Introduction

Streptomyces species are best known for their propensity to produce secondary metabolites for use as antibiotics, antitumor agents, immunomodulators, anthelmintic agents, and insect control agents. More recently they have become an important source of genetic tools applicable to a variety of biological systems. This stems from fundamental work on actinomycete bacteriophages (actinophages), particularly on ϕC31, a temperate phage for Streptomyces species. Among the temperate actinophages, there are two distinct mechanisms for integration: the coliphage λ-like tyrosine recombinases that integrate into tRNA genes [1, 28, 99]; and the serine recombinases (used by ϕC31 and others discussed here) that integrate into diverse, unrelated structural genes. ϕC31 was first developed as a means to insert cloned DNA into streptomycete genomes, but the unique nature of the Att/Int system rendered it desirable for universal use in diverse cellular systems, including eukaryotes and archaea. The universal utility derives from the unidirectionality [90, 96], and subsequent stability, imparted by the serine recombinase mechanism. ϕC31 is one of several streptomycete temperate phages that have integration mechanisms catalyzed by large serine recombinases. In all cases, the pairs of attP and attB sites share little sequence identities, and the integration and excision mechanisms differ from those of temperate actinophages that utilize tyrosine recombinases. In the present review, I describe the discovery and development of ϕC31 and other streptomycete temperate phages that utilize large serine recombinases, and discuss their applications for stable cloning and expression of genes in streptomycetes and other actinomycetes. The integration systems employing ϕC31, R4, and other serine recombinases have also been used to engineer human cells (e.g., see [32, 45, 56, 60, 80, 104]), and ϕC31 Int has been applied broadly in other eucaryotes, including lower mammals, Drosophila melanogaster, Xenopus laevis, zebrafish, Asian tiger mosquito, Arabidopsis thaliana, Nicotiana tabacum, and Schizosaccharomyces pombe (e.g., see [4, 17, 25, 39, 41, 53, 57, 63, 64, 74, 83, 94, 95, 102, 105]), and in the methanogenic archaean Methanosarcina acetivorans [16, 34], but the details of these studies are beyond the scope of this review. Also, readers are referred to excellent reviews on the molecular mechanisms of integration catalyzed by tyrosine and serine recombinases which are not reviewed here [18, 33, 89, 90].

Temperate bacteriophages that utilize large serine recombinases

At least four temperate bacteriophages that utilize large serine recombinases have been isolated on different Streptomyces species. The best studied phages are ϕC31, R4, TG1, and ϕBT1 (see below). Temperate phages utilizing large serine recombinases have been described from other Gram-positive microorganisms, including Mycobacterium and Lactococcus species [29, 33]. A hallmark of these bacteriophages is that the large serine recombinases require no additional phage or host functions for site-specific integration, and that integration is unidirectional in the absence of additional factors [33, 89]. Recombination between attP and attB sites generates hybrid attL and attR sites which are generally not substrates for excision by Int alone. The excision process has been studied in detail with the mycobacteriophage Bxb1 [29]. In this case excision requires a phage-encoded protein called recombination directionality factor (RDF). Although there is no homolog of the mycobacteriophage RDF in the ϕC31 genome, an RDF has recently been characterized that binds to ϕC31 Int to change its specificity from insertion to excision [46]. A key feature of the Streptomyces phage integration systems is that each has a unique attB site, and the individual attB sites are located in unrelated genes (Table 1).

Table 1 Genes that contain primary bacteriophage attB sites and their gene products in S. roseosporus
Full size table

Bacteriophage ϕC31

ϕC31 is a temperate bacteriophage originally described by Lomovskaya and colleagues [61, 62] that displays broad host specificity for Streptomyces species (but not for other actinomycetes) [47, 100]. As with many other streptomycete bacteriophages, the ϕC31 host range within streptomycetes is limited primarily by type II restriction endonuclease barriers [24, 36, 37, 100]. The biology of ϕC31 and its interaction with streptomycete hosts has been widely studied, but most of these studies are beyond the scope of this review.

