bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1 Shotgun proteomic profiling of dormant, ‘non-culturable’ Mycobacterium tuberculosis
2
3 Nikitushkin V.1, Shleeva M.1, Loginov D.2,3,4, Dycka F.2, Sterba J.2,3, Kaprelyants A.1
4
5 1A.N. Bach Institute of Biochemistry, Federal Research Centre ‘Fundamentals of Biotechnology’
6 of the Russian Academy of Sciences, Moscow, Russia
7 2 Faculty of Science, University of South Bohemia, České Budějovice, Czech Republic
8 3 Institute of Parasitology, Biology Centre of the Czech Academy of Sciences, České Budějovice,
9 Czech Republic
10 4 Orekhovich Institute of Biomedical Chemistry, Moscow, Russia
11
12
13 Abstract
14
15 Dormant cells of Mycobacterium tuberculosis, in addition to low metabolic activity and a high level
16 of drug resistance, are characterized by ‘non-culturability’ – a specific reversible state of the inability
17 of the cells to grow on solid media. The biochemical characterization of this physiological state of the
18 pathogen is only superficial, pending clarification of the metabolic processes that may exist in such
19 cells. In this study, applying LC-MS proteomic profiling, we report the analysis of proteins
20 accumulated in dormant, ‘non-culturable’ M. tuberculosis cells in an in vitro model of self-
21 acidification of mycobacteria in the post-stationary phase, simulating the in vivo persistence
22 conditions. This approach revealed the accumulation of a significant number of proteins (1379) in
23 cells after 4 months of storage in dormancy; among them, 468 proteins were significantly different
24 from those in the actively growing cells and bore a positive fold change (FC). Differential analysis
- 1 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
25 revealed the proteins of the pH-dependent regulatory system phoP and allowed the reconstruction of
26 the reactions of central carbon/glycerol metabolism, as well as revealing the salvaged pathways of
27 mycothiol and UMP biosynthesis, establishing the cohort of survival enzymes of dormancy. The
28 annotated pathways mirror the adaptation of the mycobacterial metabolic machinery to life within
29 lipid-rich macrophages, especially the involvement of the methyl citrate and glyoxylate pathways.
30 Thus, the current in vitro model of M. tuberculosis self-acidification reflects the biochemical
31 adaptation of these bacteria to persistence in vivo. Comparative analysis with published proteins with
32 antigenic properties makes it possible to distinguish immunoreactive proteins (40) among the proteins
33 bearing a positive FC in dormancy, which may include specific antigens of latent tuberculosis.
34 Additionally, the biotransformatory enzymes (oxidoreductases and hydrolases) capable of prodrug
35 activation and stored up in the dormant state were annotated. These findings may potentially lead to
36 the discovery of immunodiagnostic tests for early latent tuberculosis and trigger the discovery of
37 efficient drugs/prodrugs with potency against non-replicating, dormant populations of mycobacteria.
38
39
40 1 Introduction
41 Tuberculosis (TB), caused by the pathogenic microorganism Mycobacterium tuberculosis (Mtb),
42 currently surpasses HIV as the most common cause of mortality from an infectious disease worldwide
43 (WHO global report).
44 In the course of infection establishment, Mtb passes through a series of intra- and extracellular
45 locations – from alveolar macrophages to extracellular lesions, being exposed to various stress
46 conditions including lowered pH values, nutrient and cofactor limitation, and peroxynitrite stress (1).
47 Nonetheless, due to its unique metabolic plasticity, the bacterium survives these attacks, gradually
48 transiting into a state of low metabolic activity – dormancy, associated with treatment failure for
- 2 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
49 latent TB infection (2, 3). Bacterial internalization within the host is commanded by virulence factors
50 on the one hand and reprogrammed metabolic networks, balancing the anabolic and catabolic
51 processes with sufficient ATP generation to support survival (4–7).
52 Disseminated in the host, dormant bacilli may persist life-long; however, approx. 10% of latently
53 infected TB carriers develop active disease (3). Dormant bacteria obtained from in vitro models are
54 characterized by altered morphological traits (such as a thickened cell wall or cell-size diminishment),
55 or resistance to antibiotics (8, 9). However, the cells obtained in such ‘quick’ models do not reflect
56 mycobacterial physiology in vivo, where bacilli transit to a state of ‘non-culturability’ – a specific
57 reverse state of inability to grow on solid media, and dependent on a reactivation procedure in
58 selective liquid medium (10).
59 Whether metabolic pathways are active in such a state of ‘non-culturable’ dormancy of Mtb is not
60 known and any knowledge in this respect will be valuable for understanding the survival mechanism
61 and for selection and finding a remedy against latent TB. To elucidate such mechanisms, knowledge
62 of the enzymes stored in long-term stored dormant cells is essential.
63 In the case of dormant cells that are metabolically inert, and therefore tolerant to common antibiotics,
64 there is an attractive idea to consider ubiquitous enzymatic processes stored in dormancy as targets for
65 the conversion of prodrug compounds into substances with unspecific antibacterial activity (like
66 reactive oxygen and nitrogen species or antimetabolites) (11, 12).
67 Therefore, analysis of the enzymes stored up in dormancy may be prospective in terms of further
68 development of follow-up prodrugs.
69 In the current study we analysed the proteomic profiles of Mtb cells in the dormant state (after 4
70 months of storage) in an in vitro dormancy model developed earlier (13). The advantages of this
71 model over the widely exploited model of oxygen depletion (Wayne’s model (8)) comprise: the cells
72 reveal a low metabolic activity (judging by the negligible rate of radiolabelled uracil incorporation,
- 3 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
73 low level of transmembrane potential and undetectable respiration rate (14) as well as antibiotic
74 tolerance); the cells demonstrate inability to grow on typical solid media without a procedure of
75 reactivation – which is considered to be an intrinsic stage of genuine dormancy (15), developing
76 therefore the state of ‘non-culturability’ (13).
77 In our recent publication, using 2D electrophoresis-based proteomics, we found that 1-year-old
78 dormant, ‘non-culturable’ cells contain a significant amount of diverse proteins, including enzymes
79 belonging to different biochemical pathways (16). However, this approach has evident limitations
80 connected, firstly, with only partial coverage of the whole Mtb proteome and, secondly, with the
81 inability to quantitatively estimate the differences in the abundancy of particular proteins between the
82 groups of cells analysed (active vs dormant cells). LC-MS-based proteomic technologies are ideally
83 suited for such comparative analysis, evading the above-mentioned problems of a 2D approach.
84
85 2 Materials and Methods
86 2.1 Bacterial strains, growth media and culture conditions
87 Inoculum was initially grown from frozen stock stored at −70 °C. Mtb strain H37Rv was grown for
88 8 days (up to OD600 = 2.0) in Middlebrook 7H9 liquid medium (HiMedia, India) supplemented with
89 0.05% Tween 80 and 10% ADC (albumin, dextrose, catalase) growth supplement (HiMedia, India).
90 One millilitre of the initial culture was added to 200 mL of modified Sauton medium containing (per
91 litre): KH2PO4, 0.5 g; MgSO4.7H2O, 1.4 g; L-asparagine, 4 g; glycerol, 2 mL; ferric ammonium
92 citrate, 0.05 g; citric acid, 2 g; 1% ZnSO4.7H2O, 0.1 mL; pH 6.0–6.2 (adjusted with 1 M NaOH), and
93 supplemented with 0.5% BSA (Cohn Analog, Sigma), 0.025% tyloxapol and 5% glucose. Cultures
94 were incubated in 500 mL flasks containing 200 mL of modified Sauton medium at 37 °C with
95 shaking at 200 rpm (Innova, New Brunswick) for 30–50 days, and pH values were periodically
96 measured. In log phase, the pH of the culture reached 7.5–8.0 and then decreased in the stationary
- 4 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
97 phase. When the medium in post-stationary phase Mtb cultures reached pH 6.0–6.2 (after 30–45 days
98 of incubation), cultures (50 mL) were transferred to 50 mL plastic tightly capped tubes and kept under
99 static conditions, without agitation, at room temperature for up to 5 months post-inoculation. At the
100 time of transfer, 2-(N-morpholino)-ethane sulphonic acid (MES) was added at a final concentration of
101 100 mM to dormant cell cultures to prevent fast acidification of the spent medium during long-term
102 storage.
103
104 2.1.1 Evaluation of cell viability by CFU and MPN assays
105 Bacterial samples were serially diluted in fresh Sauton medium supplemented with 0.05% tyloxapol;
106 three replicates of each sample from each dilution were spotted on agar plates (1.5%) with
107 Middlebrook 7H9 medium (HiMedia, India), supplemented with 10% (v/v) ADC (HiMedia, India).
108 Plates were incubated for 30 days at 37 C, after which the number of CFU was counted.
109 The most probable number (MPN) assay is a statistical approach based on diluting a bacterial
110 suspension until reaching the point when no cell is transited into any well. The growth pattern in three
111 tubes is compared to the published statistical tables in order to estimate the total number of bacteria
112 (either viable or dormant) present in the sample (17).
113 The MPN assay was carried out in 48-well plastic plates filled with 0.9 mL of special medium for the
114 most effective reactivation of dormant Mtb cells. This medium contains 3.25 g of nutrient broth
115 (HiMedia, India) dissolved in 1 L of a mixture of Sauton medium (0.5 g KH2PO4; 1.4 g
116 MgSO4.7H2O; 4 g L-asparagine; 0.05 g ferric ammonium citrate; 2 g sodium citrate; 0.01% (w/v)
117 ZnSO4.7H2O per litre, pH 7.0), Middlebrook 7H9 liquid medium (HiMedia, India) and RPMI
118 (Thermo Fisher Scientific, USA) (1 : 1 : 1) supplemented with 0.5% v/v glycerol, 0.05% v/v Tween 80
119 and 10% ADC (HiMedia, India) (16). Appropriate serial dilutions of the cells (100 µL) were added to
120 each well. Plates were sealed and incubated at 37 C statically for 30 days. The MPN values, based on
- 5 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
121 the well pattern originating from the visible bacterial growth at the corresponding dilution point, were
122 calculated (17).
