Protein Engineering E-Book
153,99 €
Mehr erfahren.
- Herausgeber: Wiley-VCH GmbH
- Kategorie: Fachliteratur
- Serie: Advanced Biotechnology
- Sprache: Englisch
A one-stop reference that reviews protein design strategies to applications in industrial and medical biotechnology
Protein Engineering: Tools and Applications is a comprehensive resource that offers a systematic and comprehensive review of the most recent advances in the field, and contains detailed information on the methodologies and strategies behind these approaches. The authors—noted experts on the topic—explore the distinctive advantages and disadvantages of the presented methodologies and strategies in a targeted and focused manner that allows for the adaptation and implementation of the strategies for new applications.
The book contains information on the directed evolution, rational design, and semi-rational design of proteins and offers a review of the most recent applications in industrial and medical biotechnology. This important book:
- Covers technologies and methodologies used in protein engineering
- Includes the strategies behind the approaches, designed to help with the adaptation and implementation of these strategies for new applications
- Offers a comprehensive and thorough treatment of protein engineering from primary strategies to applications in industrial and medical biotechnology
- Presents cutting edge advances in the continuously evolving field of protein engineering
Written for students and professionals of bioengineering, biotechnology, biochemistry, Protein Engineering: Tools and Applications offers an essential resource to the design strategies in protein engineering and reviews recent applications.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 817
Veröffentlichungsjahr: 2021
Ähnliche
Table of Contents
Cover
Title Page
Copyright
Part I: Directed Evolution
1 Continuous Evolution of Proteins
In Vivo
1.1 Introduction
1.2 Challenges in Achieving
In Vivo
Continuous Evolution
1.3 Phage‐Assisted Continuous Evolution (PACE)
1.4 Systems That Allow
In Vivo
Continuous Directed Evolution
1.5 Conclusion
References
2
In Vivo
Biosensors for Directed Protein Evolution
2.1 Introduction
2.2 Nucleic Acid‐Based
In Vivo
Biosensors for Directed Protein Evolution
2.3 Protein‐Based
In Vivo
Biosensors for Directed Protein Evolution
2.4 Characteristics of Biosensors for
In Vivo
Directed Protein Evolution
2.5 Conclusions and Future Perspectives
Acknowledgments
References
3 High‐Throughput Mass Spectrometry Complements Protein Engineering
3.1 Introduction
3.2 Procedures and Instrumentation for MS‐Based Protein Assays
3.3 Technology Advances Focusing on Throughput Improvement
3.4 Applications of MS‐Based Protein Assays: Summary
3.5 Conclusions and Perspectives
Acknowledgments
References
Notes
4 Recent Advances in Cell Surface Display Technologies for Directed Protein Evolution
4.1 Cell Display Methods
4.2 Selection Methods and Strategies
4.3 Modifications of Cell Surface Display Systems
4.4 Recent Advances to Expand Cell‐Display Directed Evolution Techniques
4.5 Conclusion and Outlook
References
5 Iterative Saturation Mutagenesis for Semi‐rational Enzyme Design
5.1 Introduction
5.2 Recent Methodology Developments in ISM‐Based Directed Evolution
5.3 B‐FIT as an ISM Method for Enhancing Protein Thermostability
5.4 Learning from CAST/ISM‐Based Directed Evolution
5.5 Conclusions and Perspectives
Acknowledgment
References
Part II: Rational and Semi‐Rational Design
6 Data‐driven Protein Engineering
6.1 Introduction
6.2 The Data Revolution in Biology
6.3 Statistical Representations of Protein Sequence, Structure, and Function
6.4 Learning the Sequence‐Function Mapping from Data
6.5 Applying Statistical Models to Engineer Proteins
6.6 Conclusions and Future Outlook
References
Note
7 Protein Engineering by Efficient Sequence Space Exploration Through Combination of Directed Evolution and Computational Design Methodologies
7.1 Introduction
7.2 Protein Engineering Strategies
7.3 Conclusions and Future Perspectives
Acknowledgments
References
8 Engineering Artificial Metalloenzymes
8.1 Introduction
8.2 Rational Design
8.3 Engineering Artificial Metalloenzyme by Directed Evolution in Combination with Rational Design
8.4 Summary and Outlook
Acknowledgment
References
Notes
9 Engineered Cytochromes P450 for Biocatalysis
9.1 Cytochrome P450 Monooxygenases
9.2 Engineered Bacterial P450s for Biocatalytic Applications
9.3 High‐throughput Methods for Screening Engineered P450s
9.4 Engineering of Hybrid P450 Systems
9.5 Engineered P450s with Improved Thermostability and Solubility
9.6 Conclusions
Acknowledgments
References
Part III: Applications in Industrial Biotechnology
10 Protein Engineering Using Unnatural Amino Acids
10.1 Introduction
10.2 Methods for Unnatural Amino Acid Incorporation
10.3 Applications of Unnatural Amino Acids in Protein Engineering
10.4 Outlook
10.5 Conclusions
References
11 Application of Engineered Biocatalysts for the Synthesis of Active Pharmaceutical Ingredients (APIs)
11.1 Introduction
11.2 Conclusions
References
12 Directing Evolution of the Fungal Ligninolytic Secretome
12.1 The Fungal Ligninolytic Secretome
12.2 Functional Expression in Yeast
12.3 Yeast as a Tool‐Box in the Generation of DNA Diversity
12.4 Bringing Together Evolutionary Strategies and Computational Tools
12.5 High‐Throughput Screening (HTS) Assays for Ligninase Evolution
12.6 Conclusions and Outlook
Acknowledgments
References
13 Engineering Antibody‐Based Therapeutics: Progress and Opportunities
13.1 Introduction
13.2 Antibody Formats
13.3 Antibody Discovery
13.4 Therapeutic Optimization of Antibodies
13.5 Manufacturability of Antibodies
13.6 Conclusions
Acknowledgments
References
14 Programming Novel Cancer Therapeutics: Design Principles for Chimeric Antigen Receptors
14.1 Introduction
14.2 Metrics to Evaluate CAR‐T Cell Function
14.3 Antigen‐Recognition Domain
14.4 Extracellular Spacer
14.5 Transmembrane Domain
14.6 Signaling Domain
14.7 High‐Throughput CAR Engineering
14.8 Novel Receptor Modalities
References
Part IV: Applications in Medical Biotechnology
15 Development of Novel Cellular Imaging Tools Using Protein Engineering
15.1 Introduction
15.2 Cellular Imaging Tools Developed by Protein Engineering
15.3 Application in Cellular Imaging
15.4 Conclusion and Perspectives
References
Index
a
b
c
d
e
f
g
h
i
k
l
m
n
o
p
q
r
s
t
u
v
y
z
End User License Agreement
List of Tables
Chapter 1
Table 1.1 Comparison among approaches for
in vivo
continuous evolution.
Chapter 2
Table 2.1 List of
in vivo
biosensors associated with directed protein evolution.
Chapter 3
Table 3.1 Recent, select applications of MS‐based protein assays.
Chapter 4
Table 4.1 Specifications of cell display systems for directed evolution of prote...
Chapter 5
Table 5.1 Oversampling needed in order to ensure 95% library coverage in saturat...
Table 5.2 Selected examples of recent CAST/ISM studies.
Chapter 7
Table 7.1 Summary of computational methods on rational protein design approaches...
