Recent Advances in the Evolutionary Engineering of Industrial Biocatalysts James D. Winkler, Katy C. Kao PII: DOI: Reference:

S0888-7543(14)00183-9 doi: 10.1016/j.ygeno.2014.09.006 YGENO 8664

To appear in:

Genomics

Received date: Accepted date:

11 July 2014 16 September 2014

Please cite this article as: James D. Winkler, Katy C. Kao, Recent Advances in the Evolutionary Engineering of Industrial Biocatalysts, Genomics (2014), doi: 10.1016/j.ygeno.2014.09.006

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

PT

Recent Advances in the Evolutionary Engineering of

NU

SC

RI

Industrial Biocatalysts

AC CE P

TE

D

MA

James D. Winkler1 , Katy C. Kao1

1

Department of Chemical Engineering, Texas A&M University, College Station, TX, United States

of America

2

ACCEPTED MANUSCRIPT Running Head: Recent Advances in Evolutionary Engineering

PT

Keywords: evolutionary engineering, mutagenesis, automation, selection, adaptation, complex

RI

phenotypes

SC

Corresponding author: Katy C. Kao

NU

Department of Chemical Engineering, Texas A&M University,

MA

College Station, TX, United States of America

D

Phone 979 845 5571

AC CE P

TE

Email: [email protected]

3

ACCEPTED MANUSCRIPT Abstract

1

Evolutionary engineering has been used to improve key industrial strain traits, such

3

as carbon source utilization, tolerance to adverse environmental conditions, and resis-

4

tance to chemical inhibitors, for many decades due to its technical simplicity and effec-

5

tiveness. The lack of need for prior genetic knowledge underlying the phenotypes of

6

interest makes this a powerful approach for strain development for even species with

7

minimal genotypic information. While the basic experimental procedure for labora-

8

tory adaptive evolution has remained broadly similar for many years, a range of recent

9

advances show promise for improving the experimental workflows for evolutionary

10

engineering by accelerating the pace of evolution, simplifying the analysis of evolved

11

mutants, and providing new ways of linking desirable phenotypes to selectable charac-

12

teristics. This review aims to highlight some of these recent advances and discuss how

13

they may be used to improve industrially relevant microbial phenotypes.

AC CE P

TE

D

MA

NU

SC

RI

PT

2

4

ACCEPTED MANUSCRIPT Introduction

15

Biocatalyst robustness is an important parameter of microbial strains used for industrial fermenta-

16

tion. Cells in these processes are subjected to variety of different stresses, including thermal, acid,

17

and osmotic stresses, in addition to toxic effects from growth substrates and both desired and side

18

products. Improving the tolerance of strains to these physiochemical insults is therefore of great

19

interest for industrial applications. Rational methods for improving these strain characteristics have

20

been largely stymied by the fact that the underlying genetic determinants for most tolerance phe-

21

notypes are largely unknown. One way of circumventing this obstacle is to use inverse engineering

22

approaches, which involve generating strains with the desired phenotype through random methods

23

and then analyzing the genetics, transcriptomics, proteomics, and/or metabolomics of the mutants

24

to identify the causative mutations underlying the phenotype. These approaches do not require

25

a priori genetic knowledge of the organism, and can therefore be used with any organism if the

26

phenotypes-of-interest are growth-coupled.

TE

D

MA

NU

SC

RI

PT

14

Evolutionary engineering, also known as adaptive laboratory evolution or whole-cell directed

28

evolution, is the pre-eminent method for both improving industrial strains and analyzing these

29

complex tolerance phenotypes due to its simplicity and effectiveness. The technique involves the

30

continued propagation of a microbial population under a desired selective pressure. Fitter mutants

31

naturally arise from random mutations during DNA replication or other sources, and will increase

32

in frequency in proportion to their fitness (relative to the mean fitness of the population). Evolu-

33

tionary engineering can also be combined with other methods such as random mutagenesis in order

34

to generate more genetic diversity for selection. Adaptive evolution has been successfully utilized

35

to generate strains with traits such as enhanced substrate utilization, tolerance to fermentation con-

36

ditions, and resistance to toxic compounds [1, 2, 8, 10, 11, 16, 37, 42, 52, 56, 57]. Combined with

37

the increasing availability of affordable high-throughput genomic technologies, molecular charac-

38

terization of beneficial mutants isolated from the evolved populations is beginning to advance our

39

fundamental knowledge on the genetic determinants and molecular mechanisms associated with

40

industrially relevant complex phenotypes. This growing body of knowledge will be of great use

AC CE P

27

5

ACCEPTED MANUSCRIPT for future rational engineering of tolerance and other growth-linked traits. In this review, we focus

42

on recent advances in evolutionary engineering that help to simplify strain improvement and anal-

43

ysis, as well as systems biology approaches for analyzing the changes associated with phenotype

44

improvement.

45

Recent Innovations in Evolutionary Engineering

46

Several novel new thrusts towards changing the standard paradigm of evolutionary engineering

47

have recently been demonstrated in the literature. These advances can be loosely grouped into sev-

48

eral distinct categories: methods for increasing, identifying, and exploiting diversity within evolv-

49

ing populations, new schema for evolving metabolite overproducing strains, parallelization and

50

automation of cultivation systems, and novel tools for understanding the genetic bases of evolved

51

phenotypes. Figure 1 shows how improvements in these areas would affect the execution of adap-

52

tive evolution experiments. Based on improvements in these areas, we speculate about future re-

53

search directions in evolutionary engineering and how these new tools may change our use of this

54

established strain improvement tool.

