Bacillus subtilis at near-zero specific growth rates Overkamp, Wout
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2015
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Overkamp, W. (2015). Bacillus subtilis at near-zero specific growth rates: adaptations to extreme caloric restriction. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
5
Transcriptome-wide analysis of Bacillus subtilis at near-zero specific growth rates
Part of this chapter was published in:
Wout Overkamp, Onur Ercan, Martijn Herber, Antonius J. A. van Maris, Michiel Kleerebezem and Oscar P. Kuipers. Physiological and cell morphology adaptation of Bacillus subtilis at near-zero specific growth rates: a transcriptome analysis.
Environmental Microbiology 17, Issue 2, pages 346–363 (2015).
1
Abstract
2B. subtilis was cultured in a retentostat at extremely low growth rates and the 3
adaptation of B. subtilis to these near-zero growth conditions was studied by analysis 4
of the changes in the transcriptome and genome. During retentostat culturing the 5
specific growth rate decreased to a minimum of 0.00006 h-1. Transcriptome analysis 6
revealed that cellular responses to near-zero growth conditions share several 7
similarities with those of cells during the stationary phase of batch-growth. However, 8
fundamental differences between these two non-growing states are apparent by their 9
high viability and absence of stationary phase mutagenesis under near-zero growth 10
conditions. Stress resistance mechanisms were only mildly induced in response to the 11
progressively decreasing growth-rate and while glucose was still supplied, cells were 12
ready for utilization of alternative carbon sources. Genome resequencing, of samples 13
taken 40 days after inoculation to reach zero-growth conditions, indicated that only 14
minor changes in the genome occurred and that these most likely did not play a role 15
in the transcriptional responses. 16
Introduction
17Nutrient availability generally limits growth of microorganisms in natural 18
environments. Hence, high microbial growth rates as achieved in laboratory batch 19
cultures are probably rare in nature (Brock, 1971; Ferenci, 2001; Koch, 1997). 20
Therefore, to understand microbial life in natural environments, obtaining 21
knowledge about physiology at near-zero growth rates is very relevant. Additionally, 22
near-zero growth rates might contribute to uncoupling of product formation from 23
growth in industrial biotechnology. This to improve product yields, since biomass is 24
often an undesired byproduct in industrial processes. 25
Zero-growth is a metabolically active, non-growing state of a microorganism 26
and is fundamentally different from starvation encountered during stationary phase, 27
which involves deterioration of physiological processes. It is based on the idea, called 28
maintenance energy (Pirt, 1965), that a cell uses a specific minimum amount of 29
energy to fuel basal household processes and to remain viable. Thus, when the 30
amount of energy substrates available for the individual cell becomes limiting and 31
decreases to a point where it equals the maintenance energy requirement, 32
theoretically a state of zero-growth should be reached (van Verseveld et al., 1986). 33
In batch cultivations, where highly metabolically active cells proliferate at their 34
maximum growth rates until nutrients are depleted and starvation conditions are 35
induced, the transition from exponential to stationary phase is rapid and transient. 36
This makes it difficult to specifically study cells at near-zero growth rates. 37
Alternatively, a chemostat culture allows direct manipulation of the growth rate by 38
varying the dilution rate, and provides the additional advantage of a controlled and 39
constant environmental condition (Herbert et al., 1956; Novick and Szilard, 1950). 40
However, extremely low specific growth rates cannot be achieved in chemostats, due 41
to ‘feast and famine’ dynamics caused by the dropwise feeding of medium (Boender 42
et al., 2011; Daran-Lapujade et al., 2009; Herbert et al., 1956). 43
To be able to study microbes at extremely low specific growth rates, retentostat 44
cultivation has been developed (Herbert, 1961; van Verseveld et al., 1986). A 45
retentostat, or recycling fermentor, is a chemostat in which all the biomass is retained 46
by a filter in the effluent tube. Growing a culture at a fixed dilution rate on an 47
energy-limited medium leads to accumulation of biomass and a progressive decrease 48
of energy substrate availability per biomass. Consequently the substrate consumption 49
rate will asymptotically approach the substrate requirement for maintenance 50
processes, ultimately resulting in near-zero growth rates. Meanwhile, starvation is 51
prevented in this setup because substrate supply continues (Boender et al., 2009; 52
Chesbro et al., 1979; Goffin et al., 2010; Herbert, 1961; Tappe et al., 1996; van 53
Verseveld et al., 1986). 54
B. subtilis possesses many strategies to survive fluctuating environmental 55
conditions, for example, the ability to develop natural competence and motility, 56
secrete exoproteases, form biofilms and eventually form highly resistant spores 57
(Branda et al., 2001; Dubnau, 1991; Errington, 2003; Kearns and Losick, 2005; 58
Msadek, 1999; Veening et al., 2008). Many survival responses are triggered by 59
nutrient scarcity, as is determined by studies on the transcriptional response of B. 60
subtilis to glucose starvation encountered in stationary phase batch cultures (Blom et 61
al., 2011; de Jong et al., 2012; Koburger et al., 2005; Otto et al., 2010). However, 62
retentostat cultures with near-zero specific growth rates caused by glucose limitation 63
are an unexplored area. 64
Previously, we have implemented retentostat cultivation for growth of B. subtilis 65
under aerobic, glucose-limited conditions and demonstrated that a specific growth 66
rate of 0.00006 h-1 could be reached reproducibly while the cells remained viable 67
(Chapter 4). The specific growth rate reached corresponds to a doubling time of 68
470 days. The aim of this study is to investigate the transcriptional response of B. 69
subtilis at near-zero specific growth rates. Therefore, we have analyzed the 70
transcriptome of retentostat cultures during the decrease of the specific growth rate 71
and compared it to faster-growing chemostat cultures. The caloric restriction 72
encountered during retentostat cultivation clearly was reflected in the B. subtilis 73
transcriptome, which established that adaptations to extremely low growth rates 74
display similarity to cells that are progressing from growing to non-growing growth 75
phases in a batch culture. However, the transcriptome analyses also indicated that 76
the slow progressive transition towards the non-growing state in a retentostat with 77
controlled environmental conditions yields a condition fundamentally different from 78
the abrupt entry into stationary phase. In this non-growing state the assumption is 79
that cryptic growth, e.g. the lysis of cells that are replaced at the same rate by the 80