Table 7.2 Successful application of Know Volution strategy to improve enzymes pr...
Chapter 11
Table 11.1 Selected examples of enzyme engineering approaches for the preparatio...
Chapter 13
Table 13.1 Antibody‐receptor interactions of interest.
List of Illustrations
Chapter 1
Figure 1.1 A schematic illustration of a typical directed evolution setup. (a)...
Figure 1.2 Pace. Phage carrying the
selection plasmid
(
SP
) encoding the GOI pr...
Figure 1.3 Targeted mutagenesis in
E. coli
with error‐prone DNA polymeras...
Figure 1.4 Yeast systems for targeted mutagenesis of GOIs. (a) TaGTEAM is achi...
Figure 1.5 Targeted mutagenesis by somatic hypermutation (CRISPRx). A region o...
Figure 1.6 OrthoRep. In OrthoRep, GOIs encoded on the orthogonal p1 plasmid ar...
Chapter 2
Figure 2.1 Overview of
in vivo
biosensor‐mediated directed protein evolution. ...
Figure 2.2 RNA‐based
in vivo
biosensors. Ligand‐binding at a riboswitch induce...
Figure 2.3 Protein‐based
in vivo
biosensors. Ligand‐binding at a transcription...
Figure 2.4 RNA‐polymerase‐based
in vivo
biosensors. T7 RNA polymerase linked t...
Chapter 3
Figure 3.1 Typical procedures of MS‐based protein assays. (a) Solution assays ...
Figure 3.2 (a) Scheme of a mass spectrometer. (b–d) ESI, MALDI and SIMS ion so...
Figure 3.3 Correlation of MS signals and enzyme properties. (a) Enzyme convers...
Figure 3.4 Engineer transaminase ATA‐036 activity using microfluidics‐MS scree...
Figure 3.5 Optically guided MALDI MS to profile microbial colonies to engineer...
Figure 3.6 Isotopic labeling for resolving chiral substrates/products by MS to...
Chapter 4
Figure 4.1 Examples of cell surface display system. (a)
Phage surface display
....
Figure 4.2 High‐throughput screening methods. (a) Fluorescent‐activated cell s...
Figure 4.3 Microcapillary single‐cell analysis and laser extraction (μSCALE). ...
Figure 4.4 Phage‐assisted continuous evolution (PACE). Continuous selection of...
Chapter 5
Figure 5.1 Generalization of the CAST/ISM‐strategy for manipulating stereosele...
Scheme 5.1 LEH‐catalyzed hydrolytic desymmetrization of substrate
1
.
Scheme 5.2 TbSADH‐catalyzed stereoselective transformations.
Figure 5.2 Five CAST residues A85, I86, W110, L294, and C295 in contact with s...
Figure 5.3 Best mutants obtained upon generating and screening (a) TbSADH libr...
Figure 5.4 Construction of
E. coli
whole cells for producing either (
R
,
R
)...
Figure 5.5 Directed evolution of P450‐BM3 in the regio‐ and enantioselective o...
Scheme 5.3 Model reaction for P450‐BM3 catalyzed C16‐selective hydroxylation o...
Figure 5.6 Mutability landscape of P450‐BM3 towards testosterone (
16
). The fiv...
Scheme 5.4 Active mutants for C16‐selective hydroxylation of four additional s...
Figure 5.7 The Twist system of precisely controlled combinatorial library fabr...
Figure 5.8 Results of massive DNA sequencing as a reliable quality control in ...
Chapter 6
Figure 6.1 The growth of biological data. (a, b) DNA sequencing and synthesis ...
Figure 6.2 Sequence, structure, and function representations. (a) A protein's ...
Figure 6.3 A linear regression model for cytochrome P450 thermostability. This...
Figure 6.4 Unsupervised learning from protein sequences. (a) Statistical coupl...
Figure 6.5 Active machine learning. (a) Active learning involves designing max...
Chapter 7
Figure 7.1 Framework of the principle strategy of (a) FRESCO [18,19], (b) Fold...
Figure 7.2 Framework of the principle strategy of (a) CNA [23], (b) PROSS [24]...
Figure 7.3 Overview of the KnowVolution strategy. Four phases are (I) identifi...
Chapter 8
Figure 8.1 (a) Model of DFsc (top) and G4DFsc (bottom) from computational desi...
Figure 8.2 Design of SiRC
c
P starting from scaffold search followed by design o...
Figure 8.3 (a) Structure of heme and iron porphycene (top) and cartoon of the ...
Figure 8.4 (a) Structure of ruthenium photosensitizer, (b) structure of DuBois...
Figure 8.5 Schematic representation of the LmrR‐heme assembly (top), structure...
Figure 8.6 Directed evolution strategy used to obtain Ir(Me)‐mOCR‐Myo mutants ...
Figure 8.7 Artificial osmium peroxygenase from Osmium–Cupin complex..
Figure 8.8 Model reaction of SPAAC and artificial metalloenzyme structure..
Figure 8.9 Overview of artificial metalloenzyme evolution protocol..
Figure 8.10 Chemoenzymatic optimization of artificial metalloenzymes based on ...
Figure 8.11 Representative transfer hydrogenation reactions catalyzed by artif...
Figure 8.12 Synergistic effect of a basic residue introduced by site‐directed ...
Figure 8.13 Streptavidin‐based artificial metalloenzymes for
in vivo
metathesi...
Figure 8.14 Assembly of cell‐penetrating ArMs. Ruthenium complexes
1
and
2
cat...
Chapter 9
Figure 9.1 The catalytic cycle of P450 enzymes.
Figure 9.2 Model of heme domain of P450
BM3
(left) in complex with
N
‐palmitoylg...
Figure 9.3 P450 classes: (a) bacterial, class I P450 system (e.g. P450cam); (b...
Figure 9.4 Oxyfunctionalization of non‐native substrates by engineered variant...
Figure 9.5 Selective chemoenzymatic mono‐ and difluorination of ibuprofen meth...
Figure 9.6 Oxyfunctionalization of non‐native substrates by engineered variant...
Figure 9.7 Oxyfunctionalization of non‐native substrates by engineered variant...
Figure 9.8 Oxyfunctionalization of non‐native substrates by engineered variant...
Figure 9.9 Application of the engineered P450 variant aMOx for the anti‐Markov...
Figure 9.10 (a) P450
BM3
‐catalyzed intramolecular C–H amination of 2‐aminoaceta...
Figure 9.11 Development of regio‐ and stereoselective P450
BM3
‐based catalysts ...
Figure 9.12 Chemo‐and regioselective oxidation of the monocyclic diterpenoid β...
Figure 9.13 Regio‐ and stereoselective oxidation of testosterone with engineer...
Figure 9.14 Total synthesis of nigelladine A involving a C–H oxidation step ca...
Figure 9.15 Synthesis of drug metabolites using engineered P450
BM3
variants.
Figure 9.16 High‐throughput methods for engineering of regio‐ and stereoselect...
Chapter 10
Figure 10.1 In the native translation process (left), tRNA is charged with the...
Figure 10.2 F2Y NMR probe revealed arrestin mediated signaling directed by pho...
Figure 10.3 Unnatural amino acids used to probe the role of Tyr in an artifici...
Figure 10.4 (a) Overlay of the active sites of Mb His (gray, PDB ID: 1A6K) and...
