Circularized Chromosome with a Large Palindromic Structure in Streptomyces griseus Mutants

Abstract

Streptomyces linear chromosomes display various types of rearrangements after telomere deletion, including circularization, arm replacement, and amplification. We analyzed the new chromosomal deletion mutants Streptomyces griseus 301-22-L and 301-22-M. In these mutants, chromosomal arm replacement resulted in long terminal inverted repeats (TIRs) at both ends; different sizes were deleted again and recombined inside the TIRs, resulting in a circular chromosome with an extremely large palindrome. Short palindromic sequences were found in parent strain 2247, and these sequences might have played a role in the formation of this unique structure. Dynamic structural changes of Streptomyces linear chromosomes shown by this and previous studies revealed extraordinary strategies of members of this genus to keep a functional chromosome, even if it is linear or circular.

The filamentous soil bacteria belonging to the genus Streptomyces are unusual because they have an 8- to 9-Mb linear chromosome (5, 26, 37). Streptomyces genome projects have been completed for Streptomyces coelicolor A3(2), a model strain for Streptomyces genetics (3), and for Streptomyces avermitilis, an avermectin producer (14). Streptomyces species are also known to frequently have linear plasmids, which range from 12 to 1,700 kb long (10, 19, 21). For the most part these Streptomyces linear replicons have the same structural features; terminal inverted repeats (TIRs) are present at both ends, and a terminal protein is linked covalently to the 5′ end (1, 39). Intramolecular interaction of two terminal proteins may combine two telomeres in circular form in living cells (40).

Streptomyces linear chromosomes undergo telomere deletion spontaneously or after various mutagenic treatments (23, 36). The sizes of deletions sometimes are up to 2 Mb (7). The chromosomes with deletions subsequently display several types of rearrangements, including circularization, arm replacement, and amplification. Chromosomal circularization in Streptomyces was indicated by detection of a new fusion fragment (7, 25, 28) and was confirmed by cloning and sequencing of the fusion junction of the circularized chromosome (15, 17). No homology and microhomology were detected between the right and left deletion ends of the circularized chromosomes. This suggested that chromosomal circularization occurs by nonhomologous recombination between deletion ends. Furthermore, Qin and Cohen (31) reported that nonhomologous recombination also produced circular plasmids from linear plasmids with damaged telomeres. Therefore, circularization may be a common strategy of Streptomyces linear replicons when both of the telomeres are deleted.

In addition to circularization, Streptomyces chromosomes have other structural changes that allow them to survive after the loss of an extreme end that is essential for terminal replication (12, 29). Fischer et al. (8) reported that homologous recombination of two sigma factor-like open reading frames (ORFs) caused chromosomal arm replacement in Streptomyces ambofaciens, which produced unusually long TIRs in the mutants. It has also been found that in Streptomyces griseus homologous recombination between two lipoprotein-like ORFs caused chromosomal arm replacement (35). Qin and Cohen (30) reported another strategy of Streptomyces, in which duplication of the right side of the linear plasmid pSLA2 resulted in palindromic linear plasmids when the left telomere was damaged. All of these rearrangements may be strategies used by Streptomyces strains to recover an extreme end when one of the two telomeres is deleted.

S. griseus, one of the Streptomyces species that has been studied best physiologically, has a 7.8-Mb linear chromosome (24). As described above, chromosomal circularization and arm replacement have been observed in deletion mutants. In our continuing studies on S. griseus, we isolated additional deletion mutants 301-22-L and 301-22-M, which have a new type of chromosomal structure. In this paper, we describe circularized chromosomes with a large palindrome which were formed by serial rearrangements. Based on this and previous studies, we discuss dynamic structural changes in Streptomyces linear replicons below.

MATERIALS AND METHODS

Bacterial strains, cosmids, plasmids, and media.

S. griseus strain 2247, which was used for mutation, was described previously (24). A cosmid library was constructed previously for strain 2247 (24), and the cosmid order at both chromosomal ends was determined (25). Escherichia coli XL1-Blue and pUC19 were used for cloning and sequencing of DNA fragments. Glucose-meat extract-peptone medium contained 1.0% glucose, 0.4% peptone, 0.2% meat extract, 0.2% yeast extract, 0.5% NaCl, and 0.025% MgSO₄ · 7H₂O (pH 7.0). MB medium plates contained 1.0% mannitol, 0.1% yeast extract, 0.2% Polypeptone (Wako Chemicals, Osaka, Japan), 0.1% meat extract, and 1.5% agar (pH 7.0).

