Reprogramming one-carbon metabolic pathways to decouple L-serine

Subscriber access provided by READING UNIV

Article

Reprogramming one-carbon metabolic pathways to decouple Lserine catabolism from cell growth in Corynebacterium glutamicum Yun Zhang, Xiuling Shang, Shujuan Lai, Yu Zhang, Qitiao Hu, Xin Chai, Bo Wang, Shuwen Liu, and Tingyi Wen ACS Synth. Biol., Just Accepted Manuscript • DOI: 10.1021/acssynbio.7b00373 • Publication Date (Web): 09 Jan 2018 Downloaded from http://pubs.acs.org on January 11, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Synthetic Biology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Synthetic Biology

1

Reprogramming one-carbon metabolic pathways to decouple L-serine catabolism

2

from cell growth in Corynebacterium glutamicum

3 4

Yun Zhang†, Xiuling Shang†, Shujuan Lai†, Yu Zhang†,‡, Qitiao Hu†,‡, Xin Chai†, Bo

5

Wang†,‡, Shuwen Liu†, Tingyi Wen†,§,*

6 7

†CAS

8

Microbiology, Chinese Academy of Sciences, Beijing 100101, China

9

‡University

Key Laboratory of Pathogenic Microbiology and Immunology, Institute of

of Chinese Academy of Sciences, Beijing 100049, China

10

§Savaid

11

China

12

*Corresponding author: Institute of Microbiology, Chinese Academy of Sciences, 1

13

Beichen West Road, Chaoyang District, Beijing 100101, China.

14

E-mail address: [email protected] (T. Wen).

15

Phone: +86 10 64806119. Fax: +86 10 64806157.

Medical School, University of Chinese Academy of Sciences, Beijing 100049,

16 17

Graphical Table of Contents

18 19 20 1

ACS Paragon Plus Environment

ACS Synthetic Biology 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

21 22

Page 2 of 43

ABSTACT L-Serine,

the principal one-carbon source for DNA biosynthesis, is difficult for

23

microorganisms to accumulate due to the coupling of

24

microbial growth. Here, we reprogrammed the one-carbon unit metabolic pathways in

25

Corynebacterium glutamicum to decouple L-serine catabolism from cell growth. In

26

silico model-based simulation showed a negative influence on glyA-encoding serine

27

hydroxymethyltransferase flux with

28

transcription resulted in increased L-serine accumulation, and a decrease in purine

29

pools, poor growth and longer cell shapes. The gcvTHP-encoded glycine cleavage

30

(Gcv) system from Escherichia coli was introduced into C. glutamicum, allowing

31

glycine-derived

32

resulted in an increased amount of one-carbon units. Gcv introduction not only

33

restored cell viability and morphology but also increased

34

Moreover, comparative proteomic analysis indicated that abundance changes of the

35

enzymes involved in one-carbon unit cycles might be responsible for maintaining

36

one-carbon unit homeostasis. Reprogramming of the one-carbon metabolic pathways

37

allowed cells to reach a comparable growth rate to accumulate 13.21 g/L L-serine by

38

fed-batch fermentation in minimal medium. This novel strategy provides new insights

39

into the regulation of cellular properties and essential metabolite accumulation by

40

introducing an extrinsic pathway.

13

L-serine

L-serine

catabolism and

productivity. Attenuation of glyA

CH2 to be assimilated into intracellular purine synthesis, which

41 42 2


L-serine

accumulation.

Page 3 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


43

KEYWORDS:

Corynebacterium

glutamicum,

L-serine,

serine

44

hydroxymethyltransferase, glycine cleavage system, one-carbon units, in

45

silico model-based simulation

46

3



47 48

Page 4 of 43

INTRODUCTION L-Serine

is a building block for the biosynthesis of amino acids, such as glycine,

49

cysteine and tryptophan,1-3 a predominant source of one-carbon (C1) units in de novo

50

biosynthesis of purines and thymidylate and an intermediate in phospholipid

51

biosynthesis.4 Due to its importance in cell growth and division in prokaryotes and

52

eukaryotes,5 L-serine is currently used in the pharmaceutical and cosmetic industries.6

53

Industrial production of L-serine depends mainly on enzymatic or cellular conversion

54

from glycine plus formaldehyde or methanol using serine hydroxymethyltransferase

55

(SHMT).7, 8 In spite of a very high theoretical yield from glucose to L-serine, it is

56

challenging for direct fermentation to produce

57

indispensability of L-serine as an essential intermediate for cell growth.

L-serine

from glucose due to the

58

Many attempts have been made to construct L-serine-producing strains derived

59

from Escherichia coli or Corynebacterium glutamicum. The traditional approach of

60

releasing the feedback inhibition of 3-phosphoglycerate dehydrogenase and

61

co-expressing L-serine synthetic genes (serA, serC, and serB) has only led to marginal

62

accumulations of

63

mechanism for the catabolism of

64

reaction

65

5,10-methylenetetrahydrofolate (5,10-methylene-THF) from 5,6,7,8-tetrahydrofolate

66

(THF). When glyA and the sdaA, sdaB and tdcG genes encoding L-serine deaminase

67

were simultaneously knocked out to block the formation of one-carbon units, an E.

