA antibodies
Laboratory New Zealand rabbits were raised under standard conventional conditions in the approved by Polish Ministry of Science and Higher Education animal facility of the Institute Microbiology, Biotechnology and Immunology, Faculty of Biology and Environmental Protection, University of Lodz and were used for the immunization experiments. The experimental procedures were approved and conducted according to guidelines of the appropriate Polish Local Ethics Commission for Experiments on Animals No. 9 in Lodz (Agreement 54/ ŁD1/2011).

Western blotting
Bacteria were cultured in NB medium until OD 600 reached between 0.6 and 0.9. Aliquots of these seed cultures were inoculated in fresh 7H9 broth starting OD 600 = 0.05. The cultures were incubated at 37°C with vigorous shaking overnight. The following morning 5 ml of each culture were spun down at 4°C at 8000 x g and the pellet was resuspended in 1 ml TE (10 mM Tris (Sigma), 1 mM EDTA; pH 8) with the addition of 100 mM phenylmethylsulfonyl fluoride (Sigma) and 2% SDS (Sigma). The sample was then transferred to MP tube and homogenized using FastPrep-24 MP homogenizer. Subsequently the sample incubated for 30 min at 55°C and centrifuged at 14000 x g at room temperature for 15 min. The supernatant containing proteins was transferred to a new Eppendorf tube and sample was immediately used.
Cell extracts or recombinant protein solutions were subjected to acrylamide electrophoresis. Following protein electrophoresis, the gel (12% Mini-Protean Precast Gel BioRad) was placed in a plastic container and covered with PVDF membrane (ThermoScientific) washed beforehand in methanol and Transfer Buffer (0,2 M glycine (Sigma); 25 mM Tris (Sigma); 20% methanol (Sigma)). The transfer was performed overnight at 4°C at 50 mA in an electrophoresis container filled with transfer buffer. Next the membrane was blocked for 1.5 h at room temperature in PBS (137 mM NaCl (Sigma); 2.7 mM KCl (Sigma); 10 mM Na 2 HPO 4 • 2 H2O (Sigma); 2M KH 2 PO 4 (Sigma) pH 7.4) containing 10% milk (Gostynin, fat-free powdered milk). Subsequently, the membrane was incubated in PBS containing 5% milk, 0.05% Tween 20 (Sigma) and primary antibodies (rabbit anti-RnhA, obtained at Department of Immunoparasitology, University of Lodz or rabbit anti-LigA, described previously [44]) at 4°C overnight. Afterwards, the membrane was washed three times in PBS with 0.05% Tween 20 (Sigma). The membrane was then placed in PBS containing 5% milk and a secondary antibody (anti-rabbit goat IgG conjugated with peroxidase (Sigma)). The membrane was incubated for 1 h at room temperature and washed three times with PBS containing 0.05% Tween 20. The membrane was then placed in a dry plastic container and covered with mixture of ECL reagent. Following a 2 minutes incubation the membrane was covered in Saran wrap and exposed to an X-ray film (ThermoScientific) for 2 minutes. The film was developed using Kodak Medical Xray Processor.

Statistical analysis
In order to determine growth rates of the analysed strains, we fitted logistic or quadratic curves to the collected measurements of optical density of the liquid cultures. We fitted logistic curves of the form OD = A/[1 + B Ã exp(-KT)], where OD is optical density at the time T, A is an asymptotic value, B is a constant of integration, and K is the growth rate constant. Parameter K from the fitted curves was used as an indicator of growth rate for each strain. A T-test was used to test for the differences in growth rates between the strains. All statistical analyses were performed with Statistica 10.0 (StatSoft, Tulsa, OK, USA).