Mutation of S. griseus 2247.

A spore suspension of strain 2247 in distilled water was UV irradiated at a distance of 60 cm by using a germicidal lamp (15 W) for 0.5 to 5 min with constant stirring. Aliquots were taken every 30 s, kept in the dark for 2 h to prevent photoreactivation, and spread and grown on MB plates at 28°C for several days. The level of survival for the spore suspension, which was irradiated for 2.5 to 3.5 min, was 0.1 to 1%, and this suspension was used for isolation of deletion mutants.

DNA isolation and PFGE.

S. griseus strains were cultured on a reciprocal shaker at 28°C in a 500-ml Sakaguchi flask containing 100 ml of glucose-meat extract-peptone medium, and total DNAs were isolated as described by Suwa et al. (34). Gel samples for pulsed-field gel electrophoresis (PFGE) were prepared by the mycelium method as described previously (20, 24), digested in gels, and separated by using contour-clamped homogeneous electric fields (CHEF). CHEF electrophoresis was carried out with 1.0% agarose gels in 0.5× Tris-borate-EDTA buffer at 15°C. E. coli strains were cultured at 37°C in Luria-Bertani medium, and cosmid and plasmid DNAs were extracted as described by Sambrook et al. (32).

Southern hybridization.

DNA fragments were separated by CHEF electrophoresis or conventional agarose gel electrophoresis and transferred to nylon membrane filters by the capillary method. Hybridization was carried out overnight at 70°C in standard buffer by using the DIG system (Roche Diagnostics GmbH, Mannheim, Germany) according to the supplier's protocol. After hybridization, the preparations were washed twice (5 min each) in 2× wash solution at room temperature and then twice (15 min each) in 0.1× wash solution at 70°C.

DNA sequencing.

Nucleotide sequences were determined by the dideoxy termination method by using a model 4200 sequencer (LI-COR, Lincoln, Nebr.) and a Thermo Sequenase cycle sequencing kit (Amersham Pharmacia Biotech, Uppsala, Sweden) or by using an ABI-373S sequencer (PE Biosystems, Foster City, Calif.), and a DYEnamic terminator cycle sequencing premix kit (Amersham Pharmacia Biotech). Genetyx-Mac 10.2 (Software Development, Tokyo, Japan) and FramePlot 2.3.2 (16; http:/www.nih.go.jp/∼jun/cgi-bin/frameplot.pl) were used for analysis of sequence data.

Annealing experiment.

Total DNAs were digested with Alw44I or BamHI, heat denatured in Tris-EDTA buffer at 100°C for 15 min, and left in a heat block to gradually reach room temperature. To digest unannealed single-stranded DNAs, mung bean nuclease (New England Biolabs, Beverly, Mass.) was added to a final concentration of 500 U/ml and the preparations were incubated at 37°C for 15 min.

RESULTS

Chromosomal deletions in mutant 301-22.

Mutant 301-22 was obtained by UV irradiation of a spore suspension of the parent strain S. griseus 2247. Mutant 301-22 has a bald appearance on solid MB medium, but it grows normally in both solid and liquid cultures. Since plasmid pSGE1, which carries the extreme end (2.7 kb) of the 2247 chromosome (9), did not hybridize to a BamHI digest of 301-22 DNA (data not shown), we deduced that both telomeres were deleted.

To determine the sizes of deletions, ordered terminal cosmids of the 2247 chromosome (25) (Fig. 1) were hybridized to the BamHI digest of the 301-22 DNA. As shown in Fig. 2A and B, mutant 301-22 had fewer hybridizing bands than strain 2247 when the preparations were probed with cosmid 10E12 on the right arm and with cosmid 4D6 on the left arm. All the cosmids external to cosmids 10E12 and 4D6 showed no hybridizing signals, while the adjacent internal cosmids, 12D3 and 12H8, showed the same hybridization pattern as strain 2247 (data not shown). Thus, the right and left deletion ends in mutant 301-22 were located on cosmids 10E12 and 4D6, and the deletion sizes were calculated to be 250 and 130 kb, respectively (Fig. 1).

FIG. 1.