123
124 2.2 Preparation of samples for proteome analysis
125 Active and dormant cells were prepared in three biological replicates. Bacteria were harvested by
126 centrifugation at 8000 g for 15 min and washed 10 times with a buffer containing (per litre) 8 g NaCl,
127 0.2 g KCl and 0.24 g Na2HPO4 (pH 7.4). The bacterial pellet was re-suspended in ice-cold 100 mM
128 HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulphonic acid) buffer (pH 8.0) containing complete
129 protease inhibitor cocktail (Sigma, USA) and PMSF (phenylmethanesulphonyl fluoride) then
130 disrupted with zirconium beads on a bead beater homogenizer (MP Biomedicals FastPrep-24) for 1
131 min, 5 times for active cells and 10 times for dormant cells. The bacterial lysate was centrifuged at
132 25,000 g for 15 min at 4 °C. To maximize protein isolation, SDS (2% w/v) extraction was carried out.
133 The extracts were precipitated using a ReadyPrep 2D Cleanup kit (Bio-Rad, USA) to remove ionic
134 contaminants such as detergents, lipids and phenolic compounds from protein samples.
135
136 2.3 In-solution digestion
137 The protein extracts were dissolved in 20 μL of 6 M urea in 0.1 M ammonium bicarbonate. The
138 protein concentration was measured using a BCA Protein Assay Kit (Thermo Fisher Scientific,
139 Waltham, MA, USA). An equal amount of the total proteins was used for further trypsinolysis and
140 profiling.
141 Proteins were reduced using TCEP at 25 °C for 45 min, following alkylation with iodoacetamide
142 (IAA) in the dark for 30 min, having final concentrations of 5 mM and 55 mM, respectively. Excess
143 IAA was quenched with 1,4-dithiothreitol, then 50 mM ammonium bicarbonate was added to a total
144 volume of 200 μL. Trypsin was added in a protein-to-trypsin ratio of 50 : 1, and tryptic digestion was
- 6 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
145 done at 37 °C overnight. The reaction was stopped by adding formic acid (FA) to a final
146 concentration of 5%. The peptide mixtures obtained were pre-fractionated and purified using a C18
147 Empore™ disk (3M, St. Paul, USA) as described in (18).
148
149 2.4 LC-MS/MS analyses
150 Prior to analysis, peptides were dissolved in 20 µL of 3% ACN/0.1% FA. The injection volume was
151 1 µL with a flow rate of 5 µL/min. For the mobile phases, 0.1% FA in MS-safe water (A) and 0.1%
152 FA in 100% ACN (B) were used.
153 Analysis of peptides was done on a Synapt G2-Si High Definition Mass Spectrometer employing T-
154 wave ion mobility powered mass spectrometry coupled to an ACQUITY UPLC M-Class System
155 (Waters) using data-independent acquisition.
156 The UPLC system was equipped with a trapping column (nanoEase MZ Symmetry C18 Trap
157 Column, 180 µm × 20 mm, 5.0 µm particle diameter, 100 Å pore size, Waters) and an analytical
158 column (nanoEase MZ HSS T3 Column, 75 µm × 100 mm, C18, 1.8 µm particle diameter, 100 Å
159 pore size, Waters). The sample was loaded onto the trapping column for 2 min and transferred to the
160 analytical column. The peptides were eluted from the analytical column with a flow rate of 0.4
161 μL/min with a linear gradient of increasing concentration of mobile phase B from 5% to 35% for 70
162 min. Within the nanoFlowTM ESI-source, a capillary voltage of 2.5 kV was used, the sampling cone
163 voltage was 40 V and the source offset was 80 V. The source temperature was set to 80 °C. In the
164 trapping, ion mobility and transfer chambers, the wave velocity of the TriWave was set to 1000, 650
165 and 175 m/s, respectively, and ramping wave heights of 40, 40 and 4.0 V were used, respectively. In
166 the HDMSE mode, the collision energy was ramped to 22–45 eV. Data acquisition was carried out
167 using MassLynx software (Waters). HDMSE mass data were acquired in low and high energy modes.
168 Raw data were noise-reduced using the Noise Compression Tool (Waters).
- 7 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
169 The acquired data were submitted for processing and database searching using Progenesis software
170 against the database prepared in-house containing protein sequences from Mycobacterium
171 tuberculosis strain ATCC 25618/H37Rv downloaded from the UniProt database (version 20190304)
172 supplemented with sequences of common contaminants (Max Planck Institute of Biochemistry,
173 Martinsried, Germany). The low-energy and high-energy threshold values were set to 150 and 50
174 counts, respectively. The parameters used for the database search were: enzyme specificity: trypsin,
175 allowed missed cleavages: 2, fixed modification: carbamidomethylation (Cys), variable
176 modifications: N-terminal protein acetylation, and oxidation (Met); the false discovery rate was 1%;
177 minimum fragment ion matches per peptide, minimum fragment ion matches per protein and
178 minimum peptides per protein were set to 1, 4 and 2, respectively. Proteins were considered as
179 significant if they were identified in at least two independent biological replicates.
180
181 2.5 Statistical analysis
182 Data analysis and visualization were carried out in the R environment, applying scripts developed in-
183 house.
184
185 2.5.1 Data normalization
186 Before the analysis, the total protein concentration was quantified and an equal amount of protein
187 extracts was used for further trypsinolysis and profiling – pre-instrumental normalization on total
188 protein concentration was carried out (19). To make the profiles comparable for the subsequent
189 statistical analysis, post-instrumental pre-treatment methods were carried out: to deal with
190 heteroscedasticity and to make skewed distributions symmetric, log2 transformation was performed.
191 To force the samples to have the same distributions for the protein intensities, quantile normalization
192 was carried out (20).
- 8 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
193
194 2.5.2 Principle component analysis and hierarchical clustering
195 For illustration of intragroup and between-group variances, principle component analysis (PCA) was
196 carried out, applying the preliminarily normalized, log2-transformed and Pareto-scaled datasets (21–
197 23). Similarity/dissimilarity between the groups was determined by a method of hierarchical
198 clustering using the method of Ward (24).
199
200 2.5.3 Z-score calculations and heatmap visualization
201 A pre-treated protein matrix, consisting of proteins arranged in rows and groups of samples arranged
202 into columns, was centred and scaled (25). The centring represents the difference between the value
203 of a protein’s abundancy and the total mean of the protein. The scaling is the result of division of the
204 scaled metabolite by its . The resulting Z-scores, being in a similar range for the palette of proteins,
205 make feasible comparison of the variances between the variables (proteins).
206
207 2.5.4 Statistical significance analysis
208 The corresponding p-values of comparison of the two groups of cells (dormant vs active) were
209 calculated from a two-sided t-test. Since the data matrix was initially log2-transformed, the resulting
210 log2(FCdorm/mean) is the difference between the log2(meandorm) and log2(meanact).
211
212 2.6 Subtractive proteomic analysis of mycobacterial proteins prevailing in dormancy with
213 antigenic properties
214 The basic principles of subtractive analysis developed previously were used in the current work [26].
215 The significantly changed proteins revealing in dormancy a positive fold change (FC) were subjected
216 to subtractive analysis. In the first step, the paralogue proteins were identified using a CD-HIT online
- 9 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
217 platform (28, 29). The proteins with 60% identity were considered to be paralogous and discarded
218 from the further analytical steps. In the next step, the BLASTp procedure was applied to select non-
219 homologues to Homo sapiens proteins in the query set, using a threshold expectation value of 0.001
220 and similarity below 35%. The resulting list of proteins was further categorized according to their
221 subcellular localization using the CELLO online platform (30).
222 The shortened list of proteins was further compared with previously published datasets with annotated
223 Mtb proteins with antigenic properties: i) a list of 181 proteins detected exclusively in the sera of TB-
224 positive patients from a study on the dynamic antibody response to the proteome (31); ii) a list of 62
225 proteins resulting from serodiagnosis of latent Mtb infection (32); iii) a list of 22 proteins resulting
226 from the analysis of TB-specific IgG antibody profiles, detectable on two different platforms
227 (Luminex and MBio) (33).
228
229 2.7 Annotation of the list of enzymes – prodrug activators
230 KEGG API was used to prescribe EC numbers to the initially prescribed UniProt KO code (34).
231
232 3 Results
233 3.1 Dormant mycobacterial cells subjected to the proteomic analysis and the results of
234 proteomic profiling
235 Dormant Mtb cells subjected to the further proteomic profiling bore the typical traits of ‘non-
236 culturability’ (13, 16), particularly incapability to grow on rich solid medium: the experimentally
237 estimated viability by the CFU method was 1.32 ± 0.48 104 cells/mL (± SE) for dormant cells and
238 1.17 ± 0.44 108 cells/mL (± SE) for active cells. At the same time, estimation of the total number of
239 viable but ‘non-culturable’ cells after resuscitation in fresh liquid medium by the MPN assay resulted
240 in 1.21 ± 0.65 109 cells/mL (± SE) for dormant cells and 0.97 ± 0.44 109 cells/mL (± SE) for
- 10 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
241 actively growing cells (Fig. 1). Before the analysis, the total protein concentration was quantified as
242 3.56 ± 0.24 mg/g wet weight (± SE) for dormant cells and 6.01 ± 0.14 mg/g wet weight (± SE) for
243 active cells. An equal amount of the total proteins was used for further trypsinolysis and LC-MS/MS
244 profiling (19). LC-MS profiling resulted in the detection of 1379 proteins (whose encoding genes Rv
245 were further used as identifiers), covering 35% of the coding M. tuberculosis H37Rv genome.
246 However, the raw instrumental data showed skewed distributions of protein intensity, tending to
247 lower values (Fig. 2A).
248 For the subsequent statistical analysis, post-instrumental pre-treatment methods were carried out: to
249 deal with heteroscedasticity and to make skewed distributions symmetric, log2-transformation was
250 performed. To force the samples to have the same distributions for the protein intensities, quantile
251 normalization was carried out (20), resulting in comparable normally distributed profiles with equal
252 medians, making the protein profiles eligible for further statistical analyses (Fig. 2B).
253
254 3.2 PCA and hierarchical clustering
255 The results of PCA (23) depict ca. 80% of all variance in the protein profiles. To perform PCA, the
256 data matrix was preliminary Pareto-scaled (22). The results demonstrate that samples are clearly
257 distinguished by their physiological state (‘dormant’ vs ‘active’) (Fig. 3). Hierarchical cluster
258 analysis, based on calculations of the amount of within-cluster dissimilarity analogously results in two
259 distinct clusters, separating dormant cells from active ones. Ward’s minimum variance method was
260 used for the calculations (24) (Fig. 3B).
