Leveraging modern DNA assembly techniques for rapid, markerless genome modification

Leveraging modern DNA assembly techniques for rapid, markerless genome modification The ability to alter the genomic material of a prokaryotic cell is necessary for experiments designed to define the biology of the organism. In addition, the production of biomolecules may be significantly improved by application of engineered pro- karyotic host cells. Furthermore, in the age of synthetic biology, speed and efficiency are key factors when choosing a method for genome alteration. To address these needs, we have developed a method for modification of the Escherichia coli genome named FAST-GE for Fast Assembly-mediated Scarless Targeted Genome Editing. Traditional cloning steps such as plasmid transformation, propagation and isolation were eliminated. Instead, we developed a DNA assembly-based ap- proach for generating scarless strain modifications, which may include point mutations, deletions and gene replacements, within 48 h after the receipt of polymerase chain reaction primers. The protocol uses established, but optimized, genome modification components such as I-SceI endonuclease to improve recombination efficiency and SacB as a counter-selection mechanism. All DNA-encoded components are assembled into a single allele-exchange vector named pDEL. We were able to rapidly modify the genomes of both E. coli B and K-12 strains with high efficiency. In principle, the method may be applied to other prokaryotic organisms capable of circular dsDNA uptake and homologous recombination. Keywords: Genome modification; DNA assembly; I-SceI; synthetic biology Introduction Escherichia coli fall into two categories. The first set of methods is driven by homologous recombination and is reliant on pro- The ability to manipulate the genetic material of a cell has pro- teinssuchasendogenousRecA[15, 16]. A primary advantage ven to be invaluable for understanding the organism’s biology. of methods that utilize the cell’s own recombination machin- The basis of our understanding of many essential biological pro- ery is that helper plasmids are not necessary. Homologous re- cesses has been enabled by the ability to perform gene deletion combination results in a single crossover event between the and complementation studies [1–4]. Furthermore, countless incoming DNA and the chromosome resulting in a tandem ar- strains have been specifically engineered to facilitate the pro- rangement of the wild-type gene and mutant gene. Then, a duction of valuable proteins requiring modified cellular envi- second crossover event potentially leaves the mutation of in- ronments or helper factors [5–9]. terest in the chromosome. These protocols (labelled as ‘Classic A number of the genome modification methods used today allelic-exchange’ in Fig. 1) generally do not leave a scar in the were developed decades ago and have had only minor updates genome upon removal of the selection marker. However, these due to new techniques becoming available [10–16]. Most of the methods are often cumbersome, taking as much as a week for current methods for modifying common lab organisms such as Received: 17 June 2016; Revised: 6 September 2016. Accepted: 26 September 2016 V The Author 2016. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/ licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 2| Tikh and Samuelson Figure 1: Timeline comparison of the most common methods for genome modification in E. coli (h ¼ hours). During Red-mediated protocols, helper plasmid 1 contains the red exo, beta and gamma genes, which facilitate the incorporation of linear DNA into the chromosome. Helper plasmid two traditionally contains Flp recombinase to help remove the antibiotic resistance marker from the chromosome. When utilizing the CRISPR/Cas9, the first helper plasmid also contains the red exo, beta and gamma genes in addition to Cas9. The second helper plasmid contains the guide RNA, used to eliminate any WT cells which did not undergo desired recombination. Regardless of the method of construction, all modifications should be sequenced in order to verify the presence of the desired modification. a single modification. A second set of commonly used methods DNA to be eliminated by homologous recombination. Due to its is derived from the phage lambda recombination system, most ability to target multiple loci utilizing different guide RNAs, the famously described by Datsenko and Wanner, and utilizes the CRISPR/Cas9 system holds promise as a way to multiplex ge- expression of exogenous phage proteins to mediate the recom- nome editing; however, currently the ability to simultaneously bination steps [10, 11, 17, 18]. The lambda Red system allows introduce changes into multiple locations of the E. coli chromo- for faster integration and modification of a desired locus using some is limited. For an overview of current genome engineering a linear PCR product that contains short segments of homol- methods, see Fig. 1. ogy, but this method requires the presence of a helper plasmid Given the various drawbacks of current methods, we driving the expression of the phage genes and is less efficient sought to push the limits of speed, efficiency and versatility by for the replacement of large pieces of DNA. Though more rapid leveraging recent advances in molecular biology. By utilizing a than classic methods, optimized lambda Red techniques still state-of-the-art DNA assembly approach to create sealed, cir- require approximately 4 days from start to finish, counting the cular DNA in vitro, we are able to completely eliminate the time it takes to initially transform and subsequently cure the need to propagate plasmids in order to perform a genomic helper plasmids [10]. The lambda Red system has also been ap- modification in E. coli. Our work utilized a modified version of plied for DNA oligonucleotide-mediated genome modification; Gibson assembly, however, any DNA assembly technique able however, this method will not be discussed further as it is pri- to efficiently generate high levels of fully circularized DNA marily limited to the generation of point mutations or codon should be applicable in our protocol. Consequently, the de- substitutions [19]. scribed protocol allowed us to perform scarless modifications Recently, the CRISPR/Cas9 system has been adapted for of the E. coli genome in approximately 48 h from the reception modifying genomes in multiple prokaryotic organisms [18, 20– of PCR primers, significantly reducing the time relative to any 23]. Unfortunately, to take full advantage of this system, the or- other published method. This rapid turn-around allows for ganism must be highly recombinogenic, which E. coli is not [22]. fast creation of complex engineered E. coli hosts tailored for In general, the application of the CRISPR/Cas9 system in E. coli specific applications, potentially saving weeks of effort if mul- requires the use of the lambda Red system for the initial allele tiple modifications are required. Finally, given the simplicity exchange step and then the Cas9 nuclease is employed to elimi- of the protocol, we believe it should be possible to apply this nate the WT allele. This method is effective but the overall com- method to a wide range of prokaryotic organisms, so long as bined use of lambda proteins and Cas9 is subject to the same the host is capable of efficient circular DNA uptake, homolo- drawbacks of other Red-mediated methods [18, 20–22]. The ex- gous recombination, and the promoter driving the genes of the ception is a recent report that employed a Cas9 nickase to initi- pDEL vector are functional in the target organism. The method ate the generation of large genomic deletions [23]. The Cas9 described herein is named FAST-GE, an acronym chosen to in- nickase method is dependent upon naturally occurring (or pre- dicate Fast Assembly-mediated Scarless Targeted Genome engineered) repetitive DNA elements in order for intervening Editing. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 3 Materials and methods Bacterial strains, plasmids and enzymes E. coli strain NEB10-beta (New England BioLabs, Ipswich, MA) was used for all cloning steps to create the pDEL vector. Plasmids containing the R6K origin were replicated in BW23473 [24]. E. coli strains NEB Express, T7 Express and ER2744 (New England BioLabs, Ipswich, MA) [25] were utilized in genome modification experiments. All enzymes were from New England BioLabs (Ipswich, MA). Chemicals were purchased from Sigma- Aldrich (St. Louis, MO). Primers and gBlock synthetic DNAs were synthesized by IDT (Coralville, IA). The DasherGFP gene was obtained from DNA2.0 (Menlo Park, CA). Kanamycin and ampicillin were used at 40 mg/ml and 100 mg/ml, respectively, un- less otherwise stated. The rhaBAD and lac promoters were in- duced with 0.2% rhamnose and 0.5 mM IPTG, respectively. Counter-selection plates contained 5% (w/v) sucrose on plates containing 5 g/l of yeast extract, 10 g/l of tryptone and 7.5 g/l of agar, in addition to inducers as described above. Routine cell growth was performed at 37 C in lysogeny broth medium sup- plemented with 0.1% glucose. Colony PCR screens were per- formed using Quick-Load Taq 2X Master Mix, while Q5 Hot Start High Fidelity 2X Master Mix (New England BioLabs, Ipswich, MA) was used to PCR-amplify genomic DNA for sequence verification. Construction of pDEL The open reading frame for the sacB gene was amplified from Figure 2: Schematic overview of the genome modification process. In order to pRE112 [26] and inserted into pMAL-c5X using NEBuilder HiFi modify E. coli chromosomal region B, flanking regions with at least 500 bp of ho- DNA assembly to create pMAL-sacB. Subsequently, the sacB mology on each side of B (boxes A and C) are generated. PCR is the easiest gene was amplified with a lac promoter. Low-level basal expres- method to generate this linear DNA. To enable DNA assembly, the flanking re- sion of sacB is not harmful in the absence of sucrose, as such gions should also contain small homology regions of 18–24 