Mycobacterial DNA-binding protein 1 is critical for long term survival of Mycobacterium smegmatis and simultaneously coordinates cellular functions

Bacteria can proliferate perpetually without ageing, but they also face conditions where they must persist. Mycobacteria can survive for a long period. This state appears during mycobacterial diseases such as tuberculosis and leprosy, which are chronic and develop after long-term persistent infections. However, the fundamental mechanisms of the long-term living of mycobacteria are unknown. Every Mycobacterium species expresses Mycobacterial DNA-binding protein 1 (MDP1), a histone-like nucleoid associated protein. Mycobacterium smegmatis is a saprophytic fast grower and used as a model of mycobacterial persistence, since it shares the characteristics of the long-term survival observed in pathogenic mycobacteria. Here we show that MDP1-deficient M. smegmatis dies more rapidly than the parental strain after entering stationary phase. Proteomic analyses revealed 21 upregulated proteins with more than 3-fold in MDP1-deficient strain, including DnaA, a replication initiator, NDH, a NADH dehydrogenase that catalyzes downhill electron transfer, Fas1, a critical fatty acid synthase, and antioxidants such as AhpC and KatG. Biochemical analyses showed elevated levels of DNA and ATP syntheses, a decreased NADH/NAD+ ratio, and a loss of resistance to oxidative stress in the MDP1-knockout strain. This study suggests the importance of MDP1-dependent simultaneous control of the cellular functions in the long-term survival of mycobacteria.


Results
Impacts of MDP1-defeciency on the growth and survival of M. smegmatis. To determine the possible role of MDP1 in long-term survival of mycobacteria, we cultured MDP1-defecient M. smegmatis (MDP1-KO) in the conventional 7H9-ADC medium under normoxia and compared the growth and survival with those of the parental (wild-type) strain (WT) and MDP1-complemented MDP1-KO strain (COMP). We set the initial viable bacterial number (colony forming units, CFUs) as 7 × 10 4 . As shown in Fig. 1a, MDP1-KO had slightly longer lag phase, but then grew similarly to WT and COMP and entered the stationary phase after 7 days with a CFU count of 1 × 10 12 ± 0.0003. However, MDP1-KO died more rapidly than the other strains after entering the stationary phase and then reached a plateau after 15 days, maintaining a similar CFU count until Day 30. These data show that MDP1 slightly shortens the lag phase and prevents rapid death of M. smegmatis in the stationary and death phases. Consequently, it sustained the number of living bacteria after death phase, during it, the cell count was around 30-1000-fold lower for MDP1-KO than for WT and COMP 29 .
To confirm the upregulation of MDP1 in the stationary phase of M. smegmatis, we performed western blot analysis. Anti Hsp65 antibody was used as an internal loading control to confirm that the protein concentrations were equal among samples. Besides we added Ponceau S staining of the proteins prior to antibody reaction in order to check the protein amount of each lane is the same. In the same experimental setting as used to analyze the growth kinetics, we obtained bacteria on Day 6 (OD 600 values ~0.5 and 0.8, log phase, samples obtained at two different times on Day 6), Day 7 (OD 600 ~1.0, stationary phase), and Day 9 (OD 600 ~1.2, stationary to death phase) and monitored MDP1 expression. The amounts of MDP1 at OD 600 0.5 and 0.8 were low or below the detection limits but elevated in WT and COMP bacteria with OD 600 1.0 and 1.2. The MDP1 expression is higher in WT than COMP. This suggested the expression of MDP1 was partially complemented in COMP strain harboring a single copy of the MDP1 gene inserted in the genome-integration vector pMV306. We confirmed that MDP1 expression was absent in MDP1-KO strain at all the measured data points (Fig. 1b).