growth of others (Ryan, 1959), is very limited. Consequently, growth-related genome 81
mutation rates (mutations arising during deoxyribonucleic acid (DNA) replication 82
within actively dividing cells), are most likely not very numerous (Drake, 1991; Drake 83
et al., 1998; Barrick et al., 2009). Therefore genome sequencing was used to 84
corroborate that very limited numbers of mutations are found in the zero-growth 85
cultures. 86
Results
87Cultivation of B. subtilis at near-zero growth rates 88
As described in Chapter 4, B. subtilis 168 trp- sigF::spec amyE::PrrnB-GFP was grown 89
under aerobic retentostat conditions in chemically defined M9 medium with glucose 90
as the growth limiting substrate. Two independent retentostat cultivations were 91
successfully performed for 42 and 40 days (retentostat 1 and 2, respectively) to study 92
the transcriptional response of B. subtilis to near-zero specific growth rates. During 93
retentostat culturing the specific growth rate (μ) decreased to a minimum of 0.00006 94
h-1, corresponding to a doubling time of 470 days (Fig. 1). The energy distribution 95
between growth- and maintenance related processes showed that a state of near-zero 96
growth was reached (Chapter 4). Remarkably, during the retentostat cultivation a 97
filamentous morphology emerged (Fig. 2 and Chapter 4). 98
Figure 1. Growth of B. subtilis in retentostat cultures. Steady-state aerobic chemostat 99
cultures (D = 0.025 h-1) were switched to retentostat mode at time-point zero. Displayed are data 100
from retentostat cultivation 1 (■) and 2 (□). (A) Measured biomass concentration (gdw l-1). Data 101
points represent mean + standard deviation of duplicate samples. Additionally, the biomass 102
calculated with the fitted van Verseveld equation for retentostat 1 (---) and 2 (…) is shown, as 103
well as the corresponding calculated specific growth rates ((●) and (○), respectively). Time-points 104
analysed with transcriptomics are encircled for both retentostat cultivation 1 (0, 7, 18 and 42 105
days) and 2 (0, 6, 20 and 40 days). These encircled time-points are referred to as steady-state 106
chemostats, time-points 1, time-points 2 and time-points 3. 107
Transcriptome analysis of the retentostat cultures and overview of cellular 108
processes regulated at the transcriptional level 109
Transcriptome analysis was performed at 3 time-points of the independent duplicate 110
retentostat cultures, using chemostat-grown cells (μ = 0.025 h-1) as a reference (Fig. 111
1). The multiple time points during retentostat cultivation correspond to an 112
increasing fraction of glucose used for maintenance purposes and hence to decreasing 113
specific growth rates (Table 1). 114
Approximately 12% of the genes were found to be differentially expressed at the 115
end of retentostat cultivation; (amount of regulated genes in the individual 116
retentostats can be found in Table 1). A total of 136 genes exhibited an increased 117
relative mRNA level, whereas 377 genes exhibited a decreased relative mRNA level. 118
A complete list of differentially expressed genes, including transcript ratios and 119
statistical significance, has been deposited at Gene Expression Omnibus database 120
(GEO; GSE55690). 121
The genes most prominently induced by zero-growth conditions in all time- 122
points are involved in glutamine uptake, fatty acid degradation and glucomannan 123
uptake/utilization (Table S1). The most strongly repressed genes are involved in 124
fructose uptake, mannitol uptake and methionine salvage (Table S1). Analysis of the 125
transcriptome data on overrepresented functional categories in clusters of up- and 126
down-regulated genes revealed that transport and metabolism of carbohydrates and 127
of amino acids were enriched mostly among down-regulated genes, but also among 128
some up-regulated genes (Fig. 3). Ribosomal- and motility genes were prominent 129
categories among the repressed genes, which also encompassed many genes encoding 130
Figure 2. Morphological changes of B. subtilis during retentostat cultivation. During a period of 42 and 40 days an elongated morphology emerged in retentostat 1 and 2, respectively. First appearance of this morphology was after 18 and 20 days, respectively, coinciding with the biomass accumulation reaching a plateau. Scale bar indicates 5 μm. This figure is adapted from Chapter 4.
enzymes that are involved in glycolysis and biosynthesis of nucleotides, amino acids, 131
fatty acids and cell wall components. This is in apparent agreement with reduced 132
building block requirement in adaptation of the cells to the reduced substrate 133
availability and decreased growth rate. Some genes involved in antibiotic production 134
were up regulated. 135
Table 1. Overview of gene regulation in retentostat conditions. 136
Time in retentostat (days)
Specific growth rate (h-1)
Percentage of initial growth rate (%)
Percentage of substrate used for maintenance (%)
Number of significantly regulated genesa
Total Up-
regulated Down- regulated
Data from retentostat 1
0 0.02475 100 31 N/A (reference condition)
7 0.0026 11 56 97 31 66
18 0.0006 2.4 85 222 54 168
42 0.00006 0.24 98 236 46 190
Data from retentostat 2
0 0.02475 100 31 N/A (reference condition)
6 0.0028 11 57 65 54 11
20 0.0005 2 89 86 32 54
40 0.00006 0.24 98 339 169 170
asignificance criteria: p < 0.05; Fold-change > 2 or < -2 137
Retentostat cultivation resulted in reduced expression of central glycolytic genes 138
and relief of carbon catabolite repression 139
Although glucose starvation does not occur, a retentostat culture is consistently 140
limited for glucose. The transcriptome analysis gives indications for tuning of energy 141
generating pathways to the reduced substrate access and growth. This is illustrated by 142
repression of the CggR-regulated central glycolytic genes gapA, pgk, pgm, eno and tpiA. 143
Figure 3. Transcriptional adaptation to near-zero growth conditions. Displayed are the 144
relative expression levels of each gene. The corresponding time-points are depicted above the 145
columns. Comparisons are made with steady-state chemostat at t=0. Color indications are 146
yellow for increased expression, blue for decreased expression and black for unchanged 147
expression. 148
Additionally, the transcriptome shows that adaptation to alternative-carbon- 149
substrate utilization is occurring by relief of carbon catabolite repression (CCR). The 150
CcpA-regulated genes ctaDEF and qcrABC, coding for cytochrome c oxidase caa3 and 151
for menaquinol:cytochrome c oxidoreductase (Liu and Taber, 1998; Blencke et al., 152
2003), respectively, are mildly up-regulated. Together with induction of genes which 153
are all repressed by CcpA in the presence of glucose such as fadN (fatty acid 154
degradation) (Blencke et al., 2003; Tojo et al., 2011), gmuB (glucomannan utilization) 155
(Sadaie et al., 2008), and ara genes (arabinose utilization) (Inácio et al., 2003), this 156
suggests that CcpA repression is relieved under retentostat conditions (Fig. 3). 157
Furthermore, there is relative lower abundance of ilvB operon transcripts, of which 158
