RECOMBINEERING IN NON-MODEL BACTERIA

Abstract

The technology of recombineering, in vivo genetic engineering, was initially developed in E. coli and uses bacteriophage-encoded homologous recombination proteins to efficiently recombine DNA at short homologies (35–50 nt). Because the technology is homology driven, genomic DNA can be modified precisely and independently of restriction site location. Recombineering utilizes linear DNA substrates that are introduced into the cell by electroporation; these can be PCR products, synthetic double-strand DNA (dsDNA), or single-strand DNA (ssDNA). Here we describe the applications, challenges and factors affecting ssDNA and dsDNA recombineering in a variety of non-model bacteria, both Gram-negative and -positive, and recent breakthroughs in the field. We list different microbes in which the widely utilized phage λ Red and Rac RecET recombination systems have been used for in vivo genetic engineering. New homologous ssDNA and dsDNA recombineering systems isolated from non-model bacteria are also described. Basic Protocol 1 outlines a method for ssDNA recombineering in the non-model species of Shewanella. Alternate Protocol 1 describes the use of CRISPR/Cas as a counter-selection system in conjunction with recombineering to enhance recovery of recombinants. The Commentary provides additional background information, pertinent considerations for experimental design and parameters critical for success. The design of ssDNA oligonucleotides (oligos) and various internet-based tools for oligo selection from genome sequences are described in the commentary. The use of oligo-mediated recombination is discussed in the Commentary. This simple form of genome editing uses only ssDNA oligo(s) and does not require an exogenous recombination system. The information presented here should help researchers identify a recombineering system suitable for their microbe(s) of interest. If no system has been characterized for a specific microbe, researchers can find guidance in developing a recombineering system from scratch. We provide a flowchart of decision-making paths for strategically applying annealase-dependent or oligo-mediated recombination in non-model and undomesticated bacteria.

Basic Protocol 1:

ssDNA-RECOMBINEERING IN SHEWANELLA SPECIES

Alternate Protocol 1:

SSDNA-RECOMBINEERING COUPLED TO CRISPR/Cas9 IN SHEWANELLA SPECIES

INTRODUCTION

APPLICATION OF RECOMBINEERING TO NON-MODEL MICROORGANISMS

In the more than twenty years since recombineering was first achieved in E. coli, it has been refined to the point that nearly any desired genetic manipulation is possible. See the Current Protocols article by Thomason, Sawitzke, Li, Costantino & Court (2014) for background information. However, the state of recombineering in non-model bacteria is at a more preliminary stage. As the field of non-model bacterial genetics expands, the ability to perform a wide range of genetic manipulations in different species becomes vital to researchers. Several different approaches to recombineering in non-model bacteria have been tried. Some success has been achieved with expressing the E. coli phage generalized homologous recombination functions that promote genetic recombination using short homologies and are typically used for recombineering, Red from phage λ and RecET from the E. coli Rac prophage (the latter referred to here as RecET_Ec). Both these systems consist of two proteins: a single-strand annealing protein (SSAP) and 5’->3’ dsDNA exonuclease; the SSAP is also referred to as a recombinase, and more specifically as an annealase. In this paper we will generally use the term annealase, following the convention of K. Murphy (2016). For the λ Red system, the recombination functions include the annealase Beta and the exonuclease Exo. For RecET_Ec, the annealase is the RecT protein, and the exonuclease protein is RecE. Introduction of small genetic alterations such as point mutations and changes of a few bases can be done with ssDNA and requires only the annealase function. However, insertion of large heterologous DNA molecules, such as antibiotic resistance genes, requires a dsDNA substrate where both the annealase and exonuclease functions are required. While λ Red and RecET have been successfully used for genetic manipulation in bacteria closely related to E. coli, their functional range in other bacterial species is limited, especially for dsDNA recombination. For less closely related bacteria, identifying and adapting native phage recombination functions has yielded more successes and the protocol presented here provides an example of this approach. A third approach in establishing recombineering in non-model bacteria is to use an innate ssDNA recombination activity present in many types of bacteria, referred to here as oligo-mediated recombination. The frequencies of these types of genetic modifications are magnitudes lower than phage annealase-promoted events, and the mechanism is not known.

This paper describes the current state of the field and suggests approaches for developing recombineering in non-model bacteria. Table 1 lists the Gram-negative and Gram-positive bacterial species in which the λ Red and RecET_Ec systems have been used for ssDNA and dsDNA recombineering. Tables 2 and 3 describe homologous ssDNA and dsDNA recombineering systems that have been characterized to date in diverse species, and identify key factors that affect recombineering frequency, many of which parallel findings in E. coli. Basic Protocol 1 and Figure 1 present the steps for executing recombineering in one non-model bacterium, the Gram-negative Shewanella species. In this protocol, an endogenous Shewanella phage annealase, W3 Beta, with similarity to the phage λ Beta protein, is used to catalyze the recombination. Alternate Protocol 1 provides a method of coupling either annealase- or oligo-mediated ssDNA recombination with CRISPR/Cas, which is employed here for use as a counter-selection; Figure 2 illustrates this protocol. Table 4 lists the wild-type and engineered Shewanella strains available for both protocols. These protocols differ from the E. coli protocol (Thomason et al., 2014) with respect to cell preparation. While we have not tried both protocols in parallel for both bacteria, we believe that most of the steps in the two procedures should be largely interchangeable, unless otherwise noted.

Table 1.

Reports of ssDNA and dsDNA recombineering with λ Red or RecET_Ec in non-model bacteria.

ssDNA recombineering
Species	Recombineering system	References
Gram-negative, γ-proteobacteria
Salmonella enterica	λ Red Gam, Beta and Exo	(Simanti Datta, Costantino, & Court, 2006; Nyerges et al., 2016; Sawitzke et al., 2007)
Pantoea ananatis	λ Red Gam and Beta	(Katashkina et al., 2009)
Yersinia pestis	λ Red Gam, Beta and Exo	(Yan et al., 2017)
Citrobacter freundii	λ Red Gam, Beta and Exo	(Nyerges et al., 2016)
Gram-positive, actinobacteria
Corynebacterium glutamicum	RecTEc	(Binder et al., 2013; Li et al., 2021; Su, 2018)
Gram-positive, bacilli
Bacillus subtilis	λ Red Beta, Exo	(Wang et al., 2012)
dsDNA recombineering
Gram-negative, γ-proteobacteria
Salmonella enterica	λ Red Gam, Beta and Exo	(Blank, Hensel, & Gerlach, 2011; Bunny, Liu, & Roth, 2002; Cox et al., 2007; Czarniak & Hensel, 2015; Doublet et al., 2008; Gerlach, Holzer, Jackel, & Hensel, 2007; Gerlach, Jackel, Holzer, & Hensel, 2009; Hoffmann, Schmidt, Walter, Bender, & Gerlach, 2017; Karlinsey, 2007; Uzzau, Figueroa-Bossi, Rubino, & Bossi, 2001; Yu et al., 2011)
Klebsiella pneumoniae	λ Red Gam, Beta and Exo	(Chen et al., 2016; Huang et al., 2014; Wei, Sun, Shi, Liu, & Hao, 2013; Wei, Wang, Shi, & Hao, 2012)
Klebsiella aerogenes	λ Red Gam, Beta and Exo	(Wu et al., 2017)
Pantoea ananatis	λ Red Gam, Beta and Exo	(Katashkina et al., 2009)
Shigella flexneri/sonnei/dysenteriae	λ Red Gam, Beta and Exo	(Gerlach et al., 2007; Ohya, Handa, Ogawa, Suzuki, & Sasakawa, 2005; Ranallo, Barnoy, Thakkar, Urick, & Venkatesan, 2006)
Yersinia pseudotuberculosis	λ Red Gam, Beta and Exo	(Derbise et al., 2003)
Yersinia pestis	λ Red Gam, Beta and Exo	(Derbise et al., 2003; Sun et al., 2008; Zhao, 2018)
Yersinia enterocolitica	λ Red Gam, Beta and Exo	(Trulzsch, Sporleder, Igwe, Russmann, & Heesemann, 2004)
Citrobacter rodentium	λ Red Gam, Beta and Exo	(Deng et al., 2004)
		(Gueguen & Cascales, 2013)
Serratia marcescens	λ Red Gam, Beta and Exo	(Rossi, Paquelin, Ghigo, & Wandersman, 2003)
Pseudomonas aeruginosa	λ Red Gam, Beta and Exo	(Lesic & Rahme, 2008)
		(Liang & Liu, 2010)
Pseudomonas putida	λ Red Gam, Beta and Exo	(Chen, Ling, & Shang, 2016; Cook et al., 2018; Luo et al., 2016)
	RecETEc	(Choi et al., 2018)
Edwardsiella ictaluri	λ Red Gam, Beta and Exo	(Hossain et al., 2015)
Aeromonas hydrophila
Vibrio cholerae	λ Red Gam, Beta and Exo	(Yamamoto, Izumiya, Morita, Arakawa, & Watanabe, 2009)
Gram-negative, β-proteobacteria
Burkholderia cepacia	λ Red Gam, Beta and Exo	(Jia, Yang, Liu, Li, & Yan, 2010)
Burkholderia thailandensis and Burkholderia pseudomallei	λ Red Gam, Beta and Exo mutated and optimized version from (Nakayama & Ohara, 2005)*	(Kang et al., 2011)
Gram-negative, α-proteobacteria
Agrobacterium tumefaciens	λ Red Gam, Beta and Exo	(Hu et al., 2014)
Zymomonas mobilis	RecETEc and RecE588TEc**	(Wu, Cao, Li, Zhang. Tan, 2017)
	λ Red Gam, Beta and Exo	(Khandelwal et al., 2018)
Gram-positive, actinobacteria
Corynebacterium glutamicum	RecETEc	(Huang et al., 2017; Li et al., 2021)

^*This mutated λ-Red system contains 1) an additional 18bp 5′-UTR (CGTCGACGCTTATAAAAA) from phage λ gam gene, 2) a nonsynonymous amino acid substitution (C118R) in Red Beta, and 3) a nonsynonymous amino acid substitution in Red Exo (R137Q). When expressed from a high-copy plasmid (ColE1 origin), these modifications resulted in 145-fold more recombinants when engineering bacterial artificial chromosomes in E. coli.

^**The first 587 amino acids of full-length RecE are dispensable for exonuclease activity.

Table 2.

Homologous systems for ssDNA-based recombineering in different species.

Species	Recombineering system	Frequencies*	References
Gram negative, γ-proteobacteria
Pseudomonas putida	Ssr (from P. putida DOT-T1E, gene T1E_1405, homolog of λ-Beta)	~10−5–10−3 for small and large deletions, small insertions and single-bp mutations with 40–50nt homology arms, ~3ug ssDNA	(Aparicio, de Lorenzo, & Martinez-Garcia, 2020; Aparicio et al., 2016; Ricaurte et al., 2018)
	Rec2 (from P. putida CSV86, homolog of λ-Beta)	~10−5–10−3 for large deletions, small insertions and single-bp mutations with 40–50nt homology arms, ~3ug ssDNA Up to ~10−2 for single-bp mutations using cI857-PL system with 30–45nt homology arms Up to ~10−1 for single-bp changes after 5–10 recombineering cycles with 29–45nt homology arms	(Aparicio, Nyerges, Martinez-Garcia, et al., 2020; Aparicio, Nyerges, Nagy, et al., 2020; Ricaurte et al., 2018)
Pseudomonas syringae	RecTPsy (homolog of RecTEc)	~10−4 for single bp mutations using 40nt homology arms, ~1ug ssDNA	(Bao et al., 2012; Swingle, Bao, et al., 2010; Swingle, Markel, Costantino, et al., 2010)
Shewanella oneidensis	W3 Beta (homolog of λ-Beta)	~10−2 for single bp mutations and small insertions using 40–80nt homology arms, 2.5ug ssDNA	(Corts et al., 2019a, 2019b)
Shewanella amazonensis		~10−5 for single bp mutations using 40nt homology arms, 2.5ug ssDNA	(Corts et al., 2019a)
Vibrio natriegens	SXT-β and -Exo (λ Red analogs found in an SXT mobile element) coupled to λ- Gam	Up to ~104 total recombinants for small bp changes using 40–45nt homology arms, and saturating ssDNA amounts	(Lee et al., 2017)
Gram negative, β-proteobacteria
Burkholderiales	Redβ7029 (from Burkholderiales strain 7029, homolog of λ-Beta) coupled to λ-Gam	~103 total recombinants for a large 21.2kbp deletion using 80nt homology arms. (Unknown ssDNA amount; dsDNA was treated with λ Exonuclease to make ssDNA)	(Wang et al., 2018)
Gram positive, Clostridia
Clostridium acetobutylicum	Cpf0939 (RecTEc homolog) from C. perfringens	~101–102 total recombinants for point mutations using <35–40nt homology arms and 10ug ssDNA	(Dong, Tao, Gong, Li, & Zhang, 2014)
Gram positive, Actinobacteria
Mycobacterium smegmatis	Gp61 from Mycobacteriophage Che9c (homolog of RecTEc)	~10−1 to 10−3 for single bp mutations using 40nt homology arms, 100ng ssDNA	*Frequency determined as recombinants over total competent cells from a plasmid transformation (van Kessel & Hatfull, 2008)
Mycobacterium tuberculosis
Corynebacterium glutamicum	WP -010191133.1 from C. aurimucosum (homolog of RecTEc)	~102 total recombinants for point mutations using (unknown homology arms) 2μg ssDNA	(Chang et al., 2021)
Rhodococcus opacus	RrRecT	~103 total recombinants for one point mutation using ~30nt homology arms and 5–25μg ssDNA	(Liang, Wei, Jiao, & Yu, 2021)
Gram positive, Bacilli
Bacillus subtilis	GP35 from the native phage SPP1 (RecTEc homolog)	~10−2–10−1 for large deletions with 500nt homology arms and 1ug ssDNA	(Sun et al., 2015)
Lactobacillus reuteri	RecT1 from L. reuteri ATCC PTA 6475 (RecTEc homolog)	~10−3–10−1 for point mutations with ~40nt homology arms and 100μg ssDNA	(van Pijkeren & Britton, 2012, 2014; van Pijkeren et al., 2012)
Lactobacillus plantarum		~10−6 for point mutations with ~40nt homology arms and 100μg ssDNA
Lactobacillus gasseri		~10−5 for point mutations with ~40nt homology arms and 100μg ssDNA
Lactococcus lactis		~10−2–10−1 for point mutations with ~40nt homology arms and 500μg ssDNA
	EF2132 from E. faecalis (RecTEc homolog)	~103 total recombinants for point mutations using ~40nt homology arms and 100μg ssDNA (<103 recombinants using 10 μg ssDNA)	(Guo, Xin, Zhang, Gu, & Kong, 2019; van Pijkeren & Britton, 2014)
Lactobacillus casei	LCABL 13050 (homolog of λ-Beta) from prophage PLE3 in L. casei BL23	~101 total recombinants for point mutations using ~40nt homology arms and 100μg ssDNA	(Xin, Guo, Mu, & Kong, 2017)
Staphylococcus aureus	EF2132 from E. faecalis (RecTEc homolog)	~10−6–10−3 (depending on the strain) for point mutations with ~40–45nt homology arms and ~6μg ssDNA	(Penewit et al., 2018; Penewit & Salipante, 2020)
Enterococcus faecium	EFWG_RS08525 from E. faecium prophage (RecTEc homolog)	Up to 100% editing efficiency (using a selection marker) for large insertions using 1,000nt homology arms (ssDNA obtained through a dsDNA digestion reaction)	(Chen, Griffin, Maguin, Varble, & Hang; V. Chen et al., 2021)
Enterococcus durans		*Efficiency not reported using annealase alone (with CRISPR/Cas9 counter-selection, and using short homology arms, editing efficiency was high)
Enterococcus hirae
Gram positive, Mollicutes
Mycoplasma pneumoniae	GP35 from B. subtilis bacteriophage SPP1 (RecTEc homolog)	~10−2 for 1bp change, ~10−3–10−5 for large deletions with 40nt homology arms and ~12μg ssDNA	(Pinero-Lambea et al., 2020)
Yeast
Saccharomyces cerevisiae	RAD51(K342E) (strand exchange recombinase) coupled to RAD54 (helicase)	10−3–10−2 for single bp mutations using ~40–50nt homology arms, 2.5μM ssDNA	(DiCarlo et al., 2013)

^*Highest frequency reported when targeting the genome. Unless otherwise specified, any reported frequencies herein are expressed as described by Murphy (Murphy, 1998) as the portion of recombinants relative to the total viable cells after transformation of the donor linear DNA.