55

Increasing Genetic Diversity

56

Since intrinsic mutation rates are typically low for most organisms (on the order of 10−9 -10−10

57

bp−1 generation−1 ) under most growth conditions [30, 33], even a relatively small and transient in-

58

crease in mutation rate can significantly improve the probability of generating a chance beneficial

59

mutation. Mutagenesis, using chemicals or radiation, is often used to increase the genetic diversity

60

of a population prior to or during an evolution experiment [32, 46]. The use of mutator strains

61

with defective DNA repair systems can also be used to increase genetic diversity [17]. Although

62

higher mutation rates increase the frequency of beneficial mutations in the population, it also leads

63

to an increase in deleterious mutations (often in the same genetic backgrounds); thus, methods to

64

control mutation rate during adaptive evolution have the benefit of transiently increasing genetic

AC CE P

TE

D

MA

NU

SC

RI

PT

41

6

ACCEPTED MANUSCRIPT diversity while reducing the risk of accumulating deleterious mutations. The GREACE (Genome

66

Replication Engineering Assisted Continuous Evolution) method [34] was recently developed for

67

this purpose and relies on the introduction of a library of defective error-prone polymerases mutD

68

(dnaQ) into a target strain. After evolution in the condition of interest, the plasmid-borne defective

69

mutD can be cured by short-term propagation in the absence of antibiotics for plasmid maintenance,

70

returning the mutation rate of the evolved strain to normal levels for subsequent analyses. Another

71

example of mutation rate manipulation that couples mutation rate to desired product formation was

72

recently introduced by Chou and Keasling, where the strain mutation rate is inversely dependent

73

on the intracellular concentrations of particular metabolites [9]. Both approaches may be useful to

74

transiently raise mutation rates until the desired phenotypic objectives have been reached. Subse-

75

quent analysis of the evolved strains is simplified by returning the mutation rate to the wild-type

76

baseline, ensuring the evolved phenotypes are sufficiently stable for downstream analysis.

77

Interrogating Diverse Populations

78

Although the true objective of industrial evolutionary engineering experiments is to generate supe-

79

rior production strains, knowledge of the underlying genotypes is crucial for downstream rational

80

strain engineering and for improving our fundamental knowledge of biological systems. However,

81

growth under selection generally results in genetically heterogeneous populations that contain a

82

variety of adaptive, neutral, and deleterious adaptive mutations [4], making it difficult to identify

83

all genotypes present with typical, low-throughput, and often arbitrary isolation procedures. Com-

84

petition between these diverse lineages makes it difficult to detect all potentially adaptive mutations

85

from within the population, reducing our ability to understand the genetic changes that occurred

86

over the history of evolution experiments. Our laboratory recently introduced a tool for visual-

87

izing evolutionary dynamics in real time (VERT) [25, 43] that allows for the direct detection of

88

clonal interference in a population. Competing clones, expressing different fluorescent proteins,

89

can therefore be isolated based on the observed population dynamics. Computational tools have

90

also been developed to standardize the isolation procedure as well [58]. VERT has been suc-

AC CE P

TE

D

MA

NU

SC

RI

PT

65

7

ACCEPTED MANUSCRIPT cessfully employed to study several tolerance phenotypes, including solvent, drug, and biomass

92

inhibitor tolerance [1, 21, 43]. Other labeling systems, such as DNA barcodes, can be used to

93

a similar effect with significantly higher resolution; however, the isolation of single mutants-of-

94

interest and mutation linkage determination will be more challenging.

PT

91

Visualization of clonal interference, though effective as an evolutionary tool, cannot provide

96

data on all genotypes present in a population since even labeled subpopulations may be heteroge-

97

neous. With the advent of affordable whole genome sequencing, however, evolutionary engineers

98

now have the power to uncover all mutations present above a certain frequency within an evolving

99

population by direct sequencing of the entire population. Two recent studies used this approach

100

to study population dynamics and types of adaptive mutations that may occur during evolution

101

[29, 31] in ways that would be impossible with more traditional interrogation methods. The prin-

102

cipal limitation of this approach stems from properties in common Next Generation Sequencing

103

(NGS) platforms, namely their short read lengths (100-500 bp) and the need for deep sequencing

104

to identify rare genotypes in the population. Genomic rearrangements and copy number variations

105

are difficult to identify in heterogeneous population samples, and linkage information between

106

widely separated mutations cannot be extracted; for example, cohorts of mutations were observed

107

to expand and contract together in evolving populations of yeast, but not all are beneficial when

108

combined [31]. Even with these restrictions in mind, knowledge of all detectable mutations over the

109

duration of an experiment will provide unprecedented amount of information on the evolutionary

110

dynamics in the population, and with data from parallel experiments to help narrow down neutral or

111

deleterious hitchhikers from beneficial mutations will likely reveal novel mechanisms that confer

112

the phenotypes-of-interest.

113

Exploiting Genetic Diversity for Phenotypic Improvement

114

Beyond creating, visualizing, and quantifying diversity within an evolving population, additional

115

focus has recently been given to help harness this diversity to speed adaptive evolution. One ob-

116

stacle reducing the effectiveness of evolution experiments is clonal interference, which arises from

AC CE P

TE

D

MA

NU

SC

RI

95

8

ACCEPTED MANUSCRIPT the competition of genetically distinct mutants within a population. One approach to reducing this

118

effect is to permit genetic exchange between the lineages (i.e., sex) [35]. This process allows adap-

119

tive mutations to spread throughout an evolving population, and can potentially combine different

120

mutations from competing mutants into a single genetic background. In asexual reproduction,

121

multiple, independent mutations acquired sequentially would be required to combine multiple ben-

122

eficial mutations, significantly increasing the amount of time required to achieve the same level

123

of improvement expected in sexual populations. In addition, any deleterious mutation present

124

in adaptive mutants cannot be readily removed in asexual evolution (known as Muller’s ratchet,

125

[18]), which can be alleviated by sexual recombination. Classic means of recombination include

126

genome shuffling by protoplast fusion [40, 47, 63] and by exploiting sexual cycles in some labora-

127

tory species (S. cerevisae) [19]. These methods, though generally successful in their stated goals

128

of increasing diversity and permitting interclonal gene flow, are low-throughput and are not very

129

efficient, making them difficult to apply to highly parallel evolutionary engineering projects.