ϕC31 integrates via a large serine recombinase into an attB site located in a pirin-like gene (Table 1) located about 85 and 92 kb to the right of the oriC in S. avermitilis and S. coelicolor, respectively (Fig. 1). This lies in the center of an approximately 6.5-Mb region of the linear chromosomes that contain mainly highly conserved genes dedicated to primary metabolism, stress responses, macromolecule biosynthesis, and developmental biology including sporulation [13, 21, 42, 49]. (The S. coelicolor orientation [13] has been reversed to line up with the S. avermitilis genome in Fig. 1). The mechanism of integration of ϕC31 into streptomycete chromosomes has been characterized [51, 52, 59, 69, 70, 85, 86, 96] and reviewed [18, 33, 89, 90], and the ϕC31 genome sequence is known [88]. The minimal attP and attB sites comprise 39 and 34 bp, respectively [32], and they share a 3-bp common sequence at the site of conservative crossing-over [51].

Fig. 1
figure 1

Genetic map locations of the attB sites for bacteriophages R4, ϕBT1, ϕC31, and TG1 in S. avermitilis (Sav) and S. coelicolor (Sco) relative to oriC. All four sites reside in the 6.5-Mb core regions that contain genes highly conserved across Streptomyces sp. [13, 21, 42, 49]

Full size image

BLASTP analysis with ϕC31 Int was carried out in October 2011, and five full-length hits were obtained (Table 2). Two of the hits were to actinophages TG1 and ϕBT1 Int proteins, and three were to proteins encoded by Kitasatospora setae, Streptomyces violaceusniger, and Streptomyces zinciresistens. The Int homolog encoded by K. setae shows highest sequence similarity to ϕC31 Int (53.2%), and its gene maps to a region central to the linear chromosome (about 4.59 Mb into the 8.78-Mb genome) located just downstream of a truncated pirin-like gene. Just downstream of the int homolog is a gene that encodes a potential Xis function that shows 37.7% amino acid identity to the gp3 protein encoded by ϕC31 ([46]; see below). Further downstream of these genes is a large portion of the pirin-like gene missing in the truncated gene upstream of int that may have been generated by an integration event. Adjacent to the downstream truncated pirin-like gene is a complete pirin homolog which may contain the target for pSET152 integration [20]. BLASTP analysis with ϕC31 gp11, a DNA polymerase that has homologs encoded by the streptomycete phages ϕBT1 [30] and phiSASD1 [101], and by many mycobacterial phages, revealed no homolog in K. setae. BLASTP analysis of S. violaceusniger and S. zinciresistens also revealed no homologs to ϕC31 gp3 or gp11. In summary, there is no evidence for complete prophage insertions in the vicinity of the int homologs in K. setae, S. violaceusniger, or S. zinciresistens.

Table 2 Homology relationships between streptomycete bacteriophage large serine recombinases or integrases (Int) determined by BLASTP
Full size table

Recent studies have characterized protein gp3 encoded by ϕC31 as the RDF or Xis protein required for excision of integrated ϕC31. Protein gp3 binds directly to Int in 1:1 stoichiometry and changes the recombinational specificity from attP and attB to attL and attR [46]. The gp3–Int complex also catalyzes recombination between two attL or two attR sites. These findings should further extend the utility of the ϕC31 integration (and now excision) system for genetic engineering applications [91].

A number of cloning vectors employing ϕC31 have been developed [47], and those employing only the att/int functions coupled with oriT from RP4 for conjugation from E. coli were developed by Bierman et al. [14]. Notably, pSET152, which lacks replication functions for streptomycetes, has gained wide acceptance as an insertion vector to generate stable recombinants. More recently, bacterial artificial chromosome (BAC) vectors containing ϕC31 att/int and oriT functions have been used to stably insert large secondary metabolite gene clusters into the chromosomes of heterologous hosts [2, 8, 9, 71, 82]. pSET152 and other ϕC31-based conjugal insertion vectors have utility in many streptomycetes and other actinomycetes. The frequencies of transconjugant formation range from 1.6 × 10−4 to 1.4 × 10−2 in many Streptomyces species (Table 3). In some cases where the recipient host restricts modified DNA, conjugation requires the use of an E. coli host defective in Dam/Dcm methylation. The generally high transconjugant frequencies in streptomycetes can be attributed to three factors: (1) conjugation bypasses type II restriction enzyme barriers [8, 14, 67]; (2) the ϕC31 attB site is located in a gene encoding a pirin-like protein that is widely distributed within Streptomyces sp. [23] (Tables 4, 5); and (3) ϕC31 integration is generally very efficient. In addition to the primary attB site, Streptomyces sp. can have pseudo-attB sites for ϕC31 integration. The frequency of transconjugant formation in S. coelicolor dropped from 1.5 × 10−3 to 5 × 10−6 when the primary attB site was deleted (Table 3), and the insertions mapped to three pseudo-attB sites that showed some sequence homology to authentic attB sites [23].

