Content uploaded by Chris Greening
Author content
All content in this area was uploaded by Chris Greening on Sep 13, 2023
Content may be subject to copyright.
An obligate aerobe adapts to hypoxia by hybridising
1
fermentation with carbon storage
2
David L Gillett1 *, Tess Hutchinson1,2, Manasi Mudaliyar4, Thomas D. Watts1, Wei Wen Wong2,
3
Jake Locop1, Luis Jimenez1, Iresha Hanchapola3, Han-Chung Lee3, Erwin Tanuwidjaya3, Joel
4
R. Steele3. Ralf B. Schittenhelm4, Christopher K. Barlow4, Rhys Grinter1, Debnath Ghosal4,5,
5
Perran L. M. Cook2, Chris Greening1 *
6
1. Department of Microbiology, Monash University, Clayton, VIC, Australia
7
2. School of Chemistry, Monash University, Clayton, VIC, Australia
8
3. Monash Proteomics and Metabolomics Facility, Monash University, Clayton, VIC, Australia
9
4. Department of Biochemistry and Pharmacology, Bio21 Institute, The University of
10
Melbourne, Parkville, VIC, Australia
11
5. ARC Centre for Cryo-electron Microscopy of Membrane Proteins, Bio21 Molecular Science
12
and Biotechnology Institute, University of Melbourne, Parkville, VIC, Australia
13
* Correspondence can be addressed to chris.greening@monash.edu and
14
david.gillett@monash.edu
15
16
17
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Abstract
18
In soil ecosystems, obligately aerobic bacteria survive oxygen deprivation (hypoxia) by
19
entering non-replicative persistent states. Little is known about how these bacteria rewire their
20
metabolism to stay viable in these states. The model obligate aerobe Mycobacterium
21
smegmatis maintains redox homeostasis during hypoxia by mediating fermentative hydrogen
22
production. However, the fate of organic carbon during fermentation, and the associated
23
remodeling of carbon metabolism, is unresolved. Here we systematically profiled the
24
metabolism of M. smegmatis during aerobic growth, hypoxic persistence, and the transition
25
between these states. Using differential isotope labelling, and paired metabolomics and
26
proteomics, we observed rerouting of central carbon metabolism through the pentose
27
phosphate pathway and Entner-Doudoroff pathway during hypoxia. We show that M.
28
smegmatis excretes high levels of hydrogen concurrently with upregulating triacylglyceride
29
synthases and accumulating glycerides as carbon stores. Using electron cryotomography
30
(cryo-ET), we observed the presence of large spheroid structures consistent with the
31
appearance of lipid droplets. Thus, in contrast to obligately and facultative anaerobic
32
fermentative bacteria, M. smegmatis stores rather than excretes organic carbon during
33
hypoxia. This novel hybrid metabolism likely provides a competitive advantage in resource-
34
variable environments by allowing M. smegmatis to simultaneously dispose excess reductant
35
during hypoxia and maintain carbon stores to rapidly resume growth upon reoxygenation.
36
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Introduction
37
Most soil bacteria exist in non-replicating persistent states, also known as dormancy. In these
38
states, bacteria do not expend energy on growth and replicate, but still require some minimal
39
metabolic activity to sustain cellular integrity and maintenance. Limitation or variability for key
40
resources, for example organic carbon, nitrogen, or oxygen, is the primary reason microbes
41
transition from growing to dormant states13. In many soil environments, oxygen
42
concentrations sharply vary across space and time, and influence the biological structure at
43
both microscopic and macroscopic scales. Many soil bacteria are facultative anaerobes that
44
adapt to oxygen limitation (hypoxia) by growing through anaerobic respiration or fermentation
45
instead. However, the dominant bacteria in most surface soils are seemingly obligate aerobes
46
that can only survive rather than grow during hypoxia4. For example, Mycobacterium is a
47
cosmopolitan genus of obligate aerobes that comprise approximately 1% of soil bacteria and
48
are notorious for their capacity to survive extended periods of hypoxia46. To date, relatively
49
little is known about how mycobacteria and other aerobic soil bacteria adapt their metabolism
50
to maintain viability under hypoxia.
51
Recent studies suggest soil mycobacteria depend on metabolic flexibility to adapt to variations
52
in resource limitation79. shown that M. smegmatis switches between growth through
53
aerobic respiration and persistence by fermentation in response to variations in oxygen
54
availability. When oxygen is limited, this bacterium disposes of excess electrons by reducing
55
protons to H2 using the fermentative group 3b [NiFe]-hydrogenase Hyh7. M. smegmatis is
56
proposed to recycle H2 produced during hypoxia using the respiratory group 2a [NiFe]-
57
hydrogenase Huc when electron acceptors become available. Production of both
58
hydrogenases is activated under hypoxia by the oxygen- and redox-sensing DosS/T-DosR
59
two-component system7. Deletion of both respiratory and fermentative hydrogenases disrupts
60
redox homeostasis and decreases survival7,10. While this study provided the first report of
61
fermentation in an obligate aerobe, it did not address the fate of the carbon end-products.
62
Most previous studies have investigated fermentation in the context of growth rather than
63
persistence1113; for example, in the well-known examples of ethanol and lactic acid
64
fermentation, the fermentative end-product is secreted outside of the cell, preventing
65
intracellular accumulation of organic waste molecules12,13. However, it is unclear whether M.
66
smegmatis also excretes end-products or instead stores them. Organic carbon is still
67
potentially available for anabolic metabolism after it has been partially oxidised to drive ATP
68
production. Moreover, anabolic storage pathways also serve as a sink of reductant that can
69
be used to regenerate electron carriers, but it is difficult for cells to maintain redox balance
70
through this process alone. In this manner, fermentative hydrogenases may serve as a redox
71
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
valve by allowing the cell to release excess reductant and regenerate redox cofactors as
72
required.
73
The synthesis and accumulation of triacylglycerols (TAGs) is an important part of
74
mycobacterial persistence1419. TAGs are composed of three fatty acid chains joined by ester
75
linkages to a glycerol backbone; a lack of any hydrophilic moiety makes them exceedingly
76
hydrophobic so that they readily form droplets, and the highly reduced fatty acid chains can
77
be readily oxidised to extract energy. These properties make TAGs an ideal cellular store of
78
energy and carbon, as lipid droplets are self-compartmentalising and extremely energy dense.
79
Lipid droplets, also referred to as intracellular inclusions, were first reported in mycobacteria
80
more than 70 years ago and have since been consistently observed in mycobacteria via light
81
and electron microscopy15,2024. These droplets are strongly associated with stress responses
82
and persistence in mycobacteria15,18,25 and TAGs have been identified as their dominant
83
constituent18,26,27. Mycobacterial genomes encode multiple TAG synthases (Tgs), which
84
catalyse the final step of TAG synthesis (the acylation of diacylglycerols) and these are among
85
the most upregulated genes in mycobacteria during hypoxia, with Tgs1 the most active of the
86
synthases part of the regulon of the hypoxic response regulator DosR in both M. tuberculosis
87
and M. smegmatis 7,2830. As well as a storage molecule, TAGs are also abundant within and
88
on the surface of the mycobacterial cell envelope31,32, as are TAG synthases and TAG export
89
proteins17,32,33 and their hydrophobicity is thought to contribute to the highly impermeable
90
nature of the mycomembrane. Our understanding of TAG metabolism in mycobacteria
91
predominately arises from the study of M. tuberculosis, which builds substantial stores of
92
TAGs during infection, via both biosynthesis and the import of TAGs and fatty acids from host
93
lipid-loaded macrophages1517,19. In contrast, the relationship between TAG metabolism and
94
the persistence of environmental mycobacteria is poorly studied.
95
Building on these previous studies, we hypothesised that these processes are coupled: M.
96
smegmatis may adapt to hypoxia by hybridising hydrogenogenic fermentation with TAG
97
synthesis. We systematically tested this by measuring carbon degradation pathways and
98
hydrogen production, conducting paired metabolomics and proteomics, and visualising lipid
99
droplets in cells before, during, and following the transition to hypoxia. Together, we provide
100
strong evidence that M. smegmatis partially oxidises organic carbon during hypoxia by
101
disposing excess reductant as hydrogen and storing organic carbon primarily as TAGs. We
102
propose that this novel hybrid mode of metabolism is ideally suited for bacteria adapting to
103
frequent oxic-hypoxic transitions.
104
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Results
105
M. smegmatis partially oxidises carbon during fermentative H2 production.
106
M. smegmatis produces H2 during oxygen depletion using the fermentative hydrogenase Hyh7,
107
though the timing of this process and the extent of H2 accumulation have not been established.
108
To understand the dynamics of hydrogen production during the transition to hypoxia, we
109
cultured M. smegmatis de Bont (HdeB) minimal media with 0.4% w/v glucose in
110
sealed serum vials with ambient air and measured the concentration of H2 and O2 in the
111
headspace as O2 was gradually depleted (Fig. 1A). ODmax was reached when headspace O2
112
concentration reached ~2%, and O2 was gradually consumed to below the detection limit
113
(~0.10%) over the next 24 hours, indicating transition to hypoxia-induced dormancy. This is
114
broadly consistent with existing experimental models of hypoxia-induced mycobacterial
115
dormancy34,35. The onset of net hydrogen production coincided with the consumption of O2
116
below detection limits. H2 concentration steadily rose over the next 48 hours before plateauing
117
between 500-1000 ppmv, indicating that organic carbon catabolism is sustained through the
118
transition to hypoxia and is supported by fermentative H2 production once respiratory electron
119
acceptors are exhausted (Fig. 1A). The concentration of accumulated H2 is a magnitude
120
higher than previously reported using similar methodology7, and suggests that fermentative
121
H2 production is a major process during hypoxia in M. smegmatis.
