Introduction

Bacteriophages (phages) are viruses that infect bacterial cells and use the host cell machinery to generate progeny. Phages are divided into two major categories: “virulent” and “temperate”. After infection of a sensitive host cell, lytic phages immediately propagate themselves by using the host cell molecular machinery, releasing thousands of new phage particles into the environment upon killing of the infected cells. In contrast, temperate bacteriophages “make a decision” between lytic growth and lysogeny, a latent state, whereby the genome is passively replicated until the prophage is induced (Groth and Calos 2004). Temperate bacteriophages that integrate their genome into the host chromosome in the establishment of lysogeny facilitate integration by encoding an integrase enzyme that mediates unidirectional site-specific recombination between two DNA recognition sites, the bacteriophage attachment site, attP, and the bacterial attachment site, attB. The lysogen in this case is flanked by two hybrid sites, attL and attR, which consist of half attP and half attB sequences that become the substrates for extreme recombination by the excisionase. Several unfavorable environmental factors affect the excisionase reaction, which is followed by lytic phage release and the initiation of a new infection cycle. The ability of phage integrases to unidirectionally recombine two short DNA sequences offers utility toward many genetic engineering applications (Groth and Calos 2004). As such, phage integrases are of growing importance in the genetic manipulation of living eukaryotic cells, particularly mammals that possess large genomes, for which there are few tools for precise genetic manipulation. Integrases of different phages have been shown to work efficiently in mammalian cells, carrying out efficient integration at introduced native att sites or pseudo sites available on the host genome with partial identity to att sequence. This reaction may be exploited in biotechnical applications such as manipulation of prokaryotic and eukaryotic cells, bacterial engineering, construction of transgenic higher eukaryotes, or cell/gene therapy. Each phage-encoded integrase recognizes specific individual sequences. Some integrases act autonomously, while others act with the help of other phage proteins and/or bacterially encoded host factors (Groth et al. 2000).

Homologous recombination

The process of homologous recombination involves the alignment of similar DNA sequences to form mobile forks between four strands of DNA termed Holliday junctions that are highly conserved in both prokaryotic and eukaryotic cells. Homologous recombination requires energy and cofactors from host cells (Thyagarajan et al. 2000). The frequency of integration by homologous recombination appears to generally occur at a frequency of about 10−6 for most mammalian cells, although it was reported that this frequency may be increased by up to 20-fold using completely isogenic DNA or up to 1,000-fold following the introduction of a double-strand break at the target site; however, the resources necessary to generate such a break into endogenous sequences remains a challenge (Porteus 2012).

The highly efficient genetic engineering system, “recombineering”, is based on a homologous recombination, but exploits a phage-encoded recombination system (Thomason et al. 2001). Recombineering and homologous recombination now represent the methods of choice for knock-in and knock-out strategies to either manipulate bacterial and mammalian genome or as an alternative to traditional restriction enzyme-mediated cloning with PCR products (Thomason et al. 2001; Clokie et al. 2009; Sharan et al. 2009). Homologous recombination systems occur via either the host RecA-dependent or RecA-independent pathways or Escherichia coli RecA-dependent recombination pathway.

In prokaryotes, homologous recombination plays an important role in the generation of genetic diversity and in DNA repair processes. One of the best studied mechanisms is the RecA-dependent or the RecBCD pathway that is useful in the repair of chromosomal DNA double-strand breaks (Dillingham and Kowalczykowski 2008). The RecBCD recombination complex recognizes specific sequences within the bacterial chromosome called “chi sites” that prevent DNA degradation. Foreign DNA segments that lack chi sites are thus degraded to protect the E. coli cell from invading nucleic acids. RecBCD, also known as Exonuclease V (Exo V), is an enzyme that recognizes a double-stranded (ds) DNA break and initiates recombination and repair. The enzyme complex is composed of three different subunits called RecB, RecC, and RecD, hence the complex name, RecBCD. Both the alpha (RecD) and beta (RecB) subunits are energy-dependent helicases that unwind and separate the strands of DNA, where the beta subunit is also a nuclease that introduces single-stranded (ss) nicks into the DNA. The gamma subunit (RecC) recognizes the 5′-GCTGGTGG-3′ (chi) sequence to initiate recombination (Wigley 2013).