Table 3 Actinomycete host range of ϕC31-based vectors
Full size table
Table 4 Homologs in streptomycetes and other actinomycetes to genes containing bacteriophage attB sites and glnA from S. roseosporus
Full size table
Table 5 Homologs in streptomycetes and other actinomycetes to proteins encoded by genes containing bacteriophage attB sites and GlnA from S. roseosporus
Full size table

Some other actinomycetes are recipients for transconjugation, protoplast transformation, or electroporation with pSET152 or other ϕC31-based integration vectors (Table 3). In some cases transconjugant frequencies in non-streptomycetes were high (e.g., in Actinoplanes teichomyceticus and Nonomuraea sp. 40027), but in other cases they were very low. For instance, in Saccharopolyspora spinosa, which lacks a pirin-like gene, transconjugants were obtained at a frequency of 10−7, and integrations occurred in two pseudo-attB sites [67]. In Saccharopolyspora erythraea, which also lacks a pirin-like gene [81] and is normally a poor recipient for conjugation, insertion of a portable streptomycete attB site converted it into a high frequency recipient for the integration of transgenes [84]. In Mycobacterium smegmatis, Mycobacterium bovis, and Mycobacterium tuberculosis, low frequencies of recombinants were obtained by electroporation with pIJ8600 [77]. Mycobacterium smegmatis MC2-155 has a pirin-like gene, but the recombinant analyzed by Murry et al. [77] localized the insertion in a pseudo-attB site. Mycobacterium tuberculosis and M. bovis do not have pirin-like genes, and insertions were in pseudo-attB sites.

BLASTP and BLASTN surveys of ten streptomycetes identified pirin-like genes in each case. In typical streptomycete orthologs, the ratio of the number of mutations causing non-synonymous amino acid substitutions (dN) to the number causing synonymous amino acid substitutions (dS) is about 0.4–0.9 ([10]; this report). The dN/dS ratios for paralogs tend to be about 1.0 or higher. Because of the high G+C content of streptomycete genes, dN/dS ratios for orthologs translate into a situation where the percent change in amino acid identities diverges at nearly the same rate as the percent change in nucleotide identities. Inspection of the amino acid and nucleotide identities for the pirin-like homologues in the ten streptomycetes indicates that both are drifting at about the same rates; the dN/dS ratio calculated for the average of all ten was 0.4. By comparison, the dN/dS ratio calculated for the average of ten glnA genes (Tables 4, 5) was also 0.4. Thus it appears that the pirin-like genes are orthologs. Inspection of several pirin-like gene sequences indicated that the actual 45-nucleotide attB sequence is present in all cases, and that it is generally even more conserved than the overall gene sequence. For instance, the S. avermitilis attB shared 93.3% nucleotide identities with the S. roseosporus attB, whereas the complete genes showed 84.4% nucleotide identities (Table 4). Likewise, the S. griseus attB showed 100% nucleotide identities to S. roseosporus attB and their genes shared 95.7% identities. It is clear from these data that the presence of ϕC31 attB sites can be surveyed efficiently by initially doing BLASTP analysis, followed by confirmation at the gene and attB site level using BLASTN.

A BLASTP survey of ten non-streptomycete actinomycetes genomes identified ϕC31 attB potential targets in six strains. In all cases, pirin-like genes were present, and attB sites were confirmed in the two strains examined in detail. In Frankia sp. EAN1pec, the attB site showed 84.4% nucleotide identities with the S. roseosporus attB, and the M. smegmatis attB showed 73.3% identities. The average dN/dS for the six pirin homologs was 0.6, suggesting that most or all are orthologs to the S. roseosporus pirin-like gene. For comparison, the glnA genes from the non-streptomycetes have diverged from the glnA gene of S. roseosporus at an average dN/dS ratio of 0.9.

