An att site-based recombination reporter system for genome engineering and synthetic DNA assembly

The ability to precisely and directly manipulate DNA is important for functional studies and the synthetic assembly of large genetic constructs. Site-specific recombinase (SSR) systems are widely used as tools to rearrange, insert, remove, and join DNA with virtually no upper limit in size. For biotechnology purposes, this can include the insertion of exogenous DNA into chromosomes, the fusing of DNA molecules, or the construction of synthetic gene networks [1]. The tyrosine (Y-rec) and serine (S-rec) recombination families are named for the catalytic residue of their respective integrase (Int) protein. Important members of the Y-rec family include the λ-like phage recombination systems, which include λ and the closely related phage HK022 (hereafter referred to as HK). The ΦC31 recombinase system is an important member of the S-rec family [2]. Both HK and ΦC31 systems comprise attB/attP attachment sites that serve as points of recombination, and the recombinases that catalyze recombination. In each family, DNA exchange requires host-encoded proteins for recombination that differ between systems. These systems are attractive due to their directionality and stability, and both systems are functional in prokaryotic and eukaryotic organisms [3–5].

Mechanistically, attB and attP integrative recombination forms attL and attR sites. The reverse attL x attR excisive reaction also requires Int as well as a recombination directionality factor (RDF), named Xis in the HK system and gp3 in the ΦC31 system [6], typically supplied in trans from a helper plasmid, a non-replicating DNA molecule, or as mRNA [7]. Structurally, HK and ΦC31 att sites differ in size, with the HK attB sites being generally shorter than the HK attP sites, 21 base pairs (bp) vs 234 bp [8, 9]; in addition, attP contains binding sites for Int and Xis along with host-encoded proteins Fis and IHF [8–11]. ΦC31 attB and attP sites are similar in size (~50 bp) and do not require additional proteins to carry out recombination [12].

The use of SSRs generally involves selecting the recombination event through the use of a marker gene within the inserted sequence whose presence or absence would indicate successful integration [1]. Genes can be activated following recombination through either removal of blocking DNA sequences or by bringing together physically separated congruous sequences, with the recombination site embedded within the gene or between the promoter and coding sequence. This approach has long been used with the popular CRE/loxP [13] and Flp/FRT [14] systems. The β-lactamase (bla) gene is an attractive marker, as it is a useful reporter gene for both pro- and eukaryotic applications [15]. Protein chimeras of β-lactamase demonstrate tolerance to exogenous peptide insertions [16], even for domains of unknown function [17]. A split gene reassembly approach using bla has also been developed to discover directed evolution-modified SSR enzymes capable of recombining designer sequences [18]. The bla signal peptide is an attractive region for peptide insertion [19], as insertions between the signal peptide sequence and the rest of the coding gene have minimal interference with protein function [20]. As we wished to expand the available molecular toolbox, we created a set of recombination reporters consisting of the attB of HK and ΦC31 inserted in frame with bla, allowing expression of the gene and enabling the direct selection of recombination events. The selective agent is not expressed when the att sites are in attL and attR form, as the reporter gene fragments are physically separated (Fig. 1a).

This approach has been used to explore the physical structure of the E. coli genome [21, 22]. Genome engineering of the two Vibrio cholerae chromosomes used this tool to understand the evolutionary and genetic implications of multi-chromosomal bacteria [23]. We have used HK recombination in tandem with the λ-lacZ system from [21] to exchange DNA between the two V. cholerae chromosomes in a recombination-mediated cassette exchange (RMCE), resulting in large-scale chromosomal rearrangements [23]. Because the lacZ reporter allows the observation of recombination events but not to select for them, we developed a reporter system for HK recombination based on antibiotic selection. We have used an HK attB site placed in-frame within the β-lactamase (bla) gene to carry out relocation of the S10-spec-α ribosomal locus in V. cholerae in order to study the consequences of essential gene positioning as it relates to dosage [24]. We further used HK-bla to carry out large-scale genome inversions around the origin region (ori) of V. cholerae chromosome one (Chr1) to shift the timing of the initiation of chromosome two (Chr2) replication relative to Chr1 in order to study the mechanisms involved in bacterial chromosome replication timing [25].

Here, we describe the construction and validation of HK-bla and a similar tool using the serine ΦC31 att system (ΦC31-bla). We placed attB sites from each system immediately downstream of the bla signal peptide coding sequence, which directs transport of β-lactamase to the periplasm and is removed in the mature protein. β-lactamase is generally tolerant of insertions into this region. When each system is present as attL and attR sites, they are associated with fragment sequences bla’ (the 5′ region upstream of the cleavage site including the promoter and signal sequence) and ‘bla (the 3′ region comprising the mature protein sequence), respectively (Fig. 1a). In addition, the cognate att site partners show high recombination frequencies without the presence of bla-resistant background from the fragmented bla gene. These systems are extremely useful due to their ability to directly select for recombination through resistance to β-lactam antibiotics. They also have the potential to be used within synthetic biology frameworks for constructing and precisely inserting large genetic assemblies, making them useful additions to the molecular biology toolbox for both synthetic and molecular applications.

In this study, we describe the construction of two site-specific recombination tools useful for DNA manipulation applications. The utility of this attB-bla tool is based on its incorporation of the widely used HK and ΦC31 recombination systems. In the case of HK, the removal of sequences flanking the BOB’ core region reduced attB x attP recombination. This reduction could be due to the removal of bases outside of the attB core that have homology with the attP sequence, which may act to stabilize the attB/attP complex. However, obtaining the highest possible recombination frequency was not critical for the design of this system, as our main concern was β-lactamase function following insertion of the att sites into the bla coding frame.