261
262 3.3 Differential analysis
263 The results of differential analysis are listed in the Supplement and graphically depicted as a volcano
264 plot in Fig. 4A. Generally, the abundancy of 894 proteins (out of 1379) was found to change
- 11 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
265 significantly (p < 0.05) in the transition to the dormant state – among them 468 with a positive FC
266 (‘stored in dormancy proteins’) and 426 with a negative FC (‘decreased in dormancy proteins’). The
267 rest of the proteins (485) lay beyond the scope of differential analysis – there was greater variance in
268 their fluctuation in transition to dormancy and consequently their changes were insignificant. Analysis
269 of the first 30 proteins stored significantly differently in dormancy, bearing the maximum log2FC,
270 discloses a cohort of regulatory proteins and virulence factors (and not enzymatic proteins, dictating
271 biochemical adaptation to the stress conditions), despite the in vitro nature of the dormancy model
272 (Fig. 4B). According to the main stress factor – pH decrease – we observed upregulation of the
273 transcriptional regulator phoP (Rv0757, p < 0.05, log2FC = 1.43) of the two-component phoP–phoR
274 system, conferring the fitness of mycobacteria to the lowered pH values [16,35,36]. In addition,
275 phoPR controls secretion of the antigens Ag85A (Rv3804c, p < 0.05, log2FC = 1.28) and Ag85B
276 (Rv0129c, p > 0.05, log2FC = 1.11) as well as ESAT-6 EsxA (Rv3875, p < 0.05, log2FC = 4.14) – a
277 secreted strong human T-cell antigen protein that plays a number of roles in modulating the host’s
278 immune response to infection (37) (Fig. 4C). Tightly associated with the ESX-1 and ESX-5 secretion
279 systems are proteins of the PE/PPE family (38). Out of 95 KEGG-annotated PE/PPE proteins we
280 detected six: PPE18 (Rv1196, p < 0.05, log2FC = 2.44), PPE19 (Rv1361c, p < 0.05, log2FC = 2.11),
281 PPE32 (Rv1808, p < 0.05, log2FC = 3.15), PPE60 (Rv3478, p < 0.05, log2FC = 1.05), PE15 (Rv1386,
282 p < 0.05, log2FC = 3.64) and PE31 (Rv3477, p > 0.05, log2FC = 0.39). The low discovery rate may be
283 explained by the paucity of trypsin-cleavage sites in these proteins and their structural homology (39).
284 For the proteins of this class, a variety of functions is assumed: many PE and PPE genes have been
285 shown to be upregulated in starvation and during stress conditions (40). A chaperone function of PE
286 proteins has been reported, particularly in transporting of MPT64 antigen protein (Rv1980c, p < 0.05,
287 log2FC = 0.69) (41). Additionally, the phoP transcriptional regulator has been shown to downregulate
288 devR (Rv3133c, p > 0.05, log2FC = 0.01), a regulator of the DosR ‘dormancy’ regulon (35, 42).
- 12 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
289 Correspondingly, out of ca. 50 postulated genes of the DosR regulon, only 19 proteins were detected
290 in the current study, 14 of which were found significantly stored in the dormant state (Fig. 4D).
291 Similarly, we observed a down-shift of a number of other transcriptional regulatory proteins, e.g.,
292 Rv0275c, Rv1556, Rv1019, Rv0818, Rv3058c, Rv3249c, etc. Analogously, in the absence of
293 translational processes, the abundancy of translation initiation factors (Rv3462c, Rv2839c) as well as
294 30s and 50s ribosomal proteins (Rv0717, Rv2875c, Rv0641) was similarly decreased.
295 There was no consistency in behaviour of proteins of the ABC trehalose transporter complex LpqY–
296 SugA–SugB–SugC (Rv1235, Rv1236, Rv1237, Rv1238): out of four annotated in the Mtb genome
297 only Rv1235 and Rv1238 were detected in the current study, albeit with a different sign of FC in
298 dormancy, rather indicating a decrease in their abundancy and efficacy in the current stress conditions
299 and confirming the dependency of the dormant cells on functioning trehalose to replenish intracellular
300 pathways (43). On the contrary, the components of the phosphate ABC transporter pstS1-3 (Rv0928,
301 Rv0934, Rv0932c) were found concordantly to be stored in dormancy, indicating a possible
302 dependency of dormant cells on extracellular sources of inorganic phosphate. Part of the ABC efflux
303 pump Rv1218c, conferring the cells the resistance to a wide range of drugs, was found to be similarly
304 stored.
305 For the analysis of biochemical adaptation to the transition to dormancy, the 666 detected enzymes
306 (both with positive and negative log2FC, based on KEGG orthology annotation) were placed on the
307 metabolic map and the corresponding processes which they catalyse were highlighted: in orange for
308 those enzymes with a positive FC (enzymes stored in dormancy) and in blue for those demonstrating a
309 negative FC (a relative decrease in dormancy). The width of the highlighted lines corresponds to the
310 p-value (the thicker lines are for p < 0.01, the thinnest lines are for insignificant results (p > 0.05))
311 (Fig. 5A). A glance at the general metabolic visualization (Fig. 5A) gives an impression that, in
312 dormancy, a scattered ‘mosaic’ intactness of enzymes can be seen: the flawless functionality of the
- 13 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
313 many pathways is hindered by enzymes significantly decreased in dormancy, e.g., the functionality of
314 the urea cycle may be broken because of the decreased level of argininosuccinate synthase (Rv1658,
315 p < 0.05, log2FC = −0.69). To annotate precisely the metabolic pathways differentially preserved in
316 dormancy, pathway enrichment analysis was carried out on the enzymatic proteins with FC > 0 and
317 p < 0.05 (44, 45) (Fig. 5B). Detailed pathway analysis revealed nine upregulated proteins of TCA;
318 however, isocitrate dehydrogenase (Rv0066c, p < 0.05, log2FC = −0.41) and subunits of -
319 ketoglutarate dehydrogenase (Rv2454 and Rv2455) demonstrated negative values of log2FC,
320 impeding the normal flow of the TCA cycle (Fig. 5C). The subunits of membrane-bound fumarate
321 reductase were not technically detected; however, subunits of the succinate dehydrogenase complex
322 (Rv3318 and Rv0247c) were detected, potentially enabling the reverse flow of the processes in the
323 ‘reductive branch’ of the TCA cycle (part of the Arnon–Buchanan cycle) from pyruvate to succinate,
324 providing the cell with reoxidized NAD+. The intactness of enzymes acting in the -oxidation of fatty
325 acids may provide the cell with a surplus of acetyl-CoA (in the case of oxidation of even-chained fatty
326 acids) and propionyl-CoA (in the case of oxidation of odd-chained fatty acids). Dormant Mtb cells
327 demonstrated the presence of enzymes of the methylcitrate cycle, which may allow the conversion of
328 propionyl-CoA and oxaloacetate to succinate and pyruvate. The latter can be converted to fumarate by
329 the detected succinate dehydrogenase (Rv3318, Rv0247c). If this pathway is functioning, the
330 accumulation of organic acids (contributing to sustainment of the energized level of transmembrane
331 potential in dormancy (46, 47)) in the cells and medium should be expected; that, however, was found
332 experimentally for dormant cells of M. smegmatis, obtained in the same model of self-acidification of
333 growth medium (14). Contributing to the accumulation of these acids and to detoxification of surplus
334 acetyl-CoA is the glyoxylate pathway, whose enzymes isocitrate lyase (Rv0467) and malate synthase
335 (Rv1837c) were found to be accumulated in dormant cells.
- 14 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
336 The classical flow of the glycolytic pathway is hampered by deficient glyceraldehyde-3-phosphate
337 dehydrogenase (Rv1436, p < 0.05, log2FC = −0.43) and phosphoglycerate kinase (Rv1437, p < 0.05,
338 log2FC = −0.61); however, there is a bypass, consisting of the formation of glycerone phosphate,
339 which can further spontaneously convert into methylglyoxal, which in further detoxification pathways
340 leads to pyruvate through lactate formation. Similarly, all enzymes of glycerol metabolism were
341 detected – the catabolism of glycerol or glycerol-3-phosphate is conjugated with the toxic carbonyl
342 compounds methylglyoxal and lactaldehyde (the latter was detected experimentally for M. smegmatis
343 (14)). However, functioning aldehyde dehydrogenases (Rv0147, Rv2858c) or glyoxylase (Rv0577)
344 detoxify these intermediates, ultimately to lactate. The further transformation of forming lactate by a
345 quinone-dependent lactate dehydrogenase finally supplies the cells with pyruvate capable of further
346 catabolism in the TCA, glyoxylate and methylcitrate pathways. Mtb used to have annotated two
347 lactate dehydrogenases, lldD2 (Rv1872c, p < 0.05, log2FC = 1.04) and lldD1 (Rv0694, p < 0.05,
348 log2FC = 2.67); however, lactate dehydrogenase activity was confirmed for lldD2 only (48). On the
349 other hand, upregulation of Rv0694 was observed in the current conditions of dormancy in glycerol-
350 glucose medium. Rv0694 is a member of the Rv0691–Rv0694 gene cluster, involved in biosynthesis
351 of a recently discovered peptide-derived electron carrier – mycofactocin (49).
352 The decreased detection of particular enzymes may be due to the preservation of proteolytic
353 peptidases abundant in the dormant proteome: Rv0319 (p < 0.05, log2FC = 1.1), Rv0457c (p < 0.05,
354 log2FC = 0.82), Rv3883c (p < 0.05, log2FC = 2.53) and Rv0983 (p < 0.05, log2FC = 0.64).
355 The enzymes of glycerol catabolism that are stored unprocessed in dormancy contribute to lipid
356 metabolism and may serve as a source of reoxidation of reductive equivalents (Fig. 5C). The gapped
357 glycolytic pathway fails to serve substrate-level ATP generation (if any): however, accumulated
358 polyphosphate (the transport of inorganic phosphate remains intact) may serve a phosphagen – a
- 15 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
359 reservoir of phosphoryl groups, that can be used to generate ATP, thus we detected a reversible ppk1
360 (Rv2984, p < 0.05, log2FC = 0.57).