Notably, a pirin-like gene was absent from S. erythraea, and the closest homolog encoded a protein with only 31.7% amino acid identity to the pirin-like protein of S. roseosporus (Table 5). The combined genetic and bioinformatic data indicate that ϕC31-based vectors are widely applicable for streptomycetes, and suggest that they may be useful in certain other actinomycetes. The potential utility can be determined a priori by genome sequencing to determine if a pirin-like gene is present. If a ϕC31 attB site is not present, then a portable attB site might be inserted to increase the efficiency of genetic manipulations, as demonstrated in S. erythraea [84] and Micromonospora griseorubida [98]. This concept has already been generalized to engineer eucaryotes (e.g., see [18, 56, 60, 74, 103]) and archaea [16, 34], and should be applicable to any organism that is amenable to genetic manipulation.

Bacteriophage ϕBT1

ϕBT1 is a temperate phage related to ϕC31 [30]. Like ϕC31, it integrates via a large serine recombinase, and its 73-nucleotide attP and attB sites are quite different from each other. However, they have core 12-nucleotide sequences nearly identical to each other (11 of 12 identities) where crossing-over occurs. Importantly, ϕBT1 integrates into a gene annotated to encode an integral membrane protein unrelated to the pirin-like gene for ϕC31 integration (Table 1). The ϕBT1 Int is distantly related to ϕC31 Int, showing only 26% amino acid identities in reciprocol BLASTP analyses (Table 2). It has similar low sequence identities to the three other proteins that gave significant hits to ϕC31 Int, and had no other significant hits (Table 2). In S. coelicolor and S. avermitilis, the ϕBT1 attB site is located about 1 Mb to the left of oriC, and within the 6.5 Mb core region (Fig. 1). The mechanism of insertion was studied in vitro where it was shown that the minimal attB and attP sites comprise 36 and 48 bp, respectively [106]. The integration process was very efficient with attB and attP substrates, but was also measurable with attL and attR sequences, implying that Int might excise ϕBT1 in vivo at some frequency in the absence of other factors. Further mechanistic studies have been reported recently [107]. Although no studies have been carried out to characterize an Xis or RDF protein, BLASTP analysis with the 244-aa gp3 RDF from ϕC31 gave a top hit of 84.8% amino acid identities to a 247-aa gp3 protein from ϕBT1 (this report). It is likely that this protein serves an Xis or RDF function for ϕBT1. If so, it could extend the potential utility of the ϕBT1 integration system.

Gregory et al. [30] constructed vectors derived from pSET152 by replacing the ϕC31 att/int with ϕBT1 att/int, and by exchanging antibiotic resistance genes. They showed that the ϕBT1-based vectors conjugated from E. coli into S. coelicolor and integrated at frequencies comparable to those of pSET152 (3.5 × 10−3 per recipient). Importantly, they demonstrated that an S. lividans transconjugant containing a ϕBT1-based vector inserted in the chromosome was an efficient recipient for conjugal transfer of pSET152. Since the ϕC31 and ϕBT1 systems are compatible, they can be used to add genes sequentially to genetically engineer S. lividans, and other streptomycetes. Gregory et al. [30] investigated conjugation into other streptomycetes, and recovered transconjugants from S. avermitilis, S. cinnamonensis, S. fradiae, S. lincolnensis, S. nogalater, S. roseosporus, and S. venezuelae. ϕBT1-based vectors also function efficiently in the rapamycin-producing Streptomyces hygroscopicus [31, 50], where it was shown that insertions are neutral under the prevailing fermentation conditions (i.e., they cause no reduction in rapamycin production). This property is important for the genetic engineering of industrial production strains.