the production is positively controlled in the presence of glucose by interference of 159
CcpA with CodY regulation (Shivers and Sonenshein, 2005). 160
Mild induction of the stringent response when specific growth rate decreases 161
About one third of the genes known to be under negative control of the stringent 162
response by the pppGpp synthase RelA (Eymann et al., 2002; Bernhardt et al., 2003) 163
are down-regulated under retentostat conditions (Fig. 3). The observed down- 164
regulation increased in strength as the specific growth rate decreased. Many genes 165
coding for components of the translational apparatus are found to be mildly down- 166
regulated. Among these genes are 26 ribosomal proteins, including the very large rpsJ 167
operon, and the initiation factor infA. Some genes, the products of which are involved 168
in other processes typically associated with growing cells, are also found to be down- 169
regulated. A number of these are also known to be under RelA-dependent negative 170
regulation (Eymann et al., 2002; Bernhardt et al., 2003), including genes functioning 171
in RNA synthesis (rpoA), DNA replication (dnaA), nucleotide metabolism (adk, pyrH) 172
and cell wall synthesis (dltB, murA, murD). The secY gene, coding for one of the Sec 173
preprotein translocase subunits and known to be under stringent regulation (Eymann 174
et al., 2002; Bernhardt et al., 2003) is down-regulated, as well as genes functioning in 175
energy metabolism such as the F1F0-ATPase encoding atp-genes. In addition to the 176
genes mentioned above, some growth-associated genes whose regulation is not 177
dependent on RelA (Eymann et al., 2002) are also found to be down-regulated under 178
retentostat conditions (e.g. pur and pyr genes involved with nucleotide metabolism). 179
Although many RelA-dependent genes were down-regulated, none of the genes 180
known to be induced during the stringent response (Eymann et al., 2002) were up- 181
regulated under retentostat conditions. Biosynthesis of branched chain amino acids 182
coded by the ilv operon was even found to be repressed at near-zero growth rates. In 183
our retentostat cultures a repression of CodY-regulated genes is observed. 184
There is no evidence to be found of σB-dependent general stress response 185
(Hecker et al., 2007; Flórez et al., 2009) induction during retentostat cultivation when 186
compared to chemostat conditions. No genes involved in the SOS response, which is 187
activated upon DNA damage and other stresses that affect integrity of the genome 188
(for a review see Lenhart et al., 2012), are induced. There are no indications for 189
stationary phase mutagenesis as no repression of DNA repair systems such as the 190
mismatch repair system (mutSL) and oxidized guanine (GO) system (ytkD, mutM and 191
yfhQ) is observed (Pedraza-Reyes and Yasbin, 2004; Robleto et al., 2007; Vidales et 192
al., 2009). Sporulation genes that are not sigF-dependent (spoIIE, spoIIGA, sigE 193
(spoIIGB), and sigH (spo0H); Fawcett et al., 2000; Steil et al., 2005; Wang et al., 2006) 194
are not induced. 195
Down-regulation of motility- and morphology-associated genes 196
The transcriptomics data show that many genes belonging to the σD regulon are 197
down-regulated in the retentostat sample (Fig. 3). Consistently, down-regulation of 198
the sigD gene and the activator of σD-dependent gene transcription, swrB (Kearns et 199
al., 2004), was observed. Genes in the σD regulon code for proteins involved with 200
flagella synthesis, chemotaxis and autolysis (Márquez et al., 1990). Most prominent 201
among the down-regulated σD-dependent genes are the large flgB operon and the hag 202
gene, both involved in motility (Aizawa et al., 2002). Also found in this group is the 203
operon lytABC, with lytC coding for a cell wall hydrolase (Blackman et al., 1998; 204
Vollmer et al., 2008; Chen et al., 2009). This autolysin is mainly involved in motility 205
and has a minor function in cell separation (Chen et al., 2009). Interestingly there 206
was a difference between the two retentostats in expression of two other cell wall 207
hydrolases, lytE and lytF. These were down-regulated only in retentostat 2 and 1, 208
respectively. Both have an important function in cell separation because of their DL- 209
endopeptidase activity (Yamamoto et al., 2003; Fukushima et al., 2006). For 210
localization to the cell wall, LytE interacts with the cytoskeletal protein MreBH 211
(Carballido-López et al., 2006; Domínguez-Cuevas et al., 2013), of which the 212
expression is down-regulated in retentostat 2 only. Furthermore, the transcriptome 213
revealed the repression in both retentostats of the genes mreD and rodZ which are part 214
of the cell wall biosynthetic complex and involved in morphogenesis (Domínguez- 215
Escobar et al., 2011; Garner et al., 2011; Muchová et al., 2013). 216
Induction of genes for antibiotics and secondary metabolites production 217
The sigY gene and the 6 other genes in the SigY regulon were up-regulated under 218
retentostat conditions. The sigma factor Y is found to be important for maintenance 219
of the SPβ prophage which contains genes necessary to produce and resist killing by 220
the antibiotic sublancin (Mendez et al., 2012). The gene encoding the precursor of 221
sublancin, sunA, is located on the SPβ prophage (Paik et al., 1998) and was up- 222
regulated in our retentostat cultures. 223
Some genes coding for enzymes with industrial application were induced, 224
including the alkaline protease encoding aprE, and the mannose-6-phosphate 225
isomerase encoding gmuF. The latter is involved in glucomannan utilization by 226
catalizing the conversion of among others L-ribulose to L-ribose, which is employed 227
as a primary building block for the synthesis of various pharmaceutical compounds 228
(Yeom et al., 2009). 229
Whole-genome resequencing 230
Sequencing of population samples revealed that in retentostat 1, a total of 60 SNPs 231
were formed between day 18 and 42. Of these SNPs, 10 were in intergenic regions 232
and 50 in coding regions. Of these 50 SNPs, 34 were determined to result in silent 233
mutations. The remaining 16 resulted in missense mutations, affecting 9 genes. None 234
of the genes with SNPs in their upstream elements were differentially expressed in the 235
transcriptome data. Additionally 227 base pairs were deleted in the gene ldh. In 236
retentostat 2, a total of 8 SNPs were formed between day 20 and day 40, of which 237
none were in intergenic regions. Of the total SNPs found, 6 were determined to be 238
silent and 2 to result in missense mutations, affecting 1 gene. The majority of the 239
mutations lie between 2,059,833 and 2,280,665 bp, known as the SPβ prophage 240
region (Lazarevic et al., 1999). For a list of SNPs resulting in missense mutations see 241
Table S2. Similar to that of the population samples, the genomes of the single-colony 242
isolates harboured SNPs almost exclusively in the SPβ prophage region (results not 243
shown). 244
Discussion
245Retentostat cultivation allowed comparison of the B. subtilis transcriptome at near- 246
