AEM Accepts, published online ahead of print on 26 February 2010 Appl. Environ. Microbiol. doi:10.1128/AEM.01992-09 Copyright © 2010, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved.

1

Metabolomic characterization of the salt stress response in Streptomyces

2

coelicolor

3 4

Stefan Kol1+, M. Elena Merlo1,2+, Richard A. Scheltema2+, Marcel de Vries3, Roel J.

5

Vonk3, Niels A. Kikkert2, Lubbert Dijkhuizen1, Rainer Breitling2*, and Eriko Takano1*

6 7

1

8

University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands

9

2

Microbial Physiology, Groningen Biomolecular Sciences and Biotechnology Institute,

Groningen Bioinformatics Centre, Groningen Biomolecular Sciences and Biotechnology

10

Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands

11

3

12

Groningen, The Netherlands

Centre for Medical Biomics, University Medical Centre Groningen, 9713 AV

13 14

*

15

+31-50-3632154 (Streptomyces physiology) or R. Breitling, [email protected], Tel. +31-

16

50-3638088 Fax. +31-50-3637976 (metabolomics)

17

+

Address correspondence to: E. Takano, [email protected], Tel. +31-50-3632143 Fax.

Equally contributing authors

18 19

Journal section: physiology

20

Running title: Metabolomics of S. coelicolor

21

1

22

Abbreviations used: LC-MS, liquid chromatography mass spectrometry; ASA, L-

23

aspartate-β-semialdehyde; DABA, L-2,4-diaminobutyrate; ADABA, N-γ-acetyl-L-2,4-

24

diaminobutyrate, Ect, ectoine; EctOH, hydroxyectoine; HILIC, hydrophobic interaction

25

liquid chromatography.

26 27

Keywords: osmoadaptation, metabolomics, LTQ-Orbitrap, ectoine, salt stress,

28

Streptomyces coelicolor

2

29

Abstract

30

The humicolous actinomycete, Streptomyces coelicolor, routinely adapts to a wide

31

variety of habitats and rapidly changing environments. Upon salt stress the organism is

32

also known to increase the levels of various compatible solutes. Here we report the

33

results of the first high resolution metabolomics time series analysis on various strains of

34

S. coelicolor exposed to salt-stress: wild-type, progressive knockouts of the ectoine

35

biosynthesis pathway, and two stress regulator mutants (disruptions of the sigB and osaB

36

genes). Samples were taken from cultures at 0, 4, 8 and 24 h after salt-stress treatment

37

and analyzed by liquid chromatography mass spectrometry on the LTQ-Orbitrap XL. The

38

results suggest that a large fraction of amino acids is upregulated in response to the salt

39

stress, as well as proline/glycine-containing di- and tripeptides. Additionally we found

40

that 5’-methylthioadenosine, a known inhibitor of polyamine biosynthesis, is

41

downregulated upon salt stress. Strikingly, no major differences between the wild-type

42

cultures and the two stress regulator mutants were found, indicating considerable

43

robustness of the metabolomic response to salt stress, compared to the more volatile

44

changes in transcript abundance reported earlier.

45

[180 / 250]

3

46

Introduction

47

Salt stress conditions are known to cause major changes in the primary and secondary

48

metabolism of bacterial cells (26, 34). This is also the case for the Gram-positive soil

49

bacteria of the genus Streptomyces, which are well-known for their complex secondary

50

metabolism and production of a wide range of industrially relevant metabolites, including

51

various antibiotics (21). Upon salt stress, these species are known to increase the levels of

52

various compatible solutes, including ectoine, alanine, glutamine and proline (6, 11, 16,

53

28). The genome sequence of Streptomyces indicates a largely unexplored capacity for

54

secondary metabolite production (2), some of which could also be involved in the salt

55

stress response. Here we conducted an untargeted metabolomics analysis of salt-stressed

56

Streptomyces coelicolor, the genome-sequenced model species of the genus (2), to

57

investigate the complexity of the metabolite changes associated with the salt stress

58

response in more detail and to increase our understanding of the mechanisms of

59

osmoadaptation by metabolomic rearrangement. For this purpose we used a high

60

resolution

61

chromatography (23). The accurate mass determination and high resolving power of the

62

Orbitrap (22, 40) have made the analysis of complex metabolite mixtures feasible,

63

making it an attractive option for untargeted metabolomics screens (5). The use of HILIC

64

liquid chromatography ensures that the resolving power of the mass spectrometer is

65

further extended by fractionation of the sample, resulting in a less complex mixture being

66

injected in the mass spectrometer at each moment in time. In addition to wild type S.

67

coelicolor we also studied strains disrupted in various steps of the biosynthesis of ectoine

68

and hydroxyectoine (Figure 1), the two best-characterized osmoprotectants of

LTQ-Orbitrap

mass

spectrometer

(MS)

coupled

to

HILIC

liquid

4

69

Streptomyces (6, 7), as well as mutants of two stress regulators (OsaB and σB) under

70

continuous salt stress and during a time course following salt shock. The two stress

71

regulators were previously shown to be involved in the salt shock response (3, 32). This

72

study provides the first attempt at a global metabolomic characterization of the salt

73

adaptation process in bacteria.

74

Materials and methods

75

Strains and media conditions

76

Escherichia coli strain ET12567/pUZ8002 was used for conjugation of cosmids into S.

77

coelicolor (14). Salt shock or osmoadaptation experiments using S. coelicolor M145

78

parent strain (27), LW35 (ectA::Tn) (this study), LW36 (ectC::Tn) (this study), LW37

79

(ectD::Tn) (this study), osaB::Tn (3) and sigB::aac(3)IV (∆sigB) (31) were performed in

80

Difco Nutrient Broth (DNB), whereas solid growth experiments were performed on

81

Supplemented Minimal Media (SMMS) supplemented with the indicated amounts of

82

NaCl or ectoines (20). For the liquid cultures, a high inoculum of spores was used to

83

reduce the formation of clumps.