68

coli strain exhibited poor cell growth.11,

that

L-serine

converts

due to the catabolism of

L-serine

L-serine

to

L-serine.

9, 10

One important

is the glyA-encoded SHMT-catalyzed

glycine

12

and

simultaneously

generates

Physiological observation showed

4


Page 5 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


69

abnormal shapes and sizes of E. coli ∆sdaA∆sdaB∆tdcG cells, which was attributed to

70

the inhibition of peptidoglycan synthesis and cell division due to

71

accumulation, which triggered the starvation of one-carbon units.13,

72

illustrated why serine, even at low concentrations, is toxic to E. coli, representing a

73

bottleneck in

74

bottleneck, random mutagenesis or adaptive laboratory evolution (ALE) has been

75

implemented to increase the tolerance of E. coli towards L-serine for improving

76

L-serine

L-serine

14

L-serine

This result

production by microbial fermentation. To overcome this

production.12, 15

77

In C. glutamicum, the glyA-encoded SHMT-catalyzed reaction is the only way to

78

generate 5, 10-methylene-THF.117 The transcription of glyA is activated in the

79

stationary phase by two regulators, RamB and GlyR, to supply one-carbon units for

80

cell metabolism.16 The presence of the glyA gene only led to a marginal accumulation

81

of L-serine even when serACB was overexpressed.10 In spite of the feasibility of

82

deleting the glyA gene in E. coli,12 inactivation of SHMT makes C. glutamicum

83

folate-auxotrophic, resulting in high L-serine productivity combined with reduced

84

growth at low folate concentrations.18 Strong inhibition of glyA mRNA expression

85

triggers spontaneous mutations in the genome and genetic instability in C. glutamicum,

86

which is unfavorable for industrial fermentation.10,18 To control SHMT activity by

87

coenzyme THF availability, the pabAB and pabC genes of THF synthesis were

88

deleted, resulting in dose-dependent L-serine accumulation in response to an external

89

folate supply.18 However, cell growth was significantly impaired with an increase in

90

L-serine

accumulation.18, 19 Therefore, L-serine accumulation is bound to impair cell 5



91 92

Page 6 of 43

growth due to interference with one-carbon unit metabolism. L-serine-derived

5, 10-methylene-THF is an important form of THF derivatives in

93

one-carbon transfer reactions. Conversion between the forms of THF derivatives

94

drives

95

10-Methylene-THF can be oxidized to form 5, 10-methyl-THF via the enzymatic

96

action of FolD and can continue to be hydrolyzed to 10-formyl-THF, or can be

97

converted

98

2’-deoxythymidine-5’-phosphate synthesis.20

99

10-methylene-THF in C. glutamicum; however, glycine is also a donor for 5,

100

10-methylene-THF generation through the glycine cleavage system in other bacteria,

101

such as E. coli and Bacillus subtilis. The glycine cleavage system (Gcv) is composed

102

of four proteins, GcvT, GcvH, GcvP and GcvL, which catalyze the oxidized cleavage

103

of glycine to CO2, ammonia, and a methylene group that is accepted by THF.21 The

104

first reaction of glycine cleavage is the decarboxylation catalyzed by GcvP to convert

105

the carboxyl carbon of glycine to CO2 and the remaining aminomethyl moiety is

106

transferred to the lipoyl prosthetic group of GcvH. Then, the aminomethyl moiety

107

attached to GcvH is subjected to GcvT-catalyzed degradation by transferring the

108

methyl group to THF for the synthesis of 5, 10-methylene-THF, which releases

109

ammonia.22 The reduced lipoic group of GcvH is reoxidated by GcvL, which

110

functions as a component of 2-oxoacid (e.g., pyruvate, 2-oxoglutarate) dehydrogenase

111

multi-enzyme, to form the disulfide.21 However, the gcv operon from E. coli contains

112

no L-protein gene, and no specific GcvL for the Gcv system has been found. In E. coli,

one-carbon

to

unit

cycles

dihydrofolate

in

cellular

to

metabolism

donate L-Serine

6


methyl

(Figure

1a).

groups

5,

for

is the only donor for 5,

Page 7 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


113

the Gcv system provides only a secondary mechanism for 5, 10-methylene-THF

114

formation under tight regulation to maintain a homeostatic balance between glycine

115

and one-carbon units.23, 24 Therefore, L-serine and glycine availability is essential for

116

the metabolism of one-carbon units and cell viability.

117

In our previous study, the correlation between the transient dynamics of L-serine

118

and cell growth demonstrated that L-serine pool management is fundamental for

119

sustaining the viability of C. glutamicum.4 Interestingly, enhancement of extracellular

120

glycine indicated that redundant glycine was exported to maintain intracellular

121

homeostasis. In this study, to overcome the bottleneck between L-serine accumulation

122

and one-carbon unit generation for cell growth, we present a novel engineering

123

strategy for recycling glycine by introducing Gcv systems from E. coli into C.

124

glutamicum to reprogram the one-carbon unit metabolic pathways. This strategy

125

efficiently partially decoupled cell growth with L-serine catabolism and improved

126

L-serine

accumulation.