Proteomic analysis shows that MDP1 is a general repressor of gene expression. To determine how MDP1 prevents the rapid bacterial death in the stationary phase, we employed proteomic analysis. Based on the kinetics of growth and the upregulation of MDP1 in the stationary phase ( Fig. 1), we extracted proteins from bacteria with OD 600 = 1.2 and analyzed them. Although we started with an equal culture volume grown to the same OD, the quantification of total protein extracts by BCA assay showed a higher level in the MDP1-KO culture relative to WT and COMP, 740 ± 2 µg, 440 ± 1.7 µg, and 480 ± 1.7 µg, respectively. These data might be consistent with the suppressive activity of MDP1 on the gene expression 17 , although we cannot exclude the possibility that this was caused by the discrepancy between OD values and the real amounts of bacteria. Ten microgram protein extracts from each sample were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) ( Supplementary Fig. S1a). Densitometric comparison of band intensities between samples revealed a characteristic difference in the fractionation pattern of the MDP1-KO strain, especially between 50 kDa and 75 kDa markers ( Supplementary Fig. S1b). We also observed the expected difference in MDP1 itself. Since band differences might be attributable to several proteins with similar molecular weights, we performed more precise identification, using two-dimensional electrophoresis (2-DE) with isoelectric focusing as the first dimension.
Triplicate 2-DE images for a wide isoelectric non-linear range (pI: 3-10) are presented in Supplementary  Fig. S2a. Progenesis SameSpot software (Totallab) initially identified 897 dysregulated spots in the MDP1-KO gel images compared to those of WT and COMP. The spot pattern distribution was not well resolved in the acidic part of the gels. To better fractionate the acidic region, we ran additional triplicate gels using a narrower range (pI: 4-7) (Supplementary Fig. S2b). This allowed us to detect 152 more different spots.
Initial proteomic analysis of these spots was performed as shown in the chart in Supplementary Fig. S3. Briefly, we applied criteria for filtering the automatically retained results from Progenesis SameSpot software. This step is essential to eliminate biased false positives spots generated from staining artifacts. First, filter 1 (F1) depended on the spot volume intensity to retain spots significantly differ by ≥3 fold. Second, filter 2 (F2) was based on ANOVA to choose spots significantly different between in the WT and MDP1-KO gels (P ≤ 0.05). Third, filter 3 (F3) was a manual filtering step to eliminate suspected background and noise spots. These iterations finally resulted in the identification of 24 and 5 significant spots with over 3-fold difference between MDP1-KO and WT/COMP from the pI 3-10 and 4-7 gel images, respectively (total = 29 spots) (Fig. 2). Interestingly, all the proteins identified bar Identification of significantly differently expressed proteins due to MDP1-defeciency. We performed mass spectrometric identification of all 29 spots. After removing repetitions, we finally identified 22 proteins (Table 1). Three-dimensional views of intensities of the selected protein spots (spot numbers 1, 9, 11, and 25) are shown in Fig. S4, respectively. These data reconfirmed the previously reported enhanced expression of KatG in the MDP1-KO strain 23 . The effects of MDP1 on the other 21 proteins have not been reported so far.
Enhancement of DNA synthesis on MDP1-defeciency. Several proteins that mediate major cellular functions were included in the protein list which are upregulated on MDP1-defeciency (Table 1). Among them, DnaA is a critical enzyme that initiates DNA replication by binding to a DnaA box near to the replication origin of the bacterial genome and opening it to initiate replication [29][30][31] . The average normalized abundance of DnaA was significantly higher in MDP1-KO than in WT and COMP (3.9-fold difference) (Fig. S4b). We therefore postulated that MDP1 suppresses DnaA expression and MDP1-deficiency leads to a greater DNA synthesis.
To address this hypothesis, we chased DNA synthesis in each growth phase, with cells at OD 0.5, 0.8, 1.0 and 1.2, by using 3 H-labeled uracil and a method established previously 32 . We normalized values by dividing each count per minute (CPM) by its corresponding number of living bacteria (CFU). As shown in Fig. 3, DNA synthesis was higher in bacteria at OD 0.5 and 0.8 (logarithmic growth phase) than in bacteria at OD 1.0 and 1.2 (stationary or death phase). MDP1-KO exhibited elevated DNA-synthesis, demonstrating the suppressive role of MDP1 in replication in the growth and stationary/death phases.