The multifunctional protein, RecA, is a significant contributor to recombinational repair of damaged DNA in E. coli, with a functional homolog present in every species. RecA activity is energy-dependent, where the protein strongly binds to and coats the single-stranded DNA generated by the exonuclease activity of Exo V to form a nucleoprotein filament. Since it has more than one DNA-binding site, RecA can hold single- and double-stranded DNA together, a feature that makes it possible to catalyze a DNA synapse reaction between a DNA double helix and a homologous region of single strand (ssDNA). This reaction initiates the exchange of strands between two recombining DNA double helices, which progresses down the strand by DNA branch migration (Dillingham and Kowalczykowski 2008).

We previously reported on the application of RecA+ E. coli K-12 (W3110) cells to process homologous recombination-mediated insertions of bacteriophage-derived recombinase genes into the lac operon of E. coli using the pBRINT-cat integrating plasmid (Nafissi and Slavcev 2012). The pBRINT family of integrating vectors facilitates the homologous recombination-mediated chromosomal integration of cloned sequence of interest into the lacZ gene of E. coli in a RecA-dependent manner (Le Borgne et al. 1998). In contrast, a recA mutant host is often employed for cloning and plasmid maintenance to ensure the stability of extrachromosomal DNA.

The RecA-independent bacteriophage λ Red recombination system

The recombination genes of the bacteriophage λ Red system were among the earliest genes to be described. Following the isolation of recA, recB, recC, or recD recombination-deficient mutants, it was found that phage λ could effectively initiate homologous recombination independent of any of the bacterial recombination pathways due to the encoding of its own recombination enzymes. The λ Red recombination was thus considered to be a separate pathway (Christensen 2001). The red system is comprised of the exo (α), bet (β), and gam (γ) genes, all clustered in the pL-governed operon of the λ genome and is regulated by the CI repressor (Stahl 1998). The exo gene encodes the 24-kDa 5′-3′ exonuclease (Exo) that targets dsDNA ends; bet encodes the 28-kDa ssDNA-binding protein (Bet) capable of annealing complementary ssDNA strands; and gam encodes the 16-kDa polypeptide (Gam) that inhibits Exo V activity and confers protection against nuclease attack and digestion of the λ linear dsDNA genome. During the λ lytic cycle, λ Red recombination is stimulated by endonuclease activity of the λ terminase on the phage’s terminal cohesive ends (cos) sites, thereby introducing the dsDNA break following the 5′-3′ exonuclease activity of Exo that provides the 3′ overhang to initiate recombination (Fig. 1). In this case, if an intact dsDNA with no free ends is the only available partner, recombination proceeds through the E. coli RecA-dependent mechanism by strand invasion. In contrast, if the available partner is a damaged/nicked dsDNA or a replicating chromosome with a free double-strand end, then recombination can proceed in a RecA-independent manner through Bet-mediated annealing. The Red recombination system is facilitated by the λ Gam function that inhibits E. coli RecBCD exonuclease activity to prevent further digestion of λ phage DNA by host cell, thus promoting Exo and Bet access to DNA ends to promote Red-mediated recombination (Poteete 2001).

Fig. 1
figure 1

The molecular mechanism of Red-mediated recombination

Full size image

The Red recombination system was applied for the first time in 1998 for efficient and simplified genetic engineering of bacterial cells (Murphy 1998). The system was exploited by expressing the λ red genes, either via plasmid or chromosomally inserted, to promote recombination between the bacterial chromosome and a nonreplicative linear dsDNA molecules introduced into the cell. Most of the recombination systems that had been developed previous to this study required long homology arms (>500 bp), whereas the plasmid-encoded red system facilitated insertion and recombination events between PCR products with short homology arms (>50 bp) into the bacterial chromosome with high efficiency. This technology has since been applied for genome editing in different prokaryotic cells, inserting genes for long term and stable production of recombinant proteins and enzymes (Murphy et al. 2012).

Site-specific recombination

Permanent modification of the target genome is of great utility in genetic engineering and is generally carried out by three different approaches: (1) random integration (illegitimate recombination), (2) homologous recombination, and (3) site-specific recombination. Although random integration can be used to place the introduced gene into the genome, the lack of control over the position of introduced DNA can result in undesirable side effects, including unpredictable expression of the introduced gene and potential mutagenesis of neighboring genes. As such, it is a method best suited for insertional mutagenesis. A strategy that can exclusively yield efficient site-specific integration into safe locations in the target genome would be ideal. However, while homologous recombination provides excellent specificity in integration sites, it occurs at too low frequency to be optimal for genetic engineering in multicellular organisms (Kuhlman and Cox 2010).