zero growth rates versus higher growth rates. Very slow growth as studied here is 247
fundamentally different from the widely studied stationary phase. In a retentostat the 248
non-growing cells are glucose-limited instead of glucose-starved and in contrast to the 249
sudden transition in batch cultures the transition from growing to non-growing in a 250
retentostat is more gradual. Near the end of the approximately 40 day retentostat 251
culture experiments, specific growth rates had decreased to 0.00006 h-1. At this stage, 252
a transcriptional reprogramming involving more than 500 genes has taken place. 253
The progressive increase of downregulated genes coinciding with the decreasing 254
specific growth rates suggests that many processes are being shut down in response to 255
the limited availability of glucose (Table 1). The transcript profiles of the retentostat 256
cultures show similarities to those previously reported for batch culture cells 257
experiencing glucose starvation upon transition to stationary phase (de Jong et al., 258
2012; Koburger et al., 2005), but also indicate some fundamental differences. 259
The progressively increasing number of (down)regulated genes during the course 260
of the retentostat cultivation indicates that transcriptional reprogramming took place 261
to adapt the cellular physiology to limited carbon- and energy availability. An 262
example is the down-regulation of relA-dependent genes, which suggests that the 263
almost non-growing B. subtilis culture is subject to reduction of the translational 264
apparatus by the stringent response in at least part of the culture (Eymann et al., 265
2002; Bernhardt et al., 2003). Furthermore, the repression of amino acid pathways, 266
most likely by CodY (Molle et al., 2003), is a reflection of reduced building block 267
requirement in non-growing cells. It is suggested that CodY plays a role in (p)ppGpp- 268
mediated gene regulation: In stationary phase cells, genes regulated by the guanosine 269
triphosphate (GTP)-binding protein CodY are de-repressed upon reduction of GTP 270
levels. This can be due to conversion to (p)ppGpp or due to depletion of precursors 271
necessary for guanine nucleotide synthesis (Geiger and Wolz, 2014). Repression of 272
CodY-regulated genes observed in our retentostat cultures might suggest that GTP 273
pools have remained at levels high enough to activate CodY repression. Biosynthetic 274
pathway repression is illustrative for the fact that resources are diverted away from 275
growth, parallel with the decreasing growth rates. This reaches a climax in extremely 276
low growth rates and coincides with the strong redirection of substrate energy 277
towards maintenance. 278
The decreased glucose consumption rate is reflected in the transcriptome as 279
mild repression of central glycolytic genes, suggesting a fine-tuning of glycolytic 280
capacity in response to limited glucose availability. This repression is most likely due 281
to low levels of fructose 1,6-bisphosphate (FBP) which consequently relieve the 282
blockage on repression by CggR (Doan and Aymerich, 2003). In B. subtilis, carbon 283
catabolism-related gene expression is regulated by FBP- and glucose 6-phosphate- 284
stimulated global regulator CcpA (Stülke and Hillen, 2000; Sonenshein, 2007). 285
Reduced glycolytic-capacity is known to result in a lower pool of HPrSer46P, which 286
is a required co-regulator together with CcpA to mediate carbon catabolite 287
repression (Stülke and Hillen, 2000). Although the residual glucose concentration in a 288
glucose-limited chemostat is already very low and CCR is expected to be relieved, 289
the transcriptome indicates that CCR is even further relieved under retentostat 290
conditions. As described by Monod kinetics the glucose concentration drops further 291
when the growth rate decreases (Monod, 1949; Senn et al., 1994), resulting in further 292
relief of CCR under retentostat conditions. The consequential induction of genes 293
involved in the utilization of alternative carbohydrates indicates that the adaptation 294
of these cells to the severely limiting amounts of glucose leads to a prominent 295
expansion of their active metabolic repertoire. At very low carbohydrate 296
concentrations, the ability to simultaneously utilize various carbon sources, i.e., 297
mixed substrate growth, most likely gives cells advantages over single substrate 298
growth (Egli, 2010). Egli (2010) proposes ‘improved metabolic/physiological 299
flexibility’ and ‘improved kinetic performance’, as cells growing in chemostats on 300
mixed-substrate were able to utilize the carbon sources at concentrations lower than 301
observed in single-substrate chemostats (Lendenmann et al., 1996). Interestingly, the 302
induction of genes involved in fatty acid degradation is described as essential for 303
survival of non-growing B. subtilis cells by Koburger et al. (2005). Thus, the induction 304
of fatty acid degradation genes in retentostat-cultivated cells potentially serves the 305
goal of generating energy from an alternative source, e.g. phospholipids. There is no 306
indication for lysis, but possibly membrane turnover could provide very low 307
concentrations of fatty acids. The observed induction of fatty acid degradation genes 308
is relatively strong and could indicate a more specific activation mechanism by fatty 309
acids (Matsuoka et al., 2007), rather than solely CCR relief. Alternatively, this 310
response could be an adaptation to expand the active metabolic repertoire, 311
irrespective of the availability of the corresponding substrate, to be prepared for 312
utilization once these metabolites occur in the environment. 313
The effects of reduced glucose availability are similar to those previously 314
reported for batch culture cells experiencing glucose starvation upon transition to the 315
stationary phase of growth (Koburger et al., 2005; de Jong et al., 2012). However, 316
retentostat cultivation does prevent starvation by a continuous supply of glucose, 317
which possibly explains the absence of reactions characteristic for stationary phase 318
starvation such as the activation of the σB-dependent general stress response (Völker 319
et al., 1995; Petersohn et al., 2001; Zhang and Haldenwang, 2005) and repression of 320
DNA repair mechanisms characteristic for stationary phase mutagenesis (Robleto et 321
al., 2007; Vidales et al., 2009). On the other hand, the σB general stress response, is 322
only transiently activated (Völker et al., 1995; Holtmann et al., 2004) and this time- 323
window is possibly missed with the sample points taken. 324
B. subtilis initiates the formation of endospores for survival under challenging 325
conditions. The asporogenous sigF mutant used in this study is only able to express 326
genes for sporulation initiation and thereby still allows us to see if sporulation is one 327
of the responses B. subtilis applies. The fact that these genes are not differentially 328
expressed under retentostat conditions in comparison with the chemostat reference 329
condition, suggests that sporulation is not initiated under retentostat conditions. 330