Deletions (dashed lines) at the right and left ends of the chromosomes of the 301-22 mutants. Restriction and cosmid maps of both ends of the parent strain 2247 chromosome (25) provide an explanation. TIR-R and TIR-L, right and left TIRs, respectively; ORF-R and ORF-L, lipoprotein-like ORFs involved in arm replacement; Af, AflII; As, AseI; Sp, SpeI; Ss, SspI.

FIG. 2.

Southern hybridization analysis of the chromosomal deletions in the 301-22 mutants. Total DNAs of strain 2247 and the mutants were digested with BamHI, separated by conventional agarose gel electrophoresis, and hybridized with the deletion end cosmids for each mutant. λ/Hd, λ phage DNA digested with HindIII; Ba, BamHI digestion.

Chromosomal arm replacement in mutant 301-22.

The right deletion end was analyzed more precisely. A BamHI digest of the 301-22 DNA probed with right deletion end cosmid 10E12 had a new 6.6-kb fragment in place of the 9.0-kb fragment of strain 2247 (Fig. 2A). This hybridization pattern is exactly same as that of the previously described mutant MM9 with arm replacement (35). Therefore, it was hypothesized that homologous recombination between two lipoprotein-like ORFs on the right and left arms also caused chromosomal arm replacement in this mutant. To confirm this, the 6.6-kb BamHI fragment (ORF-J2) was cloned and sequenced (Fig. 3).

FIG. 3.

Nucleotide and amino acid sequences of the three lipoprotein-like ORFs located on the right and left chromosomal arms in strain 2247 (ORF-R and ORF-L) and on the fusion junction in mutant 301-22 (ORF-J2). The nucleotides of each ORF are numbered from the start codon to the stop codon. Identical nucleotides are indicated by asterisks between two sequences. Homologous recombination between ORF-R and ORF-L might have occurred in the region flanked by two arrows.

In mutant MM9, the right arm was replaced by the left arm by homologous recombination between two lipoprotein-like ORFs, ORF-R on cosmid 10E12 and ORF-L on cosmid F2B4, which resulted in the formation of long (450-kb) TIRs at both ends (35). Sequence comparison of ORF-R, ORF-L, and ORF-J2 revealed that homologous recombination occurred in mutant 301-22 and resulted in a fused ORF-J2 at the junction on the new right arm (Fig. 3). It should be noted that recombination occurred between the same ORFs in mutants 301-22 and MM9 but in different regions; it occurred between nucleotides 580 and 648 in mutant 301-22 (Fig. 3), while it occurred between nucleotides 307 and 578 in mutant MM9 (35). The fact that chromosomal arm replacement occurred between the same nonallelic ORFs in two independently isolated mutants suggests that homologous recombination is not rare in S. griseus.

Mutants generated by deletion of the long TIRs.

As described above, the BamHI digest of the 301-22 DNA showed no hybridizing signals when it was probed with the cosmids on the left arm external to cosmid 4D6. This result indicated that the two 450-kb left arms formed by arm replacement were deleted at least to this region. However, during our analysis of the deletion ends, we noticed that cosmid 4D6 did not hybridize to the 301-22 DNA. Therefore, we hypothesized that deletions of the two left arms were extended during our experiments. To study this, a macrorestriction analysis of the 301-22 mutant was performed. As shown in Fig. 4A, an SspI digest had hybridizing signals at 310, 250, 125, and 30 kb when it was probed with SspI linking cosmid 16C1 (see Fig. 1). The 250-kb fragment was the left arm fragment internal to cosmid 16C1, and the 310-kb fragment was a new internal fragment formed at the right junction by arm replacement. Thus, the 125- and 30-kb fragments might have been external fragments formed by terminal deletions. Two possibilities can be considered; one is that the 301-22 strain retained a linear chromosome with different right and left arms, and the other is that the preparation was a mixture of mutants.

FIG. 4.

Macrorestriction and hybridization analysis of the chromosomal arms in the 301-22 mutants. Total DNAs in gels of strain 2247 and the mutants were digested with SspI (Ss), separated by CHEF electrophoresis, and hybridized with the SspI linking cosmid 16C1. CHEF electrophoresis was performed at 150 V with 24-s pulses for 30 h.