361 Additionally, we were able to reconstruct a de novo unimpaired uridine monophosphate (UMP)
362 pathway, serving the precursor of all pyrimidine nucleotides.
363 Similarly to the known involvement of protective redox systems as a response to ROS attacks in
364 macrophages, upregulation of superoxide dismutase was observed in our dormancy conditions (sodA,
365 Rv3846, p < 0.05, log2FC = 1.04). Since mycobacteria lack glutathione, the ubiquitous low-molecular
366 weight thiol in other organisms, Mtb is dependent on mycothiol for antioxidant activity; the enzymes
367 encoding its biosynthesis were found to be stored in the dormant state (Fig. 5).
368 To the oxidative stress response system belong F420-dependent glucose-6-phosphate dehydrogenase
369 fgd1 (Rv0407, p > 0.05, log2FC = 0.71) and F420H2-dependent menaquinone reductase Rv1261c (p >
370 0.05, log2FC = 0.52), contributing to the protection of bacteria from the formation of toxic
371 semiquinones (50). In shaping the response to ROS and RNS, the accumulation of the regulator sigH
372 was observed (Rv3223c, p < 0.05, log2FC = 0.48).
373
374 3.4 Annotation of immunogenic proteins accumulated in dormant cells
375 It was interesting to search for immunogenic proteins among detected proteins with a positive FC in
376 dormant Mtb, as they are potentially applicable for latent TB diagnosis. In the current study we
377 enriched the possible candidates, based on the subtractive proteomic concept [26]. The essence of this
378 approach presumes the identification of non-paralogue proteins followed by the identification of non-
379 homologues to H. sapiens proteins and the analysis of subcellular localization (Fig. 6A). Thus, the
380 initially applied 468 proteins significantly represented in dormancy with a positive FC resulted in 273
381 non-homologue proteins after the first two filters, further localization analysis of which allowed the
382 identification of 53 membrane proteins, 175 cytoplasmic proteins and 45 extracellular proteins.
- 16 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
383 The short-listed data set was compared with the reference lists from published works on studies of the
384 sera of human TB patients (31–33).
385 The comparative analysis reference lists comprised the most reactive proteins (in total 265 antigens)
386 as described in the Material and Methods.
387 Comparative analysis of the proteins in these reference sets with the selected proteins in the current
388 study with a positive FC in dormancy revealed 40 potentially immunogenic proteins specific for
389 dormancy (Fig. 6A). Remarkably, these studies (including our data set) demonstrate only one shared
390 protein, Rv1284 – a carbonic anhydrase (p < 0.05, log2FC = 0.38), which was previously annotated
391 among the latency-related antigens of Mtb (51) (Fig. 6B).
392
393 3.5 Annotation of proteins accumulated in dormancy – potential prodrug activators
394 The accumulation of a significant number of enzymatic proteins in the dormant state may imply the
395 potency of ‘non-culturable’ Mtb cells to perform biotransformation reactions.
396 Analysis of the distribution of enzyme classes in dormancy revealed the preservation of a bulk bundle
397 of oxidoreductases and hydrolases (Fig. 7A). The list of oxidoreductases (73) demonstrated in
398 dormancy with an FC > 0 is summarized in the Supplement. The majority of the annotated enzymes
399 of this class are NAD- and FAD-dependent, bearing relatively low standard reduction potentials
400 (−0.32 and −0.22 V correspondingly). There are, however, two F420-dependent enzymes, Rv0407
401 (F420-dependent glucose-6-phosphate dehydrogenase Fgd1) and deazaflavin (F420)-dependent
402 nitroreductase (Ddn), to detect. F420-dependent enzymes are known to bear a lower reduction potential
403 (−0.32 V) and have been shown to participate in bicyclic nitroimidazole compounds (50, 52).
404 Similarly, the FAD-dependent decaprenylphosphoryl-beta-D-ribose-2’-epimerase (catalysing the
405 conversion of decaprenyl-phosphoryl-D-ribose to decaprenyl-phosphoryl-D-arabinose – the sole
- 17 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
406 arabinose precursor in the pathway of arabinogalactan biosynthesis) is capable of nitroreducing
407 recently developed benzothiazinones (BTZ) (53).
408 The component of the non-proton pumping NADH:quinone oxidoreductase (Rv0392c, p < 0.05,
409 log2FC = 3.81) accumulated in dormancy is hypothesized to reduce clofazimine, an antibiotic whose
410 mode of action is mediated through ROS formation (54).
411 Hydrolytic enzymes (esterases and lipases) are considered to play a crucial role in mycobacterial
412 persistence (55, 56). Among the enzymes detected with a positive FC in the dormant state, 48 could
413 be classified as hydrolytic enzymes (Supplement). This list was compared with a list of annotated
414 serine hydrolases and esterases (57) (Fig. 7B). In the query data set of enzymes with a positive FC,
415 three enzymes potentially keep their activity in dormancy (Rv2970c, Rv2224c and Rv1683) (57). The
416 list of enzymes of other EC classes can be found in the Supplement.
417
418 4 Discussion
419 The observation that the majority of proteins in dormancy are significantly abundant (in comparison
420 to the active state), and bear a positive FC, corresponds to the previously reported observation on the
421 ability of Mtb to preserve protein diversity in the dormant state (16). Comparison of the data
422 originating from 2D electrophoresis analysis of D1 cells (4.5 months of dormancy) (16) with the same
423 gene set from the current study (262 genes) after a common ranking procedure, followed by the
424 Mann–Whitney test, reveals similarity between the data sets (Ho: ULC-MS = U2D, p-value = 0.8894)
425 (Fig. 7). Similar to the previously published 2D proteome analysis, only a limited number of
426 accumulated proteins belong to the annotated DOS regulon, which demonstrates the differences
427 between the anoxic Wayne’s model and the current model of self-acidification in the post-stationary
428 phase. Analogously, the increased abundancy of Rv0757 – a component of the pH-dependent
- 18 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
429 regulator phoP which is upregulated in D1 and D2 dormancy states – may contribute to the low
430 discovery rate of proteins of the DOS system (16).
431 phoP is a virulence regulator which is conserved in a wide range of bacteria and controls the
432 transcription of more than 600 genes (58). Besides its function of regulating the transcription of
433 virulence factors (see the results of differential analysis - §3.3), it has been hypothesized that phoP
434 coordinates gene expression with changes in membrane potential, and encodes the proteins that
435 control the state of membrane energization (58). In Salmonella, phoP has been shown to control the
436 production of superoxide dismutase, SOD (59), which was found to be abundant in dormancy in the
437 current study. Correspondingly, the protective systems were detected in the current study –
438 accumulation of SODs (sodA – Rv3846, sodC – Rv0432) and the DNA-binding proteins hupB
439 (Rv2986) and iniB (Rv0341), as well as several chaperone proteins (DnaK – Rv0350, dnaJ1 –
440 Rv0352). However, 2D analysis, being qualitative in its nature, could not disclose the changes in
441 quantitative abundance of these proteins. Thus, e.g., catalase G (katG, Rv1908c) could be detected in
442 D2 (16), the abundancy of which in the dormant state in the current study, however, was slightly
443 decreased in comparison to the active state.
444 Moreover, the main drawback of the 2D approach is its inability to reach a meaningful coverage of
445 biochemical processes – thus, GO enrichment analysis based on the enzymes detected in D1 (16)
446 revealed only three statistically feasible pathways. Manual reconstruction of the glycolysis and TCA
447 pathways demonstrated only a limited coverage of these pathways. Moreover, those remaining
448 undisclosed were the transformation of glycerol, propionyl-CoA metabolism and the glyoxylate
449 pathway (Fig. 8).
450 The high-throughput LC-MS-based reconstruction of central carbon metabolism significantly
451 increased the coverage of metabolic processes, and additionally disclosed mycothiol and a de novo
452 UMP biosynthetic pathway (Fig. 5). The annotated pathways resemble the adjustment of
- 19 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
453 mycobacterial metabolic machinery for life inside lipid-rich macrophages, particularly involvement of
454 the methylcitrate and glyoxylate pathways to utilize the excess acetyl- and propionyl-CoA (resulting
455 from the catabolic oxidation of fatty acids, cholesterol and branched amino-acid catabolism) (60–63).
456 It is worth noting that such biochemical adaptation is tightly linked to mycobacterial virulence; thus,
457 isocitrate lyase, besides its biochemical functions of participation in the glyoxylate cycle, in the
458 methyl citrate cycle of propionyl-CoA metabolism or in succinate-associated generation for
459 maintenance of proton-motive force in dormancy (14, 64), is an important mycobacterial virulence
460 factor (60, 65). The decrease in virulence in icl mutants is precisely related to the enzymatic function
461 of Icl and reflects the importance of the glyoxylate shunt for bacterial life in vivo.
462 The prevalence of catabolic enzymes in the absence of respiration may result in the generation of an
463 excess of reduced FADH2 and NADH, which should be reoxidized for the continuation of metabolism
464 (66). In the absence of respiration, a proposed mechanism for NAD+ replenishment and maintaining
465 the energized level of transmembrane potential consists of exploiting the reductive branch of the TCA
466 cycle (a part of the Arnon–Buchanan cycle) (47, 64), which is corroborated by the recently published
467 data on succinate accumulation in the same model by dormant M. smegmatis cells (14).
468 Nonetheless, the current data on relative protein changes may provide only speculative information on
469 real flux flows inside the cells (67), albeit speculatively envisaging metabolic adaptation of the cells
470 to persistence alongside the activation of virulence factors, favouring the host’s immune escape. At
471 the same time, enzymes stored in dormancy that are involved in particular metabolic pathways may be
472 exploited by dormant cells during reactivation, when macromolecular biosynthetic processes are
473 initiated (68). For example, the salvage of enzymes involved in the metabolism of purine and
474 pyrimidine, the key precursors of DNA and RNA, in the current study (similarly to the observation of
475 accumulation of the corresponding metabolites in the metabolism of dormant M. smegmatis (14))
476 makes these precursors immediately accessible after the onset of reactivation.