127 128

RESULTS

129

Metabolic profile prediction of SHMT flux disturbance by iCW773

130

The relationship between SHMT flux and biomass as well as L-serine productivity

131

was investigated by constraint-based metabolic flux balance analysis using a new

132

genome-scale metabolic network model, iCW773.25 Variations in SHMT flux

133

exhibited a prominent influence on L-serine productivity. A high SHMT flux was

134

properly adjusted to supply 5, 10-methylene-THF for maximal biomass synthesis, 7



Page 8 of 43

135

leading to marginal L-serine productivity (Figure 1a). Conversely, maximum L-serine

136

productivity by in silico prediction resulted in a SHMT flux close to zero (Figure 1a),

137

indicating a negative influence of SHMT flux on L-serine productivity. A low SHMT

138

flux leads to moderate biomass synthesis and

139

Furthermore, decreased SHMT flux also led to decreases in pentose phosphate

140

pathway (PPP) and tricarboxylic acid cycle (TCA) fluxes (Figure 1b). 5,

141

10-methylene-THF generated from SHMT plays a critical role in the one-carbon unit

142

cycle, which generates 10-formyl-THF for purine synthesis (Figure 1a). Therefore, a

143

low flux through the SHMT node triggered the inefficient formation of 5,

144

10-methylene-THF, resulting in decreased flux towards the purine synthesis pathway

145

(Figure 1b). The growth rate simultaneously reduced with a decrease in SHMT flux.

146

This result suggested that SHMT had a significant impact on growth rate by

147

controlling flow toward the purine biosynthetic pathway, indicating that reasonable

148

control of SHMT flux is necessary for L-serine productivity and cellular growth.

149

Attenuation of SHMT flux by replacing the promoter of glyA

L-serine

productivity (Figure 1a).

150

Guided by model-predicted metabolic flux distributions, attenuation of SHMT flux

151

was implemented by regulator depletion and promoter replacement. In our previous

152

study, deficiency of GlyR removed the transcriptional activation of the glyA gene in

153

strain SER-3 (∆sdaA∆glyRserAr).4 Based on SER-3, cysE (encoding serine

154

O-acetyltransferase) was deleted to prevent the conversion of

155

O-acetyl-L-serine,1 resulting in strain SER-9 (Table 1). Compared to the wild-type

156

strain, the mRNA level of glyA and specific enzymatic activity of SHMT decreased by 8


L-serine

to

Page 9 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


157

30% and 48% in glyR-deficient SER-9, respectively (Figure 2a). To further attenuate

158

the transcription of glyA, the weak promoter Phom was used as a replacement for the

159

native promoter of glyA.26 The resultant strain (SER-10) showed 50% and 48%

160

decreases in the transcriptional level of the glyA gene and the specific enzymatic

161

activity of SHMT, respectively (Figure 2a). When 24 mg/L cysteine was added for

162

shake flask cultivation,

163

SER-10 (Table S1). In contrast, the specific growth rate and maximum cell dry weight

164

of SER-10 decreased by 17% and 11% compared to those of SER-9, respectively

165

(Table S1), which was consistent with the model-predicted change in growth rate in

166

response to SHMT flux variation (Figure 1).

167

Introduction of GcvTHP to recycle glycine for one-carbon unit generation

168

L-serine

accumulation significantly improved 2-fold in

To investigate whether an increase in one-carbon unit formation could restore

169

growth

rate,

three

170

10-methylene-THF generation from E. coli were added to iCW773 to simulate

171

intracellular metabolic flux distributions. A significantly increased flux toward the

172

purine synthesis pathway was observed with a simultaneous increase in Gcv flux and

173

decrease in SHMT flux (Figure 1b). As expected, the growth rate also improved with

174

an increase in Gcv flux. Due to the deficiency of gcvTHP homologs in C.

175

glutamicum,27,

176

glutamicum chromosome.

28

gcvTHP-catalyzed

reactions

for

glycine-derived

5,

we introduced the gcvTHP system from E. coli into the C.

177

GcvTHP genes, which act as one transcriptional unit in E. coli, were designed in

178

two independent expression cassettes for integration into the SER-10 genome (Figure 9



Page 10 of 43

179

2b). The co-transcribed gcvTH gene region under the control of a moderate P45

180

promoter was inserted into the deleted glyR region to construct the resultant SER-11

181

strain (Figure 2b). In the same manner, a P45-gcvP expression cassette was inserted

182

into the downstream region of the gcvTH genes to construct the resultant SER-12

183

strain (Figure 2b). Protein bands at the corresponding molecular weights of GcvT,

184

GcvH and GcvP proteins were detected by SDS-PAGE. Western blotting analysis

185

demonstrated the expressions of GcvT, GcvH and GcvP proteins in the crude cell

186

extract of SER-12 (Figure 2c). To validate that the Gcv system incorporates glycine-derived one-carbon units into

187

13

188

intracellular metabolites,

C-glycine-labeling tracing was performed by cultivating

189

SER-12 in CGX medium supplemented with [2-13C] glycine or unlabeled glycine. In

190

HPLC-Q-TOF-MS analysis, total ion chromatograms of the eluted components in the

191

intracellular extracts were obtained (Figure 3a and 3b). Purines and their derivatives,