84

Ectoine mutant construction and characterization

85

Three cosmids containing Tn5062 transposon insertions in ectA (SCO1864), ectC

86

(SCO1866) and ectD (SCO1867) were used to disrupt the ectoine biosynthesis cluster

87

genes in S. coelicolor. Conjugation to S. coelicolor was performed as previously reported

88

(3). Allelic replacements were screened by selecting for AprR and KanS clones. To

89

confirm the genomic replacement of the ectA, ectC, and ectD genes by the Tn5062-

90

containing copies, genomic DNA from each mutant was isolated and used as a template 5

91

for PCR using either the transposon primers EZR1 and EZL2 (3) and ect gene primers

92

which anneal to the start and stop codons of ectA, ectC and ectD. PCR conducted with

93

wild type strain M145 total DNA as a template gave rise to amplified products which

94

corresponded in size with the predicted products of 633 (ectA), 1250 (ectB), 414 (ectC)

95

and 888 (ectD) basepairs using the ect gene primers (Supplementary Figure A.b, lane 1–

96

4). Due to the insertion of the large transposon, no PCR product was amplified when only

97

the ect gene primers were used on genomic DNA from the mutants (Supplementary

98

Figure A.b, lane 7, 10 and 13). However, the transposon primers in combination with the

99

ect gene primers yielded products corresponding in size with the predicted distance from

100

the ect start and stop codons to the location of the transposon (Supplementary Figure A.b,

101

lane 5, 6, 8, 9, 11, 12). These data confirmed the genomic disruption of the ectA, ectC,

102

and ectD genes. This was further validated by a complementation analysis using a drop

103

dilution assay. Cells were grown on SMMS with 1M NaCl, with and without additional

104

20µM ectoine and hydroxyectoine, and monitored for growth (Figure 2).

105

Growth conditions and metabolite extraction

106

Flasks containing 50 ml DNB were inoculated with 1·108 S. coelicolor spores of M145,

107

ectA::Tn, ectC::Tn, ectD::Tn, osaB::Tn and ∆sigB and grown for 24 hours, at which

108

point 5 ml of 5 M NaCl or 5 ml of MilliQ water was added for salt shock samples. Cells

109

exposed to continuous salt stress were inoculated and grown in 50 ml DNB supplemented

110

with 0.5 M NaCl. Samples of 5 ml (OD450 of 1.0), or an equivalent thereof to obtain

111

identical cell mass, were harvested at 0, 4, 8 and 24 hours after addition of salt for the salt

112

shock cultures or at 24 and 48 hours for the continuously salt-stressed cultures. Cells

113

were collected by centrifugation for 10 minutes at 4500 rpm and 4°C, the supernatant was 6

114

removed completely and the cell pellet was resuspended in 500 µl methanol. After

115

incubation on ice for 10 min with occasional mixing, cells were spun down at 4°C, 13000

116

rpm, and the supernatant was collected and stored at –80°C.

117

Liquid Chromatography Mass Spectrometry

118

Data were acquired with a Luna 3µ HILIC 200A HPLC (150 x 2.0 mm; Bester,

119

Rotterdam, The Netherlands), coupled to an LTQ-Orbitrap XL (Thermo Fisher Scientific,

120

Bremen, Germany) in positive ionization mode at a resolution of 30.000. The LC-MS

121

system was run in binary gradient mode with a flow-rate of 400 µl/min; a volume of 20 µl

122

methanol samples was injected. The solvents used were A: 90/10 acetonitrile with 5 mM

123

ammonium acetate and B: 50/50 acetonitrile with 5 mM ammonium acetate. The used

124

gradient was set to hold 90% acetonitrile isocratic for 2.5 minutes, after which it was

125

moved to 50% in 7.5 minutes, and finally held isocratic for 2.5 minutes. The same

126

conditions were maintained for the MS/MS measurements. Analysis of the fragmentation

127

patterns was done in XCalibur 2.0.5. Extracts were supplemented before injection with

128

solvent A in a ratio of 1:10, improving the LC peak shape. Even though molecules were

129

diluted 10-fold, the peak intensity levels of compounds of interest were not affected

130

negatively due to sharper peaks. Ectoine and hydroxyectoine standards (Fluka,

131

Zwijndrecht) were used as controls.

132

Data processing

133

Data processing was performed using a configurable software pipeline, capable of

134

extracting and automatically matching peaks from multiple measurements into peak

135

groups (40). Retention time alignment of the extracted mass chromatograms was done

7

136

using the COW-CoDA algorithm (9). Noise resulting from the greedy peak picking

137

approach was reduced by application of an RSD (relative standard deviation) filter set to

138

35% for matched replicates, as quantification is expected to be at least 20% accurate over

139

multiple runs (41). The selected mass chromatograms were putatively identified by

140

matching the masses progressively to metabolite specific databases. First ScoCyc (2),

141

LipidMAPS (12) and a contaminant database (25) were used. Unidentified peak groups

142

were then matched to KEGG (24) and the remaining unidentified to Metlin (42) and the

143

Human Metabolome Database (47). This iterative process was used in order to restrict the

144

number of potential matches to the most likely (39). As a last step all peak groups related

145

to another peak group (e.g., isotopes, ion adducts, and fragments) were removed from the

146

set based on a correlation analysis of the intensity patterns and peak shape (46).

147

Hierarchical clustering was performed to identify metabolites that showed similar

148

dynamic changes after salt shock in various mutants. To achieve a robust clustering, the

149

distance score was calculated for data with sufficient variation (RSD > 0.2) based on the

150

Pearson correlation of the discretized slopes of the z-score normalized time courses,

151

rather than on the absolute values. By this approach, instead of determining if the original

152

signal intensities correlate, we determined the correlation of the slopes, i.e. the pattern of

153

changes over time, and in order to reduce the influence of noise on the slopes, they were

154

discretized to three possible values (–1, 0, or 1) depending on whether signal intensities

155

were decreasing, stable, or increasing. The resulting hierarchical tree was consequently

156

cut into 4 clusters. For each cluster a quality score was calculated, based on the average

157

distance of each observation of each strain to the average profile of the cluster; this

158

values indicates the coherence of expression profiles within the cluster (smaller values

8

159

indicate better quality). For convenient visualization of the time courses of individual

160

metabolites, the intensity values are displayed as fold-changes relative to the initial time

161

point (i.e., for each strain they are divided by the intensity value of the first time point of

162

that strain).