To distinguish between these two possibilities, we prepared and regenerated protoplasts to isolate single colonies. As shown in Fig. 4B, SspI digests of two typical colonies contained either the 125- or 30-kb external fragment in addition to the 310- and 250-kb internal fragments. Thus, the 301-22 strain was a mixture of at least two mutants with deletions of different sizes. We designated the mutant with the 125-kb fragment 301-22-L, and we designated the mutant with the 30-kb fragment 301-22-M. In addition, we obtained one mutant with additional deletions (301-22-S) and another mutant with DNA amplification (301-22-Amp), both of which gave no hybridizing signals when they were probed with cosmid 16C1. Therefore, in these mutants, there were chromosomal deletions beyond cosmid 16C1. However, all four mutants were confirmed to have the chromosomal arm replacement.

Large palindromic structure in mutants 301-22-L and 301-22-M.

We first analyzed mutant 301-22-L with the 125-kb external fragment. Southern hybridization analysis of the 301-22-L DNA by using the ordered terminal cosmids revealed that the deletion end was located on cosmid 17E10 (Fig. 2C). A new 15-kb BamHI fragment appeared in place of an 8.0-kb fragment. Finally, the deletion end was located on a 2.0-kb BamHI-XhoI fragment of cosmid 17E10 (Fig. 5A). Thus, in this mutant there were approximately 100-kb terminal deletions from cosmid 4D6 to cosmid 17E10 (Fig. 1).

FIG. 5.

Analysis of the deletion ends in mutants 301-22-L (A to C) and 301-22-M (D to F) by restriction, self-annealing, and hybridization. (A and D) Comparison of the restriction maps at the deletion ends of the mutants and the corresponding regions of strain 2247. (B and E) Southern hybridization analysis of restriction sites. (C and F) Self-annealing of the extreme end fragments. Al, Alw44I; Ba, BamHI; Bg, BglII; Kp, KpnI; Ps, PstI; Xh, XhoI; Ec, EcoRI; Al-1 and Ba-1, digested with Alw44I and BamHI, respectively; Al-2 and Ba-2, heat denatured and renatured after digestion; Al-3, digested with mung bean nuclease after renaturation.

To determine what happened at the deletion end in mutant 301-22-L, total DNA was digested with Alw44I, KpnI, and XhoI, separated by conventional agarose gel electrophoresis, and hybridized with probe 1 (Fig. 5B). Surprisingly, all of the recognition sites for Alw44I, KpnI, and XhoI were symmetrical with the center at the deletion end (Fig. 5A). This result suggested that a palindromic structure was formed at the deletion end.

This hypothesis was proved by performing self-annealing experiments. Total DNAs of strains 2247 and 301-22-L were digested with Alw44I, heat denatured, renatured, and subjected to Southern hybridization analysis. As shown in Fig. 5C, heat denaturation and renaturation changed the size of the 2.0-kb terminal Alw44I fragment (fragment L/Al-1) to 1.0 kb (fragment L/Al-2). The 1.0-kb fragment that was generated was resistant to mung bean nuclease digestion, while a broad band at 0.7 kb disappeared (fragment L/Al-3). Therefore, we concluded that the 1.0-kb fragment was formed by self-annealing of the 2.0-kb palindromic terminal fragment and that the 0.7-kb band was due to unannealed single-stranded DNA.

Mutant 301-22-M had the 30-kb external SspI fragment, suggesting that the deletion end was present on linking cosmid 16C1 itself. This hypothesis was proved by Southern hybridization of the BamHI digest probed with cosmid 16C1 (Fig. 2D). The deletion end was located on a 0.85-kb EcoRI-BamHI fragment of cosmid 16C1 (Fig. 5D). Southern hybridization analysis (Fig. 5E) also revealed symmetrical locations of restriction sites for BglII, KpnI, and PstI around the deletion end (Fig. 5D). Heat denaturation and renaturation converted the 1.0-kb BamHI fragment at the deletion end (Fig. 5F, lane M/Ba-1) to a 0.5-kb fragment (lane M/Ba-2), confirming that a palindromic structure was also formed in this mutant.