192

adenine, guanine, adenosine monophosphate (AMP), and guanosine monophosphate

193

(GMP), were detected at the corresponding retention times (Figure 3c and 3e). In the

194

13

195

different

196

incorporation of [2-13C] glycine into intracellular purine derivatives (Figure 3d and

197

3f). The mass isotopomer distributions of AMP in the 13C-labeled cell extract showed

198

the unlabeled isotopomer at ~ 50%, whereas the remaining isotopomer represented

199

incorporation of [2-13C] glycine-derived one-carbon units. In contrast, the non-labeled

200

cell extract exhibited the natural distribution of corresponding isotopes from the

C-labeled cell extract, the labeling fractions in the m + 1 mass isotopomers of purine

derivatives

were

dominantly

detected,

10


demonstrating

the

Page 11 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


201

natural presence of 13C (1%). The same distribution patterns of the mass isotopomers

202

of adenine, guanine and GMP were observed for AMP, suggesting that

203

GcvTHP-mediated one-carbon unit generation from glycine plays the same role as

204

GlyA-mediated one-carbon unit generation from L-serine for purine synthesis.

205

Shake flask fermentation showed that the

L-serine

accumulation of SER-12

206

improved from 0.70 ± 0.11 mM to 1.21 ± 0.15 mM. Compared to SER-10, SER-12

207

exhibited 15% increased specific growth rate and 8% increased biomass (Table S1),

208

which validated the simulated effect of Gcv introduction on growth rate by iCW773.

209

Cellular growth and metabolite profiles in response to reprogramming

210

one-carbon metabolic pathways

211

To increase the flux towards L-serine synthesis, pWYE1151, in which the artificial

212

assembled operon consisting of serArBC genes under the control of Ptac was induced

213

to express feedback-resistant SerAr, SerB and SerC, was transformed into SER-9,

214

SER-10 and SER-12 to construct the corresponding plasmid-carrying strains.

215

Fermentation profiles showed the effects of reprogramming one-carbon metabolic

216

pathways on

217

attenuated-transcriptional glyA gene) exhibited a 41% increase in

218

accumulation, accompanied by a 26% decrease in specific growth rate and an 11%

219

reduction in biomass compared to the basic strain SER-13. SER-15 (GcvTHP

220

introduction) showed a specific growth rate and biomass that were nearly comparable

221

to strain SER-13 (Figure 4a). Remarkably, the final L-serine accumulation of strain

222

SER-15 significantly increased by 42% and 100% compared to those of strains

L-serine

accumulation and cell growth (Table S1). SER-14 (the

11


L-serine


223

SER-14 and SER-13, respectively (Figure 4b). Notably, SER-15 showed a 33%

224

increase in the yield of L-serine to glucose (Table S1). The assay of extracellular

225

byproducts showed that glycine concentrations remarkably decreased in strain

226

SER-15 (Figure 4c). The major byproducts were pyruvate-derived L-alanine and

227

L-valine,

which is similar to previously reported results.29

228

Differences were noted in intracellular metabolites related to one-carbon units.

229

SER-14 exhibited a 95% increase in its intracellular L-serine pool and a 32% decrease

230

in its glycine pool in response to the downregulated expression of glyA (Figure 4d). In

231

addition, the intracellular pools of adenine and guanine decreased by 20% and 35%,

232

respectively (Figure 4d). This result demonstrated that the reduction of

233

L-serine-derived

234

consistent with the model-predicted decrease in the purine synthetic pathway flux. By

235

introducing the Gcv system into SER-14, the L-serine and glycine pools decreased by

236

37% and 28%, whereas the adenine and guanine pools improved 1.58- and 1.69-fold

237

(Figure 4d), respectively, demonstrating that glycine supplied one-carbon units for the

238

synthesis of purine nucleotides.

239

Cell morphology restoration by improving the supply of glycine-derived one-

240

carbon units

one-carbon units interfered with the synthesis of purines, which was

241

Because L-serine is an intermediate in the biosynthesis of phospholipids, which

242

are one of the components of membranes, morphological observation was performed

243

to investigate whether one-carbon unit disturbances affect cell shape. Microscopic

244

analysis and cell size determination of C. glutamicum showed that the deletion of 12


Page 12 of 43

Page 13 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


245

sdaA (the only L-serine deaminase) had a negligible effect on cell length and shape

246

compared to the wild-type strain (Figure 5a). In contrast, reprogramming of the

247

one-carbon metabolic pathways affected cell morphology (Figure 5b). When glyA

248

transcription was attenuated, the average cell size of SER-14 was 18% longer than

249

that of SER-13. The average cell size of SER-15 decreased by 11% compared to

250

SER-14 and was comparable to that of SER-13 (Figure 5b), indicating that GcvTHP

251

introduction could restore cell length when its variation was triggered by attenuating

252

glyA transcription.

