This is an open access article published under an ACS AuthorChoice License, which permits
`copying and redistribution of the article or any adaptations for non-commercial purposes.
`
`Research Article
`
`pubs.acs.org/synthbio
`
`A Highly Characterized Yeast Toolkit for Modular, Multipart
`Assembly
`Michael E. Lee,†,‡,⊥ William C. DeLoache,†,‡,⊥ Bernardo Cervantes,†,⊥ and John E. Dueber*,†,§,∥
`†
`Department of Bioengineering, University of California, Berkeley, California 94720, United States
`‡
`The UC Berkeley & UCSF Graduate Program in Bioengineering, Berkeley, California 94720, United States
`§QB3: California Institute for Quantitative Biological Research, Berkeley, California 94720, United States
`∥
`Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States
`*S Supporting Information
`
`ABSTRACT: Saccharomyces cerevisiae is an increasingly attractive host for synthetic biology
`because of its long history in industrial fermentations. However, until recently, most synthetic
`biology systems have focused on bacteria. While there is a wealth of resources and literature
`about the biology of yeast, it can be daunting to navigate and extract the tools needed for
`engineering applications. Here we present a versatile engineering platform for yeast, which
`contains both a rapid, modular assembly method and a basic set of characterized parts. This
`platform provides a framework in which to create new designs, as well as data on promoters,
`terminators, degradation tags, and copy number to inform those designs. Additionally, we
`describe genome-editing tools for making modifications directly to the yeast chromosomes,
`which we find preferable to plasmids due to reduced variability in expression. With this
`toolkit, we strive to simplify the process of engineering yeast by standardizing the physical
`manipulations and suggesting best practices that together will enable more straightforward translation of materials and data from
`one group to another. Additionally, by relieving researchers of the burden of technical details, they can focus on higher-level
`aspects of experimental design.
`KEYWORDS: yeast, toolkit, characterized parts, golden gate, MoClo
`
`S ynthetic biology is driven by the desire to engineer novel
`
`biological functions that push the boundaries of what can
`be accomplished within living cells. Unfortunately, the potential
`power of the cell also brings with it a level of complexity that
`makes engineering biological
`systems extremely difficult.
`Synthetic biologists have sought ways to abstract the layers of
`complexity into components with predictable interactions,
`making it more feasible to undertake large engineering projects.
`Despite these efforts, the inner workings of the cell continue to
`elude understanding, and while certain elements can be highly
`predictable, the system behavior as a whole is difficult to
`anticipate. These challenges have led to an additional, and
`equally important, aspect to synthetic biology: rapid prototyp-
`ing.1−4 Because manipulations to the cell often lead to
`unexpected results, progress is best made by rapidly iterating
`through highly parallelized experiments to explore a wide
`space.5,6 It
`parameter
`is
`the combination of
`these two
`principlespredictable parts and rapid prototypingthat
`give synthetic biologists the ability to approach difficult
`problems in energy,7,8 agriculture,9 and human health.10−12
`Saccharomyces cerevisiae is growing in popularity as a chassis for
`synthetic biology due to its powerful genetic tools,13−15
`extensively studied biology,16−19 and long history of industrial
`applications.20−22 In this work, we present a synthetic biology
`toolkit for engineering yeast that simplifies and accelerates
`experimentation in this important model organism.
`
`Abstraction is a fundamental principle in any engineering
`discipline. It allows an engineer to focus on an individual
`component with the assurance that it will interface correctly
`with other components, both existing and future. When applied
`to synthetic biology, abstraction typically refers to the level of
`complexity of the DNA that is being built or introduced into
`cells. “Parts” are often thought of as one of the most basic DNA
`sequence elements that can be assigned a function. For
`example, a coding sequence, a transcriptional terminator, and
`an origin of replication could all be described as parts. Although
`these parts can be broken down furtherthey contain, among
`other things, a start and stop codon, a hairpin, and a protein
`binding site, respectivelythe benefit of abstraction is the
`ability to ignore those lower level details and work with a part
`based solely on its reported function. Extensive efforts by others
`in the field have contributed to the Registry of Standard
`Biological Parts, a catalog of DNA sequences and character-
`ization data that continues to grow each year (http://
`partsregistry.org).23 The Registry, however, is notably biased
`toward working in bacterial systems, particularly Escherichia coli,
`and with growing interest in yeast as a synthetic biology host, it
`is becoming apparent that the field needs a more extensive set
`of
`standard yeast parts. For
`this
`toolkit, we collected,
`
`Received: December 19, 2014
`Published: April 14, 2015
`
`© 2015 American Chemical Society
`
`975
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`Downloaded via 73.229.7.22 on July 4, 2024 at 15:58:50 (UTC).