However, if sporulation is initiated under both retentostat and chemostat conditions, 331
no differential expression is observed as well. Sporulation has been previously 332
reported in carbon-limited chemostats (Dawes and Mandelstam, 1970) and is 333
regarded as a risk-spreading strategy (Fujita and Losick, 2005; Veening et al., 2008; 334
de Jong et al., 2010), therefore it is very likely that sporulation is initiated under 335
retentostat conditions. 336
The observed morphological heterogeneity and appearance of cell chains most 337
likely is related to the regulation of the σD regulon. Expression of sigD is subject to 338
stochasticity (Cozy and Kearns, 2010) and exponentially growing B. subtilis cultures 339
are found to be heterogeneous in cell morphology, as they are epigenetically 340
differentiated into two subpopulations with cells either ON or OFF for σD- 341
dependent gene expression (Kearns and Losick, 2005; Chai et al., 2010). The former 342
subpopulation grows as single motile cells, while the latter grows in non-motile 343
chains. The decrease in the number of transcripts from σD-dependent genes in 344
retentostat versus chemostat suggests that during retentostat cultivation the 345
proportion of cells that are ON for σD-dependent gene expression has decreased. 346
The repression of genes coding for autolysins (lytC, lytE and lytF) and other members 347
of the cell wall biosynthetic complex (mreBH, mreD, rodZ) are possibly the direct cause 348
of the changed morphology under retentostat conditions (Ishikawa et al., 1998; 349
Carballido-López et al., 2006; Chen et al., 2009). The observed curved long-chained 350
morphology is very similar to that of lytE (Ishikawa et al., 1998) and lytF mutants 351
(Ohnishi et al., 1999; Chen et al., 2009). Mutation of these cell wall hydrolase- 352
encoding genes prevents the appropriate rate of digestion of peptidoglycan strands, 353
which is a prerequisite for normal cell separation, and thereby results in long cell- 354
chains. It has been suggested that lytE and lytF have overlapping functions here 355
(Ishikawa et al., 1998; Ohnishi et al., 1999; Carballido-López et al., 2006). Down- 356
regulation of lytE and lytF in retentostat 2 and 1, respectively, is therefore likely to 357
have contributed to the chained-cell morphology, whereas the higher number of cell- 358
chains in retentostat 2 may be related to the observed co-repression of mreBH in this 359
culture, next to the lytE repression. The cytoskeletal protein MreBH is responsible for 360
localization of LytE to the lateral cell wall, and mreBH mutants have a curved long- 361
chained morphology very similar to that observed in this study (Carballido-López et 362
al., 2006). Based on these findings we propose that the observed morphology is 363
caused by the repression of the specific functions in cell wall metabolism, and that the 364
magnitude of the morphological effect may relate to the degree of repression of 365
particular subsets of these functions. The few slightly shorter cells observed mainly in 366
the beginning of retentostat culturing seem similar to phenomena observed in E. coli 367
such as dwarving or reductive cell division (Nyström, 2004b). Both processes occur in 368
stationary phase and result in size reduction. Whether the shorter cells under 369
retentostat conditions are indeed a result of these phenomena remains to be 370
elucidated. 371
Growth-related genome mutation rates are proportional to the number of DNA 372
replications within a growing culture (Drake, 1991; Drake et al., 1998; Barrick et al., 373
2009). Because only approximately 4 generations are formed during the course of 40 374
days retentostat cultivation as estimated from specific growth rates, these mutations 375
most likely are not very numerous. However, lysis of cells and regrowth of others at 376
the same rate (cryptic growth; Ryan, 1959; Finkel, 2006) could lead to higher growth 377
rates than estimated. As growth is a requirement for propagation of genetic variants 378
throughout a population (Berg et al., 2002; Finkel, 2006), the number of mutations in 379
the genome could reveal whether substantial growth has taken place or not. A low 380
number of SNPs, almost entirely confined to the SPβ prophage region, confirm that 381
limited cryptic growth has taken place in the retentostat. The SPβ prophage is known 382
as a region where deletions occur, sometimes at high frequency (Spancake and 383
Hemphill, 1985). This suggests it is dispensible, as shown by Westers et al. (2003). 384
Our study reveals that retentostat cultivation has many characteristics in 385
common with stationary-phase cultures, but also several fundamental differences are 386
apparent. High cell viability and no significant induction of systems involved with 387
DNA and protein repair, indicate that deterioration of cellular functions as observed 388
in stationary phase is absent in retentostat cultures. Moreover, lack of mismatch 389
repair repression, characteristic for stationary phase mutagenesis, together with a low 390
number of SNPs, underpins the difference between retentostat cells and stationary 391
phase cells and confirms that retentostat cultivation provides a method to achieve 392
extremely slow growth rates without the loss of cell integrity and function. As 393
transcriptome analysis has provided us an image of the whole population, a 394
transcriptional fusion of a GFP with a heterogeneous promoter could reveal much 395
about population dynamics of B. subtilis in a retentostat. With the rise of single-cell 396
mRNA profiling this could be a very interesting combination to map the response of 397
B. subtilis to near-zero growth conditions in more detail. 398
Material and methods
399Strain, growth conditions and media 400
B. subtilis 168 trpC2 sigF::spec amyE::PrrnB-gfp+ was used for the retentostat 401
experiments in this study. This strain carries a green fluorescent protein (GFP) fusion 402
to the promoter of the constitutively expressed ribosomal ribonucleic acid (RNA) 403
operon rrnB (Krásný and Gourse, 2004; Veening et al., 2009), and is defective in 404
sporulation, caused by a disruption in the sigF gene. Precultures for chemostat and 405
retentostat cultivations were prepared by inoculating a single colony from an 406
lysogeny broth (LB) agar plate into 10 ml LB medium (Sambrook et al., 1989). This 407
culture was grown at 37˚C until an optical density at 600 nm (OD600) of 0.3 was 408
reached. Subsequently 1000x dilutions were made in 60 ml M9 medium (Miller, 409
1972) supplemented with 27.75mM glucose and 0.1mM Tryptophan. The M9 410
minimal medium contained, per liter of deionized water, 8.5 g of Na2HPO4 · 2H2O, 411
3.0 g of KH2PO4, 1 g of NH4Cl, and 0.5 g of NaCl. The following components were 412
sterilized separately and added per liter: 1 ml of 0.1 M CaCl2, 1 ml of 1 M MgSO4, 1 413
ml of 50 mM FeCl3, and 10 ml of M9 trace salts solution. The M9 trace salts solution 414
contained (per liter) 0.1 g MnCl2 · 4H2O, 0.17 g of ZnCl2, 0.043 CuCl2 · 2H2O, 0.06 415
CoCl2 · 6H2O, 0.06 Na2MoO4 · 2H2O. The cultures were grown overnight and used 416