To investigate whether MDP1 controls the transcription of DnaA, we carried out quantitative reverse transcription polymerase chain reaction (qRT-PCR). Expression levels were determined by the comparative threshold cycle method with normalization to those of sigA and 16S rRNA. The transcription of dnaA was 3.4-± 1.4 -fold upregulated by MDP1-deficiency, suggesting that MDP1 controls DnaA expression, mainly at the transcriptional level ( Supplementary Fig. S5).
Decrease of NADH/NAD + ratio on MDP1-deficiency. NADH dehydrogenase (NDH) was one of the upregulated proteins in the MDP1-KO strain (Table 1). According to two-and three-dimensional views of the NDH protein spot, the average normalized abundance of this protein was significantly higher in MDP1-KO than in WT and COMP (3.8-fold difference) (Fig. S4c). qRT-PCR analysis showed that the mRNA level was increased 3.6-± 0.5 -fold in MDP1-KO in comparison with WT ( Supplementary Fig. S5); thus, MDP1 controls NDH expression, mainly at the transcriptional level.
Considering the proteomic and transcriptomic data showing upregulation of NDH, which oxidizes NADH into NAD + , we hypothesized that the MDP1-KO strain with relatively high NDH activity would contain an altered NADH/NAD + ratio. To test this hypothesis, the intracellular levels of NADH and NAD + were measured in WT, MDP1-KO, and COMP strains at different growth stages (i.e., OD 600 values) (see Methods) by bioluminescence assay. The NADH/NAD + ratio increased as the culture developed. The NADH/NAD + ratio was found to significantly decrease in the stationary phase of growth in MDP1-KO in comparison with WT and COMP (Fig. 4a, Supplementary Table S1). At OD = 1.2, the NADH/NAD + ratio was 0.59 ± 0.01 in KO and 0.85 ± 0.01 and 1.02 ± 0.04 in WT and COMP, respectively. The ratios at different ODs are summarized in Supplementary  Table S1. These data suggested that in the WT, MDP1-dependent inhibition of NDH-expression suppresses NADH dehydrogenase activity in the stationary phase.
Increase of cellular ATP level on MDP1-defeciency. The NADH/NAD + ratio is a reliable metric that reflects bacterial ATP synthesis 33 . Therefore, we tested the notion that MDP1-KO might alter ATP production. As expected, we found an elevated level of ATP in MDP1-KO cells in the stationary phase compared with WT and COMP (Fig. 4b).
Reduced resistance to reactive oxygen species (ROS) on MDP1 deficiency. Expression of two major antioxidant genes of mycobacteria, ahpC, alkyl hydroperoxide reductase C, and katG, catalase-peroxidase [34][35][36][37] , were increased in the MDP1-KO strain in the proteomics analysis (Table 1 and Fig. S4a and d). MDP1 itself is also an antioxidant protein; it has ferroxidase activity and suppresses generation of hydroxyl radicals, the most aggressive ROS, through prevention of the Fenton reaction 38 . Accordingly, there are two general possibilities: either MDP1-deficient M. smegmatis becomes more sensitive to ROS through loss of MDP-1, or it becomes more resistant to ROS by upregulation of AhpC and KatG. To determine which, we investigated the effect of growth inhibition by ROS using the disk method 39 . The diameter of growth inhibition produced by exposure to H 2 O 2 or menadione (vitamin K 3 ) was higher in MDP-KO than in WT and COMP, indicating that the MDP1-KO strain is more sensitive to ROS than WT and COMP (Fig. 5). These data demonstrated that MDP1 plays a major role in resistance to ROS.
We confirmed that the expression of ahpC and katG was mainly regulated at the transcriptional level by MDP1, since qRT-PCR analysis showed that their mRNAs were elevated 6.3 ± 0.4 and 7.6-± 0.4-fold, respectively in MDP1-KO compared with the WT (Supplementary Fig. S5), although the complemented strain did restore the transcription of the AhpC gene, probably due to the partial complementation of MDP1 in COMP strain.