Results and Discussion
The genome of M. smegmatis encodes two predicted RNases H type I Through performing an in silico analysis we identified two genes of M. smegmatis mc 2 155 which encode proteins containing RNase HI domain-rnhA and MSMEG4305. BLAST analysis of RNase HI domains of RnhA and MSMEG4305 of M. smegmatis mc 2 155 revealed that with 50% of query cover of RnhA, they share 36% of protein sequence identity. Next, MSMEG4305 was shown to be homologous to Rv2228c of M. tuberculosis H37Rv. Both proteins share 72% sequence identity with 100% query cover of MSMEG4305. RNase HI domains of MSMEG4305 and Rv2228c share 72% identity with 97% of MSMEG4305 query cover, while acid phosphatase domains share 75% sequence identity with 100% query cover of MSMEG4305. The alignment between protein sequences between RNases H type I of E. coli K12_MG1655, M. smegmatis mc 2 155 and M. tuberculosis H37Rv is presented in Fig 2. A possible reason for the coexistence of two RNase H type I encoding genes within the genome of M. smegmatis is the neofunctionalization of MSMEG4305. Apart from encoding RNase HI domain it also encodes CobC domain involved in vitamin B12 biosynthesis. Such neofunctionalization is thought to increase retention of RNase H type I genes [21]. The coexistence of MSMEG4305 and rnhA might be expected to lead to future inactivation of RNase HI domain in one of them and generation of a pseudogene [45]. In fact, the genome of M. tuberculosis, which is thought to have undergone genome reduction when compared with freeliving mycobacteria, contains only a homolog of MSMEG4305.

Essentiality of RNases H type I in M. smegmatis
We used the technique of gene replacement through homologous recombination to construct a series of genetic mutants with large deletions introduced within the sequence of predicted RNase H type I encoding genes. We were able to identify single mutants of both rnhA and MSMEG4305 (Fig 3). The ability to obtain the mutants signifies that none of these genes itself is essential for survival of M. smegmatis.
We were unable to identify a mutant deficient in ΔrnhA/Δ4305. Therefore, we introduced a functional copy of either MSMEG4305 or rnhA at the attB site of ΔrnhA/SCOMSMEG4305 or SCOrnhA/ΔMSMEG4305, respectively. Following selection, we generated two strains ΔrnhA/ Δ4305attB::Pami-rnhA and ΔrnhA/Δ4305attB::Pami-4305 deficient for both native versions of RNase HI encoding genes, but complemented with either rnhA or MSMEG4305 at the attB site (Fig 4). We used these strains to perform ExEV. This experiment greatly increases the number of analyzed cells. It is therefore used as a final confirmation of the essentiality of the gene [46][47][48]. It also limits drawbacks of gene replacement technique. In our case it was the enrichment of the medium by the substances used as selection markers. Enrichment of medium might be a source of confusion between contradictory reports regarding RNase HI essentiality in E. coli [29][30][31]. The authors that have obtained RNase HI deficient mutants stated that the mutants were rich broth sensitive [30,31]. This is thought to be related to high speed of metabolism. High metabolism requires efficient transcription, which leads to generation of many R-loops. R-loops, in turn, cannot be efficiently resolved in the absence of RNase HI and lead to replication fork collapse. In minimal medium, however, the accumulation of R-loops is limited by a slowed down metabolism. Hence they can be either tolerated in the genome or efficiently removed by proteins other than RNase HI [49]. Even after performing ExEV we did not identify any clones devoid of both genes encoding RNases H type I.
The difference between the essentiality of RNase H type I in M. smegmatis and E. coli may be explained by genetic background of these species. In E. coli, rnhA encoding RNase H type I has been shown to be synthetically lethal with a number of genes, namely polA [50], recB [51], recC [51] and recG [52]. These genes can be found in mycobacterial genome. However, the mutation in rnhA has also been shown to be synthetically lethal with mutation in rep [53], a homolog of which is absent in the genome of M. smegmatis. Rep is a helicase involved in restarting DNA replication forks [54], facilitates reannealing of the parental strands during replication [55] and has been shown essential for efficient replication across highly transcribed regions [56,57]. The precise role of Rep remains unknown. Sandler suggested that Rep might be involved in R-loop metabolism and this function, due to synthetic lethality, may be connected with the function of RNase HI [53]. R-loops are linked with transcription and other helicases of E. coli-RecG [52] and Cas3 [58] are involved in R-loop removal. R-loops, if unresolved, may block DNA replication in several ways. First, unrepaired lesions in displaced DNA strand may become a source of double strand ends and consequent replication fork collapse. Second, RNA hybridized to the DNA may be an obstacle for replication fork progression. Paused replication forks can be cleaved by endonucleases known to cleave recombination intermediates again We used gene replacement through homologous recombination to obtain mutants deficient in both rnhA and MSMEG4305. We were unable to identify a mutant ΔrnhA/Δ4305. Therefore, we introduced a functional copy of either MSMEG4305 or rnhA at the attB site of ΔrnhA/SCOMSMEG4305 or SCOrnhA/ΔMSMEG4305, respectively. Following selection, we generated two strains ΔrnhA/Δ4305attB::Pami-rnhA and ΔrnhA/Δ4305attB::Pami-4305 deficient for both native versions of RNase HI encoding genes, however complemented with either rnhA or MSMEG4305 at the attB site. For more information regarding plasmid construction, gene replacement procedure and complementation please refer to the text. Schematic representation of analyzed genomic regions, including enzymes used for digestion, size of restriction fragments following digestion and the site of hybridization of hybridization probe, is presented in the upper part of the figure. Photographic films presenting results of Southern blot analysis are presented in lower part of the figure. Bands corresponding to wild type genotype (wt), mutant genotype (mut) and complementation genotype (compl) are marked on the right side of each photograph. creating double strand ends [59]. Finally, transcription complex linked to an R-loop might become attached to the hybrid and collide with progressing replication fork. All of these scenarios would result in replication fork collapse and could potentially lead to subsequent cell death [60]. Although the idea that RNase H substrates other than R-loops might be responsible for the lethal phenotype of ΔrnhA/ΔMSMEG4305 mutant cannot be entirely excluded, it seems unlikely. In other bacteria Okazaki primers have been shown to be removed by other proteins, which are also present within M. smegmatis, PolI [61][62][63] and RNase HII [25].