Site-specific DNA recombination systems are derived from prokaryotes and unicellular yeasts, and among them, bacterial viruses provide a repertoire of recombinational systems, a number of which have been exploited to facilitate efficient DNA exchange in human cells. As previously mentioned, temperate bacteriophages are often integrated into the host chromosome at a specific site by recombination, a process that requires specialized “attachment sites”, attP and attB in the phage and the host chromosomes, respectively. In the prophage, the enzymes generate two junction sites, attL and attR, consisting of hybrid sequences of attP and attB, and the integration reaction strictly conserves the original sequences in the recombined product sequences. Lysogeny is reversible, and a minority of cells within a population of lysogens will spontaneously lyse, releasing progeny phage. The integrated prophage is flanked by attL and attR, which are the substrates for excisionase recombination, a process that reforms attP and attB and releases the prophage. During the initial stage of the conversion to the lytic phase, a phage-encoded excisionase catalyzes the excision of the prophage genome by cooperating with the phage integrase, and the original attP and attB sequences are regenerated via site-specific recombination between attL and attR (Turan et al. 2011).

Site-specific recombination facilitates integration, excision, and inversion of defined DNA segments that usually requires no DNA synthesis or high-energy cofactor (Groth and Calos 2004). Site-specific recombinases are classified into two major families of tyrosine or serine, which were named based on the amino acid residue that forms a covalent protein-DNA linkage in the intermediate reaction. Although enzymes of the site-specific recombinase families share high specificity and efficiency, their recombination mechanisms are distinctly different. In fact, all recombinases typically mediate efficient “cut and paste”-type DNA exchange between recognition sites in the range of 30–40 bp or longer (Groth and Calos 2004). Recombination is conservative and is characterized by the following events: (1) recombination occurs at a specific site on the interacting DNA molecules, (2) expression and synthesis of the recombinase enzyme, (3) strand exchange occurs at small regions of DNA homology within the recombining sites, (4) pairing of the interacting recombination sites followed by strand exchange resulted in structural intermediates, and finally (5) resolution of intermediates followed by strand migration (Groth and Calos 2004). However, tyrosine recombinases form “Holliday intermediates” by cleaving one strand of each duplex at a time; each crossover site is nicked and must be joined to its partner before the second strand can be cut. In contrast, serine recombinases cut all strands in advance at both crossover sites before any exchange, following 180° rotation and rejoining of cleaved substrate (Sauer and Henderson 1988). The tyrosine-type recombinases are subdivided into two families based on the directionality of site-specific recombination: (1) unidirectional tyrosine-type phage integrases and (2) bidirectional tyrosine-type simple recombinases. Generally, the tyrosine family of integrases tends to recognize longer attP sequences and requires helper proteins encoded by the phage or the host bacteria for their activity. In contrast, serine recombinases recognize shorter attP sequences that are still long enough to be specific on a genomic scale and do not require host cofactors. Serine-type recombinases are subdivided into two families based on protein size: (1) small serine-type resolvase/invertases and (2) large serine-type phage integrases (Hirano et al. 2011).

Some integrases function with no requisite for cofactors or helper proteins and therefore offer improved activity in heterologous environments. From this group, bacteriophage P1-derived Cre (cyclization recombinase), bacteriophage lambda-derived Int (integrase), and Saccharomyces cerevisiae-derived Flp have been applied widely in gene manipulation of higher eukaryotic organisms and in the production of transgenic species (Groth et al. 2000). Recombinases such as Cre, Flp, and Int perform both integration and excision at the same target sites, and thus, their net integration frequency in mammalian cells is generally low. An ideal application would see that the transgene remain inexcisable following integration (Groth and Calos 2004); a property shown by some phage-derived recombinases such as the Streptomyces phage-encoded ΦC31 integrase (Chavez and Calos 2011), Escherichia phage-encoded N15 TelN protelomerase (Deneke et al. 2000), and Yersinia phage-encoded PY54 Tel protelomerase (Hammerl et al. 2007) (Table 1).

Table 1 Characteristics of site-specific recombinases active in mammalian cells
Full size table

Here, we examine the application of phage-derived enzymes toward (1) human genome engineering strategies and regenerative medicine and (2) vectorology and in engineering of superior DNA transgene delivery vectors.