Discussion
In the current study, we studied the potential role of MDP1 in the long-term survival of M. smegmatis. We showed that expression of MDP1 is increased in the stationary phase 13,15,23 and prevents rapid cell death after bacteria enter this phase, subsequently keeps the surviving numbers of bacteria after death phase, during this time living bacteria acquire nutrients from dead siblings 40 (Fig. 1). We performed proteomic analysis to elucidate the precise role of MDP1 in prevention of rapid death at stationary to death phases. One of the proteins controlled by MDP1 was DnaA, an initiator of replication. Downregulation of DnaA indicates that M. smegmatis actively, rather than passively, terminates replication in stationary phase. This action must be advantageous, since bacteria can save energy by eliminating irrelevant DNA synthesis. Besides interestingly MDP1-dependent suppression of DNA synthesis is not limited to stationary phase (Fig. 3). This implies the involvement of MDP1 in the slow growth itself in log phase, and is consentient with Lewin's finding that partial reduction of MDP1 expression makes the slow growing M. bovis BCG grow faster 20 .
Stationary phase is generally characterized by equilibrium between the numbers of proliferating and dying bacteria, rather than cessation of bacterial growth. However, in this study, the level of DNA synthesis in the stationary phase was as low as <1% of that in the growth phase, even for a similar number of CFUs (1 × 10 11 ± 0.02 at OD 0.8 and 1.5 × 10 12 ± 0.04 at OD 1.0 for the WT). This indicates that growth arrest occurred in a large proportion of the bacteria in the stationary phase, rather than being an equilibration of growth and death. The  growth status of long term living bacteria without replication is considered to be dormant, implying that dormant bacteria arise in the stationary phase of M. smegmatis, as proposed in a previous study 2 .
Proteomic and biochemical analyses further defined the roles of MDP1. We found MDP1 suppressed the expression of Type II NDH, which transfers electrons from NADH to a common quinol pool 33 . Although type II NDH is essential in both M. smegmatis and Mtb 41,42 , it neither translocates protons nor conserves energy, whereas non-essential type I NDH is a proton-translocating enzyme 42 . However NAD + generated by NDH is used in other metabolic reactions to generate ATP. In fact, the ATP level was observed to be higher in the MDP1-KO strain than in the WT and COMP strains (Fig. 4b). There is possibility that the suppression of NDH is involved in the inhibition of ATP production, although it should be clarified by a further study.
Accumulation of toxic metabolites causes bacterial death. ROS are constantly generated as byproducts of aerobic respiration, are generated to a greater extent in stressful conditions including bacterial stationary phases, and are also absorbed from the external environment 43 . Thus, antioxidant defense systems are essential for all aerobic organisms. We found that MDP1 downregulates antioxidant molecules such as AhpC and KatG, while the loss of MDP1 resulted in higher sensitivity to ROS (Fig. 5). This might be explained by the function of MDP1 itself in inhibiting the generation of ROS by preventing the Fenton reaction 38 . Another possible explanation is changed ratio of NADH/NAD + , which is important for controlling the redox balance. A lower NADH/NAD + ratio in the stationary phase (Fig. 4) causes a redox imbalance and may cause the higher sensitivity of MDP1-KO to ROS (Fig. 5).
Other molecules controlled by MDP1 have not yet been extensively analyzed, but may also contribute to its function. For example, we found that Fas1, a critical enzyme for cellular fatty acid synthesis, is controlled by MDP1 (Table 1). This implies that MDP1 may suppress membrane and cell wall biosynthesis in the stationary phase by suppression of Fas1 expression.
Taken together, MDP1 has multifunctional properties, such as, suppression of DNA synthesis, respiration, and possibly fatty acid synthesis, and protection against ROS 19 . We consider that MDP1-dependent pleiotropic gene expression control is feasible for long term-survival, because bacteria are able to not only save energy to stay alive but also prevent the generation of toxic metabolites simultaneously. Because MDP1 has iron storage activity 38 and recently proposed as iron-transporter molecule from ferricarboxymycobactin to mycobactin in the cell envelope of mycobacteria 44 , exhausted iron in the stationary phase might be involved in the loss of viability of MDP1-KO in stationary phase. In this study we used 7H9-ADC medium, which is rich of iron (150 µM). The role of MDP1 related in the iron acquisition should have more impact on the survival of Mycobacterium under iron-limited condition, such as, in the macrophages as shown previously 27 . These considerations and reports suggest that MDP1 is a drug target to eradicate persisting mycobacterial pathogens, although further experiments are required to elucidate its role in the pathogenic Mycobacterium.