for inoculation of the bioreactors. Chemostat- and retentostat media were acidified to 417
pH 5 by addition of H2SO4 (95 to 97%) to avoid precipitation of medium 418
components. During the cultivation in the bioreactors the pH was maintained at 7.0 419
by automatic addition of NaOH 5M. 420
Chemostat cultivation 421
Duplicate chemostat cultures were performed at a dilution rate, D (defined as the 422
ratio of the medium feed rate (L h-1) and culture volume (L)) of 0.025 h-1. 2.0 L 423
bioreactors (Infors Benelux BV, the Netherlands) with 1.4 L working volume were 424
inoculated with an exponentially growing preculture to start the chemostats. The 425
bioreactors were operated at 37˚C under aerobic conditions. An airflow of 0.1 l.min-1 426
and a stirring speed of 800 r.p.m. was set to keep oxygen levels above 50% of air- 427
saturation. 428
The working volume was kept constant by means of a conductivity sensor placed 429
at the surface of the culture, activating a peristaltic pump that removed effluent. To 430
prevent foam formation, 5 ml of a 5% (wt.wt-1) solution of the antifoaming agent 431
Struktol J673 (Schill and Seilacher AG, Hamburg, Germany) was added per 24 432
hours, automatically spread over intervals of 13 minutes. Steady state was defined as 433
the condition in which culture parameters were constant for at least 5 volume 434
changes and when optical density at 600 nm (OD600) and cell dry weight (CDW) had 435
remained constant (<5% and <10% variation, respectively) for at least two volume 436
changes. Culture purity was routinely checked by phase-contrast- and fluorescence 437
microscopy. The PrrnB-GFP fusion allowed for identification of fluorescent cells as 438
being the inoculated B. subtilis. Additionally, cells were plated on LB agar plates to 439
check for possible contaminations. 440
Retentostat cultivation 441
A 2.0 L bioreactor (Infors Benelux BV, the Netherlands) was equipped with an 442
autoclavable polyethersulfone cross-flow filter with a pore size of 0.22 μm (Spectrum 443
Laboratories, CA, USA) to retain biomass in the reactor. The filter was connected to 444
the bioreactor via an external loop, through which culture was circulated. 445
Two individual retentostat experiments were initiated from chemostat cultures 446
at dilution rates of 0.025 h-1. After reaching steady state in the chemostat, the 447
bioreactors were switched to retentstat mode by withdrawing the effluent through the 448
filter instead of through the standard effluent tube. The retentostat cultivations were 449
operated under the same conditions (temperature, pH, medium flow rate, 450
oxygenation, stirring rate, anti-foam addition) as the chemostats. Since withdrawal of 451
biomass from the culture influences the kinetics of biomass accumulation, sampling 452
volumes and -frequency were kept to a minimum. The super safe sampler ports 453
(Infors Benelux BV, the Netherlands) that were used for fast and aseptic sample 454
withdrawal, allowed for accurate control of the sample volume. 455
Determination of biomass, substrate and metabolites 456
During chemostat- and retentostat cultivation, samples were withdrawn from the 457
bioreactor to determine biomass-, glucose- and organic acid concentrations. Cell dry 458
weight was determined by cooled centrifugation of 5 mL of culture in pre-weighted 459
tubes, washing with 0.9% NaCl and drying at 105˚C for 24 h to constant weight. 460
This was carried out in duplicates. Additionally, optical density of the culture was 461
determined by measuring absorbance at 600 nm. Glucose and organic acid 462
concentrations in culture supernatants were determined by high-performance liquid 463
chromatography (Shimadzu Scientific Instruments, MD, USA) using LC Solutions 464
SP1 software from Shimadzu (Kyoto, Japan). Culture supernatants were obtained by 465
centrifugation (10,000g for 10 min at 4˚C), filter sterilized and stored at -20˚C until 466
HPLC analysis. Samples were separated using an Aminex HPX-87H anion- 467
exchange column (Bio-rad Laboratories Inc., Richmond, CA) with sulphuric acid (5 468
mM; 0.6 ml · min-1) as mobile phase at 55˚C. Detection was done by a refractive 469
index detector and UV wavelength absorbance detector (Shimadzu Scientific 470
Instruments, MD, USA). 471
Microscopy and analysis of cell morphology 472
Cell morphology was analyzed by phase-contrast- and fluorescence microscopy. In 473
order to visualize the cell membrane, cells were incubated for 1 minute with an ice- 474
cold 5 µμg/mL FM5-95 membrane dye solution (Invitrogen, UK) prior to microscopy 475
analysis. Images were taken with Deltavision (Applied Precision) IX71Microscope 476
(Olympus) using a CoolSNAP HQ2 camera (Princeton Instruments) with a 100× 477
phase-contrast objective. Fluorescence filter sets used to visualize GFP (excitation at 478
450/90 nm; emission at 500/50 nm) and red dyes (excitation at 572/35 nm, emission 479
632/60 nm) were from Chroma Technology Corporation (Bellows Falls, USA). 480
Exposure time was between 0.2 and 1 s with 32% transmission xenon light (300 W). 481
Exposure time for phase-contrast images was 0.05 s. Softworx 3.6.0 (Applied 482
Precision) software was used for image capturing. 483
Calculation of retentostat growth kinetics 484
The biomass accumulation during retentostat cultivation can be described by the van 485
Verseveld equation (van Verseveld et al., 1986) (equation 1). 486
𝐶! 𝑡 = 𝐶!,!−! !!,!"!!!!
! ∙ 𝑒!!!∙!!"!"#∙!+!(!!,!"!!!!)
! (1) 487
This equation assumes an ideal situation in which no loss of viability occurs and 488
growth rate independent maintenance-energy requirements. 489
The specific growth rate (μ) of the retentostat cultivations is calculated with 490
equation 2: 491
𝜇 =!!!,!"!#$/!"
!!,!"#$%& (2) 492
In order to determine the derivative of the biomass accumulation data (dCx,total/dt), the 493
measured total biomass concentrations (viable and non-viable cells) were fitted with 494
the equation Cx = A · eB · t + C, which is of the same shape as equation 1. This was 495
done using GraphPad Prism 6 (GraphPad Software Inc., USA), minimizing the sum 496
of squares of errors by varying A, B and C. With A, B and C known, the derivative 497
(dCx,total/dt) could be determined. Because only viable biomass can replicate, this in 498
incorporated in the equation. 499
DNA microarray experiments and analysis 500
Transcriptome analysis on 4 time-points of independent duplicate retentostat cultures 501
was performed as follows. For RNA isolation, cell culture samples were quickly 502
centrifuged for 2 min at 6,000 × g, and frozen in liquid nitrogen. Cells were broken 503
using 500 mg of glass beads, 500 μl of phenol-chloroform, 30 μl of 3 M sodium 504
acetate, and 15 μl of 20% sodium dodecyl sulfate. RNAs were isolated using the High 505
Pure RNA isolation kit (Roche, Mannheim, Germany) according to the 506
manufacturer's instructions. After a quality check of the isolated RNA using a Agilent 507
Bioanalyzer 2100 with RNA 6000 LabChips (Agilent Technologies, the 508
Netherlands), 20 μg of total RNA was used for cDNA synthesis and incorporation of 509