Phage lambda integrase

Phage λ-derived integrase (Int) is a conservative site-specific recombinase that belongs to the tyrosine-type integrase family of recombinases. The natural function of Int is to carry out integration of the λ genome into the E. coli genome. In combination with the Xis (excisionase) protein, Int is also involved in excision of λ upon prophage induction. Although the Int-att system was considered as a potential candidate for human cell genome manipulation, it was not selected due to certain obstacles encountered in clinical studies. Int requires the host-encoded proteins (integration host factor, IHF) for efficient integration, which limits its integrative recombination activity in heterologous cells. However, several attempts to overcome the problems associated with the use of these integrases in heterologous cells have been made (Sanger et al. 1982). For example, IHF-independent mutant λ integrases have been generated that catalyze DNA intermolecular integration and excision reactions at low levels in the absence of accessory proteins in mammalian cells (Lorbach et al. 2000). We previously used the conditional replication, integration, and modular (CRIM) integrating plasmids (Haldimann and Wanner 2001) possessing the λ attP site-specific sequence to insert a foreign operon into a Rec+ E. coli K12 (W3110) genome (Nafissi and Slavcev 2012). These modified “CRIM” plasmids were integrated into the host bacterial attachment (attB) site by supplying phage λ integrase (Int) from a helper plasmid. This system proved very efficient for site-specific integration of the gene of interest (GOI) into the host genome. However, λ integrase usually catalyzes bidirectional site-specific recombination in cells other than the natural E. coli host, which limits its application in clinical studies where the strict unidirectional integration of foreign DNA into a heterologous genome is desirable (Lorbach et al. 2000; Hirano et al. 2011). As such, λ Int is not an ideal recombinase for stable and long-term human genomic modifications.

Phage P1 Cre recombinase and S. cerevisiae Flp recombinase

Phage P1-derived Cre (“causes recombination”) and S. cerevisiae-derived Flp (“flipase”) recombinases were discovered by Austin et al. (1981) and Broach et al. (1982), respectively. These recombinases catalyze reversible site-specific recombination events between two short, identical sequences: a 34-bp loxP target site for Cre and the 48-bp FRT site for Flp. Both Cre and Flp belong to the simple/bidirectional tyrosine-type recombinase family and, as such, do not require host-encoded accessory proteins or specific DNA structures (Hirano et al. 2011). The simplicity of the Cre/loxP and Flp/FRT systems has led to their extensive use as tools for in vivo genomic engineering such as DNA rearrangements and gene knock-in and knock-outs in a wide variety of heterologous environments, including human cells. Both recombinases have been extensively used to activate or switch gene expression in clinical applications, and integration of foreign DNA sequences into chromosomally integrated loxP or FRT sites is reversible since wild-type Cre and Flp recombinases predominantly catalyze re-excision of the integrated DNA molecule via the reverse reaction. This feature has been widely applied in knock-out studies when excision of a particular gene is required.

Cre recombinase differs from typical tyrosine-type temperate phage integrases in that it is not involved in prophage establishment, but rather resolution. Cre is essential for P1 lysogeny as the prophage does not integrate into the host chromosome, but rather replicates as an episomal replicon in its host. The P1 lysogenic cycle begins with circularization of the injected linear P1 DNA into infected cells by site-specific recombination at loxP sites in the phage genome. P1 plasmids replicate independently from the host chromosome, and Cre/loxP activity resolves dimeric P1 plasmids into monomers, thereby facilitating stable partitioning of P1 into daughter cells during cell division (Austin et al. 1981). Cre activity in cointegrate resolution limits the utility of the Cre/loxP system for applications involving stable genomic integration. The Flp recombinase from yeast with 2-μm circle, similar to Cre, also differs from typical temperate phage integrases in its physiological function. Initially, the Flp recombinase leads to amplification of the 2-μm circle by inverting the DNA segment flanked by the two FRT sites; this rearrangement initiates rolling-circle replication by changing the direction of the replication fork. Next, the enzyme resolves the multimeric array generated by rolling-circle replication into the 2-μm circle monomers (Broach et al. 1982). Thus, the physiological function of Flp recombinase is the inversion and excision of a DNA segment flanked by two FRT sites. As such, Flp/FRT, like Cre/loxP, is not ideal for applications requiring stable integrative recombination.