TE

D

MA

NU

SC

RI

PT

117

In an effort to circumvent this limitation of conventional recombination tools, a conjugation-

131

based evolution system was recently introduced by Winkler and Kao [59], built on previous work by

132

Cooper that demonstrated that conjugation proficient E. coli evolved more quickly than conjugation

133

incompetent controls [12]. The key innovation of this new method is the design of a strain capable

134

of efficient, bidirectional mating between strains, allowing for interclonal chromosomal DNA ex-

135

change, significantly increasing the rate of recombination within a population during growth. The

136

F conjugation system used by Winkler and Kao is capable of continuous mating in liquid culture,

137

allowing evolution and recombination to take place simultaneously, even under selection. Rates of

138

phenotypic improvement during evolution can be significantly increased as a result, depending on

139

the fitness landscape describing the phenotype of interest [59]. In contrast, genome shuffling by

140

protoplast fusion must be performed in the absence of any selection due to the fragility of the re-

141

combinants generated by the procedure, imposing a large bottleneck due to the loss of cells during

142

the shuffling and re-inoculation process. Recombinants generated by conjugation are immediately

143

subjected to selection, simplifying the generation of strains with improved growth phenotypes. The

AC CE P

130

9

ACCEPTED MANUSCRIPT combination of continuous mating under selection with the improved efficiency of conjugation is a

145

significant improvement over previous methods, and the system has recently been demonstrated to

146

be an effective evolutionary tool by improving the E. coli osmotolerance phenotype [60]. A similar

147

mass mating system, based on the natural yeast mating cycle, has also been developed by Hughes et

148

al. [22]. Since the theoretical advantages of recombination are significant [13, 18], we expect that

149

further refinements of these systems and development of additional tools to enable recombination

150

will play a substantial role in the evolutionary engineering field for the foreseeable future.

151

Connecting Fitness to Production

152

One limitation in the use of adaptive evolution is the required coupling between growth and desired

153

phenotypes, such as consumption of a non-ideal carbon source or tolerance of various inhibitors.

154

A fundamental, though largely unstated, assumption underlying the previous sections is that there

155

exists a simple linkage between the selective pressure of interest and growth rate or viability. The

156

desired mutants are therefore those that grow faster or survive more frequently in the selective

157

environments. However, industrial strains are used to synthesize products-of-interest that are gen-

158

erally not growth-linked, so adaptive laboratory evolution cannot be applied to evolve strains with

159

enhanced productivity. To circumvent this limitation, strain engineering efforts that force the cou-

160

pling between product formation and growth by manipulation of redox balancing in the host strain

161

have been used [15, 24, 50]. In some cases, the selective pressure used in evolutionary engineer-

162

ing can also be designed to provide a growth advantage to better producers. A recent effort by

163

our group successfully demonstrated the use of this approach in using oxidative stress to improve

164

productivity of carotenoids in yeast [44]. The use of anti-metabolites or metabolite analogs have

165

also been applied in evolutionary engineering as selective pressures to either improve productiv-

166

ity [6, 49] or improve substrate utilization [38]. The overriding problem with these approaches is

167

that each selection experiment must be uniquely designed for the pathway and desired metabolite,

168

assuming a viable selective pressure exists at all.

169

AC CE P

TE

D

MA

NU

SC

RI

PT

144

Though engineering growth-product linkages will remain a formidable problem in the near fu10

ACCEPTED MANUSCRIPT ture, significant progress has been made in designing more generic systems that can link product

171

formation to strain fitness. An innovative approach linking a metabolite-sensing riboswitch with a

172

tetracycline resistance cassette (tetA) was recently employed to evolve E. coli for improved lysine

173

and tryptophan productivity [62]. In this case, expression of tetA, the tetracycline resistance gene,

174

was inversely proportional to the concentration of the amino acid of interest. Yang and coworkers

175

took advantage of the fact that high levels of tetA expression confer a Ni2+ -sensitive phenotype,

176

so simple propagation in the presence of Ni2+ was sufficient to select for mutants with large im-

177

provements in amino acid accumulation. The main challenge in generalizing this method is the

178

engineering of suitable riboswitches for each metabolite, though this area has recently seen great

179

strides [5, 7]. An alternative approach called FREP (feedback-regulated evolution of phenotype) by

180

Chou and Keasling has also been successfully applied to increase production of isoprenoids and ty-

181

rosine in E. coli through adaptive evolution [9]. Their method relied on making the expression of a

182

dominant mutator allele (mutD5) inversely proportional to the concentration of a target metabolite,

183

so that low producers would rapidly accumulate mutations that may improve metabolite produc-

184

tion. High producers, in contrast, are more genetically stable due to decreased expression of the

185

mutator allele. Both methods represent the first steps towards a more general method for linking

186

strain productivity to fitness, and allow metabolic engineers to combine the stochasticity of adap-

187

tive evolution with large library sizes permitted by selection (∼ 1010 individuals) for more rapid

188

strain improvement. As additional sensors are designed, evolved, or discovered in nature, it is likely

189

that the evolutionary engineering approach to producer development will become a complementary

190

alternative to the more rational strain engineering approaches used at the moment.