122
In contrast to respiration where organic carbon is completely oxidised to CO2, during
123
fermentation organic carbon is incompletely oxidised (Fig. 1B). To determine if the degree of
124
oxidation of organic carbon is consistent with fermentation during the transition to hypoxia-
125
induced dormancy, we incubated M. smegmatis cultures with 1-13C-glucose, 2-13C-glucose, 3-
126
13C-glucose and 13C6-glucose for 5 hours during exponential growth, stationary phase, and
127
during the onset of hypoxia (OD600= ~2.5), and then measured the ratio of 13CO2:12CO2 using
128
isotope ratio mass spectrometry (IRMS)36. If all six carbons of glucose are oxidised, as during
129
respiration, each carbon will contribute 1/6th of the total 13CO2 signal produced by 13C6-glucose,
130
while incomplete oxidation will result in individual carbon positions contributing
131
disproportionately to the total 13CO2 signal (Fig. 1B). During exponential growth, the proportion
132
of signal attributable to 1-13C, 2-13C and 3-13C was 0.21, 0.14 and 0.20 of the total 13C6 signal
133
respectively, indicating that aerobic respiration was the dominant catabolic process, as
134
expected, while during the hypoxic transition there was a disproportionate increase in the
135
13CO2 signal attributable to 1-13C glucose (0.49 of 13C-6) (Fig. 1C). This is consistent with
136
glucose being incompletely oxidised as it passes through either the pentose phosphate
137
pathway (PPP) or the Entner-Doudoroff pathway (EDP), two alternative pathways for the
138
oxidation of glucose to pyruvate (Fig. 1B). Substantial 13CO2 signal was still attributable to 2-
139
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
140
Figure 1: M. smegmatis fermentatively produces hydrogen and partially oxidises glucose via the
141
pentose phosphate pathway during oxygen depletion. (A) Density (OD600), O2 and H2 of M. smegmatis
142
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
cultures (n=4) during gradual oxygen depletion were monitored. Density and O2 share the same axis
143
and range. Growth arrest (ODmax) is reached at [O2] = ~2% while H2 accumulation coincides with onset
144
of anoxia at ~70 hours post inoculation. Error bars represent standard deviation from the mean. (B)
145
Catabolism of glucose through either glycolysis, the Pentose Phosphate Pathway (PPP) or the Entner-
146
Doudoroff Pathway (EDP) results in carbons at different positions being oxidised to CO2. Treatment
147
with differentially isotope labelled glucose and comparing the ratios of 13CO2/12CO2 by using isotope
148
ratio mass spectrometry (IRMS) indicates contributions of different catabolic pathways to CO2
149
production. (C) M. smegmatis (n=4) was treated with differentially 13C-labelled glucose at mid-
150
exponential phase (OD600 ~0.8) and the hypoxic transition (late-exponential phase; OD600 ~2.0).
151
152
13C and 3-13C, indicating that respiration was also occurring, which is unsurprising considering
153
O2 is still present before ODmax.
154
We also treated hypoxic mid-stationary phase M. smegmatis cultures (3 days post ODmax) with
155
differentially labelled 13-C glucose but detected no 13CO2 signal above background, even after
156
an extended incubation of 5 days. This indicates that M. smegmatis stopped consuming
157
exogenous glucose in the media after prolonged anoxia, possibly because of reduced
158
metabolic activity or a switch to the catabolism of internal carbon stores.
159
160
Early proteome remodelling drives a sustained shift in metabolism during hypoxia
161
To gain further insights into the fate of partially oxidised carbon during fermentation, and
162
associated changes to broader metabolism, we conducted paired comparative metabolomics
163
and proteomics on lysates from M. smegmatis cultures harvested during aerated exponential
164
growth, the transition to hypoxia (0.5% O2), and sustained hypoxia (mid-stationary phase, 3
165
days post ODmax). We observed large changes across all comparisons: 397 metabolites and
166
310 proteins were differentially abundant between hypoxic transition vs exponential phase,
167
610 metabolites and 307 proteins between stationary phase vs exponential phase and 429
168
metabolites and just two proteins between stationary phase vs hypoxic transition (Fig. 2A and
169
2C). The changes in the metabolome were consistently larger than those for the proteome,
170
and with just two proteins reaching the cut-off criteria (>2 fold change, P < 0.01), virtually no
171
changes to the proteome were observed between the hypoxic transition and stationary phase.
172
This indicated that proteome remodelling is completed early in the response of M. smegmatis
173
to oxygen depletion, but drives a prolonged and sustained shift in metabolism. This is
174
somewhat unexpected, as large transcriptional changes are sustained throughout the
175
response of M. tuberculosis to oxygen depletion28,37.
176
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Consistent with the broader literature, the most highly upregulated proteins during oxygen
177
deprivation were members of the DosR regulon (Fig. 2A), with the most upregulated proteins
178
being HspX and the FAD-sequestering protein Fsq38β and yh were all
179
upregulated between 2.8-fold and 38-fold during the hypoxic transition (hypoxic transition vs
180
exponential phase), consistent with our finding that net hydrogen production is observed
181
shortly after oxygen is depleted below the detection threshold. Two DosR-regulated
182
triacylglycerol synthases that drive TAG synthesis in M. tuberculosis during hypoxia29 were
183
upregulated 16- to 26-fold β subunits of the uptake hydrogenase
184
Huc were mildly upregulated (~2.5 fold), which is consistent with previous studies that
185
hydrogen produced by Hyh can be recycled by Huc if a terminal electron acceptor becomes
186
available7.
187
For each comparison, we broadly categorised metabolites and proteins by predicted function
188
according to KEGG and IDEOM pathway annotations (Fig. 2A and 2C). For metabolites, we
189
observed large changes in all comparisons for every category examined. For energy
190
metabolism, amino acid metabolism, cofactor lipid metabolism, and carbohydrate metabolism,
191
multiple metabolites increased and decreased in abundance by at least 8-fold. We detected
192
an overall increase in lipid metabolism in the stationary vs exponential phase comparison
193
(mean of 2.1-fold) and carbohydrate and cofactor metabolism in the hypoxic transition vs
194
exponential phase comparison (means of 2.1- and 2.0-fold respectively) and an overall
195
decrease in nucleotide metabolism, amino acid metabolism and peptides in the stationary
196
phase vs exponential phase comparison (means of -1.5, -2.1 and -1.6 fold). Changes in the
197
proteome to metabolic pathways during oxygen depletion were generally much milder than for
198
the metabolomes, and except for the strong upregulation of the DosR regulon, we discerned
199
no pattern in the comparisons (Fig. 2 and Fig. 3).
200
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Figure 2: Comparative proteomic (top panels) and metabolomic (bottom panels) analysis of M.
203
smegmatis cultures (n=4) during mid-exponential phase (EXP), hypoxic transition (TR) and hypoxic
204
stationary phase (ST). Comparisons are separated into TR vs EXP (black), ST vs EXP (dark pink) and
205
ST vs TR (dark green). Proteins and metabolites with statistically significant abundance across
206
comparisons (proteomics, p<0.05 (A-B); metabolomics, p < 0.01 (C-D)) are represented with large dark
207
coloured dots. Proteins and metabolites that were detected but were not statistically different are
208
represented with smaller dots in lighter shades. The fold change (FC) threshold is represented with
209
thick dotted lines (proteomics, FC > 1 (log2) (A-B); metabolomics, FC > 2 (log2). All error bars represent
210
standard deviation from the mean. Differentially abundant proteins are separated into predicted
211
metabolic functional groups based on KEGG pathways annotations (A) and KEGG module annotations
212
(B). Differentially abundant metabolites are separated into predicted metabolic function based on
213
IDEOM annotations (C) and KEGG modules annotations (D). Predicted annotations are superseded
214
with experiment-based annotations where available.
215
216
M. smegmatis drives TAG synthesis during hypoxia
217
To investigate changes to metabolic pathways relevant to fermentation and TAG synthesis,
218
we examined changes to central carbon metabolism and fatty acid metabolism. Large
219
accumulations of monoacylglycerols (MAGs) and diacylglycerols (DAGs) occurred during the
220
hypoxic transition and hypoxic stationary phase (Fig. 2B). This is consistent with the strong
221
upregulation of two TAG synthases, though it may also reflect the biosynthesis or degradation
222
of other glycerolipids such membrane phospholipids. We observed a large decrease in
223
phospholipids, indicating that accumulation of MAGs and DAGs is not due to an upregulation
224
of phospholipid biosynthesis, and suggesting that phospholipid degradation could be a major
225
source of the MAGs and DAGs. While we detected no TAGs, this reflects their highly
226
hydrophobic nature is incompatible with our extraction and LC-MS protocol.