The ϕBT1 Att/Int system can also be used in conjugal BAC vectors for site-specific insertion in Streptomyces chromosomes. Liu et al. [58] developed a ϕBT1-based BAC and used it to clone and express the meridomycin biosynthetic gene cluster in S. lividans. Alexander et al. [2] modified a BAC vector to accommodate the engineering of lipopeptide biosynthetic genes in E. coli followed by conjugal transfer and insertion into the ϕBT1 attB site in S. fradiae strains. This system was coupled with the use of ϕC31-based vectors to set up an ectopic transcomplementation system that allows lipopeptide biosynthetic genes to be expressed from three different locations in the chromosome to facilitate combinatorial biosynthesis [2, 3, 12, 79]. They also demonstrated that insertions into ϕBT1 and ϕC31 attB sites are neutral with respect to antibiotic production in S. fradiae, and that the complete set of A54145 biosynthetic genes can be expressed more efficiently from either attB site than from the native locus which is located in a potentially unstable subteleomeric region containing IS and transposase sequences [11, 72]. This approach might be applied to other streptomycetes where antibiotic biosynthetic genes are located in unstable subteleomeric regions of linear chromosomes. The compatibility of the two integration systems presents possibilities for doubling and tripling of complete secondary metabolite gene clusters for heterologous expression and strain improvement in streptomycetes [8, 9].

The apparent broad host specificity of the ϕBT1 Att/Int system is supported by recent genome sequencing studies. The integral membrane protein gene containing the attB site for S. roseosporus has apparent orthologs in all ten Streptomyces surveyed by BLASTN and BLASTP analyses (Tables 4, 5). Although the average dN/dS ratio for all ten was 0.8, BLASTN analysis of the S. griseus and S. ghanaensis genomes using the original 73-nucleotide attB site described by Gregory et al. [30] picked up full-length sequences with 92 and 96% identities to attB in the target genes. The attB sites are located in a highly conserved region in the first one-third of the gene.

Of the ten non-streptomycete actinomycetes surveyed, only four have homologous genes encoding integral membrane proteins. These include Amycolatopsis mediterranei and S. erythraea, both of which lack ϕC31 attB sites. The average dN/dS ratio for the four genes is 1.0, suggesting that one or more may have been under selection to evolve a paralogous function. A closer inspection of the attB regions in these four genes indicated that the A. mediterranei and S. viridis attB sequences, which have only 1 mismatch in the 12-nucleotide crossover region, are more highly conserved than those of S. erythraea and C. acidiphila, which have four and five mismatches in the crossover region, respectively. The lower amino acid sequence homologies relative to the S. roseosporus target gene product observed with the last two strains also suggests that the corresponding genes are paralogs to the streptomycete genes, and that purifying selection [10] to maintain the usually highly conserved attB region is relaxed in both cases.

The biological data and bioinformatic analyses indicate that ϕBT1-based vectors should have broad applicability for engineering of streptomycetes; bioinformatic data also suggest limited potential utility in other actinomycetes. As demonstrated with the ϕC31 integration system, a portable ϕBT1 attB site could be inserted into non-streptomycete chromosomes for genetic engineering purposes. A portable ϕBT1 att site has been used to demonstrate that ϕBT1 Int functions efficiently in vertebrate cells and Schizosaccharomyces pombe. Morover, this system has been used in conjunction with Cre to build a transgenic human–Chinese hamster hybrid cell line containing 400 kb of contiguous transgenic DNA [105].

Bacteriophage R4

R4 is a broad-host-range streptomycete temperate bacteriophage isolated from soil on Streptomyces albus J1074, a mutant of S. albus G defective in SalI restriction and modification [19]. Like many other Streptomyces bacteriophages, its host range is limited primarily by type II restriction enzyme barriers [19, 24, 36, 37]. R4 integrates site-specifically into the chromosome of Streptomyces parvulus (and presumably in other streptomycetes) to establish lysogeny [87]. Matsuura et al. [68] demonstrated that integration is catalyzed by a large serine recombinase that recognizes attP and attB sites for integration, but not attL and attR sites for excision. The 50-nucleotide attB site contains a 12-nucleotide common core that is also found in the attP site, and serves as the region for site-specific recombination [80]. The 50-nucleotide attB site was used to carry out BLASTN analysis in S. roseosporus, and a highly conserved 41-nucleotide segment containing the 12-nucleotide common core was located in a gene that encodes an acyl-CoA synthetase (Table 1). This gene and its product were used to carry out BLASTN and BLASTP analyses against ten Streptomyces and ten other actinomycete genomic sequences: highly conserved apparent orthologs (average dN/dS = 0.6) were observed in all ten Streptomyces species (Tables 4, 5). The acyl-CoA synthetase apparent orthologs containing attB sites in S. coelicolor and S. avermitilis mapped to similar locations within the 6.5-Mb core regions of the linear chromosomes (Fig. 1). Homologs were also observed in nine of ten other actinomycetes, but these appear to be a mixture of orthologs and paralogs. The average dN/dS ratio for seven of the gene/protein pairs (excluding Frankia sp. EAN1pec and M. aurantiaca) was 0.9. The 50-nucleotide R4 attB sites were compared for four of the strains. C. acidiphila, S. erythraea, and S. viridis have authentic attB sites showing 94, 96, and 92% nucleotide identities, respectively, to the S. roseosporus attB site. Furthermore, the same strains had 12, 12, and 11 nucleotide identities to the 12-nucleotide crossing-over region. On the other hand, Frankia sp. EAN1pec, which encodes an acyl-CoA synthetase homolog that shows only 40.6 amino acid identities to the S. roseosporus counterpart, has a 50-nucleotide attB site that is only 54% identical to the attB of S. roseosporus, and it has only 6 of the conserved 12 nucleotides for crossing-over. This gene appears to be a paralog to the streptomycete R4 target genes, and probably would not serve as an efficient target for R4 integration.