Table 3.

Homologous systems for dsDNA-based recombineering in different bacterial species.

Species	Recombineering system	Frequencies*	References
Gram negative, γ-proteobacteria
Pseudomonas aeruginosa	BAS from P. aeruginosa phage Ab31 (λ Redβα) RecETPsy (homolog of RecETEc) coupled to λ Gam or Plu Gam	~102 total recombinants for large deletions/insertions using 75 bp homology arms and 2–5 μg dsDNA with PluγTEPsy	(Yin et al., 2019)
Pseudomonas fluorescens		~20 total recombinants for large deletions/insertions using 75 bp homology arms and 2–5 μg dsDNA (targeting the genome) with RedγTEPsy
Pseudomonas putida		Up to ~103 total recombinants for large deletions/insertions using 100–150 bp, 1.5 μg (saturating amount) dsDNA with BAS (targeting a plasmid, not tested for targeting the chromosome)
Pseudomonas syringae		~102 total recombinants for large insertions using 75 bp homology arms and 2–5 μg dsDNA with BAS
	RecETPsy (homolog of RecETEc)	~10−6 for point mutations using ~90 bp homology arms, ~10−7 for large insertion/deletions using 80 bp homology arms (500–1000 bp gives 100x higher frequency), 100–500 ng dsDNA (1–2μg reduced frequency) with RecTEPsy	(Bao et al., 2012; Swingle, 2014; Swingle, Bao, et al., 2010)
Pseudomonas protegens		Efficiency not determined. Tested large deletion/insertion using 75 bp homology arms, 300 ng dsDNA RecTEPsy	(Yu et al., 2019)
Pseudomonas parafulva	RecETPsy (homolog of RecETEc) coupled to λ Gam or Plu Gam	Frequency not reported for targeting the genome. Tested using 75–100 bp, 1μg (saturating amount)	(Zheng et al., 2021)
Photorhabdus luminescens	Pluγβα (λ Redγβα homolog from P. luminiscens prophage)	>20 total recombinants for large insertions using 75 bp homology arms and 1 μg dsDNA	(Yin et al., 2015)
Xenorhabdus stockiae		>20 total recombinants for large insertions using 75 bp homology arms, 1 μg dsDNA
Vibrio natriegens	SXT Beta and Exo (λ Redβα homologs found in an SXT mobile element) coupled to λ Gam	Frequency not determined. Tested for a large insertion-deletion using 500 bp homology arms, 1μg dsDNA	(Lee et al., 2017)
Acinetobacter baumannii	RecAB (λ Redβα homolog) from A. baumannii IS-123 for dsDNA recombineering	~20–200 total recombinants (80–100% correct clones) for large deletion-insertions using 125 bp homology arms, 5 μg dsDNA	(Tucker et al., 2014; Tucker, Powers, Trent, & Davies, 2019)
Gram negative, β-proteobacteria
Burkholderiales	Redαβ7029 (λ Redβα) from Schlegelella brevitalea (previously known as Burkholderiales strain) DSM 7029	Up to ~400 total recombinants for large deletions of 20– 200kb (by insertion of a marker) and using 80–100 bp homology arms, 1 μg dsDNA in strain DSM7029. Also applied to other strains (frequency not reported)	(Wang et al., 2018)
Paraburkholderia megapolitana		~102 total recombinants for large deletions of 17 kb (by insertion of a marker) and using 100–200 bp homology arms, 1 μg dsDNA	(Zheng et al., 2020)
Burkholderia gladioli	BAS from P. aeruginosa phage Ab3 (homolog of λ-Redβα) coupled to λ Gam	~103 total recombinants for deletion-insertions using 50–100 bp homology arms (dsDNA amount not reported)	(H. Chen et al., 2021)
Burkholderia glumae	RecEThTJI49 from Burkholderia sp. TJI49 (homolog of RecETEc)	Up to ~103 total recombinants for insertions using 125 bp homology arms and 500 ng dsDNA	(Li et al., 2021)
Gram negative, α-proteobacteria
Rhizobium rhizogenes	ETRHI483 from Rhizobium sp. Root483D2 (homolog of RecETEc) coupled to Pluγ	<100 recombinants/μg DNA for large deletion-insertions using 80 bp homology arms and 4 μg dsDNA	(Bian et al., 2022)
Agrobacterium tumefaciens	ETRHI145 from Rhizobium sp. LC145 or RecETh1h2h3h4AGROB6 from A. tumefaciens str. B6 depending on the strain (homologs of RecETEc) coupled to Pluγ	<100 recombinants/μg DNA in A. tumefaciens C58 and <400 in A. tumefaciens EHA105, for large deletion-insertions using 80 bp homology arms and 1.5–2 μg dsDNA (saturating amount)
Gram positive, Actinobacteria
Mycobacterium smegmatis	Gp60 and Gp61 from Mycobacteriophage Che9c (homologs of RecETEc)	>100 total recombinants for large deletion-insertions using ~500 bp homology arms, 100 ng dsDNA	(van Kessel & Hatfull, 2007; van Kessel, Marinelli, & Hatfull, 2008)
Mycobacterium tuberculosis
Corynebacterium glutamicum	WP 010191133.1 and WP 010191134.1 from C. aurimucosum (homologs of RecETEc)	~103 total recombinants for large deletion-insertions using 800 bp homology arms, 1 μg dsDNA. Also tested other protein pairs, but the one from C. aurimucosum resulted in higher recombinants	(Chang et al., 2021)
Gram positive, Bacilli
Lactobacillus casei	LCABL 13040–50–60 (homolog of λ-Redγβα) coupled to Redγ	~100–102 total recombinants for large insertion/deletion (depending on the strain) using ~1 kbp homology arms and 4 μg dsDNA	(Xin et al., 2017)
Lactobacillus paracasei
Lactobacillus plantarum	Lp_0640–41–42 from prophage P1 operon from L. plantarum WCFS1 (Lp_0641 is a homolog of RecTL.reuteri)	~101–103 total recombinants for large insertion/deletions (depending on the strain and modification size) using ~1–1.4 kbp homology arms and 1 μg dsDNA	(Yang, Wang, & Qi, 2015)

^*Highest frequency reported when targeting the genome. Unless otherwise specified, any reported frequencies herein are expressed as described by Murphy (Murphy, 1998) as the portion of recombinants relative to the total viable cells after transformation of the donor linear DNA.

Figure 1. Overview of ssDNA recombineering in Shewanella species.

Design and procure ssDNA oligos to make the desired genetic changes. Obtain the plasmid pX2SW3Bet, which expresses the annealase W3 Beta under the arabinose inducible P_BAD promoter. Make electrocompetent Shewanella cells induced for the annealase and introduce the mutagenic ssDNA oligos by electroporation, to generate the targeted mutation/s. Following the necessary cell recovery period, select bacterial colonies on kanamycin, and analyze possible recombinants using PCR. Verify candidates with DNA sequencing. This process can be repeated for additional recombination cycles.

Figure 2. Overview of ssDNA recombineering coupled to CRISPR/Cas9 in Shewanella species.

For ssDNA recombineering coupled to a CRISPR/Cas9 counter-selection, design and procure ssDNA oligos to make the desired genetic changes. Obtain plasmid pX2C9pLacW3Beta, which encodes the W3 beta annealase under control of the constitutive P_Lac promoter, and Cas9 under control of the arabinose inducible P_BAD promoter. Modify the pACYC’ plasmid by inserting the desired sgRNA sequence for Cas9 targeting. Make electrocompetent cells and introduce the mutagenic ssDNA oligos and the sgRNA plasmid by electroporation. The annealase, which does not require induction, will generate the targeted mutations. Following the necessary cell recovery period, which allows time for the recombination to occur, plate the cells on kanamycin solid media supplemented with arabinose. The arabinose will induce the Cas9 protein, which will cleave non-recombinant DNA molecules, providing a counter-selection. Recombinants are analyzed using PCR and sequence verified. This process can be repeated for additional recombination cycles.

Table 4.

Bacterial Strains Used for Recombineering with W3 Beta & for using CRISPR/Cas9 as a counter-selection.

Strain ID (Gralnick Lab)	Description	Relevant Genotype	References
Shewanella strains
JG274	S. oneidensis MR-1	NA	Gralnick Lab
JG2150	S. oneidensis, lacZ (integrated downstream of glmS)
JG98	S. amazonensis SB2B
JG239	Shewanella sp. W3–18–1
JG3653	S. oneidensis, pBTBX2	Empty plasmid, KanR, PBAD	(Corts et al., 2019a, 2019b)
JG4140	S. oneidensis lacZ, pBTBX2		(Corts et al., 2019b)
JG4239	S. amazonensis, pBTBX2		(Corts et al., 2019a)
JG4123	S. oneidensis, pX2SW3Beta	KanR, PBAD-W3 Beta	(Corts et al., 2019b)
JG4127	S. oneidensis, lacZ, pX2SW3Beta		(Corts et al., 2019b)
JG4240	S. amazonensis, pX2SW3Beta		(Corts et al., 2019a)
JG3514	S. oneidensis, pX2Cas9	KanR, PBAD-Cas9	(Corts et al., 2019a)
JG3675	S. oneidensis, lacZ, pX2Cas9		(Corts et al., 2019a)
JG4234	S. amazonensis, pX2Cas9		(Corts et al., 2019a)
JG4250	S. oneidensis, pX2C9pLacW3Bet	KanR, PBAD-Cas9, PLAC-W3 Beta	(Corts et al., 2019a)
JG4251	S. oneidensis, lacZ, pX2C9pLacW3Bet		(Corts et al., 2019a)
JG4271	S. amazonensis, pX2C9pLacW3Bet		(Corts et al., 2019a)
JG4285	S. oneidensis, lacZ, pX2C9pLac26W3Bet**		(Corts et al., 2019a)
E. coli strains
JG3582	E. coli GM2163	dam− dcm−chloramphenicol resistant	CGSC# 6581
JG3680	E. coli GM1674	dam− dcm−	CGSC# 7971
JG4236	GM2163, [pX2Cas9] KmR	PBAD-Cas9 KmR	Gralnick lab
JG4249	GM2163, [pX2C9pLacW3Bet] KmR	PBAD-Cas9 PLac-W3Beta KmR
JG4229	GM1674 [pACYC’] CmR	plasmid vector for cloning gblocks encoding sgRNAs
JG4244	GM1674 [pACYC’-mtrAgRNA] CmR	mtrA sgRNA (for net deletion of mtrA in S. oneidensis MR-1) CmR
JG4246	GM1674 [pACYC’-mtrDgRNA] CmR	mtrD sgRNA (for mutating mtrD in S. amazonensis SB2B, see text) CmR

^*All plasmids shown here were maintained on cultures supplemented with antibiotic (see Table 8). Methylation-proficient and -deficient E. coli strains with the plasmids listed in this table can also be obtained from the Gralnick Lab.

^**This plasmid contains a shorter sequence of 26 bp between P_Lac and the RBS for better transcription. Note that cell viability may be reduced.

When successful, these recombination methods will allow introduction of single point mutations and small genetic alterations into the genomes of non-model bacteria. When used in combination with a CRISPR/Cas system employed as a counterselection against unmodified DNA sequences (Thomason et al., 2014), the repertoire of possible changes becomes much broader (see Alternate Protocol 1). Development of better genetic engineering tools for these less well characterized but important bacteria will improve our ability to modify their genomes for clinical and industrial use, as well as broaden our understanding of their biology.

The flowchart in Figure 3 provides guidance for the researcher in determining what steps to take, depending on the genetics available in the undomesticated bacterium of interest. Before undertaking recombineering, the researcher should determine recent advances of any genetic manipulations available for their microbe of interest. Figure 4 shows the genetic components and tools needed to successfully apply ssDNA and dsDNA recombineering to diverse species. Clearly, only microbes where the genome has been sequenced can be modified by recombineering, so that researchers can design linear DNAs as substrates for recombination and electroporation. The many references cited in this paper will help the reader with this step. Figure 5 depicts a systematic experimental approach for identifying and validating non-model recombineering systems. If a characterized phage recombination system is already available for the bacterium of interest, request it from the laboratory that developed the system and follow their suggested procedures. If a phage recombination system is unavailable, first try the λ Red and/or RecET_Ec systems; however, it is important to be aware of their limitations. An electro-transformation method is critical for successful recombineering, as is the availability of genetic tools for expression of phage recombineering systems. However, if these are not available, consider using oligo-mediated recombination initially, to help optimize electroporation conditions for linear DNA transfer into the bacterial cell. Table 5 lists the Gram-negative and Gram-positive strains in which this approach has been effective, including commonly made mutations, genes targeted, and recombination efficiency. Online resources to assist researchers in designing ssDNA oligos for recombineering or oligo-mediated recombination are provided in Table 6 and addressed in the Commentary.

Figure 3. Flow chart for determining approach to take for developing recombineering in a non-model bacterium.

Systems for species in which a recombineering system has been characterized should be prioritized for testing. Test the annealase alone for ssDNA recombineering before testing the annealase/exonuclease pair for dsDNA recombineering. Other distantly related systems can be tested as well if resources allow. For species with no known or functional recombineering system, researchers should first attempt to characterize a homologous system from or associated with the strain or species of interest. If no electroporation protocol is available for the host of interest, researchers may be able to develop a protocol either using plasmids or ssDNA. If genetic parts (inducible promoter, transcription terminator, low-copy plasmid ori, selective marker) are not available for the host of interest, the researchers will need to identify and characterize them. See Figure 4 for an overview of the genetic parts and tools needed.