Several strategies have been employed to overcome the issues associated with application of these recombination systems to unidirectional integrative recombination. For instance, the recombination-mediated cassette exchange (RMCE) and the “LE/RE mutant system” have led to the development of two types of mutant loxP sequences to facilitate unidirectional site-specific recombination by Cre. The Cre recombinase recognizes 34-bp loxP sites, each of which consists of two 13-bp inverted repeats that bind to Cre and a central 8-bp spacer region where strand exchange occurs. If one nucleotide in the central 8-bp spacer region is modified, Cre recombinase is still able to recognize the modified target site and carry out the integration event, but cannot execute its excision function. This was shown in heterologous genomes, where a pre-inserted DNA sequence flanked by two mutant loxP sites containing modified spacer sequences, resulted in the loss of Cre endonuclease capacity (Bouhassira et al. 1997). Modified loxP sites have been used to integrate foreign DNAs into a wide variety of heterologous host genomes including those of plants (Albert et al. 1995), mammals (Araki et al. 1997), and bacteria (Suzuki et al. 2007).

RMCE-type recombination based on the Flp/FRT system has also been utilized to facilitate unidirectional integrative recombination since FRT target sequences are longer and more complex than loxP sites (Turan et al. 2011). Each 48-bp FRT site consists of three 13-bp repeats and an 8-bp spacer region, in which strand exchange occurs. Two nearly identical 13-bp inverted repeats bind the Flp recombinase and flank the 8-bp spacer, while the additional 13-bp repeat follows a 1-bp gap forming a perfect direct repeat with the repeat at one side of the spacer. While the inversion and excision reactions occur with a 34-bp core site consisting of the two 13-bp inverted repeats and 8-bp spacer region, Flp-mediated recombination is more efficient with the intact FRT site that contains the third repeat serving as an extra Flp-binding sequence (Hirano et al. 2011).

Both λ and P1 phage-derived site-specific recombination systems have been exploited toward the production of modern plasmid-derived DNA transgene delivery vectors (Table 2).

Table 2 Systems developed for production of bacterial sequence-depleted DNA vectors
Full size table

The first system described in 1997 for bacterial sequence-free DNA “minicircle” production, employed the λ Int/attP site-specific recombination system. In this study, recombinant Int was expressed endogenously from E. coli, and two attP target sequences were inserted into a plasmid DNA vector flanking the eukaryotic expression cistron. Gene int expression resulted in site-specific excision of att sites and generation of two mini circular constructs, one of which contained the bacterial sequence to form the “miniplasmid” and the other encoding the gene of interest (GOI) cistron to form the desired “minicircle” (Darquet et al. 1997). Despite the obvious advantage, the system was limited by the toxicity of phage λ Int recombinase in bacterial cells, thus necessitating tight control over the recombinase induction process in order to produce the minicircle in culture, which led to inefficient recombination (40–70 %) (Darquet et al. 1997, 1999). As such, the project instead turned to an alternative strategy of using the Cre-loxP recombination system. Similar to the original design, Cre was expressed either endogenously or episomally (Bigger et al. 2001). Cre would then catalyze site-specific recombination between loxP sites available on standard plasmid DNA vector (pDNA) leaving the bacterial sequence-free minicircle for further application (Bigger et al. 2001). Until recently, the Cre-loxP recombination system was the method of choice to generate DNA minicircle vectors, despite its limitation in processing pure DNA minicircles with no conventional plasmid residual contamination. The in vivo efficiency of this system was <100 %, and the undigested parent pDNA was one of the most challenging steps in purification. Other groups have since applied different recombination systems and tactics to improve DNA minicircle production leading to the achievement of highly pure minicircles with very low parent plasmid or miniplasmid residual contamination (Kay et al. 2010) (Fig. 2).

Fig 2
figure 2

The in vivo production of minicircle DNA vectors. Vectors are generated by site-specific recombination via two excision sequences (blue rounded rectangle), which flank the mammalian gene of interest expression cassette (green rounded rectangle) and the bacterial amplification unit (black rounded rectangle) in a recombinant plasmid vector. Arrows indicate the direction of the recombination reaction. System employing the phage P1 Cre, or ΦC31 and λ Int, recombinase supplied by the bacterial host strain and the corresponding target sequences loxP or attB and attP, respectively