- 20 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
477 At the same time, individual enzymes preserved in dormancy can retain functionality, in particular
478 those with cytoprotective functions.
479 Thus, under excess exogeneous glycerol one would expect the accumulation of carbonyl compounds,
480 e.g., lactaldehyde (14) or methylglyoxal (Fig. 5). Therefore, it is to be expected that mycobacteria
481 have developed mechanisms of protection from such reactive electrophilic agents.
482 Functioning NAD(P)H-dependent aldehyde dehydrogenases ensure the flow of metabolic
483 transformations of carbonyl compounds and, on the other hand, may represent valuable targets for TB
484 drug discovery (69). The glyoxylase Rv0577, presumably ensuring the detoxification of
485 methylglyoxal, has similarly been considered as a target for pyrimidine imidazoles (70).
486 Whereas a majority of organisms (including human beings) exploit glutathione as a carbonyl
487 detoxification system, mycobacteria lack glutathione, being dependent on mycothiol, a conjugate of
488 N-acetylcysteine with myo-inosityl-2-amino-deoxy-α-D-glucopyranoside (71). Aside from its
489 antioxidant properties, mycothiol may act as a cofactor for aldehyde dehydrogenase and participate in
490 detoxification of rifampicin and isoniazid (71, 72). Correspondingly, pathway enrichment analysis
491 disclosed the salvage of biosynthetic enzymes of mycothiol biosynthesis (Fig. 5E).
492 Part of the adaptation to oxidative stress is the accumulation of F420-dependent enzymes: glucose-6-
493 phosphate dehydrogenase – fgd1 (Rv0407) and F420H2-dependent quinone reductase (deazaflavin-
494 dependent nitroreductase, Rv1261c). The first passes electrons from the resulted in gluconeogenesis
495 glucose-6-phosphate on F420H2, which further participates in specific two-electron reduction of
496 quinones by the F420H2-dependent quinone reductase (deazaflavin-dependent nitroreductase,
497 Rv1261c), preventing the one-electron reduction pathway and the formation of cytotoxic
498 semiquinones, notorious for formation of superoxide radical. The evolution of this self-protective
499 mechanism was necessary, considering menaquinone as a main quinone of mycobacteria, which due
500 to its more negative redox potential than ubiquinones tends to generate superoxide (73).
- 21 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
501 Thus, alongside the individually functioning SODs, catalases and chaperones, mycobacterial cells are
502 armoured with efficient defence systems, protecting the bacteria against various stress factors.
503 Therefore, the current model of self-acidification of Mtb in the post-stationary phase being an in vitro
504 model, it mirrors the virulence and biochemical adaptation of these bacteria to in vivo persistence.
505 Comparative analysis of proteins the stored in dormancy found in the present study with the recently
506 established immunogenic Mtb proteins resulted in 40 proteins with immunogenic properties (Fig. 6).
507 Despite the serodiagnosis of latent TB being very obscure, the work of Zhou et al. (32), where
508 serodiagnostic analysis was carried out using the sera of patients with confirmed latent TB at the
509 proteome level, is of particular interest. Among 62 seropositive TB proteins reported by Zhou et al.,
510 12 proteins out of 40 were found in the current study. Additionally, experiments recently carried out
511 on the sera of 42 TB patients allowed the detection of a set of 27 potentially immunoreactive proteins
512 (74), six of them (Rv2623, Rv2018, Rv2145c, Rv2744c, Rv1837c, Rv0341) similarly detected in the
513 curent study with a positive FC in the dormant state. It is to be anticipated that the list of selected
514 antigenic proteins includes those that may be of interest for identifying the specific
515 ‘immunosignature’ of latent TB, which can later be used for immunodiagnosis of latent TB. However,
516 this assumption requires further experimental verification.
517 The currently available therapy with the commonly exploited ‘first- and second-line’ TB drugs is
518 directed at a rather narrow spectrum of Mtb targets, controlling the processes of DNA replication and
519 transcription translation or targeting the enzymes of cell-wall biosynthesis/remodelling. However, all
520 these ‘targets’ require the active flow of physiological pathways in the bacterial cells, sparing
521 therefore the population of metabolically inactive dormant cells. Therefore, accounting for the urgent
522 need for new chemicals efficient against drug-resistant mycobacteria and drug-tolerant dormant
523 populations, it is an attractive idea to consider the enzymes available in dormancy capable of
- 22 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
524 biotransformation, which could activate prodrugs and result in the accumulation of substances with
525 general toxicity (ROS, RNS and chemically reactive metabolites and antimetabolites) (11, 12).
526 Some prodrugs such as clofazimine or delamanid which are activated by intracellular specific
527 cofactor-dependent enzymes may serve as an illustration of this approach. The first molecule
528 generates intracellular ROS through NADH dehydrogenase (54). The latter undergoes a metabolic
529 conversion by mycobacterial deazaflavin-dependent nitroreductase followed by the release of
530 intracellular NO, eradicating therefore a dormant Mtb subpopulation (75, 76).
531 Knowledge of the rare cofactors and cofactor-dependent oxidoreductases accumulated in dormancy
532 may contribute to the development of new redox compounds (Supplement). Additionally, the
533 bioactivation of prodrug compounds in hydrolytic processes, saved intact in dormancy (Supplement),
534 is potent for the development of new anti-latent TB chemicals. Preliminary success of this concept has
535 been demonstrated for nitazoxanide and methotrexate, albeit the particular bioactivating hydrolases
536 were not disclosed (11). Malfunctioning of the reactive carbonyl detoxification systems under a
537 surplus of glycerol may lead to the accumulation of (methyl)glyoxal/lactaldehyde in dormant
538 mycobacteria (14). In vivo, a fat-rich diet may result in the accumulation of similar toxic carbonyl
539 compounds within the cells: glyceraldehyde, glyceronphosphate, methylglyoxal, crotonaldehyde,
540 malondialdehyde, 4-hydroxynonenal and, therefore, the application of inhibitors of the carbonyl
541 detoxifying process may lead to inactivation of the bacteria (77, 78). On the other hand, lipid
542 catabolism may elicit ‘ketone body’ intracellular acidosis (in the case of surplus acetyl-CoA and
543 malfunction), in particular of malate synthase (Rv1837c) (79). A possible malate synthase inhibitor,
544 mimicking the structure of methyl glyoxal – phenyl-diketo acid inhibitor – was recently proposed (80)
545 and its applicability for the inactivation of dormant mycobacteria looks promising.
546
- 23 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
547 In conclusion, the current study demonstrated the biochemical adaptation of Mtb to the conditions of
548 ‘non-culturability’ and visually demonstrated involvement of the pH-dependent regulator phoP in the
549 suppression of components of the DOS regulon and in the activation of a number of processes crucial
550 for cellular survival. Further, biochemical processes essential for the development and support of
551 dormancy (methylcitrate and glyoxylate pathways, the salvage of mycothiol and UMP biosynthetic
552 pathways as well as protective systems) were revealed. In the present study we, for the first time,
553 annotate enzymes significantly abundant in dormant Mtb cells (oxidoreductases and hydrolases),
554 which potentially may be further considered as biotransformatory enzymes – prodrug activators
555 suitable for the elimination of dormant mycobacteria.
556
557 Acknowledgements
558
559 This work was funded by the Russian Science Foundation – Grant 19-15-00324
560
561 Author Contributions
562
563 VN carried out statistical and bioinformatics analysis: analysed and visualized data, prepared the
564 figures, wrote the manuscript. MS – designed the project, supervised the project, provided the cells
565 and cellular extracts for the analysis; DL, FD, JS – carried out proteomic profiling; annotated
566 mycobacterial proteins. AK – supervised the project, corrected the draft.
567 The Authors read, commented and approved the final version of the manuscript.
568
- 24 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
569
CFU MPN 10
8
)
e
u
l
a
v
( g
o 6 l
4
akt dorm akt dorm 570
571 Fig. 1. Viability assays (CFU and MPN) of M. tuberculosis H37Rv cells subjected to further
572 proteomic analysis. Active cells were probed in the late log phase – the 10th day of cultivation.
573 Viability of the dormant cells was analysed after 4 months of storage after gradual acidification of the
574 bacterial culture in the post-stationary phase (81). Estimation based on probing of 6–9 independent
575 samples of each type of cell.
576
577
578
579
580
581
582
583
584
585
- 25 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
586
A B
80000
60000 dorm
e u
l 40000
a V
20000 akt
0 0 5 10 15 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ I_ _ _ _ I I I II II II _ _ _ I I I II II II Value ______t t t t_ t_ t_ _ _ _ rm rm rm m m m k k k k k k t t t o o o r r r rm rm rm a a a a a a k k k d d d o o o o o o a a a d d d d d d
587
588 Fig. 2. Normalization of proteomic profiles. A) Distribution of raw protein intensities; B) average
589 median intensity of nine samples (biological and technical replicates) for each physiological condition
590 after log2 transformation and quantile normalization of the protein profiles.
591
592
593
594
595
596
597
598
599
600
- 26 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A B
akt_II_2 akt_II_1 20 akt_II_3 akt_III_2 akt_II_3 dorm_III_3 akt_III_1 akt_II_2 akt_II_1 dorm_II_2 dorm_III_2 akt_I_2 akt_III_1 akt_III_2 akt_I_1 dorm_III_1
2 akt_I_2 akt_III_3
C dorm_II_3 0 akt_I_3
P dorm_I_3 dorm_II_1 dorm_III_3 dorm_I_1 dorm_III_2 dorm_I_2 dorm_III_1 akt_I_1 dorm_II_3 akt_III_3 −20 dorm_II_2 akt_I_3 dorm_I_2 dorm_I_1 dorm_II_1 dorm_I_3
−20 0 20 200 150 100 50 0 PC1 Distance
601 Fig. 3. Results of the principal component analysis (A) and hierarchical clustering (B) of M.
602 tuberculosis cells used for proteomic profiling. PC1 with the largest variance of each protein level
603 separates the samples into two distinct clusters. The elliptical boundary is analogous to the 95% CI for
604 both bivariate distributions. Ward’s method (24) of hierarchical clustering similarly demonstrates
605 segregation of the two types of physiological group.