191

Parellelization and Automation of Evolution

192

One of the factors limiting the utility of adaptive laboratory evolution is the capital expense and

193

manual effort required. In our experience, one person can feasibly handle an experiment consisting

194

of 24-30 replicate batch cultures, while performing the necessary maintenance and quantification

195

procedures (e.g., archival storage, contamination tests, adjustment of selective pressure, pheno-

AC CE P

TE

D

MA

NU

SC

RI

PT

170

11

ACCEPTED MANUSCRIPT type improvement analysis, etc). This scale is fairly limiting, as it is difficult to draw statistical

197

conclusions about strain improvement or potential adaptive mechanisms from a small number of

198

replicate cultures. The amount of manual intervention required also introduces an element of bias

199

into the experiment, especially when trying to adjust selective pressures in a consistent manner as

200

population fitness improves.

RI

PT

196

Several new cultivation systems with increased parallelism and automation have recently been

202

developed as a means of overcoming these limitations. For example, control systems and reac-

203

tors themselves can now be inexpensively constructed and customized according to user needs

204

[36, 54]. One particularly innovative tool is the "morbidostat" developed by Toprak et al. [55],

205

which combines a continuous bioreactor array with a feedback control system for adjusting in situ

206

selective pressure automatically. The population growth rate is, therefore, maintained at a constant

207

level throughout the experiment without manual adjustment. The advantages of the system were

208

demonstrated by evolving E. coli for resistance to several antibiotics to show that starkly different

209

selective pressure regimes were needed to maintain the desired level of inhibition. They also in-

210

ferred structural properties of the underlying fitness landscapes [39] based on the ramping schemes

211

used by the feedback controller in the system, providing additional insight as to how adaptation can

212

proceed in response to antibiotic challenges. The use of this approach will clearly improve con-

213

sistency of selective pressures within experiments while enabling researchers to obtain quantitative

214

data concerning the rate of adaptation for target inhibitors. As the evolutionary process is stochas-

215

tic, these high-throughput, small-scale, and low-cost bioreactor designs allow even small labs the

216

ability to run multiple parallel evolutionary experiments in order to identify possible evolutionary

217

trajectories.

AC CE P

TE

D

MA

NU

SC

201

218

Liquid handling robots and microfluidics have also been increasingly used to dramatically im-

219

prove experimental throughput. A fully automated evolution system that propagates microbial cul-

220

tures in microplates was recently introduced by Horinouchi and coworkers [20] and successfully

221

used to evolve strains with improved tolerance of a range of industrially relevant stressors. Other

222

automated systems, also based on batch growth in microtiter plates, has also been used elsewhere

12

ACCEPTED MANUSCRIPT [28, 31]. While these approaches rely on robotics to handle liquid transfers and inoculations, mi-

224

croscale chemostat systems based on the maintenance of populations inside liquid droplets have

225

also been introduced [23], potentially allowing thousands of populations to be maintained in paral-

226

lel. The decreasing cost of liquid handling robots and microfluidic devices will help enable future

227

development of high throughput approaches. It is important to note that the small population sizes

228

imposed by typical plate-based high throughput evolutionary systems may not be desirable in all

229

situations, as the genetic diversity and evolutionary trajectories are more limited compared to larger

230

population sizes. However, the advantages of extensive replication and automated handling are nu-

231

merous: it facilitates detection of divergent or antagonistic adaptation pathways, and it can provide

232

data for theoretical analyses of evolutionary dynamics, clonal interference, or other aspects of pop-

233

ulation genetics. Combined with advances in NGS and genetic analysis, these new high throughput

234

cultivation tools promise to simplify and streamline the process of evolutionary engineering, while

235

providing extensive information on parallel adaptation routes. Knowledge gained from the resulting

236

adaptive mutants can be used to improve existing industrially strains.

237

Understanding Evolved Genotypes

238

Despite advances in other areas, characterization of evolved strains remains a laborious and low-

239

throughput process. The most common approach for uncovering genotype-phenotype linkages

240

involves reconstructing each potentially adaptive mutation into the unevolved parental strain, fol-

241

lowed by the appropriate phenotypic characterization (growth rate, tolerance, etc) to confirm its

242

putative effect. Traditional reconstruction techniques, such as selection-counterselection cassette

243

replacement, transduction, site-directed mutagenesis, and others [14, 48, 53], while powerful, re-

244

quire a significant amount of time to produce a single reconstructed mutant. When the number of

245

mutations per strain is large, or when it is necessary to examine interactions (positive or negative

246

epistasis) between adaptive mutations [26, 37], then reconstruction quickly becomes the limiting

247

step for understanding evolved mutants. The introduction of collections of defined deletion strains

248

for E. coli and S. cerevisae [3, 61], in addition to the ASKA overexpression plasmid collection

AC CE P

TE

D

MA

NU

SC

RI

PT

223

13

ACCEPTED MANUSCRIPT for E. coli [27], has significantly improved the ability of researchers to more rapidly analyze cer-

250

tain types of evolved mutations (e.g. inactivating mutations or mutations resulting in increased

251

gene expression). However, for those mutations occurring in non-model organisms or those that do

252

not result in gene disruption or increased expression, each potentially adaptive mutation must be

253

manually reconstructed using the appropriate tools.