Full size image

Phage ΦC31 integrase

In the 1990s, actinophage ΦC31 integrase was discovered and found to be homologous to resolvase/invertases (Kuhstoss and Rao 1991). This enzyme belongs to the large integrase serine-type recombinase family and catalyzes unidirectional site-specific recombination between attP and attB sites, both of which are normally ~40 bp in size and do not require any supplementary host factors or specific DNA structures. In general, both att sites have an identical 2-bp core sequence flanked by 20–25-bp imperfect, inverted repeats (P and P′ in attP; B and B′ in attB), where there is little sequence similarity between attP and attB sequences. This integrase requires the slightly longer attP sequence for efficient integration, and attP encodes more perfect inverted repeats than attB. The integrase possesses highly homologous N-terminal catalytic domains (140 amino acids (a.a.)) and diverse C-terminal DNA-binding domains (300–500 a.a.), and the DNA-binding domains confer sequence-specific DNA binding to distinct att-integrase pairs (Kuhstoss and Rao 1991). These integrases function on a recombination mechanism similar to that of the resolvase/invertase, serine-type recombinases, involving the 180° rotation and rejoining of cleaved att DNAs, except that they recognize two different sites and catalyze a tightly regulated unidirectional recombination. The unidirectional recombination is mediated by the larger C-terminal DNA-binding domains of the integrase that possess a motif controlling recombinational direction. The excision reaction requires the phage-encoded excisionase, Xis, which binds tightly to the attP- and attB-bound dimers, inhibiting the integration reaction, and binds less strongly to the attL- and attR-bound dimers to facilitate the excision reaction. It is this characteristic in particular that makes integrases the most ideal recombination enzyme of choice in clinical applications where there is a need for insertion of a specific gene into a heterologous genome and for provision of a long-term and stable expression of the protein product (Keravala et al. 2011).

The relative simplicity of serine-type phage integrase-mediated recombination has spurred the development of heterologous gene integration systems based on these enzymes. Genomic integration systems based on ΦC31 integrase have been developed for human genome editing (Brown et al. 2011). This integrase efficiently catalyzes the site-specific integration of an attP-containing DNA vector into preexisting (engineered) attB or attP sites in the mammalian genome. The second strategy (attP) imparted with better outcome owing to the complexity and lower accessibility of the chromosomal DNA compared to that of the pDNA structure (Rice et al. 2010). In addition, “pseudo attP” sites with only 40 % identical sequence to wild-type attP have been identified that can bind to ΦC31 integrase and mediate ΦC31 integrase-dependent integration at 10-fold higher rates than random sites. Interestingly, these “pseudo attP” sites are often present in transcriptionally active regions of genome and euchromatin chromosomal DNA, and thus, the insertion of a vector into pseudo attP sites often provides higher expression levels than those inserted randomly into the human genome. These findings indicate that multiple distinct site-specific recombination systems, based on various serine-type phage integrases, may be employed to simultaneously or sequentially introduce multiple foreign DNAs into att sites pre-inserted into heterologous genomes.

The integration frequency mediated by the ΦC31 integrase in mammalian cells is approximately 10- to 100-fold higher than Cre-mediated integration at an inserted wild-type loxP site, higher than FLP-mediated integration at an inserted FRT site, and more efficient than phage integrases of the tyrosine-catalyzed site-specific recombinase family, such as λ Int (Thyagarajan et al. 2001). Furthermore, recent advances in recombinase engineering showed promising results to target the GOI into a specific site in the mammalian genome, for example, by fusing a DNA binding domain into the serine family recombinase, serin-type-mediated recombination rates were dramatically improved in mammalian cells (Gaj et al.).

Phage N15 TelN protelomerase

The E. coli bacteriophage, N15, discovered by Ravin in 1964 in Moscow, has some unique features compared to other temperate bacteriophages with respect to its genetic composition and organization (Ravin et al. 2000). N15 not only represents the first example of a linear prophage genome with covalently closed ends, but still remains the only known naturally occurring example of a linear plasmid in E. coli since its discovery (Valentin and Rybchin 1999). While N15 virions are very similar to phage lambda in morphology, burst size, plaque morphology, and length of genome, N15 prophage is very unusual in that it exists and autonomously replicates as a linear covalently closed (LCC) DNA plasmid in the lysogenic state. Like λ, the linear N15 mature packaged phage genome also possesses ssDNA cohesive ends, cosL and cosR. The linear plasmid prophage termini, telR and telL (telRL), are located near the center at position 24.8 kb on the mature N15 DNA map. The telRL site is located in a similar site that is occupied by the attachment site, attP, in λ and defines the boundary between the left and right arms (Ravin et al. 2000). In the mature N15 genome, the genes responsible for formation and maintenance of the linear plasmid in the host cell (protelomerase, anti-repressor, replication, and partitioning genes) are located on the left (0–10 kb) arm of the genome. This cluster has some overlap with genes involved in the lytic growth of phage although most lytic genes are located on the right arm. Upon deletion of the sequences between the left and right arms, forming a 5.2-kb mini plasmid product, the derivative is still capable of replication (Ravin 2011).