nucleotides to each the lac repressor is not required. If the strain encodes lacI, then other as well as to the linearized deletion cassette (amplified from pDEL). If re- IPTG is necessary during the resolution step. Plasmid pKD4 was gion B is being mutated or replaced, as opposed to deleted, an additional frag- ment containing that modification must be included and will be assembled used as the source of the kanamycin resistance gene and asso- between A and C. Linear DNA fragments are subsequently assembled into a cir- ciated promoter [11]. The R6K origin of replication was amplified V R cular construct using NEBuilder HiFi DNA assembly, transformed into electro- from pCD13PKS [27]. The region containing the I-SceI endonu- competent cells, and the initial integration event is selected for by kanamycin clease site as well as the I-SceI gene under the control of rhaBAD resistance. The integration location is verified by PCR and colonies containing promoter was ordered as a gBlock ; the sequence is shown in the desired insertion are subcultured into media containing the inducers for I- Supplementary Table S1. The rhamnose promoter was chosen SceI and SacB, and subsequently on counter-selection plates with the same in- ducers. Expression of I-SceI promotes homologous recombination in response to as it should be tightly repressed in the absence of the inducer, a double-strand break in the genome and SacB is used to select against cells that but will also generate sufficient levels of I-SceI transcripts in the failed to remove the integration cassette. The final colonies are screened by PCR majority of E. coli strains. To generate the pDEL vector, the fol- for the presence of the desired genome modification. V R lowing PCR fragments were assembled using an NEBuilder HiFi DNA assembly reaction: the R6K origin, kanamycin resistance V R modification was gene replacement, the PCR product of the new gene, I-SceI gBlock and the Plac-sacB fragment. The final circular gene fragment was assembled between the upstream and down- vector is shown in Supplementary Fig. S1 and the entire pDEL stream regions of homology. Typical assembly reactions con- vector sequence is included in the Supplementary Material. tained 250–300 ng of total DNA, at manufacturer’s recommended ratios in a total volume of 10 ml. Sequencing of individual clones Genome modification experiments generated from each of the HiFi DNA assemblies confirmed a very low error rate. Results are shown in Supplementary Table S2. For genome modification experiments, electrocompetent E. coli For transformation, 2 ml of assembled DNA was mixed with were prepared as follows: cells were grown in 3 ml cultures of electrocompetent cells and electroporated in a 1 mm cuvette ac- LB until late exponential phase (OD600  0.8). Cells were har- cording to manufacturer’s instructions (BioRad). Electroporated vested by centrifugation and washed 3 times with 500 ml of ice cells were resuspended in 950 ml of SOC and allowed to recover cold 10% glycerol. Finally, cells were resuspended in 50 ml of 10% at 37 C for 1.5–2 h. Following the recovery step, 100 and 900 ml glycerol. aliquots were plated on fresh LB-Kan plates and incubated at Genome modification constructs were made by assembling 37 C overnight. Colonies from the LB-Kan plate were picked, upstream and downstream homology regions (Fig. 2,boxes Aand transferred into a liquid LB-Kan culture and simultaneously C, respectively), created using site-specific primers containing re- screened by colony PCR to confirm the location of the initial re- gions of microhomology to each other as well as to a PCR product combination event. Cultures containing the desired integration of pDEL amplified with primers located between the R6K origin of were allowed to reach early log phase (approx. OD ¼ 0.2), at replication and the I-SceI gene. In cases when the desired 600 Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 4| Tikh and Samuelson which point they were diluted 1:500 into fresh LB containing Rapid, targeted genome modification 100 mM IPTG and 0.2% w/v rhamnose but lacking kanamycin. Three changes were made to non-essential chromosomal re- Post induction, cultures were allowed to grow for an additional gions: deletion of the lac operon (Fig. 3A), a point mutation in the 3 h at 37 C, after which they were plated on counter-selection lacZ gene (Supplementary Fig. S1A), and insertion of the T7 RNA plates containing 5% sucrose, 100 mM IPTG and 0.2% rhamnose. polymerase gene (Fig. 4). Changes were introduced into both E. Following an overnight incubation at 37 C, colonies from the coli B and K-12 backgrounds. These modifications were chosen to counter-selection plates were picked and grown in LB to be ana- demonstrate the capability of the FAST-GE method to easily gen- lysed by PCR with locus-specific primers for removal of the inte- erate a large deletion, a point mutation and an insertion, respec- gration cassette. In order to confirm the desired changes, the tively. The detailed illustration of the construct used to generate PCR products were column purified and sequenced using the the deletion of the lac operon in ER2523 (an E. coli B derivative) is Sanger method. shown in Fig. 3A. Electrocompetent ER2523 cells were trans- formed with 150 ng of DNA from the assembly reaction. This pro- Results tocol was sufficient to result in several correctly integrated transformants (Fig. 3B). Two different primer pairs are used to Composition of the insertion cassette identify strains with the pDEL construct integrated correctly at When designing the pDEL vector, several critical factors were the desired locus. All eight colonies analysed contained a pDEL considered. First, the vector size was maintained as compact as integration at the lac locus when combining the results of the F1- possible in order to optimize amplification yield and fidelity. R1 and F2-R2 PCR analyses. Each primer pair has two potential This led us to critically evaluate the components currently used products depending on whether recombination occurred via the 0 0 with other genome modification approaches. We determined 5 flank homology region or the 3 flank homology region. In each that at the minimum an antibiotic selection marker and a coun- diagnostic PCR reaction, the extension time is set to amplify the ter-selection marker were required. The frequency of homolo- shorter of the two possible PCR products. We found that this ex- gous recombination in wild-type E. coli is low; however, perimental approach was acceptable as shown in Fig. 3B. recombination can be increased locally by the presence of a Recombinant number one was chosen for the resolution step. double-strand break. Thus, an I-SceI restriction site and the cor- After reaching noticeable turbidity grown in the presence of responding endonuclease gene were added to the construct to kanamycin (OD ¼ approx. 0.2), the culture from recombinant encourage homologous recombination by inducing a double- number one was diluted 1:500 into fresh LB medium lacking strand break. Additionally, we customized the promoters of the kanamycin but containing rhamnose and IPTG, for induction of sacB and I-SceI genes, to improve the overall efficiency of the I-SceI and sacB, respectively. After 2 h of incubation in antibiotic- protocol. The final composition of the deletion cassette is free medium, dilutions were plated on sucrose agar plates. shown in Fig. 2. Following counter-selection on sucrose, final resolution and re- Kanamycin resistance was chosen as the primary selection moval of the integration cassette was highly effective. Table 1 marker as growth on kanamycin plates has proven to be a con- presents a summary of the process for deletion of the lac operon. sistent indicator of single-copy genome integration [11, 16]. For The removal of the counter-selection cassette relies upon homol- counter-selection SacB was chosen as its activity and toxicity in ogous recombination and depending on the recombination site E. coli has been well documented over the past several decades the process can lead to one of two outcomes. First, the entire in- [28, 29]. The sacB gene originates from B. subtilis and encodes a tegration cassette may be removed such that the original genome periplasmic levansucrase. The exact mechanism of toxicity by sequence is restored (henceforth referred to as WT sequence). SacB is still not well understood, but it is thought that periplas- Alternatively, the integration cassette may be removed resulting mic SacB creates large levan polymers when cells are grown in in the desired genomic modification (Fig. 3A). Blue colony colour the presence of sucrose [29, 30]. In order to optimize sacB ex- (X-gal conversion) was used as an indicator of lac operon function pression and periplasmic localization, we replaced the native and Fig. 3C shows that 9 of 16 resolved clones had regained b-ga- Bacillus promoter with a lac promoter. Counter-selection on su- lactosidase activity, suggesting resolution to WT sequence. PCR crose proved to be very robust as a large majority of surviving analysis of the same clones showed that six of the recombinants colonies were free of the sacB gene (Supplementary Table S3). had the desired deletion of the lac operon. Several of the resulting Inclusion of the I-SceI homing endonuclease as well as the 1.3 kb F1-R2 PCR products were sequenced to find that no unin- cognate recognition site was based on several recent papers tended mutations were introduced. demonstrating the increased recombination frequency during The described method of genome engineering is not biased genome alteration by the generation of a unique double-strand according to the modification type. To demonstrate the ability of break in the E. coli chromosome by I-SceI. In order to augment I- the FAST-GE method to introduce large insertions in the E. coli SceI cleavage efficiency, two terminators were introduced im- genome, the T7 RNA polymerase (T7 gene 1) was inserted into mediately upstream of the I-SceI cut site as active transcription the chromosome of MC1061, a common E. coli K-12 strain. T7 through the recognition sequence was recently reported to gene 1 may be inserted into the lac locus to facilitate IPTG- lower I-SceI cleavage efficiency [10]. inducible expression of the T7 RNA polymerase for the purpose In addition to the components actively involved in the re- of recombinant protein expression. Accordingly, we chose to re- combination process, we chose to include the R6K origin of rep- create the lacZ-T7gene1 operon fusion (approx. 