RI

One promising approach towards reducing the time required for strain reconstruction and genotype-

SC

254

PT

249

phenotype analysis in bacteria is a new method named REGRES (recursive genomewide recombi-

256

nation and sequencing) introduced by Quandt and colleagues [41]. REGRES is based on the F con-

257

jugation system, in which strains containing the mutations of interest are converted into Hfr (high

258

frequency recombinant-forming) strains using integrative F plasmids, and subsequently mated to

259

F- recipients. The donor is then selected against using antibiotics, yielding a library of transcon-

260

jugants that contain heterozygous genomes of donor and recipient DNA. These transconjugants

261

can be screened or selected again for the sought after phenotype, followed by NGS or Sanger se-

262

quencing to identify the causative mutations underlying the observed phenotypes. This approach

263

was used to identify the genetic bases for the aerobic citrate utilization trait that arose during the

264

long-term evolution experiment in the Lenski laboratory [4] without the need for manual strain re-

265

construction, which poses a significant challenge given the large number of accumulated mutations

266

in the isolates after more than 50,000 generations of evolution. Further improvements in recom-

267

bineering tools, including the rapid disruption of multiple genes simultaneously [45, 51], will also

268

help to decrease the burden of strain analysis in the future.

269

Conclusions

270

These improvements in evolutionary engineering techniques, though enormously beneficial from

271

both scientific and engineering perspectives, have focused primarily on refinements of the same ex-

272

perimental procedure. In effect, a scientist evolving E. coli for an improved phenotype in the 1950s

273

or the 2000s would have very similar workflows during the actual evolution experiment, despite

AC CE P

TE

D

MA

NU

255

14

ACCEPTED MANUSCRIPT the seismic changes in other areas of the biotechnology field. It is no longer sufficient to merely

275

improve the phenotype of one organism in a single condition using evolution; given that most in-

276

dustrial environments involve a range of stressors (temperature, osmolarity, and so forth) using

277

different microbial species and strains, we must be able to quickly evolve for new traits, dissect

278

their tolerance mechanisms, and then transfer these findings to other organisms to increase their

279

industrial utility. While it is likely that adaptive laboratory evolution will continue to be performed

280

in much the same way for the foreseeable future, we speculate that evolutionary engineering will

281

eventually be able to accomplish this seminal goal as a result of continuing improvement in the

282

tools and methodologies available to evolutionary engineers.

283

Acknowledgements

284

We gratefully acknowledge support from the Norman Hackerman Advanced Research Program

285

(grant number 000512-0004-2011) for this work.

286

Literature Cited

AC CE P

TE

D

MA

NU

SC

RI

PT

274

287

[1] María P Almario, Luis H Reyes, and Katy C Kao. Evolutionary engineering of Saccharomyces

288

cerevisiae for enhanced tolerance to hydrolysates of lignocellulosic biomass. Biotechnology

289

and Bioengineering, 110(10):2616–2623, 2013.

290

[2] Shota Atsumi, Tung-Yun Wu, Iara MP Machado, Wei-Chih Huang, Pao-Yang Chen, Matteo

291

Pellegrini, and James C Liao. Evolution, genomic analysis, and reconstruction of isobutanol

292

tolerance in Escherichia coli. Molecular Systems Biology, 6(1), 2010.

293

[3] Tomoya Baba, Takeshi Ara, Miki Hasegawa, Yuki Takai, Yoshiko Okumura, Miki Baba, Kir-

294

ill A Datsenko, Masaru Tomita, Barry L Wanner, and Hirotada Mori. Construction of Es-

295

cherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Molecular

296

Systems Biology, 2(1), 2006. 15

ACCEPTED MANUSCRIPT

299

300

Nature Reviews Genetics, 14(12):827–839, 2013. [5] Chase L Beisel and Christina D Smolke. Design principles for riboswitch function. PLoS

PT

298

[4] Jeffrey E Barrick and Richard E Lenski. Genome dynamics during experimental evolution.

Computational Biology, 5(4):e1000363, 2009.

RI

297

[6] Jeanne Bonomo, Michael D Lynch, Tanya Warnecke, James V Price, and Ryan T Gill.

302

Genome-scale analysis of anti-metabolite directed strain engineering. Metabolic Engineer-

303

ing, 10(2):109–120, 2008.

NU

SC

301

[7] Andrew L Chang, Joshua J Wolf, and Christina D Smolke. Synthetic rna switches as a tool

305

for temporal and spatial control over gene expression. Current Opinion in Biotechnology,

306

23(5):679–688, 2012.

MA

304

[8] Kian-Kai Cheng, Baek-Seok Lee, Takeshi Masuda, Takuro Ito, Kazutaka Ikeda, Akiyoshi

308

Hirayama, Lingli Deng, Jiyang Dong, Kazuyuki Shimizu, Tomoyoshi Soga, et al. Global

309

metabolic network reorganization by adaptive mutations allows fast growth of Escherichia

310

coli on glycerol. Nature Communications, 5, 2014.

312

TE

AC CE P

311

D

307

[9] Howard H Chou and Jay D Keasling. Programming adaptive control to evolve increased metabolite production. Nature Communications, 4, 2013.

313

[10] Tom M Conrad, Michael Frazier, Andrew R Joyce, Byung-Kwan Cho, Eric M Knight,

314

Nathan E Lewis, Robert Landick, and Bernhard Ø Palsson. Rna polymerase mutants found

315

through adaptive evolution reprogram Escherichia coli for optimal growth in minimal media.

316

Proceedings of the National Academy of Sciences, 107(47):20500–20505, 2010.

317

[11] Tom M Conrad, Andrew R Joyce, M Kenyon Applebee, Christian L Barrett, Bin Xie,

318

Yuan Gao, and Bernhard Ø Palsson. Whole-genome resequencing of Escherichia coli K-12

319

MG1655 undergoing short-term laboratory evolution in lactate minimal media reveals flexible

320

selection of adaptive mutations. Genome Biology, 10(10):R118, 2009.