606
607
608
609
610
611
612
613
614
615
- 27 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A B
Significance 20 p<0.05 p>0.05
Rv3855 15
Rv3137 Rv0634A Rv0838
) Rv0932c Rv0341 p
( Rv2604c Rv1689 Rv1198 g
l Rv1530
− 10 Rv0729 Rv1465 Rv1987 Rv1192 Rv1184c Rv2861c Rv2574 Rv1657 Rv1480 Rv2785c Rv2623 Rv1196 Rv2991 Rv2004c Rv0098 Rv0717 Rv1361c Rv0315 Rv2588c Rv1404 Rv2552c Rv3154 Rv2347c Rv1178 Rv0645c 5 Rv1546 Rv3502c Rv0818 Rv0064A Rv1159A Rv1113 Rv2069 Rv2313c
Rv0827c 0
−5 0 5 log2(FCdorm/act)
C D
Rv3133c devR; two component transcriptional regulator DevR Rv3870 eccCa1 Rv1998c hypothetical protein Rv3879c Rv3871 espK eccCb1
Rv3878 1 Rv1996 universal stress protein Rv3616c espJ espA Rv2006 otsB1; trehalose−6−phosphate phosphatase OtsB
Rv3872 Rv0571c hypothetical protein Rv1411c Rv0287 PE35 lprG esxGRv3881cRv3873 espB PPE68 Rv3615c Rv3841 bfrB; bacterioferritin BfrB espC Rv2629 hypothetical protein Rv3763 Rv0475 lpqH hbhA Rv3019c Rv2624c universal stress protein Rv3874 esxR Rv1860 Rv2875 esxB Z−score apaRv1926c mpt70 Rv1196 mpt63 PPE18 (−1,−0.8] Rv2031c hspX; alpha−crystallin Z−score Rv3803c Rv3875 fbpD Rv0934 1 pstS1 Rv1984c esxA (−0.8,−0.6] Rv2028c universal stress protein 0.5 cfp21 Rv0757 (−0.4,−0.2] Rv1980c phoP 0 (−0.2,0] Rv0350 mpt64 Rv2873 Rv0288 Rv1738 hypothetical protein Rv1270c mpt83 esxH Rv2608 Rv0125 −0.5 lprA dnaK Rv2660c pepA (0,0.2] PPE42 −1 Rv3418c Rv2660c (0.2,0.4] Rv2030c hypothetical protein
Rv0440groES Rv1886c 2 groEL2 Rv2031c fbpB (0.4,0.6] hspX Rv3127 hypothetical protein Rv0667 (0.6,0.8] rpoB (0.8,1] Rv2004c hypothetical protein Rv3846 sodA NA Rv0005 Rv0447c gyrB Rv2032 acg; NAD(P)H nitroreductase Rv3133c ufaA1 devR Rv2029c Rv2957 Rv2623 TB31.7; universal stress protein pfkB Rv2957 Rv0001 Rv2626c Rv2626c hrp1; hypoxic response protein dnaA hrp1 Rv3245c Rv2958c mtrB Rv2958c Rv3130c tgs1; diacyglycerol O−acyltransferase Rv2032 acg Rv2005c universal stress protein Rv3244c lpqB m kt Rv3246c Dor A mtrA Rv2623 TB31.7
616 Fig. 4. General statistical results of comparative proteomic analysis of dormant vs active
617 M. tuberculosis cells. A) Results of differential comparison of dormant cells vs active cells reveal 894
618 statistically significantly changed proteins, among them 468 proteins prevailing in dormancy (with a
619 positive FC value) and 426 proteins whose abundancy decreased in the transition to dormancy (with a
620 negative FC value); B) visualization of the 30 proteins most abundant in dormancy with the maximum
621 FC; C) STRING-based interaction network of phoP and devR regulators of DosR regulon.
- 28 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
622 Upregulation of pH-dependent regulator phoP suppresses devR, a regulator of the DosR system,
623 however activating the expression of mycobacterial virulence factors: ESAT and PE/PPE proteins;
624 D) heat map of the detected proteins of DosR regulon (14 upregulated out of 50 postulated for
625 Wayne’s model of dormancy) (82).
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
- 29 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A C
B
D
E
646 Fig. 5. Proteomic composition of enzymatic proteins in dormant M. tuberculosis. A) Pathway
647 visualization, based on KO annotation: the detected enzymes (either stored in the dormant state, or
648 those whose abundancy decreased) were placed on a metabolic map and the corresponding processes
- 30 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
649 highlighted – orange for enzymes stored in dormancy (FC > 0), dark blue for degraded enzymes
650 (FC < 0); line thickness corresponds to statistical significance – the thickest lines depict statistically
651 valuable processes; B) GO enrichment analysis for the pathways assembled by the enzymes stored in
652 dormancy; C) reconstruction of central carbon metabolism pathways of dormant M. tuberculosis cells
653 – Z-scores are mapped on the corresponding Rv coding enzymes; D) mycothiol salvage pathway;
654 E) reconstruction of ‘de novo’ UMP pathway. UMP accumulation in the dormant state was similarly
655 recently observed in the same model conditions in M. smegmatis cells (14).
656
- 31 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
657 A
658
659
660 B
Zhou_2017 Current_study
Kunnath_2010 Broger_2017 45 233
10 2 5
142 0 1 54
Rv1284 20 2 3 0
12
661
- 32 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
662 Fig. 6. Determination of potential antigenic proteins in dormant M. tuberculosis proteome.
663 A) General workflow for establishment of potential antigenic proteins. The complete list of antigenic
664 proteins is available in the supplementary file; bold-annotated proteins are those with immunogenic
665 properties selected by the sera of latently infected patients (32); B) Venn analysis of the reference
666 antigenic proteins annotated from previously published works (31–33), with the list of proteins
667 resulting from the subtractive proteomic analysis of the current data set.
668
- 33 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
669 A
670
Stored 4 9.5 7.5 13.9 15.4 31.3 18.4
Whole_genome 3.7 9.3 5 10.7 14.6 37.1 19.6
Contribution of the EC classes, %
Oxidoreductases Hydrolases Isomerases Translocases EC Transferases Lyases Ligases 671
672
673 B
674
Rv2970c Rv2224c Rv1683 Esterases 22 Query set 3 (FC>0) A 37 :Ser B 11 hydrolases C 154
Rv1263 Rv0457c Rv2579 Rv2296 Rv2068c Rv2911 Rv3569c Rv1833c Rv2970c Rv2224c Rv1683 675
676 Fig. 7. Enzymes of dormancy as potential intracellular prodrugs activators. A) Distribution of
677 the enzyme classes within the enzymes stored in dormancy compared to whole-genome enzyme
678 classes; B) overlap of the detected 48 hydrolytic enzymes with the annotated group of serine
679 hydrolases and esterases (57).
680
- 34 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A C
B
681 Fig. 8. Comparison of the current LC-MS data with the previously published work on 2D
682 proteomics of dormant M. tuberculosis (16). A) Visualization of the results of Mann–Whitney
683 comparison of data acquired in the current study with the previously published data on 2D-proteomic
684 analysis; B) GO enrichment analysis for the pathways assembled by the enzymes stored in D1;
685 C) coverage of central carbon metabolism in D1.
686
687
688
689
- 35 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
690 References
691
692 1. Fang FC. 2004. Antimicrobial reactive oxygen and nitrogen species: concepts and controversies.
693 10. Nature Reviews Microbiology 2:820–832.
694 2. Oliver JD. 2010. Recent findings on the viable but nonculturable state in pathogenic bacteria.
695 FEMS Microbiology Reviews 34:415–425.
696 3. Lenaerts A, Barry CE, Dartois V. 2015. Heterogeneity in tuberculosis pathology,
697 microenvironments and therapeutic responses. Immunol Rev 264:288–307.
698 4. Garber ED. 1960. The Host as a Growth Medium*. Annals of the New York Academy of
699 Sciences 88:1187–1194.
700 5. Brown SA, Palmer KL, Whiteley M. 2008. Revisiting the host as a growth medium. Nat Rev
701 Microbiol 6:657–666.
702 6. Eisenreich W, Dandekar T, Heesemann J, Goebel W. 2010. Carbon metabolism of intracellular
703 bacterial pathogens and possible links to virulence. Nat Rev Microbiol 8:401–412.
704 7. Schnappinger D, Ehrt S, Voskuil MI, Liu Y, Mangan JA, Monahan IM, Dolganov G, Efron B,
705 Butcher PD, Nathan C, Schoolnik GK. 2003. Transcriptional Adaptation of Mycobacterium
706 tuberculosis within Macrophages. J Exp Med 198:693–704.
707 8. Wayne LG. 1994. Dormancy of Mycobacterium tuberculosis and latency of disease. Eur J Clin
708 Microbiol Infect Dis 13:908–914.
- 36 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
709 9. Deb C, Lee C-M, Dubey VS, Daniel J, Abomoelak B, Sirakova TD, Pawar S, Rogers L,
710 Kolattukudy PE. 2009. A Novel In Vitro Multiple-Stress Dormancy Model for Mycobacterium
711 tuberculosis Generates a Lipid-Loaded, Drug-Tolerant, Dormant Pathogen. PLOS ONE 4:e6077.
712 10. Dhillon J, Lowrie DB, Mitchison DA. 2004. Mycobacterium tuberculosis from chronic murine
713 infections that grows in liquid but not on solid medium. BMC Infect Dis 4:51.
714 11. Awasthi D, Freundlich JS. 2017. Antimycobacterial Metabolism: Illuminating Mycobacterium
715 tuberculosis Biology and Drug Discovery. Trends in Microbiology 25:756–767.
716 12. Kaprelyants AS, Salina EG, Makarov VA. 2018. How to kill dormant Mycobacterium
717 tuberculosis. Int J Mycobacteriol 7:399–400.
718 13. Shleeva MO, Kudykina YK, Vostroknutova GN, Suzina NE, Mulyukin AL, Kaprelyants AS.
719 2011. Dormant ovoid cells of Mycobacterium tuberculosis are formed in response to gradual
720 external acidification. Tuberculosis (Edinb) 91:146–154.
721 14. Nikitushkin VD, Trenkamp S, Demina GR, Shleeva MO, Kaprelyants AS. 2020. Metabolic
722 profiling of dormant Mycolicibacterium smegmatis cells’ reactivation reveals a gradual
723 assembly of metabolic processes. Metabolomics 16:24.
