Engineering the Genome of Thermus thermophilus Using a Counterselectable Marker

ABSTRACT

Thermus thermophilus is an extremely thermophilic bacterium that is widely used as a model thermophile, in large part due to its amenability to genetic manipulation. Here we describe a system for the introduction of genomic point mutations or deletions using a counterselectable marker consisting of a conditionally lethal mutant allele of pheS encoding the phenylalanyl-tRNA synthetase α-subunit. Mutant PheS with an A294G amino acid substitution renders cells sensitive to the phenylalanine analog p-chlorophenylalanine. Insertion of the mutant pheS allele via a linked kanamycin resistance gene into a chromosomal locus provides a gene replacement intermediate that can be removed by homologous recombination using p-chlorophenylalanine as a counterselective agent. This selection is suitable for the sequential introduction of multiple mutations to produce a final strain unmarked by an antibiotic resistance gene. We demonstrated the utility of this method by constructing strains bearing either a point mutation in or a precise deletion of the rrsB gene encoding 16S rRNA. We also used this selection to identify spontaneous, large-scale deletions in the pTT27 megaplasmid, apparently mediated by either of the T. thermophilus insertion elements ISTth7 and ISTth8. One such deletion removed 121 kb, including 118 genes, or over half of pTT27, including multiple sugar hydrolase genes, and facilitated the development of a plasmid-encoded reporter system based on β-galactosidase. The ability to introduce mutations ranging from single base substitutions to large-scale deletions provides a potentially powerful tool for engineering the genome of T. thermophilus and possibly other thermophiles as well.

IMPORTANCE Thermus thermophilus is an extreme thermophile that has played an important part in the development of both biotechnology and basic biological research. Its suitability as a genetic model system is established by its natural competence for transformation, but the scarcity of genetic tools limits the kinds of manipulations that can currently be performed. We have developed a counterselectable marker that allows the introduction of unmarked deletions and point mutations into the T. thermophilus genome. We find that this marker can also be used to select large chromosomal deletions apparently resulting from aberrant transposition of endogenous insertion sequences. This system has the potential to advance the genetic manipulation of this important model organism.

INTRODUCTION

Thermus thermophilus is an extremely thermophilic bacterium, found globally in high-temperature environments. Thermus, Deinococcus, and several other recently described genera form a phylum branching deep within the Bacteria domain (1). T. thermophilus is one of a few favored laboratory model thermophiles, having a number of distinct advantages as an experimental system (reviewed in reference 2). Its aerobic growth and optimum growth temperature allow easy cultivation in the laboratory. Most importantly, T. thermophilus is amenable to genetic manipulation by virtue of its natural competence for transformation with either chromosomal or plasmid DNA (3), using the most efficient and rapid DNA uptake machinery yet described (4). The most widely used strain of T. thermophilus, HB27, has a genome consisting of a 1,894,877-bp main chromosome carrying 1,988 protein-coding genes and a 232,605-bp megaplasmid, pTT27, carrying 230 protein-coding genes (5). A sizeable number of these genes, 20% of the chromosomal genes and 39% of the pTT27 open reading frames, have no known function. As demonstrated by the discovery of Taq DNA polymerase (6) and, more recently, a DNA-dependent DNA interference system (7), the functional characterization of these genes has the potential to significantly impact both biotechnology and basic biological research.

A number of tools for the genetic manipulation of T. thermophilus have been developed. There exist several shuttle vectors (8–10) and host integration systems (11, 12), allowing the expression of foreign or mutated genes. There are also several gene expression reporter systems, including two based on carbohydrate hydrolases such as β-galactosidase (13) and β-glucosidase (14) and a third based on the crtB gene encoding phytoene synthase, an enzyme involved in carotenoid biosynthesis (15). Such systems are built using the few antibiotic resistance markers that have been artificially evolved to function at physiological temperature for T. thermophilus. These include kat and its derivative htk encoding kanamycin (Kan) adenyltransferases (16, 17), hph encoding hygromycin B (Hyg) phosphotransferase (18), and ble encoding a bleomycin-binding protein (19). Given the limited number of such markers and the significant effort involved in developing them, methods for genome manipulation that ultimately remove antibiotic resistance genes are highly advantageous. Such methods use counterselectable markers and allow the reuse of antibiotic resistance genes for sequential genetic manipulations or for the maintenance of plasmid replicons for protein expression and reporter systems. They also have the potential for use in deletion analysis or large-scale genome engineering.

Several counterselectable schemes have been devised for thermophiles. The scheme that is perhaps the most widely used is based on sensitivity to the uracil analog 5-fluoroorotic acid (5-FOA) as a counterselection against the pyrE and pyrF pyrimidine biosynthesis genes. This has been adapted for use in T. thermophilus (11) and other thermophilic bacteria and archaea (20–22). The toxic indoxyl product resulting from cleavage of the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-d-glucopyranoside (X-Glc) by β-glucosidase acts as a counterselection against the bgl gene (23) and could presumably be applied to other sugar hydrolases. Another utilizes a streptomycin-dependent allele of rpsL encoding ribosomal protein S12 (24).