The essentiality of RNase HI domain in M. smegmatis suggests that similar phenomenon might be present in M. tuberculosis. Data obtained by high density transposon mutagenesis in M. tuberculosis seem to confirm this hypothesis [64]. In studies assessing the essentiality of mycobacterial genes by high-density transposon mutagenesis the authors observed a small level of transposon insertions within the gene encoding a homolog of MSMEG4305-Rv2228c. The number of insertions was significantly smaller from what would be expected. Perhaps this observation could be related to insertions within CobC domain encoded by the gene and lethal phenotype in the case of insertions within the region encoding RNase HI domain.

RNase HI mutants-RnhA level, growth rates and constitutive stable DNA replication
We wanted to investigate how deletion of MSMEG4305 influences the level of RnhA. Therefore, we obtained recombinant His-tagged RnhA protein expressed in E. coli. We used this protein to immunize a rabbit and to obtain polyclonal anti-RnhA antibodies. These antibodies were used to perform detection of RnhA in protein extracts isolated from M. smegmatis strains used in this study. Further, we quantitated the amount of RnhA in M. smegmatis mc 2 155 and compared it with the amount of LigA, NAD + dependent ligase [44]. We observed that the level of RnhA in wild type strain was low (Fig 5). There were 200 ng of LigA detected in approximately 30 μg of total protein extract, while 10 ng of RnhA was identified in approximately 200 μg of total protein extract. Therefore we estimate that LigA is approximately 133 times more abundant than RnhA.
Since the level of RnhA was low, we wanted to see whether deletion of one RNase HI genes influenced growth rate of the mutants. For this purpose, we assessed growth rates by measuring optical density of the liquid cultures at determined intervals of time and fitted appropriate growth curves [65]. We did not observe any differences in growth rates or asymptotical values between the ΔrnhA and M. smegmatis mc 2 155 on 7H9 medium supplemented with OADC (t = 1.90, df = 5, p = 0.12; t = 0.35, df = 5, p = 0.74, respectively, data presented in S1 Table) or Δ4305 and M. smegmatis mc 2 155 on 7H9 medium supplemented with OADC and vitamin B12 (t = 0.84, df = 4, p = 0.84; t = 0.49, df = 4, p = 0.65, respectively, data presented in S2 Table) ( Fig 6A and 6B). The morphology of the mutant cells grown in before mentioned conditions, in terms of cell length was not altered (t = 1.83, df = 198, p = 0.07 for ΔrnhA and M. smegmatis mc 2 155, data presented in S3  7A and 7B).
In summary, these results suggest that mycobacterial cells may produce more RNase H type I than would be sufficient for optimal growth, as deletion of one of the genes encoding this enzyme did not seem to affect the viability of the mutant. A similar phenomenon was observed for other replication related proteins of mycobacteria-namely LigA [44] or DnaG [48].