724 15. Kaprelyants AS, Gottschal JC, Kell DB. 1993. Dormancy in non-sporulating bacteria. FEMS
725 Microbiology Reviews 10:271–285.
726 16. Trutneva KA, Shleeva MO, Demina GR, Vostroknutova GN, Kaprelyans AS. 2020. One-Year
727 Old Dormant, “Non-culturable” Mycobacterium tuberculosis Preserves Significantly Diverse
728 Protein Profile. Front Cell Infect Microbiol 10.
- 37 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
729 17. de Man JC. 1975. The probability of most probable numbers. European J Appl Microbiol 1:67–
730 78.
731 18. Rappsilber J, Mann M, Ishihama Y. 2007. Protocol for micro-purification, enrichment, pre-
732 fractionation and storage of peptides for proteomics using StageTips. Nat Protoc 2:1896–1906.
733 19. O’Rourke MB, Town SEL, Dalla PV, Bicknell F, Koh Belic N, Violi JP, Steele JR, Padula MP.
734 2019. What is Normalization? The Strategies Employed in Top-Down and Bottom-Up Proteome
735 Analysis Workflows. Proteomes 7.
736 20. Zhao Y, Wong L, Goh WWB. 2020. How to do quantile normalization correctly for gene
737 expression data analyses. 1. Scientific Reports 10:15534.
738 21. Jolliffe I.T. 2002. Principal Component Analysis. Springer-Verlag New York, Inc. 2002,
739 Department of Mathematical Sciences King’s CollegeUniversity of AberdeenAberdeenUK.
740 22. Eriksson L, Johansson E, Kettaneh-Wold N, Trygg J, Wikstrom C, Wold S. 2006. Chapter 3:
741 PCA. Multi- and Megavariate Data Analysis 43–70.
742 23. Lever J, Krzywinski M, Altman N. 2017. Principal component analysis. 7. Nature Methods
743 14:641–642.
744 24. Murtagh F, Legendre P. 2014. Ward’s Hierarchical Agglomerative Clustering Method: Which
745 Algorithms Implement Ward’s Criterion? J Classif 31:274–295.
746 25. Key M. 2012. A tutorial in displaying mass spectrometry-based proteomic data using heat maps.
747 BMC Bioinformatics 13:S10.
- 38 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
748 26. Uddin R, Siddiqui QN, Azam SS, Saima B, Wadood A. 2018. Identification and characterization
749 of potential druggable targets among hypothetical proteins of extensively drug resistant
750 Mycobacterium tuberculosis (XDR KZN 605) through subtractive genomics approach. European
751 Journal of Pharmaceutical Sciences 114:13–23.
752 27. Maurya S, Akhtar S, Siddiqui MH, Khan MKA. 2020. Subtractive Proteomics for Identification
753 of Drug Targets in Bacterial Pathogens: A Review. International Journal of Engineering
754 Research & Technology 9.
755 28. Li W, Jaroszewski L, Godzik A. 2001. Clustering of highly homologous sequences to reduce the
756 size of large protein databases. Bioinformatics 17:282–283.
757 29. Huang Y, Niu B, Gao Y, Fu L, Li W. 2010. CD-HIT Suite: a web server for clustering and
758 comparing biological sequences. Bioinformatics 26:680–682.
759 30. Yu C-S, Chen Y-C, Lu C-H, Hwang J-K. 2006. Prediction of protein subcellular localization.
760 Proteins: Structure, Function, and Bioinformatics 64:643–651.
761 31. Kunnath-Velayudhan S, Salamon H, Wang H-Y, Davidow AL, Molina DM, Huynh VT, Cirillo
762 DM, Michel G, Talbot EA, Perkins MD, Felgner PL, Liang X, Gennaro ML. 2010. Dynamic
763 antibody responses to the Mycobacterium tuberculosis proteome. PNAS 107:14703–14708.
764 32. Zhou F, Xu X, Wu S, Cui X, Pan W. 2017. ORFeome-based identification of biomarkers for
765 serodiagnosis of Mycobacterium tuberculosis latent infection. BMC Infectious Diseases 17:793.
766 33. T B, R BR, A F, Ch G, S R, J H, D D, N S-M, Cm G, M S, I R, P A, Gm H, To J, Ml G, Mj L,
767 Cm D, Md P. 2017. Diagnostic Performance of Tuberculosis-Specific IgG Antibody Profiles in
768 Patients with Presumptive Tuberculosis from Two Continents. Clin Infect Dis 64:947–955.
- 39 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
769 34. Tenenbaum D, Volkening J, Maintainer BP. 2021. KEGGREST: Client-side REST access to the
770 Kyoto Encyclopedia of Genes and Genomes (KEGG). Bioconductor version: Release (3.13).
771 35. García EA, Blanco FC, Bigi MM, Vazquez CL, Forrellad MA, Rocha RV, Golby P, Soria MA,
772 Bigi F. 2018. Characterization of the two component regulatory system PhoPR in
773 Mycobacterium bovis. Veterinary Microbiology 222:30–38.
774 36. Bansal R, Kumar VA, Sevalkar RR, Singh PR, Sarkar D. 2017. Mycobacterium tuberculosis
775 virulence-regulator PhoP interacts with alternative sigma factor SigE during acid-stress
776 response. Molecular Microbiology 104:400–411.
777 37. Augenstreich J, Arbues A, Simeone R, Haanappel E, Wegener A, Sayes F, Le Chevalier F,
778 Chalut C, Malaga W, Guilhot C, Brosch R, Astarie-Dequeker C. 2017. ESX-1 and phthiocerol
779 dimycocerosates of Mycobacterium tuberculosis act in concert to cause phagosomal rupture and
780 host cell apoptosis. Cell Microbiol 19.
781 38. Brennan MJ. 2017. The Enigmatic PE/PPE Multigene Family of Mycobacteria and Tuberculosis
782 Vaccination. Infection and Immunity 85.
783 39. Ates LS. 2020. New insights into the mycobacterial PE and PPE proteins provide a framework
784 for future research. Molecular Microbiology 113:4–21.
785 40. Betts JC, Lukey PT, Robb LC, McAdam RA, Duncan K. 2002. Evaluation of a nutrient
786 starvation model of Mycobacterium tuberculosis persistence by gene and protein expression
787 profiling. Molecular Microbiology 43:717–731.
788 41. Sali M, Sante GD, Cascioferro A, Zumbo A, Nicolò C, Donà V, Rocca S, Procoli A, Morandi M,
789 Ria F, Palù G, Fadda G, Manganelli R, Delogu G. 2010. Surface Expression of MPT64 as a
- 40 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
790 Fusion with the PE Domain of PE_PGRS33 Enhances Mycobacterium bovis BCG Protective
791 Activity against Mycobacterium tuberculosis in Mice. Infection and Immunity 78:5202–5213.
792 42. Voskuil MI, Visconti KC, Schoolnik GK. 2004. Mycobacterium tuberculosis gene expression
793 during adaptation to stationary phase and low-oxygen dormancy. Tuberculosis (Edinb) 84:218–
794 227.
795 43. Shleeva MO, Trutneva KA, Demina GR, Zinin AI, Sorokoumova GM, Laptinskaya PK,
796 Shumkova ES, Kaprelyants AS. 2017. Free Trehalose Accumulation in Dormant Mycobacterium
797 smegmatis Cells and Its Breakdown in Early Resuscitation Phase. Front Microbiol 8:524.
798 44. Jiao X, Sherman BT, Huang DW, Stephens R, Baseler MW, Lane HC, Lempicki RA. 2012.
799 DAVID-WS: a stateful web service to facilitate gene/protein list analysis. Bioinformatics
800 28:1805–1806.
801 45. Yu G, Wang L-G, Han Y, He Q-Y. 2012. clusterProfiler: an R Package for Comparing
802 Biological Themes Among Gene Clusters. OMICS: A Journal of Integrative Biology 16:284–
803 287.
804 46. Rao SPS, Alonso S, Rand L, Dick T, Pethe K. 2008. The protonmotive force is required for
805 maintaining ATP homeostasis and viability of hypoxic, nonreplicating Mycobacterium
806 tuberculosis. Proc Natl Acad Sci U S A 105:11945–11950.
807 47. Watanabe S, Zimmermann M, Goodwin MB, Sauer U, Barry 3rd CE, Boshoff HI. 2011.
808 Fumarate Reductase Activity Maintains an Energized Membrane in Anaerobic Mycobacterium
809 tuberculosis. PLOS Pathogens 7:e1002287.
- 41 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
810 48. Billig S, Schneefeld M, Huber C, Grassl GA, Eisenreich W, Bange F-C. 2017. Lactate oxidation
811 facilitates growth of Mycobacterium tuberculosis in human macrophages. 1. Scientific Reports
812 7:6484.
813 49. Haft DH. 2011. Bioinformatic evidence for a widely distributed, ribosomally produced electron
814 carrier precursor, its maturation proteins, and its nicotinoprotein redox partners. BMC Genomics
815 12:21.
816 50. Greening C, Ahmed FH, Mohamed AE, Lee BM, Pandey G, Warden AC, Scott C, Oakeshott
817 JG, Taylor MC, Jackson CJ. 2016. Physiology, Biochemistry, and Applications of F420- and Fo-
818 Dependent Redox Reactions. Microbiol Mol Biol Rev 80:451–493.
819 51. Serra-Vidal MM, Latorre I, Franken KLCM, Díaz J, de Souza-Galvão ML, Casas I, Maldonado
820 J, Milà C, Solsona J, Jimenez-Fuentes MÁ, Altet N, Lacoma A, Ruiz-Manzano J, Ausina V, Prat
821 C, Ottenhoff THM, Domínguez J. 2014. Immunogenicity of 60 novel latency-related antigens of
822 Mycobacterium tuberculosis. Front Microbiol 5:517.
823 52. Singh R, Manjunatha U, Boshoff HIM, Ha YH, Niyomrattanakit P, Ledwidge R, Dowd CS, Lee
824 IY, Kim P, Zhang L, Kang S, Keller TH, Jiricek J, Barry CE. 2008. PA-824 Kills Nonreplicating
825 Mycobacterium tuberculosis by Intracellular NO Release. Science 322:1392–1395.
826 53. Makarov V, Manina G, Mikusova K, Möllmann U, Ryabova O, Saint-Joanis B, Dhar N, Pasca
827 MR, Buroni S, Lucarelli AP, Milano A, Rossi ED, Belanova M, Bobovska A, Dianiskova P,
828 Kordulakova J, Sala C, Fullam E, Schneider P, McKinney JD, Brodin P, Christophe T, Waddell
829 S, Butcher P, Albrethsen J, Rosenkrands I, Brosch R, Nandi V, Bharath S, Gaonkar S, Shandil
830 RK, Balasubramanian V, Balganesh T, Tyagi S, Grosset J, Riccardi G, Cole ST. 2009.
- 42 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
831 Benzothiazinones Kill Mycobacterium tuberculosis by Blocking Arabinan Synthesis. Science
832 324:801–804.