Figure 10.5 Some metal‐coordinating UAAs and their potential coordinating mode...
Chapter 11
Figure 11.1 Transaminase catalyzed reductive amination to form sitagliptin.
Figure 11.2 Enzymatic processes for atorvastatin intermediates.
Figure 11.3 KRED catalyzed reduction of prochiral ketone for ticagrelor interm...
Figure 11.4 KRED catalyzed reduction of tetrahydrothiophene‐3‐one.
Figure 11.5 Biocatalytic synthesis of key fragment for vibegron.
Figure 11.6 AADH catalyzed production of unnatural amino acids.
Figure 11.7 Rationale in re‐tasking of P450 to mediate carbene transfer.
Figure 11.8 Panel of hydroxylated substrates synthesized using A328 variants o...
Figure 11.9 Application of an engineered BVMO for the synthesis of esomeprazol...
Figure 11.10 MAO‐N variants for the preparation of different APIs.
Figure 11.11 Hydrogen borrowing cascade employing an engineered imine reductas...
Figure 11.12 Esterase catalyzed resolution of α‐hydroxy acids for the synthesi...
Figure 11.13 Bioremediation of TCP with haloalkane dehalogenase to afford enan...
Figure 11.14 Five‐step cascade employing five engineered enzymes (in red) and ...
Chapter 12
Figure 12.1 Interrelationships for the ligninolytic consortium of enzymes. Lac...
Figure 12.2 General scheme for a directed evolution round of ligninases. From ...
Figure 12.3 Mapping of the mutations (highlighted in blue) for functional expr...
Figure 12.4 Library creation methodologies based on
S. cerevisiae
: (a) IV...
Chapter 13
Figure 13.1 Therapeutic antibody formats. Several antibody‐based therapeutics ...
Figure 13.2 Details of the receptor binding sites on human IgG1 Fc. The human ...
Figure 13.3 Mechanisms of antibody‐mediated therapeutic effects. (a) Fc‐indepe...
Chapter 14
Figure 14.1 Metrics of CAR‐T cell performance. CAR engineering typically begin...
Figure 14.2 Basic structure of a chimeric antigen receptor. Starting from the ...
Figure 14.3 Programming CARs capable of executing molecular logic. (a) Single‐...
Figure 14.4 Novel receptor modalities for engineering anti‐tumor T cells. (a) ...
Chapter 15
Figure 15.1 The basic principle of localization‐based super‐resolution microsc...
Figure 15.2 A schematic representation of different biosensors and their activ...
Figure 15.3 The design principle of MT1‐MMP FRET biosensors. (a) The schematic...
Figure 15.4 Imaging and quantification of basal level of apparent MT1‐MMP acti...
Figure 15.5 Schematic illustrations of the design and activation mechanism of ...
Guide
Cover Page
Title Page
Copyright
Table of Contents
Begin Reading
Index
WILEY END USER LICENSE AGREEMENT
Pages
iii
iv
1
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
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
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
124
125
123
126
127
128
129
130
131
132
133
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
288
284
285
286
287
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
377
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
Protein Engineering
Tools and Applications
Edited byHuimin Zhao
Volume Editor
Professor Huimin ZhaoUniversity of Illinois at UrbanaChemical & Biomolecular Engineering600 South Mathews Avenue215 Roger Adams Laboratory61801 Urbana ILUSA
Series Editors
Prof. Dr. Sang Yup LeeKAIST373‐1; Guseong‐Dong291 Daehak‐ro,Yuseong‐gu305‐701 DaejonSouth Korea
Prof. Dr. Jens NielsenChalmers UniversityDepartment of Biology and Biological EngineeringKemivägen 10412 96 GöteborgSweden
Prof. Dr. Gregory StephanopoulosMassachusetts Institute of TechnologyDepartment of Chemical EngineeringMassachusetts Ave 77Cambridge, MA 02139USA
Cover Culture Flasks in microbiologicallaboratory / science photo, fotolia
All books published by WILEY‐VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.
Library of Congress Card No.:applied for
British Library Cataloguing‐in‐Publication DataA catalogue record for this book is available from the British Library.
Bibliographic information published by the Deutsche NationalbibliothekThe Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at <http://dnb.d-nb.de>.
© 2021 WILEY‐VCH GmbH, Boschstr. 12, 69469 Weinheim, Germany
All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law.
Print ISBN: 978‐3‐527‐34470‐3
ePDF ISBN: 978‐3‐527‐81509‐8
ePub ISBN: 978‐3‐527‐81511‐1
oBook ISBN: 978‐3‐527‐81512‐8
Cover Design Adam-Design, Weinheim, Germany
Part IDirected Evolution
1Continuous Evolution of Proteins In Vivo
Alon Wellner1, Arjun Ravikumar1, and Chang C. Liu1,2,3
1University of California, Department of Biomedical Engineering, 3201 Natural Sciences II, Irvine, CA, 92697, USA
2University of California, Department of Chemistry, 1102 Natural Sciences 2, Irvine, CA, 92697, USA
3University of California, Department of Molecular Biology and Biochemistry, 3205 McGaugh Hall, Irvine, CA, 92697, USA
1.1 Introduction
Directed evolution is a powerful approach for engineering new biomolecular and cellular functions [1–3]. In contrast to rational design approaches, directed evolution exploits diversity and evolution to shape the behavior of biological matter by applying the Darwinian cycle of mutation, selection, and amplification of genes and genomes. By doing so, the field of directed evolution has generated important insights into the evolutionary process [4–6] as well as useful RNAs, proteins, and systems with wide‐ranging applications across biotechnology and medicine [7–11].
To mimic the evolutionary process, classical directed evolution approaches carry out cycles of ex vivo diversification on genes of interest (GOIs), transformation of the resulting gene libraries into cells, and selection of the desired function (Figure 1.1). Each iteration of this cycle is defined as a round of evolution, and as selection stringency increases over rounds, either automatically through competition or manually through changing conditions (or both), this process can lead GOIs closer and closer to the desired function. This overall process makes practical sense for a number of reasons, especially for the goal of protein engineering (i.e. GOI encodes a protein). First, ex vivo diversification is appropriate, because test tube molecular biology techniques such as DNA shuffling, site‐directed saturation mutagenesis, and error‐prone (ep) polymerase chain reaction (PCR) [2] are capable of generating exceptionally high and precise levels of sequence diversity for any GOI. Second, transforming diversified libraries of the GOI into cells is appropriate, because each GOI variant needs to be translated into a protein in order to express its function, and cells, especially model microbes, are naturally robust hosts for protein expression. Third, carrying out selection inside cells is appropriate, because (i) cells automatically maintain the genotype–phenotype connection between the GOI and expressed protein that is necessary for amplification of desired variants, (ii) we often care about a GOIs function within the context of a cell, especially as metabolic engineering and cell‐based therapy applications mature, and (iii) the use of cell survival as the output for a desired protein function allows millions or billions of GOI variants to be simultaneously tested – it is easy to culture billions of cells under selection conditions – in contrast to ex vivo screens that are much lower throughput. Survival‐based selections are not always immediately available, but one can often find a way to reliably link the desired function of a protein to cellular fitness.