227
Intermediates of fatty acid metabolism were overall more abundant during oxygen depletion,
228
with several free fatty acid species increasing in abundance by >20 fold (26:0, 19:1, 17:0, 14:0
229
and 16:1), indicating that either glycerolipids were being degraded to provide free fatty acids,
230
or new fatty acids were being synthesised de novo. Malonyl-CoA, the precursor of malonyl-
231
ACP which is used by fatty acid synthase I (FAS I) to initiate fatty acid chain synthesis (C16-
232
C24) and by fatty acid synthase II (FAS II) to produce long chain fatty acids (up to C56) as
233
precursors for mycolic acids, major components of the mycobacterial cell wall39, increased in
234
abundance by 36-fold in hypoxic transition and 10-fold in hypoxic stationary phase compared
235
to exponential phase. We observed a mild decrease in proteins involved in mycolic acid
236
synthesis during oxygen depletion (<2-fold), but did not identify any mycolic acid synthesis
237
intermediates, although this is likely be due to their highly hydrophobic nature preventing mass
238
spectrometric detection. Overall, the accumulation of MAGs and DAGs is likely driven by a
239
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
combination of both de novo fatty acid synthesis and the degradation of phospholipids.
240
Although we cannot exclude the possibility of increased mycolic acid synthesis, the
241
metabolomic and proteomic changes are overall consistent with a dramatic upregulation of
242
TAG biosynthesis during hypoxia concurrent with fermentative H2 production.
243
244
Rewiring of central carbon metabolism and reduction of cofactor levels during hypoxia
245
Changes to central carbon metabolism during hypoxia were characterised by an overall
246
decrease in abundance of intermediates of glycolysis and the TCA cycle, and an increase in
247
intermediates of the PPP and EDP, consistent with our differential isotope experiment.
248
Interestingly, the abundance of glycolysis and TCA cycle intermediates only decreased in the
249
hypoxic stationary phase vs exponential phase comparison, whereas the abundance of PPP
250
and EDP intermediates increases in the hypoxic transition vs exponential phase (Fig. 3). The
251
most dramatic changes were observed in gluconate and 6-phosphogluconate, precursors of
252
both the PPP and EDP, both of which were markedly more abundant during hypoxic transition
253
(7.4- and 7.7-fold, respectively) and hypoxic stationary phase phases (74-fold and 37-fold,
254
respectively) (Fig. 3). Compared to exponential phase, PPP-intermediates sedoheptulose 7-
255
phosphate and erthryose 4-phosphate were more abundant during the hypoxic transition (12-
256
fold and 5.6-fold, respectively) and hypoxic stationary phase (5-fold and 2.2-fold, respectively),
257
while the EDP intermediate 2-dehydro-3-deoxy-gluconate was less abundant during the
258
hypoxic transition (3.5-fold) and hypoxic stationary phase (17-fold). This suggests increased
259
flux through the PPP, rather than the EDP. We were unable to identify the key EDP
260
intermediate 2-dehydro-3-deoxy-gluconate-6-phosphate (Fig. 3).
261
We saw a marked accumulation of gluconate, gluconolactone and gluconate-6-phosphate
262
during oxygen depletion, early intermediates in both the EDP and PPP. Interestingly, the
263
accumulation of gluconate and the transient accumulation of gluconolactone suggests an
264
active gluconate shunt during oxygen depletion, which circumvents the first committed step of
265
the PPP and ED: the oxidation of glucose-6-phosphate to glucono-1,5-lactone-6-phosphate
266
by the glucose-6-phosphate dehydrogenase (G6PDH)40 (Fig. 3). The gluconate shunt is poorly
267
studied in bacteria, and while M. smegmatis has genes predicted to encode the enzymes of
268
the gluconate shunt (MSMEG_0655: glucose 1-dehydrogenase (GDH); MSMEG_1274 and
269
MSMEG_2561: gluconolactonase (GNL); MSMEG_0453: gluconokinase), they have not been
270
biochemically characterised. M. smegmatis encodes both a canonical NADP-dependent and
271
a F420- dependent G6PDH, both of which were mildly upregulated (1.4- to 1.6-fold) during
272
hypoxia, suggesting that the conventional oxidative PPP was active during hypoxia in addition
273
to the gluconate shunt. Regardless, both routes produce reduced cofactors (i.e., NADPH,or
274
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
F420H2) that supply reducing equivalents for anabolism, which is consistent with metabolic
275
remodelling to support large-scale TAG synthesis. Furthermore, the accumulation of
276
gluconate and gluconate-6-phosphate may reflect that metabolic flux is primarily rerouted
277
through the PPP / EDP to produce reducing equivalents to support TAG synthesis, rather than
278
intermediates for nucleotide and amino acid synthesis, for which we observed either no
279
change or a decrease in abundance during oxygen depletion.
280
Based on the proteome, there were minimal changes to central carbon metabolism during
281
oxygen depletion, with most enzymes exhibiting either no change or less than 2-fold change
282
in abundance (Fig. 2B and Fig. 3). The exceptions to this were two products of the DosR
283
regulon, MSMEG_3934 (upregulated 70-fold and 69-fold, hypoxic transition and hypoxic
284
stationary phase, vs exponential phase, respectively), which encodes a putative
285
phosphoenolpyruvate synthase (PEPS), and MSMEG_3947 (upregulated 9-fold for both
286
hypoxic transition and hypoxic stationary phase, vs exponential phase), which encodes the
287
reversible phosphofructokinase PfkB (Fig. 3). We also observed a 3.2- and 2.7-fold down
288
regulation of isocitrate lyase (icl), a key enzyme of the glyoxylate shunt and methylcitrate
289
pathways (Fig. 3), in contrast to studies in M. tuberculosis. The potential roles of these
290
enzymes are further discussed in Supplementary Note 1. We also observed a 4- to 8- fold
291
upregulation of MSMEG_4710-4712, homologs of the subunits of the branched-chain keto
292
acid dehydrogenase (BCKADH) complex for valine, leucine, and isoleucine degradation,
293
consistent with the 5-fold increase in abundance of products ketoleucine and ketovaline during
294
the hypoxic transition vs exponential phase comparison. Beyond these changes, the
295
proteomes did not reflect the more dramatic changes in metabolite abundance. This potentially
296
reflects that regulation of central carbon metabolism is largely post-transcriptional, for example
297
through allosteric regulation, for example with studies in M. tuberculosis showing that PPP
298
intermediates inhibit pyruvate kinase (PK) to further increase flux towards the PPP 41.
299
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
300
Figure 3: Summary of metabolic changes during oxygen depletion in M. smegmatis, as determined by comparative proteomics and metabolomics (Fig. 2). As
301
described in legend (bottom right), heat map displayed as circles (metabolites) and squares (proteins) represents fold change in abundance for features
302
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
meetings their respective cut-off threshold (proteomics, P < 0.05, FC > 2 ; metabolomics, p < 0.01, FC > 4). Features not identified in experiment are marked
303
with X and white background. For clarity, oxidised cofactors (e.g., NAD(P)+) and ADP are not included. Abbreviations: ACN: aconitase; ACP: acyl-carrier protein;
304
ADL: fructose-bisphosphate aldolase; ADP: adenosine diphosphate; ATP: adenosine triphosphate; CIT: citrate synthase; CoA: Coenzyme A; EDA: 2-dehydro-
305
3-deoxyphosphogluconate aldolase; EDD: phosphogluconate dehydratase; EDP: Entner-Doudoroff pathway; ENL: enolase; FAD: flavin adenine dinucleotide;
306
FAS: fatty acid synthase; Fdred/ox: ferredoxin (reduced / oxidised) FGD: F420-dependent G6PDH; FUM: fumarase; GAPDH: glyceraldehyde phosphate
307
dehydrogenase; G6PDH: glucose-6-phosphate dehydrogenase; GDH: glucose-1 dehydrogenase; GNK: gluconokinase; GNL: gluconolactonase; GPDH:
308
glycerol-3-phosphate dehydrogenase; HHY: respiratory group 1h [NiFe]-hydrogenase HK: hexokinase; HUC: respiratory group 2a [NiFe]-hydrogenase; HYH:
309
fermentative group 3b [NiFe]-hydrogenase; ICD/IDH: isocitrate dehydrogenase; ICL: isocitrate lyase; KDGK: 2-dehydro-3-deoxygluconokinase; KDH:
310
ketoglutarate dehydrogenase; PGK: phosphoglycerate kinase; PGM: phosphoglycerate mutase; PK: pyruvate kinase; MCS: methylcitrate synthase; MCD:
311
methylcitrate dehydratase; MCM: methylmalonyl-CoA mutase; MCP: methylcitrate pathway; MDH: malate dehydrogenase; MMCE: methylmalonyl-CoA
312
epimerase; MMP: methylmalonyl pathway; MS: malate synthase; NAD(H): nicotinamide adenine dinucleotide; NADP(H): nicotinamide adenine dinucleotide
313
phosphate: PCC: propionyl-CoA carboxylases; PDH: pyruvate dehydrogenase; PEPCK: phosphoenolpyruvate carboxykinase; PEPS: phosphoenolpyruvate
314
synthase PFK: phosphofructokinase; PGI: phosphoglucoisomerase; PPP: Pentose Phosphate Pathway; PYC: pyruvate carboxylase; RPI: ribulose-5-phosphate
315
isomerase; SCS: succinyl-CoA synthetase; SDH: succinic dehydrogenase; TALDO: transaldolase; TCA: tricarboxylic acid cycle; TKT: transketolase; TGS:
316
triacylglycerol synthase; TPI: triosephosphate isomerase; 6PGD: 6-phosphogluconate dehydrogenase;
317
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
318
M. smegmatis produces large droplets resembling lipid inclusions during hypoxia
319
To investigate if TAGs accumulated during hypoxia in M. smegmatis either within the
320
cytoplasm or cell envelope, we used cryo-ET to investigate changes to the ultrastructure of M.