Growing evidence shows that slow growth rate, low metabolism and antioxidant properties are determinants of the longevity in multicellular eukaryotes [45][46][47][48][49] . Interestingly these activities are shared with the functions of MDP1. The present study is likely to provide the novel knowledge to understand the long-term living of the organisms from bacteria to multicellular organisms.
Measurement of growth kinetics. The growth kinetics of the three tested strains were monitored in terms of CFU per ml for 1 month. On certain days (as shown in Fig. 1), 10-fold dilutions from 10 −1 to 10 −7 were made and appropriate dilutions were plated in triplicate on Middlebrook 7H9 agar. The plates were incubated at 37 °C for 5 days before calculating the CFU/ml. The experiment was performed in duplicate.
Extraction and preparation of bacterial proteins. Mycobacterial cultures were harvested by centrifugation for 10 min at 4,000 × g at 4 °C three times. Cellular proteins were prepared from the cell pellets according to Lanigan et al. with slight modifications, using bead beating at 4 °C for 30 s intermittently and then adding acetone 51 . Cellular debris was removed by centrifugation and protein lysate was collected and stored at −80 °C. Protein assay of the extracts was carried out using the bicinchoninic acid (BCA) assay kit (Pierce, Thermo Scientific) with bovine serum albumin as the standard 52 .

SDS-PAGE and western blotting.
Seven micrograms of protein extract from each sample were electrophoresed in duplicate on 12.5% SDS-polyacrylamide gels 53 . One gel was stained with Coomassie Brilliant Blue stain (CBB R-250, Wako, Japan) and the second was used for western blotting; the bands were transferred to a polyvinylidene difluoride membrane. For primary antibody incubation, 1:10000 diluted mouse anti-MDP1 antibody (7C monoclonal antibody) was used and anti-heat shock protein Hsp65 (GroEL2) antibody was used as an internal control 23 . Reactions were visualized using ECL Prime western blotting detection reagent (GE Healthcare Lifesciences, Japan). The fractionation patterns were compared using the public domain software ImageJ (a Java image processing program developed by the National Institutes of Health).

Isoelectric focusing (IEF) and second dimension gel electrophoresis (2-DE). First dimensional
IEF was performed using the Ettan IPGphor 3 ™ system (Amersham Biosciences, UK) with 7 cm Immobiline ™ DryStrips Non Linear pH 3-10 and pH 4-7 (GE Healthcare, UK). Protein extracts were diluted in DeStreak rehydration buffer and focused at 9400 V.h at a maximum voltage of 5000 V. Following IEF, proteins were reduced and alkylated by soaking the strips in equilibration solution (6 M urea, 50 mM Tris-HCl pH 8.8, 30% glycerol, 2% SDS, 0.004% bromophenol blue) containing 10 mM dithiothreitol and 55 mM iodoacetamide 54 . Equilibrated strips were directly used for the second dimension SDS-PAGE as described before 55,56 . The 2-DE was run in at least in three replicates per sample to ensure maximum accuracy and reproducibility.
Gel staining, spot imaging and visualization. Gels were stained with silver stain. The 2-DE patterns were recorded as digitalized images and compared using Progenesis SameSpots software (Totallab) for detection of significantly different spots. The software was programmed to select significant spots, ANOVA with P ≤ 0.05 and cutoff spots ≥3-fold different. Spots with significantly altered levels were excised for mass analysis.