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 1 of 12
`
`

`

`ACS Synthetic Biology
`
`constructed, and characterized a starter set of useful parts to lay
`the foundation for a standardized engineering platform, and
`these parts are available from Addgene.
`Prototyping is a more necessary step in synthetic biology
`than in other engineering fields, as synthetic biologists lack the
`ability to accurately predict behavior, even of devices made
`from parts of known function.24−26 When working in fast-
`growing cells such as yeast, cloning is often the bottleneck step
`in an experimental cycle. The lag between having a DNA design
`and actually obtaining the physical DNA is far too long to
`support a robust prototyping workflow. The solution that many
`groups have developed is standardization of cloning.27−32 For
`example, the BioBrick standard (and its relatives) defines a set
`of restriction enzyme sites that are used to flank each part in a
`vector.27,30 When those restriction enzymes are used to join
`two parts, the junction contains an assembly “scar”, and the
`resulting plasmid reconstitutes the sites external to the newly
`combined parts (an idempotent operation). This enables an
`endless number of cycles of pairwise assembly. More recently,
`Golden Gate assembly based methods have increased in
`popularity due to the added flexibility provided by the use of
`Type IIs
`restriction enzymes, which cut outside their
`recognition sequence and provide unique cohesive ends to
`enable directional, multi-insert, one-pot cloning.33 One example
`is the MoClo (modular cloning) system, which categorizes
`parts as “types” based on their function and location in a
`completed device (e.g., promoter types or coding sequence
`types) and designates particular overhangs that flank each type,
`allowing all parts of a particular type to be interchangeable.32 In
`this work, we adapted the MoClo strategy specifically to build
`yeast expression devices. The major advantage of using a
`standardized system such as MoClo is that once parts are
`constructed, they are immediately available for incorporation
`into devices and no longer require synthesis of oligonucleo-
`tides, PCR amplification and purification, or verification by
`sequencing. This allows us to construct from parts, a plasmid
`carrying multiple gene expression devices in as little as 2 days.
`
`■ RESULTS AND DISCUSSION
`Definition of an Assembly Standard for Yeast. Our
`standard for assembling DNA for expression in yeast is a
`bottom-up hierarchical approach to DNA construction (Figure
`1). A description of the assembly scheme, part types, and
`overhang sequences are discussed briefly here and in more
`detail in the Supporting Information. For brevity, Golden Gate
`assemblies using either BsaI or BsmBI are referred to as “BsaI
`assembly” and “BsmBI assembly”.
`for
`Our workflow for assembling complex plasmids
`expressing multiple genes in yeast has multiple steps that
`correspond to our abstraction layers. First, source DNA is
`obtained through PCR, synthesis or another user-preferred
`method. That source DNA is “domesticated” via BsmBI
`assembly into a universal entry vector, resulting in a “part”
`plasmid. Part plasmids come in different Types, numbered 1
`through 8 (with some optional subtypes). Each part Type is
`defined by the sequences of the upstream and downstream
`flanking overhangs generated when digested by BsaI. All parts
`of a particular Type are interchangeable, which lends the
`system well to combinatorial experiments. Part plasmids are
`joined in a BsaI assembly to form a “cassette” plasmid that, in
`most cases,
`is used to express a single gene in yeast (a
`transcriptional unit, TU, comprised of a promoter, coding
`sequence, and terminator). These cassettes can optionally be
`
`Research Article
`
`Figure 1. Standardized, hierarchical assembly strategy based on
`MoClo. (A) Source DNA is obtained via PCR, DNA synthesis, or
`oligonucleotides, then assembled using BsmBI into a part plasmid
`entry vector. (B) Part plasmids of a particular Type have unique
`upstream and downstream BsaI-generated overhangs. All part plasmids
`of the same Type are therefore interchangeable. Plasmids at this stage
`typically confer chloramphenicol resistance in E. coli. One part plasmid
`of each Type is assembled using BsaI to form a cassette plasmid. (C)
`Cassette plasmids contain a complete transcriptional unit (TU), and
`can be transformed directly into yeast. Plasmids at this stage typically
`confer ampicillin resistance in E. coli. Alternatively, cassette plasmids
`can be further assembled using BsmBI to form a multigene plasmid.