In directly comparing the two systems, the ΦC31 site appears to recombine at a similar frequency to the 51 bp HK sites and the 23 bp HK ORFs incorporated into bla have a lower recombination frequency (Fig. 2). This decrease is likely due to the reduction of size of the attB _HK site, as the recombination frequencies for the smaller HK site tested independently of bla insertion are not different from the frequencies obtained when they are embedded in bla (Fig. 1). Reported differences between recombination systems in the literature may result from differences in protocols and practices. A recent review of ΦC31 found a wide range of reported recombination frequencies for this recombinase [33]. To our knowledge, the only information comparing HK and ΦC31 recombination frequencies reports HK recombining at a higher frequency than ΦC31 [34]. However, this study used a clonetegration technique where constructs were recombined into native att sites on either the E. coli chromosome for attP _HK or Salmonella typhimurium for attP _ΦC31.

While testing bla expression with inserted ORFs, we observed that ΦC31 ORF 5 interfered with bla expression, while ΦC31 recombination was not affected (Table 1, Fig. 2b). The bla gene used for our system originates from pBR322 and belongs to the TEM-1 class of β-lactamases. The signal sequence is recognized by the Sec export pathway that transports unfolded proteins across the cytoplasmic membrane [26, 35]. DNA secondary structures could be a source of transcription interference, as ORF 5 forms a 30 bp hairpin (∆G at 37 °C = −9.09 kcal/mol). However, hairpins are formed in all 6 ORFs at similar ∆G, making it unlikely that this factor alone prevents bla expression. At the translation level, the overall charge of the first 5 amino acids following the signal sequence can influence cleavage and cross-membrane transport, as they generally have an overall negative charge [36]. For ORF 5, the overall negative charge of this region is +2. Again, however, this is unlikely to explain the loss of bla expression, as only ORF 1 has an overall negative charge, at −1. The amino acids in the 1 and 2 position after the cleavage site can also influence protein function [37, 38]. For ORF 5, the first two amino acids are glycine and serine. Analysis of 307 proteins from the SPdb database [39] found that in Gram-negative bacteria, glycine occurs in the 1st position in 6.19% of proteins, and serine appears in the 2nd position in 5.54% of proteins. [40]. Additionally, two of the 307 Gram-negative proteins analyzed in this study begin with glycine-serine. Thus it is unlikely that the first two residues of the ORF 5 sequence alone interfere with protein transport. More experimental and analytical work is needed to determine the source of bla expression interference.

The high tolerance of bla to in-frame DNA sequence insertion downstream of the bla promoter and leader peptide sequence allows for further modifications of this system through insertion of potentially large ORFs. This approach has already been proposed as an “ORF-trap” to capture DNA encoding protein fragments [41]. Indeed, large ORFs in frame with bla may not greatly reduce β-lactamase function, although export to the periplasm can be inhibited [42]. Additionally, as attB and attP site reactivity can be modified through mutations to their respective core sequences, variable non-reacting “synthetic” att sites can be designed for sequential introduction into the bacterial chromosome [43].

Integration of exogenous DNA sequences into genomes by SSRs generally involves the recombination of an attP site on the inserted sequence with an endogenous chromosomal attB or pseudo-attB site [1]. The use of genome editing technologies allows the insertion of recombination sites that differ from native sites in location and sequence. Native att sites may be located in undesirable regions of the genome, for example, in an active gene locus, or a locus subject to silencing. Additionally, dosing effects can be observed in bacterial species dependent on a gene’s location in the chromosome [24]. Engineering att site recognition by Int proteins allows the creation of semi-synthetic partner sites [27, 43]. This would avoid recombination with other native att sites, and could allow rapid construction of synthetic gene networks. The addition of FRT sites flanking the bla-attB cassette would further allow for removal of the resistance selection marker gene. Similarly, gene-editing technologies could allow the targeted insertion of att sites to serve as landing pads for insertion. In this way, the bla’-attL sequence from our system can be inserted into a genome, into which a sequence containing the partner attR-‘bla can be inserted through attL x attR recombination. This framework has already been proposed for the construction and insertion of metabolic networks into eukaryotic cell lines [44]. Our system adds the advantage of avoiding marker expression until recombination, making it versatile for synthetic applications as well as genome-scale engineering.

We describe here the construction of new tools based on two different site-specific recombination systems, the tyrosine recombinase HK, and the serine recombinase ΦC31. Recombination for each system is reported based on the reconstitution of the bla ampicillin resistance gene, providing resistance to β-lactam antibiotics as a selective agent. Both HK-bla and ΦC31-bla are useful for selecting recombination events in a genomic context due to a high rate of recombination frequency, directionality based on the recombination proteins supplied in trans, and the ability to carry out in vivo genomic rearrangements. We have previously used this tool in our lab to carry out large-scale reorganization of the V. cholerae chromosomes to study the importance of chromosome size in multi-chromosomal bacteria [23], the relevance of genome position and chromosome location for gene dosage and its evolutionary importance [24], and the timing of V. cholerae chromosome replication [25]. The importance of these tools lie in their capacity to exist simultaneously in the cell at two separate loci without expression of the marker gene until expression of the recombination proteins is induced.