253

Changes in enzymes involved in one-carbon unit cycle are responsible for the

254

homeostasis of one-carbon units

255

The total soluble proteins of strains SER-13, SER-14 and SER-15 were separated

256

by two-dimensional electrophoresis to investigate proteome responses to one-carbon

257

metabolic disturbances. The proteome profiles exhibited considerable similarity in

258

terms of the quantity and distribution of proteins (Figure S1). The expression

259

abundances of twenty-one proteins involved in different cellular processes changed

260

(Table S2). In the protein profile of SER-14, down-regulation of GlyA expression was

261

consistent with the decrease in glyA mRNA expression and specific activity of GlyA

262

compared with SER-13 (Figure 2a). Remarkable differences in the abundances of

263

proteins involved in one-carbon unit metabolism were observed (Figure 6). The folD

264

gene, encoding bifunctional 5, 10-methylene-tetrahydrofolate dehydrogenase/5,

265

10-methylene-tetrahydrofolate cyclohydrolase increased 1.88-fold. Moreover, AICAR

266

transformylase/IMP cyclohydrolase (PurH), which catalyzes AICAR to form 13



Page 14 of 43

267

intermediates of purine synthesis using 10-formyl-THF as a donor of formyl group,

268

also improved 1.82-fold. The abundance of SAM-dependent methyltransferase

269

(MTase) increased 1.55-fold, indicating that cells might strengthen the methyl donors

270

derived from SAM to regulate one-carbon unit metabolism. The decreased abundance

271

of bifunctional PyrR suggested that pyrimidine synthesis might be repressed to a

272

certain extent to deal with decreased one-carbon pools. Significantly, the abundances

273

of FolD, PurH and MTase in strain SER-15 were much lower than those in SER-14

274

and were comparable to those in SER-13 (Figure 6). In addition, the elongation

275

factors TufP and NusA and single-strand DNA-binding protein Ssp also exhibited

276

reduced abundances in response to glyA down-regulation, which could be related to

277

the retarded cell proliferation of SER-14 (Table S2). The increased abundances of

278

these proteins in SER-15 might contribute to increases in cell viability. Interestingly,

279

one-carbon unit disturbance incurred a response similar to the stress response, which

280

triggered increased abundances of catalase and Hsp70.

281

The effect of reprogramming

282

accumulation in fed-batch fermentation

one-carbon unit pathways

on

L-serine

283

Due to the significant increase in L-serine accumulation by Gcv introduction,

284

production performance of the SER-15 strain was investigated in a 7.5-L bioreactor

285

with minimal medium containing 40 g/L glucose and 24 mg/L cysteine. As shown in

286

Figure 7, the maximal specific growth rate of SER-15 was 0.22 h-1, and the maximal

287

cell dry weight reached 22.68 gCDW/L. The specific glucose consumption rate of

288

SER-15 was 3.48 mmol/gCDW/h. L-serine accumulated at a marginal level in the 14


Page 15 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


289

exponential growth phase due to the requirement of

L-serine-derived

5,

290

10-methylene-THF for growth, but its level rapidly increased in the stationary phase.

291

The glycine concentration was maintained at a relatively low level during the whole

292

fermentation. SER-15 produced up to 13.21 g/L L-serine with a yield of 0.22 mol/mol

293

glucose and a productivity of 0.24 g/L/h (Figure 7). The fermentation performance of

294

SER-15 was much better than that of SER-8 in the previous report,4 demonstrating

295

that reprogramming of the one-carbon metabolic pathways can improve

296

production by partially decoupling L-serine catabolism from cell growth.

L-serine

297 298

DISCUSSION

299

One-carbon units are incorporated into the 2nd and 8th positions of the purine ring

300

for DNA and RNA biosynthesis, and are required for the conversion of homocysteine

301

to methionine, which is the precursor of s-adenosylmethionine for the methylation of

302

DNA, RNA, proteins, and lipids.5 Because L-serine is a principal source of one-carbon

303

units, SHMT-catalyzed L-serine catabolism occupied a key position in central carbon

304

metabolism, and followed a complicated regulatory mode, such as PurR-mediated

305

repression and MetR-mediated activation in E. coli and RamB- and GlyR-mediated

306

activation in C. glutamicum.16, 30 SHMT was an abundant protein and distributed 60%

307

of the metabolic flux over L-serine in C. glutamicum;17 therefore, SHMT flux was a

308

key factor in controlling L-serine accumulation. Deletion of GlyR merely decreased

309

the flux towards

310

resulted in increased L-serine accumulation and a decrease in the intracellular purine

L-serine

to glycine by 24%.4 Attenuation of glyA transcription

15



Page 16 of 43

311

pool, which agreed with the model-simulated downward flux towards the purine

312

synthesis pathway. Decreases in the specific growth rate and biomass were consistent

313

with model-simulation and were observed in a previous report.10 Similarly, the

314

engineered E. coli strain exhibited obvious

315

growth.11,

316

growth.

12

Taken together,

L-serine

L-serine

accumulation and poor cell

accumulation occurs at the expense of cell

317

The export of redundant glycine inspired us to introduce Gcv systems, which

318

enable glycine to be recycled to generate 5, 10-methylene-THF, into C. glutamicum.

319

As expected, one-carbon units derived from 13C-labeled glycine could be assimilated

320

in purine synthesis, resulting in an increase in intracellular purine pools. This result

321

was consistent with the model-predicted increase in the flux towards purine synthesis.

