Molecular Microbiology (1998) 28(5), 905–916
The telomeres of Streptomyces chromosomes contain
conserved palindromic sequences with potential to
form complex secondary structures
Chih-Hung Huang, Yi-Shing Lin,† Ya-Ling Yang,‡
Shu-wen Huang and Carton W. Chen*
Institute of Genetics, National Yang-Ming University,
Shih-Pai, Taipei 112, Taiwan.
Summary
The chromosomes of the Gram-positive soil bacteria
Streptomyces are linear DNA molecules, usually of
about 8 Mb, containing a centrally located origin of
replication and covalently bound terminal proteins
(which are presumably involved in the completion of
replication of the telomeres). The ends of the chromosomes contain inverted repeats of variable lengths.
The terminal segments of five Streptomyces chromosomes and plasmids were cloned and sequenced. The
sequences showed a high degree of conservation in
the first 166–168 bp. Beyond the terminal homology,
the sequences diverged and did not generally crosshybridize. The homologous regions contained seven
palindromes with a few nucleotide differences. Many
of these differences occur in complementary pairs,
such that the palindromicity is preserved. Energyoptimized modelling predicted that the 38 strand of
the terminal palindromes can form extensive hairpin
structures that are similar to the 38 ends of autonomous parvovirus genomes. Most of the putative hairpins have a GCGCAGC sequence at the loop, with the
potential to form a stable single C-residue loop closed
by a sheared G:A pairing. The similarity between
the terminal structures of the Streptomyces replicons
and the autonomous parvoviral genomes suggests
that they may share some structural and/or replication features.
Introduction
The linearity of chromosomes in the streptomycetes was
first discovered in Streptomyces lividans 66 (Lin et al .,
1993). Pulsed-field gel electrophoresis (PFGE) also
Received 11 December, 1997; revised 28 February, 1998; accepted
5 March, 1998. Present addresses: †Food Industry Research and
Development Institute, Hsin-Chu, Taiwan; ‡Pathology and Clinical
Laboratory, Veterans General Hospital, Taipei, Taiwan. *For correspondence. E-mail [email protected]; Tel. (2) 2826 7040; Fax (2)
2826 4930.
Q 1998 Blackwell Science Ltd
revealed linear chromosomes in six other species: S. moderatus ATCC 23443, S. parvulus ATCC 12434, S. rochei
7434-AN4, S. antibioticus IMRU 3720, S. coelicolor
A3(2) strain M130 and S. lipmanii ATCC 27357. These
results prompted Lin et al . (1993) to propose that chromosomal linearity was common among the streptomycetes,
and this proposal has been supported by the characterization of linear chromosomes in other species (Gravius et al .,
1994; Lezhava et al ., 1995; Leblond et al ., 1996; Pandza
et al ., 1997). Most Streptomyces chromosomal DNA molecules are about 8 Mb long, with terminal inverted repeats
(TIRs) and covalently bound terminal proteins (TPs) supposedly at the 58 ends. The sizes of the TIRs range widely,
from 24 to 550 kb (Lin et al ., 1993; Lezhava et al ., 1995;
Leblond et al ., 1996; Pandza et al ., 1997). The terminal
regions of the Streptomyces chromosomes correspond
to the so-called ‘unstable regions’, in which extensive deletions and amplifications frequently occur (for reviews, see
Cullum et al ., 1988; Schrempf et al ., 1989; Birch et al .,
1990; Leblond and Decaris, 1994; Chen, 1995).
The location of the replication origin (oriC ) in the centre
of the Streptomyces chromosomes (Calcutt and Schmidt,
1992; Kieser et al ., 1992; Zakrzewska-Czerwinska and
Schrempf, 1992) indicates that replication proceeds
bidirectionally towards the telomeres, consistent with the
observation of Musialowski et al . (1994). This internally
initiated replication would leave single-strand gaps on the
38 strands at the termini, which are patched by an unknown
mechanism supposedly involving the TPs (Chang and
Cohen, 1994).
In the most distal 120 bp sequence of the S. lividans
chromosome, Lin et al . (1993) found four palindromes,
each consisting of a 13–25 bp inverted repeat flanking a
3 bp gap. Palindrome I starts at the first basepair of the
chromosome and is complementary to a 13 bp subset of
palindrome IV, thus forming a superpalindrome. The published terminal sequences of three linear plasmids of Streptomyces also contain abundant palindromic sequences
(see also Discussion ). The terminal sequence of the S. lividans chromosome is identical to that of the right terminus
of the linear plasmid SLP2 (Chen et al ., 1993), in agreement with the observed homology of 15.3 kb between
these two segments (Lin et al ., 1993).