During the establishment of N15 lysogeny, the genome circularizes by annealing and ligating its terminal cohesive (cos) ends. The telomere-forming telRL site on the circular plasmid prophage becomes the target for N15-encoded protelomerase cleaving-joining enzyme (Deneke et al. 2002). The telRL sequence contains a 56-bp palindrome with two different base pairs at the 12th and 14th positions on both sides of the palindrome center (telO), which forms a 14-bp perfect palindrome recognized and cleaved by TelN protelomerase to create ssDNA termini. After cleavage, the self-complementary ssDNA ends reanneal and create perfect hairpin structures at each end (Valentin and Rybchin 1999).

TelN is a slightly acidic 630 a.a. protein that is about 72.2 kDa in size, and while essential for the generation and replication of a linear plasmid prophage, it also plays a significant role in the lytic growth of N15 phage (Deneke et al. 2002). TelN and λ Int play a similar role in the creation of lysogeny. As previously mentioned, the C-terminal segments of integrases are highly conserved, particularly their catalytically active domains. Analytical studies on TelN structure revealed a region (residues 390–427) that is paralogous to the C-termini of integrases that belong to the tyrosine recombinase family. TelN possesses a tyrosine at the same position as integrases, which suggests that protelomerases and tyrosine recombinases possess similar catalytic mechanisms for DNA cleavage and ligation and likely evolved from a common ancestor. Both generate a 3′-phosphotyrosine DNA intermediate that enables the covalent rejoining of cleaved DNA strands without the use of a high-energy cofactors (Valentin and Rybchin 1999). TelN cleaving-joining activity is functional in the absence of any other N15-encoded factors, and purified TelN is the only protein required to convert circular DNA substrates possessing the telRL sequence to the linear molecule with covalently closed ends (Heinrich et al. 2002). In vitro studies on the functionality of purified TelN demonstrate that in the presence of the 56-bp telRL target site, and in particular the 22-bp telO target site, purified TelN processes circular and linear pDNA in vitro with the same covalently closed end DNA topology as was observed in vivo (Deneke et al. 2000). In addition, it was shown that the central telomerase occupancy site (tos), within the telO perfect palindrome, is occupied by two TelN molecules to create a stable TelN-telRL complex. The telO palindrome alone is not sufficient for specific stable complex formation and requires the additional sequences of telRL to stabilize the TelN-target complexes that process the binding reaction of TelN homodimers to the 3′-phosphoryl end of the cleaved strands and the generation of LCC DNA (Deneke et al. 2002).

TelN protelomerase represents a significant addition to the repertoire of exploitable enzymes in biotechnology with respect to its unique property in linearizing the plasmids carrying its target telRL site. The “pJAZZ” series of transcription-free, linear cloning vectors and “linear miniplasmids” are N15-based linear cloning systems that facilitate cloning of up to 30-kb DNA sequences, including AT-rich inserts or inverted repeats, which are very difficult to clone by standard methods into plasmid vectors (Ravin et al. 2000). TelN/telRL cleaving-joining activity has also been applied to enhance in vitro production of LCC DNA vectors (Fig. 3), where LCC DNA vectors demonstrate heightened and sustained expression of a gene of interest both in vitro and in vivo (Heinrich et al. 2002).

Fig. 3
figure 3

The in vitro application of TelN/telRL recombination system. a Schematic presentation of eGFP expressing plasmid DNA carrying the telRL target site. b Schematic representation of the processing of the telRL substrate by recombinant TelN protein. TelN cuts the telRL site and produces a linear covalently closed (LCC) plasmid DNA (modified from Heinrich et al. 2002)

Full size image

Phage PY54 Tel protelomerase

Bacteriophage PY54 was first isolated in 2003 from Yersinia enterocolitica as a temperate phage, belonging to the Siphoviridae family with a 46-kb dsDNA genome (Stefan et al. 2003). Like N15, PY54 also exhibits a λ-like morphology with similar genome size and is a temperate phage that coexists with the infected host as a linear plasmid with similar telomere-like hairpin ends. However, despite these similarities, these two phages are evolutionary quite distant, where N15 is far more closely related to λ than to PY54 (Stefan et al. 2003).