321
smegmatis during oxygen depletion. We harvested M. smegmatis cultures at exponential
322
phase (OD600 = ~0.5, O2 = ~18%), the hypoxic transition (OD600 = 2-2.5, O2 = 0.5 1%) and
323
hypoxic stationary phase (5 days post ODmax). During hypoxic stationary phase we
324
consistently observed large and electron dense spheroid structures primarily clustered
325
towards the bacterial cell poles, which are consistent with appearance of intracellular lipid
326
inclusions in previous studies18,26,27 (Fig. 4AB). In contrast, during exponential phase, the
327
cytoplasm of M. smegmatis cells were homogenous, except for density we attributed to
328
ribosomes clustered towards the cytoplasmic membrane and the nucleoid (Fig. 4CD). We
329
were unable to obtain tomograms of cells at the hypoxic transition. The distance between the
330
inner membrane (IM) and outer membrane (OM) did not change substantially between
331
exponential phase (~24 nm) and hypoxic stationary phase (~26 nm). This is consistent with
332
the accumulation of DAGs and MAGs due to the biosynthesis and accumulation of TAGs,
333
rather than the large-scale deposition of TAGs or other glycerolipids into the cell envelope.
334
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
335
336
Figure 4: Cryo-ET of hypoxic and aerated M. smegmatis cultures. (A) Representative tomographic
337
slice of hypoxic stationary phase (3 days post ODmax) culture of M. smegmatis. White box indicates a
338
region with several lipid droplets. (B) Segmentation analysis of the same tomogram showing lipid
339
droplets (green), inner membrane (blue), outer membrane (red). (C) Representative tomographic slice
340
of aerated mid-exponential phase culture of M. smegmatis, showing homogenous cytoplasm. (D)
341
Segmentation analysis of the same tomogram shows inner membrane (blue) and outer membrane (red)
342
but no lipid droplets. Scale bars 100 nm.
343
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Discussion
344
In this study, we characterised the metabolic remodelling associated with the transition of M.
345
smegmatis to hypoxia. When exposed to a gradual onset of hypoxia, we found that M.
346
smegmatis remodelled its proteome extensively during hypoxia but before oxygen was fully
347
depleted (~0.5%), and ceased growth between 2-3% oxygen, indicating that this remaining
348
oxygen is used to support the transition to dormancy. Consistent with previous studies7,30,34,
349
the most upregulated proteins during hypoxia were members of the DosR regulon. This
350
included the fermentative hydrogenase Hyh. Consistently, we observed substantial
351
accumulations of H2 in M. smegmatis cultures, with net H2 production beginning almost
352
immediately after the depletion of oxygen below detection limits. This indicates that Hyh is
353
highly active during oxygen depletion and could support the bulk of cofactor regeneration
354
during ATP production through substrate-level phosphorylation. It should be noted that H2
355
accumulation is only observed after the complete depletion of oxygen. Previous studies show
356
that Hyh is also active during early hypoxia, but the uptake hydrogenase Huc immediately
357
recycles the H2, preventing its accumulation, indicating that M. smegmatis balances both
358
aerobic respiration and fermentative H2 production to maintain redox homeostasis7. This
359
agrees with our differential isotope labelling experiment, which indicated respiration and
360
fermentation are both active during the early stages of hypoxia. The expression of Hyh during
361
hypoxia could also be a proactive response to ensure fermentation begins as soon as all
362
oxygen is depleted, therefore minimising any disruption to central carbon metabolism.
363
Our study strongly supports our hypothesis that M. smegmatis directs carbon partially oxidised
364
during fermentative H2 production towards large reservoirs of TAGs during oxygen depletion.
365
This is due to the observed accumulation of MAGs and DAGs, the strong upregulation of TAG
366
synthases, the appearance of intracellular lipid inclusions during oxygen depletion, and the
367
rerouting of carbon through the PPP and EDP to provide NADPH for fatty acid synthesis.
368
These findings are broadly consistent with the accumulation of TAGs in M. tuberculosis during
369
hypoxia28. The metabolic scheme shown in Figure 3 shows how hypoxic cells rewire their
370
carbon metabolism to dispose excess reductant as H2 and organic carbon as TAGs. Crucially,
371
while TAG synthesis serves as a large sink of reducing equivalents for M. smegmatis during
372
hypoxia, the simultaneous activity of hydrogen fermentation indicates that there is still an
373
excess of reducing equivalents as the cell builds TAG stores, likely because the cell is only
374
able to generate ATP with pathways that also produce NADPH or NADH, i.e., the PPP, EDP
375
or glycolysis. To our knowledge, this is the first report of this hybrid mode of metabolism,
376
though it is likely that other obligate aerobes adopt comparable strategies to adapt to resource
377
variability.
378
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
The rapid and complex response to hypoxia, including hydrogen cycling and carbon storage,
379
likely reflects the dynamic aeration conditions that obligate aerobes like M. smegmatis are
380
exposed to. Soil constitutes a complex and irregular physical matrix with uneven distributions
381
of oxygen and water at the microscopic level, and bacteria may be frequently exposed to
382
oxygen depletion within their microenvironment4244. Alongside genes encoding for aerobic
383
respiration, fermentative [NiFe]-hydrogenase genes are prevalent within wetland, forest and
384
grassland soils microbial communities, suggesting that fermentative H2 production could be a
385
widespread strategy that aerobic bacteria use in response to oxygen depletion4. Moreover, in
386
environments that are transiently rather than persistently deoxygenated, the production of
387
reduced carbon stores rather than excretion of fermentative endproducts (e.g. volatile fatty
388
acids) is likely to be adaptive for multiple reasons, namely: it provides a mechanism to acquire
389
organic compounds from the environment even when oxygen is limiting; it enables rapid
390
resumption of metabolism and growth upon reoxygenation; and it avoids release of organic
391
compounds that may provide competitors (including facultative anaerobes) with an advantage.
392
It is likely that these observations also extend to other ecosystems. For example, in coastal
393
permeable sediments (i.e. sand) which frequently experience dramatic changes to redox
394
conditions over small spatial scales45, fermentative H2 production and carbon storage appear
395
to be biogeochemically and ecologically significant processes46,47. Building on these findings,
396
future studies should further investigate the roles of fermentation and carbon storage in the
397
context of dynamically aerated environments and microbial persistence.
398
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Material and Methods
399
Culturing of M. smegmatis
400
M. smegmatis mc2155 was grown from frozen glycerol stocks on lysogeny broth agar plates
401
supplemented with 0.05% (w/v) Tween80 (LBT) for 3-4 days at 37 C°. Starter cultures were
402
inoculated with colonies from LBT plates and were grown in LBT media at 37 C° and shaking
403
at 200 rpm overnight. Minimal media cultures were inoculated from turbid starter cultures to
404
OD600 = 0.003 - 0.015 and grown in 30 mL
405
supplemented with 0.4% (w/v) glucose and 0.05% tyloxapol. Minimal media cultures and
406
incubated at 37 °C with shaking at 150 rpm in 120 mL serums vials (Sigma, #33111-U) sealed
407
with rubber stoppers (Sigma, #Z166065) and aluminium caps (Sigma, #Z114146)7.
408
Measurement of hydrogen and oxygen concentrations during oxygen depletion
409
Measurements of hydrogen, oxygen and OD600 were taken from sealed cultures of M.
410
smegmatis mc2155 from early stationary phase (~24 hours post inoculation, OD600 = ~0.4) to
411
mid-stationary phase (3 days post ODmax, OD600 = 2.5 - 3). Separate sets of replicate cultures
412
were used for each type of measurement. To measure headspace O2 concentrations, a
413
Retractable Needle-Type Oxygen Minisensor (PyroScience, #OXR430) was used with a
414
FireSting Optical Oxygen meter (PyroScience, # FSO2-C4). The oxygen minisensor was
415
calibrated using 12 mL exetainers purged with N2 and ambient air in sealed 120 mL serums
416
vials incubated at 37°C. To measure headspace H2 concentrations, 200 µL of headspace was
417
sampled using a gas-tight Luer lock 0.25 mL syringes (PhaseSep, # 050051-LL) flushed with
418
N2 and immediately diluted via injection into a N2-purged 3 mL exetainer. To prevent
419
contamination of oxygen depleted cultures with oxygen, sampling was conducted while culture
420
vials were submerged in sterile milliQ water, Headspace H2 concentration was measured by
421
injecting 2 mL of diluted headspace into a pulsed discharge helium ionisation detector (model
422
TGA-6791-W-4U-2, Valco Instruments Company Inc.). A three-point calibration (100, 10 and
423
0 ppmv) curve was constructed and used to interpolate H2 concentrations.
424
Differential isotope labelling
425
Sealed cultures of M. smegmatis were grown to mid exponential (OD600 = ~0.7), hypoxic
426
transition (OD600 = ~2.5, approx. 5 hours pre ODmax) and hypoxic stationary phase (3 days post
427
ODmax) in quadruplicate for each timepoint and condition. At each timepoint, cultures were
428
treated with 50 µM of either 1-13C-, 2-13C-, 3-13C- or 13C6-glucose. To prevent contamination
429
of oxygen depleted cultures with oxygen, sampling was conducted while culture vials were
430
submerged in sterile milliQ water, and solutions of isotope labelled glucose and milliQ H2O
431
were degassed via purging with N2 for at least 30 minutes in 12 mL gastight exetainers (Labco
432
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Exetainer) and treated using a gas-tight Luer lock 0.25 mL syringes (PhaseSep, # 050051-LL)
433
flushed with N2. Exponential and hypoxic transition cultures were incubated for 5 hours post-
434
treatment, and mid-stationary phase cultures were incubated for 5 days. Immediately after
435
incubation, 12 mL of each culture was transferred to a gastight glass vial (12 mL, Labco
436
Exetainer) with no headspace and immediately terminated via treatment with 20 µL 6% HgCl2
437
solution. Prior to 13CO2 analysis, 4 mL of sample in the 12 mL vial was replaced with helium.
