Use of rpsL for dominance selection and gene replacement in Streptomyces roseosporus

We developed a gene replacement system using the rpsL gene of Streptomyces roseosporus and demonstrated its utility by constructing a deletion in the S. roseosporus glnA gene. A 1.3-kb BamHI fragment that hybridized to the Mycobacterium smegmatis rpsL gene was subcloned from an S. roseosporus cosmid library and sequenced. Plasmid pRHB514 containing the rpsL gene conferred streptomycin sensitivity (Sm(S)) to the Sm(r) S. roseosporus TH149. The temperature-sensitive plasmid pRHB543 containing rpsL and the S. roseosporus glnA gene disrupted with a hygromycin resistance (Hm(r)) gene was introduced into S. roseosporus TH149, and recombinants containing single and double crossovers were obtained after a temperature increase. Southern hybridization analysis revealed that single crossovers occurred in the glnA or rpsL genes and that double crossovers resulted in replacement of the chromosomal glnA gene with the disrupted glnA. Glutamine synthetase activity was undetectable in the recombinant containing the disrupted glnA gene.

Streptomycetes are commercially important microorganisms that produce secondary metabolites with diverse biological activities. Much is known about the organization and regulation of many secondary metabolic pathway genes, and these studies have been aided by gene disruption and gene replacement techniques (3). Gene replacement methods have also been used to insert genes into the chromosome to improve product yields and to produce novel products (3). Most of the procedures used for gene replacement or gene addition select for the first crossover integrating a plasmid, then screen for a second crossover that exchanges or inserts a gene and eliminates the remaining plasmid sequences. To develop a more versatile gene replacement system, we have explored use of the rpsL gene in Streptomyces roseosporus.
The rpsL gene, which encodes ribosomal protein S12, has been used as a counterselectable marker since it confers dominant streptomycin sensitivity (Sm s ) in an Sm r background (17). Gene replacement using rpsL was first demonstrated in Escherichia coli (24) and more recently in Mycobacterium smegmatis and Yersinia pestis (27,31). E. coli rpsL has also been used as a counterselectable marker for gene replacement by heterologous expression in Vibrio cholerae (30). A potential advantage of the rpsL system in streptomycetes is that Sm r strains are not defective in primary metabolic functions and are readily isolated in most strains of interest, thus facilitating the use of the rpsL system in many actinomycetes. We report here the cloning and use of the S. roseosporus gene for counterselection in S. roseosporus to select double crossovers that replace the chromosomal glnA gene with a disrupted glnA gene.

MATERIALS AND METHODS
Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Table 1.
Media and growth conditions. S. roseosporus strains were grown in CSM media (13) at 29 or 39ЊC, fragmented into individual CFU by ultrasonic vibration (2), and grown on B agar (32). E. coli strains were grown in TY media (26) or Circlegrow media (Bio 101). Antibiotics were added to appropriate media for streptomycete strains at 30 g/ml for apramycin (AM), 200 g/ml for hygromycin (HM), 50 g/ml for nalidixic acid, and 50 g/ml for streptomycin (SM). For E. coli strains, antibiotics were added at 100 g/ml for AM and HM. Auxotrophy was determined as described elsewhere (12) on CDA agar (32).
DNA techniques and plasmid constructions. Standard methods were used for plasmid isolation, restriction enzyme digestion, random priming, and Southern hybridization analysis (12,26). Restriction endonucleases and other enzymes were used according to the manufacturer's recommendations. DNA fragments and vectors used for subcloning and radiolabeling were isolated from 0.8% SeaKem agarose (FMC) gels following electrophoresis using GenecleanII (Bio 101). DNA was concentrated and desalted by using Microcon 30 microconcentrators (Amicon) prior to ligation or electrotransformation. S. roseosporus genomic DNA was isolated as described previously (12). DNA sequence was determined by using a Taq Dye Deoxy Terminator Cycle sequencing kit and a model 373A DNA sequencing system (Applied Biosystems). PCR amplification was performed with a Perkin-Elmer GeneAmp 9600 using standard PCR conditions and a Touchdown PCR amplification program (9). Oligonucleotides were synthesized by Genosys (Woodlands, Tex.). Primers for PCR amplification were designed with the Oligo Primer Analysis software (National Biosciences, Inc.). Plasmid map constructions were performed with the Gene Constructor Kit version 1.2 (Textco, Inc., West Lebanon, N.H.).