Figure 4. Genetic parts and tools needed for annealase-independent oligo-mediated recombination, ssDNA recombineering, and dsDNA recombineering.

a) A means for introducing linear recombinogenic ssDNA or dsDNA into the cells is needed, preferably electroporation. b) The simplest approach is oligo-mediated recombination, which requires only ssDNA oligos. c) For ssDNA recombineering, expression of a phage annealase is also required. d) In addition to recombinogenic dsDNA, recombineering with dsDNA necessitates the expression of both an annealase and its related exonuclease. More details may be found in the text.

Figure 5. Experimental workflow for testing multiple recombineering systems in a non-model species.

Genomic sequence of the host of interest is required for designing recombinogenic ssDNA oligos and dsDNA donor templates. An efficient transformation protocol is also needed for introducing the linear DNA. Once a recombineering system has been identified using available databases and homology search functions such as BLAST, the researcher should first test the functionality of ssDNA recombination with the annealase (also indicated here as SSAP) of choice. Once ssDNA recombineering is established, the annealase (SSAP)/exonuclease pair can be co-expressed and tested for dsDNA recombination by attempting to introduce a selectable marker such as an antibiotic resistance gene. Confirm the insertion using PCR. As always, recombinants to be used in downstream applications should be sequenced. When multiple rounds of recombineering are performed, Illumina-based whole genome sequencing is ideal since prolonged laboratory handling can result in single nucleotide polymorphisms.

Table 5.

Hosts in which annealase-independent oligo-mediated recombination has been tested.

Species		Biosafety Level (BSL)*	Target genes	Frequencies**	Mutation	References
	Gram-negative, γ-proteobacteria
Escherichia coli K-12		1	rpsL, rpoB, galK	~10−7 for rpsL (0.3 μg oligo, 76 nt long) ~10−6 for rpoB (0.3 μg oligo, 75 nt long) ~10−4 for galK using saturating oligo concentration (10 μg, 70 nt long), ~10−5 with non-saturating amount (0.1μg, 70 nt long)	rpsLK88R oligo for StR, and rpoBP564L for RifR, galk amber codon 145 TAG>TAC for Gal+	(Swingle, Markel, Costantino, et al., 2010)
Salmonella typhimurium		2	rpsL, rpoB	~10−6 for both rpsL and rpoB (0.3 μg oligo, 76 and 75 nt long, respectively)
Shigella flexneri		2	rpsL, rpoB	~10−6 for both rpsL and rpoB (0.3 μg oligo, 76 and 75 nt long, respectively)
Pseudomonas syringae		2	rpsL	~10−5 with saturating oligo concentration (5 μg, 84 nt long), ~10−6 with non-saturating amount (0.5–1 ug). A similar efficiency was obtained with 1 ug rpsLK43R oligo and 4 ug of a non-homologous “carrier” oligo	rpsLK43R oligo for StR (4 bp change)	(Swingle, Markel, & Cartinhour, 2010; Swingle, Markel, Costantino, et al., 2010)
Pseudomonas putida		2	pyrF	~10−5 (3 μg oligo, 90–100 nt long)	Small insertions (9 bp), small deletions (100 bp) and 1 bp substitution with low MMR affinity: pyrFE50Stop	(Aparicio et al., 2016)
Shewanella oneidensis		1	lacZ	~10−4 (2.5 μg oligo, 170 nt long)	Small 10 bp changes to introduce premature stop codons	(Corts et al., 2019b)
Legionella pneumophila		2	rpoB	~10−5 (1nmol [~22ug], 71 nt long)	Point mutation: rpoBH541Y for RifR	(Bryan, 2011)
	Gram positive, Actinobacteria
Brevibacterium lactofermentum		1	rpsL	~10−5 with saturating oligo concentration (6 μg, 80 nt long), ~10−6 with non-saturating oligo concentration (0.6 μg, 80 nt long). A similar efficiency was obtained with 0.6 ug rpsLK43R oligo and 5.4 ug of a non-homologous “carrier” oligo	rpsLK43Rfor StR (4 bp change)	(Krylov et al., 2014)
Corynebacterium glutamicum
	Gram positive, Bacilli
Staphylococcus aureus		2	rpoB	~10−6 (200 pmol, unknown oligo length)	rpoBH481Y for RifR (6 bp changes)	(Penewit et al., 2018)

^*https://ehs.stanford.edu/reference/biosafety-levels-biological-agents (has some but not all)

^**Highest frequency reported when targeting the genome with lagging-strand targeting oligos (except for S. aureus, where the targeting strand was not specified). Unless otherwise specified, any reported frequencies herein are expressed as described by Murphy (1998) as the portion of recombinants relative to the total viable cells after transformation of the linear donor DNA.

Table 6.

Tools and resources useful for oligo design.

Tool	Purpose	Website	References
MODEST MAGE	Automatically designs single and multiple oligos *Useful for MMR inactivated background host only *Any genome can be analyzed	modest.biosustain.dtu.dk	(Bonde et al., 2014)
DoriC	oriC database *Also provides information about dif sites	tubic.org/doric/public/index.php/browse/bacteria	(Luo & Gao, 2019; Luo, Quan, Peng, & Gao, 2019)
OriFinder	oriC predictor *Also contains gene predictor ZCURVE 1.0 *Can predict oriC regions in some draft genomes *Any genome can be analyzed	tubic.tju.edu.cn/Ori-Finder/	(Guo, Ou, & Zhang, 2003; Luo et al., 2019)
ZCURVE 3.0	Gene predictor for annotating genes, which can help determine the leading/lagging strand of a genome *Any genome can be analyzed	guolab.whu.edu.cn/zcurve/	(Hua et al., 2015)
Geptop 2.0	Essential genes predictor, which can help in determining the leading strand of a genome *Any genome can be analyzed	guolab.whu.edu.cn/geptop/	(Wen et al., 2019)
‘leading-strand essential genes’	Database of the leading strand essential genes for 10 bacterial species	tubic.tju.edu.cn/ls-deg/index.html	(Lin, Gao, & Zhang, 2010)
UNAFold	Nucleic acid folding (secondary structure) prediction, useful for determining the folding energy of an oligo	idtdna.com/unafold/	(Markham & Zuker, 2008)
Snapgene and Benchling	For visualizing genomic DNA and differentiating between leading and lagging strands	snapgene.com benchling.com	N/A

CAUTION:

Some of the bacteria mentioned in this unit are Biosafety Level [2, 3] ([BSL-2, -3]) pathogens. The biosafety levels for some bacteria are indicated in Table 5. Follow all appropriate guidelines and regulations for the use and handling of pathogenic microorganisms.

STRATEGIC PLANNING

Choice of recombineering systems:

Adaptation of the λ Red and RecET_Ec systems for use in other microbes

λ Red has been broadly tested in other microbes (Table 1). Oligo-promoted recombineering mediated by either λ Beta or RecT_Ec has been successfully adapted for various γ-proteobacteria, as well as in Gram-positives from the class of actinobacteria and bacilli (see Table 1 and references therein). λ Red mediated dsDNA recombineering has also been successful in several hosts, mostly γ-proteobacteria, but also β-proteobacteria (see Table 1 and references therein). On the other hand, dsDNA-recombineering mediated by RecET_Ec has been demonstrated only in P. putida and Z. mobilis and was inefficient for the latter (Table 1). The λ Red system is ~10-fold better than the RecET_Ec system for ssDNA and dsDNA recombineering targeting circular replicons in E. coli (Datta, Costantino, Zhou, & Court, 2008; Fu et al., 2012; Thomason, Costantino, & Court, 2016; Wang et al., 2016; Wang et al., 2018), while the RecET_Ec system excels at intracellular assembly of linear dsDNA molecules (Fu et al., 2012; Sawitzke, Costantino, Hutchinson, Thomason, & Court, 2022). In general, for adaptation of the well-known E. coli λ Red or RecET systems, modifications such as longer homology arms or higher dsDNA concentrations were needed to obtain high numbers of recombinants (Chang, Wang, Su, & Qi, 2019; Hossain, Thurlow, Sun, Nasrin, & Liles, 2015; Lesic & Rahme, 2008). Even with those changes, the recombineering frequency was still poor in some cases (Hossain et al., 2015; Hu et al., 2014). As a general criterion, the probability of the well-characterized λ Red or RecET_Ec systems working in a host of interest will most likely depend on the organism’s relatedness to E. coli.

Novel recombineering systems from non-model species

Efforts have been made to identify and characterize native phage homologous recombination systems in both Gram-positive and -negative bacteria for use in recombineering. Recent advances in this area are accelerating our understanding of ssDNA recombineering and its application in non-model hosts. Most of these recently discovered systems have similarities to either the Red or RecET systems. (See Tables 2 and 3 for ssDNA- and dsDNA-based recombineering, respectively). The γ-proteobacteria and Bacilli constitute the largest groups for which new systems have been developed. Native systems also exist for the Gram-negative β-proteobacteria Burkholderiales and α-proteobacteria Rhizobiales, and Gram-positive species such as Clostridia, Mollicutes and Actinobacteria (including Mycobacteria) (Tables 2 and 3).

Choice of linear DNA substrate:

Test ssDNA first:

When attempting to develop recombineering in a non-model bacterium, it makes sense to first test ssDNA recombineering (Thomason et al., 2005). Recombineering with ssDNA oligos has been successfully used to make point mutations and small genetic alterations at high efficiency in various non-model bacteria (see Table 2). Although large deletions can also be made with ssDNA oligos, the frequency of recombination is much lower, and a selection/counter-selection system may be necessary for successful target modification. As in E. coli, ssDNA recombineering in other microbes requires only the annealase function (Aparicio, de Lorenzo, & Martinez-Garcia, 2019; Corts, Thomason, Gill, & Gralnick, 2019b; S. Datta et al., 2008; Swingle, Bao, Markel, Chambers, & Cartinhour, 2010; van Pijkeren & Britton, 2012; Van Pijkeren, Neoh, Sirias, Findley, & Britton, 2012), and involves a simpler mechanism than dsDNA recombineering (Ellis, Yu, DiTizio, & Court, 2001). The successful recombineering of non-model bacteria with ssDNA oligonucleotides vs. dsDNA substrates (see discussion below), may be because the SSAPs that promote recombination with ssDNA oligos have greater flexibility for use in non-native systems. Datta, Costantino, Zhou & Court (2008) demonstrated functionality of a variety of annealases from diverse sources for ssDNA recombineering in E. coli, and annealases from various species share more sequence similarity than do their exonuclease counterparts (Binder, Siedler, Marienhagen, Bott, & Eggeling, 2013). An annealase derived from a phage or prophage naturally resident in the bacterium to be modified is most likely to yield high recombination frequencies. However, it is also worthwhile to test multiple annealases since some do perform quite well in heterologous hosts.

As mentioned above, if no phage annealase-dependent system is available, the use of oligo-mediated recombination should be tested, especially when coupled with a CRISPR/Cas counter-selection (see Alternate Protocol 1). Observations made in E. coli pertaining to oligo recombination such as lagging-leading strand bias and the importance of evading mismatch repair also apply in non-model organisms (Corts, Thomason, Gill, & Gralnick, 2019b; van Kessel & Hatfull, 2008; Wang et al., 2018). The section entitled Critical Parameters below contains a discussion of these issues.

Modifying with dsDNA:

In general, the paired exonuclease-SSAP functions of λ Red and RecET required for dsDNA recombination have a relatively narrow species range, probably due to host-specific interactions between the recombination proteins and endogenous host replisome factors (Datta et al., 2008; Li, Thomason, Sawitzke, Costantino & Court, 2013; Poteete, 2013; Thomason, Costantino & Court, 2016). Attempts at dsDNA recombination using these SSAP/exonuclease pairs in non-native organisms have been largely unsuccessful outside of the γ-proteobacteria (see Table 1 and references therein). An additional complexity of dsDNA recombineering is that intracellular linear dsDNA is subject to in vivo degradation by cellular nucleases such as RecBCD; therefore, inclusion of the anti-RecBCD Gam function may improve recombinant frequencies (Datta et al., 2008). Thus, we strongly recommend that a researcher start by testing ssDNA recombination using available annealases before attempting dsDNA recombineering.

Other important parameters to consider before attempting recombineering in non-model organisms include validation of adequate expression of the particular annealase in vivo, designing the linear DNA substrates, and optimization of the electroporation parameters of the non-model bacterium. These topics are covered in the Commentary.

BASIC PROTOCOL 1

ssDNA-RECOMBINEERING IN SHEWANELLA SPECIES

This protocol describes ssDNA-recombineering in Shewanella oneidensis MR-1 and Shewanella amazonensis SB2B which are both BSL-1 organisms. Note that this approach has not been tested with other Shewanella species. The annealase W3 Beta from Shewanella sp. W3–18–1 is used for making point mutations or small insertions/deletions (indels), and is expressed from plasmid pX2SW3Beta, under control of the inducible arabinose promoter, P_BAD, regulated by the arabinose repressor (AraC). This plasmid contains a broad host range origin of replication (pBBR1) and a kanamycin (Km^R) marker for selection (Corts et al., 2019b). An example of a control recombineering reaction, modification of the recessive rpsL gene encoding the ribosomal subunit S12, to streptomycin resistance, is included (Corts et al., 2019b). Figure 1 depicts an overview of our approach.

While Basic Protocol 1 is similar to that presented in the E. coli recombineering Current Protocols article (Thomason et al., 2014), it differs in some respects. Recently, it was shown that electroporation and recombineering may be performed at room temperature in E. coli without compromising the effectiveness of DNA transformation or recombineering efficiency (Tu et al., 2016), and we have confirmed this observation for Shewanella (Corts et al., 2019a, b). We found an ~10–30-fold increase in plasmid transformation efficiencies when Shewanella electrocompetent cells were prepared at room temperature rather than preparing competent cells on ice. Thus, the steps for competent cell preparation and electroporation are performed at room temperature, unlike the procedure in Thomason et al. (2014). The method of washing the cells also differs; in this protocol, the cells are washed in smaller volumes using microcentrifuge tubes. The washing steps in the two protocols are likely interchangeable and the choice of which to use may depend on what equipment is available (see Commentary for additional considerations). The protocol presented here can also be used for oligo-mediated recombination experiments by omitting the recombineering plasmid (Corts et al. 2019a; 2019b).

To test ssDNA recombineering and/or oligo-mediated recombination in S. oneidensis MR-1 and S. amazonensis SB2B, researchers may modify the rpsL gene using ssDNA oligos listed in Table 7, as was done in Corts et al., (2019b). These oligo sequences are designed to introduce four bp changes containing the K43R mutation (AAA-> CGG for S. oneidensis, AAG->CGG for S. amazonensis) and a synonymous mutation in P42 (CCT- > CCA for both species). This recessive K43R mutation allows a positive selection for streptomycin resistance. The base changes to make the mutation are centered in the oligo, with ~40 nt of homology on either side of the modification. The additional P42 wobble mutation allows evasion of the bacterial host methyl-directed mismatch repair (MMR), which can substantially lower the recombination efficiency of single-strand DNA recombination (Costantino & Court, 2003; Sawitzke et al, 2011). The effect of MMR on ssDNA recombination is discussed in the Commentary, in the section “Critical parameters affecting oligonucleotide-promoted recombineering.” Both oligos are designed to target the lagging strand of the S. oneidensis and S. amazonensis genomes, respectively, and their sequences are given in Table 7, indicated as SO and SAMA. We also provide an oligo sequence that researchers can use to delete the mtrA gene in S. oneidensis MR-1. This lagging-strand oligo contains 160 nt of flanking homology, 80bp on either side of the open reading frame.