438
Phosphoric acid (12.5 mM) was added to the sample to convert dissolved inorganic carbon to
439
CO2 before being analysed on a Hydra 20-22 Continuous Flow Isotope Ratio Mass
440
Spectrometer (CF-IRMS; Sercon Ltd., UK).
441
Comparative metabolomics
442
Sealed cultures of M. smegmatis were harvested during exponential phase (EXP; O2 = 13%,
443
OD600 = ~1.0), the hypoxic transition (TR; O2 = 0.5%, OD600 = 2.6, 12 hours post ODmax) and
444
hypoxic stationary phase (ST; O2 = 0%, OD600 = 2.2) in quadruplicate. 15 mL of culture were
445
pelleted via centrifugation (4,500 ×g, 15 min, 4°C) and washed once via resuspension in 15
446
mL of 1X PBS, pelleting via centrifugation as before, discarding supernatant and flash freezing
447
the pellet in liquid N2. Pellets were stored at -80C. Pellets were resuspend thoroughly in 200
448
µL extraction solvent (2:6:1 chloroform:methanol:water v/v/v and spiked with 2 µM generic
449
internal standard (CHAPS/CAPS/PIPES and Tris) at 4 °C, and subject to three freeze-thaw
450
cycles by snap freezing in liquid N2 and then thawing on ice. Samples were mixed on a
451
vibrating mixer (PCV-3000, 1500 RMP, 1 sec, Vortex = Hard, 10, cycle = 60, stop) at 4 °C and
452
then centrifuged at 20,000 ×g, 10 min, 4 °C). 180 µL of supernatant was transferred into new
453
1.5 mL tube and frozen at -80 °C. Immediately prior to LC-MS analysis samples were thawed
454
on ice and centrifuged at 20,000 ×g, before being transferred to sample vial inserts.
455
LC-MS analysis was performed using a Dionex RSLC3000 UHPLC coupled to a Q-Exactive
456
Plus Orbitrap MS (Thermo). Samples were analysed by hydrophilic interaction liquid
457
chromatography (HILIC) following a previously published method48. The chromatography
458
utilized a ZIC-p(HILIC) column 5 µm 150 x 4.6 mm with a 20 x 2.1 mm ZIC-pHILIC guard
459
column (both Merck Millipore, Australia) (25 °C). A gradient elution of 20 mM ammonium
460
carbonate (A) and acetonitrile (B) (linear gradient time-%B: 0 min-80%, 15 min-50%, 18 min-
461
5%, 21 min-5%, 24 min-80%, 32 min-80%) was utilised. Flow rate was maintained at 300
462
is. MS
463
was performed at 70,000 resolution operating in rapid switching positive (4 kV) and negative
464
465
auxiliary gas flow rate 20; sweep gas 2; probe temp 120 °C). For accurate metabolite
466
identification, a standard library of ~500 metabolites were analysed before sample testing and
467
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
accurate retention time for each standard was recorded. This standard library also forms the
468
basis of a retention time prediction model used to provide putative identification of metabolites
469
not contained within the standard library49. Acquired LC-MS/MS data was processed in an
470
untargeted fashion using open source software IDEOM, which initially used msConvert
471
(ProteoWizard)50 to convert raw LC-MS files to mzXML format and XCMS to pick peaks to
472
convert to .peakML files51. Mzmatch was subsequently used for sample alignment and
473
filtering52. IDEOM was utilised for further data pre-processing, organisation and quality
474
evaluation51. Data was further analysed using MetaboAnalyst53 (www.metaboanalyst.ca/): no
475
data filtering was performed (<5000 features). Samples were normalised by median ion peak
476
ion intensity, subject log10 transformation and Pareto scaling. Volcano plots for each
477
comparison were generated (fold change threshold = 2.0, p-value (FDR) threshold: 0.01) and
478
.csv were further analysed in Microsoft Excel.
479
Map annotations from IDEOMv21 were used to broadly group metabolites by predicted
480
function. For more specific groupings (PPP, EDP, glycolysis, TCA), KEGG pathway
481
annotations (https://www.genome.jp/kegg/pathway.html) were used. Data visualisation was
482
performed in GraphPad Prism 9. Spreadsheets used for the processing and analysis of
483
metabolomics data are available in the supplementary.
484
Comparative proteomics
485
Sealed cultures of M. smegmatis were harvested in tandem with cultures harvested for
486
comparative metabolomics. 15 mL of culture were pelleted via centrifugation (4,500 x g, 15
487
min, 4°C) and washed three times via resuspension in 15 mL of 1X PBS, pelleting via
488
centrifugation as before, and discarding the supernatant. After the final wash, pellets were
489
resuspended in 1 mL PBS and transferred to a 1.5 mL tube, pelleted via centrifugation (21,000
490
×g, 10 min, 4 °C) before discarding the supernatant and flash freezing the pellet in liquid N2.
491
Cell pellets were solubilised in 5% SDS 100 mM Tris-HCl with heating at 95°C for 10 min to
492
denature enzymes. Subsequently, DNA was sheared using probe sonication, then protein
493
494
manuf
495
using the S- 54. Eluted peptides were
496
acidified to 1% tri-flouro acetic acid and purified using Stage-tips packed with SDB-RPS
497
(Empore)55, iRT peptides were spiked into all samples before LC-MS/MS analysis.
498
LC MS/MS data was acquired using a Dionex UltiMate 3000 RSLCnano for peptide separation
499
and analysed with an Orbitrap Eclipse Tribrid mass spectrometer (Thermo Scientific) with an
500
Acclaim PepMap RSLC analytical column (75 µm x 50 cm, nanoViper, C18, 2 µm, 100Å;
501
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Thermo Scientific) and an Acclaim PepMap 100 trap column (100 µm x 2 cm, nanoViper, C18,
502
5 µm, 100Å; Thermo Scientific) using a 120 minute linear gradient for separation with two
503
FAIMS compensation voltages (-45, -65)56. This was performed by the Monash Proteomics
504
and Metabolomics Facility.
505
The raw data files were analysed with the Fragpipe software suite 18.0 (MSFragger version
506
3.5, Philosopher version 4.1.1)57,58. The standard label free quantification match-between-runs
507
(LFQ-MBR) workflow was applied with no changes to workflow, employing IonQuant59 and the
508
MaxLFQ method of protein abundance calculations60. Searches were performed against the
509
M. smegmatis proteome database (September 2022) and with common contaminants. The
510
proteomics data were further analysed using LFQ-Analyst which performed data manipulation
511
and statistical tests with standard parameters61. Contaminant proteins, reverse sequences
512
converted to log2 scale and
513
missing values were imted using the Missing not At Random (MNAR) method. Protein-wise
514
linear models combined with empirical Bayesian statistics were used for differential expression
515
analyses. The limma package from R Bioconductor was used to generate a list of differentially
516
expressed proteins for each pair-wise comparison. A fold-change threshold of 2 was used and
517
an adjusted p-value threshold of 0.05 (Benjamini-Hochberg method) was used to identify
518
differentially abundant proteins62.
519
UniProt accession IDs were used to match M. smegmatis gene numbers (MSMEG_XXXX)
520
with KEGG Ortholog (KO) identifiers and KEGG module numbers. KEGG module numbers
521
were used to annotate M. smegmatis genes with associated KEGG pathways. Spreadsheets
522
used for the processing and analysis of proteomics data are available in the supplementary.
523
The mass spectrometry proteomics data have been deposited to the ProteomeXchange
524
Consortium via the PRIDE63 partner repository with the dataset identifier PXD045129.
525
Electron cryotomography data collection and processing
526
Sealed cultures of M. smegmatis were harvested during mid-exponential phase (EXP; O2 =
527
17%, OD600 = 0.5), the hypoxic transition (TR; O2 = 0.7 %, OD600 = 2.5) and mid-stationary
528
phase (ST; O2 = 0%, OD600 = 1.9, 5 days post ODmax). About 4 µl of aerated and hypoxia
529
samples were pipetted on glow-discharged Quantifoil Au extra thick EM-grids (R2/2, 200
530
mesh, Electron Microscopy Sciences). The extra fluid was blotted (blot force of 7 s and blot
531
time 5 sec) using Whatman filter paper and plunge-frozen in liquid ethane using a FEI Vitrobot
532
Mark IV, set to 22 °C and 100% humidity.
533
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Data were collected using a 300 kV FEI Titan Krios TEM equipped with a Gatan K3 direct
534
. Tilt-series were collected in movie mode
535
2, defocus of -8 µm, and pixel size of 3.4 Å.
536
Motioncor was used for motion correction64. Tilt series were aligned with IMOD and
537
subsequent 2K binned micrographs were reconstructed using Tomo 3D65,66. Tomograms were
538
segmented using Dragonfly ((Object Research Systems (ORS) Inc, Montreal, Canada,
539
software available at www.theobjects.com/dragonfly), an U-Net convolutional neural network-
540
based software. Tomograms were loaded into Dragonfly and filtered using the histogram
541
equalization filter followed by a 3D Gaussian filter to boost contrast. The feature of interest
542
was initially hand segmented to generate a multi ROI training output. This was then used for
543
unsupervised and unbiased segmentation.