`(D) Multigene plasmids contain multiple TUs, the order of which is
`dictated by the Assembly Connector parts used to flank the individual
`cassettes. Plasmids at this stage typically confer kanamycin resistance
`in E. coli.
`
`joined in a final BsmBI assembly to form “multi-gene” plasmids
`that, as the name suggests, are used to simultaneously express
`multiple genes. The multigene assembly is enabled by the use
`of Assembly Connectors (Type 1 and 5) that, in similar fashion
`to each part plasmid’s unique BsaI overhangs, contain unique
`BsmBI overhangs that flank each cassette. At each round of
`assembly,
`the antibiotic selection is changed to minimize
`background (typically, chloramphenicol → ampicillin →
`kanamycin). Using this workflow, we can construct a multigene
`plasmid from PCR templates in only 3 days. This construction
`time is typically reduced to only 2 days, since, in most cases, the
`final multigene plasmids are built from existing parts.
`There are many benefits to the standard we defined, which
`should prove useful to synthetic biologists with a wide range of
`needs. First,
`the cloning protocols are extremely simple,
`requiring no PCR amplification or purification steps after the
`initial part creation. Second, the standardized Golden Gate
`assemblies are highly robust. It was previously shown that for a
`10-part assembly with an optimized set of overhangs, 97% of
`isolated transformants contained a correctly assembled
`
`976
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 2 of 12
`
`

`

`ACS Synthetic Biology
`
`Research Article
`
`Figure 2. Yeast toolkit starter set of 96 parts and vectors. Note that the eight primary part Types can be further divided into subtypes (e.g., 3a/3b), or
`combined to make composite types (e.g., 234). Each Type has a unique upstream and downstream overhang pair, and a complete cassette can be
`assembled when a complete path can be drawn from left to right (1 to 8). For example, the preassembled integration vector is assembled from
`ConLS′ (1), GFP dropout (234), ConRE′ (5), URA3 (6), URA3 3′ Homology (7), KanR-ColE1 (8a), and URA3 5′ Homology (8b). A
`transcriptional unit (promoter, coding sequence, terminator) can be assembled into this vector, replacing the BsaI-flanked GFP dropout. A set of
`cassettes can also be assembled into this vector, due to the special Assembly Connectors ConLS′ and ConRE′ that have the BsmBI recognition sites
`in the reverse orientation (Supporting Information). The part plasmid entry vector is used for constructing new parts. A table of plasmid names,
`parts, and Types is included in Supporting Table S1.
`
`plasmid.34 We observed comparable efficiencies in this work
`and screened only one transformant for almost all plasmid
`assemblies described here. Because PCR- and oligonucleotide-
`derived point mutations cannot occur after the construction of
`part plasmids, we do not
`typically sequence downstream
`assemblies and instead use simple restriction mapping to verify
`size. Third, our workflow supports a simple method for
`chromosomal
`integration in which plasmids designed for
`integration can be transformed directly after being linearized
`via a NotI digestion.35 Fourth, our design specification includes
`unique restriction enzyme sites that make cassettes both
`BioBrick- and BglBrick-compatible, and multigene plasmids
`BioBrick-compatible. While a variety of
`restriction sites
`
`(BamHI, BbsI, BglII, BsaI, BsmBI, EcoRI, NotI, PstI, SpeI,
`XbaI, and XhoI) have been removed from all parts in the
`toolkit for increased flexibility, only BsaI, BsmBI, and NotI
`must be removed from new parts to conform to the complete
`assembly scheme described here. Finally,
`the Assembly
`Connectors,
`in addition to harboring BsmBI sites, can also
`act as homology sequences for recombination-based cloning,
`such as sequence and ligation-independent cloning (SLIC),36
`Gibson assembly,37,38 ligase cycling reaction (LCR),39 or yeast
`in vivo assembly,40 if those methods are preferred.
