Expanding the Toolkit for Metabolic Engineering Yao Zong (Andy) Ng Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Graduate School of Arts and Sciences COLUMBIA UNIVERSITY 2016 © 2016 Yao Zong (Andy) Ng All rights reserved ABSTRACT Expanding the Toolkit for Metabolic Engineering Yao Zong (Andy) Ng The essence of metabolic engineering is the modification of microbes for the overproduction of useful compounds. These cellular factories are increasingly recognized as an environmentally-friendly and cost-effective way to convert inexpensive and renewable feedstocks into products, compared to traditional chemical synthesis from petrochemicals. The products span the spectrum of specialty, fine or bulk chemicals, with uses such as pharmaceuticals, nutraceuticals, flavors and fragrances, agrochemicals, biofuels and building blocks for other compounds. However, the process of metabolic engineering can be long and expensive, primarily due to technological hurdles, our incomplete understanding of biology, as well as redundancies and limitations built into the natural program of living cells. Combinatorial or directed evolution approaches can enable us to make progress even without a full understanding of the cell, and can also lead to the discovery of new knowledge. This thesis is focused on addressing the technological bottlenecks in the directed evolution cycle, specifically de novo DNA assembly to generate strain libraries and small molecule product screens and selections. In Chapter 1, we begin by examining the origins of the field of metabolic engineering. We review the classic “design–build–test–analyze” (DBTA) metabolic engineering cycle and the different strategies that have been employed to engineer cell metabolism, namely constructive and inverse metabolic engineering. This is followed by a discussion of how combinatorial methods that draw on both approaches can improve the speed, cost and efficiency of metabolic engineering. Next, we discuss how enabling technologies such as DNA assembly methods and genome-wide mutagenesis techniques have served as a driver of developments in the field, as well as some current limitations in the area of screens and selections for natural products. Finally, we offer some perspectives on how the field of metabolic engineering could be advanced in the future. In order to re-engineer the metabolism of a cell, its genetic program has to be modified, either by mutating its DNA or introducing foreign genes and pathways. Furthermore, because metabolic engineering is a combinatorial optimization problem, it could be advantageous to create a library of strain variants in parallel for testing. However, a key bottleneck is the lack of tools to build long multi-gene pathways and libraries, especially in genetically intractable host cells. In Chapter 2, we report the de novo assembly of the ~ 100 kb meridamycin Type I modular polyketide synthase (mPKS) pathway using Reiterative Recombination. The mPKS pathways are difficult to manipulate due to their long, GC-rich and highly repetitive sequences. Our strategy was to harness the power of yeast recombination to assemble and engineer the pathway in the chromosome of the yeast Saccharomyces cerevisiae, followed by transfer of the pathway into a heterologous Streptomyces lividans host for polyketide production. We also developed FLIP, which is based on site-specific recombination, for the efficient recovery of the pathway/library from the yeast chromosome onto shuttle plasmids. Lastly, we describe the construction of a promoter library in yeast using CRISPR in order to optimize the production of the polyketide meridamycin, a compound with promising neuroprotective activities. Our approach can also be potentially applied for combinatorial biosynthesis to produce novel analogs. Although there has been much progress in the development of techniques for the de novo assembly of pathways, libraries and even whole genomes, as well as large-scale mutagenesis approaches, it is widely held that the current bottleneck to applying combinatorial approaches to metabolic engineering is a lack of general and high-throughput screens and selections for the product. State-of-the-art methods such as liquid/gas chromatography – mass spectrometry (LC/GC-MS) can detect a diverse range of compounds, but are too low-throughput (< 102 samples/day) to screen libraries of ≥ 103. Other methods that rely on detecting a product’s fluorescence, color, bioactivity or reactivity cannot be generally applied to the majority of compounds. Thus in Chapter 3, we established a high-throughput screen (104 – 105 samples/day) based on the Fluorescence Polarization (FP) assay, which in theory can be adapted as a screen for a wide range of small molecules and natural