3 kb), as found in lication on the pDEL plasmid. The R6K origin is unique in that it T7 Express cells, between the yahF and mhpT genes (Fig. 4A). The requires the presence of the pir gene product for the plasmid to integration and counter-selection steps were followed by colony replicate; common E. coli strains lack this gene. The conditional PCR analysis as described above (Fig. 4). As with other experi- origin was included as an alternative strategy, in cases where ments, we were able to obtain desired clones in 48 h from the be- plasmid isolation is a more practical first step. In our experi- ginning of the process. Using a similar procedure, we were also ence, the DNA assembly products were successfully trans- able to replace a single nucleotide in order to create an E462A ac- formed and integrated directly in all strains tested. tive site substitution within LacZ (Supplementary Fig. S1). Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 5 Figure 3: Deletion of the lac operon. (A) Overview of the genomic locus around the lac operon in strain ER2523 as well as the composition of the deletion cassette. The scheme demonstrates the two potential integration outcomes as well as the location of primer pairs F1-R1 and F2-R2 used to determine the location of the integration event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a representative subset of PCR products obtained with the F1-R1 and F2-R2 primer pairs. Expected sizes for the F1-R1 PCR products are either 1.6 kb if the crossover event occurred in the mhpR gene or 8.5 kb if the first crossover event occurred in the cynX gene. The expected sizes for the PCR products from the F2-R2 primer pair are either 7 kb if the crossover event occurred in the mhpR gene or 1.7 kb if the first crossover event occurred in the cynX gene. A DNA molecular ladder (Quick-load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1-8 served as a template for both F1-R1 and F2-R2 PCR reactions. (C) An agarose gel of PCR results utilizing the primer pair F1 and R2 from individual colonies post counter-selection on sucrose. Expected PCR product size from a successful deletion of the lac operon is 1.3 kb. Figure 4: Insertion of the T7 RNA polymerase gene into K-12 strain MC1061. (A) Overview of the genomic locus around the yahF-mhpT locus as well as the composition of the T7 RNA polymerase gene insertion cassette. The scheme demonstrates the two potential integration outcomes after the first crossover event and the location of primer pairs F1-R1 and F2-R2 used to determine the location of the integration event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a representative subset of PCR products obtained with the F1-R1 and F2-R2 primer pairs. The expected sizes for the F1-R1 PCR products are either 4.5 kb if the crossover event occurred in the yahF gene or 2.5 kb if the first crossover event occurred in the mhpT gene. The expected sizes for the PCR products from the F2-R2 primer pair are either 2.6 kb if the crossover event occurred in the yahF gene or 4.7 kb if the first crossover event occurred in the mhpT gene. A DNA molecular ladder (Quick-load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1-6 served as a template for F1-R1 and F2-R2 PCR reactions. (C) An agarose gel showing PCR results from 16 representative colonies post counterselection on sucrose (1–16) from primer pair F1 and R2. The ex- pected PCR product size for successful T7 gene 1 insertion is 5.2 kb. Resolution to WT yahF-mhpT locus yields a PCR product of 2 kb. Lack of a PCR product would suggests that the integration cassette was not successfully removed from the genome. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 6| Tikh and Samuelson Table 1: The efficiency of the genome modification protocol at various loci and modification types Allele Integration efficiency Resolution efficiency Desired modification LacZ E462A 3/8 5/16 3/5 Dlac (B strain) 8/8 13/16 6/13 T7 gene 1 5/6 16/16 7/16 rplI-Dasher 1/4 7/16 0/7 The ratio of isolated colonies reverting to WT sequence as op- When we attempted to fuse a fluorescent protein posed to the desired modification varied depending on the mod- (DasherGFP) to the C-terminus of the ribosome L9 subunit, cor- ification and the integration locus (Table 1). responding to the rplI gene, colonies containing the initial cross- over event were isolated (Fig. 5B). Out of the four colonies tested, only one returned the expected fragment size for the Working with essential genes two diagnostic PCR reactions (Fig. 5B). The colony showing the expected PCR pattern was chosen for resolution. However, none The method described in this article is also beneficial for deter- of the colonies tested after sucrose selection contained the de- mining whether a gene is essential for growth of the bacterium sired gene fusion (Table 1)(Fig. 5C). This led us to conclude that or if a specific mutation destroys function. Since a fully func- this fusion protein is not tolerated by E. coli due to disruption of tional copy of the target gene is maintained during the integra- ribosome function. tion event, integration should always be possible as the present method separates the integration and resolution steps. If all of the resolved colonies in a sufficiently large sample have re- Best practices for genome modification success verted to WT sequence, then one can reasonably assume that the desired modification is not tolerated by the cell (Fig. 5). In When optimizing the FAST-GE protocol, we have noted several contrast, in lambda Red-derived methods, the absence of viable points where special care should be practiced to achieve best re- transformants does not necessarily indicate that a gene is es- sults and we offer several guidelines to improve the likelihood sential or that the modification being attempted is not tolerated of success for first-time users. First, as with other homology- by the cell. An additional advantage of this method is that reso- driven genome modification methods, the size of the flanking lution can be carried out under different culture conditions, homology regions will affect the efficiency of the initial recom- such as varied temperatures or growth medium composition. bination event. Following standard practice for RecA-mediated Thus, one can determine if a gene’s activity is essential for sur- homologous recombination, we recommend designing homol- vival under varied experimental conditions. ogy regions of at least 500 bases on both sides of the desired Figure 5: Working with an essential gene. (A) Overview of the rplI genomic locus as well as the composition of the rplI-dasher fusion cassette. The scheme demonstrates the two potential integration outcomes after the first crossover event and the location of primer pairs F1-R1 and F2-R2, used to determine the location of the integra- tion event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a sample of a PCR product obtained with the F1-R1 and F2-R2 primer pairs. The expected sizes for the F1-R1 PCR products are either 2 kb if the crossover event occurred in the rplI gene or 1.3 kb if the first crossover event occurred in the yjfZ ORF. Similarly, the expected sizes for the PCR products produced by the F2-R2 primer pair are either 1.4 kb if the crossover event occurred in the rplI gene or 2.1 kb if the first crossover event occurred in the yjfZ ORF. A DNA molecular ladder (Quick-Load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1–4 served as a template for both F1-R1 and F2-R2 PCR reactions. (C) An agarose gel of PCR results from 16 representative col- onies post counter-selection on sucrose from primer pair F1 and R2. The expected PCR product size for successful rplI-dasher fusion is 1.6 kb. Resolution to WT rplI se- quence yields a PCR product of 1 kb. Lack of a PCR product suggests that the integration cassette was not successfully removed from the genome. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 7 modification [16]. Secondly, researchers may choose to trans- accessible to anyone familiar with PCR and electroporation- form the assembled genome modification construct into a pir based transformation protocols. strain in parallel with transformation into the desired host. This While we were able to generate all of our desired mutants in optional, extra step ensures that a suicide plasmid is available both K-12 and B derivatives using direct transformation of the to use, in case direct transformation of the assembly reaction assembly reaction, we do understand that some researchers does not yield colonies or if electroporation is not an option for may prefer to use a purified plasmid when working with diffi- transformation. Thirdly, if multiple changes to the genome are cult to transform strains. For those preferring to work with an desired, we suggest that all of the necessary genome modifica- isolated assembly clone we have included the conditional R6K tion constructs be assembled in parallel in a pir host and after origin of replication on the pDEL vector, which allows transfor- the first modification, purified suicide plasmids can be used to mation of the assembly reaction into a pir host for multi-copy perform subsequent modifications, simplifying and expediting replication. generation of the desired strain. In a drive to improve the overall speed of the protocol, versa- Fourthly, when attempting direct transformations, we tility was not compromised as insertions, deletions and point strongly recommend that a high concentration of DNA be used mutations are all equally possible. The only limitation of this