Another counterselection that has been applied to mesophilic systems but has yet to be used with thermophiles takes advantage of the toxicity of the amino acid analog p-chlorophenylalanine (p-Cl-Phe) when incorporated into proteins via misacylation of ${\text{tRNA}}_{\text{GAA}}^{\text{Phe}}$ by a mutant phenylalanyl-tRNA synthetase (PheRS). This analog is normally excluded as a substrate for PheRS by the same structural features of the α-subunit active site that prevent misacylation of ${\text{tRNA}}_{\text{GAA}}^{\text{Phe}}$ with tyrosine. However, single amino-acid substitutions, including A294G in the active site of the synthetase α-subunit (encoded by pheS), allow misacylation to produce p-Cl-Phe-${\text{tRNA}}_{\text{GAA}}^{\text{Phe}}$ (25, 26). This property has led to the use of mutant pheS and p-Cl-Phe sensitivity for negative selection in Escherichia coli (27) and Streptococcus (28). In this approach, a mutant pheS copy is inserted at the point of interest by virtue of a linked antibiotic resistance gene. The final desired mutant allele, whether a deletion or a point mutation, is cloned onto a plasmid vector and then used to transform the pheS insertion mutant. Colony formation on plates containing p-Cl-Phe results from replacement of the pheS and antibiotic resistance gene with the mutant allele. This final mutant strain is unmarked by any antibiotic resistance genes. Here we describe the application of the p-Cl-Phe selection for introducing unmarked deletions and point mutations in the genome of T. thermophilus. This system consists of a cassette containing a p-Cl-Phe-sensitive allele of pheS followed by the htk gene encoding a kanamycin adenyltransferase (17). To minimize the potential for homologous recombination and gene conversion with the native chromosomal pheS locus, we developed a synthetic pheS gene composed of synonymous codons.

We tested the potential utility of this counterselection by applying it to the engineering of mutants of T. thermophilus HB27 unmarked by drug-resistance genes. In one example, we constructed a strain bearing a clean deletion of rrsB, one of two 16S rRNA gene copies in the genome. In a second example, we demonstrated the ability to perform site-directed mutagenesis of chromosomal genes by introducing a single-base antibiotic resistance mutation into rrsB. Unexpectedly, we also found that this counterselection is capable of generating large-scale deletions, apparently resulting from aberrant transposition by native IS elements, including T. thermophilus ISTth7 and ISTth8. As indicated by these results, this counterselectable marker will likely be a valuable asset to future genome engineering efforts involving T. thermophilus.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The strains used in this study are listed in Table 1. All experiments were conducted using T. thermophilus HB27 (ATCC BAA-163) (29), which was cultivated in Thermus enhanced medium (TEM; ATCC medium 1598). All cultures were grown at 65°C with shaking at 180 rpm in a New Brunswick Innova 4200 incubator shaker or on plates with TEM solidified with 2.8% Bacto agar (Difco).

TABLE 1.

Bacterial strains used in this study^a

Strain	Genotype	Source
HB27	Wild type	ATCC BAA-163
MD9	ΔrrsB::pheS1-htk	From HB27 by gene replacement
MD33	ΔrrsB::pheS2-htk	From HB27 by gene replacement
MD42	ΔrrsB	From MD33 by gene replacement
JC1792	rrsB-U1495C	From MD33 by transformation with genomic DNA from HG309
MD132	TT_P0041::pheS2-htk	From HB27 by gene replacement
MD136	ΔcrtB::pheS2-htk	From HB27 by gene replacement
MD156	Δ(TT_P0229-TT_P0063)	Spontaneous p-Cl-Phe-resistant derivative of MD132
MD158	Δ(TT_P0189-TT_P0079)	Spontaneous p-Cl-Phe-resistant derivative of MD132
JC1851	Δ(TT_P0179-TT_P0162)	Spontaneous p-Cl-Phe-resistant derivative of MD136
JC1852	Δ(TT_P0170-TT_P0062)	Spontaneous p-Cl-Phe-resistant derivative of MD136

^aAll strains are derived from T. thermophilus HB27. The deletion endpoints for JC1851 and JC1852 have not been precisely determined.

Preparation of T. thermophilus genomic DNA.

Genomic DNA was prepared by pelleting cells from 3 ml of saturated overnight cultures and resuspending them in 250 μl of 50 mM Tris-HCl (pH 8.0)–50 mM EDTA and freezing overnight. A 25-μl volume of lysozyme solution (10 mg/ml lysozyme, 250 mM Tris-HCl [pH 8.0]) was added to frozen cell suspensions, which were then thawed at room temperature and kept on ice for 15 min. Samples were treated with 50 μl of STEP solution (0.5% SDS, 50 mM Tris-HCl [pH 7.5], 50 mM EDTA, 1 mg/ml proteinase K) and incubated at 65°C for 15 min. RNase A (5 μl) was added, and the reaction mixture was incubated at 37°C for 30 min. Tubes were chilled on ice for 5 min, and protein precipitation solution (PPS; Promega catalog number A7953) was added. The solution was subjected to vortex mixing, chilled on ice for 5 min, and then centrifuged for 10 min to pellet the precipitate. The supernatant was mixed with 0.4 ml isopropanol to precipitate DNA, which was then pelleted, washed with 70% ethanol, dried, and dissolved in 100 μl Tris-EDTA (TE). This method purified both the chromosomal DNA and the pTT27 megaplasmid.