aminoallyl-dUTP using SuperscriptIII reverse transcriptase (Invitrogen, Life 510
Technologies Europe BV, the Netherlands). Subsequently, the cDNA was labeled 511
with Dylight 550 or Dylight 650 Dyes (Thermo Scientific Pierce, Rockford, USA) as 512
described before (van Hijum et al., 2005; Lulko et al., 2007). Hybridization was 513
performed on Bacillus subtilis 168 Agilent 8x15k DNA microarrays (GEO platform 514
GPL18393) at 65°C as described in the Agilent Two-Color microarray manual 515
(v1.3). These slides contained 2-3 probes of each gene. For hybridization the 516
following cDNA comparisons were made: (a) chemostats with the retentostat time- 517
points 1 (indicated in Fig. 1), (b) time-points 1 with time-points 2, (c) time-points 2 518
with time-points 3, (d) time-points 3 with chemostats, (e) chemostats with time-points 519
2, (f) time-points 1 with time-points 3. This resulted in a total of 24 slides used for this 520
study. Slides were scanned using a confocal laser scanner (GenePix Autoloader 521
AL4200, Molecular Devices Ltd., Sunnyvale, USA). Fluorescent signal intensity data 522
were quantified using GenePix 6.1 (Molecular Devices Ltd., Sunnyvale, USA). The 523
data sets were Lowess normalized and a statistical analysis was performed using the 524
LimmaR software package (Smyth, 2004). Following the Limma R pipeline, the 525
multiple values of genes with a multi probe design are merged to one value by taking 526
the average ln(ratio) and the e(average ln(p-values)). Genes showing a fold change higher 527
then 1.5 and a Benjamini Hochberg corrected p-value (Benjamini and Hochberg, 528
1995) of <0.05 were considered to be significantly altered in expression. Functional 529
analysis on http://server.molgenrug.nl was used to calculate which functional classes 530
were overrepresented in the DNA microarray data for each of the time points. 531
Various annotation sources were used in this enrichment analysis: metabolic 532
pathways from Kyoto Encyclopedia of Genes and Genomes (KEGG; Kanehisa et al., 533
2004), categories from Gene Ontology (GO; Ashburner et al., 2000) and Cluster of 534
Orthologous Groups (COG; Tatusov et al., 1997) and regulons from Database of 535
Transcriptional regulation in Bacillus subtilis (DBTBS; Sierro et al., 2008). 536
The microarray data have been deposited in the Gene Expression Omnibus 537
database (GEO; http://ncbi.nlm.nih.gov/geo/) under the accession number 538
GSE55690. 539
Whole-genome sequencing 540
In order to resequence the genome of retentostat grown cells, genomic DNA was 541
isolated by phenol/chloroform extraction using Phase Lock Gel Heavy 2 mL tubes (5 542
PRIME, Hilden, Germany). The full genomes of the following population samples 543
were sequenced: 1) the strain used for the initial inoculum; 2) retentostats at time- 544
point 2; and 3) the endpoint of both retentostat cultures at time-point 3 (See Fig. 1). 545
In addition, the genomes of two single colony isolates of retentostat culture 1 at time- 546
point 3 were sequenced. Genome sequencing was performed using paired-end 547
sequencing with 100 bp runs on an Illumina HiSeq 2000 using a library of 500 bp 548
fragments. Data from the genome sequencing was analysed with the BRESEQ 549
software pipeline using default settings (Barrick et al., 2009). Single Nucleotide 550
Polymorphisms (SNPs) and insertions/deletions (indel) of retentostat strains were 551
identified by comparison with the strain used for inoculation. 552
Acknowledgements
553We thank Bert van der Bunt, Marjo Starrenburg and Erik de Hulster for valuable 554
help with the bioreactors; Mark Bisschops for valuable help with calculations; Anne 555
de Jong for valuable help with micro-array analysis, and members of the joint zero- 556
growth project group (Kluyver Centre, the Netherlands) for support and valuable 557
discussions. 558
This work was carried out within the research programme of the Kluyver 559
Centre for Genomics of Industrial Fermentation which is part of the Netherlands 560
Genomics Initiative / Netherlands Organization for Scientific Research. 561
References
562Aizawa, S., Zhulin, I.B., Márquez-Magaña, L., and Ordal, G.W. (2002). Chemotaxis 563
and Motility. In Bacillus subtilis and Its Closest Relative: From Genes to Cells, 564
A.L. Sonenshein, J.A. Hoch, and R. Losick, eds. (Washington, DC: ASM 565
Press), pp. 437–452. 566
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H., Cherry, J.M., Davis, 567
A.P., Dolinski, K., Dwight, S.S., Eppig, J.T., et al. (2000). Gene ontology: 568
tool for the unification of biology. The Gene Ontology Consortium. Nat. 569
Genet. 25, 25–29. 570
Barrick, J.E., Yu, D.S., Yoon, S.H., Jeong, H., Oh, T.K., Schneider, D., Lenski, 571
R.E., and Kim, J.F. (2009). Genome evolution and adaptation in a long-term 572
experiment with Escherichia coli. Nature 461, 1243–1247. 573
Benjamini, Y., and Hochberg, Y. (1995). Controlling the False Discovery Rate: A 574
Practical and Powerful Approach to Multiple Testing. J. R. Stat. Soc. Ser. B 575
Methodol. 57, 289–300. 576
Berg, J.M., Tymoczko, J.L., and Stryer, L. (2002). Biochemistry, Fifth Edition (W.H. 577
Freeman). 578
Bernhardt, J., Weibezahn, J., Scharf, C., and Hecker, M. (2003). Bacillus subtilis 579
during feast and famine: visualization of the overall regulation of protein 580
synthesis during glucose starvation by proteome analysis. Genome Res. 13, 581
224–237. 582
Blackman, S.A., Smith, T.J., and Foster, S.J. (1998). The role of autolysins during 583
vegetative growth of Bacillus subtilis 168. Microbiol. Read. Engl. 144 ( Pt 1), 584
73–82. 585
Blencke, H.-M., Homuth, G., Ludwig, H., Mäder, U., Hecker, M., and Stülke, J. 586
(2003). Transcriptional profiling of gene expression in response to glucose in 587
Bacillus subtilis: regulation of the central metabolic pathways. Metab. Eng. 5, 588
133–149. 589
Blom, E.-J., Ridder, A.N.J.A., Lulko, A.T., Roerdink, J.B.T.M., and Kuipers, O.P. 590
(2011). Time-resolved transcriptomics and bioinformatic analyses reveal 591
intrinsic stress responses during batch culture of Bacillus subtilis. PloS One 6, 592
e27160. 593
Boender, L.G.M., de Hulster, E.A.F., van Maris, A.J.A., Daran-Lapujade, P.A.S., 594
and Pronk, J.T. (2009). Quantitative physiology of Saccharomyces cerevisiae at 595
near-zero specific growth rates. Appl. Environ. Microbiol. 75, 5607–5614. 596
Boender, L.G.M., Maris, A.J.A., Hulster, E.A.F., Almering, M.J.H., Klei, I.J., 597
Veenhuis, M., Winde, J.H., Pronk, J.T., and Daran‐Lapujade, P. (2011). 598
Cellular responses of Saccharomyces cerevisiae at near‐zero growth rates: 599
transcriptome analysis of anaerobic retentostat cultures. FEMS Yeast Res. 600
11, 603–620. 601
Branda, S.S., González-Pastor, J.E., Ben-Yehuda, S., Losick, R., and Kolter, R. 602
(2001). Fruiting body formation by Bacillus subtilis. Proc. Natl. Acad. Sci. 98, 603
11621–11626. 604
Brock, T.D. (1971). Microbial growth rates in nature. Bacteriol. Rev. 35, 39–58. 605
Carballido-López, R., Formstone, A., Li, Y., Ehrlich, S.D., Noirot, P., and 606
Errington, J. (2006). Actin Homolog MreBH Governs Cell Morphogenesis 607
by Localization of the Cell Wall Hydrolase LytE. Dev. Cell 11, 399–409. 608