163

Results and discussion

164

Validation of metabolite quantification

165

The relative quantification accuracy of the mass spectrometry data was assessed by

166

quantifying the S. coelicolor antibiotic undecylprodigiosins, which are visibly present in

167

the extracts (the extracts are dark red when high amounts of antibiotics are produced).

168

The traditional method of measuring the abundance of undecylprodigiosins by optical

169

density (OD) measurement at 533nm (44) was used as a second quantification method.

170

Supplementary Figure B shows the total detected signal for undecylprodigiosin (mass

171

391.26236: butylcycloheptylprodigiosin, metacycloprodiginine and methyl-cyclo-decyl-

172

prodiginine; 393.27801: undecylprodiginine) compared to the OD533nm values. The

173

Pearson correlation is 0.95, indicating that the relative quantification achieved by the

174

HILIC-Orbitrap combination is surprisingly precise. Especially considering that these

175

compounds do not bind well to the HILIC column and are flushed out with a large

176

number of other compounds (mostly lipids), making exact quantification challenging due

177

to ionization suppression effects (23). For the large number of more polar compounds in

178

our metabolome screen, which are better separated by the HILIC column, we can

179

therefore expect similarly reliable quantitation.

9

180

Metabolomics of continuous salt exposure

181

To validate our methodology, we first focused on a targeted metabolomics screen of wild

182

type and ectoine biosynthesis mutants of S. coelicolor grown under continuous salt stress.

183

Under these conditions, we expect very characteristic abundance profiles for ectoine,

184

hydroxyectoine and their precursors; these critical osmoprotectants should be absent in

185

the mutants that show a clear growth defect under salt-stress conditions (Figure 2). This

186

is indeed the case in our measurements, as the found intensity levels closely matched the

187

expected phenotype of the knockouts (Figure 3): (1) The early precursor metabolite

188

DABA was found to be accumulating in the ectA::Tn strain, where its conversion to

189

ADABA is deficient, and was detected nowhere else. (2) The later precursor metabolite

190

ADABA was found to accumulate strongly in the ectC::Tn strain and, to a lesser degree,

191

in the ectD::Tn strain. The presence of this metabolite in the ectD::Tn strain can be

192

explained by the reversibility of its conversion to ectoine. (3) Ectoine was found to

193

accumulate strongly in the ectD::Tn strain, where its conversion to hydroxyectoine is

194

deficient. Low intensity levels of ectoine were also detected for the M145 strain,

195

suggesting that ectoine acts as an intermediate in S. coelicolor and is converted almost

196

quantitatively to hydroxyectoine. Very low levels of the same mass and similar retention

197

time as ectoine were also detected for the ectA::Tn and ectC::Tn strains; this could

198

indicate the existence of an alternative, low-yield biosynthesis pathway for ectoine or the

199

presence of an unrelated isomer of low abundance. In both cases, the detected levels were

200

so low that we do not expect this to be of biological relevance for osmoprotection. (4)

201

Finally, hydroxyectoine was found to accumulate strongly in the M145 strain, indicating

202

that it is acting as the prime osmoprotectant. Retention time and tandem MS

10

203

fragmentation patterns in comparison with the standards confirmed the identity of the

204

detected compounds as ectoine and hydroxyectoine in the strains ectD::Tn and M145

205

respectively (Supplementary Figure C and D).

206

Metabolomics of the salt shock response

207

Global metabolome screen

208

To obtain a global metabolomic characterization of the salt shock response, we measured

209

time series of metabolite profiles in salt shocked cells, using the same S. coelicolor parent

210

(M145) and mutant strains (ectA::Tn, ectC::Tn, ectD::Tn, osaB::Tn and ∆sigB) as before.

211

1247 distinct peak groups were quantified, which were combined into 363 peak groups of

212

related peaks (each corresponding to a potential metabolite). Of the peak groups, 229

213

could be assigned putative identities based on exact mass matching (for an overview of

214

the identifications see Figure 4 and supplementary Table A). This compares well to the

215

predicted amount of metabolites based on genome annotations reported by Borodina et al

216

(4).

217

Dynamic responses to salt exposure

218

The dynamic response to salt exposure was visualized by unsupervised clustering of the

219

metabolite levels for the time series of the salt-stressed cultures. Two coherent clusters of

220

metabolites were revealed with consistently increasing and decreasing abundance (Figure

221

5, cluster 1 and 4 respectively). All other clusters are much less coherent as indicated by

222

their quality scores. This coherence shows that the global metabolomic response to salt

223

stress is clearly dominating in our samples and involves a reasonably large group of 52

224

peak groups (33 increasing; 19 decreasing). Comparison of the time trends of all 11

225

putatively identified metabolites from the salt shock to the control non-salt shocked

226

cultures revealed reproducible results for 15 metabolites accumulating in response to the

227

salt shock, and 3 metabolites with the reverse response (Table 1). The remaining

228

metabolites either did not give reproducible results or showed equal behavior in the salt

229

shock and control cultures. Figure 6 shows exemplary time courses with and without salt

230

shock, including the putative molecular identities of the compounds involved. A large

231

part of the upregulated compounds are also contained in the medium used and further

232

analysis is required to determine whether they are synthesized or taken up by the

233

organism. The most important accumulating metabolite is proline, which is a well-known

234

osmoprotectant that was previously reported for Streptomyces (11). The strong

235

accumulation of proline after the salt shock is the same in all the strains, while in the non-

236

shocked cultures proline levels remain stable or slightly decrease over time. The strong

237

and immediate change indicates that proline is used for acute osmoprotection.