16

ACCEPTED MANUSCRIPT

323

324

mutations in populations of Escherichia coli. PLoS Biology, 5(9):e225, 2007. [13] James F Crow and Motoo Kimura. Evolution in sexual and asexual populations. American

PT

322

[12] Tim F Cooper. Recombination speeds adaptation by reducing competition between beneficial

Naturalist, pages 439–450, 1965.

RI

321

[14] Hilary M. Ellis, Daiguan Yu, Tina DiTizio, and Donald L. Court. High efficiency mutage-

326

nesis, repair, and engineering of chromosomal DNA using single-stranded oligonucleotides.

327

Proceedings of the National Academy of Sciences, 98(12):6742–6746, 2001.

NU

SC

325

[15] Stephen S Fong, Anthony P Burgard, Christopher D Herring, Eric M Knight, Frederick R

329

Blattner, Costas D Maranas, and Bernhard O Palsson. In silico design and adaptive evolu-

330

tion of Escherichia coli for production of lactic acid. Biotechnology and Bioengineering,

331

91(5):643–648, 2005.

D

MA

328

[16] Hani Goodarzi, Bryson D Bennett, Sasan Amini, Marshall L Reaves, Alison K Hottes,

333

Joshua D Rabinowitz, and Saeed Tavazoie. Regulatory and metabolic rewiring during lab-

334

oratory evolution of ethanol tolerance in E. coli. Molecular Systems Biology, 6(1), 2010.

336

337

338

339

340

AC CE P

335

TE

332

[17] Alan Greener, Marie Callahan, and Bruce Jerpseth. An efficient random mutagenesis technique using an E. colimutator strain. Molecular Biotechnology, 7(2):189–195, 1997. [18] John Haigh. The accumulation of deleterious genes in a population: Muller’s ratchet. Theoretical population biology, 14(2):251–267, 1978. [19] Ira Herskowitz. Life cycle of the budding yeast saccharomyces cerevisiae. Microbiological Reviews, 52(4):536, 1988.

341

[20] Takaaki Horinouchi, Teruaki Minamoto, Shingo Suzuki, Hiroshi Shimizu, and Chikara Furu-

342

sawa. Development of an Automated Culture System for Laboratory Evolution. Journal of

343

Laboratory Automation, page 2211068214521417, 2014.

17

ACCEPTED MANUSCRIPT 344

345

[21] Mian Huang, Mark McClellan, Judith Berman, and Katy C Kao. Evolutionary dynamics of Candida albicans during in vitro evolution. Eukaryotic Cell, 10(11):1413–1421, 2011. [22] Stephen R Hughes, Ronald E Hector, Joseph O Rich, Nasib Qureshi, Kenneth M Bischoff,

347

Bruce S Dien, Badal C Saha, Siqing Liu, Elby J Cox, John S Jackson, David E Sterner,

348

Tauseef R Butt, Joshua LaBaer, and Michael A Cotta. Automated yeast mating protocol

349

using open reading frames from Saccharomyces cerevisiae genome to improve yeast strains

350

for cellulosic ethanol production. Journal of Laboratory Automation, 14(4):190–199, 2009.

NU

SC

RI

PT

346

[23] Slawomir Jakiela, Tomasz S Kaminski, Olgierd Cybulski, Douglas B Weibel, and Piotr

352

Garstecki. Bacterial Growth and Adaptation in Microdroplet Chemostats. Angewandte

353

Chemie, 125(34):9076–9079, 2013.

MA

351

[24] Kaemwich Jantama, MJ Haupt, Spyros A Svoronos, Xueli Zhang, JC Moore, KT Shan-

355

mugam, and LO Ingram. Combining metabolic engineering and metabolic evolution to

356

develop nonrecombinant strains of Escherichia coli C that produce succinate and malate.

357

Biotechnology and Bioengineering, 99(5):1140–1153, 2008.

AC CE P

TE

D

354

358

[25] Katy C Kao and Gavin Sherlock. Molecular characterization of clonal interference during

359

adaptive evolution in asexual populations of Saccharomyces cerevisiae. Nature Genetics,

360

40(12):1499–1504, 2008.

361

[26] Aisha I Khan, Duy M Dinh, Dominique Schneider, Richard E Lenski, and Tim F Cooper.

362

Negative epistasis between beneficial mutations in an evolving bacterial population. Science,

363

332(6034):1193–1196, 2011.

364

[27] Masanari Kitagawa, Takeshi Ara, Mohammad Arifuzzaman, Tomoko Ioka-Nakamichi, Eiji

365

Inamoto, Hiromi Toyonaga, and Hirotada Mori. Complete set of ORF clones of Escherichia

366

coli ASKA library (a complete set of E. coli K-12 ORF archive): unique resources for biolog-

367

ical research. DNA Research, 12(5):291–299, 2006.

18

ACCEPTED MANUSCRIPT [28] Sergey Kryazhimskiy, Daniel P. Rice, Elizabeth R. Jerison, and Michael M. Desai. Global

369

epistasis makes adaptation predictable despite sequence-level stochasticity. 344(6191):1519–

370

1522, 2014.

PT

368

[29] Daniel J Kvitek and Gavin Sherlock. Whole Genome, Whole Population Sequencing Reveals

372

That Loss of Signaling Networks Is the Major Adaptive Strategy in a Constant Environment.

373

PLoS Genetics, 9(11):e1003972, 2013.

375

SC

[30] Gregory I Lang and Andrew W Murray. Estimating the per-base-pair mutation rate in the

NU

374

RI

371

yeast Saccharomyces cerevisiae. Genetics, 178(1):67–82, 2008. [31] Gregory I Lang, Daniel P Rice, Mark J Hickman, Erica Sodergren, George M Weinstock,

377

David Botstein, and Michael M Desai. Pervasive genetic hitchhiking and clonal interference

378

in forty evolving yeast populations. Nature, 500(7464):571–574, 2013.