In E. coli R-loops have been shown to be responsible for alternative pathway of initiation of DNA replication. The phenomenon was termed constitutive stable DNA replication (cSDR). cSDR was identified in thermosensitive E. coli mutants with inactivated gene encoding RNase HI [66]. Unlike the classical pathway, cSDR initiates from regions termed oriK instead of oriC [67]. The initial strand opening involves RecA dependent hybridization of the RNA transcript to dsDNA [68] (Fig 8A and 8B). The resulting R-loop is not resolved by RNase HI and therefore RNA persists on DNA strand and serves as a primer for elongation by PolI [50] (Fig 8C). When the loop opens sufficiently, primosome is loaded on the leading strand. As replication continues, PolI removes persisting RNA transcript and primosome is loaded on the lagging strand [50] (Fig 8D and 8E). cSDR in E. coli is independent of DnaA and oriC. Hence, even though dnaA and oriC are essential for viability of wild type E. coli, it is possible to obtain double mutants bearing mutations inactivating either dnaA and rnhA or oriC and rnhA [69].
We wanted to investigate whether unaltered growth rate in RNase HI mutants of M. smegmatis was a consequence of induced cSDR. In order to investigate the presence of cSDR in M. smegmatis we used the technique of gene replacement through homologous recombination. We introduced gene replacement plasmid for dnaA, with Gm resistance cassette cloned within the sequence of the gene to facilitate screening, into ΔrnhA and Δ4305. In order to differentiate cell's metabolism and hence potential level of R-loop formation, we intended to remove the native version of dnaA gene on rich medium at 37°C and on minimal medium at 28°C. We were not able to identify a mutant deficient for dnaA. To confirm that the plasmid that we used for gene replacement was capable of creating a deletion within the native sequence of dnaA, we introduced the complete version of the dnaA gene under a native promoter at the attB site of ΔrnhA strain. We generated a strain, ΔrnhA/dnaASCOattB::dnaA, which was further subjected to gene replacement protocol. We were therefore able to generate a strain deficient for a native copy of dnaA, however possessing a copy of the dnaA gene at the attB site, ΔrnhA/ΔdnaAattB:: dnaA (Fig 9). Finally, we used the latter strain for ExEV experiment and confirmed that dnaA cannot be removed from ΔrnhA strain.
The results of this study confirmed that the dnaA gene is essential in M. smegmatis cells in which the formation of R-loops is limited by the presence of both RNases HI, and suggest that dnaA gene may be required in cells with an increased tendency to form these loops. However, it cannot be entirely excluded that cSDR is not present under the conditions of our experiment. Further decrease in RNase HI level could affect the ability to remove dnaA. Therefore, the existence of cSDR within mycobacteria needs to be further evaluated.  RNases HI as a new antimycobacterial target?
The products of the genes that are essential for survival of the pathogen are potential targets for novel antibiotics. The possibility that RNase HI might be essential for survival of M. tuberculosis is particularly interesting as inhibitors of RNase H are currently intensively studied as potential antiviral drugs in HIV therapy [70]. As reported by the World Health Organization, TB  is the major cause of death among people living with HIV. Though the idea that one compound might be limiting for both M. tuberculosis and HIV is wonderful, it is rather utopian. Suboptimal concentrations of RNase H inhibitors of HIV have been shown to decrease susceptibility to antiretroviral drug used for the treatment of HIV, zidovudine [71], and therefore they may be questioned as a novel component of HIV therapy. Finding an RNase H inhibitor that would inhibit both viral and mycobacterial RNase H, without affecting human RNase H, though theoretically possible, does not seem achievable. As an advantage, we observed that in M. smegmatis RNase HI level was low. Therefore, perhaps even a small amount of an inhibitor could be sufficient to effectively kill these cells. This is particularly important in the case of mycobacteria due to their intracellular life niche (in the case of M. tuberculosis) and composition of their unusual and thick cell wall. Therefore the question of whether mycobacterial RNases HI may be used for drug development remains open for future research. We used gene replacement through homologous recombination to obtain mutants deficient in rnhA. Further, we used the same procedure to generate ΔrnhA/dnaASCO strain where SCO signifies an intermediate step of gene replacement procedure. Next, through complementation procedure, we introduced an additional version of dnaA gene at the attB site of mycobacterial genome. Finally, we removed the native version of dnaA, thereby generating a mutant ΔrnhA/dnaASCOattB::dnaA. For more information regarding plasmid construction and gene replacement procedure please refer to the text. Schematic representation of analyzed genomic region, including enzymes used for digestion, size of restriction fragments following digestion and the site of hybridization of hybridization probe, is presented in the upper part of the figure. Photographic film presenting results of Southern blot analysis is presented in the lower part of the figure. Bands corresponding to wild type genotype (wt) and mutant genotype (mut) are marked on the right side of the photograph. Supporting Information S1