238

Interestingly, this response is independent of the major regulators OsaB and σB. In

239

addition to this classical salt-responsive metabolite, the global metabolome screen also

240

identified additional potential osmoprotectants. Arginine, phenylalanine, methionine,

241

tryprophan and iso-/leucine show the same strong accumulation as proline, and thus

242

could also play a supplementary role in acute osmoprotection. Like all amino acids, the

243

listed compounds are zwitterions that could plausibly act as osmoprotectants. A similar

244

dynamic, seen for guanosin, adenosine, adenine, and hypoxanthine, nucleotide

245

derivatives, is less obvious to explain. While affecting a large number of amino acids, the

246

accumulation in response to the salt shock is not a general phenomenon for all amino

247

acids, as can be seen in the time courses for glutamate and valine.

12

248

In addition to the amino acids, we detected five di- and tripeptides that also

249

accumulated during salt stress (Figure 4). All of these peptides contained a proline and a

250

glycine residue. These proline/glycine containing peptides were significantly enriched

251

compared to the protein composition of both S. coelicolor and Saccharomyces cerevisiae

252

(p-value < 0.01, based on random sampling of peptide sequences). Glycine, glutamine

253

and alanine individually were observed at similar significance levels, while other amino

254

acids were represented at the same frequency in proteins and the accumulating peptides.

255

This indicates that proline/glycine-containing peptides are either produced by specific

256

proteolysis or taken up preferentially from the peptone containing medium. A possible

257

candidate for a di- and tripeptide uptake system is SCO3064, which has homology with

258

DtpT from Lactococcus lactis (18). This is in agreement with earlier studies in other

259

organisms that also showed an involvement of proline-containing peptides in the salt

260

stress response (1, 30, 37).

261

One of the few putatively identified metabolites that showed a slight, but

262

consistent decrease in abundance in salt shock, but not in control conditions was 5’-

263

methylthioadenosine. This metabolite acts as an inhibitor of polyamine biosynthesis in

264

vitro and in vivo (19, 36). As polyamines are known to protect against salt stress in plants

265

(48), an active degradation of the inhibitor 5’-methylthioadenosine could be part of the

266

salt response in Streptomyces as well. The pfs gene encoding 5'-methylthioadenosine/S-

267

adenosylhomocysteine nucleosidase (10) is increased in expression by salt stress in E.

268

coli (29). Alternatively, the decreased levels of 5’-methylthioadenosine could indicate a

269

decreased activity of the polyamine biosynthesis pathway under salt stress.

13

270

Distinguishing these two alternatives will require targeted measurements of polyamines,

271

which were not detected in our experimental setup.

272

Other metabolites that decreased in abundance in high-salt conditions also did so

273

in the control cultures. For instance, the amino acid histidine and its catabolite urocanate

274

decrease in both cultures, the only amino acid-related compounds which show this

275

behavior.

276

Not surprisingly, ectoine and hydroxyectoine are absent from the core salt shock

277

response cluster, due to the disruption of the ectoine pathway in several of the strains.

278

The detected compounds are thus those that show a consistent immediate response to salt

279

shock independent of genotype, indicating that the salt shock metabolome extends far

280

beyond the ectoines. The only putative osmoprotectant that we detected, but which

281

showed no response to salt stress was alanine, which was previously described as an

282

osmoprotectant for Streptomyces (11). The time-series for all strains show stable behavior

283

of this metabolite for the salt shocked samples and the non-salt shocked control cultures,

284

indicating that it is in fact not used as osmoprotectant in the conditions tested.

285

Conclusions

286

Streptomyces species are known to be able to withstand considerable levels of osmotic

287

stress (34) and to accumulate ectoine and hydroxyectoine under salt adaptation (7). We

288

have studied the global metabolic response of S. coelicolor to salt stress, using the high

289

resolution LTQ-Orbitrap mass spectrometer. We did not aim at a complete coverage of

290

the entire metabolome, but rather intended to achieve an unbiased, comprehensive

291

assessment of the moderately polar osmolytes most similar to ectoine, in particular amino

292

acids, nucleotides and their derivatives. Sugars and polyols, the remaining major group of 14

293

osmoprotectants which were not detectable with the liquid chromatography mass

294

spectrometry conditions used, will therefore be the target of future studies using

295

complementary analytical technologies with higher selectivity towards these compounds.

296

It is very likely that among the described compounds, we have detected all that are

297

osmotically relevant, as osmoprotectants need to accumulate in large amounts to be

298

effective. However, the mass spectrometry data do not provide absolute quantification of

299

the compounds involved. Instead, our results are focusing on the dynamic changes of the

300

metabolome. This relative quantification, which is standard practice in metabolomics

301

studies, clearly identifies the temporal trends in a large number of potentially

302

osmoprotective compounds.

303

The current study leads to the putative identification of potential novel

304

osmoprotectants in S. coelicolor. Strikingly, we do not observe a major difference

305

between the metabolomic response to salt stress in wild-type cultures and the two

306

mutants in the major osmotic stress regulatory genes, osaB and sigB (32). This indicates a

307

considerable robustness of the metabolome, compared to the much more volatile

308

transcriptome, in agreement with recent studies in plants (15). Robustness of metabolic

309

fluxes under environmental and genetic perturbations had been found earlier (13, 45), and

310

seem to be a general feature of microbial metabolic networks (43). Our findings extend

311

this observation to the robustness of steady-state metabolite levels. This robustness could

312

be due to redundancy in the regulatory input leading to metabolic adaptation and has

313

obvious evolutionary benefits. Elucidating the cellular networks that enable this

314

metabolic robustness will require additional experimentation at multiple molecular levels,

15

315

including transcriptomics, proteomics and, most importantly, the salt-dependent kinetics

316

of the enzymes involved.

317

Furthermore, the strong accumulation of ectoine in the EctD::Tn strain is an

318

indication that S. coelicolor could potentially be used as a producer of ectoine in (large

319

scale-) bacterial milking (38), but the absolute levels of production would need to be

320

quantified to determine whether this is an attractive option.