for the DNA assembly reaction, up to 250–300 ng of DNA can be method (and many other methods) is that the strain to be modi- assembled in a 10 ml reaction, with subsequent transformation fied must be capable of homologous recombination. For exam- of 2 ml of this reaction. If working with strains known to have ple, typical cloning strains (e.g. DH5a) containing a recA low competence, larger assemblies on the scale of 500–1000 ng mutation are not suitable for modification by this method. of DNA in a 50 ml reaction may be necessary. Then, concentrate Colonies containing the desired initial crossover event are easy the DNA before the transformation step. to identify by colony PCR and require at most the screening of eight colonies, which is on par or better than any other protocols reported to date. Additionally, no difference in the frequency of Discussion integration was apparent regardless of whether the target gene was essential for E. coli growth. This combination of efficiency For decades, genome editing in bacteria has relied upon the use and speed is unique and is of especially high value to re- of plasmids, either as a template for recombination or as the searchers interested in making multiple genome modifications. source of helper proteins to facilitate the recombination event. The reliance on replicative plasmids, requires an initial transfor- mation step to introduce them into the hosts, as well as a curing Author Contributions procedure to remove them after the genome modification proto- col is completed. In some methods, plasmid curing is not I.B.T. and J.C.S. designed research. I.B.T. performed experi- straightforward. In allele exchange methods where the plasmid ments. I.B.T. and J.C.S. analysed data and wrote the paper. is carrying the WT gene after resolution, if plasmid curing fails, then the researcher is left wondering whether the WT gene Acknowledgements might be essential for viability. Some allele exchange methods We would like to thank Elisabeth A. Raleigh and William are particularly time-consuming and may require 2–3 weeks Jack for critical reading of the manuscript and Don Comb from construct design to verification of the final strain [16]. and Jim Ellard for creating and fostering a creative research In this study, we demonstrate the ability to modify a bacterial genome without the need for a replicative plasmid, thus drasti- environment. cally reducing protocol time. This feat is accomplished by leveraging recent advances in DNA assembly technologies in or- Funding der to transform the cells with a non-replicative, circularized piece of DNA. The assembled construct may be immediately Funding for open access charge: New England BioLabs, Inc. transformed into the desired strain without the need for se- Conflict of interest statement. The authors of this manuscript quence verification due to high fidelity of the assembly process. are employees of New England Biolabs, which produces mo- Importantly, only a small number of resolved strains (typically lecular biology reagents, including NEBuilder HiFi DNA fewer than eight) need to be analysed by focused sequencing re- Assembly Master Mix. actions to verify the genome alteration as well as fidelity of the assembly process. Eliminating the requirement for helper plasmid(s) enables Supplementary data the generation of markerless genomic modifications within 48 h from receipt of the necessary primers. Minimum requirements Supplementary data is available at Biology Methods and for this protocol are fragments of DNA sequence flanking the Protocols online. site of the desired modification and linear DNA encoding the pDEL deletion cassette, with all pieces containing small homol- V References ogy regions sufficient to allow assembly via the NEBuilder HiFi DNA assembly method (or other high-efficiency DNA assembly 1. Trinh CT, Li J, Blanch HW et al. Redesigning Escherichia coli me- methods). A high fidelity DNA polymerase should be employed tabolism for anaerobic production of isobutanol. Appl Environ to obtain the necessary DNA fragments and bacterial colonies Microbiol 2011;77:4894–904. may serve as the source of chromosomal template DNA to gen- 2. Baba T, Ara T, Hasegawa M et al. Construction of Escherichia coli erate fragments for the DNA assembly reaction. As a result, K-12 in-frame, single-gene knockout mutants: the Keio collec- time-consuming plasmid or genomic DNA purification proce- tion. Mol Syst Biol 2006;2:2006.0008. doi:10.1038/msb4100050. dures may be bypassed. To our knowledge, this protocol is at 3. Liu H, Yu C, Feng D et al. Production of extracellular fatty acid least 2 days faster, from start to finish, than any other method using engineered Escherichia coli. Microb Cell Fact currently described. In addition, the FAST-GE method is 2016;11:41–54. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 8| Tikh and Samuelson 4. Snyder WB and Silhavy TJ. Enhanced export of beta- 18. Pyne ME, Moo-Young M, Chung DA et al. Coupling the CRISPR/ galactosidase fusion proteins in prlF mutants is Lon depen- Cas9 system to lambda Red recombineering enables simpli- dent. J Bacteriol 1992;174:5661–68. fied chromosomal gene replacement in Escherichia coli. Appl 5. Lobstein J, Emrich CA, Jeans C et al. SHuffle, a novel Escherichia Environ Microbiol 2015;81:5103–14. coli protein expression strain capable of correctly folding disul- 19. Wang HH, Xu G, Vonner AJ et al. Modified bases enable fide bonded proteins in its cytoplasm. Microb Cell Fact 2012;11:56. high-efficiency oligonucleotide-mediated allelic replace- 6. Ajikumar PK, Xiao WH, Tyo KE et al. Isoprenoid pathway opti- ment via mismatch repair evasion. Nucleic Acids Res mization for Taxol precursor overproduction in Escherichia 2011;39:7336–47. coli. Science 2010;330:70–4. 20. Touchon M and Rocha EP. The small, slow and specialized 7. Lee SJ, Lee DY, Kim TY et al. Metabolic engineering of CRISPR and anti-CRISPR of Escherichia and Salmonella. PLoS Escherichia coli for enhanced production of succinic acid, One 2010;5:e11126. based on genome comparison and in silico gene knockout 21. Jiang Y, Chen B, Duan C et al. Multigene editing in the simulation. Appl Environ Microbiol 2005;71:7880–87. Escherichia coli genome via the CRISPR-Cas9 system. Appl 8. Kolisnychenko V, Plunkett G, Herring CD et al. Engineering a Environ Microbiol 2015;81:2506–14. reduced Escherichia coli genome. Genome Res 2002;12:640–47. 22. Jiang W, Bikard D, Cox D et al. RNA-guided editing of bacterial 9. Robichon C, Luo J, Causey TB et al. Engineering Escherichia coli genomes using CRISPR-Cas systems. Nat Biotechnol BL21 (DE3) derivative strains to minimize E. coli protein con- 2013;31:233–39. tamination after purification by immobilized metal affinity 23. Standage-Beier KS, Zhang Q and Wang X. Targeted large- chromatography. Appl Environ Microbiol 2011;77:4634–46. scale deletion of bacterial genomes using CRISPR-Nickases. 10. Kim J, Webb AM, Kershner JP et al. A versatile and highly effi- ACS Synth Biol 2015;4:1217–25. cient method for scarless genome editing in Escherichia coli 24. Haldimann A, Prahalad MK, Fisher SL et al. Altered recogni- and Salmonella enterica. BMC Biotechnol 2014;14:84. tion mutants of the response regulator PhoB: a new genetic 11. Datsenko KA and Wanner BL. One-step inactivation of chro- strategy for studying protein-protein interactions. Proc Natl mosomal genes in Escherichia coli K-12 using PCR products. Acad Sci USA 1996;93:14361–366. Proc Natl Acad Sci USA 2000;97:6640–45. 25. Samuelson JC, Zhu Z and Xu S. The isolation of strand- 12. Yu BJ, Kang KH, Lee JH et al. Rapid and efficient construction specific nicking endonucleases from a randomized SapI ex- of markerless deletions in the Escherichia coli genome. Nucleic pression library. Nucleic Acids Res 2004;32:3661–71. Acids Res 2008;36:e84. 26. Edwards RA, Keller LH and Schifferli DM. Improved allelic ex- 13. Herring CD and Blattner FR. Conditional lethal amber muta- change vectors and their use to analyze 987P fimbria gene ex- tions in essential Escherichia coli genes. J Bacteriol pression. Gene 1998;207:149–57. 2004;186:2673–81. 27. Platt R, Drescher C, Park SK and Phillips GJ. Genetic system 14. Yang J, Sun B, Huang H et al. High-efficiency scarless genetic for reversible integration of DNA constructs and lacZ gene fu- modification in Escherichia coli by using lambda Red recombi- sions into the Escherichia coli chromosome. Plasmid nation and I-SceI cleavage. Appl Environ Microbiol 2000;43:12–23. 2014;80:3826–34. 28. Gay P, Le Coq D, Steinmetz M et al. Cloning structural gene 15. Link AJ, Phillips D and Church GM. Methods for generating sacB, which codes for exoenzyme levansucrase of Bacillus sub- precise deletions and insertions in the genome of wild-type tilis: expression of the gene in Escherichia coli. J Bacteriol Escherichia coli: application to open reading frame characteri- 1983;153:1424–31. zation. J Bacteriol 1997;179:6228–37. 29. Blomfield I, Vaughn V, Rest R et al. Allelic exchange in 16. Hamilton CM, Aldea M, Washburn B et al. New method for Escherichia coli using the Bacillus subtilis sacB gene and a generating deletions and gene replacements in Escherichia temperature-sensitive pSC101 replicon. Mol Microbiol coli. J Bacteriol1989;171:4617–22. 1991;5:1447–57 17. Posfai  G, Kolisnychenko V, Bereczki Z et al. Markerless gene 30. Reyrat JM, Pelicic V, Gicquel B et al. Counterselectable replacement in Escherichia coli stimulated by a double-strand markers: untapped tools for bacterial genetics and pathogen- break in the chromosome. Nucleic Acids Res 1999;27:4409–15. esis. Infect Immun 1998;66:4011–17. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Biology Methods and Protocols Oxford University Press

Leveraging modern DNA assembly techniques for rapid, markerless genome modification

Free
8 pages

Loading next page...
 
/lp/ou_press/leveraging-modern-dna-assembly-techniques-for-rapid-markerless-genome-l3BjtV64Gh
Publisher
Oxford University Press
Copyright
© The Author 2016. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
eISSN
2396-8923
D.O.I.