Gene synthesis.

A codon-optimized pheS coding sequence was commercially synthesized by GeneWiz based on locus tag TT_C1594 from the published genome sequence of T. thermophilus HB27 (GenBank accession number AE017221, gene identification no. [ID] 46197523; 5). The annotated sequence of the synthetic pheS2 gene is presented in Fig. S1 in the supplemental material.

Molecular biology techniques.

The plasmids used in this study are listed in Table 2. Plasmids for gene replacement were constructed using either standard cloning methods or a Gibson assembly cloning kit from New England BioLabs (catalog number E5510S), following the recommendations of the manufacturer. Plasmid DNA was prepared using a Zymo Research miniprep kit from Genesee Scientific. Details of plasmid constructions are given in Materials and Methods in the supplemental material.

TABLE 2.

Plasmids used in this study

Plasmid	Characteristic(s)	Source
pUC18	Cloning vector	S. T. Gregory collection
pUC18ΔrrsB::htk	pUC18 with htk inserted between 500-bp chromosomal sequences flanking rrsB	S. T. Gregory collection
pUC18ΔrrsB::htk-NoKpnI	pUC18ΔrrsB::htk with the KpnI site in the MCSa mutated by PCR mutagenesis	This study
pUC18ΔrrsB::pheS1-htk	pUC18ΔrrsB::htk-NoKpnI with pheS1 inserted upstream of htk	This study
pMD32	pUC18ΔrrsB::htk-NoKpnI with pheS2 inserted upstream of htk	This study
pMD56	pUC18 with the rrsB UHR and DHR assembled into the MCS	This study
pMD123	pUC18 with bgl UHR, pheS2-htk, and bgl DHR assembled into the MCS	This study
pMD125	pUC18 with crtB UHR, pheS2-htk, and crtB DHR assembled into the MCS	This study
pBGAA1	Plasmid expressing β-galactosidase	This study

^aMCS, multiple cloning site.

Reporter plasmid pBGAA1 was constructed by Gibson assembly using PCR products derived from several templates as follows. The pTT8 ori and repA sequences, as well as the R6Kγori and hph genes, were amplified from plasmid pJC1111, obtained by in vitro transposition of the hph gene as part of a synthetic transposon into pTT8 (30). The bgaA gene was amplified from total genomic DNA from T. thermophilus IB-21 (31). The annotated sequence of pBGAA1 is presented in Fig. S9 in the supplemental material. Oligonucleotides were synthesized by and purchased from Integrated DNA Technologies (IDT), and their sequences are listed in Tables S1 and S2 in the supplemental material. Q5 PCR mutagenesis was performed using a kit from New England BioLabs (catalog number E0554S).

Transformation and selection of p-Cl-Phe-resistant recombinants.

T. thermophilus was transformed with plasmid or genomic DNA according to the method of Koyama et al. (3); transformants were selected on TEM plates containing 30 μg/ml kanamycin sulfate (Sigma) or 15 mM p-Cl-Phe (Sigma; catalog number C6506) and incubated at 65°C. Because of the extremely poor water solubility of p-Cl-Phe, the compound was added directly to TEM agar medium prior to autoclaving. Initial tests for sensitivity to p-Cl-Phe were performed using a disc assay in which a 6-mm-diameter Whatman filter disc (catalog no. 2017-006) was infused with a total of 100 μg p-Cl-Phe and placed on a TEM plate after spreading 100 μl of a saturated overnight culture.

Assays for enzyme activity.

β-Galactosidase enzyme activity was measured using p-nitrophenyl-β-d-galactopyranoside (PNPG; Sigma catalog number N1252) as a substrate. Cultures were grown to mid-log phase and chilled on ice. Cell density was adjusted to an optical density at 600 nm (OD₆₀₀) of 0.5 with TEM. A 5-ml volume of each culture was mixed with 5 ml of Z-Buffer (60 mM Na₂HPO₄·7H₂O, 40 mM NaH₂PO₄·H₂O, 10 mM KCl, 1 mM MgSO₄·7H₂O, 0.0027% β-mercaptoethanol) (32). Alternatively, 1 ml of cells was mixed with 9 ml of Z-Buffer. Cells were permeabilized by adding 80 μl chloroform and 20 μl of 10% SDS and vortex mixing for 10 s. A 2-ml volume of 4 mg/ml PNPG was added to cells, and 1-ml aliquots were dispensed to Eppendorf tubes and incubated at 65°C in an Eppendorf thermostat. Aliquots were removed at various times, and reactions were stopped by addition of 0.5 ml of 1 M Na₂CO₃. Cells were pelleted by centrifugation, and the OD₄₂₀ was measured spectrophotometrically. In vivo activity of β-galactosidase was detected histochemically in bacterial colonies by plating on TEM on which was spread the chromogenic substrate X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; Sigma catalog number B9146).