Spot processing and peptide mass mapping by TripleTOF MS/MS. Two-dimensional gels were stained using the Silver Stain Kit for Mass Spectrometry (Piece, Thermo Scientific). The spots of interest were excised and reduced with 10 mM dithiothreitol, carbamide methylated with 55 mM iodoacetamide, and subjected to in-gel trypsin digestion. The peptides from each sample were finally dissolved in 15 μl of 0.3% formic acid, and 5 μl of each sample was injected into a nano-flow-LC (Eksgent nanoLC 415 with ekspert cHiPLC, AB Sciex) coupled with a tandem MS (TripleTOF5600, AB Sciex). Analysis was conducted in duplicate for each sample in trap and elute mode using a ChromeXP C18 Chip column (200 μm × 0.5 mm) as a trap column and a ChromeXP C18 Chip column (75 μm × 150 mm) as an analytical column. Mobile phases A and B were 0.1% formic acid and 0.1% formic acid in acetonitrile, respectively. Peptides were eluted by 20-min gradients from 2% B to 32% B at 300 nl/ min. MS spectra (250 msec) followed by 10 MS/MS spectra (100 msec each) were acquired in the data-dependent mode.
Bioimformatic analysis. Product ion output data were searched against the Mycobacterium smegmatis mc 2 155 UniProt KB database (9,855 entries) with concatenated decoy database using a locally stored copy of the Mascot search engine (version 2.4, Matrix Science, London, UK). A protein was accepted if peptides passed the identity and homology thresholds of the Mascot algorithm. The false discovery rate against the decoy database was <1%. Gene Ontology annotation including protein molecular function and biological processes were searched using the SmegmaList database (http://mycobrowser.epfl.ch/smegmalist.html) and UniProt database (http://www.uniprot.org/).

RNA extraction.
Total RNA was extracted from three samples with TRIzol reagent (Invitrogen, Carlsbad, CA) following the manufacturer's recommendations. Cultures were harvested by centrifugation at 4,000 × g. Cells were resuspended in 1 ml TRIzol reagent and disrupted using glass in a Bead Smash 12 (Wakenyaku, Japan). The samples were incubated at room temperature, chloroform was added, and finally RNA pellets were air-dried and redissolved in ultrapure distilled RNase free water (Invitrogen). RNA quantity and quality were assessed.
Determination of NADH to NAD + ratio. Mycobacterial pellets harvested at different OD values were lysed in a basic buffer (0.2 M NaOH and 1% dodecyltrimethyl ammonium bromide). Two aliquots of each sample (100 µl) were taken, one for NADH measurement (base-treated) and the second, mixed with 0.4 M HCl, for measurement of NAD + (acid-treated). Both aliquots were heated at 60 °C for 15 min. Neutralization with 0.4 M HCl/0.5 M Trizma and 0.5 M Trizma was done for base-and acid-treated aliquots, respectively. The assay was carried out using NAD/NADH-Glo ™ (Promega) as described by the manufacturer. Luminescence was recorded with a luminometer (Filter Max F5 Multi-Mode Microplate reader) with SoftMax Pro Easy software, and the NADH/NAD + ratio was calculated by comparing the relative light units.
Scientific REpORts | 7: 6810 | DOI:10.1038/s41598-017-06480-w Measurement of ATP level. Cellular ATP levels were determined in the three tested strains at different OD values by a luciferase bioluminescence assay. Briefly, 100 µl of each sample was mixed with an equal volume of BacTiter-Glo reagent (Promega) in a 96-well plate, incubated for 5 min, and then the luminescence was measured using the previously described luminometer.
Analysis of oxidative stress sensitivity. Sensitivity to H 2 O 2 and the superoxide generator menadione were examined by zone inhibition assays, as previously described 39 . Briefly, M. smegmatis strains were grown to an OD 600 of 0.8 in 7H9-ADC medium, and approximately 1 × 10 5 bacteria were plated onto 7H11-oleic albumin dextrose catalase complex. An 8-mm paper disc saturated with 30 µl of a solution containing 500 mM hydrogen peroxide or 40 mM menadione was placed onto the center of the plate. After incubation for 2 days, the diameter of the zone of growth inhibition generated by each compound was measured.
Statistical analysis. Data were analyzed using one-way ANOVA and the Kruskal-Wallis non-parametric test followed by Dunn's multiple comparison test as a post-hoc test. Differences were considered significant when the P-value was ≤0.05.