Figure 1.1 A schematic illustration of a typical directed evolution setup. (a) A GOI is diversified ex vivo, typically by applying an error‐prone PCR to generate a GOI library. (b) The library is then cloned into an expression vector and transformed/transfected into cells that are subjected to (c) outgrowth and selection for enhanced protein activity. (d) Plasmid DNA that is enriched for library members with increased properties is extracted and (e) subjected again to diversification and selection. The directed evolution cycle is iterated until the desired outcome is achieved or until diminishing returns (a plateau is reached).
While sensible, the practical requirement that diversification should occur in vitro but expression and selection should occur in vivo in this classical directed evolution pipeline creates significant suboptimalities. First, the number of steps that can be taken along an adaptive path becomes few, since each round of in vitro mutation, transformation, and in vivo selection takes several days or weeks to carry out. Second, limited DNA transformation efficiencies result in strong bottlenecking of diversity that can mitigate the probability of finding the most optimal solutions in sequence space. Third, the number of evolution experiments that can be run simultaneously is minimal, because in vitro mutagenesis, cloning, and transformation are experimentally onerous, demanding extensive researcher intervention [12]. These shortcomings keep two highly promising categories of experiments largely outside the grasp of classical methods: first is the directed evolution of genes towards highly novel functions that likely require long mutational paths to reach (e.g. the optimization of multi‐gene metabolic pathways or the de novo evolution of enzyme activity); and second is the large‐scale replication of directed evolution experiments, needed in cases when many different functional variants of a gene are desired (e.g. the evolution of multiple synthetic receptors for a collection of ligands) or when statistical power is required in order to understand outcomes in experimental evolution (e.g. probing the scope of adaptive trajectories leading to resistance in a drug target).
An emerging field of in vivo continuous directed evolution seeks to overcome these shortcomings by performing both continuous diversification of the GOI and selection entirely within living cells [13]. In this way, GOIs can be rapidly and continuously evolved through basic serial passaging of cells under selective conditions. This removes the labor‐intensive cycling between in vitro and in vivo steps and the DNA transformation bottlenecks associated with the classical pipeline, creating a new paradigm for directed evolution that is limited only by the generation time of the host cell and the number of cells that can be cultured. These limitations are usually negligible – in most host organisms for directed evolution such as Escherichia coli and Saccharomyces cerevisiae, generation time is fast (20–100 minutes) and the number of cells that can be cultured is massive (108–109 ml−1) – so the potential power of continuous systems is enormous. Moreover, in vivo continuous directed evolution is amenable to high‐throughput experiments, because serial passaging is straightforward and can be automated at scale or converted to continuous culture using bioreactors [14–16]. In this chapter, we discuss various systems that partially or fully achieve in vivo continuous evolution (ICE).
1.2 Challenges in Achieving In Vivo Continuous Evolution
Before discussing how ICE can be realized, we shall first clarify why this is a challenging problem. The difficulty of achieving ICE of GOIs lies in the fundamental relationship between how fast one can mutate an information polymer and its length. Several theories predict that organisms face an “error threshold” at mutation rates on order 1/L (where L is the length of the genome), near which selection cannot maintain fitness, leading to gradual decline towards low fitness, or above which one is nearly guaranteed a lethal mutation every cycle of replication, leading to rapid extinction [17–20]. Because cellular genomes are large (e.g. ∼5 × 106 in E. coli, ∼1.2 × 107 in S. cerevisiae, and ∼ 3 × 109 in humans), this implies that evolution strongly favors low genomic mutation rates (e.g. ∼5 × 10−8 substitutions per base [s.p.b.] in E. coli, ∼10−10 s.p.b. in S. cerevisiae and ∼3 × 10−9 in human somatic cells) [21–23]. Experiment confirms this prediction. Drake observed empirically that mutation rates scale as 1/L across many organisms [17]; evolution experiments have shown that when mutator phenotypes do arise, they are accompanied by fitness costs and only transiently persist [19,24–26]; and more direct tests in yeast find that there is indeed a mutation‐induced extinction threshold at ∼1/L, above which yeast cannot propagate [18]. Yet individual GOIs are small in comparison with genomes, so they are capable of tolerating much higher error rates. In fact, they require much higher per base error rates than genomes to generate the same amount of total mutational diversity, because they have fewer bases. Following the 1/L scaling, a typical 1 kb GOI should be able to tolerate mutation rates on order ∼10−3 s.p.b.
Therefore, the primary challenge in achieving rapid ICE is how to develop molecular machinery or other strategies that target rapid mutagenesis to only GOIs, allowing the host genome to replicate at mutation rates below its low error thresholds but driving the GOI at the high mutation necessary for fast generation of sequence diversity. When considering the level of targeting in the ideal case, the formidability of this challenge becomes quite apparent. Ideally, one should continuously mutate GOIs at rates close to their error threshold (∼10−3) to maximize diversification but leave the genomic error rate completely unchanged, as the genome's error rate is evolutionarily optimized for host fitness. In E. coli, S. cerevisiae, and human cells, this means that on‐target versus off‐target mutagenesis must differ by 106‐fold, 107‐fold, and 107‐fold, respectively, which is much more than the 10‐ to 1000‐fold targeting required in most synthetic biology problems involving molecular recognition. How can we achieve such extreme precision in mutational targeting in the cell?
There is yet another hard challenge in realizing ICE, which has to do with the durability of mutagenesis. Ideally, one wants a high rate of mutagenesis on the GOI to persist indefinitely (or at least for as long as the experimenter cares), so that a protein can traverse long mutational pathways towards desired functions. Because one needs to achieve mutational targeting to the GOI, there is almost always a risk to durability: any mechanism for targeting the GOI over the rest of the genome will necessarily rely on some cis‐elements in or surrounding the GOI to mediate the targeting. If these cis‐elements become mutated, which is quite likely since they are usually in or near the GOI undergoing rapid mutation, then mutagenesis will slow or stop. Ideally, a continuous evolution system will limit the chance that a cis‐element for mutational targeting gets degraded. In the case that it does, an ideal system will remove the GOI containing the mutated cis‐element from the population so that it can't fix in the population (through gradual mutational accumulation or a selective sweep if mutagenesis comes with a fitness cost) and end the continuous evolution process prematurely. How do we achieve architectures for durability?
Other challenges for ICE include generality across host organisms, the ability to mutate many genes simultaneously, and fine control over mutation rate and spectra; but the most defining ones are targeting and durability. In the remainder of this chapter, we review several in vivo continuous directed evolution platforms within the framework of these challenges. We highlight in Section 1.4.4 and note here that our recently developed orthogonal DNA replication (OrthoRep), among systems for ICE, seems uniquely capable of complete precision in mutational targeting (as far as we can tell), and is a highly durable architecture for enforcing prolonged mutagenesis in GOIs. We also highlight, in Section 1.3, phage‐assisted continuous evolution (PACE), which has been remarkably successful for continuous biomolecular evolution. Although PACE is not an entirely in vivo system, it achieves complete precision in mutational targeting and durability – in fact by not being entirely in vivo, as we will explain. We do not discuss several powerful technologies for non‐continuous in vivo diversification or streamlined diversification methods, such as MAGE [27], CREATE [28], DiVERGE [29], and CPR [30], but note that these are also promising approaches to protein evolution as they address some of the constraints of classical directed evolution methods. A summary of various characteristics of the systems we discuss is provided in Table 1.1.
Table 1.1 Comparison among approaches for in vivo continuous evolution.