A paralog of N15 telN, the PY54 tel gene encodes a 77-kDa TelN-like protelomerase with observably identical function and is able to process recombinant plasmids containing the 42-bp telRL-like palindrome (pal) (Stefan et al. 2003). The pal site (Fig. 4) is a 42-bp perfect palindrome, that unlike N15 telRL, possesses only partial function in vivo in the absence of adjacent sequences (Stefan et al. 2003). Sequencing and nucleotide BLAST of telRL and pal indicates that the 10-nucleotide sequence, 5′-TACGCGCGTA-3′, in the center of the paralogous palindromes are 100 % identical. Furthermore, the protein BLAST alignment of TelN and Tel also reveals 60 % similarity (Fig. 5).

Fig. 4
figure 4

The phage PY54 pal sequence and the mechanism of conversion into closed ends. The 42-bp palindrome is flanking, inverted repeat and the putative start codon are boxed (modified from Hertwig et al. 2003)

Full size image
Fig. 5
figure 5

Protein sequence alignment of the PY54 and N15 protelomerases. The amino acid alignment of N15 TelN and PY54 Tel proteins showing ~60 % a.a. identity (black box) and similarity in gray box

Full size image

To date, there are no in vivo applications of the Tel-pal system of PY54 such as the construction of Tel-expressing cells in genome engineering, or in the production of new pDNA vectors that exploit this system. While the generated LCC and mini LCC DNA vectors have been employing the TelN-telRL system, enzyme production was limited to in vitro use and LCC DNA processing. We are currently applying the N15 and particularly the PY54 recombination systems to develop a novel in vivo technology for production of superior bacterial sequence-free mini DNA vectors for transgene delivery. LCC DNA vectors may be superior to conventional CCC DNA vectors in that they are more efficient and likely safer. The obvious drawback to LCC DNA vectors is the multistep approach to production that adds expense and makes them less desirable in clinical trials due to the fact that pDNA needs to be purified from bacterial cells, digested, chemically modified to produce the LCC DNA, and then finally purified as LCC vector. We have developed an in vivo, cost-effective, one-step, and robust technology for the production of LCC transgene expression cassettes, called “DNA ministring” vectors. This technology exploits the phage PY54 recombination system in comparison to N15 (Nafissi and Slavcev 2012). The PY54-derived Tel-pal system relies only upon protelomerase activity and the palindromic target site, offering the major advantage of a single enzyme and a single-step cleaving-joining enzymatic reaction that can be manipulated for unlimited in vivo production of mini LCC DNA vectors.

Conclusion and perspective

In recent decades, several site-specific recombinases have been discovered from various microorganisms and applied to in vivo genome engineering for a wide variety of heterologous cells. These enzymes have been classified into tyrosine-type and serine-type recombinases according to their catalytic mechanisms, where both families are further subdivided into two subgroups identified (see Table 1). Tyrosine-type phage integrases, such as λ integrase, catalyze tightly regulated unidirectional recombination, but due to their characteristic bidirectional recombination in heterologous cells, their application to chromosomal insertions in eukaryotes remains limited. In contrast, the tyrosine-type simple recombinases, such as Cre and Flp, catalyze bidirectional recombination and have been widely used as tools for gene integration into heterologous genomes by development of RMCE and LE/RE mutant site systems. Serine-type phage integrases, such as ΦC31 integrase catalyze unidirectional recombination between short attachment sites with no need for additional host-encoded proteins, cofactors, or specific DNA structures, which, due to simplicity, favors their use in the development of human gene integration systems. Homologous recombination is generally favored for site-specific introduction of foreign DNAs into a wide variety of heterologous genomes, but the efficiency of homologous recombination is dependent on the host cell features and DNA structure of targeting region. In addition to the abovementioned methods, the “restriction-modification (R-M) and CRISPR-Cas systems” could be a method of choice in the future for more policed recombination in mammalian cells. The R-M and CRISPR-Cas systems is a defending mechanism for bacterial cells against invading phages (Dupuis et al. 2013).