544
Footnotes
545
546
Competing interests: The authors declare no competing interests.
547
548
Acknowledgements: This work was supported by three National Health & Medical Research
549
Council Emerging Leader Fellowships (APP1178715 to C.G.; APP1197376 to R.G. and
550
APP1196924 to D.G.), three Australian Research Council Discovery Project grants
551
(DP200103074 to C.G. and R.G.; DP230103080 to C.G. and R.G.; DP210101595 to C.G. and
552
P.L.M.C.), and an Australian Government Research Training Stipend Scholarship (to D.L.G.).
553
We thank Matthew Johnson and Ian Holmes Imaging Center (Bio21 Institute, University of
554
Melbourne) for help with cryo-ET data collection.
555
556
Author contributions: C.G. conceived and oversaw the study. C.G., D.L.G., P.L.M.C., R.G.,
557
and D.G. designed experiments. D.L.G. and J.L. was responsible for growth, oxygen
558
consumption, and hydrogen production experiments. D.L.G., T.H., J.L., W.W.W., and P.L.M.C.
559
conducted isotopologue analysis. D.L.G., H.L., J.R.S., I.H., E.T., R.B.S., and C.B. conducted
560
proteomic and metabolomic analysis. M.M, D.G., and D.L.G. conducted cryo-ET analysis.
561
T.D.W. and L.J. conducted lipid analysis. D.L.G. and C.G. wrote the manuscript with input
562
from all authors.
563
564
565
References
566
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
1. Greening, C., Grinter, R., and Chiri, E. “Uncovering the Metabolic Strategies of the Dormant
567
Microbial Majority: towards Integrative Approaches,” mSystems, V. 4, No. 3, 2019, pp. e00107-
568
19.
569
2. Lennon, J. T., and Jones, S. E. “Microbial seed banks: the ecological and evolutionary
570
implications of dormancy,” Nature Reviews Microbiology, V. 9, No. 2, 2011, pp. 119–30.
571
3. Blagodatskaya, E., and Kuzyakov, Y. “Active microorganisms in soil: Critical review of estimation
572
criteria and approaches,” Soil Biology and Biochemistry, V. 67, 2013, pp. 192–211.
573
4. Bay, S. K., Dong, X., Bradley, J. A., et al. “Trace gas oxidizers are widespread and active members
574
of soil microbial communities,” Nature Microbiology, V. 6, No. 2, 2021, pp. 246–56.
575
5. Smeulders, M. J., Keer, J., Speight, R. A., et al. “Adaptation of Mycobacterium smegmatis to
576
Stationary Phase,” Journal of Bacteriology, V. 181, No. 1, 1999, pp. 270–83.
577
6. Walsh, C. M., Gebert, M. J., Delgado-Baquerizo, M., et al. “A Global Survey of Mycobacterial
578
Diversity in Soil,” Applied and Environmental Microbiology, V. 85, No. 17, 2019, pp. e01180-19.
579
7. Berney, M., Greening, C., Conrad, R., et al. “An obligately aerobic soil bacterium activates
580
fermentative hydrogen production to survive reductive stress during hypoxia,” Proceedings of
581
the National Academy of Sciences, V. 111, No. 31, 2014, pp. 11479–84.
582
8. Cordero, P. R. F., Bayly, K., Man Leung, P., et al. “Atmospheric carbon monoxide oxidation is a
583
widespread mechanism supporting microbial survival,” The ISME Journal, V. 13, No. 11, 2019,
584
pp. 2868–81.
585
9. Greening, C., Villas-Bôas, S. G., Robson, J. R., et al. “The Growth and Survival of Mycobacterium
586
smegmatis Is Enhanced by Co-Metabolism of Atmospheric H2,” PLoS ONE, V. 9, No. 7, 2014, p.
587
e103034.
588
10. King, G. M. “Uptake of Carbon Monoxide and Hydrogen at Environmentally Relevant
589
Concentrations by Mycobacteria†,” Applied and Environmental Microbiology, V. 69, No. 12,
590
2003, pp. 7266–72.
591
11. Nakano, M. M., and Zuber, P. “Anaerobic growth of a ‘strict aerobe’ (Bacillus subtilis),” Annual
592
Review of Microbiology, V. 52, 1998, pp. 165–90.
593
12. Giraffa, G. “Overview of the ecology and biodiversity of the LAB.” In: Holzapfel, W. H., Wood, B.
594
J. B., eds. Lactic Acid Bacteria. Chichester, UK, John Wiley & Sons, Ltd, 2014. pp. 45–54.
595
13. Dashko, S., Zhou, N., Compagno, C., et al. “Why, when, and how did yeast evolve alcoholic
596
fermentation?,” FEMS Yeast Research, V. 14, No. 6, 2014, pp. 826–32.
597
14. Watanabe, S., Zimmermann, M., Goodwin, M. B., et al. “Fumarate Reductase Activity Maintains
598
an Energized Membrane in Anaerobic Mycobacterium tuberculosis,” PLoS Pathogens, V. 7, No.
599
10, 2011, p. e1002287.
600
15. Vijay, S., Hai, H. T., Thu, D. D. A., et al. “Ultrastructural Analysis of Cell Envelope and
601
Accumulation of Lipid Inclusions in Clinical Mycobacterium tuberculosis Isolates from Sputum,
602
Oxidative Stress, and Iron Deficiency,” Frontiers in Microbiology, V. 8, 2018, p. 2681.
603
16. Muñoz-Elías, E. J., and McKinney, J. D. “Mycobacterium tuberculosis isocitrate lyases 1 and 2 are
604
jointly required for in vivo growth and virulence,” Nature Medicine, V. 11, No. 6, 2005, pp. 638–
605
44.
606
17. Martinot, A. J., Farrow, M., Bai, L., et al. “Mycobacterial Metabolic Syndrome: LprG and Rv1410
607
Regulate Triacylglyceride Levels, Growth Rate and Virulence in Mycobacterium tuberculosis,”
608
PLOS Pathogens, V. 12, No. 1, 2016, p. e1005351.
609
18. Low, K. L., Rao, P. S. S., Shui, G., et al. “Triacylglycerol Utilization Is Required for Regrowth of In
610
Vitro Hypoxic Nonreplicating Mycobacterium bovis Bacillus Calmette-Guerin,” Journal of
611
Bacteriology, V. 191, No. 16, 2009, pp. 5037–43.
612
19. Low, K. L., Shui, G., Natter, K., et al. “Lipid Droplet-associated Proteins Are Involved in the
613
Biosynthesis and Hydrolysis of Triacylglycerol in Mycobacterium bovis Bacillus Calmette-
614
Guérin*,” Journal of Biological Chemistry, V. 285, No. 28, 2010, pp. 21662–70.
615
20. Burdon, K. L. “Disparity in Appearance of True Hansen’s Bacilli and Cultured ‘Leprosy Bacilli’
616
When Stained for Fat.,” Journal of bacteriology, V. 52, No. 6, 1946, pp. 679–80.
617
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
21. Knaysi, G., Hillier, J., and Fabricant, C. “THE CYTOLOGY OF AN AVIAN STRAIN OF
618
MYCOBACTERIUM TUBERCULOSIS STUDIED WITH THE ELECTRON AND LIGHT MICROSCOPES,”
619
Journal of Bacteriology, V. 60, No. 4, 1950, pp. 423–47.
620
22. Schaefer, W. B., and Lewis, C. W. “Effect of Oleic Acid on Growth and Cell Structure of
621
Mycobacteria,” Journal of Bacteriology, V. 90, No. 5, 1965, pp. 1438–47.
622
23. Boxtel, R. M., Lambrecht, R. S., and Collins, M. T. “Effect of polyoxyethylene sorbate compounds
623
(Tweens) on colonial morphology, growth, and ultrastructure of Mycobacterium
624
paratuberculosis,” APMIS, V. 98, Nos. 7–12, 1990, pp. 901–8.
625
24. Peyron, P., Vaubourgeix, J., Poquet, Y., et al. “Foamy Macrophages from Tuberculous Patients’
626
Granulomas Constitute a Nutrient-Rich Reservoir for M. tuberculosis Persistence,” PLoS
627
Pathogens, V. 4, No. 11, 2008, p. e1000204.
628
25. Garton, N. J., Waddell, S. J., Sherratt, A. L., et al. “Cytological and Transcript Analyses Reveal Fat
629
and Lazy Persister-Like Bacilli in Tuberculous Sputum,” PLoS Medicine, V. 5, No. 4, 2008, p. e75.
630
26. Viljoen, A., Blaise, M., de Chastellier, C., et al. “MAB_3551c encodes the primary triacylglycerol
631
synthase involved in lipid accumulation in M ycobacterium abscessus: TAG accumulation in
632
Mycobacterium abscessus,” Molecular Microbiology, V. 102, No. 4, 2016, pp. 611–27.
633
27. Maurya, R. K., Bharti, S., and Krishnan, M. Y. “Triacylglycerols: Fuelling the Hibernating
634
Mycobacterium tuberculosis,” Frontiers in Cellular and Infection Microbiology, V. 8, 2019, p. 450.