321

322

Acknowledgements

323

The authors would like to thank Prof. Dr. Erhard Bremer for his critical reading of the

324

manuscript. The work was supported by grants from: EU-FP6 ActinoGEN (LD, SK), the

325

Northern Netherlands Collaboration Initiative (SNN EZ/KOMPAS RM119; LD), the

326

University of Groningen (Ubbo Emmius Fellowship, MEM; Rosalind Franklin

327

Fellowship, ET), and the Netherlands Organization for Scientific Research NWO (Vidi

328

grant, RB; medium investment grant, RJV).

16

329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

References 1.

2.

3.

4. 5. 6.

7.

8.

9.

10.

11. 12. 13.

Amezaga, M. R., I. Davidson, D. McLaggan, A. Verheul, T. Abee, and I. R. Booth. 1995. The role of peptide metabolism in the growth of Listeria monocytogenes ATCC 23074 at high osmolarity. Microbiology 141 ( Pt 1):41-9. Bentley, S. D., K. F. Chater, A. M. Cerdeno-Tarraga, G. L. Challis, N. R. Thomson, K. D. James, D. E. Harris, M. A. Quail, H. Kieser, D. Harper, A. Bateman, S. Brown, G. Chandra, C. W. Chen, M. Collins, A. Cronin, A. Fraser, A. Goble, J. Hidalgo, T. Hornsby, S. Howarth, C. H. Huang, T. Kieser, L. Larke, L. Murphy, K. Oliver, S. O'Neil, E. Rabbinowitsch, M. A. Rajandream, K. Rutherford, S. Rutter, K. Seeger, D. Saunders, S. Sharp, R. Squares, S. Squares, K. Taylor, T. Warren, A. Wietzorrek, J. Woodward, B. G. Barrell, J. Parkhill, and D. A. Hopwood. 2002. Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature 417:141-7. Bishop, A., S. Fielding, P. Dyson, and P. Herron. 2004. Systematic insertional mutagenesis of a streptomycete genome: a link between osmoadaptation and antibiotic production. Genome Res. 14:893-900. Borodina, I., P. Krabben, and J. Nielsen. 2005. Genome-scale analysis of Streptomyces coelicolor A3(2) metabolism. Genome Res. 15:820-9. Breitling, R., A. R. Pitt, and M. P. Barrett. 2006. Precision mapping of the metabolome. Trends Biotechnol. 24:543-8. Bursy, J., A. U. Kuhlmann, M. Pittelkow, H. Hartmann, M. Jebbar, A. J. Pierik, and E. Bremer. 2008. Synthesis and uptake of the compatible solutes ectoine and 5-hydroxyectoine by Streptomyces coelicolor A3(2) in response to salt and heat stresses. Appl. Environ. Microbiol. 74:7286-96. Bursy, J., A. J. Pierik, N. Pica, and E. Bremer. 2007. Osmotically induced synthesis of the compatible solute hydroxyectoine is mediated by an evolutionarily conserved ectoine hydroxylase. J. Biol. Chem. 282:31147-55. Canovas, D., C. Vargas, M. I. Calderon, A. Ventosa, and J. J. Nieto. 1998. Characterization of the genes for the biosynthesis of the compatible solute ectoine in the moderately halophilic bacterium Halomonas elongata DSM 3043. Syst. Appl. Microbiol. 21:487-97. Christin, C., A. K. Smilde, H. C. Hoefsloot, F. Suits, R. Bischoff, and P. L. Horvatovich. 2008. Optimized time alignment algorithm for LC-MS data: correlation optimized warping using component detection algorithm-selected mass chromatograms. Anal. Chem. 80:7012-21. Cornell, K. A., and M. K. Riscoe. 1998. Cloning and expression of Escherichia coli 5'-methylthioadenosine/S-adenosylhomocysteine nucleosidase: identification of the pfs gene product. Biochim. Biophys. Acta 1396:8-14. Empadinhas, N., and M. S. da Costa. 2008. Osmoadaptation mechanisms in prokaryotes: distribution of compatible solutes. Int. Microbiol. 11:151-61. Fahy, E., M. Sud, D. Cotter, and S. Subramaniam. 2007. LIPID MAPS online tools for lipid research. Nucleic Acids Res. 35:W606-12. Fischer, E., and U. Sauer. 2005. Large-scale in vivo flux analysis shows rigidity and suboptimal performance of Bacillus subtilis metabolism. Nat. Genet. 37:63640. 17

375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420

14.

15.

16. 17.

18.

19.

20. 21. 22. 23.

24. 25.

26.

27. 28.

29.

Flett, F., V. Mersinias, and C. P. Smith. 1997. High efficiency intergeneric conjugal transfer of plasmid DNA from Escherichia coli to methyl DNArestricting streptomycetes. FEMS Microbiol. Lett. 155:223-9. Fu, J., J. J. Keurentjes, H. Bouwmeester, T. America, F. W. Verstappen, J. L. Ward, M. H. Beale, R. C. de Vos, M. Dijkstra, R. A. Scheltema, F. Johannes, M. Koornneef, D. Vreugdenhil, R. Breitling, and R. C. Jansen. 2009. Systemwide molecular evidence for phenotypic buffering in Arabidopsis. Nat Genet 41:166-7. Galinski, E. A. 1995. Osmoadaptation in bacteria. Adv. Microb. Physiol. 37:272328. Garcia-Estepa, R., M. Argandona, M. Reina-Bueno, N. Capote, F. IglesiasGuerra, J. J. Nieto, and C. Vargas. 2006. The ectD gene, which is involved in the synthesis of the compatible solute hydroxyectoine, is essential for thermoprotection of the halophilic bacterium Chromohalobacter salexigens. J. Bacteriol. 188:3774-84. Hagting, A., J. vd Velde, B. Poolman, and W. N. Konings. 1997. Membrane topology of the di- and tripeptide transport protein of Lactococcus lactis. Biochemistry 36:6777-85. Hibasami, H., and A. E. Pegg. 1978. Differential inhibition of mammalian aminopropyltransferase activities. Biochem. Biophys. Res. Commun. 81:1398405. Hopwood, D. A. 1967. Genetic analysis and genome structure in Streptomyces coelicolor. Bacteriol Rev. 31:373-403. Hopwood, D. A. 2007. Streptomyces in Nature and Medicine: The Antibiotic Makers. Oxford University Press. Hu, Q., R. J. Noll, H. Li, A. Makarov, M. Hardman, and R. Graham Cooks. 2005. The Orbitrap: a new mass spectrometer. J. Mass Spectrom. 40:430-43. Kamleh, M. A., J. A. Dow, and D. G. Watson. 2009. Applications of mass spectrometry in metabolomic studies of animal model and invertebrate systems. Brief. Funct. Genomic. Proteomic. 8:28-48. Kanehisa, M., and S. Goto. 2000. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28:27-30. Keller, B. O., J. Sui, A. B. Young, and R. M. Whittal. 2008. Interferences and contaminants encountered in modern mass spectrometry. Anal. Chim. Acta. 627:71-81. Kempf, B., and E. Bremer. 1998. Uptake and synthesis of compatible solutes as microbial stress responses to high-osmolality environments. Arch. Microbiol. 170:319-30. Kieser, T., M. J. Bibb, M. J. Butter, K. F. Chater, and D. A. Hopwood. 2000. Practical Streptomyces genetics. The John Innes Foundation, Norwich. Killham, K., and M. K. Firestone. 1984. Salt Stress Control of Intracellular Solutes in Streptomycetes Indigenous to Saline Soils. Appl. Environ. Microbiol. 47:301-306. Kim, Y., C. M. Lew, and J. D. Gralla. 2006. Escherichia coli pfs transcription: regulation and proposed roles in autoinducer-2 synthesis and purine excretion. J. Bacteriol. 188:7457-63. 18