`A Toolkit of Yeast Parts. Although an assembly standard
`has some inherent value, its utility is determined in large part by
`the availability of parts. To this end, we have compiled a
`
`977
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 3 of 12
`
`

`

`ACS Synthetic Biology
`
`Research Article
`
`Figure 3. Characterization of promoters. (A) The relative strength of 19 constitutive promoters is consistent across two coding sequences, mRuby2
`and Venus. Three promoters (strong pTDH3, medium pRPL18B, and weak pREV1) that are used throughout this work are highlighted. The
`horizontal and vertical bars represent the range of four biological replicates, and the intersection represents the median value. (inset) A third
`fluorescent protein, mTurquoise2, was also tested, and a larger plot can be found in Supporting Figure S1. (B) The mating-type-specific promoter,
`pMFA1, is only active in the MATa haploid; pMFα2 is only active in MATα haploids; neither promoter is active in the opposite haploid or in the
`diploid. The expression level of pRPL18B in the three strains is shown for reference. The height of the bars represents the median value of four
`biological replicates, and the error bars show the range. (C) Galactose induction of pGAL1 increases expression from background levels up to the
`highest expressing constitutive promoter, pTDH3. All solid line data were collected from a Δgal2 strain. The dashed line shows a much more
`sensitive response to galactose induction in a wild type strain. Points represent the median value of four biological replicates, and error bars show the
`range.
`
`collection of 96 parts compatible with this standard for
`efficiently engineering yeast strains (Figure 2 and Supporting
`Table S1). This starter collection contains an assortment of
`promoters,
`terminators, fluorescent proteins, peptide tags,
`selectable markers, and origins of replication, as well as a part
`entry vector into which new parts can be cloned. Additionally,
`we have included sequences targeting chromosomal
`loci for
`integration, and genome-editing tools for introducing double-
`strand breaks to stimulate homologous recombination. Finally,
`rather than provide a large array of different vectors,
`the
`assembly standard enables construction of custom vectors
`directly from parts in the toolkit, and one such vector is
`included as an example (Supporting Information).
`Characterization of Promoters. We have characterized 19
`constitutive promoters,
`two mating-type-specific promoters,
`and two inducible promoters, all cloned from the yeast genome
`(although synthetic promoters41,42 could easily be ported into
`the system as well). The promoters were selected to span a
`wide range of
`transcriptional strengths while minimizing
`variability between growth conditions.43 In general,
`they
`constitute the 700 bp directly upstream of the native start
`codon, although in some cases where another ORF was less
`than 700 bp away, we cloned only the intergenic noncoding
`region. To examine the strength of each promoter, we cloned it
`
`upstream of a fluorescent reporter (mRuby2, Venus, and
`mTurquoise2) and measured bulk fluorescence on a plate
`reader.
`It was previously shown that the strength of constitutive
`promoters cloned from the yeast genome was
`largely
`the downstream coding sequence,44 an
`independent of
`important distinction between controlling expression in
`bacteria and yeast. This held true for the 19 constitutive
`promoters characterized in this work (which include some
`overlap with Lee et al., 2013)44 (Figure 3A and Supporting
`Figure S1). The promoters span a range of up to 3 orders of
`magnitude, and there are also some promoters that have very
`similar expression strengths, allowing them to be interchanged
`so as
`to reduce the risk of undesired homologous
`recombination in multigene plasmids due to repeated
`sequences. Although we only tested these promoters in one
`type of media, the majority of native yeast promoters have been
`shown to maintain their
`relative expression strengths in
`different growth conditions, although the absolute strengths
`may change.45
`It is sometimes useful to have genes under dynamic control,
`and for this we provide two tools: mating-type-specific and
`inducible promoters. We tested pMFA1 and pMFα2 and found
`that they have very close to background levels of fluorescence
`
`978
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 4 of 12
`
`

`

`ACS Synthetic Biology
`
`Research Article
`
`Figure 4. Characterization of terminators. Six terminators were cloned behind three fluorescent proteins, each driven by three promoters. The
`relative expression levels for this set of terminators are largely independent of the coding sequence and the promoter. The height of the bars
`represents the median value of four biological replicates, and the error bars show the range.