635
28. Galagan, J. E., Minch, K., Peterson, M., et al. “The Mycobacterium tuberculosis regulatory
636
network and hypoxia,” Nature, V. 499, No. 7457, 2013, pp. 178–83.
637
29. Daniel, J., Deb, C., Dubey, V. S., et al. “Induction of a Novel Class of Diacylglycerol
638
Acyltransferases and Triacylglycerol Accumulation in Mycobacterium tuberculosis as It Goes into
639
a Dormancy-Like State in Culture,” Journal of Bacteriology, V. 186, No. 15, 2004, pp. 5017–30.
640
30. Park, H.-D., Guinn, K. M., Harrell, M. I., et al. “Rv3133c/dosR is a transcription factor that
641
mediates the hypoxic response of Mycobacterium tuberculosis: DosR and the M. tuberculosis
642
hypoxic response,” Molecular Microbiology, V. 48, No. 3, 2003, pp. 833–43.
643
31. Bansal-Mutalik, R., and Nikaido, H. “Mycobacterial outer membrane is a lipid bilayer and the
644
inner membrane is unusually rich in diacyl phosphatidylinositol dimannosides,” Proceedings of
645
the National Academy of Sciences, V. 111, No. 13, 2014, pp. 4958–63.
646
32. Hayashi, J. M., Luo, C.-Y., Mayfield, J. A., et al. “Spatially distinct and metabolically active
647
membrane domain in mycobacteria,” Proceedings of the National Academy of Sciences, V. 113,
648
No. 19, 2016, pp. 5400–5.
649
33. Rastogi, S., Singh, A. K., Chandra, G., et al. “The diacylglycerol acyltransferase Rv3371 of
650
Mycobacterium tuberculosis is required for growth arrest and involved in stress-induced cell
651
wall alterations,” Tuberculosis, V. 104, 2017, pp. 8–19.
652
34. Berney, M., and Cook, G. M. “Unique Flexibility in Energy Metabolism Allows Mycobacteria to
653
Combat Starvation and Hypoxia,” PLoS ONE, V. 5, No. 1, 2010, p. e8614.
654
35. Wayne, L. G., and Hayes, L. G. “An in vitro model for sequential study of shiftdown of
655
Mycobacterium tuberculosis through two stages of nonreplicating persistence,” Infection and
656
Immunity, V. 64, No. 6, 1996, pp. 2062–9.
657
36. Hutchinson, T. F., Kessler, A. J., Wong, W. W., et al. “Microorganisms oxidise organic carbon
658
through distinct pathways in permeable and cohesive sediments,” (in revision, ISME Journal),
659
2023.
660
37. Peterson, E. J. R., Abidi, A. A., Arrieta-Ortiz, M. L., et al. “Intricate Genetic Programs Controlling
661
Dormancy in Mycobacterium tuberculosis,” Cell Reports, V. 31, No. 4, 2020, p. 107577.
662
38. Harold, L. K., Antoney, J., Ahmed, F. H., et al. “FAD-sequestering proteins protect mycobacteria
663
against hypoxic and oxidative stress,” Journal of Biological Chemistry, V. 294, No. 8, 2019, pp.
664
2903–5814.
665
39. Dulberger, C. L., Rubin, E. J., and Boutte, C. C. “The mycobacterial cell envelope — a moving
666
target,” Nature Reviews Microbiology, V. 18, No. 1, 2020, pp. 47–59.
667
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
40. Corkins, M. E., Wilson, S., Cocuron, J.-C., et al. “The gluconate shunt is an alternative route for
668
directing glucose into the pentose phosphate pathway in fission yeast,” Journal of Biological
669
Chemistry, V. 292, No. 33, 2017, pp. 13823–32.
670
41. Zhong, W., Guo, J., Cui, L., et al. “Pyruvate Kinase Regulates the Pentose-Phosphate Pathway in
671
Response to Hypoxia in Mycobacterium tuberculosis,” Journal of Molecular Biology, V. 431, No.
672
19, 2019, pp. 3690–705.
673
42. Tiedje, J. M., Sexstone, A. J., Parkin, T. B., et al. “Anaerobic processes in soil,” Plant and Soil, V.
674
76, Nos. 1–3, 1984, pp. 197–212.
675
43. Sexstone, A. J., Revsbech, N. P., Parkin, T. B., et al. “Direct Measurement of Oxygen Profiles and
676
Denitrification Rates in Soil Aggregates,” Soil Science Society of America Journal, V. 49, No. 3,
677
1985, pp. 645–51.
678
44. Fan, Z., Neff, J. C., Waldrop, M. P., et al. “Transport of oxygen in soil pore-water systems:
679
implications for modeling emissions of carbon dioxide and methane from peatlands,”
680
Biogeochemistry, V. 121, No. 3, 2014, pp. 455–70.
681
45. Ahmerkamp, S., Winter, C., Krämer, K., et al. “Regulation of benthic oxygen fluxes in permeable
682
sediments of the coastal ocean: Regulation of benthic oxygen fluxes,” Limnology and
683
Oceanography, V. 62, No. 5, 2017, pp. 1935–54.
684
46. Chen, Y.-J., Leung, P. M., Cook, P. L. M., et al. “Hydrodynamic disturbance controls microbial
685
community assembly and biogeochemical processes in coastal sediments,” The ISME Journal, V.
686
16, No. 3, 2022, pp. 750–63.
687
47. Kessler, A. J., Chen, Y.-J., Waite, D. W., et al. “Bacterial fermentation and respiration processes
688
are uncoupled in anoxic permeable sediments,” Nature Microbiology, V. 4, No. 6, 2019, pp.
689
1014–23.
690
48. Stoessel, D., Nowell, C. J., Jones, A. J., et al. “Metabolomics and lipidomics reveal perturbation of
691
sphingolipid metabolism by a novel anti-trypanosomal 3-(oxazolo[4,5-b]pyridine-2-yl)anilide,”
692
Metabolomics, V. 12, No. 7, 2016, p. 126.
693
49. Creek, D. J., Jankevics, A., Breitling, R., et al. “Toward Global Metabolomics Analysis with
694
Hydrophilic Interaction Liquid Chromatography–Mass Spectrometry: Improved Metabolite
695
Identification by Retention Time Prediction,” Analytical Chemistry, V. 83, No. 22, 2011, pp.
696
8703–10.
697
50. Chambers, M. C., Maclean, B., Burke, R., et al. “A cross-platform toolkit for mass spectrometry
698
and proteomics,” Nature Biotechnology, V. 30, No. 10, 2012, pp. 918–20.
699
51. Creek, D. J., Jankevics, A., Burgess, K. E. V., et al. “IDEOM: an Excel interface for analysis of LC-
700
MS-based metabolomics data,” Bioinformatics (Oxford, England), V. 28, No. 7, 2012, pp. 1048–9.
701
52. Scheltema, R. A., Jankevics, A., Jansen, R. C., et al. “PeakML/mzMatch: A File Format, Java
702
Library, R Library, and Tool-Chain for Mass Spectrometry Data Analysis,” Analytical Chemistry, V.
703
83, No. 7, 2011, pp. 2786–93.
704
53. Pang, Z., Chong, J., Zhou, G., et al. “MetaboAnalyst 5.0: narrowing the gap between raw spectra
705
and functional insights,” Nucleic Acids Research, V. 49, No. W1, 2021, pp. W388–96.
706
54. HaileMariam, M., Eguez, R. V., Singh, H., et al. “S-Trap, an Ultrafast Sample-Preparation
707
Approach for Shotgun Proteomics,” Journal of Proteome Research, V. 17, No. 9, 2018, pp. 2917–
708
24.
709
55. Humphrey, S. J., Karayel, O., James, D. E., et al. “High-throughput and high-sensitivity
710
phosphoproteomics with the EasyPhos platform,” Nature Protocols, V. 13, No. 9, 2018, pp.
711
1897–916.
712
56. Ayers, K. L., Eggers, S., Rollo, B. N., et al. “Variants in SART3 cause a spliceosomopathy
713
characterised by failure of testis development and neuronal defects,” Nature Communications,
714
V. 14, No. 1, 2023, p. 3403.
715
57. Kong, A. T., Leprevost, F. V., Avtonomov, D. M., et al. “MSFragger: ultrafast and comprehensive
716
peptide identification in mass spectrometry–based proteomics,” Nature Methods, V. 14, No. 5,
717
2017, pp. 513–20.
718
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
58. Teo, G. C., Polasky, D. A., Yu, F., et al. “Fast Deisotoping Algorithm and Its Implementation in the
719
MSFragger Search Engine,” Journal of Proteome Research, V. 20, No. 1, 2021, pp. 498–505.
720
59. Yu, F., Haynes, S. E., and Nesvizhskii, A. I. “IonQuant Enables Accurate and Sensitive Label-Free
721
Quantification With FDR-Controlled Match-Between-Runs,” Molecular & Cellular Proteomics, V.
722
20, 2021.
723
60. Cox, J., Neuhauser, N., Michalski, A., et al. “Andromeda: A Peptide Search Engine Integrated into
724
the MaxQuant Environment,” Journal of Proteome Research, V. 10, No. 4, 2011, pp. 1794–805.
725
61. Shah, A. D., Goode, R. J. A., Huang, C., et al. “LFQ-Analyst: An Easy-To-Use Interactive Web
726
Platform To Analyze and Visualize Label-Free Proteomics Data Preprocessed with MaxQuant,”
727
Journal of Proteome Research, V. 19, No. 1, 2020, pp. 204–11.