Chai, Y., Norman, T., Kolter, R., and Losick, R. (2010). An Epigenetic Switch 609
Governing Daughter Cell Separation in Bacillus Subtilis. Genes Dev. 24, 754– 610
765. 611
Chen, R., Guttenplan, S.B., Blair, K.M., and Kearns, D.B. (2009). Role of the 612
sigmaD-dependent autolysins in Bacillus subtilis population heterogeneity. J. 613
Bacteriol. 191, 5775–5784. 614
Chesbro, W., Evans, T., and Eifert, R. (1979). Very slow growth of Escherichia coli. J. 615
Bacteriol. 139, 625–638. 616
Cozy, L.M., and Kearns, D.B. (2010). Gene position in a long operon governs 617
motility development in Bacillus subtilis. Mol. Microbiol. 76, 273–285. 618
Daran-Lapujade, P., Daran, J.-M., van Maris, A.J.A., de Winde, J.H., and Pronk, 619
J.T. (2009). Chemostat-based micro-array analysis in baker’s yeast. Adv. 620
Microb. Physiol. 54, 257–311. 621
Dawes, I.W., and Mandelstam, J. (1970). Sporulation of Bacillus subtilis in Continuous 622
Culture. J. Bacteriol. 103, 529. 623
Doan, T., and Aymerich, S. (2003). Regulation of the central glycolytic genes in 624
Bacillus subtilis: binding of the repressor CggR to its single DNA target 625
sequence is modulated by fructose-1,6-bisphosphate. Mol. Microbiol. 47, 626
1709–1721. 627
Domínguez-Cuevas, P., Porcelli, I., Daniel, R.A., and Errington, J. (2013). 628
Differentiated roles for MreB-actin isologues and autolytic enzymes in Bacillus 629
subtilis morphogenesis. Mol. Microbiol. 89, 1084–1098. 630
Domínguez-Escobar, J., Chastanet, A., Crevenna, A.H., Fromion, V., Wedlich- 631
Söldner, R., and Carballido-López, R. (2011). Processive movement of 632
MreB-associated cell wall biosynthetic complexes in bacteria. Science 333, 633
225–228. 634
Drake, J.W. (1991). A constant rate of spontaneous mutation in DNA-based 635
microbes. Proc. Natl. Acad. Sci. U. S. A. 88, 7160–7164. 636
Drake, J.W., Charlesworth, B., Charlesworth, D., and Crow, J.F. (1998). Rates of 637
spontaneous mutation. Genetics 148, 1667–1686. 638
Dubnau, D. (1991). Genetic competence in Bacillus subtilis. Microbiol. Rev. 55, 395– 639
424. 640
Egli, T. (2010). How to live at very low substrate concentration. Water Res. 44, 641
4826–4837. 642
Errington, J. (2003). Regulation of endospore formation in Bacillus subtilis. Nat. Rev. 643
Microbiol. 1, 117–126. 644
Eymann, C., Homuth, G., Scharf, C., and Hecker, M. (2002). Bacillus subtilis 645
functional genomics: global characterization of the stringent response by 646
proteome and transcriptome analysis. J. Bacteriol. 184, 2500–2520. 647
Fawcett, P., Eichenberger, P., Losick, R., and Youngman, P. (2000). The 648
transcriptional profile of early to middle sporulation in Bacillus subtilis. Proc. 649
Natl. Acad. Sci. U. S. A. 97, 8063–8068. 650
Ferenci, T. (2001). Hungry bacteria--definition and properties of a nutritional state. 651
Environ. Microbiol. 3, 605–611. 652
Finkel, S.E. (2006). Long-term survival during stationary phase: evolution and the 653
GASP phenotype. Nat. Rev. Microbiol. 4, 113–120. 654
Flórez, L.A., Roppel, S.F., Schmeisky, A.G., Lammers, C.R., and Stülke, J. (2009). A 655
community-curated consensual annotation that is continuously updated: the 656
Bacillus subtilis centred wiki SubtiWiki. Database J. Biol. Databases Curation 657
2009, bap012. 658
Fujita, M., and Losick, R. (2005). Evidence that entry into sporulation in Bacillus 659
subtilis is governed by a gradual increase in the level and activity of the master 660
regulator Spo0A. Genes Dev. 19, 2236–2244. 661
Fukushima, T., Afkham, A., Kurosawa, S.-I., Tanabe, T., Yamamoto, H., and 662
Sekiguchi, J. (2006). A new D,L-endopeptidase gene product, YojL (renamed 663
CwlS), plays a role in cell separation with LytE and LytF in Bacillus subtilis. J. 664
Bacteriol. 188, 5541–5550. 665
Garner, E.C., Bernard, R., Wang, W., Zhuang, X., Rudner, D.Z., and Mitchison, T. 666
(2011). Coupled, circumferential motions of the cell wall synthesis machinery 667
and MreB filaments in B. subtilis. Science 333, 222–225. 668
Geiger, T., and Wolz, C. (2014). Intersection of the stringent response and the CodY 669
regulon in low GC Gram-positive bacteria. Int. J. Med. Microbiol. IJMM 670
304, 150–155. 671
Goffin, P., van de Bunt, B., Giovane, M., Leveau, J.H.J., Höppener-Ogawa, S., 672
Teusink, B., and Hugenholtz, J. (2010). Understanding the physiology of 673
Lactobacillus plantarum at zero growth. Mol. Syst. Biol. 6, 413. 674
Hecker, M., Pané-Farré, J., and Völker, U. (2007). SigB-dependent general stress 675
response in Bacillus subtilis and related gram-positive bacteria. Annu. Rev. 676
Microbiol. 61, 215–236. 677
Herbert, D. (1961). A theoretical analysis of continuous culture systems. In 678
Continuous Culture of Micro-Organisms, (London: Society of chemical 679
industry), pp. 21–53. 680
Herbert, D., Elsworth, R., and Telling, R.C. (1956). The continuous culture of 681
bacteria; a theoretical and experimental study. J. Gen. Microbiol. 14, 601– 682
622. 683
van Hijum, S.A.F.T., García de la Nava, J., Trelles, O., Kok, J., and Kuipers, O.P. 684
(2003). MicroPreP: a cDNA microarray data pre-processing framework. 685
Appl. Bioinformatics 2, 241–244. 686
van Hijum, S.A.F.T., de Jong, A., Baerends, R.J.S., Karsens, H.A., Kramer, N.E., 687
Larsen, R., den Hengst, C.D., Albers, C.J., Kok, J., and Kuipers, O.P. 688
(2005). A generally applicable validation scheme for the assessment of factors 689
involved in reproducibility and quality of DNA-microarray data. BMC 690
Genomics 6, 77. 691
Holtmann, G., Brigulla, M., Steil, L., Schutz, A., Barnekow, K., Volker, U., and 692
Bremer, E. (2004). RsbV-Independent Induction of the SigB-Dependent 693
General Stress Regulon of Bacillus subtilis during Growth at High 694
Temperature. J. Bacteriol. 186, 6150–6158. 695
Inácio, J.M., Costa, C., and de Sá-Nogueira, I. (2003). Distinct molecular 696
mechanisms involved in carbon catabolite repression of the arabinose 697
regulon in Bacillus subtilis. Microbiol. Read. Engl. 149, 2345–2355. 698
Ishikawa, S., Hara, Y., Ohnishi, R., and Sekiguchi, J. (1998). Regulation of a new 699
cell wall hydrolase gene, cwlF, which affects cell separation in Bacillus subtilis. 700
J. Bacteriol. 180, 2549–2555. 701
de Jong, I.G., Veening, J.-W., and Kuipers, O.P. (2010). Heterochronic 702
Phosphorelay Gene Expression as a Source of Heterogeneity in Bacillus 703
subtilis Spore Formation. J. Bacteriol. 192, 2053–2067. 704
de Jong, I.G., Veening, J.-W., and Kuipers, O.P. (2012). Single cell analysis of gene 705
expression patterns during carbon starvation in Bacillus subtilis reveals large 706
phenotypic variation. Environ. Microbiol. 14, 3110–3121. 707
Kanehisa, M., Goto, S., Kawashima, S., Okuno, Y., and Hattori, M. (2004). The 708
KEGG resource for deciphering the genome. Nucleic Acids Res. 32, D277– 709
D280. 710
Kearns, D.B., and Losick, R. (2005). Cell population heterogeneity during growth of 711
Bacillus subtilis. Genes Dev. 19, 3083–3094. 712
Kearns, D.B., Chu, F., Rudner, R., and Losick, R. (2004). Genes governing 713
swarming in Bacillus subtilis and evidence for a phase variation mechanism 714
controlling surface motility. Mol. Microbiol. 52, 357–369. 715