To understand the structure and function of the Streptomyces telomeres, we isolated and sequenced terminal
906 C.-H. Huang et al.
Fig. 1. Identification of putative terminal Ase I
fragments in various Streptomyces species. Left:
PFGE of Ase I-digested genomic DNA. Genomic
DNA of each species was isolated in agar plugs
after proteinase K treatment and digested with
Ase I. The Ase I digests were separated by PFGE
and stained with ethidium bromide. The sizes (in
kb) of the Ase I fragments of S. lividans ZX7
(Kieser et al ., 1992) are indicated. Right:
Hybridization of the Ase I fragments with the
2.6 kb terminal DNA of the S. lividans
chromosome (Lin et al ., 1993).
DNA from four other Streptomyces species, as described
here. We found that these nucleotide sequences are well
conserved in the first 166–168 nucleotides, which consist
of seven palindromes. The 38 strands of these palindromes
have the potential to form extensive secondary structures
that resemble the 38 ends of the autonomous parvovirus
genomes.
Results
Lack of long sequence homology among the
Streptomyces termini
Using PFGE analysis of undigested chromosomal DNA,
we found linear chromosomes in four more species: S.
argenteolus ATCC 11009, S. clavuligerus ATCC 27064,
S. hydrogenans ATCC 19631 and S. rimosus R6 strain
539 (Yang, 1994). The sizes of the chromosomes in these
four species were similar to that of S. lividans chromosomes (8 Mb), except for that of the S. clavuligerus chromosome, which was about 6.5 Mb. These chromosomes
also appeared to contain TP.
Two DNA clones spanning the terminal regions of the S.
lividans chromosome were used to detect possible sequence homology in seven other species: S. argenteolus,
S. clavuligerus , S. hydrogenans , S. rimosus , Streptomyces
rochei , Streptomyces lipmanii and Streptomyces parvulus .
The Ase I digests of total DNA from these species were
hybridized with the terminal 2.6 kb Bgl II fragment (cloned
in pLUS450) of the S. lividans chromosome (Fig. 1). The
two terminal fragments (2.6 Mb and 190 kb) of the S. lividans ZX7 chromosome hybridized as expected. Out of
all the other species, only two fragments (about 1.5 Mb
and 400 kb) in S. lipmanii exhibited significant hybridization. A low level of hybridization was also detected with a
700 kb fragment in S. clavuligerus (and possibly a small
fragment in S. argenteolus), indicating a relatively low or
short homology. No hybridization was detected to the
Ase I digests of the other four species.
Next, cosmid 3B-11 containing a 40 kb insert spanning
the region next to the 2.6 kb Bgl II terminal fragment of S.
lividans was used to detect homologous sequences in
the seven species. Essentially, no significant hybridization
was detected (data not shown). These results indicated
that the chromosomal termini of most Streptomyces
species did not share extensive homology in the first
tens of kilobases spanning the TIR.
Cloning of terminal DNA fragments
The absence of extensive homology among the terminal
sequences of chromosomes from various Streptomyces
species did not preclude the possibility that short stretches
of homology at the termini had escaped detection. To
investigate this possibility, we attempted to clone terminal
DNA from these species.
The glass-binding procedure of Lin et al . (1993) was
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Terminal sequences of Streptomyces chromosomes and plasmids 907
Fig. 2. Strategy for cloning the terminal DNA.
Lysozyme-treated mycelium was lysed without
proteolytic treatment. Total DNA was purified
by ethidium bromide–CsCl gradient
centrifugation and digested with Pst I
(abbreviated ‘P’) or another enzyme that cut
the terminal DNA into suitably sized
fragments. Fragments containing the TP were
purified by binding to glass beads in the
presence of 0.3 M NaCl and eluted by SDS
(Coombs and Pearson, 1978; Thomas et al .,
1979). The TP–DNA complex was incubated
in 10% piperidine at 558C for 1 h (Peñalva
and Salas, 1982). The released DNA was
purified with a GeneClean II kit and ligated
with pBluescript-II vector DNA predigested
with Pst I plus an enzyme (typically Hin cII,
abbreviated ‘H’) that gave a blunt end. The
recombinant plasmids were purified from the
transformants and analysed by restriction
digestion. Those that contained the terminal
DNA were selected based on three criteria:
(i) Pst I digestion should cut uniquely, giving
rise to a single linear DNA species, so the
absence of a Pst I site or the presence of
more than one Pst I sites would indicate that
deletions or multiple insertions had occurred
during cloning; (ii) Pst I plus a restriction
enzyme cutting at the multiple cloning site on
the other side of the insert (e.g. Kpn I,
abbreviated ‘K’) should release a fragment of
the size of the vector and one or more
fragments representing the insert, so the
absence of the vector-sized fragment would
also indicate rearrangements during cloning;
(iii) the insert should hybridize to a genomic
Pst I fragment of the same size. Only
recombinant plasmids that fulfilled these
criteria were subjected to further studies.