10.1093/biomethods/bpw004
Publisher site
See Article on Publisher Site

Abstract

The ability to alter the genomic material of a prokaryotic cell is necessary for experiments designed to define the biology of the organism. In addition, the production of biomolecules may be significantly improved by application of engineered pro- karyotic host cells. Furthermore, in the age of synthetic biology, speed and efficiency are key factors when choosing a method for genome alteration. To address these needs, we have developed a method for modification of the Escherichia coli genome named FAST-GE for Fast Assembly-mediated Scarless Targeted Genome Editing. Traditional cloning steps such as plasmid transformation, propagation and isolation were eliminated. Instead, we developed a DNA assembly-based ap- proach for generating scarless strain modifications, which may include point mutations, deletions and gene replacements, within 48 h after the receipt of polymerase chain reaction primers. The protocol uses established, but optimized, genome modification components such as I-SceI endonuclease to improve recombination efficiency and SacB as a counter-selection mechanism. All DNA-encoded components are assembled into a single allele-exchange vector named pDEL. We were able to rapidly modify the genomes of both E. coli B and K-12 strains with high efficiency. In principle, the method may be applied to other prokaryotic organisms capable of circular dsDNA uptake and homologous recombination. Keywords: Genome modification; DNA assembly; I-SceI; synthetic biology Introduction Escherichia coli fall into two categories. The first set of methods is driven by homologous recombination and is reliant on pro- The ability to manipulate the genetic material of a cell has pro- teinssuchasendogenousRecA[15, 16]. A primary advantage ven to be invaluable for understanding the organism’s biology. of methods that utilize the cell’s own recombination machin- The basis of our understanding of many essential biological pro- ery is that helper plasmids are not necessary. Homologous re- cesses has been enabled by the ability to perform gene deletion combination results in a single crossover event between the and complementation studies [1–4]. Furthermore, countless incoming DNA and the chromosome resulting in a tandem ar- strains have been specifically engineered to facilitate the pro- rangement of the wild-type gene and mutant gene. Then, a duction of valuable proteins requiring modified cellular envi- second crossover event potentially leaves the mutation of in- ronments or helper factors [5–9]. terest in the chromosome. These protocols (labelled as ‘Classic A number of the genome modification methods used today allelic-exchange’ in Fig. 1) generally do not leave a scar in the were developed decades ago and have had only minor updates genome upon removal of the selection marker. However, these due to new techniques becoming available [10–16]. Most of the methods are often cumbersome, taking as much as a week for current methods for modifying common lab organisms such as Received: 17 June 2016; Revised: 6 September 2016. Accepted: 26 September 2016 V The Author 2016. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/ licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 2| Tikh and Samuelson Figure 1: Timeline comparison of the most common methods for genome modification in E. coli (h ¼ hours). During Red-mediated protocols, helper plasmid 1 contains the red exo, beta and gamma genes, which facilitate the incorporation of linear DNA into the chromosome. Helper plasmid two traditionally contains Flp recombinase to help remove the antibiotic resistance marker from the chromosome. When utilizing the CRISPR/Cas9, the first helper plasmid also contains the red exo, beta and gamma genes in addition to Cas9. The second helper plasmid contains the guide RNA, used to eliminate any WT cells which did not undergo desired recombination. Regardless of the method of construction, all modifications should be sequenced in order to verify the presence of the desired modification. a single modification. A second set of commonly used methods DNA to be eliminated by homologous recombination. Due to its is derived from the phage lambda recombination system, most ability to target multiple loci utilizing different guide RNAs, the famously described by Datsenko and Wanner, and utilizes the CRISPR/Cas9 system holds promise as a way to multiplex ge- expression of exogenous phage proteins to mediate the recom- nome editing; however, currently the ability to simultaneously bination steps [10, 11, 17, 18]. The lambda Red system allows introduce changes into multiple locations of the E. coli chromo- for faster integration and modification of a desired locus using some is limited. For an overview of current genome engineering a linear PCR product that contains short segments of homol- methods, see Fig. 1. ogy, but this method requires the presence of a helper plasmid Given the various drawbacks of current methods, we driving the expression of the phage genes and is less efficient sought to push the limits of speed, efficiency and versatility by for the replacement of large pieces of DNA. Though more rapid leveraging recent advances in molecular biology. By utilizing a than classic methods, optimized lambda Red techniques still state-of-the-art DNA assembly approach to create sealed, cir- require approximately 4 days from start to finish, counting the cular DNA in vitro, we are able to completely eliminate the time it takes to initially transform and subsequently cure the need to propagate plasmids in order to perform a genomic helper plasmids [10]. The lambda Red system has also been ap- modification in E. coli. Our work utilized a modified version of plied for DNA oligonucleotide-mediated genome modification; Gibson assembly, however, any DNA assembly technique able however, this method will not be discussed further as it is pri- to efficiently generate high levels of fully circularized DNA marily limited to the generation of point mutations or codon should be applicable in our protocol. Consequently, the de- substitutions [19]. scribed protocol allowed us to perform scarless modifications Recently, the CRISPR/Cas9 system has been adapted for of the E. coli genome in approximately 48 h from the reception modifying genomes in multiple prokaryotic organisms [18, 20– of PCR primers, significantly reducing the time relative to any 23]. Unfortunately, to take full advantage of this system, the or- other published method. This rapid turn-around allows for ganism must be highly recombinogenic, which E. coli is not [22]. fast creation of complex engineered E. coli hosts tailored for In general, the application of the CRISPR/Cas9 system in E. coli specific applications, potentially saving weeks of effort if mul- requires the use of the lambda Red system for the initial allele tiple modifications are required. Finally, given the simplicity exchange step and then the Cas9 nuclease is employed to elimi- of the protocol, we believe it should be possible to apply this nate the WT allele. This method is effective but the overall com- method to a wide range of prokaryotic organisms, so long as bined use of lambda proteins and Cas9 is subject to the same the host is capable of efficient circular DNA uptake, homolo- drawbacks of other Red-mediated methods [18, 20–22]. The ex- gous recombination, and the promoter driving the genes of the ception is a recent report that employed a Cas9 nickase to initi- pDEL vector are functional in the target organism. The method ate the generation of large genomic deletions [23]. The Cas9 described herein is named FAST-GE, an acronym chosen to in- nickase method is dependent upon naturally occurring (or pre- dicate Fast Assembly-mediated Scarless Targeted Genome engineered) repetitive DNA elements in order for intervening Editing. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 3 Materials and methods Bacterial strains, plasmids and enzymes E. coli strain NEB10-beta (New England BioLabs, Ipswich, MA) was used for all cloning steps to create the pDEL vector. Plasmids containing the R6K origin were replicated in BW23473 [24]. E. coli strains NEB Express, T7 Express and ER2744 (New England BioLabs, Ipswich, MA) [25] were utilized in genome modification experiments. All enzymes were from New England BioLabs (Ipswich, MA). Chemicals were purchased from Sigma- Aldrich (St. Louis, MO). Primers and gBlock synthetic DNAs were synthesized by IDT (Coralville, IA). The DasherGFP gene was obtained from DNA2.0 (Menlo Park, CA). Kanamycin and ampicillin were used at 40 mg/ml and 100 mg/ml, respectively, un- less otherwise stated. The rhaBAD and lac promoters were in- duced with 0.2% rhamnose and 0.5 mM IPTG, respectively. Counter-selection plates contained 5% (w/v) sucrose on plates containing 5 g/l of yeast extract, 10 g/l of tryptone and 7.5 g/l of agar, in addition to inducers as described above. Routine cell growth was performed at 37 C in lysogeny broth medium sup- plemented with 0.1% glucose. Colony PCR screens were per- formed using Quick-Load Taq 2X Master Mix, while Q5 Hot Start High Fidelity 2X Master Mix (New England BioLabs, Ipswich, MA) was used to PCR-amplify genomic DNA for sequence verification. Construction of pDEL The open reading frame for the sacB gene was amplified from Figure 2: Schematic overview of the genome modification process. In order to pRE112 [26] and inserted into pMAL-c5X using NEBuilder HiFi modify E. coli chromosomal region B, flanking regions with at least 500 bp of ho- DNA assembly to create pMAL-sacB. Subsequently, the sacB mology on each side of B (boxes A and C) are generated. PCR is the easiest gene was amplified with a lac promoter. Low-level basal expres- method to generate this linear DNA. To enable DNA assembly, the flanking re- sion of sacB is not harmful in the absence of sucrose, as such gions should also contain small homology regions of 18–24 nucleotides to each the lac repressor is not required. If