The large palindromic structure was also supported by the results of physical mapping of cosmids 17E10 and 16C1, which carried the deletion end of each mutant. As shown in Fig. 6A, the deletion ends were located 63 and 15 kb from the SspI site in mutants 301-22-L and 301-22-M, respectively. The calculated sizes of the large palindromic fragments (63 kb × 2 = 126 kb and 15 kb × 2 = 30 kb) agree with the observed sizes for mutants 301-22-L and 301-22-M (125 and 30 kb) (Fig. 4B). The macrorestriction patterns of undeleted parts of the mutant chromosomes are identical to the pattern of parent strain 2247 (data not shown).

FIG. 6.

(A) Locations of the deletion ends on the physical map of cosmids 17E10 and 16C1. (B) Open and closed racket frame structures of the 2247 and mutant chromosomes. The right and left arm regions are shown enlarged in panel B, and the sizes (in kilobases) of important SspI fragments are indicated. As, AseI; Ec, EcoRI; Ss, SspI; Xh, XhoI; ORF-R, ORF-L, ORF-J, and ORF-J2, lipoprotein-like ORFs involved in and formed by arm replacement.

The chromosomal structures of mutants 301-22-L and 301-22-M, together with those of parent strain 2247 and mutant MM9 with arm replacement, are indicated by the racket frame structures in Fig. 6B. A racket frame structure was proposed previously for linear plasmid pSLA2 (11), in which the TIR sequences at both ends were speculated to be juxtaposed by the interaction of terminal proteins and other binding proteins. Since the ends of the racket frame are closed in mutants 301-22-L and 301-22-M, we called this unique structure a closed racket frame. We previously constructed an AseI restriction map for the 2247 chromosome. The expected fused AseI fragment of the 301-22 mutants, which contained both an extremely large palindrome and the flanking right and left arms, did not enter PFGE gels (data not shown), perhaps due to its unusual structure. Instead, the SspI fragment patterns completely agreed with the deduced structures (Fig. 4B).

Possible mechanism of formation of the large palindromic structure.

Qin and Cohen (30) reported formation of palindromic linear plasmids from defective pSLA2 plasmids which had a deletion at one of the two telomeres. In these plasmids, the defective side was completely deleted, and the intact side was duplicated symmetrically. Nucleotide sequencing revealed that short palindromic sequences were originally present in pSLA2, which formed the center of the palindrome in the mutants. Based on these results, Qin and Cohen proposed a formation mechanism: the 3′ overhang at the deletion end forms a hairpin structure and is replicated in a strand-displacing manner until the 3′ end forms a large hairpin structure, and complete duplication of this large single-stranded hairpin DNA produces a palindromic linear plasmid.

To study a similar possibility, we tried to clone the center of the palindrome of mutants 301-22-L and 301-22-M using several combinations of cloning vectors and hosts, including pUC19, pBR322, pACYC184, M13mp19, and E. coli XL1Blue and SURE2. However, in accordance with the findings of Kieser and Melton (18), we found in all cases that plasmids carrying long uninterrupted perfect palindromes were not viable. We also tried PCR amplification and direct sequencing by using the DNA fraction extracted from agarose gels, but we were not successful with these methods either. Thus, we determined the nucleotide sequences of the 2247 chromosome which correspond to the deletion ends in mutants 301-22-L and 301-22-M, the 1.4-kb BamHI-Alw44I fragment (Fig. 5A) and the 0.85-kb EcoRI-BamHI fragment (Fig. 5D).

As shown in Fig. 7A, two short palindromic sequences were identified for mutant 301-22-L. Palindrome 1 contains 13-bp palindromic sequences with an 8-bp core, and palindrome 2 contains 14-bp sequences with a 9-bp core. One of these palindromes, palindrome 2, might have contributed to the formation of mutant 301-22-L, because the distance between the center of the palindrome and the Alw44I site (1,022 bp) (not shown in Fig. 7A) agrees with the observed size of the annealed fragment (1.0 kb) (Fig. 5C). Furthermore, digestion with BsrI, which had a recognition site between palindromes 1 and 2, did not affect formation of the 2.0-kb palindromic Alw44I fragment. Palindrome 3 with 24-bp sequences was found for mutant 301-22-M (Fig. 7B). The distance between the center of this palindrome and the BamHI site (521 bp) (not included in Fig. 7B) also agrees with the size of the observed annealed fragment (0.5 kb) (Fig. 5F).

FIG. 7.

Nucleotide sequences of the 2247 DNA which correspond to the chromosomal deletion ends of mutants 301-22-L (A) and 301-22-M (B). Two and one palindromic sequences were originally present near the deletion ends of mutants 301-22-L and 301-22-M, respectively.