`
`in both the opposite mating-type haploid and diploid strains
`and a 6- to 10-fold induction in the appropriate haploid (Figure
`3B). We also tested pGAL1 in varying concentrations of
`galactose and observed a 100-fold induction (Figure 3C).
`Although the promoter can be used in wild-type strains, the
`response is very sensitive to low concentrations of galactose; a
`strain with the GAL2 transporter knocked out should be used
`for more graded control overexpression.46 Finally, we tested
`pCUP1 in varying concentrations of copper(II) sulfate
`(CuSO4) and observed a 55-fold induction (Supporting Figure
`S2). This promoter exhibits leaky expression under basal
`conditions, with approximately 7-fold fluorescence over back-
`ground when CuSO4 is not added to the media. This may be
`due in part to the CuSO4 that is present at 250 nM in the yeast
`nitrogen base commonly used to make defined media.
`Characterization of Terminators. The impact of different
`transcriptional
`terminators on gene expression can vary
`considerably, and could provide a secondary mode of control
`to complement the promoters. However, for simplicity, we
`opted in this toolkit
`to provide terminators that yielded
`approximately the same expression output. Using expression
`data from the whole yeast genome,43 we selected six of the
`most highly expressed genes and cloned the 225 bp
`immediately downstream of the stop codon. We assembled
`these terminators with each of our three fluorescent reporters
`and each using three promoters. The largest difference in
`expression we observed between terminators for a given
`promoter and fluorescent protein was 3.6-fold (Figure 4). In
`general, the fold-changes produced by different promoters were
`greater than those effected by the terminators, but this was not
`always the case. If applications are sensitive to small
`fold-
`changes of expression, we advise characterizing individual
`promoter-terminator pairs to ensure that the desired levels of
`expression are obtained.
`Protein Degradation Tags. In addition to controlling
`transcript
`levels, protein levels can be tuned by fusing
`degradation tags to the N-terminus. We have included three
`such tags of varying strengths, Ubi-M (weak), Ubi-Y (medium),
`and Ubi-R (strong), which can be used to adjust the rate of
`protein turnover.47 We fused these tags to the N-terminus of
`mRuby2, and expressed them using a strong, moderate, and
`weak promoter, pTDH3, pRPL18B, and pREV1, respectively.
`The strong degradation tag (Ubi-R) resulted in no detectable
`fluorescence at any expression level, while the medium strength
`degradation tag (Ubi-Y) resulted in detectable levels of
`fluorescence at only the highest expression level (Figure 5).
`Copy Number, Gene Expression, and Single-Cell
`Variability. When engineering yeast strains expressing multi-
`
`Figure 5. Protein degradation tags. Three N-terminal degradation tags
`were fused to mRuby2 and expressed using three different promoters.
`Steady-state fluorescence levels are dependent on the difference
`between the strength of
`the promoter and the strength of
`the
`degradation tag. The height of the bars represents the median value of
`six biological replicates, and the error bars show the range.
`
`to consider the
`it is important
`ple heterologous proteins,
`relative expression of
`those proteins. As described above,
`protein levels can be controlled by changing promoters,
`terminators, or degradation rates. However, another important
`consideration is the copy number of the gene(s). Typically, one
`of three systems is used to express genes in yeast: single-copy
`integrations into the chromosome,
`low-copy CEN6/ARS4
`plasmids, and high-copy 2micron plasmids. One could easily
`assume that the differences in copy number simply titrate gene
`expression accordingly, but we observed that there are subtle,
`but important, effects that could influence the decision to use
`one system over another.
`We cloned cassettes expressing either mRuby2 or Venus
`under
`strong, moderate, and weak promoters (pTDH3,
`pRPL18B, and pREV1, respectively). Versions of these cassettes
`were made for each of the three copy numbers. Finally, each of
`the nine possible combinations of the three promoters and two
`genes were either assembled in tandem onto a single
`chromosomal
`locus/plasmid or kept separate in two loci/
`plasmids. We measured bulk fluorescence of both fluorescent
`proteins to compare protein expression levels of
`the cell
`populations at
`the three copy-numbers (Figure 6 and
`Supporting Figure S3).