the strain encodes lacI, then other as well as to the linearized deletion cassette (amplified from pDEL). If re- IPTG is necessary during the resolution step. Plasmid pKD4 was gion B is being mutated or replaced, as opposed to deleted, an additional frag- ment containing that modification must be included and will be assembled used as the source of the kanamycin resistance gene and asso- between A and C. Linear DNA fragments are subsequently assembled into a cir- ciated promoter [11]. The R6K origin of replication was amplified V R cular construct using NEBuilder HiFi DNA assembly, transformed into electro- from pCD13PKS [27]. The region containing the I-SceI endonu- competent cells, and the initial integration event is selected for by kanamycin clease site as well as the I-SceI gene under the control of rhaBAD resistance. The integration location is verified by PCR and colonies containing promoter was ordered as a gBlock ; the sequence is shown in the desired insertion are subcultured into media containing the inducers for I- Supplementary Table S1. The rhamnose promoter was chosen SceI and SacB, and subsequently on counter-selection plates with the same in- ducers. Expression of I-SceI promotes homologous recombination in response to as it should be tightly repressed in the absence of the inducer, a double-strand break in the genome and SacB is used to select against cells that but will also generate sufficient levels of I-SceI transcripts in the failed to remove the integration cassette. The final colonies are screened by PCR majority of E. coli strains. To generate the pDEL vector, the fol- for the presence of the desired genome modification. V R lowing PCR fragments were assembled using an NEBuilder HiFi DNA assembly reaction: the R6K origin, kanamycin resistance V R modification was gene replacement, the PCR product of the new gene, I-SceI gBlock and the Plac-sacB fragment. The final circular gene fragment was assembled between the upstream and down- vector is shown in Supplementary Fig. S1 and the entire pDEL stream regions of homology. Typical assembly reactions con- vector sequence is included in the Supplementary Material. tained 250–300 ng of total DNA, at manufacturer’s recommended ratios in a total volume of 10 ml. Sequencing of individual clones Genome modification experiments generated from each of the HiFi DNA assemblies confirmed a very low error rate. Results are shown in Supplementary Table S2. For genome modification experiments, electrocompetent E. coli For transformation, 2 ml of assembled DNA was mixed with were prepared as follows: cells were grown in 3 ml cultures of electrocompetent cells and electroporated in a 1 mm cuvette ac- LB until late exponential phase (OD600  0.8). Cells were har- cording to manufacturer’s instructions (BioRad). Electroporated vested by centrifugation and washed 3 times with 500 ml of ice cells were resuspended in 950 ml of SOC and allowed to recover cold 10% glycerol. Finally, cells were resuspended in 50 ml of 10% at 37 C for 1.5–2 h. Following the recovery step, 100 and 900 ml glycerol. aliquots were plated on fresh LB-Kan plates and incubated at Genome modification constructs were made by assembling 37 C overnight. Colonies from the LB-Kan plate were picked, upstream and downstream homology regions (Fig. 2,boxes Aand transferred into a liquid LB-Kan culture and simultaneously C, respectively), created using site-specific primers containing re- screened by colony PCR to confirm the location of the initial re- gions of microhomology to each other as well as to a PCR product combination event. Cultures containing the desired integration of pDEL amplified with primers located between the R6K origin of were allowed to reach early log phase (approx. OD ¼ 0.2), at replication and the I-SceI gene. In cases when the desired 600 Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 4| Tikh and Samuelson which point they were diluted 1:500 into fresh LB containing Rapid, targeted genome modification 100 mM IPTG and 0.2% w/v rhamnose but lacking kanamycin. Three changes were made to non-essential chromosomal re- Post induction, cultures were allowed to grow for an additional gions: deletion of the lac operon (Fig. 3A), a point mutation in the 3 h at 37 C, after which they were plated on counter-selection lacZ gene (Supplementary Fig. S1A), and insertion of the T7 RNA plates containing 5% sucrose, 100 mM IPTG and 0.2% rhamnose. polymerase gene (Fig. 4). Changes were introduced into both E. Following an overnight incubation at 37 C, colonies from the coli B and K-12 backgrounds. These modifications were chosen to counter-selection plates were picked and grown in LB to be ana- demonstrate the capability of the FAST-GE method to easily gen- lysed by PCR with locus-specific primers for removal of the inte- erate a large deletion, a point mutation and an insertion, respec- gration cassette. In order to confirm the desired changes, the tively. The detailed illustration of the construct used to generate PCR products were column purified and sequenced using the the deletion of the lac operon in ER2523 (an E. coli B derivative) is Sanger method. shown in Fig. 3A. Electrocompetent ER2523 cells were trans- formed with 150 ng of DNA from the assembly reaction. This pro- Results tocol was sufficient to result in several correctly integrated transformants (Fig. 3B). Two different primer pairs are used to Composition of the insertion cassette identify strains with the pDEL construct integrated correctly at When designing the pDEL vector, several critical factors were the desired locus. All eight colonies analysed contained a pDEL considered. First, the vector size was maintained as compact as integration at the lac locus when combining the results of the F1- possible in order to optimize amplification yield and fidelity. R1 and F2-R2 PCR analyses. Each primer pair has two potential This led us to critically evaluate the components currently used products depending on whether recombination occurred via the 0 0 with other genome modification approaches. We determined 5 flank homology region or the 3 flank homology region. In each that at the minimum an antibiotic selection marker and a coun- diagnostic PCR reaction, the extension time is set to amplify the ter-selection marker were required. The frequency of homolo- shorter of the two possible PCR products. We found that this ex- gous recombination in wild-type E. coli is low; however, perimental approach was acceptable as shown in Fig. 3B. recombination can be increased locally by the presence of a Recombinant number one was chosen for the resolution step. double-strand break. Thus, an I-SceI restriction site and the cor- After reaching noticeable turbidity grown in the presence of responding endonuclease gene were added to the construct to kanamycin (OD ¼ approx. 0.2), the culture from recombinant encourage homologous recombination by inducing a double- number one was diluted 1:500 into fresh LB medium lacking strand break. Additionally, we customized the promoters of the kanamycin but containing rhamnose and IPTG, for induction of sacB and I-SceI genes, to improve the overall efficiency of the I-SceI and sacB, respectively. After 2 h of incubation in antibiotic- protocol. The final composition of the deletion cassette is free medium, dilutions were plated on sucrose agar plates. shown in Fig. 2. Following counter-selection on sucrose, final resolution and re- Kanamycin resistance was chosen as the primary selection moval of the integration cassette was highly effective. Table 1 marker as growth on kanamycin plates has proven to be a con- presents a summary of the process for deletion of the lac operon. sistent indicator of single-copy genome integration [11, 16]. For The removal of the counter-selection cassette relies upon homol- counter-selection SacB was chosen as its activity and toxicity in ogous recombination and depending on the recombination site E. coli has been well documented over the past several decades the process can lead to one of two outcomes. First, the entire in- [28, 29]. The sacB gene originates from B. subtilis and encodes a tegration cassette may be removed such that the original genome periplasmic levansucrase. The exact mechanism of toxicity by sequence is restored (henceforth referred to as WT sequence). SacB is still not well understood, but it is thought that periplas- Alternatively, the integration cassette may be removed resulting mic SacB creates large levan polymers when cells are grown in in the desired genomic modification (Fig. 3A). Blue colony colour the presence of sucrose [29, 30]. In order to optimize sacB ex- (X-gal conversion) was used as an indicator of lac operon function pression and periplasmic localization, we replaced the native and Fig. 3C shows that 9 of 16 resolved clones had regained b-ga- Bacillus promoter with a lac promoter. Counter-selection on su- lactosidase activity, suggesting resolution to WT sequence. PCR crose proved to be very robust as a large majority of surviving analysis of the same clones showed that six of the recombinants colonies were free of the sacB gene (Supplementary Table S3). had the desired deletion of the lac operon. Several of the resulting Inclusion of the I-SceI homing endonuclease as well as the 1.3 kb F1-R2 PCR products were sequenced to find that no unin- cognate recognition site was based on several recent papers tended mutations were introduced. demonstrating the increased recombination frequency during The described method of genome engineering is not biased genome alteration by the generation of a unique double-strand according to the modification type. To demonstrate the ability of break in the E. coli chromosome by I-SceI. In order to augment I- the FAST-GE method to introduce large insertions in the E. coli SceI cleavage efficiency, two terminators were introduced im- genome, the T7 RNA polymerase (T7 gene 1) was inserted into mediately upstream of the I-SceI cut site as active transcription the chromosome of MC1061, a common E. coli K-12 strain. T7 through the recognition sequence was recently reported to gene 1 may be inserted into the lac locus to facilitate IPTG- lower I-SceI cleavage efficiency [10]. inducible expression of the T7 RNA polymerase for the purpose In addition to the components actively involved in the re- of recombinant protein expression. Accordingly, we chose to re- combination process, we chose to include the R6K origin of rep- create the lacZ-T7gene1 operon fusion (approx. 