322

Moreover, Gcv introduction had an obvious impact on L-serine accumulation and

323

biomass. There are several explanations for the increases in L-serine accumulation and

324

biomass. One possibility is that the cleavage of glycine mediated by the Gcv system

325

in SER-15 increases the amount of 5, 10-methylene-THF compared to SER-14, which

326

triggers the cellular response to variations in one-carbon unit concentrations. It is well

327

known that L-serine is a physiological substrate for SHMT in the THF-dependent

328

reaction, and the regulation of SHMT activity depends on the concentrations of two

329

substrates (L-seine and THF) and two products (glycine and 5, 10-methylene-THF).31,

330

32

331

amount of 5, 10-methylene-THF to a certain degree. The cleavage of glycine leads to

332

a large pool of 5, 10-methylene-THF, which might feedback inhibit SHMT activity

We can assume that feedback regulation for SHMT activity is dependent on the

16


Page 17 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


333

and give rise to a decrease in the amount of L-serine cleavage. On the other hand, an

334

increasing amount of 5, 10-methylene-THF might decrease the demand for the

335

cleavage of L-serine to generate 5, 10-methylene-THF. We assume that an unknown

336

regulatory effect would play a role to enable cellular export of L-serine to decrease the

337

intracellular pool of

338

decrease the reaction rate of SHMT to maintain metabolic homeostasis of one-carbon

339

units. Therefore, a decrease in L-serine cleavage and an increase in L-serine export

340

result in an increase in the yield of L-serine to glucose. Similarly, the increase in the

341

yield of L-serine to biomass can be explained in the same way. The recycling of

342

glycine elevates the capacity of one-carbon unit synthesis to meet the demand of cell

343

growth, which enables cell accumulation of L-serine instead of cleavage of L-serine

344

and boosts the yield of L-serine to biomass. Consequently, the recycling of glycine

345

partially decoupled cell growth from

346

improving L-serine accumulation and cell growth at the same time.

L-serine.

The decline in the concentration of substrate will

L-serine

catabolism, which contributed to

352

Morphology analysis showed the changes in cell shapes in response to one-carbon

353

unit disturbance. Attenuation of glyA transcription together with the deletion of sdaA

354

led to an increase in the intracellular L-serine pool and longer cell shape. Similarly,

355

the inactivation of three

356

accumulation and abnormally shaped and sized E. coli.14 On the basis of the decrease

357

in the intracellular purine pool, it could be deduced that intracellular

358

accumulation might trigger the deficiency of one-carbon units, which has been

359

demonstrated to be the primary cause of defects in cell shape by inhibiting

L-serine

deaminases resulted in intracellular

17


L-serine

L-serine


Page 18 of 43

360

peptidoglycan synthesis and cell division in E. coli.13,

14

361

system into C. glutamicum restored cell shape and size, which is consistent with

362

previous observations in the E. coli triple mutant.13 This result demonstrated that the

363

cleavage of glycine could restore the supply of one-carbon units to maintain the

364

overall shapes of cells. Alternatively, the unusual sizes and shapes of triple mutant E.

365

coli could be recovered by exogenous supplementation of SAM.13 Therefore, both

366

L-serine-derived

367

generate sufficient one-carbon units to maintain cell morphology.

Introduction of the gcv

and glycine-derived one-carbon pathways make C. glutamicum

368

Interestingly, comparative proteomic analysis indicated self-regulation of the

369

enzymes involved in the one-carbon unit cycle in response to one-carbon metabolic

370

disturbance in C. glutamicum. When the attenuation of glyA expression led to the

371

insufficient supply of 5, 10-methylene-THF, the upregulation of FolD and PurH might

372

enhance the reciprocal interconversion of THF derivatives to maintain the one-carbon

373

unit cycle. When glycine could be used as a donor to generate 5, 10-methylene-THF,

374

the abundances of FolD and PurH were regulated to their original levels. One-carbon

375

unit metabolism was interconnected with purine anabolism through the intermediate

376

10-formyl-THF, which acts as a donor of formyl groups.20 Recently, ZTP/THF

377

riboswitch, an ancient riboswitch class in bacteria, has been found to be directly

378

involved in de novo purine synthesis and the generation of one-carbon units

379

associated with folate coenzymes.33 Our results indicated that C. glutamicum might

380

respond specifically to changing levels of THF derivatives through potential

381

riboswitches to regulate the expression of FolD and PurH to maintain metabolic 18


Page 19 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


382

homeostasis of one-carbon units.

383 384

MATERIALS AND METHODS

385

Bacterial strains, plasmids, and cell growth. All bacterial strains and plasmids

386

used in this study are listed in Table 1. C. glutamicum ATCC 13032∆sdaA∆glyRserAr

387

(SER-3) was used as the parent strain for strain engineering.4 E. coli strain DH5α was

388

used as host cells for plasmid construction. pK18mobsacB and pXMJ19 were used for

389

genetic disruption and expression as described previously.34, 35 E. coli strains were

390

cultured in Luria-Bertani medium at 37 ºC. C. glutamicum strains were cultured in

391

brain heart infusion medium at 30 ºC. If necessary, antibiotics were added at the

392

following concentrations: 50 µg/mL kanamycin or 20 µg/mL chloramphenicol for E.

393

coli and 25 µg/mL kanamycin and 10 µg/mL chloramphenicol for C. glutamicum.