The mechanism proposed by Qin and Cohen (30) could not be directly applied to the 301-22-L and 301-22-M chromosomes, because they are circular in spite of containing a large palindromic structure. A possible formation mechanism is shown in Fig. 8, which shows that there may be a recBCD-like homologous recombination system (6, 33). A hairpin loop, which is formed at the 3′ deletion end of the long TIR of a mutant with arm replacement (Fig. 8b), invades the deletion end of the opposite TIR and is ligated (Fig. 8c). The complementary strand of the invaded strand is replicated by using the hairpin DNA as a template (Fig. 8d), reaches the deletion end of the first TIR strand, and is finally ligated to give a large palindromic structure (Fig. 8e).

FIG. 8.

Possible mechanism of formation of the 301-22-L and 301-22-M chromosomes.

DISCUSSION

In this study, we analyzed the structures of the deleted chromosomes of S. griseus 301-22-L and 301-22-M and discovered a new type of chromosomal structure, a circularized chromosome with a large palindrome. Previously, large palindromic structures have been found only in linear replicons in Streptomyces. Before large palindromic plasmids were found by Qin and Cohen (30), we isolated pSLA2-L1, a deletion derivative of pSLA2-L, in Streptomyces rochei mutant KE32 (22). pSLA2-L1 may also belong this group of linear plasmids. Recently, Wenner et al. (38) reported end-to-end fusion of two deleted chromosomes in an inverted orientation in S. ambofaciens. The chromosomes generated did not have a complete palindromic structure, because they contained a unique sequence at the junction of two fused chromosomes. Therefore, this study may be the first report of a large palindromic structure in circular replicons.

The fact that the 301-22 mutants grow normally indicates that large palindromic sequences in the circularized chromosomes are replicated without difficulty in S. griseus. This is very surprising because one would not expect that bacteria can replicate such palindromes. Taken together with the fact that palindromic linear plasmids were found in Streptomyces, it was suggested that the replication machinery of this genus is somewhat different from that of other bacteria. However, careful analysis of Fig. 4B revealed that there are still deletions in large palindromic structures. There were several faint hybridizing signals in the SspI digests of both parent strain 2247 and mutant 301-22-L. Some of these signals seemed to be identical to each other based on their sizes. These signals may have been due to fragments of additional mutants, which were generated by progressive deletions and were present at low levels in the sample. Thus, we may have observed a process of chromosomal deletion in which parent strain 2247 was converted to mutants 301-22-L and 301-22-M.

Simple circularization (15, 17) and arm replacement (35) of the S. griseus linear chromosome have been reported previously. Qin and Cohen (31) analyzed similar structural changes of pSLA2 derivatives and discussed the strategies of Streptomyces linear replicons after telomere damage. In addition, they reported that an intact telomere was reproduced by recombination between linear plasmids and between a linear chromosome and a linear plasmid. The latter phenomenon has also been reported for Streptomyces rimosus (27) and Streptomyces lividans (13), and it caused exchange of the ends of the chromosome and a plasmid. Thus, Strepomyces linear chromosomes and plasmids have two strategies to deal with telomere deletions: (i) recovery of a deleted telomere by intramolecular or intermolecular recombination, when one of the two telomeres is deleted; and (ii) circularization by nonhomologous recombination of deletion ends, when both telomeres are deleted.

In addition to these two strategies, Streptomyces linear chromosomes display DNA amplification after deletion. Birch et al. (4) found complex rearrangements which occurred adjacent to DNA amplification regions in deletion mutants of S. lividans. Recently, Bao and Cohen (2) reported that disruption of the tap (telomere-associated protein) gene caused chromosomal circularization, as well as amplification, in S. lividans. They found an insertion of adventitious DNA at the junction of amplified DNA. Thus, complex structural changes that occur in amplified mutants have not been clarified well. We speculate that DNA amplification may be an intermediate state before the final stable state, such as arm replacement and circularization.

Accumulated data on dynamic structural changes of Streptomyces linear replicons demonstrated that this genus has extraordinary strategies to keep a functional replicon, whether it is linear or circular and even if it contains an extremely large palindromic structure. Similar studies should provide further insight into the evolution of chromosomes in general.