D

MA

376

[32] Dae-Hee Lee, Adam M Feist, Christian L Barrett, and Bernhard Ø Palsson. Cumulative num-

380

ber of cell divisions as a meaningful timescale for adaptive laboratory evolution of Escherichia

381

coli. PLoS One, 6(10):e26172, 2011.

383

AC CE P

382

TE

379

[33] Peter A Lind and Dan I Andersson. Whole-genome mutational biases in bacteria. Proceedings of the National Academy of Sciences, 105(46):17878–17883, 2008.

384

[34] Guodong Luan, Zhen Cai, Yin Li, Yanhe Ma, et al. Genome replication engineering as-

385

sisted continuous evolution (GREACE) to improve microbial tolerance for biofuels produc-

386

tion. Biotechnology for Biofuels, 6(1):137, 2013.

387

388

389

390

[35] Gabriel Marais and Brian Charlesworth. Genome evolution: recombination speeds up adaptive evolution. Current Biology, 13(2):R68–R70, 2003. [36] Aaron W Miller, Corrie Befort, Emily O Kerr, and Maitreya J Dunham. Design and use of multiplexed chemostat arrays. Journal of Visualized Experiments, (72), 2013.

19

ACCEPTED MANUSCRIPT [37] J. Minty, A. Lesnefsky, F. Lin, Y. Chen, T. Zaroff, A. Veloso, B. Xie, C. McConnell, R. Ward,

392

D. Schwartz, J.M. Rouillard, Y. Gao, E. Gulari, and X.N. Lin. Evolution combined with

393

genomic study elucidates genetic bases of isobutanol tolerance in Escherichia coli. Microbial

394

Cell Factories, 10(1):18, 2011.

PT

391

[38] Ali Mohagheghi, Jeff Linger, Holly Smith, Shihui Yang, Nancy Dowe, and Philip T Pienkos.

396

Improving xylose utilization by recombinant Zymomonas mobilis strain 8b through adaptation

397

using 2-deoxyglucose. Biotechnology for Biofuels, 7(1):19, 2014.

SC

NU

399

[39] H Allen Orr.

Fitness and its role in evolutionary genetics.

10(8):531–539, 2009.

Nature Reviews Genetics,

MA

398

RI

395

[40] Ranjan Patnaik, Susan Louie, Vesna Gavrilovic, Kim Perry, Willem PC Stemmer, Chris M

401

Ryan, and Stephen del Cardayré. Genome shuffling of lactobacillus for improved acid toler-

402

ance. Nature Biotechnology, 20(7):707–712, 2002.

TE

D

400

[41] Erik M Quandt, Daniel E Deatherage, Andrew D Ellington, George Georgiou, and Jeffrey E

404

Barrick. Recursive genomewide recombination and sequencing reveals a key refinement step

405

in the evolution of a metabolic innovation in Escherichia coli. Proceedings of the National

406

Academy of Sciences, page 201314561, 2013.

407

408

AC CE P

403

[42] Amparo Querol, M Teresa Fernández-Espinar, Eladio Barrio, et al. Adaptive evolution of wine yeast. International Journal of Food Microbiology, 86(1):3–10, 2003.

409

[43] Luis H Reyes, Maria P Almario, James Winkler, Margarita M Orozco, and Katy C Kao. Visu-

410

alizing evolution in real time to determine the molecular mechanisms of n-butanol tolerance

411

in Escherichia coli. Metabolic Engineering, 14(5):579–590, 2012.

412

413

[44] Luis H Reyes, Jose M Gomez, and Katy C Kao. Improving carotenoids production in yeast via adaptive laboratory evolution. Metabolic Engineering, 21:26–33, 2014.

414

[45] Owen W Ryan, Jeffrey M Skerker, Matthew J Maurer, Xin Li, Jordan C Tsai, Snigdha Poddar,

415

Michael E Lee, Will DeLoache, John E Dueber, Adam P Arkin, and Jamie H D Cate. Selection 20

ACCEPTED MANUSCRIPT 416

of chromosomal DNA libraries using a multiplex CRISPR system. eLife, pages e03703–

417

e03703, 2014.

420

421

PT

Metabolic Engineering, pages 129–169. Springer, 2001.

RI

419

[46] Uwe Sauer. Evolutionary engineering of industrially important microbial phenotypes. In

[47] R. Schnettler, U. Zimmermann, and C. C. Emeis. Large-scale production of yeast hybrids by electrofusion. FEMS Microbiology Letters, 24(1):81–85, 1984.

SC

418

[48] Shyam K Sharan, Lynn C Thomason, Sergey G Kuznetsov, Donald L Court, et al. Recombi-

423

neering: a homologous recombination-based method of genetic engineering. Nature Proto-

424

cols, 4(2):206–223, 2009.

MA

development in Escherichia coli. Metabolic Engineering, 13(6):674–681, 2011.

D

426

[49] Kevin M Smith and James C Liao. An evolutionary strategy for isobutanol production strain

TE

425

NU

422

[50] Marco Sonderegger and Uwe Sauer. Evolutionary engineering of Saccharomyces cerevisiae

428

for anaerobic growth on xylose. Applied and Environmental Microbiology, 69(4):1990–1998,

429

2003.

430

431

AC CE P

427

[51] Chan Woo Song and Sang Yup Lee. Rapid one-step inactivation of single or multiple genes in escherichia coli. Biotechnology Journal, 8(7):776–784, 2013.