Table 7.

DNA substrates used in BASIC PROTOCOL 1 and ALTERNATE PROTOCOL 1

ssDNA oligos used for recombineering in BASIC PROTOCOL 1
Oligo designation	DNA sequence	Intended use
SO-rpsLK43R	5’ gac gca cac gag cta ctt tac gta gtg cag agt tag gtt tCC GTg ggg tag ttg tgt aca cac gtg tac aaa cac cac gct ttt gt 3’	lagging strand oligo to introduce streptomycin resistant rpsL K43R mutation and the P42 wobble mutation into S. oneidensis MR-1
SAMA-rpsLK43R	5’ aac gca cac gag cta ctt tac gca gtg cag agt ttg gtt tCC GTg ggg cgg tgg tgt aca cac gag tac aaa cac cac gct tct gt 3’	lagging strand oligo to introduce streptomycin resistant rpsL K43R mutation and the P42 wobble mutation into S. amazonensis SB2B
SO-mtrA-KO	gtg aaa atg taa ttt gcc caa gca ggg gga gct cgc tcc ccc ttt ctt gaa ttt tgt tgg gac aaa ttg gga agc cta tt/g gag acg aga aaa tga aat tta aac tca att tga tca ctc tag cgt tat tag cca aca cag gct tgg ccg tcg ctg ctg a 3’	lagging strand oligo to delete mtrA gene in S. oneidensis MR-1
gblocks used for cloning in ALTERNATE PROTOCOL 1
SO-mtrA-gRNA	5’ ATG TTC CGG ATC TGC ATC GCt tta cac ttt atg ctt ccg gct cgt atg ttg tgt gga acg agc cga tga tca ctt tgt ttt aga gct aga aat agc aag tta aaa taa ggc tag tcc gtt atc aac ttg aaa aag tgg cac cga gtc ggt gct ttt ttt CAG GCA TTT GAG AAG CAC ACG G 3’	gblock sequence to target S. oneidensis MR-1 mtrA gene for Cas9 cutting
SAMA-mtrD-gRNA	5’ ATG TTC CGG ATC TGC ATC GCt tta cac ttt atg ctt ccg gct cgt atg ttg tgt gga ctg gag cgc ttg tgg ata tgt ttt aga gct aga aat agc aag tta aaa taa ggc tag tcc gtt atc aac ttg aaa aag tgg cac cga gtc ggt gct ttt ttt CAG GCA TTT GAG AAG CAC ACG G 3’	gblock sequence to target S. amazonensis SB2B mtrD gene for Cas9 cutting
Cloning primers used in ALTERNATE PROTOCOL 1 *
ADC7	5’ ATG TTC CGG ATC TGC ATC GC 3’	forward primer to amplify sgRNA gblocks, Tm=68°C
ADC8	5’ CCG TGT GCT TCT CAA ATG CCT G 3’	reverse primer to amplify sgRNA gblocks Tm=68°C
ADC9	5’ GCG ATG CAG ATC CGG AAC AT 3’	forward primer to amplify pACYC’ for sgRNA cloning Tm=68°C
ADC10	5’ CAG GCA TTT GAG AAG CAC ACG G 3’	reverse primer to amplify pACYC’ for sgRNA cloning Tm=69°C
ADC11	5’ TGA GAG TCA ACG CCA TGA GC 3’	For sequencing sgRNAs after cloning Tm=68°C

^*Tm determined using NEB Tm calculator (https://tmcalculator.neb.com/#!/main).

Materials

Oligonucleotide primers with ~40 bases flanking homology to the target site suspended in dH2O to a concentration of 0.5 μg/μl and stored frozen at −20°C until used. Longer homology arms may be needed for larger edits (Corts et al., 2019a).

Primers needed for control recombinations are given in Table 7.

Shewanella strains expressing W3 Beta annealase from plasmid pX2SW3Beta and control strains containing a control plasmid lacking W3 Beta (pBTBX2) (see Table 4, strains available from the Gralnick Laboratory).

Chemically competent dam⁻ dcm⁻E. coli cells from New England Biolabs (NEB) (https://www.neb.com) Catalog # C2925I. Alternatively, obtain GM2163 (CGSC#6581, chloramphenicol resistant) and GM1674 (CGSC#7971) from the Gralnick laboratory.

30°C incubator for petri plates

30°C shaking incubator (either air or water)

20 ml sterile glass culture tubes (Thomas Scientific, cat. no. 1160X21)

1.7 ml sterile microcentrifuge tubes (Fisher Scientific, cat. no. 07–200–184)

2 ml sterile microcentrifuge tubes (Fisher Scientific, 14–222–180)

QIAprep Spin Miniprep Kit (Qiagen, cat. no. 27104)

DNA quantification instrument, for example Nanodrop or Qubit

UV/Vis Spectrophotometer (ie. Thermo Scientific Genesys 20, cat. no. 4001)

0.1-cm electroporation cuvettes (Bio-Rad, cat. no. 1652083)

Electroporator (Bio-Rad MicroPulser Electroporator, cat. no. 1652100)

P100, P100 and P10 pipettes and sterile tips. Aerosol resistant tips are highly recommended.

Table-top centrifuge (ie. Eppendorf^™ Centrifuge 5425, Fisher Scientific, cat. no. 13–864–455).

Sterile Glass beads, diam. ~5 mm for plating (or sterile spreaders) (Fisher Scientific, cat. no. 50–444–635).

Parafilm for sealing the tubes when recovering cells (Fisher Scientific, S37441).

Materials for agarose gel DNA analysis

Antibiotics and Supplements (see Table 8)

Table 8.

Antibiotic and Supplements Used for Recombineering in Shewanella.*

Antibiotic/Supplement	Concentration	Solvent	Sterilization method	References
Kanamycin (Fisher Scientific, cat. no. AC450811000)	50 μg/ml for both S. oneidensis and S. amazonensis	Sterile water	Using a 0.2 μm syringe filter	(Corts et al., 2019a, 2019b)
Chloramphenicol (Fisher Scientific, cat. no. AC227921000)	10 μg/ml for both S. oneidensis and S. amazonensis	70–100% ethanol	na
5-Bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal, Fisher Scientific, cat. no. FERR0402)	60 μg/ml for S. oneidensis	Dimethyl-formamide (DMF)	na
Spectinomycin (Fisher Scientific, cat. no. ICN15899305)	50 μg/ml for S. oneidensis and 100 μg/ml for S. amazonensis	Sterile water	Using a 0.2 μm syringe filter
L-Arabinose (Fisher Scientific, cat. no. AAA1192130)	20 mM for W3 Beta expression 0.8 M for Cas9 expression (for both S. oneidensis and S. amazonensis)	Sterile water	Using a 0.2 μm syringe filter

^*Usually, concentrated antibiotic stock solutions are made at 1000-fold the working concentration and used at 1μl/ml. Store all antibiotic stocks in the dark at −20 °C. Tetracycline in stock solution, broth, or agar must always be stored in the dark and covered with aluminum foil. Antibiotics dissolved in ethanol do not need sterilization. Do not use antibiotics that have passed their expiration dates to make stocks, as these may be unstable and give variable results. Antibiotic concentrations were reduced to Km 25 μg/ml and Cm 6.5 μg/ml when Shewanella strains harbored two plasmids (Corts et al., 2019a).

Also needed are the following reagents (see recipes in Reagents and Solutions):

LB liquid medium

M9 minimal salts solution

Washing buffer, 1 M sorbitol pH 7.6

LB solid agar plates with desired antibiotic for selection, and without antibiotic to enumerate viable cells

Prepare DNA for transformation

• 1
  Design and obtain the ssDNA oligos. See the section in the Commentary entitled Oligo Design Tools for Maximal Recombination Efficiency for help in designing oligos. Oligos of up to 100 bp can be ordered lyophilized from Integrated DNA Technologies (IDT) (https://www.idtdna.com/pages). Longer oligos, up to 200 bp, can be purchased as ultramers.

  For larger genomic changes in S. oneidensis, such as 1 kbp gene deletions, longer homology arms of 80 bp improve recombination efficiency (Corts et al., 2019a).

• 2Suspend oligo DNA in dH₂O to a concentration of 0.5 μg/μl. Store at 4°C for short-term storage, or −20°C for long-term storage.
• 3Prepare aliquots of 2.5 μg (5 μL of the 0.5 μg/ul DNA stock) of the ssDNA oligo into as many 2 ml microcentrifuge tubes needed.

Prepare Shewanella cultures

• 4
  Inoculate desired Shewanella strains containing the pX2SW3Beta plasmid from a single colony into 5 ml LB medium in 20 ml sterile glass culture flasks. Shake cultures overnight (~16 hours) at 30°C, ~200 rpm. Make sure to add kanamycin at 50 μg/ml (i.e. Kan-50) to the culture for maintaining the plasmid.

  If desired, a strain with plasmid pBTBX2, which lacks the annealase, can be used in parallel as a control reaction. The pBTBX2 plasmid will not give W3 Beta-mediated recombinants. For oligo-mediated recombination lacking any recombinase function, use the desired strain.

  The researcher can transform any desired Shewanella strains with the plasmids themselves. While not strictly necessary, it is preferable to isolate plasmids destined for transformation into Shewanella either from another Shewanella strain or from a methylation-deficient (dam⁻ dcm⁻) E. coli strain. Either approach will yield plasmid DNA lacking E. coli methylation, thus improving subsequent transformation of Shewanella substantially (as much as 10³-fold) (Corts et al., 2019b). (However, transformants should still be obtained even if the plasmid DNA has E. coli methylation). Suitable chemically competent dam- dcm- E. coli cells are available from New England Biolabs (NEB) (https://www.neb.com) Catalog # C2925I. To obtain unmodified plasmid DNA, use the commercially available dam⁻ dcm⁻ E. coli and follow the manufacturer’s instructions for transformation. Plate suggested dilutions of the transformed cells on an L Kan-50 petri plate and incubate the plate at 30°C 24–36 h until colonies appear. When colonies arise on the plate, inoculate 5 ml L Kan-50 broth with a single colony, grow the culture to saturation, and isolate the unmethylated plasmid DNA using a plasmid DNA purification kit such as Qiagen.

• 5
  Dilute overnight cultures to a starting OD₆₀₀ of ~0.08 in LB medium supplemented with 20 mM arabinose to induce expression of W3 Beta. To do this, first measure the optical cell density (OD₆₀₀) of the overnight cultures using the UV/Vis spectrophotometer. (Make sure to dilute your samples (~10x) before measuring the optical density of the overnight cultures, since inaccurate readings will be obtained at an OD₆₀₀>1). Depending on the initial OD₆₀₀, the overnight cultures may need to be diluted ~50 to 100-fold for growing the cell culture. Normalizing the OD₆₀₀ ensures that the various cultures (of the same strain) grow at the same rate and density. Dilute the cultures at least 50-fold, so that the strains will undergo logarithmic growth.

  Also prepare a separate culture without arabinose as a control reaction. Grow enough cell volume for at least two transformations each; with and without ssDNA (control). One electroporation requires 3 ml cell culture, and monitoring cell density will require additional volume. For oligo-mediated reactions, arabinose is not required because no annealase will be expressed.

• 6Place cultures in a shaking incubator at 30°C until cells reach an OD₆₀₀ ~0.4–0.5 (approximately 65–75 min). Make sure not to overgrow the cells, since recombineering efficiency declines at higher cell densities (see Critical Parameters for Successful Recombineering). Be aware that a larger dilution in step 5 may mean a somewhat longer incubation time to reach an optimal OD₆₀₀.

Make Shewanella electrocompetent cells

• 7
  Once cells reach an OD₆₀₀~0.4–0.5, transfer 3 ml cells into 3 separate and appropriately labeled 1.5 ml centrifuge tubes (1 ml per microcentrifuge tube). Pellet cells by centrifugation at 7607 × g for 1 min (higher speeds may cause cell lysis).

  Induced and uninduced cultures should be made in parallel for comparison. The conical shape of the 1.5 ml microcentrifuge tubes helps to avoid loss of the pellet. Alternatively, as described in Thomason et al. (2014), cells can be processed in larger volumes for multiple transformations.

• 8
  Pour off supernatant and combine the 3 cell pellets together in one ml of sorbitol washing buffer. Suspend the pellets using the larger P1000 tips. Centrifuge the samples again as in step 7.

  All materials i.e., electroporation cuvettes and washing buffer should be kept at room temperature. For S. oneidensis MR-1, when the cells are harvested in exponential phase, we found that washing them with Sorbitol washing buffer yields a ~250-fold higher transformation efficiency than does washing them with 10% glycerol (Corts et al., 2019a). When suspending cells in the Sorbitol buffer, use gentle pipetting with a P1000 ~10 times. When decanting the buffer after centrifugation, take care to not lose the soft pellet.

• 9Pour off supernatant and add another 1 ml of the sorbitol washing buffer. Gently resuspend as in step 8. Centrifuge the samples again as in step 7.
• 10Repeat step 9 again for a total of three washes. Decant the sorbitol and leave ~60–70 μl in which to suspend the cells.
• 11Suspend cells gently by shaking the tube and pipetting with a P100 several times.

Introduce ssDNA oligos by electroporation

• 12
  Immediately add the suspended cells into the DNA sample prepared in the 2 ml tube (you may keep the tubes and use them for cell recovery). Mix cells thoroughly with the DNA by swirling gently with a pipette tip. Be sure to include the necessary control reactions of uninduced cells with and without ssDNA. These critical controls are used to determine recombination efficiency. (See E. coli. Basic Protocol 1 in the Current Protocols article by Thomason et al. (2014), Introduce DNA by electroporation, for more information on necessary controls).

  It is important to mix the cells and DNA adequately. Inadequate mixing will lower recombinant frequency.

• 13
  Rapidly transfer the mixture into a 0.1-cm cuvette without introducing bubbles and electroporate at 1.2 kV.

  A 0.1 cm cuvette holds ~100 μl maximum volume. 0.2 cm cuvettes require higher voltages and we do not recommend them. Time constants should be ~5 msec. If the time constant is not at least 4, recombineering efficiency may be impaired. Electroporator settings: 10 μF, 600 Ω. Tap the cuvette gently on the benchtop prior to pulsing the sample to get rid of any bubbles.

• 14
  Quickly add 1 ml of sterile LB into the cuvette and mix with gentle pipetting. Transfer the cells into the empty 2 ml microcentrifuge tubes.

  The cells are very fragile after electroporation, and it is critical to add the LB immediately. Even a short delay will impair cell viability and reduce the chance of recovering recombinants.

  2 ml microcentrifuge tubes allow for more aeration than 1.5 ml tubes, which helps obtain a higher transformation efficiency in S. oneidensis MR-1 (data not shown). Alternatively, the cultures can also be transferred to larger culture tubes as in E. coli Thomason et al. (2014).