Although R4 has not been used widely as a general tool for insertion of genes in actinomycetes, it has been shown to be a useful tool for engineering human cells (e.g., see [56, 60, 80]).

Bacteriophage TG1

TG1 is a temperate bacteriophage isolated on Streptomyces cattleya, the thienamycin producer [26]. It has a broad host range for Streptomyces species, but did not form plaques on S. coelicolor or S. lividans [26, 27]. Analysis of multiple lysogens indicated that it inserted into a single attB site in S. cattleya [26]. TG1 was developed as a bifunctional vector that could be engineered in E. coli, transfected into a streptomycete host, then transduced into other streptomycete hosts where it formed relatively stable lysogens [27].

TG1 was recently shown to integrate site-specifically by a large serine recombinase mechanism [75, 76]. BLASTP analysis with TG1 Int gave significant hits only to ϕC31 Int (49.7%) and to the four other proteins identified in BLASTP analyses with ϕC31 and ϕBT1 integrases (Table 2). In vitro studies demonstrated that the TG1 Int does not require host factors for insertion, and that it does not catalyze excision [76]. The minimal attP and attB sites were shown to comprise 43 and 39 nucleotides, respectively, and share a common dinucleotide (TT) at the site for crossing-over [76]. Recent studies have demonstrated that TG1 Int can drive efficient integration of attB-containing circular plasmid DNA into E. coli containing an attP sequence inserted into the chromosome by EZ-Tn5 transposition [38], a technique that might be applicable to other bacteria and other serine integrases.

The TG1 attB site is located in a dapC-like gene which may encode an N-succinylaminopimelate aminotransferase [75]. However, TG1 lysogens of S. avermitilis did not require lysine or diaminopimelate for growth, suggesting that the dapC annotation may be incorrect, and that the gene may encode an aminotransferase with a different function. The TG1 attB site in the dapC-like gene is located about 230 kb to the right of oriC in both S. coelicolor and S. avermitilis, or about 140 kb to the right of the ϕC31 attB site (Fig. 1). Apparent orthologs of the dapC-like gene were observed in all ten streptomycetes (average dN/dS = 0.70) (Tables 4, 5). Homologs of the dapC-like gene were observed in all ten other actinomycetes surveyed, but some of these are likely to be paralogs. For instance, the dN/dS ratios for Frankia sp. EAN1pec and M. smegmatis are 1.4 and 1.2, respectively. The bioinformatic data suggest that the TG1 integration system may be directly applicable to many streptomycetes and possibly to some other actinomycetes.