421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465

30.

31.

32.

33.

34.

35.

36.

37.

38. 39.

40.

41.

42.

43.

Le Marrec, C., E. Bon, and A. Lonvaud-Funel. 2007. Tolerance to high osmolality of the lactic acid bacterium Oenococcus oeni and identification of potential osmoprotectants. Int. J. Food Microbiol. 115:335-42. Lee, E. J., Y. H. Cho, H. S. Kim, B. E. Ahn, and J. H. Roe. 2004. Regulation of sigmaB by an anti- and an anti-anti-sigma factor in Streptomyces coelicolor in response to osmotic stress. J. Bacteriol. 186:8490-8. Lee, E. J., N. Karoonuthaisiri, H. S. Kim, J. H. Park, C. J. Cha, C. M. Kao, and J. H. Roe. 2005. A master regulator sigmaB governs osmotic and oxidative response as well as differentiation via a network of sigma factors in Streptomyces coelicolor. Mol. Microbiol. 57:1252-64. Louis, P., and E. A. Galinski. 1997. Characterization of genes for the biosynthesis of the compatible solute ectoine from Marinococcus halophilus and osmoregulated expression in Escherichia coli. Microbiology 143 ( Pt 4):1141-9. Malin, G., and A. Lapidot. 1996. Induction of synthesis of tetrahydropyrimidine derivatives in Streptomyces strains and their effect on Escherichia coli in response to osmotic and heat stress. J. Bacteriol. 178:385-95. Ono, H., K. Sawada, N. Khunajakr, T. Tao, M. Yamamoto, M. Hiramoto, A. Shinmyo, M. Takano, and Y. Murooka. 1999. Characterization of biosynthetic enzymes for ectoine as a compatible solute in a moderately halophilic eubacterium, Halomonas elongata. J. Bacteriol. 181:91-9. Pegg, A. E., R. T. Borchardt, and J. K. Coward. 1981. Effects of inhibitors of spermidine and spermine synthesis on polyamine concentrations and growth of transformed mouse fibroblasts. Biochem. J. 194:79-89. Piuri, M., C. Sanchez-Rivas, and S. M. Ruzal. 2003. Adaptation to high salt in Lactobacillus: role of peptides and proteolytic enzymes. J. Appl. Microbiol. 95:372-9. Sauer, T., and E. A. Galinski. 1998. Bacterial milking: A novel bioprocess for production of compatible solutes. Biotechnol. Bioeng. 57:306-13. Scalbert, A., L. Brennan, O. Fiehn, T. Hankemeier, B. Kristal, B. van Ommen, E. Pujos-Guillot, E. Verheij, D. Wishart, and S. Wopereis. 2009. Mass-spectrometry-based metabolomics: limitations and recommendations for future progress with particular focus on nutrition research. Metabolomics. Scheltema, R. A., A. Kamleh, D. Wildridge, C. Ebikeme, D. G. Watson, M. P. Barrett, R. C. Jansen, and R. Breitling. 2008. Increasing the mass accuracy of high-resolution LC-MS data using background ions: a case study on the LTQOrbitrap. Proteomics 8:4647-56. Shah, V. P., K. K. Midha, J. W. Findlay, H. M. Hill, J. D. Hulse, I. J. McGilveray, G. McKay, K. J. Miller, R. N. Patnaik, M. L. Powell, A. Tonelli, C. T. Viswanathan, and A. Yacobi. 2000. Bioanalytical method validation--a revisit with a decade of progress. Pharm. Res. 17:1551-7. Smith, C. A., G. O'Maille, E. J. Want, C. Qin, S. A. Trauger, T. R. Brandon, D. E. Custodio, R. Abagyan, and G. Siuzdak. 2005. METLIN: a metabolite mass spectral database. Ther. Drug Monit. 27:747-51. Stelling, J., U. Sauer, Z. Szallasi, F. J. Doyle, 3rd, and J. Doyle. 2004. Robustness of cellular functions. Cell 118:675-85.

19

466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486

44. 45.

46. 47.

48.