• 15
  Parafilm the microcentrifuge tube caps (sometimes they pop open) and transfer the tubes into an Erlenmeyer flask (to avoid losing the tubes in the shaker). Put the flask in the incubator for the cells to recover at 30°C with shaking at ~200 rpm.

  Alternatively, place the tubes in a test tube rack, wrapped in saran wrap, and position the tube rack such that the tubes are nearly horizontal during shaking, which improves aeration.

• 16
  Let the cells recover for the appropriate time. If the intention is to plate total viable cells on solid agar and screen for recombinants, such as when targeting lacZ, use a 30 minute outgrowth. If targeting rpsL or rpoB, allow the cells to recover for at least ~3.5–4 hours for S. oneidensis MR-1 (Corts et al., 2019b). Also see discussion of outgrowth time in Critical Parameters. Note that outgrowth times may need to be adjusted to optimize for strains with a slow growth rate.

  A 30-minute outgrowth allows time for the cells to recover from the electric shock but does not allow complete chromosome segregation of the multiple chromosomes within individual cells. This short outgrowth time gives a somewhat higher apparent recombination frequency, since a single colony will initially contain both parental and recombinant chromosomes. This is easily detectable using an assay such as the white-blue screen with lacZ, when blue/white sectored colonies can be seen on X-gal solid medium.

  A long outgrowth allows additional time for the cells to replicate and completely segregate their chromosomes, which is necessary to allow expression of the recessive rpsL or rpoB allele that has been introduced with the oligo. Longer recovery times will decrease the apparent recombineering efficiency about six-fold (Costantino & Court, 2003; Pines, Freed, Winkler & Gill, 2015).

Determine cell titer

• 17Make 1:10 serial dilutions in LB or a M9 minimal salts solution out to 10⁻⁵.
• 18
  Spread cells on agar plates with appropriate antibiotics/supplements.

  If targeting a gene that provides a screen for isolation of edited cells, such as lacZ, plate all cells from the 10⁴-fold dilution (100 ul per 10 cm plate, ~10 plates) to identify the recombinants. When altering lacZ, 5-Bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-Gal) must be added to the plates (see Table 8). The wild type lacZ gene gives blue colonies on X-gal solid agar, while the mutant gene gives white colonies. Sectored colonies may arise on the agar plates due to the short recovery time. If so, re-streak these on an agar plate to obtain pure genotypes with the desired phenotype. When targeting a gene that allows a positive selection after the modification is made, such as rpsL or rpoB, plate 100 ul of the 10⁻¹ to 10⁻⁴ serial dilutions to yield single colonies. The antibiotics spectinomycin and rifampin, respectively, are used to select these mutations (Table 8).

• 19Incubate plates at 30°C for ~24–36 h until colonies appear. Pick single colonies and purify them on the same solid medium. Four colonies can easily be purified per petri plate by dividing the plate into quadrants. Number the quadrants and retain these plates for colony analysis.

Analyze and confirm recombinants

• 20Once recombinant clones are identified, the genomic modification can be confirmed using colony PCR. See Support Protocol 3, Screening for oligo recombinants by PCR in Thomason et al, (2014). Obtain a pair of primers (~24-mers) with similar annealing temperatures that will allow amplification of the recombinant region as a reasonably sized PCR product (ideally ~0.5–2 kb). Use these flanking primers and colony PCR to make a PCR product of the region encompassed by the ssDNA oligo sequence and the genetic modification. Use the same primer pair to create a control PCR product from the parental (nonrecombinant) strain. Analyze the PCR products by agarose gel electrophoresis and compare the sizes of the recombinant band with the control band. Researchers should see the expected difference in size between the recombinants and the non-recombinant control. If the PCR product from the control does not give a product different in size than the recombinant, researchers can perform PCR verification using each flanking primer paired with an internal primer that only anneals to the recombinant sequence. In this case, the two recombinant flanks are amplified, and only recombinant colonies should produce PCR products. An example of this type of analysis is MAMA PCR (Cha, Zarbl, Keohavong, & Thilly, 1992) which can be used for confirming small insertions or point mutations. Here the 3’ end of one primer is designed to anneal to the change that was introduced by recombineering. This will ensure that this primer will amplify a PCR product from the edited genomes but not from the wild-type cells. (It may be necessary to optimize PCR conditions so that this primer pair does not give a PCR product with non-recombinant bacteria).
• 21it is important to sequence the final construct, especially those regions spanned by the oligo, since unwanted mutations arise during chemical synthesis of the ssDNA oligos that may be transferred to the genome during recombination (Oppenheim, Rattray, Bubunenko, Thomason, & Court, 2004).

ALTERNATE PROTOCOL 1

SSDNA-RECOMBINEERING COUPLED TO CRISPR/Cas9 IN SHEWANELLA SPECIES

This protocol describes ssDNA-recombineering coupled to CRISPR/Cas9 used as a counter-selection tool in Shewanella species. It is based on the procedure in the Basic Protocol 1, with the additional complexity of CRISPR/Cas9 expression following recombineering. The combined procedure facilitates isolation of recombinants at a high efficiency, optimally nearly 100%, and is useful when making larger genetic modifications or those for which there is not a selection. In this protocol, the recombination occurs first, and the CRISPR/Cas9 system is targeted to cleave non-recombinant chromosomes by plating the recombinant cells on solid agar containing arabinose, which induces Cas9. Before beginning this procedure, it is important for the researcher to familiarize themselves with the essential elements of CRISPR/Cas targeting (Newman & Ausubel, 2016). Recombination is mediated by the W3 Beta annealase, which is required for making large genome modifications. Figure 2 depicts an overview of the approach. For making small changes of a few bases, the endogenous oligo-mediated recombination activity present in Shewanella can also be used (Corts et al., 2019a). If the researcher chooses to use oligo-mediated recombination, a Shewanella strain expressing only Cas9 is sufficient (i.e. strains JG3514 and JG4234 in Table 4). See the section in the Commentary entitled Oligo-mediated Recombination in Bacteria. It is critical that the DNA used for recombineering remove the three nt PAM site (NGG), which is required for Cas9 targeting. If the PAM site is not removed in the recombinants, Cas9 will be targeted to cleave recombinant chromosomes as well as non-recombinants.

The annealase-mediated recombination protocol uses plasmid pX2C9pLacW3Bet. This plasmid expresses the cas9 gene from Streptococcus pyogenes (spCas9) under the inducible arabinose promoter (P_BAD), and the W3 Beta annealase gene, expressed constitutively from the P_Lac promoter (Corts et al., 2019a). pX2C9pLacW3Bet has a broad host range origin of DNA replication, pBBR1, and a kanamycin (KmR) marker for selection.

A second plasmid, pACYC-xxxgRNA, expresses the sgRNA that targets Cas9 to non-recombinant chromosomes; this plasmid is introduced into the cells by co-electroporation with the ssDNA recombinogenic oligos. pACYC-xxxgRNA contains the single guide RNA (sgRNA) chimera, which is expressed constitutively from the P_Lac promoter. The ‘xxx’ represents the intended gene being targeted. The chimera includes a 20 nt region for binding to the desired genomic target, the Cas9 binding hairpin, and a transcriptional terminator from S. pyogenes. Note that the researcher must design and insert the sgRNA into the vector pACYC’, which contains a low copy origin of replication (p15a) and a chloramphenicol (Cm^R) marker for selection (see Table 4). The pACYC’ plasmid is a modified version of pACYC184 (Corts et al., 2019a). For guidance in designing sgRNA, see (Brazelton et al., 2015; Mohr et al., 2016; Newman & Ausubel, 2016). The articles from Takara Bio on “How to design sgRNA sequences” and “Choosing the right tool for designing guide RNAs” are also excellent resources.

Various bacterial strains for this protocol are available from the Gralnick laboratory. S. oneidensis and S. amazonensis strains JG4250 and JG4271, respectively, contain the pX2C9pLacW3Bet plasmid. A strain of S. oneidensis with an in-frame insertion of E. coli lacZ containing the recombineering plasmid is also available (JG4251). Several control strains harboring the pX2Cas9 plasmid that lacks W3 Beta are also available; these are S. oneidensis (JG3514); S. oneidensis with lacZ (JG3675) and S. amazonensis (JG4234).

As examples of using CRISPR-Cas9 as a counter-selection, we have provided DNA sequences for gblocks (see Table 7) enabling Cas9 targeting of the wild type mtrA gene in S. oneidensis MR-1 and the mtrD gene in S. amazonensis SB2B. To couple the Cas9 targeting with recombineering, for S. oneidensis MR-1 use the SO-mtrA-KO oligo to delete the mtrA gene; and for S. amazonensis SB2B, use the SAMA-mtrD oligo to make a mutation in mtrD that replaces 5 bp (5’ CCGAT 3’) of the mtrD gene with a 10 bp sequence (5’ TAATAAGTAA 3’).

Additional Materials (also see Materials from BASIC PROTOCOL 1.)

Example DNA substrates are detailed below and in Table 7. gblocks are obtained from IDT.

Bacterial strains are available from the Gralnick laboratory (see Table 4):

Shewanella strains expressing Cas9 and W3 Beta annealase from the pX2C9pLacW3Bet plasmid, and control strains expressing Cas9 but lacking the W3 Beta function (plasmid pX2Cas9)

The E. coli dcm- dam- strain JG4229 harboring pACYC’, used for cloning of the sgRNA sequence.

Q5® High-Fidelity DNA Polymerase (NEB, cat. no.M0491S)

PCR purification kit (ie. Qiagen QIAquick PCR Purification Kit, 28104)

Enzyme DpnI (NEB, cat. no. R0176S)

NEBuilder® HiFi DNA Assembly Master Mix (NEB, E2621S/L/X)

PCR Thermocycler (ie. Eppendorf Mastercycler® X50, cat. no. 2231000852)

Obtain DNAs required for the procedures

• 1Follow instructions in BASIC PROTOCOL 1 to obtain the recombinogenic ssDNA oligos. Remember to design the oligos so that the PAM site is removed in the recombinant species.
• 2
  Order sgRNA sequences as dsDNA gblocks from IDT and clone them into pACYC’ by Gibson Assembly, as described by the manufacturer.

  Design the sgRNA, using the references provided above for guidance. To allow Cas9-mediated recognition of the target, SpCas9 needs the Protospacer Adjacent Motif (PAM) site, NGG, positioned at the 3′ end of the DNA target, in addition to the sgRNA (20 nt target guide sequence-NGG). The PAM site is not included in the sgRNA (Jinek et al., 2012; Cui & Bikard, 2016). Before deciding on a final sgRNA design, a comprehensive BLAST search for the 10 nt seed sequence next to the PAM site (full sequence: 5′-N10-NGG-3′, N being any nt) should be done to determine possible off-targets throughout the genome. Any candidate sgRNA that matches more than one genomic locus should be eliminated.

The unique 20 nt spacer sequences for the mtrA and mtrD genes are shown below.

SO-mtrA-gRNA

20nt spacer sequence + PAM (5’-3’): AAC GAG CCG ATG ATC ACT TTT GG

SAMA-mtrD-gRNA

20nt spacer sequence + PAM (5’-3’): ACT GGA GCG CTT GTG GAT ATC GG

The gblock DNA sequences that can be used in control reactions to target the mtrA gene in S. oneidensis MR-1 and the mtrD gene in S. amazonensis SB2B are provided in Table 7. These sequences contain the user-programmable 20 nt spacer (shown underlined and bold) complementary to the target locus with its promoter and transcription terminator, as well as the SpCas9 binding scaffold to form the sgRNA-SpCas9 complex. Additionally, these sequences contain 20 bp homology arms (shown in uppercase) for cloning into the pACYC’ plasmid backbone.

Primers needed for sgRNA cloning and sequencing verification are listed in Table 7. In order to perform Gibson Assembly, amplify a linear portion of the pACYC′ plasmid with PCR primers ADC 9 and ADC10, using NEB Q5® High-Fidelity DNA Polymerase as directed by the manufacturer. The product should be 2887 bp. Digest the linear product with DpnI to eliminate parental plasmid background. The gblocks should be suspended following manufacturer’s instructions. Assembly can be performed without amplification of the gblock fragments, however, if amplification is desired, use primers ADC7 and ADC8 (Table 7), utilize the smallest quantity of starting template possible according to the PCR kit used and restrict the number of PCR cycles to 10–12. The gblock pCR products should be ~180 bp. All PCR products should be purified prior to assembly. Follow manufacturer’s conditions for Gibson Assembly, using the NEBuilder® HiFi DNA Assembly Master Mix.

• 3
  Once the gblocks are assembled in the pACYC’ plasmid, transform and purify the final recombinant plasmid pACYC-xxxgRNA, either from a Shewanella strain or from an E. coli dcm- dam- strain.

  See Step 4 in BASIC PROTOCOL for comments concerning plasmid transformation efficiency into Shewanella.

• 4
  Prepare 250 ng of sgRNA plasmid and 1 μg ssDNA (2 μL of the 0.5 μg/mL DNA stock) in a 2 ml microcentrifuge tube for each recombination. All sgRNA expressing plasmids are co-electroporated with ssDNA donor template at a ratio of 1/4.

  Also prepare a control containing the sgRNA plasmid DNA but lacking ssDNA for recombineering in each experiment.

Prepare Shewanella cultures

• 5
  Inoculate desired Shewanella strains from a single colony into 5 mL LB medium in 20 mL sterile glass culture flasks. Shake cultures at 30°C, ~200 rpm overnight.

  For annealase-mediated recombination, strains should contain the pX2C9pLacW3Bet plasmid. For oligo-mediated recombination, use plasmid pX2Cas9, which lacks an annealase but expresses Cas9 for the counter-selection.

• 6
  Dilute overnight culture/s to an OD₆₀₀ of ~0.08 in LB medium. Place cultures in a shaking incubator at 30°C until cells reach an OD₆₀₀ of~0.4–0.5. Refer to Basic Protocol 1, step 5 for details.

  Grow enough cultures for at least two transformations each; the experiment, with ssDNA, and a control (without ssDNA). Alternatively, instead of a control reaction without oligo, the researchers can use a non-targeting oligo lacking homology to the genome). Be sure to use adequate volume to remove samples to monitor culture turbidity.

• 7Follow instructions in BASIC PROTOCOL 1 for making Shewanella electrocompetent cells and introducing DNA.
• 8
  Allow the cells to recover in broth for at least 4 hours, or overnight.

  This allows adequate time for the recombination to occur and for complete segregation of the recombinant chromosomes before Cas9-mediated double-stranded breaks occur at non-recombinant targets.

Determine cell titers

• 9Make 1:10 serial dilutions in LB.
• 10
  Spread cells on Km (25 μg/ml) and Cm (6.5 μg/ml) agar plates supplemented with 0.8 mM (0.01%) arabinose. Plate 100 ul of undiluted cells and a range of 10-fold serial dilutions to yield single colonies.

  In all experiments (with and without ssDNA), also plate cells on the same medium lacking arabinose. This serves as a control to determine the effect of inducing Cas9 with arabinose.

• 11
  Incubate plates at 30°C for ~24–36 h until colonies appear.