Approach
Systems
Mutation rate
Targeting of mutagenesis
Durability of mutagenesis
Number (simple estimates) and location of genes that can be evolved simltaneously
Generality across host organisms
Mutational spectrum
References
Continuous rounds of evolution with a conditionally replicating bacteriophage
PACE
Mutates GOIs at ∼10
−3
s.p.b.
Complete targeting to the bacteriophage genome, since
E. coli
are constantly replaced
Indefinitely continuous since mutagenesis is enforced. In practice, this method is typically implemented for 1–3 weeks
1–10 genes encoded on bacteriophage genome.
Currently in
E. coli
. Could be implemented with mammalian cells using non‐integrating viruses (e.g. adenovirus).
Fairly unbiased mutational spectrum.
[33]
Targeted mutagenesis in
E. coli
with error‐prone DNA polymerase I
ep Pol I/ColE1‐based systems, CRISPR‐guided DNA polymerases (EvolvR)
GOIs encoded near the ColE1 origin are mutated by ep Pol I at ∼10
−3
s.p.b. CRISPR‐guided Pol I can induce rates as high as 10
−2
, but this quickly drops off after the guide region
Targeting with unfused ep Pol I is maximally only ∼400‐fold. Fusion to nCas9 generally improves targeting to ∼1000‐fold
Durability remains to be tested. Ep Pol I/ColE1 incurs significant off‐target mutagenesis, which could quickly abrogate mutagenesis. EvolvR risks breaking down because it rapidly mutates the gRNA target region
1–5 genes encoded on a plasmid with ep Pol I/ColE1. 1–20 genes on plasmids or at their endogenous genomic loci with EvolvR, depending on how many targeting sgRNAs one can stably encode.
Both systems are currently in
E. coli
. EvolvR should be fairly general across hosts, especially with the use of Phi29 DNAP.
ep Pol I mutates ColE1 plasmids with a bias towards transition mutations. EvolvR generates substitutions of all four nucleotide types, in a relatively unbiased manner. If needed, this can be improved through DNAP engineering.
[47
–
50
,
54]
Yeast systems that do not use engineered DNA polymerases for mutagenesis
TaGTEAM, ICE
For TaGTEAM, ∼10
−4
s.b.p. at 10 kb regions on both sides of the tetO array. For ICE, 1.5 × 10
−4
, if excluding the rate of retrotransposition needed to induce mutagenesis
TaGTEAM offers targeting of genomic GOIs, however with low accuracy. ICE's targeting is theoretically good since off‐target regions are not reverse transcribed
Durability remains to be tested. Off‐target mutation and the requirement that retrotransposition occurs back into the original locus for continued evolution with ICE will likely affect durability
1–10 genes on plasmids or at engineered genomic loci
Both systems are currently in yeast. ICE has been demonstrated in several diverged yeast species. TaGTEAM should function in
E. coli
and mammalian cells. ICE could be implemented in new hosts using retrotransposable elements similar to Ty1.
TaGTEAM generates a broad spectrum of both transitions and transversions. In addition, 25% of mutations are single base deletions. In ICE there is a 1 : 1 ratio between transitions and transversions.
[
55
,
58
]
Somatic hypermutation as a means for targeted mutagenesis of GOIs
Hypermutator B cell line, Ramos cell line, dCas9‐AID fusions (such as CRISPRx), T7 RNAP‐AID fusion
CRISPRx mutates GOIs at ∼5 × 10
−4
s.p.b.
Efficient targeting. No increase in mutagenesis rate was detected in an off‐target locus. The hyperactive AID variant can create dense, highly variable point mutations within a region of 100 bp surrounding an sgRNA target site
Durability remains to be tested.
1–10 genes on plasmids or at engineered genomic loci with the hypermutator B cell line, Ramos cell line, or T7 RNAP‐AID fusion. Dozens of genes at endogenous genomic loci with dCas9‐AID fusions
Systems depending on natural SHM are limited to mammalian cells. AID‐fusions are currently available in mammalian systems or
E. coli
, depending on the system. AID fusions should be extensible to all host‐types.
AID generates point mutations rather than insertions and deletions, and it favors transitions over transversions. However, repair pathways operate at AID‐mutated loci to extend the scope of mutagenesis.
[67
–
73]
Orthogonal DNA replication
OrthoRep
Mutates GOIs at ∼10
−5
s.p.b.
Complete orthogonality (at least 100 000‐fold targeting)
Indefinitely continuous since mutagenesis is enforced. This method has been implemented for up to 300 generations without any sign of erosion
1–10 genes encoded on a special orthogonal plasmid
Currently in yeast. Should be extensible to bacteria and mammalian systems using related protein‐priming DNAPs.
TP‐DNAP1‐4‐2 strongly favors transition mutations. This can be readily improved through DNAP engineering.
[
74
,
75
,
78
]
Source: Esvelt et al. [33]; Fabret et al. [47]; Alexander et al. [49].
1.3 Phage‐Assisted Continuous Evolution (PACE)
The most successfulmethod for continuous protein evolution thus far is the PACE system developed in the lab of David Liu (Figure 1.2) [2,12,14,15,31–37]. PACE reimagines traditional “rounds” of directed evolution as generations of the M13 bacteriophage life‐cycle, thereby transforming a step‐wise and labor‐intensive procedure into a continuous biological process. In PACE, GOIs are encoded in the M13 genome, and the resulting phage continuously replicate in a vessel (termed “lagoon”) that experiences a constant influx of E. coli cells. To create a selection pressure for GOIs to evolve, the activity of interest is coupled to phage survival. This is achieved by deleting the essential gene III (gIII), encoding coat protein III (pIII), from the M13 genome. The host E. coli strain is engineered to encode gIII in a genetic circuit that makes pIII expression dose‐dependent on the desired activity of the GOI (see the following text for examples); so only phage that successfully evolves the GOI can trigger pIII expression and continue propagating. Due to the rapid generation time of M13 (∼10 minutes without selection), evolution in this manner can iterate hundreds of times in just a few days.
Figure 1.2 Pace. Phage carrying the selection plasmid (SP) encoding the GOI propagates on E. coli cells which are constantly flowing into the “lagoon” at a rate that does not permit their propagation but is longer than the phage life cycle, thus permitting phage replication. Upon infection, the SP (as well as the bacterial genome) experiences a high degree of mutagenesis due to the presence of a mutator plasmid (MP). In a PACE experiment, high GOI activity (green) is linked to drive strong gIII expression, resulting in progeny that can then infect incoming E. coli. No GOI activity (or a weak one, red) results in poor progeny production, becoming washed away from the lagoon at a larger rate (alongside bacterial cells). The system is designed to run for hundreds of generations without human intervention and result in the evolution of the GOI towards the desired activity.
Source: Packer and Liu [2]; Badran and Liu [12]; Carlson et al. [14]; Dickinson et al. [15].
A key parameter in PACE is the E. coli flowrate, which should exceed their doubling time but be slower than the phage life cycle, allowing only phage to replicate in the lagoon (on average). Consequently, only phage accumulates mutations, whereas E. coli are physically prevented from doing so. High rates of mutation on the phage (and E. coli) genome are driven by a mutator plasmid (MP) that is carried by the E. coli cells and induced in the lagoon for error‐prone M13 replication. The latest version of the MP is able to drive potent mutagenesis at >10−3 s.p.b. by combining the effects of six different mutagenesis drivers [38].