728
62. Pasternak, M., Verhoef, J. M. J., Wong, W., et al. “RhopH2 and RhopH3 export enables assembly
729
of the RhopH complex on P. falciparum-infected erythrocyte membranes,” Communications
730
Biology, V. 5, No. 1, 2022, p. 333.
731
63. Deutsch, E. W., Bandeira, N., Perez-Riverol, Y., et al. “The ProteomeXchange consortium at 10
732
years: 2023 update,” Nucleic Acids Research, V. 51, No. D1, 2023, pp. D1539–48.
733
64. Zheng, S. Q., Palovcak, E., Armache, J.-P., et al. “MotionCor2: anisotropic correction of beam-
734
induced motion for improved cryo-electron microscopy,” Nature Methods, V. 14, No. 4, 2017,
735
pp. 331–2.
736
65. Kremer, J. R., Mastronarde, D. N., and McIntosh, J. R. “Computer Visualization of Three-
737
Dimensional Image Data Using IMOD,” Journal of Structural Biology, V. 116, No. 1, 1996, pp. 71–
738
6.
739
66. Agulleiro, J.-I., and Fernandez, J.-J. “Tomo3D 2.0 – Exploitation of Advanced Vector eXtensions
740
(AVX) for 3D reconstruction,” Journal of Structural Biology, V. 189, No. 2, 2015, pp. 147–52.
741
67. Eoh, H., and Rhee, K. Y. “Multifunctional essentiality of succinate metabolism in adaptation to
742
hypoxia in Mycobacterium tuberculosis,” Proceedings of the National Academy of Sciences, V.
743
110, No. 16, 2013, pp. 6554–9.
744
68. Eoh, H., and Rhee, K. Y. “Methylcitrate cycle defines the bactericidal essentiality of isocitrate
745
lyase for survival of Mycobacterium tuberculosis on fatty acids,” Proceedings of the National
746
Academy of Sciences, V. 111, No. 13, 2014, pp. 4976–81.
747
69. McKinney, J. D., zu Bentrup, K. H., Muñoz-Elías, E. J., et al. “Persistence of Mycobacterium
748
tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase,”
749
Nature, V. 406, No. 6797, 2000, pp. 735–8.
750
70. Gould, T. A., van de Langemheen, H., Munoz-Elias, E. J., et al. “Dual role of isocitrate lyase 1 in
751
the glyoxylate and methylcitrate cycles in Mycobacterium tuberculosis,” Molecular
752
Microbiology, V. 61, No. 4, 2006, pp. 940–7.
753
71. Cooper, R. A., and Kornberg, H. L. “The mechanism of the phosphoenolpyruvate synthase
754
reaction,” Biochimica et Biophysica Acta (BBA) - General Subjects, V. 141, No. 1, 1967, pp. 211–3.
755
72. Hutchins, A. M., Holden, J. F., and Adams, M. W. “Phosphoenolpyruvate synthetase from the
756
hyperthermophilic archaeon Pyrococcus furiosus,” Journal of Bacteriology, V. 183, No. 2, 2001,
757
pp. 709–15.
758
73. Lim, J., Lee, J. J., Lee, S.-K., et al. “Phosphoenolpyruvate depletion mediates both growth arrest
759
and drug tolerance of Mycobacterium tuberculosis in hypoxia,” Proceedings of the National
760
Academy of Sciences, V. 118, No. 35, 2021, p. e2105800118.
761
74. Baek, S.-H., Li, A. H., and Sassetti, C. M. “Metabolic Regulation of Mycobacterial Growth and
762
Antibiotic Sensitivity,” PLoS Biology, V. 9, No. 5, 2011, p. e1001065.
763
75. Snášel, J., Machová, I., Šolínová, V., et al. “Phosphofructokinases A and B from Mycobacterium
764
tuberculosis Display Different Catalytic Properties and Allosteric Regulation,” International
765
Journal of Molecular Sciences, V. 22, No. 3, 2021, p. 1483.
766
76. Phong, W. Y., Lin, W., Rao, S. P. S., et al. “Characterization of Phosphofructokinase Activity in
767
Mycobacterium tuberculosis Reveals That a Functional Glycolytic Carbon Flow Is Necessary to
768
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Limit the Accumulation of Toxic Metabolic Intermediates under Hypoxia,” PLoS ONE, V. 8, No. 2,
769
2013, p. e56037.
770
77. Martini, M. C., Zhou, Y., Sun, H., et al. “Defining the Transcriptional and Post-transcriptional
771
Landscapes of Mycobacterium smegmatis in Aerobic Growth and Hypoxia,” Frontiers in
772
Microbiology, V. 10, 2019, p. 591.
773
774
775
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
Supplementary Note 1: Additional proteomic and metabolomic analyses
776
The proteomic response of M. smegmatis differs in fundamental ways to M. tuberculosis.
777
Studies in M. tuberculosis have demonstrated that rerouting of the carbon flux from the TCA
778
779
response to oxygen depletion, and contrary to our findings, isocitrate lyase (icl) is upregulated
780
during hypoxia and essential during infection14,16,6769. It has also been shown these pathways
781
drive an accumulation of succinate during oxygen depletion, which M. tuberculosis excretes
782
to maintain membrane potential14,67. Consistent with the downregulation of icl during oxygen
783
depletion in M. smegmatis, we observed no accumulation of succinate, and in fact detected a
784
60-fold reduction of succinate in the stationary phase vs exponential phase comparison (Fig.
785
3). These differences may be in part due to the fatty acid rich diet of M. tuberculosis during
786
infection, as the methylcitrate pathway is required for the catabolism of odd-chain fatty
787
acids68,70, and the glyoxylate shunt is required for the regeneration of TCA cycle intermediates
788
when relying on fatty acids for energy and carbon67. The differences may also be due to
789
additional metabolic flexibility conferred by H2 fermentation, as M. tuberculosis excretes
790
succinate during hypoxia in order to maintain membrane potential and ATP production14,67,
791
while M. smegmatis can continually produce ATP via substrate level phosphorylation by
792
dispensing of reducing equivalents via Hyh.
793
With respect to the DosR-regulated genes, MSMEG_3934 is annotated as a
794
phosphoenolpyruvate synthase (PEPS) that catalyses the gluconeogenic reaction ATP + H2O
795
+ pyruvate AMP + 2H+ + phosphate + phosphoenolpyruvate71,72. PEPS has no homolog in
796
M. tuberculosis, and its predicted function has not been experimentally investigated, and
797
considering it is the second most upregulated protein in the hypoxic transition vs exponential
798
phase comparison, it is somewhat difficult to reconcile with our data where we observe a
799
decrease in abundance in phosphoenolpyruvate (3-fold and 18-fold, during hypoxic transition
800
vs exponential phase and stationary phase vs exponential phase, respectively), while
801
observing no changes to the abundance of pyruvate (Fig. 3). Although we observed a
802
decrease in the abundance of PEP during oxygen depletion, and no change in the abundance
803
of pyruvate, PEPS mediated flux may happen very early in the response to hypoxia to arrest
804
growth. Similarly, in M. tuberculosis, which lacks a PEPS homolog, the metabolic rerouting
805
towards TAG synthesis plays a key role in growth arrest by depleting PEP73 and acetyl-CoA74,
806
and rerouting towards the PPP is mediated by pyruvate kinase41. We can only speculate about
807
possible posttranscriptional regulation when the abundance of metabolites and associated
808
enzymes do not correlate, though investigation of these mechanisms require intensive
809
biochemical characterisation75.
810
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint
The homolog of PfkB in M. tuberculosis is also upregulated during hypoxia, can catalyse the
811
reverse gluconeogenic reaction (albeit with low efficiency), and exhibits lower activity,
812
sensitivity to allosteric regulators and substrate specificity than PfkA, which is essential for
813
growth on glucose75,76. We did not identify PfkA in our proteomics dataset, but it was not found
814
to be differentially regulated in other microarray34 and transcriptomic77 datasets comparing
815
hypoxic and aerated cultures of M. smegmatis. The glycolysis enzyme PksB potentially has
816
an allosteric role given its lower activity, ability to catalyse the reverse gluconeogenic reaction,
817
and recalcitrance to metabolites in M. tuberculosis. It may allosterically regulate its isozyme
818
PksA, playing a pivotal role in redirecting glycolytic flux to the PPP / EDP, if these properties
819
are conserved in M. smegmatis.
820
We also observed changes in cofactor levels during hypoxia based on the metabolomics. Both
821
the PPP and EDP reduce NADP+ to NADPH, which in turn provides reductive power for
822
anabolism. During the hypoxic transition vs exponential phase comparison, we only observed
823
changes to NAD+/NADH, which both increased in abundance by 2-fold (Fig. 3). However, we
824
observed a dramatic decrease in the abundance of both NADP+/NADPH (80- to 92-fold
825
respectively), in addition to NAD+/NADH (38- to 13-fold, respectively) and ADP (9-fold, ATP
826
not detected), during the stationary phase vs exponential phase comparison (Fig. 3). This
827
likely reflects a dramatic downshift in overall metabolism and energy expenditure as M.
828
smegmatis enters non-replicative persistence. Due to the nature of our comparative analysis,
829
compared to absolute quantification of metabolites, we were unable to calculate the
830
stoichiometric ratios of each redox cofactor pair.
831
832
preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for thisthis version posted September 12, 2023. ; https://doi.org/10.1101/2023.09.11.557286doi: bioRxiv preprint