modified and used to isolate restriction fragments containing covalently bound TP. Some early attempts encountered deletions of the cloned DNA and the adjacent
vector sequences (Yang, 1994). A more systematic procedure for the cloning of intact terminal DNA was developed
(Fig. 2), which incorporated two modifications. First, piperidine was used instead of proteinase K to remove TPs
(Peñalva and Salas, 1982). Proteinase K treatment, which
was used in this step previously (Lin et al ., 1993), would
leave peptide fragments on the 58 ends of DNA. Secondly,
the recombinant plasmids that contained putative terminal
fragments were prescreened for the expected restriction
patterns before sequence determination. The details of
the procedure are described in Fig. 2.
Four putative terminal Pst I fragments were cloned from
three species: a 3.5 kb and a 0.9 kb fragment from S. parvulus ; a 2.1 kb fragment from S. clavuligerus; and a 1.5 kb
fragment from S. lipmanii. No putative terminal Pst I fragment was cloned from S. coelicolor after several attempts,
but a 1.45 kb putative terminal Bam HI fragment was
cloned instead. Altogether, five putative terminal DNA
fragments were isolated from four species.
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Location of the terminal fragments
These putative terminal DNA fragments were used to
hybridized to PFGE-separated intact genomic DNA to
determine whether they were chromosomally or plasmid
borne (Fig. 3). The strain of S. coelicolor (M145) from
which the putative terminal DNA was cloned does not contain any free plasmid (Kieser et al ., 1992) and, therefore,
was not tested. For S. clavuligerus, the 2.1 kb putative
terminal fragment hybridized to a 120 kb linear plasmid,
pSCL2 (Netolitzky et al ., 1995), and not to the chromosomal DNA (Fig. 3A). For S. lipmanii , the putative terminal
fragment hybridized to the 8 Mb chromosome (Fig. 3B).
For S. parvulus , the 0.9 kb putative terminal Pst I fragment
hybridized only to the chromosomal DNA, which appeared
to be about 6 Mb in size and significantly shorter than the
other Streptomyces chromosomes (Fig. 3C). In contrast,
the 3.5 kb putative Pst I fragment hybridized to the
650 kb linear plasmid pSPA1 in S. parvulus (Fig. 3D).
In the absence of proteinase K treatment, the plasmid
and chromosome DNA molecules that hybridized to these
putative terminal DNA fragments were retained at the origin
908 C.-H. Huang et al.
Fig. 3. Determination of the chromosomal versus extrachromosomal locations of the putative terminal DNA sequences. For each species,
intact DNA prepared in the presence (þ) or in the absence (¹) of proteinase K (PK) treatment from the culture was separated by PFGE (left)
and hybridized with the DNA clone isolated from that species (right).
A. S. clavuligerus . The 430 kb pSCL3 and the 120 kb pSCL2 plasmids are indicated. The 12 kb pSCL1 was not separated from nucleic aids of
similar size. Hybridization probe: the terminal 2.1 kb Pst I fragment.
B. S. lipmanii . Hybridization probe: the terminal 1.5 kb Pst I fragment.
C. S. parvulus . The chromosomal DNA was estimated to be about 6 Mb in comparison with S. pombe and other Streptomyces chromosomes.
The three linear plasmids were not separated from one another. Hybridization probe: the terminal 0.9 kb Pst I fragment.
D. S. parvulus . Hybridization probe: the terminal 3.5 kb Pst I fragment.
during electrophoresis (Fig. 3A–C). This was in agreement with the presence of TP on these replicons.
The putative terminal fragments were mapped further to
separate Ase I restriction fragments (Fig. 4). The cloned
2.1 kb putative terminal fragment of S. clavuligerus hybridized to a 120 kb Ase I fragment of S. clavuligerus DNA (Fig.
4A). This DNA corresponded to the pSCL2 plasmid present in the undigested genomic DNA (Fig. 3A). Therefore,
this plasmid appeared to lack Ase I sites, and the putative
terminal DNA was probably present on one or both ends
of it.