Uses of site-specific insertion for genetic engineering in actinomycetes

Streptomycete phage site-specific integration systems have been used for a number of applications that require stable insertion of one or more genes into the chromosome. Industrial applications include strain improvement for early to late-stage process development, heterologous expression of cryptic secondary metabolite biosynthetic gene clusters for drug discovery, and combinatorial biosynthesis to generate novel derivatives of known secondary metabolites [6–9, 12]. For strain improvement, site-specific insertion can be used to: (1) increase gene dosage to address rate-limiting primary or secondary metabolic steps; (2) change promoters to improve the expression of regulatory and other genes; (3) alter the metabolic capability of cells by adding new functions; (4) and duplicate or triplicate complete secondary metabolite gene clusters [9]. For the discovery of novel drug candidates from cryptic secondary metabolite gene clusters discovered in genome sequencing projects, candidate gene clusters can be: (1) cloned in BAC vectors that replicate in E. coli; (2) transferred by conjugation from E. coli into streptomycete expression hosts, including those derived from industrial production strains; and (3) stably inserted at appropriate attB sites. Transconjugants can then be fermented in several media and screened for the expression of novel secondary metabolites. This process and the properties of key streptomycete expression hosts are discussed in more detail elsewhere [8, 9]. The use of site-specific integration vectors for combinatorial biosynthesis has the advantage that different genes or sets of genes can be engineered separately, and then different combinations of the engineered genes can be brought together in an expression host [7, 8, 12]. Recent examples that demonstrate the power of this approach are the engineering and expression of separate nonribosomal peptide synthetase (NRPS) multi-enzymes, or other genes encoding amino acid modifying enzymes, by insertion into the S. fradiae chromosome at the ϕC31 and ϕBT1 attB sites to generate a large array of novel lipopeptide antibiotics with tridecapeptide structures derived from A54145 and daptomycin [2, 3, 79]. These site-specific integration systems can also be used in combination with other insertion systems, such as IS117 [22, 73, 78].

Discussion

The bacteriophage ϕC31 Att/Int system has made a large impact on the development of robust genetic engineering tools for the industrially important Streptomyces and other actinomycetes. This work was initiated in Russia by the Lomoskaya laboratory, and further developed in Russia and in the UK by the Keith Chater and Margaret Smith laboratories. The work on the fundamental biology of ϕC31 provided a rich starting point for the seminal work of Kustoss, Rao, and colleagues at Eli Lilly and Company, who developed the Att/Int system into a widely useful set of cloning vectors for Streptomyces species [14, 51, 52]. These and their derivatives have been applied to strain improvement, combinatorial biosynthesis, and whole pathway heterologous expression. In addition to the important applications in the native actinomycetes, the unidirectional serine recombinase systems have impacted the broader field of biotechnology, providing a robust methodology for the engineering of eucaryotic cells.

ϕC31 is just one of several streptomycete temperate phages described in the literature, most of which are poorly characterized. R4, TG1, and ϕBT1 have been studied in some detail; as with ϕC31, all three have broad host ranges within Streptomyces species, and integrate by unidirectional serine recombinases. Importantly, they integrate into different genes that are highly conserved in Streptomyces species. This represents an interesting evolutionary strategy to have the potential to lysogenize any species of the genus Streptomyces, rather than limit the host range to one or a subset of streptomycetes by inserting into genes not conserved across the species. The host range is thus maximized, and limited primarily by host type II restriction barriers.

For applications in streptomycetes, the broad host specificity and conservation of genes containing attB sites for these integration systems enables sequential addition of genes for combinatorial biosynthesis, strain development, and other applications. The host restriction barriers are often easily overcome by using conjugal transfer from E. coli [14, 67], a process that transfers linear concatemers of single-strand DNA which are not susceptible to host-encoded type II restriction endonuclease cleavage [8]. The different attB genes or sequences can also be used as portable integration sites in other actinomycetes that lack attB sites. It is conceivable that four or more different attB sites could be cloned contiguously, then inserted into an actinomycete of interest at a site that is neutral for secondary metabolite production. This would provide a target for sequential addition of any number of genes for a variety purposes. This approach has already been applied to mammalian cells [56, 60, 74, 103]. In principle, this concept could be applied to other eubacteria, archaea, plants, mammals, and other eucaryotes. The applications are limited only by our current knowledge of bacteriophages that employ large serine integrases. There are undoubtedly many more temperate bacteriophages for streptomycetes and other actinomycetes that use this mechanism. Broad-host-range temperate bacteriophages are readily isolated on Streptomyces strains, and Streptomyces griseofuscus is particularly useful for bacteriophage isolations because it is non-restricting for bacteriophage plaque formation [24], and has been used to isolate temperate bacteriophages from soil [36, 37].

Although several large serine recombinase systems have already been discovered, it is not known if the best ones have been identified. There exists an untapped wealth of additional temperate actinophages yet to be discovered, and these can be isolated inexpensively from soil.