394

Construction of plasmids and strains. General DNA manipulations were

395

performed as described previously. Primers used for gene amplification and plasmid

396

construction are listed in Table S3. The pK18mobsacB derivatives that were used for

397

promoter replacement, gene deletion and chromosomal insertion were constructed in

398

E. coli (Table 1) and transformed into C. glutamicum by electroporation. For serC,

399

serB and serAr overexpression, pWYE1116 (carrying serAr) was digested with

400

PstI/XbaI, and the serC and serB-containing insert was ligated into PstI/XbaI-treated

401

pWYE1116

402

isopropylthio-β-d-galactopyranoside (IPTG) was added to the culture media to induce

403

target gene expression.

404

to

construct

pWYE1151.

When

needed,

1

mM

A markerless homologous recombination system carrying the sacB gene was used 19



Page 20 of 43

405

for promoter replacement, gene disruption and chromosomal insertion of the E. coli

406

gcvTHP genes to construct engineered C. glutamicum strains.35 The upstream and

407

downstream homologous fragments of cysE were amplified using PCR with primers

408

WZ618/WZ619 and WZ620/WZ621 (Table 1). The amplified DNA fragments were

409

spliced

410

pK18mobsacB-∆cysE was transformed into SER-3 by electroporation to generate

411

cysE-deficient SER-9. For promoter replacement of the glyA gene with the weak Phom

412

promoter, a fusing fragment with the upstream region of the glyA gene, the Phom

413

promoter and the glyA gene were obtained by overlap PCR and ligated to

414

pK18mobsacB. The resulting plasmid was transformed into C. glutamicum cells by

415

electroporation.35,

416

conﬁrmation of the chromosomal deletion were performed as described previously.35

417

Promoter replacement was verified by DNA sequencing to obtain the SER-10 strain.

418

For chromosomal insertion of the E. coli gcvTHP genes, the strong P45 promoter was

419

used to control the constitutive expression of the gcvTHP genes, which were

420

amplified using the E. coli genome as the template. To construct SER-11, the plasmid

421

pK18mobsacB-P45-gcvTH was used to insert the gcvTH genes into the glyR deletion

422

gene

423

pK18mobsacB-P45-gcvP was used to insert the gcvP genes after the integration site of

424

the gcvTH genes, resulting in SER-12 (Figure 2).

425 426

using

position

overlap

36

extension

PCR

and

ligated

into

pK18mobsacB.

Screening for the ﬁrst and second recombination events and

of

SER-10

via

two

recombination

events.

Then,

Flux balance analysis. A new genome-scale metabolic network model of C. glutamicum ATCC13032

25

was used to perform all computations based on the 20


Page 21 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


427

COBRA toolbox v2.05 within MATLAB 2010b using the GLPK solver.37, 38 For FBA

428

analysis, simulation constraints followed the uptake rate of glucose with 4 .0

429

mmol/gDCW/h and free transport of external metabolites, such as CO2, H2O, SO3,

430

NH3, and PO4. Simulation of the impact of SHMT flux on biomass and cellular

431

metabolic flux distributions was performed with decreasing relative SHMT fluxes

432

from 1.0 to 0.2 by flux balance analysis.

433

SHMT activity assay. The cells of C. glutamicum were collected by centrifugation

434

and were resuspended with 0.02% CTAB. SHMT activity was measured by

435

β-phenylserine assay.39 The assay mixture contained 50 mM D,L-β-phenylserine, 50

436

µM PLP, 1 mM Na2EDTA, 25 mM sodium sulfate, 50 mM Hepes buffer (pH 7.5) and

437

the crude enzyme in a total volume of 0.5 mL. The rate of the appearance of the

438

product, benzaldehyde, was measured at 25 °C at 279 nm employing a molar

439

absorption coefficient of 1400 1/M/cm. D, L-β-phenylserine was used as the substrate

440

and the production of benzaldehyde was followed by its strong UV absorbance at 279

441

nm.

442

Real-time reverse transcription-PCR. The engineered C. glutamicum strains

443

were grown to the late exponential phase in CGXII minimal medium. Total RNA was

444

isolated by RNAprep Pure Cell/Bacteria Kit (Tiangen, China). Reverse transcription

445

of approximately 200 ng of RNA was performed with the specific primers listed in

446

Table 1 using the FastQuant RT Kit (Tiangen, China). Quantitative PCR was

447

performed with BRYT Green from the GoTaq qPCR master mix (Promega, USA) in a

448

20-µL mixture with a LightCycler® 96 Real-Time PCR System (Roche, Switzerland). 21



449

The RpoB gene of C. glutamicum was used as the reference gene to normalize the

450

glyA mRNA level. Negative controls were used in each PCR run to exclude DNA and

451

other contamination. The qPCR products were verified by melting curve analysis.

452

Data collection and analysis were facilitated using LightCycler® 96 software (Roche,

453

Switzerland), according to the 2-∆∆CT method.40

454

Western blotting. The crude proteins from C. glutamicum SER-10, SER-11 and

455

SER-12 were extracted using a lysis buffer (50 mM Tris-HCl (pH 7.5), 1 mM EDTA,

456

5% glycerol, 1 mM DTT) by ultrasonication. The supernatants were collected by

457

centrifugation and analyzed by SDS-polyacrylamide gel electrophoresis. Specific

458

detection of the GcvT, GcvH and GcvP proteins was performed by Western blotting.