  We routinely obtain ~10² - >10³ total recombinants on plates supplemented with arabinose and ~10⁴ - 10⁵ without arabinose, after a 2 hour recovery. Fewer recombinants are observed with shorter homology arms. Since the P_BAD promoter may be slightly leaky, some edited cells may be found on the no arabinose control.

• 12Follow instructions in BASIC PROTOCOL 1 for confirming recombinants.

REAGENTS AND SOLUTIONS

M9 Minimal Salts

1. Prepare 800 ml of MilliQ water in a suitable container.

2. Add 64 g of Sodium Phosphate Dibasic Heptahydrate (mw: 268.07 g/mol) to the solution (final concentration: 0.2387 M).

3. Add 15 g of Potassium Phosphate Monobasic (mw: 136.09 g/mol) to the solution (final concentration: 0.1102 M).

4. Add 2.5 g of Sodium chloride (mw: 58.44 g/mol) to the solution (final concentration: 0.04277 M).

5. Add 5 g of Ammonium chloride (mw: 53.49 g/mol) to the solution (final concentration: 0.09347 M).

6. Add MilliQ water until the volume is 1 L.

7. Stir until all ingredients are completely dissolved.

8. Divide the salt solution into 200 ml aliquots and sterilize by autoclaving for 15 minutes at 15 psi (1.05 kg/cm²) on liquid cycle.

9. The solution can be stored at room temperature or in a cold room.

LB medium

1. BD Difco^™ Dehydrated Culture Media: LB Broth, Miller (Luria-Bertani) (Fisher Scientific, cat. no. DF0446–17–3)

2. For each 950 ml of MilliQ water add 25 g of the pre-mixed LB formulation, as dictated in the container.

3. Mix until powder is dissolved.

4. Adjust solution with MilliQ water to a final volume of 1000 ml.

5. Mix again and divide the solution into 500-ml aliquots

6. Sterilize by autoclaving for 15 minutes at 15 psi (1.05 kg/cm2) on liquid cycle.

7. Store at room temperature.

8. Add antibiotics to LB immediately prior to use.

1 M sorbitol pH 7.6 washing buffer

1. For each 30 ml of MilliQ water add 7.29 g of D-sorbitol (Fisher Scientific, MW: 182.172, cat. no. S459–500). A sterile 50 ml falcon tube can be used.

2. Shake or use a magnetic bar to thoroughly mix the contents.

3. Adjust solution with MilliQ water to a final volume of 40 ml (final concentration: 1 M).

4. Mix again. Transfer 10ml of the solution in a separate sterile falcon tube. Do not adjust the pH.

5. Adjust the pH of the remaining 30 ml with NaOH.

6. Do not use HCl as it causes the electroporation cuvettes to arc, which destroys the samples. If the pH becomes >7.6, re-adjust it by adding more of the 1 M buffer solution in step 4 (make more if needed).

7. Filter-sterilize the solution using a 0.2 μm filtered syringe or cup.

8. Store at room temperature.

LB solid agar plates (with desired antibiotic, and without antibiotic)

1. BD Difco^™ Dehydrated Culture Media: LB Broth, Miller (Luria-Bertani) (Fisher Scientific, cat. no. DF0446–17–3)

2. Thermo Scientific^™ Agar, pure, powder (Fisher Scientific, cat. no. AC400400010)

3. Fisherbrand^™ Petri Dishes with Clear Lid (Fisher Scientific, cat. no. FB0875714)

4. 500 ml LB agar yields about one sleeve of culture plates.

5. Add 25g LB Broth per Liter to a 1–2 L beaker.

6. Add 500 ml MilliQ water and mix until clumps are gone.

7. Add 15 g agar per Liter and mix (this will not dissolve).

8. Transfer to a 1 L autoclavable glass bottle (or at least a bottle 2x the volume that you are making)

9. Autoclave for at least a 20 min liquid cycle.

10. Once the liquid is cool to the touch (≤50 °C) add appropriate antibiotic/s and/or supplement/s as needed, and mix. See Table 5 for common antibiotics/supplements used in Shewanella.

11. Pour into plates, covering the surface of the plate and avoiding bubbles.

12. Bubbles can be burned off with a Bunsen burner or popped with a sterile tip.

13. Let the plates cool, under the biosafety flow hood, with the lids ajar for approximately 15–30 min. Alternatively, leave to dry on the bench overnight.

14. Once the agar has been solidified, plates should be stored upside down at 4°C in the dark (foil wrapped) for no longer than three months to avoid degradation of antibiotics.

Commentary

Background Information

Genomic manipulation facilitates the study of bacteria for basic research and their adaptation for industrial use. Recombineering has been highly developed in E. coli (see Thomason et al. 2014), and the interested reader should familiarize themselves with this chapter since most of the information provided there is also applicable to recombineering in non-model microbes. This powerful in vivo genetic engineering technique offers distinct advantages and will continue to be extremely useful for industrially relevant microbes, since it allows the introduction of many different types of genetic changes with targeted, precise and scarless genome engineering. The repertoire of synthetic biology tools available for genetic engineering of non-model microbes is rapidly expanding beyond basic recombineering, and several such advances are mentioned later in this unit.

This chapter provides an overview of the current state of the field and an experimental technique for Shewanella species. Several different options for researchers of other non-model bacteria are suggested. Important decisions include choice of recombineering system and choice of linear DNA substrate, i.e., dsDNA or ssDNA. The E. coli recombineering functions λ Red and RecET_Ec function in a few non-native hosts and ideally will promote recombination of both linear DNA substrates (Table 1). Some phage recombination functions native to non-model bacteria have been identified and developed for use (Tables 2 and 3). With native phage functions, greater success has been achieved using ssDNA recombination, which requires only the cognate annealase. In some circumstances, oligo-mediated recombination can also be used, in combination with a selection or CRISPR/Cas counterselection. The latter has the advantage of requiring only ssDNA oligos and a means of introducing exogenous linear DNA into the cells. These options are discussed in more detail below, as is the usefulness of each type of DNA substrate.

Using Homologous DNA-based recombineering systems in non-model species

As in E. coli, recombineering with ssDNA oligos has been shown in other microbes to require only the annealase function (Aparicio, Nyerges, Martinez-Garcia, & de Lorenzo, 2020: Van Kessel, 2008; Corts et al., 2019b; Swingle, Bao, et al., 2010; van Pijkeren & Britton, 2012; Wang et al., 2018) and involves a simpler mechanism than does dsDNA recombineering (Ellis et al., 2001). Therefore, the development of ssDNA-based recombineering in new host species should be an easier task than dsDNA-based recombineering (see Table 2). As is true for E. coli (Sawitzke et al., 2007), Wang et al. (2018) showed that ssDNA recombineering is more efficient than dsDNA recombineering in Burkholderiales. Thus, in establishing recombineering in non-model bacteria, it is recommended to start by characterizing the annealase function for ssDNA recombineering first before attempting dsDNA recombineering. Compared to other genome engineering technologies, ssDNA recombineering has a lower error rate since no PCR amplification or cloning of the targeted gene is required. The presence of sporadic mutational errors in the recombinant may occur as a result of errors arising during oligonucleotide synthesis (Oppenheim et al., 2004), rather than being inherent in the recombineering mechanism.

Insertion of large DNA sequences is a common requirement for genome engineering of industrially relevant microbes. This can be accomplished by dsDNA recombineering using a phage annealase paired with its cognate exonuclease (Tables 1 and 3). The λ Redα/Redβ and Rac prophage RecE/RecT (annealase and exonuclease pairs) required for dsDNA recombination have co-evolved for efficient recombination, and the individual components of these analogous systems cannot be interchanged (Datta et al., 2008; Muyrers, Zhang, Buchholz, & Stewart, 2000; Yin et al., 2015). While the λ Red and RecET_Ec systems function well in a limited range of non-model organisms, they may fail to promote recombination in more diverse species. Even within the same bacterial species, recombineering frequencies can vary among different strains; λ Red mediated dsDNA recombineering was successfully applied in some strains of S. enterica, but was inefficient in others (Czarniak & Hensel, 2015). One cause of low recombineering frequency in non-model hosts may be a low transformation efficiency (Derbise, Lesic, Dacheux, Ghigo, & Carniel, 2003; Khandelwal, Agrawal, Singhi, Srivastava, & Bisaria, 2018; Sun, Wang, & Curtiss, 2008). When both the expression system and transformation efficiency are poor, low recombineering frequencies should be expected.

Some bacterial species may contain a host-encoded dsDNA exonuclease that rapidly degrades linear dsDNA substrates, lowering the recombineering frequency. In E. coli and other Gram-negative bacteria, this exonuclease is the potent RecBCD complex. In Gram-positive and proteobacteria, the AddAB exonuclease protein complex is a related enzyme. The phage λ Gam protein can inhibit RecBCD (Dillingham & Kowalczykowski, 2008; Murphy, 1991), and Gam expression improves the efficiency of Red-mediated dsDNA recombineering ~10-fold in E. coli (Datta et al., 2008) by preventing degradation of intracellular linear DNA. However, the Gam protein is useful only for bacteria that have RecBCD analogs. To our knowledge, there is no known Gam analog for the AddAB-type exonucleases, although chemical inhibitors of both the RecBCD and AddAB exonucleases have been identified (Amundsen et al., 2012; Bannister et al., 2010). Functional Gam analogs are rare (Tucker et al., 2014; van Kessel & Hatfull, 2007; Wang et al., 2018), although one has been identified in Photorhabdus luminescens (Pluγ) (Yin et al., 2019; Yin et al., 2015). Pluγ is ~30% identical with the 27-N terminal residues of λ Gam that contain negatively charged residues for interaction with RecBCD (Yin et al., 2015). In some (but not all) cases, co-expression of λ Gam or Pluγ with various recombineering systems in different species (Chen et al., 2016; Wang et al., 2018; Yin et al., 2019; Yin et al., 2015; Zheng et al., 2020) has been found to increase the recombineering frequency and may facilitate the use of shorter homology arms (Wang et al., 2018).

Native recombineering systems

To date, several recombineering systems homologous to λ Red or RecET_Ec have been characterized for recombineering in non-model species (Tables 2 and 3). Most systems have the exonuclease and associated SSAP genes immediately adjacent within the same operon (Chang, Wang, Su, & Qi, 2021; Datta et al., 2008; Swingle, Bao, et al., 2010; Tucker et al., 2014; van Kessel & Hatfull, 2007; Yin et al., 2015). The number of newly sequenced bacterial and phage genomes in GenBank is expanding rapidly and can be explored for phage and prophage recombination genes using various search methods such as BLAST (Altschul, Gish, Miller, Myers, & Lipman, 1990). Both Beta and RecT should be used in any search; although they are both phage annealases, they do not share close sequence identity (Iyer, Koonin & Aravind, 2002).Homology searches should be done with the protein sequence. See Datta et al. (2008), as well as references from Table 2 and 3, for examples of recombination gene identification. A dataset of >18,000 RecT/RedBeta homologous annealase genes from bacterial and phage genomes was recently published (Steczkiewicz, Prestel, Bidnenko, & Szczepankowska, 2021). Promising candidate genes should be cloned into an appropriate vector under control of a tightly regulated inducible promoter. Figure 5 illustrates the experimental workflow for this process.

Oligo-mediated recombination in bacteria.

In various bacteria, both Gram-negative and -positive, ssDNA oligos recombine with homologous regions of the bacterial chromosome in the absence of any known annealase (Bryan & Swanson, 2011; Dutra & Lovett, 2006; Murphy & Marinus, 2010) (see Table 5 and references therein). This recombination occurs at a low frequency of about 10⁻⁵. While not strictly recombineering, since no exogenous functions are necessary for this recombination, this simple ssDNA oligo mutagenesis technique may be useful for initial recombineering studies in non-model microbes. A selection or enrichment is required, however, since the recombination frequency will be extremely low. This oligo-mediated technique can be combined with a strong counterselection such as CRISPR-Cas (Corts et al., 2019b), which will substantially boost the apparent recombination frequency and allow recovery of recombinants by screening.

Oligo-mediated recombination was revealed in P. syringae when Swingle et al. (Swingle, Markel, Costantino, et al., 2010) found recombinants in control reactions lacking annealase activity. No annealase homologs were found in the genome of P. syrinagae and expression of the λ Red functions did not increase recombinant recovery in P syringae. In E. coli, removal of the Rac prophage recET genes or host recombination genes did not eliminate this type of oligo recombination (Swingle, Bao, et al., 2010). Oligo-mediated recombination has also been observed in Corynebacteria (Krylov, Kolontaevsky, & Mashko, 2014), Legionella pneumophila (Bryan & Swanson, 2011) and other non-model species (see Table 5). This capability is likely to be found in other non-model microorganisms and may be useful in bacteria without a known annealase.

Two useful assays to detect oligo-mediated recombination involve introducing modifications into either the small ribosomal protein S12P, encoded by the rpsL gene (Krylov et al., 2014; Swingle, Markel, Costantino, et al., 2010), or the β subunit of the DNA-dependent RNA polymerase, encoded by rpoB (Bryan & Swanson, 2011; Penewit et al., 2018; Swingle, Markel, Costantino, et al., 2010). Both assays allow straightforward and efficient positive selections conferring streptomycin and rifampin resistance, respectively, and are useful in both Gram-negative and Gram-positive bacteria. These assays require a long cellular recovery time after transformation (more than three hours for E. coli), since in both cases the wild-type antibiotic sensitive allele is dominant to the mutant antibiotic resistant allele. The longer outgrowth allows the wild type and mutant alleles to segregate completely, enabling mutant detection (Corts et al., 2019b; Krylov et al., 2014; Swingle, Markel, & Cartinhour, 2010). Additional information about modifying rpsL can be found in Thomason et al. (2014).

The lacZ gene is another useful target when developing an oligo-mediated recombination strategy. If the organism of interest does not carry a copy of lacZ, it may be possible to introduce a functional lacZ copy from another organism into the genome of the desired strain. Corts et al. (2019a, 2019b) used a strain of S. oneidensis containing a chromosomal integration of the E. coli lacZ gene, which facilitated detection of recombinants using X-gal supplemented agar plates. Colonies bearing a functional lacZ can metabolize X-gal resulting in blue pigmented colonies, while loss of lacZ function results in white colonies. As in annealase-mediated recombineering (Corts et al., 2019b; Sawitzke et al., 2011), when a screening assay is used to detect oligo recombinants, cells should be plated on non-selective media immediately after a 30 min recovery from electroporation. The oligo is incorporated at the replication fork during DNA replication (Pines et al., 2015; Sawitzke et al., 2011; Thomason et al., 2016). If the recombinant chromosome is not fully segregated prior to plating, some colonies may be sectored white and blue (Corts et al., 2019b; Sawitzke et al., 2011). The sectored colonies indicate incorporation of the oligo at some, but not all, of the active replication forks (Pines et al., 2015; Sawitzke et al., 2011). When colonies are counted, the observed oligo recombination frequency will be somewhat higher than if complete segregation had been allowed. It is important to purify any sectored colonies to obtain pure clonal populations and to verify the sequence of the desired modification.