Esvelt et al. first demonstrated proof of concept by evolving T7 RNA polymerase (RNAP) to initiate transcription from new promoter sequences [33]. pIII expression was bottlenecked at the level of transcription by encoding promoter sequences unrecognized by wild‐type (wt) T7 RNAP (or any E. coli RNAPs), thus driving the selection to favor T7 RNAP variants that are able to efficiently recognize the new promoters. After eight days and 200 “rounds” of PACE, new T7 RNAPs emerged that could transcribe from the distant T3 RNAP promoter as efficiently as wt T7 RNAP does from its cognate promoter [33]. Similarly, T7 RNAP variants that efficiently initiate transcription with ATP or CTP, instead of GTP, were evolved. Since that landmark study, the ability to couple T7 RNAP activity to PACE has been exploited in a number of ways, ranging from basic adaptation studies to selections for split T7 RNAP [14,15,35–38].
In principle, PACE is applicable for the evolution of any biomolecular function that can be linked to pIII expression; and in just a few years since its inception, this has been realized in a wide range of applications beyond RNAP evolution. A notable example is the evolution of new DNA binding domains. Hubbard et al. employed the classic one‐hybrid selection with PACE to evolve transcription activator‐like effector nucleases (TALENs) with broadly improved DNA cleavage specificity [34]. Although TALENs are highly promising for gene editing, their major limitation is that they require the 5′ nucleotide of the target sequence to be T [39]. New TALEs (TALENs without the fused nuclease) were evolved with PACE by fusing the DNA binding domain of the canonical CBX8‐targeting TALE to the ω subunit of E. coli RNAP. The PACE system was designed to include the TALE target sequence upstream of gIII. TALEs that successfully bind the target DNA recruit holoenzyme RNAP around the ω subunit, resulting in subsequent pIII expression. With this TALE selection, the identity of the target sequence can be custom‐tailored, in this case, to encode noncanonical 5′ nucleotides. After using an additional negative selection (see below) that inhibited variants with promiscuous substrate specificity, Hubbard et al. were able to evolve TALE variants that displayed two‐ to fourfold increases in specificity for 5′ A, 5′ C, or 5′ G versus 5′ T, relative to wt TALE.
The one‐hybrid PACE format was also used for overcoming one of Cas9's main limitations, restricted protospacer adjacent motif (PAM) compatibility. This time, Hu et al. fused a catalytically dead variant of Streptococcus pyogenes Cas9 (dCas9) to the ω subunit of E. coli RNAP [40]. Then, the authors cleverly fed the lagoon with a mixture of host E. coli cells bearing a library of target sequences that covers all 64 possible PAM sequences, to select for broadened PAM compatibility. After PACE, several variants were isolated that could efficiently recognize NG, GAA, and GAT as PAMs. Upon restoration of nuclease catalytic activity to these evolved dCas9 variants, the authors remarkably found that one of them, xCas9, exhibited greater DNA specificity than wt Cas9, even with its newly‐gained broad PAM compatibility. This result challenges the widely‐held assumption that there must be a trade‐off between editing specificity and PAM compatibility and suggests that Cas9 can be improved through laboratory evolution to meet the most demanding challenges of CRISPR‐Cas9 applications.
Another important form of PACE is its use with two‐hybrid selection for the evolution of high‐affinity protein‐binders [31]. In the bacterial two‐hybrid system, the ω subunit of E. coli RNAP is fused to a protein of interest, which is recruited to DNA through its interaction with a target protein. This target protein is fused to a DNA binding domain that localizes the complex at its cognate sequence encoded upstream of a reporter gene. If the protein of interest binds the target protein, then the RNAP holoenzyme can reconstitute around the ω subunit and drive expression of the downstream reporter. Badran et al. adapted this system for PACE using gIII as the reporter. After extensive optimization, Badran et al. were able to use this PACE format to evolve the insecticidal protein, Bacillus thuringiensis δ‐endotoxin (Bt toxin) Cry1Ac, to bind and inhibit a new receptor in the gut of the insect pest Trichoplusia ni (TnCAD) [31]. Although wt Cry1Ac did not detectably bind TnCAD, the evolved variants were able to bind with nM affinity. Significantly, this strategy could overcome widespread Bt toxin resistance, which primarily occurs through mutational changes that inhibit binding to the native receptor of wt Cry1Ac. Badran et al. demonstrated this by showing that evolved Cry1Ac is highly potent at killing T. ni that are resistant to wt Cry1Ac. An exciting possibility for the future would be to evolve TnCAD to resist the new Cry1Ac variant, and then iterate this cycle in a study of molecular co‐evolution.
Additional positive selections developed for PACE have enabled evolution of proteases that are drug resistant [32] or have altered substrate specificities [41], aminoacyl‐tRNA synthetases (aaRSs) that can accept noncanonical amino acids [42], and protein variants with improved soluble expression [43]. Negative selections are also compatible with PACE, and are useful in cases where it is desirable to evolve high specificity towards the target substrate and restrict promiscuity towards others (especially the native substrate). This can be achieved by introducing a dominant negative allele of pIII, pIII‐neg, that inhibits phage propagation [14]. The expression of pIII‐neg can then be linked to the unwanted activity (e.g. recognition of the T7 promoter by T7 RNAP) for negative selection. (This strategy was successfully employed during TALEN and aaRS evolution.) Selection stringency and mutation rate are also important determinants of PACE outcomes and can be titrated [14,35]. Lastly, we note that the Isalan lab developed a system related to PACE that accommodates the evolution of multiple genes, starting from combinatorial libraries. With this system, they were able to evolve a panel of orthogonal dual promoter‐transcription factor pairs that were used to make multi‐input logic gates [44,45].
Clearly, PACE is a powerful method for continuous protein evolution, but as noted early in this chapter, it is not an entirely in vivo system. Rather, M13 serves as a biological carrier of the GOI from one E. coli host cell to the next, with a given cell serving as a host of error‐prone replication just once (on average). This ingenious design circumvents the challenges of in vivo mutational targeting. Since mutagenesis is induced in the lagoon, where E. coli briefly reside without doubling, mutation rates can be elevated entirely through untargeted mechanisms (and temporarily induced to be as high as desired), without consideration for replication of the E. coli genome. Even if E. coli cells stochastically replicate in the lagoon and become a source of cheater mutations (e.g. constitutive gIII expression), the flow rate ensures that any progeny are quickly diluted out. What's left in the lagoon is a population of M13 that selectively undergoes error‐prone replication. In effect, targeting of mutations to the phage genome containing the GOI is complete, as the host E. coli is constantly replaced.
PACE also achieves durable mutagenesis by enforcing continuity. Replication of GOIs is intrinsically coupled to mutagenesis, through error‐prone replication of the M13 genome. Any phage that escapes mutagenesis through a mutation in the phage genome's origin of replication, for example, must do so at the expense of being replicated. Only variants that continue to accumulate mutations can survive and propagate. And since E. coli cells do not persist long enough in the lagoon to evolve, the mutation rate experienced by phage remains unchanged. The durability of PACE is best evidenced by the long mutational trajectories traversed during evolution experiments, which have yielded protein variants with up to 16 mutations [46].