products. The FP assay works by measuring the changes in fluorescence polarization values when the target molecule competes with a fluorescently-tagged reporter molecule for binding to a common receptor protein. As a proof-of- principle, we applied the FP assay to screen a UV-mutagenized library of Streptomyces tsukubaensis and identified strains that produced higher titers of FK506 (tacrolimus), an immunosuppressant drug. We also demonstrate the biosynthesis of FK506 in 96-well, deep well, microtiter plates, which is compatible with the throughput of the FP assay performed in 384-well microtiter plates. In contrast to a screen, genetic selections enable much higher throughput (> 108 samples/day limited by culture volume), since the population of cells can be assayed together in a one pot growth selection. However, this requires that the production of the target compound confer a growth advantage on the cell. In Chapter 4, we describe how the yeast three-hybrid system (Y3H) can be adapted to detect a product by displacement of one arm of a small molecule dimerizer that bridges two receptor protein fusions that reconstitute an active transcription factor, thus changing the transcription levels of a reporter gene. We report Y3H systems capable of detecting trimethoprim (TMP), meridamycin and FK506, and linking detection to the transcription of a selectable marker. This sets the stage to use the Y3H for genetic selections, if both the production and detection functions can be performed in the same cell. Although metabolic engineering can potentially be used to biosynthesize a large portfolio of compounds, there are limitations to relying completely on this approach. For instance, the final products could be toxic to the cell, or the enzymes for a reaction step could be non-existent in nature, or more often than not, the biosynthetic process from feedstock to final compound is not economically viable or would take too long to develop. Thus, many groups have developed semi-synthetic approaches that combine biosynthetic steps with chemical synthesis to generate compounds. In Chapter 5, we describe novel semi-synthetic strategies leading to Ambrox, a key component of fragrances, and the tocotrienols, key components of Vitamin E that possess anti- oxidant and anti-cancer properties, as well as their derivatives. Both strategies rely on engineering the yeast S. cerevisiae or the bacterium E. coli to produce linear terpenoid intermediates followed by modification with synthetic chemistry. To derive Ambrox, E. coli was engineered to produce the unnatural polyene intermediate Z,E-farnesol, which was cyclized by the halonium-based reagents BDSB (bromodiethylsulfonium bromopentachloroantimonate) and IDSI (iododiethylsulfonium iodopentachloroantimonate) to form halo-Ambrox, which could be readily converted to 9-epi-Ambrox and its analogs. For the tocotrienols, yeast was engineered using a triple enzyme fusion to produce E,E,E-geranylgeraniol, which was converted in one step to the tocotrienols using a novel Lewis acid – Phase Transfer Catalyst system (Type I Adanve acid). Table of Contents List of Figures v List of Tables ix List of Abbreviations x Acknowledgements xiv Dedication xvii Chapter 1 Strategies and Enabling Technologies for Metabolic Engineering 1 1.0 Chapter outlook 2 1.1 Metabolic engineering – origins 3 1.2 Classic strategies for metabolic engineering 6 1.2.1 The design–build–test–analyze cycle 8 1.2.2 Constructive metabolic engineering 10 1.2.3 Inverse metabolic engineering 16 1.3 Combinatorial directed evolution strategies and tools for metabolic engineering 18 1.3.1 Mutagenesis to create single gene libraries 20 1.3.2 Mutagenesis to create pathway libraries 22 1.3.3 Genome-wide mutagenesis 30 1.3.4 Screening and selection technologies for small molecule products 40 1.4 Conclusions and future perspectives 46 1.5 References 49 i Chapter 2 De novo Assembly and Promoter Engineering of the ~ 100 kb Type I Meridamycin Polyketide Biosynthetic Pathway 64 2.0 Chapter outlook 65 2.1 Introduction 66 2.2 Results 68 2.2.1 ~ 100 kb meridamycin PKS pathway assembly by Reiterative Recombination 68 2.2.2 Sequence editing and promoter library construction using CRISPR 74 2.2.3 Pathway recovery from the chromosome onto a shuttle vector by FLIP 77 2.2.4 Meridamycin production and promoter library screening 90 2.3 Discussion 92 2.4 Experimental methods 99 2.5 Assembly PCR, check PCR and CRISPR tables 108 2.6 Strains and plasmids 119 2.7 References 124 Chapter 3 Fluorescence Polarization Assay for High-throughput Small Molecule Screens 132 3.0 Chapter outlook 133 3.1 Introduction 134 3.2 Results 136 3.2.1 High-throughput FP assay to detect FK506 136 ii 3.2.2 Production of FK506 in 96-well microtiter plate