Plasmids were constructed as follows. A 1.3-kb BamHI fragment from cosmid pRHB545 that hybridized to the M. smegmatis rpsL gene was ligated to pBluescript II KS Ϫ digested with BamHI to yield pRHB515. The 1.3-kb BamHI fragment from pRHB515 containing the S. roseosporus rpsL gene was ligated to pKC1139 digested with BamHI to yield pRHB514. A 1.46-kb fragment containing the S. roseosporus glnA gene was PCR amplified from chromosomal DNA with primers PR170 (5Ј GCTGAGATGCCGCCCACCAC 3Ј) and PR171 (5Ј ACGCCGCCCGCCTGGAGGTA 3Ј) and ligated to pCR II (Invitrogen) to yield pRHB534. An Hm r gene from pCZA256 was inserted into a SalI site within the glnA gene on pRHB534 as follows. A 1.7-kb BamHI-to-HindIII fragment from pCZA256 was treated with mung bean nuclease and ligated to pRHB534 previously digested with SalI and mung bean nuclease to yield pRHB537. A 687-bp fragment containing rpsL was PCR amplified to introduce BglII sites at the ends with primers PR115 (5Ј GGGGAGATCTGCGTGGACAGCATTT GTT 3Ј) and PR116 (5Ј GGGGAGATCTGCCCTTACGAGGCATTCT 3Ј) and ligated with pCR II to yield pRHB541. A 0.69-kb BglII fragment from pRHB541 containing rpsL was ligated with BglII-digested pKC1139 to yield pRHB538.
A cosmid library of S. roseosporus DNA was constructed by partially digesting genomic DNA with Sau3A1 and alkaline phosphatase (Boehringer Mannheim Biochemicals). DNA of approximately 40 kb was isolated and ligated to BamHIdigested pKC1471 and packaged with a Gigapack packaging extract (Stratagene). Packaged DNA was introduced into E. coli XL1-Blue-MFRЈ, and individual clones containing cosmid DNA were stored as an ordered array in 96-well microtiter plates. Primary screening filters were prepared with a 96-well dot blot apparatus. Twelve cultures from a row of microtiter wells were pooled, and plasmid DNA was prepared, bound to nylon filters (Bio-Rad Zeta-probe GT), and probed with a 255-bp fragment of M. smegmatis rpsL DNA amplified from chromosomal DNA with primers PR99 (5Ј CCGCAAGGGCCGCCGCGACA AAGATCGCC 3Ј) and PR100 (5Ј CTTGTAACGGACACCGGGCAGGTCC TTC 3Ј) (28). Secondary screening filters were prepared in a similar manner except that DNAs from individual cosmid clones were blotted to nylon filters.
Transformation, electroporation, and conjugation. Plasmids were introduced into E. coli XL1-Blue MFRЈ by electroporation using a Bio-Rad Genepulser electroporator. Electrocompetent E. coli cells were prepared as described previously (34). E. coli S17-1 was transformed as described elsewhere (26). Plasmids pRHB514, pRHB538, and pRHB543 were introduced into S. roseosporus strains by conjugation from E. coli S17-1. S. roseosporus was grown from a frozen (Ϫ70ЊC) stock in CSM medium for 16 h, and E. coli was grown from a 1% inoculum in TY broth for 3 to 5 h at 37ЊC. S. roseosporus mycelia were homogenized and mated in 1:9, 1:1, and 9:1 ratios (100-l total volume) with E. coli S17-1 containing the appropriate plasmid and were spread on B agar plates with a glass spreader. Plates were incubated for 15 h at 29ЊC and then overlaid with nutrient soft agar containing nalidixic acid and AM to give final bottom agar concentrations of 50 and 30 g/ml, respectively. Transconjugants appearing in 4 to 6 days were picked, homogenized, and grown in CSM media containing appropriate antibiotics.