`the different
`In the chromosomally integrated strains,
`promoter combinations fill out the points of a regular grid, as
`expected. In the low-copy CEN6/ARS4 plasmid system, the
`absolute fluorescence is generally higher compared to the
`
`979
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 5 of 12
`
`

`

`ACS Synthetic Biology
`
`Research Article
`
`machinery in the cell is limiting and that having more copies of
`the DNA has little effect on increasing expression; i.e., the
`average fluorescence of cells with the strongest promoter is
`similar between low- and high-copy plasmids. In the single-
`plasmid, 2micron system, not only is the grid compressed, but
`also it appears that high expression of one gene seems to reduce
`the expression of the second gene (Supporting Figure S3). On
`the basis of flow cytometry, there appears to be a bimodal
`distribution for some of these populations (Figure 7C, e.g.,
`pTDH3-mRuby2/pRPL18B-Venus), which is consistent with
`previous studies comparing the distribution of expression in
`2micron and chromosomally integrated systems.48 Interest-
`ingly,
`this effect
`is not nearly as pronounced in the
`chromosome or on the low-copy plasmid. It is unclear why
`this would be specific to the high-copy plasmid. On the basis of
`these data, we believe that use of high-copy 2micron plasmids
`should generally be avoided, since the highest expression levels
`accessible by them are very nearly accessible by low-copy
`CEN6/ARS4 plasmids, and low-copy plasmids give greater
`access to lower expression, and in general have less erratic
`expression patterns.
`Another parameter we examined was cell-to-cell variability in
`the relative expression of two genes. While it has been shown
`that strains expressing fluorescent proteins from chromoso-
`mally integrated genes display much tighter distributions
`compared to those expressing from 2micron plasmids,48 we
`were curious about any additional effects of propagating one
`versus multiple plasmids. We took the same cultures used to
`measure bulk fluorescence and ran them on a flow cytometer to
`measure single-cell fluorescence of the two fluorescent proteins
`(Figure 7). As expected, the single-cell measurements revealed
`that the variability in fluorescence increased considerably when
`moving from the chromosome to either a low-copy or high-
`copy plasmid, indicating that the precise copy number of these
`plasmids is not tightly regulated. When expressed from a single
`locus/plasmid, the expression of the two fluorescent proteins
`
`Figure 6. Effect of copy number on gene expression. Three promoters
`(pTDH3, pRPL18B, and pREV1) drive two fluorescent proteins
`(mRuby2 and Venus) in all nine possible combinations. These nine
`combinations are integrated into the chromosome (blue), expressed
`from a low-copy plasmid (green), and expressed from a high-copy
`plasmid (red). The translucent, shaded boxes show the range of
`expression spanned by each respective copy number. For lower
`strength promoters, increasing copy number gives higher fluorescence;
`but for the strongest promoter, there is a much smaller difference
`between the low- and high-copy plasmids. Each gene is integrated in a
`separate locus or expressed from a separate plasmid. The horizontal
`and vertical bars represent the range of four biological replicates, and
`the intersection represents the median value.
`
`the range
`Interestingly,
`chromosome, again, as expected.
`between the highest and lowest expression is actually slightly
`greater in the CEN6/ARS4 plasmid system. Compared to low-
`copy plasmids,
`the high-copy 2micron plasmids
`showed
`considerably more irregular expression patterns. In the two-
`plasmid, 2micron system, the grid is preserved, but compressed
`at higher expression levels, suggesting that some expression
`
`Figure 7. Effect of copy number on single-cell gene expression. The same strains expressing mRuby2 and Venus that measured for bulk fluorescence
`in Figure 6 and Supporting Figure S3 were run on a flow cytometer: chromosomally integrated in a single locus (A) or two loci (D); on a single (B)
`or two (E) low-copy plasmids; on a single (C) or two (F) high-copy plasmids. As copy number increased, the variability of expression also increased.
`For all single-locus strains, the expression of the two fluorescent proteins was well correlated, suggesting that copy number is the main contributor to
`variation in expression. When expressed from two plasmids, correlation between fluorescent proteins is lost, suggesting that the copy number of each
`plasmid is independent of the other. Fluorescence in each channel was normalized for cell size by dividing by forward-scatter. Dot plots for each
`sample represent 5000−10000 events.
`
`980
`
`DOI: 10.1021/sb500366v
`ACS Synth. Biol. 2015, 4, 975−986
`
`LCY Biotechnology Holding, Inc.