Nucleotide sequence accession numbers.

DNA sequences have been deposited in GenBank for the synthetic pheS2 gene (accession number KM659911) and plasmid pBGAA1 (accession number KM659912).

RESULTS

The A294G amino acid substitution mutation in pheS confers p-Cl-Phe sensitivity to T. thermophilus.

In order to test the feasibility of using p-Cl-Phe counterselection, we first tested the sensitivity of T. thermophilus HB27 to p-Cl-Phe using a disc assay, in which a filter paper disc infused with p-Cl-Phe is placed on a TEM plate after a saturated overnight culture was spread on the plate. In this assay, wild-type T. thermophilus HB27 shows no zone of inhibition, consistent with the exclusion of the analog from the PheRS active site, although failure of p-Cl-Phe to enter T. thermophilus cells would also explain the lack of inhibition.

We next tested the effect of the PheS-A294G amino acid substitution on p-Cl-Phe sensitivity by mutagenizing the T. thermophilus HB27 pheS gene (locus tag TT_C1594, GeneID no. 2776131, GenBank accession no. AE017221.1; 5) and inserting it ectopically into the chromosome. The promoter responsible for pheS transcription has not been determined experimentally, and we were unable to detect any promoter-like sequence within 500 bp upstream of the start site by algorithms designed to detect bacterial promoters (33, 34). However, pheS is unlikely to be part of an operon, as its start codon overlaps that of an oppositely oriented upstream orf, TT_C1593, encoding a hypothetical protein. We therefore cloned the pheS allele along with 200 bp of upstream sequence to include any likely promoter sequence.

The cloned pheS gene was mutated to encode the A294G amino acid substitution and is referred to here as the pheS1 allele. For gene replacement, the pheS1 sequence was inserted into a plasmid construct upstream of the htk gene, encoding a thermostable kanamycin adenyltransferase, whose transcription is driven by the Geobacillus kaustophilus cytidylate kinase promoter (17). The final plasmid construct consisted of pUC18, into which was inserted the pheS1-htk cassette surrounded by chromosomal sequencing flanking the rrsB locus, one of two genes encoding 16S rRNA. This construct was used to delete rrsB by transformation and homologous recombination, selecting kanamycin resistance (Kan^r). Individual transformants were purified by streaking to single colonies and screened by PCR of the rrsB locus to confirm the replacement of rrsB with the pheS1-htk construct. Deletion of rrsB and its replacement by the antibiotic resistance htk gene are consistent with our previous observation that T. thermophilus can survive with a single rrs gene (35). The ΔrrsB::pheS1-htk allele does indeed confer sensitivity to p-Cl-Phe, consistent with expression of active, mutant phenylalanyl-tRNA-synthetase capable of misacylation to produce p-Cl-Phe-${\text{tRNA}}_{\text{GAA}}^{\text{Phe}}$. The normal colony formation of this strain further indicated that any misacylation to produce Tyr-${\text{tRNA}}_{\text{GAA}}^{\text{Phe}}$ by the mutant PheS is not sufficient to impair growth.

Design and construction of a synthetic, codon-optimized pheS2 allele.

Potentially associated with this approach is the problem of homologous recombination between the pheS1 allele used for gene replacement and the native chromosomal copy of pheS. This can potentially occur with transforming plasmid DNA either during the initial construction of the intermediate strain or after insertion into the desired chromosomal locus. This could lead either to gene conversion and loss of the A294G mutation or to chromosomal rearrangements. To minimize homologous recombination, we designed a synthetic mutant allele of pheS (here referred to as pheS2) whose coding sequence is composed of abundantly used synonymous codons, based on data obtained from the codon usage database (http://www.kazusa.or.jp/codon/; 36). Rare codons were deliberately avoided so as to facilitate expression. The pheS2 coding sequence contains 238 single-base changes in 1,053 total bases from the wild-type pheS coding sequence, including the two base changes needed to incorporate the A294G mutation conferring sensitivity to p-Cl-Phe (see Fig. S1 in the supplemental material).

Generalized scheme for gene replacement.

As has been described in previous studies using knockout constructions and antibiotic resistance markers, upstream and downstream flanking sequences (referred to here as upstream and downstream homology regions or, for brevity, UHR and DHR) extending for approximately 500 bp in either direction are inserted into E. coli cloning vector pUC18. Inserted between the UHR and DHR is the pheS2-htk cassette, which is amplified by PCR. Such constructs are most effectively made using Gibson assembly (37). As the replication origin of pUC-derived plasmids is not active in T. thermophilus, transformation with the knockout constructs to Kan^r results from homologous recombination between the UHR and DHR with their respective chromosomal sequences. In contrast to findings with E. coli (38, 39), double crossovers work very efficiently in T. thermophilus, without the need to select for single insertions and subsequent cointegrate resolution (17, 35, 40). While the latter approach does have the advantage of allowing the use of a single plasmid construct for insertion of pheS and its replacement, our preference is to construct an intermediate double-crossover strain as it allows replacement by transformation with genomic DNA as well as plasmids.