459

Proteins were electrophoretically transferred to a PVDF membrane and probed with

460

three rabbit polyclonal antibodies that were raised against the purified proteins of

461

GcvT, GcvH and GcvP, respectively. The blots were visualized with a

462

peroxidase-coupled goat anti-rabbit secondary antibody and an ECL color

463

development reagent (GE, USA).

464

13

C-labeling glycine tracing. [2-13C] glycine (99% enriched) from Sigma-Aldrich 13

465

Co. LLC was used for

C-labeling tracing. C. glutamicum stains were grown

466

aerobically to the exponential growth phase in minimal media CGX containing 40 g/L

467

glucose and 24 mg/L cysteine with the addition of [2-13C] glycine (100 mg/L) and

468

unlabeled glycine. Extraction was performed by mixing preheated perchloric acid

469

(PCA, 55 °C, final concentration of 500 mM) with the sample, followed by incubation

470

at 55 °C for 10 min.41 The supernatants were collected by centrifugation (10 min, 4°C, 22


Page 22 of 43

Page 23 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


471

8,000 x g) and stored at -20 °C .

472

High-performance liquid chromatography-quadrupole time-of-flight/mass

473

spectrometry (HPLC-Q-TOF-MS) assay. Samples were analyzed on an Agilent

474

1260 series binary HPLC system (Agilent Technologies, Waldbronn, Germany)

475

coupled with an Agilent 6520 Accurate-Mass quadrupole time-of-flight tandem mass

476

spectrometry equipped with an electrospray ionization (ESI) source. Approximately

477

10 µL of the standard or sample solution were injected by the autosampler to the

478

Agilent 1260 HPLC using a Zorbax SB-Aq column (4.6 mm×250 mm, 5 µm; Agilent)

479

with the column temperature kept at 40 ºC. The metabolites were separated by

480

isocratic elution in the mobile phase of 0.1% formic acid in water (eluent A) and

481

methanol (95%/5%, v/v) at a constant flow rate of 500 µL/min. MS data were

482

collected in positive ionization mode with nitrogen supplied as the nebulizing and

483

drying gas. Full-scan spectra were acquired over a scan range of m/z 50-1000 at 1.03

484

spectra/s. Data were acquired and evaluated via MassHunter Workstation software.

485

Optical analysis of cells and cell size determination. C. glutamicum strains were

486

grown at 30 °C in LB medium, and cells were harvested in the exponential phase.

487

After washing the cells two times with PBS buffer, the cell suspension was diluted

488

into 0.2 of OD600 on glass slides. Microscopic investigations and cell size

489

determinations were performed with a Zeiss Axio Imager A1 Microscope equipped

490

with a transmitted light illuminator and a high-resolution Zeiss Axio Cam camera.

491

Imaging was acquired with Zeiss Axio Vision imaging software (Carl Zeiss

492

MicroImaging GmbH, Jena, Germany). Cell sizes were determined with a 23



493

computationally efficient image analysis tool MicrobeJ on the platform ImageJ.42

494

Comparative proteomic analysis. Cells were harvested at the late exponential

495

phase by centrifugation for 10 min at 8,000 × g. The pellets were resuspended in lysis

496

buffer (8 M Urea, 4% CHAPS and 2% Bio-Lyte) and were broken by sonication on

497

ice. The supernatant was collected and protein concentration was determined

498

according to the method described by Bradford (Bradford, 1976). An aliquot of

499

supernatant containing 800 µg of proteins was analyzed by 2-DE. Approximately 450

500

µL of rehydration solution (8M urea, 2% CHAPS, 65 mM DTT, 2% IPG buffer (pH

501

4-7)) containing the sample were loaded onto IPG strips (pH 4-7; 24 cm in length; GE

502

Healthcare, USA) by active rehydration. IEF was performed in Ettan IPGphor III (GE

503

Healthcare, USA). After the equilibration of IPG strips, the second dimension of

504

SDS-PAGE was performed on a 13% polyacrylamide gel using the Ettan DALT six

505

Electrophoresis instrument (GE Healthcare, USA). The gels were stained with

506

Coomassie Blue R-250 (BioRad, Hercules, CA, USA) and scanned using an EPSON

507

Expression 11000XL at a resolution of 300 dpi. The samples from three biological

508

replicates were run in parallel. Image analysis was performed with ImageMasterTM 2D

509

Platinum 7.0 (GE Healthcare, USA). To compare proteome profiles of SER-13,

510

SER-14 and SER-15 strains, gel images in triplicates from each strain were grouped

511

into a class. Ratio was defined as the fold change of protein in quantity between two

512

classes to evaluate protein expression variations. The highest spot value of a protein in

513

one class was divided by the lowest spot value of the same protein in the other class to

514

calculate the ratio of each protein between two classes. The comparison between 24


Page 24 of 43

Page 25 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


515

two-class data was performed by ANOVA with a two-tailed Student’s t-test to

516

evaluate the significantly differential proteins (P

Reprogramming one-carbon metabolic pathways to decouple L-serine

Recommend Documents