However, because PACE is not entirely in vivo, it suffers two major limitations. First, it requires continuous propagation of phage in a population of freshly diluted E. coli cells, which has been achieved thus far with a chemostat or turbidostat setup. This greatly limits the throughput and accessibility of PACE experiments, typically to fewer than ten replicates or experiments especially when different selection environments are desired across replicates. Second, PACE is restricted to selections that are linked to phage propagation. This precludes selections for in vivo phenotypes like tolerance or metabolism, as well as cell‐based selections like fluorescence‐activated cell sorting (FACS) or droplet sorting. These limitations motivate the need for continuous directed evolution systems that operate entirely in vivo.
1.4 Systems That Allow In Vivo Continuous Directed Evolution
1.4.1 Targeted Mutagenesis in E. coli with Error‐Prone DNA Polymerase I
The first system that was able to perform continuous targeted mutagenesis in vivo was published in 2000 by Fabret et al. [47]. It was designed based on the developments in understanding the mechanism of ColE1 plasmid replication in E. coli. For plasmids that contain a ColE1 origin of replication, DNA polymerase (DNAP) I (Pol I) is responsible for elongating from the RNA primer that initiates replication at the origin. Pol I will extend for about 400–2000 bp, after which DNAP III (Pol III), responsible for bulk DNA replication in E. coli, replaces Pol I [48]. When using a genome‐encoded proofreading‐deficient Pol I, genes that were cloned near the ColE1 origin experienced a 6‐ to 20‐fold higher degree of mutagenesis over genes at more remote areas in the plasmid, showing targeting. The system's components were further combined with mismatch repair mutants to raise the mutation rate on GOIs yet another 20‐ to 40‐fold, although significant increases in genomic mutation rates of at least several hundred‐fold were observed. As a proof of concept, the authors evolved dominant negative variants of LacI that would outcompete a genomically‐encoded wt LacI in binding its cognate operator, LacO. After 30 generations, LacI mutants that caused complete abolishment of wt LacI's binding to LacO were isolated. These variants were altered in their DNA binding domain but still formed tetramers with wt LacI, thereby abolishing LacI's repression at LacO.
Further improvement of the Pol I/ColE1 system was demonstrated in 2003 (Figure 1.3a) [46,49]. Camps et al. modified the system to express the ep Pol I from a plasmid with a Pol I‐independent origin of replication. Then, they used a host E. coli strain (J2000) whose genomically‐encoded wt Pol I was temperature sensitive (ts) [49]. At restrictive temperatures, the ts Pol I becomes inactive such that only the ep Pol I acts, preventing the high‐fidelity ts Pol I from competing for replication at the ColE1 origin. Based on prior studies of Pol I from the same lab [50], Camps et al. engineered a Pol I variant that was exceptionally error‐prone, leading to mutation rates as high as 8.1 × 10−4 s.p.b at the GOI when the ts Pol I was inactivated. Mutagenesis expanded to about 3 kb from the ColEI origin and was evenly distributed within this region, albeit with certain biases in mutational preference. As a proof of concept experiment, Camps et al. demonstrated that their system could be used to evolve enzymes with diverged function by generating TEM‐1 β‐lactamase mutants that were able to hydrolyze a third‐generation lactam antibiotic, aztreonam.
Figure 1.3 Targeted mutagenesis in E. coli with error‐prone DNA polymerase I. (a) An ep version of Pol I is expressed from a plasmid whose replication is driven by a non‐ColE1 origin of replication (ori). The GOI is placed on the target plasmid near the ColE1 ori and thus targeted for mutagenesis. After 1–3 kb of ep replication, Pol III replaces Pol I to replicate the remainder of the plasmid with high fidelity. The genomic allele of POL I is temperature sensitive, such that enhanced mutagenesis can be induced by growth at the restrictive temperature.
Source: Alexander et al. [49]; Camps et al. [46].
(b) The EvolvR system is composed of a CRISPR‐guided nickase that nicks the target GOI, fused to ep Pol I that performs nick translation.
The ep Pol I/ColE1 system has subsequently been applied in a handful of additional directed evolution experiments. For example, Koch et al. used the system to prepare a library of terminal alkane hydroxylases with the aim of evolving variants that can oxidize butane [51]. Although they only used the system for the preparation of mutant libraries (i.e. as a mutator strain) and not for continuous evolution involving serial passaging under prolonged selection conditions, they demonstrated that one can create large libraries of GOI variants directly in vivo. In another application, an M13 phagemid with a ColE1 origin was made to encode LuxR and infect E. coli harboring the ep Pol I [52]. LuxR is a transcriptional activator and drove the transcription of an antibiotic resistance gene (β‐lactamase) controlled by the lux promoter in the E. coli. Through several cycles of infecting fresh E. coli, antibiotic selection, lysis of E. coli, and phage isolation, LuxR evolved a 17‐fold higher binding affinity to the lux promoter sequence.
While the ep Pol I/ColE1 system approaches ICE, it is limited by off‐target mutagenesis and low durability. Because Pol I is responsible for Okazaki fragment mending throughout the genome and also participates in DNA repair [53], expressing an ep Pol I causes substantial mutagenesis genome‐wide. Targeting of mutations to the GOI does occur – owing to the ColEI origin, the limited role of Pol I in lagging strand replication, and special growth conditions optimized to time ep Pol I action with growth phases where genome replication activity is low – but is maximally only ∼400‐fold. Therefore, when highly ep Pol Is are used, it is possible that off‐target mutagenesis will lower the fitness of the cell, causing fixation of suppressor mutations that abrogate the activity of ep Pol I. Still, the Pol I/ColE1 system represents a landmark development that encouraged the field to pursue new strategies for realizing ICE.
Perhaps the closest conceptual descendant of the ep Pol I/ColE1 system is a new E. coli continuous evolution system called EvolvR, which uses CRISPR‐guided ep DNAPs to continuously target mutations to GOIs (Figure 1.3b). Rather than rely on the natural targeting of Pol I to ColE1, Halperin et al. [54] fused ep Pol I variants (and other DNAPs) to a nickase Cas9 (nCas9) that would serve two purposes. First, nCas9 would bring the ep Pol I to any GOI encoded on a plasmid or the genome using a guide RNA (gRNA). Second, the nCas9 would nick the target strand, creating a free 3′‐OH substrate from which the ep Pol I could extend. Once nCas9 releases the nicked product, it is believed that ep Pol I then latches on and carries out error‐prone extension from the nick. This highly clever idea was demonstrated in E. coli with a number of ep Pol I variants spanning different mutation rates and activities, as well as with a moderately ep Phi29 DNAP with high processivity. Using the most mutagenic ep Pol I, Halperin et al. measured a mutation rate approaching 10−2 s.p.b. (a 7.7 million‐fold elevation compared to wt cells) at the first nucleotide 3′ of the nCas9‐induced nick. While this extreme mutation rate quickly dropped when moving away from the nick, other Pol I and Phi29 DNAP variants with moderate error rates could achieve mutagenesis windows up to 350 bp. With these characteristics and with the potential to use multiple gRNAs to simultaneously target multiple parts of a gene, EvolvR could readily and efficiently generate sequence diversity on a GOI in vivo to support continuous evolution. Indeed, in a proof of principle experiment, Halperin et al. used EvolvR to rapidly evolve spectinomycin resistance by targeting mutagenesis to the rpsE gene and found new resistance mutations that were previously unknown.