Use of ssDNA oligos to introduce point mutations

The most straightforward genetic engineering step is to use an oligo to make small base changes in the target genome. This reaction requires only an annealase such as Red β or Rac RecT. Many related annealases have been identified in diverse bacterial species and offer attractive alternatives for researchers to try (Table 2). As demonstrated in E. coli and S. cerevisiae, a major benefit of using ssDNA for creating genomic alterations is the simplicity of this technology, which allows researchers to rapidly perform multiple cycles of recombination to further increase the mutational genomic space (DiCarlo et al., 2013; Wang & Church, 2011). More information about ssDNA recombination is provided under Critical Parameters.

Considerations for dsDNA-promoted recombineering

Recombination with dsDNA is less efficient than ssDNA (Costantino & Court, 2003; Swingle, Bao, et al., 2010; Wang et al., 2012; Wang et al., 2018), thus, researchers should rigorously design dsDNA recombineering experiments to optimize recombination frequency. See the Strategic Planning section of this document for guidance. Basic Protocol 1 and Figure 1 of Thomason et al., (2014) have information about linear dsDNA preparation when performing experiments in E. coli; the same general guidelines apply regardless of the host being modified. Because the DNA substrates are obtained by PCR amplification, verification of the final construct by sequencing is highly recommended. Screening for point mutations in both the amplified DNA and in the recombinant junctions is important, since mistakes may be introduced by primers that contain in vitro DNA synthesis errors (Oppenheim et al., 2004).

While a homology arm length of 35–40 bp is sufficient for high λ-Red and RecET recombineering frequency in E. coli (Yu et al., 2000; Zhang, Buchholz, Muyrers & Stewart, 1998), increasing the length of the homology arms on the linear substrate can enhance the efficiency of dsDNA recombination in E. coli and in other hosts (Table 3). The saturating amount of DNA substrate will vary in different bacteria based on the efficiency of transformation by electroporation. When using E. coli expressing the Red functions, the saturating amount of dsDNA substrate is ~300 ng (Thomason et al., 2005; Yu et al., 2000), since transformation by electroporation in E. coli is extremely efficient (Costantino & Court, 2003; Thomason et al., 2005). In other organisms that transform poorly, higher amounts of DNA may be needed (Table 3). Addition of phosphorothioate bonds on the 5’ ends of the dsDNA may increase the number of recombinants obtained as they have been observed to hinder cellular exonuclease end-trimming of oligos (Lee, Ostrov, Gold, & Church, 2017), as may codon optimization of the recombineering proteins (Li, Swofford, Ruckert, & Sinskey, 2021; Wang et al., 2018). Co-expression of the single-stranded DNA binding protein (SSB), from the same host as the annealase/exonuclease pair used for recombineering was also found to enhance linear dsDNA-based recombineering in heterologous hosts. (Yin et al., 2019).

Critical Parameters for Successful Recombineering

Critical steps that can affect the outcome of the recombination are noted in the protocol itself, and key points to be aware of during the execution of the protocol are emphasized below.

It is important to use fresh, logarithmically growing bacterial cultures, and not to let the cell culture used for recombineering overgrow. Recombination occurs at active replication forks, which are not present in stationary phase (overgrown) cells. Grow the cells to a maximal A₆₀₀ of 0.6; cultures that have exceeded this density are not suitable for recombineering.

Proper controls are essential and should be included in your experiment. One control contains cells that are induced for the recombineering system but to which no DNA is added; the other control contains uninduced cells. Including these controls allows the researcher to compare titers and confirm that colonies arising from the induced culture with DNA are true recombinants rather than unwanted bacterial contamination or residual supercoiled plasmid DNA background.

In both E. coli and Shewanella, it has been reported (Tu, et al., 2016; Corts et al., 2019a, 2019b) that electrocompetent cells can be prepared at room temperature. Since other non-model microbes may have different temperature optima for recombineering, it may be necessary to determine this parameter empirically.

Gentle treatment of cells during the processing steps is critical. Centrifuge cells at the recommended relative centrifugal force (RCF) to avoid lysis during preparation. It is also important to wash the cells properly. During washing the cell pellet becomes very soft and is easily lost when decanting the supernatant. Remove tubes from centrifuge promptly and take care when removing the liquid. The washing steps are crucial to remove all salts from the cell suspension and to allow efficient electroporation. Multiple washes are required. Failure to remove salts will cause the electroporator to arc, meaning that the cuvette and its contents must be discarded. Even in the absence of arcing, residual salts can cause a lower time constant, which results in lowered cell viability, thus decreasing the likelihood of recovering recombinants. Before electroporation, mix the cells with the DNA in the microcentrifuge tube carefully and thoroughly. Inadequate mixing will reduce recombination frequency. After electroporation the cells are osmotically fragile, and recovery broth should be added to the cuvette immediately.

Depending on the type of recombination being done, outgrowth time in recovery broth will vary. In all cases, allow the cells a minimum 30-minute recovery time before transferring them to solid agar petri plates. At least ~2-hour outgrowth is required if selecting for antibiotic resistance. Still longer outgrowth times of ~3.5–4 hours may be required for complete chromosome segregation in order to see a pure phenotype in individual colonies. For example, sensitivity to streptomycin is dominant. If attempting to alter the rpsL gene to express streptomycin resistance (Str^R), the culture must be outgrown until individual cells no longer contain a copy of the wild type gene.

To ensure that pure recombinant clones are obtained, it is essential to re-streak colonies arising on the selection plates before proceeding with analysis. Normally the Court laboratory purifies 6–12 independent colonies for further analysis.

It is important to transiently express and tightly regulate production of the recombination functions (Huang et al., 2014). In some cases, expression of an SSAP, alone or paired with its cognate Exo, can cause toxic effects; this may be due to the annealase incompatibility with endogenous cellular machinery (Thomason et al., 2005). Inappropriate expression of the annealase and exonuclease can result in unwanted rearrangements (Murphy & Campellone, 2003; Sawitzke et al., 2007). Prolonged expression of λ Gam can lead to plasmid instability and toxic effects (Cohen & Clark, 1986; Sergueev, Yu, Austin, & Court, 2001). Thus, when attempting to adapt any recombineering system for a new host, it is critical to determine the appropriate level of gene induction and consequent protein expression by using different inducer concentrations and induction times.

The transformation efficiency of different bacteria will dictate whether recombineering is possible, and if so, the amount of linear DNA required for recombineering. An electroporation procedure is generally required, since the efficiency achieved with calcium-competent cells is inadequate for recombineering. A higher transformation efficiency correlates with a higher frequency of recombination; however, a low transformation efficiency should not discourage a scientist from trying recombineering (and oligo-mediated recombination) for which a positive selection exists. A complete review of transformation procedures for non-model bacterial is beyond the scope of this chapter, but the researcher is encouraged to investigate what has been reported in the literature pertaining to their bacterium of interest. It should be noted that for Gram-positive bacteria, cell-wall weakening agents are used prior to transformation, so the cultivation time and concentration of these agents in the growth medium prior to electroporation must be considered, since overgrowing the cultures will limit transformation efficiency and thus recombineering rates.

Critical Parameters affecting Oligonucleotide-promoted Recombineering

Experimental considerations for annealase-dependent ssDNA oligo recombination are also pertinent to oligo-mediated recombination. Issues to keep in mind are: (1) type of genetic modification to be introduced, (2) targeted DNA strand (lagging vs. leading), (3) mismatch repair (MMR), (4) oligo concentration, (5) homology length, (6) G:C content.

1. Type of genetic modification: The efficiency of ssDNA oligo recombination depends on the type of modification introduced. As found in E. coli, Swingle et al. (Swingle, Markel, Costantino, et al., 2010) reported that in P. syringae, making small nucleotide substitutions with ssDNA oligos is much more efficient than making large deletions, which are at least 100-fold less frequent. Aparicio, Jensen, Nielsen, de Lorenzo, & Martinez-Garcia, (2016) obtained P. putida recombinants at a frequency of ~10⁻⁵ recombinants per viable cell when small insertions or deletions were made using oligo-mediated recombination, but were unable to recover a deletion of ~700 bp.

2. Targeted DNA strand (lagging vs leading strand bias): Recombination with ssDNA oligos displays a strand bias, and oligos identical in sequence to the lagging strand result in a higher efficiency than do oligos identical to the leading strand (Bryan & Swanson, 2011; Dutra & Lovett, 2006; Krylov et al., 2014; Swingle, Markel, & Cartinhour, 2010; Swingle, Markel, Costantino, et al., 2010). In C. glutamicum, a 10-fold higher lagging strand frequency was observed when targeting rpsL (Krylov et al., 2014). The annealase-dependent strand bias arises from the direction of DNA replication through the target sequence (Pines et al., 2015; Sawitzke et al., 2011; Thomason et al., 2016). The oligo-mediated reactions also occur at the DNA replication fork, with the oligos annealing to single-stranded gaps in the lagging strand at the target as well as other genomic regions containing gapped DNA.

3. Mismatch repair: The ssDNA oligos used to alter DNA sequence result in non-paired DNA bases when bound to the otherwise complementary target. These mispairs are subject to removal by the host mismatch repair (MMR) system. Preventing MMR can increase the efficiency of annealase-dependent reactions ~100-fold in E. coli (Costantino & Court, 2003). Similar results have been obtained in oligo-mediated reactions (Aparicio, Nyerges, Nagy, et al., 2020). Some clusters of mispairs can evade MMR (Sawitzke et al., 2011), although these vary depending on the host and not all configurations of multiple mispairs allow this (Krylov et al., 2014). Note that small clusters of mispairs (≤4 bp) may not necessarily overcome the MMR system in S. oneidensis (Corts et al., 2019a).

4. Oligo concentration: Increasing oligo concentration enhances the recombination frequency, until the cells become saturated with oligo (Krylov et al., 2014; Swingle, Markel, Costantino, et al., 2010). Higher amounts of oligo may be needed for oligo-mediated recombination in the absence of an annealase (see Table 5 and references therein), probably because the ssDNA is not protected by annealase binding, and is thus subject to digestion by host ssDNA exonucleases (Dutra & Lovett, 2006; Murphy & Marinus, 2010). For oligo-mediated recombination, the amount of DNA necessary for saturation ranges between 2.5–10 μg, depending on the host (Krylov et al., 2014; Swingle, Markel, Costantino, et al., 2010). Adding saturating amounts of oligo helps titrate out host exonucleases and increase the lifetime of the ssDNA in the cell (Murphy & Marinus, 2010). Carrier oligo, of a sequence non-homologous to the genome, can be added in combination with the recombining oligo, so that the total amount of DNA equals the saturating concentration. This method resulted in a frequency equivalent to using saturating conditions of the targeting oligo in P. syringae (Swingle, Markel, Costantino, et al., 2010). Similar results were seen for Red-mediated recombination (Sawitzke et al., 2011). Mixed results have been obtained when terminal phosphorothioate linkages were added to oligos to protect them from degradation by host nucleases (Krylov et al., 2014).

5. Homology length: Increasing the length of ssDNA oligos improves the rate of Red- and RecET-dependent reactions, up to about 70 nt in E. coli (Sawitzke et al., 2011; Zhang et al., 1998). In contrast, Swingle et al. (2010) found that for oligo-mediated reactions, ~20 nt of homology gives maximal frequency of oligo-recombination for nucleotide substitutions in E. coli and P. syringae. Similar results were obtained in L. pneumophila (Bryan & Swanson, 2011).

6. G:C content: When the G/C content of the oligo and the target are low (~30%), oligo recombination rate may be reduced. For oligo-mediated recombination, under these low G/C conditions, longer oligos were needed to obtain a frequency of that obtained with shorter oligos of ~50% G/C content (Swingle, Markel, Costantino, et al., 2010).

7. Overcoming transformation issues: If only a subpopulation of the cells in the culture are transformable, a useful strategy used in E. coli (Sawitzke et al., 2011) and mycobacteria (van Kessel & Hatfull, 2008) for annealase-dependent oligo recombination may be tried. Here ssDNA oligos containing a mutation targeted to the bacterial chromosome were co-electroporated with a selectable plasmid and recombinants were scored within the population that received the plasmid and became drug resistant. 3–5% of cells that took up the selectable plasmid became recombinant for the oligo mutation. This approach allows the scientist to screen out cells that are not competent for DNA uptake.

Oligo Design Tools for Maximal Recombination Efficiency

Various helpful internet-based tools exist for designing ssDNA oligos for efficient recombination; these are listed in Table 6 below. The details of each program are too lengthy to fully cover in this discussion. As the different applications will have different features, we suggest the researcher try the various programs to find one that suits their needs. The researcher will need to design the ssDNA oligo(s) used for recombineering as well as flanking primers for PCR amplification and sequencing of the altered target.

The most advanced platform is MODEST, MAGE Oligo Design Tool, which allows for automatic design of single and multiple mutagenic oligos and can be found at www.modest.biosustain.dtu.dk (Bonde et al., 2014). MODEST was specifically developed for recombineering and multiplex automated genome engineering (MAGE), but it can be used for oligo-mediated reactions. MODEST designs oligos that will result in the highest efficiency of recombination; the oligo sequence is optimized so that the optimal configuration has the lowest secondary structure energy with the highest minimum free energy, with a limitation being that the desired changes introduced are at least 15 nt from the end of the oligo.

As discussed above, there is a significant difference in efficiency of recombination between the two complementary ssDNA oligos that can be used for recombineering (strand bias) (Ellis et al., 2001; Thomason et al., 2016). This bias arises from the mechanics of DNA replication and the fact that the lagging strand is replicated discontinuously, with gaps between the Okazaki fragments, in contrast to the (largely) continuously replicated leading strand. An abundance of experimental evidence supports the model that linear DNA substrates are incorporated at the gaps in the lagging strand (Ellis et al., 2001; Thomason et al., 2016). The ssDNA oligos corresponding to this strand (so-called lagging strand oligos) routinely give higher recombination levels than those of complementary sequence. It is therefore useful to know the direction of DNA replication with regards to the target gene. Applications to determine the DNA replication origin are also listed in Table 6. However, it is not an absolute requirement to determine direction of DNA replication, since it may be easier to simply order both complementary strands and electroporate both oligos, separately or together. If multiple changes are desired for the targeted gene, one set of complementary oligos can be ordered initially for one gene, and each electroporated separately. Whichever oligo gives the higher frequency is likely the lagging strand-targeting and subsequent oligos of this same orientation can be procured for future recombineering.

Recombineering Advances and Combination with other Genome Engineering Techniques

Targeted, precise and scarless genome editing is fundamental for exploring genotype–phenotype relationships, developing high-performing biological tools, and improving robustness in industrial bacteria. An assortment of technologies has been developed for genome engineering of E. coli, as well as other non-model microbes. Some examples are mentioned below.