Strauch, E., E. Takano, H. A. Baylis, and M. J. Bibb. 1991. The stringent response in Streptomyces coelicolor A3(2). Mol. Microbiol. 5:289-98. Tang, Y. J., H. G. Martin, P. S. Dehal, A. Deutschbauer, X. Llora, A. Meadows, A. Arkin, and J. D. Keasling. 2009. Metabolic flux analysis of Shewanella spp. reveals evolutionary robustness in central carbon metabolism. Biotechnol. Bioeng. 102:1161-9. Tautenhahn, R., C. Böttcher, and S. Neumann. 2007. Annotation of LC/ESIMS Mass Signals, p. 371-380, Bioinformatics Research and Development. Wishart, D. S., D. Tzur, C. Knox, R. Eisner, A. C. Guo, N. Young, D. Cheng, K. Jewell, D. Arndt, S. Sawhney, C. Fung, L. Nikolai, M. Lewis, M. A. Coutouly, I. Forsythe, P. Tang, S. Shrivastava, K. Jeroncic, P. Stothard, G. Amegbey, D. Block, D. D. Hau, J. Wagner, J. Miniaci, M. Clements, M. Gebremedhin, N. Guo, Y. Zhang, G. E. Duggan, G. D. Macinnis, A. M. Weljie, R. Dowlatabadi, F. Bamforth, D. Clive, R. Greiner, L. Li, T. Marrie, B. D. Sykes, H. J. Vogel, and L. Querengesser. 2007. HMDB: the Human Metabolome Database. Nucleic Acids Res. 35:D521-6. Yamaguchi, K., Y. Takahashi, T. Berberich, A. Imai, A. Miyazaki, T. Takahashi, A. Michael, and T. Kusano. 2006. The polyamine spermine protects against high salt stress in Arabidopsis thaliana. FEBS Lett. 580:6783-8.

20

487

Figures and legends

488

Figure 1: Biosynthesis pathway for ectoine. L-aspartate-β-semialdehyde (ASA) is converted into L-2,4-

489

diaminobutyrate (DABA) by the enzyme DABA aminotransferase encoded by the ectB gene. DABA is

490

converted into Nγ-acetyl-L-2,4-diaminobutyrate (ADABA) by DABA acetyltransferase encoded by the

491

ectA gene and L-ectoine (Ect) is formed from ADABA by ectoine synthetase encoded by the ectC gene (8,

492

33, 35). Ectoine can then be hydroxylated to form 5-hydroxyectoine (EctOH) by the enzyme ectoine

493

hydroxylase encoded by the ectD gene (7, 17).

494 495

Figure 2: Ectoine biosynthesis disruption mutants display a salt-sensitive phenotype, which is

496

complemented by the extracellular addition of ectoines. S. coelicolor spore stocks of strain M145,

497

ectA::Tn, ectC::Tn, ectD::Tn, ∆sigB and osaB::Tn were diluted so that a spot of 2.5 µl contained

498

approximately 105, 104, 103, 102, 101 and 100 spores (from left to right). Spots were plated on SMMS

499

medium without any addition (A, left panel), with 1 M NaCl addition (A, right panel) with 1 M NaCl and

500

20 µM ectoine addition (B, left panel) with 1 M NaCl and 20 µM hydroxyectoine addition (B, right panel)

501

or with 1 M NaCl, 20 µM ectoine and 20 µM hydroxyectoine added (C). Plates were incubated at 30°C for

502

3 days in the absence of salt and 6 days in the presence of salt.

503 504

Figure 3: Detected levels of ectoine and related compounds in continuous salt stress reflect the

505

ectoine mutants. The presence of DABA, ADABA, ectoine and hydroxyectoine was determined in wild

506

type cells and in mutants lacking ectA, ectC or ectD grown under continuous salt. DABA is only detected

507

in the etcA::Tn strain, which is expected to be deficient in the conversion of DABA into ADABA and

508

therefore accumulates the precursor. ADABA, another intermediate compound, accumulates in ectC::Tn

509

and in a lesser degree in ectD::Tn as expected. The etcD::Tn strain, which is deficient in the conversion of

510

ectoine to hydroxyectoine, accumulates large amounts of ectoine, while hydroxyectoine is accumulating in

511

the M145 as expected.

512

21

513 514

Figure 4: Overview of the putatively identified metabolites detected in the global metabolome screen.

515

A total of 229 masses could be assigned putative identities based on database matching. By far the largest

516

class of detected metabolites are lipids (107 metabolites), in particular a large number of short- and medium

517

chain fatty acids. The second largest class contains amino acids and their derivatives (66 metabolites),

518

including a large number of di- and tripeptides, many of which contained proline and glycine residues.

519 520

Figure 5: Clustering analysis uncovers the salt stress response. Clustering analysis has been used to

521

group metabolites showing similar temporal behavior into clusters. For each of the 4 clusters the mean

522

behavior is shown with error bars depicting the standard deviation. Cluster 1 contains compounds that are

523

accumulating after the salt shock, including the well known osmoprotectant proline. Cluster 4 contains

524

compounds that are decreasing after the salt shock. Interestingly, these clusters are the most homogenous of

525

all clusters as indicated by the quality score, showing that the salt-responsive metabolites are reacting as a

526

single group.

527 528

Figure 6: Comparison of time-trends of putatively identified salt stress responsive metabolites. The

529

major osmoprotectant proline, as well as several other putatively identified amino acids shows an increase

530

in response to the salt shock. The detected proline/glycine-containing di- and tripeptides also accumulate in

531

response to the salt shock (ArgGlyPro is shown as a representative example), as do a few nucleotide-related

532

compounds, including adenosine. 5’-Methylthioadenosine is one of the few compounds reproducibly

533

showing the opposite pattern, an accumulation in low-salt conditions only. Histidine and its catabolite

534

urocanate decrease in all the cultures, showing no response to the salt shock. For an overview of all of the

535

20 reproducible metabolites see Supplementary Figure E. Gray/dashed line, control non-salt shocked

536

samples; black/solid line, salt shock samples taken at 0, 4, 8, and 24 hours after salt shock from each strain

537

indicated below.