Once the pheS2-htk cassette (see Fig. S2 in the supplemental material) is inserted onto the chromosomal locus, as confirmed by PCR, a second construct, a derivative of pUC18 containing the final desired unmarked deletion allele or point mutation allele, is used to transform the intermediate strain to p-Cl-Phe resistance. We found that transformation and selection on p-Cl-Phe plates generally produce on the order of several hundred colonies, depending on the specific construct and locus being manipulated. Spontaneous p-Cl-Phe-resistant colonies arise on no-DNA control plates at a frequency of 2.5 × 10⁻⁷. Sequencing of the pheS2 gene from several such colonies failed to identify any reversion mutation, suggesting the possibility of resistance arising by some other mechanism, including mutations affecting uptake of p-Cl-Phe from the medium. This background of resistant colonies is high enough to require screening prior to purification. We find that the most effective approach is to screen 100 to 200 colonies for loss of kanamycin resistance to ensure identifying the correct clone. Several kanamycin-sensitive (Kan^s) transformants are then purified by serially streaking to single colonies and analyzed by PCR and sequencing.

Deletion of a 16S rRNA gene.

The T. thermophilus HB27 genome produces 16S rRNA from two loci (rrsA and rrsB) that are unlinked to two rRNA operons, each encoding 23S rRNA (rrlA or rrlB), 5S rRNA (rrfA), and ${\text{tRNA}}_{\text{GCC}}^{\text{Gly}}$ (glyT) as first described by Hartmann and Erdmann (41) and later indicated by the completed genome sequence (5). Our previous studies have shown that T. thermophilus is capable of surviving on a single rrs gene (35). Such deletion strains facilitate the genetic analysis of rRNA genes and allow the isolation of mutants with recessive antibiotic resistance mutations (35). Since two of the available selectable markers confer resistance to the ribosome-targeted antibiotics kanamycin and hygromycin B, having a ΔrrsB strain lacking such drug-resistance genes would be of some advantage for investigating mechanisms of antibiotic resistance due to mutations in the ribosome.

We constructed a precise, unmarked rrsB deletion in two sequential steps. First, strain MD33 was constructed by deleting rrsB and replacing it with the pheS2-htk cassette, selecting Kan^r (Fig. 1A; see also Fig. S3 in the supplemental material). In the second step, the ΔrrsB::pheS2-htk allele was replaced by homologous recombination with the final unmarked ΔrrsB allele to produce strain MD42 (Fig. 1B). Half of the p-Cl-Phe-resistant colonies tested Kan^s. The absence of rrsB was confirmed by PCR (Fig. 1D).

FIG 1.

Using pheS2 and the p-Cl-Phe counterselection to delete the rrsB gene encoding 16S rRNA. (A) Construction of MD33 (ΔrrsB::pheS2-htk), the deletion intermediate in deletion of rrsB. (B) Construction of MD42 (ΔrrsB), the final unmarked deletion strain. (C) Construction of JC1792 (rrsB-U1495C) from MD33. (D) PCR analysis of the rrsB gene before and after mutagenesis. Lane 1, 1 kb-ladder; lane 2, HB27; lane 3, MD33; lane 4, MD42; lane 5, JC1792.

Introducing point mutations by counterselection.

The p-Cl-Phe counterselection is useful not only in generating unmarked deletions but also for introducing site-specific point mutations into chromosomal genes. Starting with MD33, we replaced the ΔrrsB::pheS2-htk allele with a rrsB gene containing a single mutation, a U1495C base substitution in 16S rRNA that confers hygromycin B resistance (Hyg^r). This was done by transforming MD33 with genomic DNA from a previously described spontaneous Hyg^r mutant bearing a U1495C base substitution in 16S rRNA (42), selecting p-Cl-Phe resistance, and screening for acquisition of the Hyg^r phenotype (Fig. 1C). A 22% proportion of the p-Cl-Phe-resistant colonies tested Kan^s. The restoration of the native rrsB locus was confirmed by PCR (Fig. 1D) and sequencing. This clearly demonstrates one advantage of a system that does not leave antibiotic resistance genes at sites of deletions, as they can interfere with studies of antibiotic resistance mutations in the ribosome. It also serves to illustrate the advantage of the natural competence of T. thermophilus for transformation with chromosomal DNA and incorporation of mutations by homologous recombination.

Spontaneous large-scale deletions of megaplasmid pTT27.