Continued refinement of our understanding of the molecular mechanism of recombineering will help extend the portability of this powerful technology to additional hosts. In a notable example of this, Filsinger et al. (2021) showed that expression of a single-strand binding protein (SSB) compatible with a particular SSAP increases the efficiency of ssDNA recombineering for hosts in which the SSAP otherwise performs poorly. They used chimeric SSB proteins to demonstrate that SSAPs interact with the C-terminus of their host SSB, and that SSAPs paired with an SSB having a compatible C-terminus can be used for recombineering in a different species. For example, expressing the Pseudomonas aeruginosa RecT with its cognate SSB increased the recombineering efficiency in L. lactis ~1,000-fold, reaching that of L. lactis RecT. The Pseudomonas RecT-SSB pair also gave a high efficiency of recombineering in E. coli without toxicity (Filsinger et al., 2021).

Jiang, Bikard, Cox, Zhang & Marriffini (2013) first demonstrated that CRISPR/Cas9 could be used in conjunction with recombineering. Since their seminal work was published, a variety of modifications have been tested, and coupling recombineering to a CRISPR/Cas system used for counter-selection has proven to be a powerful tactic to isolate recombinant cells at high efficiencies (Cook et al., 2018; Corts et al., 2019a). For this technique to work, the recombination event that introduces the desired mutation must occur before CRISPR cutting is used to eliminate nonrecombinants, since single-strand annealing proteins normally cannot use a DNA double-strand break as a recombination substrate (Kuzminov, 1999). Non-recombinant DNA sequences are destroyed by Cas cutting, preserving the recombinants created during recombineering. This method works efficiently in diverse species (Corts et al., 2019a; Vento, Crook, & Beisel, 2019; Wang et al., 2019) and can drastically boost the number of recombinants recovered among total viable cells (Aparicio, de Lorenzo, & Martinez-Garcia, 2018; V. Chen, Griffin, Maguin, Varble, & Hang, 2021; Pinero-Lambea et al., 2020; Shen et al., 2019; Zhao, 2018). The use of CRISPR/Cas-assisted recombineering in various hosts has been reviewed (Vento et al., 2019). Using CRISPR/Cas as a counter-selection increased the recovery of ssDNA recombinants from a fraction of a percent with recombineering alone to more than 80% in both M. smegmatis and Shewanella (Corts et al., 2019a; Yan et al., 2017). Successful coupling of recombineering with CRISPR/Cas can allow recovery of recombinants from reactions that have a low recombination frequency, such as insertion or removal of a large heterology, which are generally non-selectable and less efficient than creating a single base change, regardless of the host. In general, the larger the modification, the lower the recombination frequency (Bao, Cartinhour, & Swingle, 2012; Corts et al., 2019a; Sawitzke et al., 2007; Tucker et al., 2014; Wang et al., 2018). A limitation of using CRISPR/Cas as a counter-selection include the requirement of a nearby PAM site. While the PAM site for the commonly used SpCas9 is small (NGG), others may limit the type of modifications one can make. The researcher should also be aware that spontaneous mutations in the target region and the large cas9 gene can contribute to undesirable background i.e., incorrect colonies arising when the counterselection is applied.

The use of CRISPR/Cas-mediated dsDNA breaks has also been combined with recombineering using plasmid substrates instead of linear dsDNA for recombination. In this case, the recombination cassette with the homology arms is embedded within the plasmid (Choi, Cho, Cho, Park, & Lee, 2018; Cook et al., 2018; Garst et al., 2017; Vento et al., 2019). The molecular mechanism of this modified reaction has not been clarified.

Retron-based recombineering (Schubert et al., 2021), is also based on oligo recombineering. This technology uses prokaryotic retroelements to produce multicopy ssDNA in vivo by reverse transcription. These endogenously made ssDNAs are recombinogenic, just like ssDNA oligos introduced by electroporation. Retron plasmids carry a donor DNA, and the retron sequences function as barcodes to identify mutants in a mixed population (Schubert et al., 2021). In an ΔmutS ΔrecJ ΔsbcB E. coli strain expressing CspRecT, between 50% and >90% of the total viable cells (depending on the gene targeted) were recombinants following 20 generations of batch growth. Further development of retron-based recombineering is needed, and it is not known whether the retrons cause off-target genomic modifications (Schubert et al., 2021).

In M. tuberculosis, the Che9 RecT-promoted recombination has been coupled with a mycobacterial phage site-specific integration system called ORBIT. In ORBIT (Oligo Recombineering followed by Bxb1 Integrase Targeting) (Murphy et al., 2018), a ssDNA oligo containing the phage site-specific attachment site (attP) is co-electroporated with a replicating plasmid containing the bacterial attachment site (attB), an antibiotic resistance gene, and a payload (such as a gfp tag). Simultaneous expression of the annealase and the phage integrase results in integration of the payload plasmid. A variety of payload plasmids are available that allow generation of various genetic changes, such as gene knockouts, replacements, or tags (Murphy et al., 2018).

Troubleshooting Recombineering in non-model bacteria:

Common problems encountered when executing recombineering and suggested solutions are given in Table 9.

Table 9.

Troubleshooting Recombineering in non-model bacteria

Problem	Possible Cause	Solution
No recombinants found	Problems with bacterial strain used for recombineering	For strains containing plasmids, confirm drug resistance. For Shewanella, perform a control recombineering reaction (i.e target rpsL or rpoB with ssDNA oligo) to confirm that recombinants can be obtained with the system. For temperature-sensitive (ts) expression systems, grow strains at lower temperature (i.e. −30–32°C for E. coli) except when inducing recombination functions. Note that for Shewanellae, it may not be feasible to use heat-inducible expression systems.
	Cell loss when supernatant is discarded during washing with H2O prior to electroporation	The cell pellet is very soft. Remove tube from centrifuge and decant supernatant promptly to avoid losing cells during wash steps. It is also possible to carefully remove the supernatant using a pipette. Titer viable cells as well as plating for recombinants; if viable cells are less than 107/ml, recombinants may not be obtained without selection.
	Problems with inducing recombination proteins	It may be necessary to optimize levels of chemicals used to induce recombination proteins. When using ts expression systems, it is best to use a shaking H2O bath for cell growth and it is essential for the 42°C induction. Air shakers will not work for 42°C induction step. Measure temperatures of H2O baths. Do not exceed 15 min 42°C induction. For other bacteria, this timing may need to be optimized; in Z. mobilis, optimal induction was 30 min for the same cI857-PL system (Khandelwal, 2018).
	Problems with linear DNA used for recombination	Check design of linear DNA, make new PCR product or obtain new ssDNA oligos
	Problems with electrocompetent cells	Wash cells well to prevent arcing during electroporation. Discard sample if cells arc. Mix cells with DNA gently but thoroughly before electroporation. Avoid introducing bubbles. Use appropriate electroporation conditions. Generally, time constant for electroporation should be >5msec, but this may vary depending on the host. Add recovery broth to cuvette immediately after electroporation. Transform cells with a supercoiled plasmid to confirm competence.
	Cells overgrown	If cultures are in stationary phase, there will be no recombination, which requires active DNA replication forks. Grow cells to 0.4–0.6 A600.
No recombinants obtained with drug selection	Drug concentration in agar petri plates too high	Be sure to use antibiotic concentrations appropriate for single copy drug resistance cassettes rather than multi-copy plasmids.
	Inadequate outgrowth time or poor outgrowth conditions	Allow a minimum 2 hr recovery time following electroporation. Grow cells in larger volumes of broth so more cell doublings occur.
No recombinants or transformants with double drug selection	Double drug selection issues	Use only one drug selection at a time; do not select for multiple drugs simultaneously. For example, do not maintain selection for recombineering plasmids when selecting drug-resistant recombinants. If two drugs must be used due to plasmid instability; reduce drug concentrations by half.
Colonies in uninduced control reactions when plasmid DNA was used as template for recombineering substrates	If plasmid is incompletely digested, circular plasmid transformants will outnumber recombinants	We do not recommend amplifying PCR products used for recombineering from a plasmid template. If absolutely necessary, linearize plasmid with appropriate restriction enzymes before using as template, and digest PCR product with DpnI following amplification, to eliminate uncut plasmid background, since DpnI only cuts when its DNA recognition site is methylated. Confirm complete digestion by transforming WT cells with the linear DNA; no colonies should be observed on selective plates.
Colonies in uninduced control reactions	Leaky selection	Modify selection i.e. increase drug concentration if necessary.
	Contamination	Use sterile technique. Test solutions for source of contamination and obtain fresh solutions if necessary.

Understanding Results:

The Basic protocol describes the creation of small genetic changes in Shewanella using an endogenous phage annealase, W3 beta, and linear ssDNA substrates. The Alternate Protocol presents use of annealase- and oligo-mediated recombination coupled with CRISPR/Cas as a counterselection to destroy non-recombinant alleles after recombination is completed. The inclusion of CRISPR/Cas9 allows large deletions made with ssDNA to be recovered at reasonable frequencies. While the protocol presented here is similar to that used for E. coli (Thomason et al., 2014), one major difference is the use of room temperature cells and reagents, which we have found to be superior for Shewanella. Tu et al. (2016) have reported a similar finding for E. coli. The researcher may need to determine optimal temperature for their bacterium empirically.

After overnight incubation at the appropriate temperature, the petri plates should be examined, and bacterial colonies enumerated. The titer of viable cells in the experiments should be determined from the rich plates, by counting the number of colony-forming units (CFUs) and multiplying that by the dilution factor applied when plating the cells. Be aware of possible contamination when examining the plates. The researcher should know the colony morphology of the bacterium under study and be on the lookout for non-standard colonies. Dilutions of viable cells should be internally consistent, and negative control plates should not have bacterial colonies. The number of recombinant colonies should be determined from the selective medium if used. If a PCR screen is used to look for recombinants, the number of positives per total colonies should be tallied. Recombinant frequencies are determined by dividing the recombinant values by the titer of the total viable cells (~10⁸ for S. oneidensis MR-1). Expected ssDNA recombineering frequencies for S. oneidensis should be 10⁶-10⁷ in 10⁸ viable cells. That means about 5% of the viable cells generally are correct recombinants). The percentage of edited cells will be much higher if CRISPR/Cas is used. However, co-electroporation of multiple DNAs (ssDNA and sgRNA plasmid together as in the Alternate Protocol 1) may reduce the transformation efficiency; a decrease in the number of transformed cells reduces the absolute number of recombinants, while the percentage of edited cells is increased due to the powerful CRISPR/Cas counter-selection (Corts et al., 2019a, 2019b).

The techniques described here will allow researchers to construct some genetic modifications in some non-model bacteria. Making small genetic alterations with a single annealase protein and linear ssDNA substrates is easier to achieve than insertion of larger heterologous dsDNAs, such as drug resistance genes, or making large deletions with ssDNA oligos. Combining ssDNA recombination with a CRISPR/Cas counterselection improves the utility of the technique and may allow the recovery of these, as well as unselected modifications. The capacity to do dsDNA recombineering in non-model bacteria will increase as additional SSPA/Exo pairs are identified and characterized.

The Shewanella work was facilitated by successful identification and adaptation of a Shewanella phage annealase, W3 beta, to promote ssDNA recombination in the bacterium. Although a probable exonuclease gene was identified, we were unable to obtain dsDNA recombination when the putative exo gene was expressed in combination with W3 beta. This failure was largely circumvented using CRISPR/Cas9 as a counterselection. Not all researchers will be fortunate enough to obtain such a facile recombination system, in which case we encourage them to explore use of known heterologous annealases and oligo-mediated techniques.

Time Considerations:

Once the intended constructs have been designed and the linear targeting DNAs have been obtained, the basic recombineering technique may be completed in one day, using bacterial cultures inoculated the previous day. The recombinants take at least a day to develop on the agar plates. Colony purification and subsequent colony PCR and DNA sequencing to verify recombinants will add a few more days. As a result, the entire process can usually be finished in a week, although additional time may be required for slow growing strains.

KEY REFERENCES:

Aparicio, T., Nyerges, A., Nagy, I., Pal, C., Martinez-Garcia, E., & de Lorenzo, V. (2020). Mismatch repair hierarchy of Pseudomonas putida revealed by mutagenic ssDNA recombineering of the pyrF gene. Environ Microbiol, 22(1), 45–58.

ssDNA recombineering is used in P. putida KT2440 to characterize the mismatch repair system hierarchy of the strain, demonstrating some variation as compared to E. coli.

Bonde, M. T., Klausen, M. S., Anderson, M. V., Wallin, A. I., Wang, H. H., & Sommer, M. O. (2014). MODEST: a web-based design tool for oligonucleotide-mediated genome engineering and recombineering. Nucleic Acids Res, 42(Web Server issue), W408–415.

Useful web-based tool to automatically design oligos for engineering of any sequenced genome (MMR inactivated background host only).

Datta, S., Costantino, N., & Court, D. L. (2006). A set of recombineering plasmids for gram-negative bacteria. Gene, 379, 109–115.

Constructed a set of plasmids carrying a minimal λ Red expression cassette under its native λ cI857 temperature-sensitive repressor system for ssDNA or dsDNA recombineering in Gram-negative bacteria.

Datta, S., Costantino, N., Zhou, X., & Court, D. L. (2008). Identification and analysis of recombineering functions from Gram-negative and Gram-positive bacteria and their phages. Proc Natl Acad Sci U S A, 105(5), 1626–1631.

Characterized the functionality of a variety of non-model annealases and exonucleases from diverse sources for ssDNA and dsDNA recombineering in E. coli. Demonstrated that SSAPs have greater flexibility for use in non-native systems than when paired with their associated exonucleases.

Murphy, K. C. (1998). Use of bacteriophage λ recombination functions to promote gene replacement in Escherichia coli. J Bacteriol, 180(8), 2063–2071.

First paper demonstrating the use of λ Red functions for targeted and precise genome engineering in E. coli.

Nyerges, A., Csorgo, B., Nagy, I., Balint, B., Bihari, P., Lazar, V., . . . Pal, C. (2016). A highly precise and portable genome engineering method allows comparison of mutational effects across bacterial species. Proc Natl Acad Sci U S A, 113(9), 2502–2507.

Developed a broad-host range plasmid recombineering system, pORTMAGE, carrying a dominant-negative mutant protein of the methyl-directed mismatch repair system for transient suppression of DNA repair system and efficient recombination. Functional in distant relatives of E. coli.

Swingle, B., Markel, E., Costantino, N., Bubunenko, M. G., Cartinhour, S., & Court, D. L. (2010). Oligonucleotide recombination in Gram-negative bacteria. Mol Microbiol, 75(1), 138–148.

Demonstrates that oligo-mediated recombination is a universal and naturally occurring method of homologous recombination that is independent of RecA.

Vento, J. M., Crook, N., & Beisel, C. L. (2019). Barriers to genome editing with CRISPR in bacteria. J Ind Microbiol Biotechnol, 46(9–10), 1327–1341.

Review of the application and challenges of using CRISPR-based genome editing techniques, including coupled to recombineering systems for in vivo genome engineering non-model bacteria.

Zhang, Y., Buchholz, F., Muyrers, J. P., & Stewart, A. F. (1998). A new logic for DNA engineering using recombination in Escherichia coli. Nature Genetics, 20(2), 123–128.

First paper demonstrating the use of RecET functions from the Rac prophage for targeted and precise genome engineering in E. coli.

DATA AVAILABILITY STATEMENT:

The datasets described in this protocol are available from the corresponding author on reasonable request.