Future studies and improvements on EvolvR will clarify how well it drives ICE for prolonged periods of time, needed to traverse long mutational pathways. Durability may be difficult in the current architecture, because the mutation rate is maximal at nucleotides within the target region of the gRNA, which if mutated, will reduce the ability of the system to continue inducing mutagenesis. Since the GOI can still be replicated (by high‐fidelity host systems) in the absence of EvolvR function, this may result in the fixation of partially adapted GOI mutants that stop mutating, leading to premature cessation of evolution. In addition, EvolvR still has off‐target elevations in mutation rate, presumably because ep Pol I or Phi29 can participate in genomic replication and/or because Cas9 has off‐target binding. Strategies that use more processive ep DNAPs with no activity in normal genome replication and alternative CRISPR systems that nick outside the critical regions for gRNA targeting may overcome potential issues of targeting and durability. We also anticipate that this system should readily transfer to cell‐types other than E. coli. Therefore, EvolvR is a highly promising new system for ICE with enormous potential, especially for the multiplexed evolution of genes at their endogenous genomic loci rather than on a plasmid.
1.4.2 Yeast Systems That Do Not Use Engineered DNA Polymerases for Mutagenesis
The first demonstration of continuous targeted mutagenesis in vivo in yeast was published in 2013 under the name TaGTEAM (Figure 1.4a), which stands for targeting glycosylases to embedded arrays for mutagenesis [55]. In TaGTEAM, mutagenesis at the GOI is initiated by recruiting a DNA glycosylase, which normally functions as the first step in the base excision repair (BER) pathway responsible for removing chemically altered DNA bases [56]. The authors adopted the yeast 3‐methyladenine glycosylase, Mag1p, and fused it to the tet repressor (tetR) that binds a 19‐bp operator sequence, tetO. By introducing a non‐recombinogenic tetO array (with each tetO site separated by 10–30 bp of random sequence), the tetR‐Mag1p fusion could be targeted to GOIs in the chromosome or plasmid. It is presumed that tetR‐Mag1p targeting generates a build‐up of unprocessed abasic sites at target loci, leading to replication fork stalling and recruitment of ep translesion polymerases [57]. This faulty repair can lead to both point mutations and frameshifts. To test their system for its ability to generate mutagenesis at a GOI, Finney‐Manchester et al. introduced a 240X tetO array upstream of a URA3 auxotrophic marker in a region of chromosome 1 that does not contain nearby essential genes. The distance between the tetO array and the marker was titrated to assess the size of the area subjected to mutagenesis. The presence of tetR‐Mag1p resulted in a >800‐fold increase in mutation rate spanning a 10 kb region. However, the off‐target mutation rate was also increased 40‐fold in the absence of the array, indicating genome‐wide mutagenesis by tetR‐Mag1p. No direct applications of the system have been published to date, but this mutagenic strategy was important for opening new avenues of thought in the field.
Figure 1.4 Yeast systems for targeted mutagenesis of GOIs. (a) TaGTEAM is achieved by fusing the yeast 3‐methyladenine DNA glycosylase, Mag1p, to a tetR DNA‐binding domain. Upon expression of the fusion from an inducible galactose promoter, the 20 kb region that is proximal to the tetO array experiences a high degree of mutagenesis. (b) In ICE, the GOI is cloned into an inducible Ty1 retrotransposon in the genome. The ICE cycle begins with inducible transcription of the retroelement followed by ep reverse transcription driven by Ty1's encoded rt. The cycle ends upon re‐integration of the mutated cDNA into the genome.
Source: Based on Crook et al. [58].
ICE is another notable example of continuous evolution in yeast (Figure 1.4b), introduced in 2016 [58]. ICE adopts a strategy for DNA diversification that is based on the mutagenic properties of the Ty1 retrotransposon element. A GOI is cloned into the Ty1 cassette, which then gets transcribed into an RNA. Next, the RNA is reverse transcribed to form cDNA and reintegrated into the chromosome [59]. The mutagenic properties of the system stem from Ty1's self‐encoded reverse transcriptase (rt), which introduces mutations at a rate of ∼2.5 × 10−5 to ∼1.5 × 10−4 per base per retrotransposition event [58,60], thus allowing rapid mutagenesis of Ty1 and its embedded GOI. However, since mutagenesis depends on retrotransposition and the retrotransposition rate of Ty1 with a GOI inserted is low, the high mutation rate of Ty1's rt is only occasionally experienced on the GOI. Therefore, the authors carried out a series of experiments to increase the retrotransposition rate. By fine‐tuning various parameters including the cargo's promoter strength, host genotype (i.e. deletions of certain host genes), cell density, temperature, initiator methionine tRNA expression (which acts to prime Ty1 replication), and inclusion of terminators, the authors were able to significantly increase retrotransposition rate. Altogether, the optimization process reached a mutation rate capable of generating up to 1.6 × 107 distinct mutants of a GOI per round per liter cultured [58]. Crook et al. then used ICE in three independent experiments to test the system's ability to evolve genetic material. In the first demonstration, URA3 was evolved for increased resistance to 5‐fluoroorotic acid (5‐FOA); in the second example, the Spt15p global transcription regulator was evolved to confer a complex cellular phenotype of butanol resistance; and in the third example, a multigene pathway spanning 4.6 kb and containing two enzymes and a regulatory region was evolved for increased xylose catabolism. Additional experiments will clarify the extent to which ICE continuously mutates GOIs, as the ability for Ty1 elements to semi‐randomly spread throughout the yeast genome [61,62] could potentially complicate analysis, reduce mutational accumulation for the GOI, and diffuse the target of evolution. These issues could potentially be solved by somehow limiting Ty1 integration to a single location in the genome, turning the retrotransposon into a “retrocisposon,” and then increasing the “retrocisposition” rate to access high levels of diversification. In fact, the ability to achieve “retrocisposition” would also be important for reaching continuous evolution in other systems based on retroelement‐mediated mutagenesis, such as a recently reported bacterial approach for in vivo genome editing and evolution [63]. Nevertheless, ICE is an important example of continuous evolution in yeast.
1.4.3 Somatic Hypermutation as a Means for Targeted Mutagenesis of GOIs
Some groups have harnessed one of nature's built‐in mechanisms for generating targeted DNA diversity, somatic hypermutation (SHM). In SHM, B cells create point mutations in their immunoglobulins (Igs) to drive antibody affinity maturation [64]. The enzyme responsible for SHM is Activation Induced cytidine Deaminase (AID), which deaminates cytidine (C) to generate uridine (U). This triggers various mismatch repair mechanisms resulting in a mutation rate of ∼10−3 s.p.b. at Ig loci [65]. Several researchers have successfully hijacked this natural mechanism for diversifying and evolving non‐antibody proteins. In 2001, Bachl et al. set the stage for SHM‐based protein directed evolution [66]. They demonstrated a high rate of reversion of a premature stop codon in a green fluorescent protein (GFP) cloned into a hypermutator B cell line (18–81) that expresses endogenous AID. They concluded that elevated reversion rates depended on AID and were rate limited by transcriptional levels of the target gene, in agreement with previous findings on SHM mechanisms [67,68]. In 2004, Wang et al. applied SHM to the directed evolution of an entire open reading frame [69] by integrating a single copy of red fluorescent protein (RFP