432

[52] Boris U Stambuk, Barbara Dunn, Sergio L Alves, Eduarda H Duval, and Gavin Sherlock.

433

Industrial fuel ethanol yeasts contain adaptive copy number changes in genes involved in

434

vitamin B1 and B6 biosynthesis. Genome Research, 19(12):2271–2278, 2009.

435

436

[53] N Sternberg and R Hoess. The molecular genetics of bacteriophage P1. Annual Review of Genetics, 17(1):123–154, 1983.

437

[54] Chris N. Takahashi, Aaron W. Miller, Felix Ekness, Maitreya J. Dunham, and Eric Klavins.

438

A Low Cost, Customizable Turbidostat for Use in Synthetic Circuit Characterization. ACS

439

Synthetic Biology, 2014. 21

ACCEPTED MANUSCRIPT [55] Erdal Toprak, Adrian Veres, Jean-Baptiste Michel, Remy Chait, Daniel L Hartl, and Roy

441

Kishony. Evolutionary paths to antibiotic resistance under dynamically sustained drug selec-

442

tion. Nature Genetics, 44(1):101–105, 2012.

PT

440

[56] Valeria Wallace-Salinas and Marie F Gorwa-Grauslund. Adaptive evolution of an indus-

444

trial strain of Saccharomyces cerevisiae for combined tolerance to inhibitors and temperature.

445

Biotechnology for Biofuels, 6(1):151, 2013.

SC

RI

443

[57] Yongze Wang, Ryan Manow, Christopher Finan, Jinhua Wang, Erin Garza, and Shengde

447

Zhou. Adaptive evolution of nontransgenic Escherichia coli kc01 for improved ethanol tol-

448

erance and homoethanol fermentation from xylose. Journal of Industrial Microbiology and

449

Biotechnology, 38(9):1371–1377, 2011.

452

453

MA

D

VERT system. Journal of Biological engineering, 6(1):1–8, 2012.

TE

451

[58] James Winkler and Katy C Kao. Computational identification of adaptive mutants using the

[59] James Winkler and Katy C Kao. Harnessing recombination to speed adaptive evolution in

AC CE P

450

NU

446

Escherichia coli. Metabolic Engineering, 14(5):487–495, 2012.

454

[60] James D Winkler, Carlos Garcia, Michelle Olson, Emily Callaway, and Katy C Kao. Evolved

455

Osmotolerant Escherichia coli Mutants Frequently Exhibit Defective N-Acetylglucosamine

456

Catabolism and Point Mutations in Cell Shape-Regulating Protein MreB. Applied and Envi-

457

ronmental Microbiology, 80(12):3729–3740, 2014.

458

[61] Elizabeth A. Winzeler, Daniel D. Shoemaker, Anna Astromoff, Hong Liang, Keith Ander-

459

son, Bruno Andre, Rhonda Bangham, Rocio Benito, Jef D. Boeke, Howard Bussey, An-

460

gela M. Chu, Carla Connelly, Karen Davis, Fred Dietrich, Sally Whelen Dow, Mohamed

461

El Bakkoury, Franc?oise Foury, Stephen H. Friend, Erik Gentalen, Guri Giaever, Johannes H.

462

Hegemann, Ted Jones, Michael Laub, Hong Liao, Nicole Liebundguth, David J. Lockhart,

463

Anca Lucau-Danila, Marc Lussier, Nasiha M’Rabet, Patrice Menard, Michael Mittmann, Chai

22

ACCEPTED MANUSCRIPT Pai, Corinne Rebischung, Jose L. Revuelta, Linda Riles, Christopher J. Roberts, Petra Ross-

465

MacDonald, Bart Scherens, Michael Snyder, Sharon Sookhai-Mahadeo, Reginald K. Storms,

466

Steeve Veronneau, Marleen Voet, Guido Volckaert, Teresa R. Ward, Robert Wysocki, Grace S.

467

Yen, Kexin Yu, Katja Zimmermann, Peter Philippsen, Mark Johnston, and Ronald W. Davis.

468

Functional Characterization of the S. cerevisiae Genome by Gene Deletion and Parallel Anal-

469

ysis. Science, 285(5429):901–906, 1999.

SC

RI

PT

464

[62] Jina Yang, Sang Woo Seo, Sungho Jang, So-I Shin, Chae Hyun Lim, Tae-Young Roh, and

471

Gyoo Yeol Jung. Synthetic RNA devices to expedite the evolution of metabolite-producing

472

microbes. Nature Communications, 4:1413, 2013.

MA

NU

470

[63] Ying-Xin Zhang, Kim Perry, Victor A Vinci, Keith Powell, Willem PC Stemmer, and

474

Stephen B del Cardayré. Genome shuffling leads to rapid phenotypic improvement in bacteria.

475

Nature, 415(6872):644–646, 2002.

AC CE P

TE

D

473

23

ACCEPTED MANUSCRIPT Figures

TE

D

MA

NU

SC

RI

PT

Figure 1: Principal variables that can be adjusted when performing adaptive evolution experiments. A) use of mutagens or mutagenic strains to increase adaptation rate to stressors. B) Detection of clonal interference (mutant competition) to observe independent mutant lineages within a population. C) modification of the applied selective pressure through various means, D) mutant characterization using high-throughput technologies, and E) increased parallelism to detect favored adaptive mechanisms. Advances that alter these experimental properties and more are explored in the text.

AC CE P

476

24

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT



Recent advances in the evolutionary engineering of industrial biocatalysts.

Evolutionary engineering has been used to improve key industrial strain traits, such as carbon source utilization, tolerance to adverse environmental ...
311KB Sizes 3 Downloads 7 Views