22

538

Table 1: Responses of putatively identified metabolites showing reproducible behavior in response to

539

salt stress. Response

Class

Observed mass

Upregulation in response to salt shock

AA + derivatives

Peptides

Nucleotide derivatives

Downregulation

No response

AA + derivatives Nucleotide derivatives AA + derivatives

115.0633285 131.0946287 174.1116757 165.0789786 149.0510493 204.0898776 172.0847923 328.1859033 300.1797553 301.1273854 243.1219061 136.0385108 135.0544952 267.0967539 283.0916686 117.0789786 147.0531578 297.0895601

Putative identification Name Proline (Iso-)/leucine Arginine Phenylalanine Methionine Tryptophan GlyPro ArgGlyPro LysGlyPro GluGlyPro AlaGlyPro Hypoxanthine Adenine Adenosine/deoxyguanosine Guanosine Valine Glutamate 5’-methylthioadenosine

Formula C5H9NO2 C6H13NO2 C6H14N4O2 C9H11NO2 C5H11NO2S C11H12N2O2 C7H12N2O3 C13H24N6O4 C13H24N4O4 C12H19N3O6 C10H17N3O4 C5H4N4O C5H5N5 C10H13N5O4 C10H13N5O5 C5H11NO2 C5H9NO4 C11H15N5O3S

155.0694765 138.0429274 89.04767847

Histidine Urocanate Alanine

C6H9N3O2 C6H6N2O2 C3H7NO2

540 541

23

EctB

EctD

EctC

EctA

O

O H

H

COO

NH 3 H

L-aspartateB-semialdehyde (ASA)

acetyl-CoA

H

glutamate H

NH3

2-oxoglutarate

COO

NH 3 H

L-2,4-diaminobutyrate (DABA)

H

H

H

N

CoA

H N

H

H COO

NH 3 H

Ng-acetylL-2,4-diaminobutyrate (ADABA)

H2O

CH 3

H

2-oxoglutarate + O2 Fe2+

COO

N H

ectoine (Ect)

succinate+CO2

H H N

CH 3

OH

COO

N H

hydroxyectoine (EctOH)

intensity [arbitrary units]

A:

:Tn

DABA

t ec

t ec

:T C:

n t ec

:T D:

n 5 14 M

intensity [arbitrary units]

A:

:Tn

t ec

ADABA

t ec

:T C:

n t ec

:T D:

n

5 14 M

intensity [arbitrary units]

A:

:T n

Ect

t ec

t ec

:T C:

n t ec

:T D:

n

5 14 M

intensity [arbitrary units]

A:

:T n

t ec

EctOH

t ec

:T C:

n t ec

:T D:

n

5 14 M

Contaminants Other

AA, AA derivatives, di- and tripeptides

18

AA derivatives

Nucleotide derivatives 66

38

12

AA 13

18

Non proteinogenic aminoacids

23 Di- and tripeptides

3 Vitamins + derivatives

33

Diverse

107 Proline Lipids

3

Other

9

11

Glycine + proline 10 Glycine

cluster=1,size=33,quality=0.50

-

-

cluster=3,size=45,quality=0.72 -

-

-

-

o

o

-

o

o

-

-

o

-

-

-

o

-

o

o

-

-

-

o

o

-

-

o

o

-

-

o

o

o o

-

o -

-

-

o -

-

o -

-

-

-

-

-

o

-

-

-

-

-

o -

-

o

-

o -

intensity

intensity

o

-

-

-

o

o

o

-

-

-

-

o

-

o

o

o

o o

o

o

o

o

-

o

o

o o

-

-

o

o

-

o

o o

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

o

o

-

ectA:Tn ectC:Tn ectD:Tn M145 osaB:Tn ΔsigB

-

-

-

-

ectA:Tn ectC:Tn ectD:Tn M145 osaB:Tn ΔsigB

cluster=2,size=46,quality=0.73

cluster=4,size=19,quality=0.58 -

-

-

-

-

-

intensity

-

-

-

-

-

-

o -

o

o o

o

o

o o

o

o

o

o

o

o

o

o o

o

o

-

-

-

-

-

o

o

-

-

-

-

-

-

-

-

-

-

o -

o

-

-

-

-

ectA:Tn ectC:Tn ectD:Tn M145 osaB:Tn ΔsigB

-

-

-

-

o

-

-

-

o

o o o

o

-

-

-

-

o

o o

o -

o

-

-

-

-

o

-

o

-

-

-

-

o

-

-

o

-

o

o

-

-

o

o

o

-

-

o

-

o

-

-

-

-

-

intensity

-

-

o

-

o

-

o -

ectA:Tn ectC:Tn ectD:Tn M145 osaB:Tn ΔsigB

6

NH

12

H

5

O

proline

4 3 2

4

0 M145

∆sig i B

osaB::Tn

EctA t ::Tn EctC::Tn EctD::Tn M145

O

osaB::Tn ∆sig i B

12

2.5

ArgGlyPro

OH

10

NH2

2

phenylalanine

1.5 1

fold change

fold change

arginine

2 EctA t ::Tn EctC::Tn EctD::Tn

8 6 4

0.5

2 EctA t ::Tn EctC::Tn EctD::Tn

M145

0

osaB::Tn ∆sig i B

ctA t ::Tn EctC::Tn EctD::Tn M145

osaB::Tn ∆sig i B

H 2N

NH 2

50

N

N O

OH

30

OH

adenosine

20

fold change

40

10

6 5

N O H3 C

S

HO

OH

5’-methylthioadenosine

4 3 2 1 0

EctA t ::Tn EctC::Tn EctD::Tn M145

N N

N HO

N

7

N

fold change

OH NH 2

6

0

0

O

N H

8

1

0

H2 N

10

fold change

fold change

OH N H

osaB::Tn ∆sig i B

ctA t ::Tn EctC::Tn EctD::Tn M145

osaB::Tn ∆sig i B O

O N N

OH OH

histidine

1 0.8 0.6 0.4

HN

urocanate

1 0.8 0.6 0.4 0.2

0.2 0

1.2

NH 2

fold change

fold change

1.2

HN

EctA t ::Tn EctC::Tn EctD::Tn

control

M145

osaB::Tn ∆sig i B

salt shock

0

EctA t ::Tn EctC::Tn EctD::Tn M145

osaB::Tn ∆sig i B