`Ex. 1067
`Page 6 of 12
`
`

`

`ACS Synthetic Biology
`
`was well correlated, as evidenced by the distribution of each
`strain along the diagonal. This result suggests that DNA copy
`number is the primary source of added variation in plasmid-
`based expression systems, a model which is further supported
`by the data from two loci/plasmids. Strains expressing mRuby2
`and Venus from two separate loci in the chromosome showed
`distributions that are nearly identical
`to the single locus,
`chromosomal strains. In contrast, the low-copy and high-copy
`plasmids
`lose their
`tight correlation between the two
`fluorescent proteins when the two genes are expressed from
`separate plasmids. Thus, not only is the copy number of a
`plasmid highly variable, the relative copy numbers of two
`plasmids in the same cell are not well correlated. Therefore, we
`would recommend that genes be integrated into the
`chromosome whenever possible. If, however, higher expression
`than can be attained from the chromosome is required, use of a
`low-copy plasmid is preferred, and all genes should be
`expressed from the same plasmid rather
`than split onto
`multiple plasmids. Accordingly,
`the assembly standard we
`provide here accommodates the facile assembly of up to six
`genes on a single plasmid or in a single chromosomal locus,
`with more possible if additional Assembly Connectors are
`designed.
`High-Efficiency Integrations into the Chromosome.
`Yeast is well suited to chromosomal modification due to its
`efficient homologous recombination machinery. This allows for
`site-specific integration of DNA into the chromosome by
`simply transforming linear DNA flanked by sequences
`homologous to the target
`locus. However, compared to
`plasmid transformation,
`integrations usually result in almost
`an order of magnitude reduction in colony counts, which is one
`reason why the use of plasmids is often preferred. Given the
`desire for chromosomal
`integrations described earlier,
`it is
`evident that a higher efficiency method for integrating into the
`chromosome is necessary, particularly when working with large
`libraries. Fortunately,
`it was previously shown that
`trans-
`formation efficiency could be dramatically improved by using a
`homing endonuclease to generate a double-strand break in the
`chromosome and stimulate recombination.49 More recently,
`the clustered regularly interspaced short palindromic repeats
`(CRISPR) and CRISPR-associated (Cas) system has been used
`for similar purposes.50−52 We tested both systems to directly
`compare their effects on chromosomal library construction.
`First, we prepared an experimental strain by integrating a
`“landing pad” using conventional homologous recombination
`of linear DNA (Supporting Figure S4A). This landing pad
`contained an I-SceI recognition site; the I-SceI recognition site
`conveniently contains an NGG protospacer adjacent motif
`(PAM) close to the I-SceI cutting site, and we added an extra
`10 bp upstream of the site to create a 20 bp targeting sequence
`for a single guide RNA (sgRNA); we also included a partial
`URA3 coding sequence and terminator
`that by itself
`is
`nonfunctional.
`Next, we designed the repair DNA we were integrating to
`contain a Venus-expressing cassette and a HIS3-expressing
`cassette, flanked by homology to the sequence upstream of the
`landing pad and to the partial URA3 marker. Thus, when the
`DNA integrated successfully, the cells would be prototrophic
`for histidine and uracil, and they would be fluorescent. If the
`DNA integrated off-target, the cells would be prototrophic for
`histidine and fluorescent, but they would remain auxotrophic
`for uracil, allowing us to measure the rate of off-target
`integration by selecting on 5-fluoroorotic acid (5-FOA).
`
`Research Article
`
`Finally, we compared the efficiency of integration when the
`repair DNA was transformed unassisted, with a transient
`“cutter” plasmid (we did not select for it) expressing Cas9 and
`an sgRNA, or with a transient cutter plasmid expressing I-SceI.
`As a control for cell competency, we transformed a circular
`version of the repair DNA that also contained an origin of
`replication (Figure 8 and Supporting Figure S4B). Compared
`
`Figure 8. High-efficiency integration into the chromosome. Integra-
`tion of
`linear DNA into the chromosome by homologous
`recombination yields 6-fold fewer colonies (compare shaded bars).
`Adding in a Cas9 or I-SceI improves transformation efficiency by 2-
`fold and 5-fold, respectively (compare white bars to unassisted
`integration). Linearizing the Cas9 or I-SceI expression plasmid prior to
`transformation further improves transf