833 54. Mirnejad R, Asadi A, Khoshnood S, Mirzaei H, Heidary M, Fattorini L, Ghodousi A, Darban-
834 Sarokhalil D. 2018. Clofazimine: A useful antibiotic for drug-resistant tuberculosis. Biomed
835 Pharmacother 105:1353–1359.
836 55. Côtes K, N’Goma JCB, Dhouib R, Douchet I, Maurin D, Carrière F, Canaan S. 2008. Lipolytic
837 enzymes in Mycobacterium tuberculosis. Appl Microbiol Biotechnol 78:741–749.
838 56. Ortega C, Anderson LN, Frando A, Sadler NC, Brown RW, Smith RD, Wright AT, Grundner C.
839 2016. Systematic Survey of Serine Hydrolase Activity in Mycobacterium tuberculosis Defines
840 Changes Associated with Persistence. Cell Chemical Biology 23:290–298.
841 57. Tallman KR, Levine SR, Beatty KE. 2016. Small-Molecule Probes Reveal Esterases with
842 Persistent Activity in Dormant and Reactivating Mycobacterium tuberculosis. ACS Infect Dis
843 2:936–944.
844 58. Alteri CJ, Lindner JR, Reiss DJ, Smith SN, Mobley HLT. 2011. The broadly conserved regulator
845 PhoP links pathogen virulence and membrane potential in Escherichia coli. Mol Microbiol
846 82:145–163.
847 59. Golubeva YA, Slauch JM. 2006. Salmonella enterica serovar Typhimurium periplasmic
848 superoxide dismutase SodCI is a member of the PhoPQ regulon and is induced in macrophages.
849 J Bacteriol 188:7853–7861.
- 43 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
850 60. Muñoz‐Elías EJ, Upton AM, Cherian J, McKinney JD. 2006. Role of the methylcitrate cycle in
851 Mycobacterium tuberculosis metabolism, intracellular growth, and virulence. Molecular
852 Microbiology 60:1109–1122.
853 61. Gould TA, Langemheen HVD, Muñoz‐Elías EJ, McKinney JD, Sacchettini JC. 2006. Dual role
854 of isocitrate lyase 1 in the glyoxylate and methylcitrate cycles in Mycobacterium tuberculosis.
855 Molecular Microbiology 61:940–947.
856 62. Peyron P, Vaubourgeix J, Poquet Y, Levillain F, Botanch C, Bardou F, Daffé M, Emile J-F,
857 Marchou B, Cardona P-J, Chastellier C de, Altare F. 2008. Foamy Macrophages from
858 Tuberculous Patients’ Granulomas Constitute a Nutrient-Rich Reservoir for M. tuberculosis
859 Persistence. PLOS Pathogens 4:e1000204.
860 63. Zimmermann M, Kogadeeva M, Gengenbacher M, McEwen G, Mollenkopf H-J, Zamboni N,
861 Kaufmann SHE, Sauer U. 2017. Integration of Metabolomics and Transcriptomics Reveals a
862 Complex Diet of Mycobacterium tuberculosis during Early Macrophage Infection. mSystems 2.
863 64. Zimmermann M, Kuehne A, Boshoff HI, Barry CE, Zamboni N, Sauer U. 2015. Dynamic
864 exometabolome analysis reveals active metabolic pathways in non-replicating mycobacteria.
865 Environmental Microbiology 17:4802–4815.
866 65. Muñoz-Elías EJ, McKinney JD. 2005. Mycobacterium tuberculosis isocitrate lyases 1 and 2 are
867 jointly required for in vivo growth and virulence. 6. Nature Medicine 11:638–644.
868 66. Cook GM, Hards K, Vilchèze C, Hartman T, Berney M. 2014. Energetics of Respiration and
869 Oxidative Phosphorylation in Mycobacteria. Microbiol Spectr 2.
- 44 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
870 67. Schubert OT, Ludwig C, Kogadeeva M, Zimmermann M, Rosenberger G, Gengenbacher M,
871 Gillet LC, Collins BC, Röst HL, Kaufmann SHE, Sauer U, Aebersold R. 2015. Absolute
872 Proteome Composition and Dynamics during Dormancy and Resuscitation of Mycobacterium
873 tuberculosis. Cell Host Microbe 18:96–108.
874 68. Warner DF, Evans JC, Mizrahi V. 2014. Nucleotide Metabolism and DNA Replication.
875 Microbiol Spectr 2.
876 69. Ferraris DM, Gelardi ELM, Garavaglia S, Miggiano R, Rizzi M. 2020. Targeting NAD-
877 dependent dehydrogenases in drug discovery against infectious diseases and cancer. Biochem
878 Soc Trans 48:693–707.
879 70. Pethe K, Sequeira PC, Agarwalla S, Rhee K, Kuhen K, Phong WY, Patel V, Beer D, Walker JR,
880 Duraiswamy J, Jiricek J, Keller TH, Chatterjee A, Tan MP, Ujjini M, Rao SPS, Camacho L,
881 Bifani P, Mak PA, Ma I, Barnes SW, Chen Z, Plouffe D, Thayalan P, Ng SH, Au M, Lee BH,
882 Tan BH, Ravindran S, Nanjundappa M, Lin X, Goh A, Lakshminarayana SB, Shoen C,
883 Cynamon M, Kreiswirth B, Dartois V, Peters EC, Glynne R, Brenner S, Dick T. 2010. A
884 chemical genetic screen in Mycobacterium tuberculosis identifies carbon-source-dependent
885 growth inhibitors devoid of in vivo efficacy. Nat Commun 1:1–8.
886 71. Buchmeier NA, Newton GL, Koledin T, Fahey RC. 2003. Association of mycothiol with
887 protection of Mycobacterium tuberculosis from toxic oxidants and antibiotics. Molecular
888 Microbiology 47:1723–1732.
889 72. Hernick M. 2013. Mycothiol: a target for potentiation of rifampin and other antibiotics against
890 Mycobacterium tuberculosis. Expert Review of Anti-infective Therapy 11:49–67.
- 45 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
891 73. Gurumurthy M, Rao M, Mukherjee T, Rao SPS, Boshoff HI, Dick T, Barry CE, Manjunatha
892 UH. 2013. A novel F420-dependent anti-oxidant mechanism protects Mycobacterium
893 tuberculosis against oxidative stress and bactericidal agents. Molecular Microbiology 87:744–
894 755.
895 74. Trutneva KA, Avdienko VG, Demina GR, Shleeva MO, Shumkov MS, Salina EG, Kaprelyants
896 AS. 2021. Immunoreactive Proteins of Dormant Mycobacterium tuberculosis Cells. Appl
897 Biochem Microbiol 57:170–179.
898 75. Liu Y, Matsumoto M, Ishida H, Ohguro K, Yoshitake M, Gupta R, Geiter L, Hafkin J. 2018.
899 Delamanid: From discovery to its use for pulmonary multidrug-resistant tuberculosis (MDR-
900 TB). Tuberculosis (Edinb) 111:20–30.
901 76. Chen X, Hashizume H, Tomishige T, Nakamura I, Matsuba M, Fujiwara M, Kitamoto R, Hanaki
902 E, Ohba Y, Matsumoto M. 2017. Delamanid Kills Dormant Mycobacteria In Vitro and in a
903 Guinea Pig Model of Tuberculosis. Antimicrob Agents Chemother 61:e02402-16.
904 77. Gupta KB, Gupta R, Atreja A, Verma M, Vishvkarma S. 2009. Tuberculosis and nutrition. Lung
905 India 26:9–16.
906 78. Esterbauer H, Cheeseman KH. 1990. Determination of aldehydic lipid peroxidation products:
907 malonaldehyde and 4-hydroxynonenal. Methods Enzymol 186:407–421.
908 79. Puckett S, Trujillo C, Wang Z, Eoh H, Ioerger TR, Krieger I, Sacchettini J, Schnappinger D,
909 Rhee KY, Ehrt S. 2017. Glyoxylate detoxification is an essential function of malate synthase
910 required for carbon assimilation in Mycobacterium tuberculosis. Proc Natl Acad Sci USA
911 114:E2225–E2232.
- 46 - bioRxiv preprint doi: https://doi.org/10.1101/2021.08.06.455493; this version posted August 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
912 80. Krieger IV, Freundlich JS, Gawandi VB, Roberts JP, Gawandi VB, Sun Q, Owen JL, Fraile MT,
913 Huss SI, Lavandera J-L, Ioerger TR, Sacchettini JC. 2012. Structure-guided Discovery of Phenyl
914 diketo-acids as Potent Inhibitors of M. tuberculosis Malate Synthase. Chem Biol 19:1556–1567.
915 81. Shleeva MO, Kudykina YK, Vostroknutova GN, Suzina NE, Mulyukin AL, Kaprelyants AS.
916 2011. Dormant ovoid cells of Mycobacterium tuberculosis are formed in response to gradual
917 external acidification. Tuberculosis 91:146–154.
918 82. Schubert OT, Mouritsen J, Ludwig C, Röst HL, Rosenberger G, Arthur PK, Claassen M,
919 Campbell DS, Sun Z, Farrah T, Gengenbacher M, Maiolica A, Kaufmann SHE, Moritz RL,
920 Aebersold R. 2013. The Mtb Proteome Library: A Resource of Assays to Quantify the Complete
921 Proteome of Mycobacterium tuberculosis. Cell Host Microbe 13:602–612.
922
- 47 -