Selection for single and double crossovers. Recombinants containing plasmid pRHB543 inserted into the chromosome were obtained as follows. Strain TH158 containing pRHB543 was grown to stationary phase at 29ЊC in CSM medium containing AM. The culture was sonicated, diluted, plated on B agar containing AM, and incubated at 29 and 39ЊC for 4 to 6 days. Colonies appearing at 39ЊC were analyzed by Southern hybridization to determine plasmid integration patterns. Single-crossover frequencies were determined as the ratio of the CFU at 39ЊC to the CFU at 29ЊC.
Double crossovers between pRHB543 and the chromosome of strain TH158 were obtained as follows. TH158 was grown at 29ЊC in CSM medium containing HM for 2 days. Cultures were shifted to 39ЊC and grown in CSM plus HM for 2 to 3 days, sonicated, diluted, and plated on B agar containing HM and SM. Colonies appearing at 39ЊC were analyzed by Southern hybridization to determine the location of crossover events and to confirm the loss of plasmid sequences. Cells were also plated on B agar plus AM and HM to determine the frequency of plasmid integrations by single crossovers.
GS assay. S. roseosporus strains were grown in CSM liquid media to stationary phase, permeabilized by treatment with cetyltrimethylammonium bromide as described elsewhere (22), and glutamine synthetase (GS) activity was determined by the ␥-glutamyltransferase assay (4). One unit equals 1 mmol of ␥-glutamylhydroxamate formed/min/mg of protein at 37ЊC.
Nucleotide sequence and data analysis. Derived amino acid sequences were analyzed with the Genetics Computer Group software package (version 8.0) (8). Amino acid sequence homology searches were performed with the BLAST server at the National Center for Biotechnology Information (Bethesda, Md.) and nonredundant protein sequence databases (1). The nucleotide sequence for rpsL, rpsG, and part of fusA has been assigned GenBank accession number U60191.

Cloning and characterization of the S. roseosporus rpsL gene.
A library of S. roseosporus chromosomal DNA prepared in cosmid pKC1471 was probed with the M. smegmatis rpsL gene, and cosmids pRHB545 and pRHB546 containing rpsL-hybridizing sequences were identified. The cosmids were digested with BamHI, EcoRI, and KpnI, and DNA fragments were separated by gel electrophoresis, blotted to nylon filters, and analyzed by Southern hybridization using a radiolabeled M. smegmatis rpsL probe. Similar patterns of hybridization were observed with the two cosmids. A 1.3-kb BamHI fragment that hybridized to the M. smegmatis rpsL probe was subcloned into pBluescript II KS Ϫ and sequenced. The DNA and predicted amino acid sequences are shown in Fig. 1. Three open reading frames were identified by codon preference analysis of the region, each having the predicted percent GϩC (ϳ70:50:90) at positions 1, 2, and 3 of codons as generally observed in streptomycete coding regions (5). These open reading frames en- code proteins highly homologous to the gene products of rpsL, rpsG, and fusA in other bacteria. The linkage relationship of these three genes is the same as that in other bacteria (14,21,23). A possible promoter sequence was located at nucleotide positions 97 to 126 upstream of the rpsL translational start site. The sequence TTGACC-16n-TACGCT is similar to the streptomycete E. coli-like promoters (SEP promoters) which have a consensus sequence of TTGACPu-16 to 18n-TAgPuPuT (35) and to a promoter upstream of the M. smegmatis rpsL gene by about the same distance (ϳ190 bp) (14). A comparison of the deduced S. roseosporus S12 protein amino acid sequence to those