The pTT27 megaplasmid contains a number of genes encoding enzymes that are of interest as potential reporters of gene expression. Among these sugar hydrolases are two β-galactosidases (TT_P0220 and TT_P0222, bga), two α-galactosidases (TT_P0221 and TT_P0072, aga), and a β-glucosidase (TT_P0042, bgl). It also carries genes involved in carotenoid biosynthesis, including phytoene synthetase (TT_P0057, crtB). We therefore began to systematically mutagenize individual genes from this group. We first inserted the pheS2-htk cassette upstream of the bgl start site in TT_P0041, without deleting any chromosomal sequence, to produce strain MD132 (Fig. 2A; see also Fig. S4 in the supplemental material). In a separate experiment, we deleted crtB (TT_P0057) and replaced it with pheS2-htk to produce MD136 (see Fig. S5 in the supplemental material). During attempts to introduce point mutations into bgl, we found, within the background of spontaneous p-Cl-Phe-resistant mutants arising in selections, mutants lacking multiple loci around the pheS2-htk insertion site. PCR analysis of the TT_P0041-to-TT_P0042 region of p-Cl-Phe-resistant colonies failed to produce products corresponding to either the original insertion allele or the expected replacement. All colonies tested Kan^s, consistent with deletions removing the pheS2-htk cassette occurring at high frequency. PCR analysis of surrounding genes also failed to produce product, suggestive of substantial deletions.

FIG 2.

Large-scale deletions of pTT27. (A) Construction of the bgl::pheS2-htk MD132 insertion strain, the source of MD156 and MD158. (B) Genetic map of pTT27 showing the deletions found in MD156 and MD158. The four ISTth7 elements and the single ISTth8 element in pTT27 are indicated (green), as are genes encoding sugar hydrolases (blue), crtB (orange), ago (yellow), and cas6 (violet). Locus tag numbers are given in parentheses. (C) Structure of the MD156 deletion junction, derived from MD132. Open reading frames deleted from MD156 are white; those retained are gray. ISTth7 is shown in green, and the inversion within ISTth7 is indicated in violet. (D) Structure of the MD158 deletion junction, derived from MD132. Open reading frames deleted in MD158 are white; those retained are gray. ISTth8 is shown in green.

An extensive PCR analysis of two derivatives, MD156 and MD158, revealed large-scale deletions in pTT27 (Fig. 2B; see also Fig. S6 in the supplemental material). The deletion endpoints were initially physically mapped by sequentially PCR amplifying surrounding open reading frames, walking outward from TT_P0041 to map the limits of each deletion. From these analyses, we determined that the MD156 deletion extends from TT_P0229 to TT_P0063 (pTT27 coordinates 230320 to 57598, or a distance of 59,883 bp) and that the MD158 deletion extends from TT_P0189 to TT_P0079 (pTT27 coordinates 186533 to 74904, or a distance of 120,976 bp). The latter deletion removes just over half the 233-kb pTT27 megaplasmid. These deletions correspond to the loss of 65 and 118 genes, respectively. Parenthetically, these data also indicate that no genes essential for either replication and maintenance of pTT27 or viability of T. thermophilus under laboratory conditions likely reside between TT_P0188 and TT_P0080, although insertion of the remaining megaplasmid into the main chromosome by recombination cannot be ruled out. Fusion of the main chromosome and the megaplasmid was observed to be the natural state of the genome of Thermus scotoductus (43).

A number of genes of interest were removed by these deletions. Both the MD156 and MD158 deletions removed crtB (TT_P0057), which encodes phytoene synthetase; this enzyme participates in the biosynthesis of carotenoid pigments that are responsible for the orange-yellow color of T. thermophilus colonies. Both deletions removed a cluster of 23 genes (from TT_P0001 to TT_P0023) encoding enzymes involved in the biosynthesis of cobalamin. The deletion in MD158 also removed cas6 (TT_P0204) encoding the Cas6 protein involved in the T. thermophilus CRISPR-Cas adaptive immunity system (44). Finally, both the MD156 and MD158 deletions removed ago (TT_P0026) encoding the Argonaute protein involved in a recently discovered DNA-guided DNA interference system; deletion of ago has been shown by others to enhance transformation efficiency (7), and we have confirmed this observation (data not shown).

The MD158 deletion also removed two genes, TT_P0196 and TT_P0197, encoding hypothetical conserved proteins more recently identified as a putative nuclease and a RecB-family exonuclease; deletion of these genes has been reported to cause a decreased growth rate at high temperature (45). While these proteins have been suggested to be involved in DNA repair, we did not find that MD158 exhibited a higher frequency of mutation to streptomycin resistance than either HB27 or MD156 (data not shown). While the genes deleted in MD156 or MD158 are not essential, they nonetheless appear to include genes advantageous for growth, at least under the laboratory conditions examined here. The doubling times for wild-type HB27, MD156, and MD158 were measured as 51 min, 54 min, and 63 min, respectively. The more substantial disadvantage of the MD158 deletion could be attributed to loss of sugar hydrolases or DNA repair enzymes, which are retained by MD156. The characterization of a larger number of large deletion mutants could provide a more detailed picture of the fitness contribution of genes on pTT27.