3 kb), as found in lication on the pDEL plasmid. The R6K origin is unique in that it T7 Express cells, between the yahF and mhpT genes (Fig. 4A). The requires the presence of the pir gene product for the plasmid to integration and counter-selection steps were followed by colony replicate; common E. coli strains lack this gene. The conditional PCR analysis as described above (Fig. 4). As with other experi- origin was included as an alternative strategy, in cases where ments, we were able to obtain desired clones in 48 h from the be- plasmid isolation is a more practical first step. In our experi- ginning of the process. Using a similar procedure, we were also ence, the DNA assembly products were successfully trans- able to replace a single nucleotide in order to create an E462A ac- formed and integrated directly in all strains tested. tive site substitution within LacZ (Supplementary Fig. S1). Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 5 Figure 3: Deletion of the lac operon. (A) Overview of the genomic locus around the lac operon in strain ER2523 as well as the composition of the deletion cassette. The scheme demonstrates the two potential integration outcomes as well as the location of primer pairs F1-R1 and F2-R2 used to determine the location of the integration event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a representative subset of PCR products obtained with the F1-R1 and F2-R2 primer pairs. Expected sizes for the F1-R1 PCR products are either 1.6 kb if the crossover event occurred in the mhpR gene or 8.5 kb if the first crossover event occurred in the cynX gene. The expected sizes for the PCR products from the F2-R2 primer pair are either 7 kb if the crossover event occurred in the mhpR gene or 1.7 kb if the first crossover event occurred in the cynX gene. A DNA molecular ladder (Quick-load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1-8 served as a template for both F1-R1 and F2-R2 PCR reactions. (C) An agarose gel of PCR results utilizing the primer pair F1 and R2 from individual colonies post counter-selection on sucrose. Expected PCR product size from a successful deletion of the lac operon is 1.3 kb. Figure 4: Insertion of the T7 RNA polymerase gene into K-12 strain MC1061. (A) Overview of the genomic locus around the yahF-mhpT locus as well as the composition of the T7 RNA polymerase gene insertion cassette. The scheme demonstrates the two potential integration outcomes after the first crossover event and the location of primer pairs F1-R1 and F2-R2 used to determine the location of the integration event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a representative subset of PCR products obtained with the F1-R1 and F2-R2 primer pairs. The expected sizes for the F1-R1 PCR products are either 4.5 kb if the crossover event occurred in the yahF gene or 2.5 kb if the first crossover event occurred in the mhpT gene. The expected sizes for the PCR products from the F2-R2 primer pair are either 2.6 kb if the crossover event occurred in the yahF gene or 4.7 kb if the first crossover event occurred in the mhpT gene. A DNA molecular ladder (Quick-load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1-6 served as a template for F1-R1 and F2-R2 PCR reactions. (C) An agarose gel showing PCR results from 16 representative colonies post counterselection on sucrose (1–16) from primer pair F1 and R2. The ex- pected PCR product size for successful T7 gene 1 insertion is 5.2 kb. Resolution to WT yahF-mhpT locus yields a PCR product of 2 kb. Lack of a PCR product would suggests that the integration cassette was not successfully removed from the genome. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 6| Tikh and Samuelson Table 1: The efficiency of the genome modification protocol at various loci and modification types Allele Integration efficiency Resolution efficiency Desired modification LacZ E462A 3/8 5/16 3/5 Dlac (B strain) 8/8 13/16 6/13 T7 gene 1 5/6 16/16 7/16 rplI-Dasher 1/4 7/16 0/7 The ratio of isolated colonies reverting to WT sequence as op- When we attempted to fuse a fluorescent protein posed to the desired modification varied depending on the mod- (DasherGFP) to the C-terminus of the ribosome L9 subunit, cor- ification and the integration locus (Table 1). responding to the rplI gene, colonies containing the initial cross- over event were isolated (Fig. 5B). Out of the four colonies tested, only one returned the expected fragment size for the Working with essential genes two diagnostic PCR reactions (Fig. 5B). The colony showing the expected PCR pattern was chosen for resolution. However, none The method described in this article is also beneficial for deter- of the colonies tested after sucrose selection contained the de- mining whether a gene is essential for growth of the bacterium sired gene fusion (Table 1)(Fig. 5C). This led us to conclude that or if a specific mutation destroys function. Since a fully func- this fusion protein is not tolerated by E. coli due to disruption of tional copy of the target gene is maintained during the integra- ribosome function. tion event, integration should always be possible as the present method separates the integration and resolution steps. If all of the resolved colonies in a sufficiently large sample have re- Best practices for genome modification success verted to WT sequence, then one can reasonably assume that the desired modification is not tolerated by the cell (Fig. 5). In When optimizing the FAST-GE protocol, we have noted several contrast, in lambda Red-derived methods, the absence of viable points where special care should be practiced to achieve best re- transformants does not necessarily indicate that a gene is es- sults and we offer several guidelines to improve the likelihood sential or that the modification being attempted is not tolerated of success for first-time users. First, as with other homology- by the cell. An additional advantage of this method is that reso- driven genome modification methods, the size of the flanking lution can be carried out under different culture conditions, homology regions will affect the efficiency of the initial recom- such as varied temperatures or growth medium composition. bination event. Following standard practice for RecA-mediated Thus, one can determine if a gene’s activity is essential for sur- homologous recombination, we recommend designing homol- vival under varied experimental conditions. ogy regions of at least 500 bases on both sides of the desired Figure 5: Working with an essential gene. (A) Overview of the rplI genomic locus as well as the composition of the rplI-dasher fusion cassette. The scheme demonstrates the two potential integration outcomes after the first crossover event and the location of primer pairs F1-R1 and F2-R2, used to determine the location of the integra- tion event. Finally, two potential outcomes of the resolution step are shown. (B) An agarose gel showing a sample of a PCR product obtained with the F1-R1 and F2-R2 primer pairs. The expected sizes for the F1-R1 PCR products are either 2 kb if the crossover event occurred in the rplI gene or 1.3 kb if the first crossover event occurred in the yjfZ ORF. Similarly, the expected sizes for the PCR products produced by the F2-R2 primer pair are either 1.4 kb if the crossover event occurred in the rplI gene or 2.1 kb if the first crossover event occurred in the yjfZ ORF. A DNA molecular ladder (Quick-Load 1 kb ladder) is designated as M and 1 and 3 kb markers are labelled. The same set of individual colonies numbered 1–4 served as a template for both F1-R1 and F2-R2 PCR reactions. (C) An agarose gel of PCR results from 16 representative col- onies post counter-selection on sucrose from primer pair F1 and R2. The expected PCR product size for successful rplI-dasher fusion is 1.6 kb. Resolution to WT rplI se- quence yields a PCR product of 1 kb. Lack of a PCR product suggests that the integration cassette was not successfully removed from the genome. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Leveraging DNA assembly for genome modification | 7 modification [16]. Secondly, researchers may choose to trans- accessible to anyone familiar with PCR and electroporation- form the assembled genome modification construct into a pir based transformation protocols. strain in parallel with transformation into the desired host. This While we were able to generate all of our desired mutants in optional, extra step ensures that a suicide plasmid is available both K-12 and B derivatives using direct transformation of the to use, in case direct transformation of the assembly reaction assembly reaction, we do understand that some researchers does not yield colonies or if electroporation is not an option for may prefer to use a purified plasmid when working with diffi- transformation. Thirdly, if multiple changes to the genome are cult to transform strains. For those preferring to work with an desired, we suggest that all of the necessary genome modifica- isolated assembly clone we have included the conditional R6K tion constructs be assembled in parallel in a pir host and after origin of replication on the pDEL vector, which allows transfor- the first modification, purified suicide plasmids can be used to mation of the assembly reaction into a pir host for multi-copy perform subsequent modifications, simplifying and expediting replication. generation of the desired strain. In a drive to improve the overall speed of the protocol, versa- Fourthly, when attempting direct transformations, we tility was not compromised as