of other S12 proteins is shown in Fig. 2. S. roseosporus S12 shows the closest similarity to M. smegmatis and Mycobacterium tuberculosis S12 (ϳ90% amino acid identity) and to Micrococcus luteus (85% identity). The S. roseosporus S12 pro-tein lacks a terminal amino acid found in many of the other bacteria and also lacks a 13-amino-acid region observed in the Streptococcus pneumoniae, Bacillus stearothermophilus, and Staphylococcus aureus S12 proteins. The predicted initiation codon (GTG) for rpsL is the same as that in M. luteus and is preceded 5 bp upstream by a potential ribosomal binding site (GGAG). The S. roseosporus S12 and S7 proteins appear to be translationally coupled with 2 nucleotides separating the S12 stop codon and the predicted S7 start codon. A potential ribosomal binding site (GGAG) is located 8 nucleotides upstream of the S7 initiation codon and within the S12 coding region. The S12 and S7 proteins of M. smegmatis, M. tuberculosis, Mycobacterium leprae, and Mycobacterium intracellulare also appear to be translationally coupled and contain identical potential ribosomal binding sites (GGAG) 8 nucleotides up- stream of the S7 initiation codon and located within the S12 coding region (14,20). The Mycobacterium species appear to contain overlapping stop and start codon sequences (TAATG or TGATG) for the S12 and S7 proteins, a configuration which was not observed in S. roseosporus since its S12 protein is 1 amino acid shorter. Potential translational coupling of S12 and S7 was not observed in the Haemophilus influenzae, E. coli, S. pneumoniae, B. stearothermophilus, and Staphylococcus aureus operons (data not shown).
The rpsL gene expresses dominant streptomycin sensitivity. Gene replacements using wild-type rpsL (Sm s ) for counterselection are performed in an Sm r background (24,27,31). Sm r maps to the rpsL gene in many different microorganisms (18). We selected spontaneous Sm r S. roseosporus mutants, and strain TH149 was further examined as a host for double-crossover analysis. To determine if the S. roseosporus rpsL gene would confer Sm s , we introduced plasmid pRHB514 into TH149 by conjugation from E. coli S17-1 and transconjugants were isolated and streaked onto B agar containing SM. TH149 containing pRHB514 was Sm s , but a small number of Sm r colonies grew in the patch, presumably due to homogenotization. The results indicated that the S. roseosporus rpsL gene expresses a dominant Sm s phenotype in an Sm r background, suggesting that the apparent promoter sequence upstream of rpsL on plasmid pRHB514 is functional (Fig. 1).
Selection for single and double crossovers using rpsL. Recombinants containing single or double crossovers between plasmids pRHB543 and the S. roseosporus chromosome were selected, and the frequency of single crossovers after a temperature shift to 39ЊC was about 10 Ϫ2 . To select for recombinants containing double crossovers, a temperature shift from 29 to 39ЊC followed by 2 to 3 days of growth in liquid culture was necessary to cure the plasmid containing the rpsL gene prior to SM selection. Therefore, a direct measurement of the double-crossover frequency was not possible. However, after 2 to 3 days of outgrowth following the temperature increase, the frequency of recombinants containing single crossovers was about 100-fold greater than the frequency of recombinants containing double crossovers. If the ratio of recombinants containing single and double crossovers remains relatively constant during growth, then the initial double-crossover frequency prior to plasmid curing was approximately 10 Ϫ4 .