Analysis of deletion endpoints.

Sequencing of the endpoints of the MD156 deletion indicated that it most likely resulted from aberrant transposition of insertion sequence ISTth7. TT_P0063 corresponds to the ISTth7 tnpA transposase gene (see Fig. S7 in the supplemental material). We have previously reported the active transposition of ISTth7 (46), as have others (7). The absence of an IS element at or near the second endpoint of the deletion argues against homologous recombination between identical copies of IS elements as the mechanism of deletion formation. In addition to the deletion, this locus carries an inversion of 321 bp of ISTth7 (Fig. 2C). The inversion is flanked by 9-bp inverted repeats, one of which is at the deletion endpoint. The right end of the inversion corresponds precisely with the ISTth7 right-end 12-bp inverted terminal repeat but does not create the 9-bp direct repeat within ISTth7 that would be expected for a canonical transposition event (46). This complex rearrangement suggests that the deletion resulted from a failed transposition event involving the right end of this copy of ISTth7.

Sequencing of the larger MD158 deletion similarly showed a deletion endpoint precisely at an IS element. While another copy of ISTth7, on the noncoding strand corresponding to TT_P0079, is located near one of the deletion endpoints in MD158, the actual deletion endpoint occurs at the adjacent open reading frame corresponding to a different IS element, ISTth8 (Fig. 2D; see also Fig. S8 in the supplemental material). ISTth8 has not been reported to be transpositionally active (ISfinder database, www-is.biotoul.fr; 47). Our results here suggest that ISTth8 does encode an active transposase, although here again the mechanism of aberrant transposition is unclear. In this case, there was no inversion in ISTth8 such as we observed with the ISTth7 involved in the MD156 deletion. There was also no direct repeat, as would be expected for a transposition event.

We determined that this type of deletion event is not dependent on a specific insertion at bgl. MD136 (ΔcrtB::pheS2-htk), in which crtB was deleted and replaced with pheS2, also gave rise to a large number of spontaneous p-Cl-Phe resistant colonies. A cursory examination by PCR indicated that these, too, contain deletions extending beyond crtB. JC1851 retains TT_P0170, TT_P0154, and TT_P0107 and has lost TT_P0179, TT_P0013, TT_P0035, and TT_P0062. A second mutant, JC1852, has lost TT_P0154 and TT_P0107 and lacks TT_P0170, TT_P0179, TT_P0013, TT_P0035, and TT_P0062. A thorough examination of a number of such events will be required to establish a model for the relationship between the inversion and the deletion and the mechanism by which they form. Though such events have not been previously reported in T. thermophilus, they must occur at a substantial frequency, given our ability to detect them using p-Cl-Phe counterselection.

An effective β-galactosidase reporter system.

Fortuitously, the large deletion arising in MD158 removed a number of sugar hydrolase-encoding genes, including two β-galactosidases (TT_P0220 and TT_P0222, bga), two α-galactosidases (TT_P0221 and TT_P0072, aga), and a β-glucosidase (TT_P0042, bgl). This provided an ideal genetic background for developing a plasmid-based reporter system using any of these enzymes. For this purpose, we constructed reporter plasmid pBGAA1 expressing a thermostable β-galactosidase (Fig. 3A). This construct replicates in T. thermophilus by virtue of the repA gene and oriV1 and oriV2 replication origins of plasmid pTT8 (48, 49), a 9-kb cryptic plasmid found natively at about 8 copies per chromosome in T. thermophilus strain HB8 but not in strain HB27 (50). pBGAA1 is maintained in E. coli by virtue of the plasmid R6Kγori replication origin (51) and carries a thermostable hph gene conferring Hyg^r at both 37°C and 65°C (18), allowing selection in either E. coli or T. thermophilus. The reporter gene of pBGAA1 is bgaA (GenBank accession number AY130259) (52) from the Icelandic T. thermophilus IB-21 strain (31). This gene is not homologous to, and therefore does not recombine with, TT_P0220 or TT_P0222 and so can also be used in wild-type T. thermophilus HB27, albeit with some background activity. Transcription of bgaA in pBGAA1 occurs by continued transcription from the hph gene located just upstream, which itself is driven by the slpA promoter. The β-galactosidase produced by this plasmid is also active in E. coli, causing the formation of blue colonies on plates containing X-Gal. The annotated sequence of pBGAA1 is given in Fig. S9 in the supplemental material.

FIG 3.

A β-galactosidase reporter system. (A) Map of the reporter plasmid pBGAA1. (B) Phenotype of MD158 with and without pBGAA1 plated on TEM plus X-Gal. (C) β-Galactosidase enzyme activity in HB27, MD158, and MD158/pBGAA1.