insertions, deletions and point strongly recommend that a high concentration of DNA be used mutations are all equally possible. The only limitation of this for the DNA assembly reaction, up to 250–300 ng of DNA can be method (and many other methods) is that the strain to be modi- assembled in a 10 ml reaction, with subsequent transformation fied must be capable of homologous recombination. For exam- of 2 ml of this reaction. If working with strains known to have ple, typical cloning strains (e.g. DH5a) containing a recA low competence, larger assemblies on the scale of 500–1000 ng mutation are not suitable for modification by this method. of DNA in a 50 ml reaction may be necessary. Then, concentrate Colonies containing the desired initial crossover event are easy the DNA before the transformation step. to identify by colony PCR and require at most the screening of eight colonies, which is on par or better than any other protocols reported to date. Additionally, no difference in the frequency of Discussion integration was apparent regardless of whether the target gene was essential for E. coli growth. This combination of efficiency For decades, genome editing in bacteria has relied upon the use and speed is unique and is of especially high value to re- of plasmids, either as a template for recombination or as the searchers interested in making multiple genome modifications. source of helper proteins to facilitate the recombination event. The reliance on replicative plasmids, requires an initial transfor- mation step to introduce them into the hosts, as well as a curing Author Contributions procedure to remove them after the genome modification proto- col is completed. In some methods, plasmid curing is not I.B.T. and J.C.S. designed research. I.B.T. performed experi- straightforward. In allele exchange methods where the plasmid ments. I.B.T. and J.C.S. analysed data and wrote the paper. is carrying the WT gene after resolution, if plasmid curing fails, then the researcher is left wondering whether the WT gene Acknowledgements might be essential for viability. Some allele exchange methods We would like to thank Elisabeth A. Raleigh and William are particularly time-consuming and may require 2–3 weeks Jack for critical reading of the manuscript and Don Comb from construct design to verification of the final strain [16]. and Jim Ellard for creating and fostering a creative research In this study, we demonstrate the ability to modify a bacterial genome without the need for a replicative plasmid, thus drasti- environment. cally reducing protocol time. This feat is accomplished by leveraging recent advances in DNA assembly technologies in or- Funding der to transform the cells with a non-replicative, circularized piece of DNA. The assembled construct may be immediately Funding for open access charge: New England BioLabs, Inc. transformed into the desired strain without the need for se- Conflict of interest statement. The authors of this manuscript quence verification due to high fidelity of the assembly process. are employees of New England Biolabs, which produces mo- Importantly, only a small number of resolved strains (typically lecular biology reagents, including NEBuilder HiFi DNA fewer than eight) need to be analysed by focused sequencing re- Assembly Master Mix. actions to verify the genome alteration as well as fidelity of the assembly process. Eliminating the requirement for helper plasmid(s) enables Supplementary data the generation of markerless genomic modifications within 48 h from receipt of the necessary primers. Minimum requirements Supplementary data is available at Biology Methods and for this protocol are fragments of DNA sequence flanking the Protocols online. site of the desired modification and linear DNA encoding the pDEL deletion cassette, with all pieces containing small homol- V References ogy regions sufficient to allow assembly via the NEBuilder HiFi DNA assembly method (or other high-efficiency DNA assembly 1. Trinh CT, Li J, Blanch HW et al. Redesigning Escherichia coli me- methods). A high fidelity DNA polymerase should be employed tabolism for anaerobic production of isobutanol. Appl Environ to obtain the necessary DNA fragments and bacterial colonies Microbiol 2011;77:4894–904. may serve as the source of chromosomal template DNA to gen- 2. Baba T, Ara T, Hasegawa M et al. Construction of Escherichia coli erate fragments for the DNA assembly reaction. As a result, K-12 in-frame, single-gene knockout mutants: the Keio collec- time-consuming plasmid or genomic DNA purification proce- tion. Mol Syst Biol 2006;2:2006.0008. doi:10.1038/msb4100050. dures may be bypassed. To our knowledge, this protocol is at 3. Liu H, Yu C, Feng D et al. Production of extracellular fatty acid least 2 days faster, from start to finish, than any other method using engineered Escherichia coli. Microb Cell Fact currently described. In addition, the FAST-GE method is 2016;11:41–54. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018 8| Tikh and Samuelson 4. Snyder WB and Silhavy TJ. Enhanced export of beta- 18. Pyne ME, Moo-Young M, Chung DA et al. Coupling the CRISPR/ galactosidase fusion proteins in prlF mutants is Lon depen- Cas9 system to lambda Red recombineering enables simpli- dent. J Bacteriol 1992;174:5661–68. fied chromosomal gene replacement in Escherichia coli. Appl 5. Lobstein J, Emrich CA, Jeans C et al. SHuffle, a novel Escherichia Environ Microbiol 2015;81:5103–14. coli protein expression strain capable of correctly folding disul- 19. Wang HH, Xu G, Vonner AJ et al. Modified bases enable fide bonded proteins in its cytoplasm. Microb Cell Fact 2012;11:56. high-efficiency oligonucleotide-mediated allelic replace- 6. Ajikumar PK, Xiao WH, Tyo KE et al. Isoprenoid pathway opti- ment via mismatch repair evasion. Nucleic Acids Res mization for Taxol precursor overproduction in Escherichia 2011;39:7336–47. coli. Science 2010;330:70–4. 20. Touchon M and Rocha EP. The small, slow and specialized 7. Lee SJ, Lee DY, Kim TY et al. Metabolic engineering of CRISPR and anti-CRISPR of Escherichia and Salmonella. PLoS Escherichia coli for enhanced production of succinic acid, One 2010;5:e11126. based on genome comparison and in silico gene knockout 21. Jiang Y, Chen B, Duan C et al. Multigene editing in the simulation. Appl Environ Microbiol 2005;71:7880–87. Escherichia coli genome via the CRISPR-Cas9 system. Appl 8. Kolisnychenko V, Plunkett G, Herring CD et al. Engineering a Environ Microbiol 2015;81:2506–14. reduced Escherichia coli genome. Genome Res 2002;12:640–47. 22. Jiang W, Bikard D, Cox D et al. RNA-guided editing of bacterial 9. Robichon C, Luo J, Causey TB et al. Engineering Escherichia coli genomes using CRISPR-Cas systems. Nat Biotechnol BL21 (DE3) derivative strains to minimize E. coli protein con- 2013;31:233–39. tamination after purification by immobilized metal affinity 23. Standage-Beier KS, Zhang Q and Wang X. Targeted large- chromatography. Appl Environ Microbiol 2011;77:4634–46. scale deletion of bacterial genomes using CRISPR-Nickases. 10. Kim J, Webb AM, Kershner JP et al. A versatile and highly effi- ACS Synth Biol 2015;4:1217–25. cient method for scarless genome editing in Escherichia coli 24. Haldimann A, Prahalad MK, Fisher SL et al. Altered recogni- and Salmonella enterica. BMC Biotechnol 2014;14:84. tion mutants of the response regulator PhoB: a new genetic 11. Datsenko KA and Wanner BL. One-step inactivation of chro- strategy for studying protein-protein interactions. Proc Natl mosomal genes in Escherichia coli K-12 using PCR products. Acad Sci USA 1996;93:14361–366. Proc Natl Acad Sci USA 2000;97:6640–45. 25. Samuelson JC, Zhu Z and Xu S. The isolation of strand- 12. Yu BJ, Kang KH, Lee JH et al. Rapid and efficient construction specific nicking endonucleases from a randomized SapI ex- of markerless deletions in the Escherichia coli genome. Nucleic pression library. Nucleic Acids Res 2004;32:3661–71. Acids Res 2008;36:e84. 26. Edwards RA, Keller LH and Schifferli DM. Improved allelic ex- 13. Herring CD and Blattner FR. Conditional lethal amber muta- change vectors and their use to analyze 987P fimbria gene ex- tions in essential Escherichia coli genes. J Bacteriol pression. Gene 1998;207:149–57. 2004;186:2673–81. 27. Platt R, Drescher C, Park SK and Phillips GJ. Genetic system 14. Yang J, Sun B, Huang H et al. High-efficiency scarless genetic for reversible integration of DNA constructs and lacZ gene fu- modification in Escherichia coli by using lambda Red recombi- sions into the Escherichia coli chromosome. Plasmid nation and I-SceI cleavage. Appl Environ Microbiol 2000;43:12–23. 2014;80:3826–34. 28. Gay P, Le Coq D, Steinmetz M et al. Cloning structural gene 15. Link AJ, Phillips D and Church GM. Methods for generating sacB, which codes for exoenzyme levansucrase of Bacillus sub- precise deletions and insertions in the genome of wild-type tilis: expression of the gene in Escherichia coli. J Bacteriol Escherichia coli: application to open reading frame characteri- 1983;153:1424–31. zation. J Bacteriol 1997;179:6228–37. 29. Blomfield I, Vaughn V, Rest R et al. Allelic exchange in 16. Hamilton CM, Aldea M, Washburn B et al. New method for Escherichia coli using the Bacillus subtilis sacB gene and a generating deletions and gene replacements in Escherichia temperature-sensitive pSC101 replicon. Mol Microbiol coli. J Bacteriol1989;171:4617–22. 1991;5:1447–57 17. Posfai  G, Kolisnychenko V, Bereczki Z et al. Markerless gene 30. Reyrat JM, Pelicic V, Gicquel B et al. Counterselectable replacement in Escherichia coli stimulated by a double-strand markers: untapped tools for bacterial genetics and pathogen- break in the chromosome. Nucleic Acids Res 1999;27:4409–15. esis. Infect Immun 1998;66:4011–17. Downloaded from https://academic.oup.com/biomethods/article-abstract/1/1/bpw004/2743810 by Ed 'DeepDyve' Gillespie user on 13 July 2018

Journal

Biology Methods and ProtocolsOxford University Press

Published: Dec 27, 2016

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off