To investigate the nature of the DNA insertions in recombinants containing single or double crossovers, chromosomal DNA was digested with BamHI and EcoRI and DNA fragments were separated by gel electrophoresis, blotted onto nylon filters, and analyzed by Southern hybridization using pRHB543 or an Hm r gene fragment as a probe. Figure 3A shows the expected DNA structures of recombinants containing single crossovers in glnA or rpsL sequences, and Fig. 3B shows the expected structure of a recombinant containing two crossovers to insert the Hm r -disrupted glnA gene in the chromosome. Figures 3C and D show the Southern hybridization patterns of several recombinants. Strains TH172 and TH174, which contained single crossovers, had identical junction fragments of 3 and 20 kb, two pRHB543 internal Hm r -hybridizing bands (1.8 and 1.4 kb), a 9.0-kb glnA-hybridizing fragment, but not the 1.3-kb rpsL-hybridizing fragment. This hybridization pattern is consistent with the insertion of pRHB543 into the rpsL gene in strains TH172 and TH174 (Fig. 3A, X3). Strains TH173 and TH175, which contained single crossovers, had identical junction fragments of 23 and 4.3 kb, two pRHB543 internal fragments (7.2 and 1.4 kb), a 1.3-kb rpsL-hybridizing fragment, but not the 9.0-kb glnA-hybridizing fragment. In this case, one internal fragment (1.4 kb) and one junction fragment (4.3 kb) hybridized to the Hm r probe. This pattern is consistent with the insertion of pRHB543 into the glnA gene (Fig. 3A,  X2). Figure 3D shows that recombinant strains TH168 and TH169, which express the appropriate phenotype for a doublecrossover gene replacement (Hm r Sm r Am s ), contain identical Hm r -hybridizing junction fragments. One fragment (4.3 kb) is identical to one Hm r -hybridizing fragment in strains TH173 and TH175. The new junction fragment generated by the second crossover was about 7 kb. Neither TH168 or TH169 contained any plasmid sequences ( Fig. 3C and D), and each contained a single intact rpsL gene identical to that in wild-type S. roseosporus. This pattern of hybridization is consistent with exchange of glnA::Hm r from pRHB543 into the chromosome of the recombinants. In summary, these results indicate that, although recombination between plasmid and chromosomal rpsL was observed during selection for single crossovers, it did not interfere significantly with the ability to directly select for the desired recombinants containing double crossovers in a target gene. Since recombination between partially homologous sequences in streptomycetes occurs at greatly reduced frequencies compared to homologous recombination (13), the S. roseosporus rpsL system should be even more useful in other heterologous streptomycetes and possibly other actinomycetes.
Determination of GS activity in recombinant strains. Streptomyces species contain two GSs (10,11). GSI, encoded by the glnA gene, is heat stable, whereas GSII, encoded by the glnII gene, is heat labile (11). Streptomyces coelicolor glnA is regulated by a positive regulatory factor encoded by glnR (10). Biochemical studies of Streptomyces viridochromogenes showed that GSI and GSII activities are expressed under different conditions and that insertional inactivation of either glnA or glnII did not cause glutamine auxotrophy (11).
To determine if recombinant strains express GSI activity, we carried out GS assays with permeabilized S. roseosporus strains grown in CSM media with and without heat treatment. S. roseosporus A21978.6 and the Sm r strain TH149 expressed GS specific activities of 3.9 and 3.1, respectively, before heat treatment. Heat treatment (57ЊC for 30 min) of these strains caused an ϳ15% reduction in GS activity, similar to that in S. viridochromogenes (11). Strain TH168 containing glnA::Hm r had no detectable GS activity on CSM medium. It is not clear why heat-labile GSII activity was not detected. TH168 was prototrophic on CDA agar, suggesting that GSII activity was expressed on this medium. The recombinant strain TH173 containing a single-crossover insertion of plasmid pRHB543 into  (15), and Staphylococcus aureus (SarpsL) (37) were aligned, and amino acids identical to the corresponding S. roseosporus amino acids are shaded (28). A consensus sequence shows identical amino acids for all nine bacterial strains. Conserved amino acid substitutions are indicated by asterisks.
the glnA gene also expressed no detectable GS activity. Since the glnA fragment in pRHB543 lacks a promoter and C-terminal glnA sequences, a single crossover should generate two nonfunctional glnA genes (Fig. 3). Strain TH172 containing pRHB543 inserted into the rpsL gene expressed normal GS activity (specific activity, 3.1), as expected.
In summary, we have demonstrated that the rpsL gene of S. roseosporus can be used as a counterselectable marker in S. roseosporus, allowing for the direct selection for recombination events that result in gene replacement. This technology should have broad applications in Streptomyces species to stably insert genes into the chromosome, to construct mutants, and to modify secondary metabolite biosynthetic pathways. This should aid both fundamental and applied research in this important genus.