MD158 containing pBGAA1 produces deep-blue colonies on plates containing X-Gal (Fig. 3B). Assays for β-galactosidase enzyme activity using PNPG as the substrate showed high levels of activity in comparison to wild-type HB27. MD158 showed even less PNPG hydrolysis, which is attributable to spontaneous hydrolysis of the substrate at 65°C (Fig. 3C). Given the number of genes of unknown function that are deleted from MD158 and its observed growth disadvantage, it will eventually be desirable to construct a derivative of HB27 bearing unmarked deletions of TT_P0220, TT_P0222, and other potential reporters, made sequentially using p-Cl-Phe counterselection.

DISCUSSION

The system described here has the potential to significantly facilitate genetic analysis and genome engineering of T. thermophilus. We have demonstrated the ability to make directed mutations ranging from single-point mutations to deletions on the order of 1 to 2 kb. In principle, this system could also be used to generate directed large-scale genome rearrangements, although we have yet to ascertain the size limit of large deletions or replacements. Nevertheless, the DNA repair system of T. thermophilus seems able to rejoin ends that are as far apart as 121 kb. While of great advantage, the remarkable efficiency with which T. thermophilus performs homologous recombination could also negatively impact the ability to perform contiguous replacements of large sections of the T. thermophilus genome. Such experiments are the next logical step in developing this system for genome engineering projects.

One of the serious challenges to T. thermophilus genome engineering is the marked prevalence of overlapping genes. In some instances, the termination codon of one gene overlaps the initiation codon of a downstream gene (e.g., 5′-TGATG-3′). This arrangement makes precise deletions difficult to construct, and in many cases partial deletions have to suffice. Annotated in the HB27 genome is a hypothetical protein encoded by TT_C1382, which is on the strand opposite that of rrsB and whose start codon is immediately upstream of the rrsB −35 promoter element. Our deletion construct removes the start codon for TT_C1382 with no apparent consequences. It seems somewhat unlikely that TT_C1382 represents an actual gene, as its promoter sequence would need to be encoded with the rrsB leader sequence or within the 16S rRNA coding sequence. The construction of individual deletions will therefore need to be assessed on a case-by-case basis.

While several other counterselections for T. thermophilus have been reported, the system described here has the advantage that it does not require genetically marked strains, as in the case of pyrE (11) and bgl (23). There are several aspects to this system that could be expanded upon in future versions. Since a number of strains constructed in other laboratories may have been previously marked with either of the kanamycin-resistance genes kat (16) and htk (17), the synthetic pheS gene can easily be joined with the hph gene (18) or the bleS gene conferring bleomycin resistance (19). Given the facility with which constructs can now be made by Gibson assembly, such constructions will not impede future progress.

Our observation of large-scale deletions generated by the p-Cl-Phe selection is thus far limited to genes on the pTT27 megaplasmid. The ability to delete regions as large as 121 kb in size probably reflects the lack of essential genes on this accessory element, consistent with its documented plasticity (45). In contrast, the greater density of potentially essential genes on the main chromosome will place constraints on the location and size of possible deletions. Large deletions on this order have been observed in Agrobacterium tumefaciens megaplasmid pAtC58 and were deemed to result from homologous recombination between short repeated sequences (53). The deletions in the T. thermophilus pTT27 megaplasmid do not appear to arise by homologous recombination but instead result from some form of aberrant transposition involving one end of an IS element (see Fig. S6 and S7 in the supplemental material). Thus, the nature of spontaneous deletions will also be influenced by the location and orientation of IS elements.

While the mechanism by which T. thermophilus IS elements stimulate deletion formation is unclear at present, the involvement of mobile genetic elements in genome rearrangements is well documented in other systems. Tn10 insertions can generate spontaneous adjacent chromosomal deletions in Salmonella (54); restriction analysis indicates that such deletions have one endpoint at or near the terminus of one of the inverted repeats (55), as we have observed with ISTth7- or ISTth8-mediated deletions in T. thermophilus. Tn10 insertions can also be used to generate adjacent deletions by selection on medium containing fusaric acid and chlortetracycline (56). Combining ISTth7 with the pheS2-htk cassette can potentially provide an analogous system.

The occurrence of spontaneous large deletions is potentially problematic for the construction of precise gene replacements in pTT27 or in regions of the main chromosome devoid of essential genes. If such deletion events are generally dependent on aberrant transposition of IS elements, they could be prevented by systematically removing IS elements where they are not wanted, using this two-step counterselection method. Since the IS element is deleted and replaced with the pheS2-htk cassette in the intermediate step, prior to the selection on p-Cl-Phe, its removal would not result in large-scale deletions.

Alternatively, the generation of large-scale deletions by the p-Cl-Phe counterselection method provides a potential tool for extensive genome engineering of T. thermophilus. For large-scale deletion analysis of the main chromosome, it should also be possible to insert either ISTth7 or ISTth8, along with an adjacent pheS2-htk cassette, into a desired position by homologous recombination, rather than transposition, and select for spontaneous deletions. Inclusion of an E. coli plasmid replication origin within the construct would allow rapid mapping and sequencing of deletion junctions by rescue cloning (30). Such a system could make possible extensive deletion analysis or engineering of the genome of this thermophile and possibly of other thermophiles as well.