The 1.45 kb putative terminal fragment from S. coelicolor
hybridized to the 1.5 Mb Ase I-A fragment and the 190 kb
Ase I-J fragments (Fig. 4B), which had been shown to contain the telomeres of the chromosome (Kieser et al ., 1992;
Lin et al., 1993). The unexpected hybridization of the probe
to the 820 kb Ase I E-fragment was presumably caused by
the presence of repeat sequences (e.g. transposable elements). Terminally located transposable elements have
been found in the Streptomyces chromosome (Lin et al .,
1993; Fischer et al ., 1996; Lin and Chen, 1997).
The 1.5 kb putative terminal Pst I fragment of S. lipmanii
hybridized to a 1.1 Mb and a 370 kb Ase I fragment (Fig.
4C). As the cloned fragment also hybridized to the intact
chromosome (Fig. 3), these two Ase I fragments presumably represented the two terminal fragments of the chromosome that hybridized to the terminal probe of S. lividans
(Fig. 1).
The 0.9 kb Pst I putative terminal fragment of S. parvulus
hybridized to a 1.2 Mb and a 420 kb Ase I fragment (Fig.
4D). These two fragments presumably contained the terminal inverted repeats. Hybridization to Ase I fragments
by the 3.5 kb Pst I putative terminal fragment from the
pSPA1 plasmid was not analysed.
The presence of TP on pSCL2 (Fig. 4A) and on the
two terminal Ase I fragments of S. coelicolor (Fig. 4B)
was demonstrated by retardation of electrophoretic mobility when proteinase K treatment was omitted.
Conserved sequences of the terminal DNA
The five putative terminal DNA fragments on these clones
were subcloned. Attempts to determine the terminal sequences by the dideoxy chain termination method of Sanger
et al . (1977) encountered several problems (such as band
compressions, non-specific premature termination and
high background signals) presumably caused by the presence of extraordinarily stable hairpin structures (Pandza
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Terminal sequences of Streptomyces chromosomes and plasmids 909
Fig. 4. Locations of the putative terminal DNA on the Ase I fragments. For each species, total DNA prepared in the presence (þ) or in the
absence (¹) of proteinase K (PK) treatment was digested with Ase I and separated by PFGE (left). The Ase I fragments were hybridized with
the DNA clone isolated from that species (right). The size (in kb) of the hybridizing fragments are shown on the right.
et al ., 1997). These difficulties were overcome by incorporating additional procedures to counter the formation
of secondary structures (Huang, 1997; see Experimental
procedures ). After repeated attempts, about the first 200
nucleotides from each of the five terminal fragments
were determined. The first 180 nucleotides of these terminal sequences are aligned (Fig. 5). The 120 nucleotide
terminal sequence of the S. lividans chromosome, determined previously (Lin et al ., 1993), was extended to 180
nucleotides and included in the comparison.
As seen in Fig. 5, there is a strong homology (yellow
blocks) for the first 166–168 nucleotides among these
sequences. This terminal homology contains seven palindromic sequences (I–VII), each consisting of a pair of
inverted repeats (of at least 5 bp) flanking a 3 bp gap.
Palindrome I begins at the first basepair of the terminal
sequence and is conserved in all the terminal sequences.
Palindrome II follows after three nucleotides. Palindromes
III and IV adjoin each other without gaps, as do palindromes V and VI. Palindrome VII ends at nucleotide 166
in the S. lipmanii chromosome, nucleotide 168 in pSPA1
plasmid and nucleotide 167 in the other four terminal
sequences. The end of palindrome VII also marks the end
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
of the terminal homology. Beyond this point, essentially
no general homology exists, consistent with the lack of
hybridization between the terminal fragment of S. lividans
and those of the others, except S. lipmanii .
A few nucleotide substitutions are present in the seven
conserved terminal palindromes. No divergence was observed in palindrome I. In palindromes II and III, all the nucleotide substitutions occur in such a way that the palindromic
nature is preserved: a nucleotide substitution in one repeat
is accompanied by a complementary substitution in the
other. On the other hand, several nucleotide substitutions
in palindromes IV, V and VI are not complementary, thus
resulting in mismatches (indicated by asterisks). In palindrome VII, the inverted repeats are interrupted by a pair
of trinucleotide TCC direct repeats (shaded in green). A
nucleotide substitution was present in one of the trinucleotide repeats in the S. coelicolor chromosomal sequence.
As noted by Lin et al . (1993), the palindrome I sequences form a superpalindrome with the central 13 bp of
palindrome IV (thick green arrows) in all the terminal
sequences determined in this study. However, a mismatch
was present in the superpalindrome of the S. parvulus
chromosome.
910 C.-H. Huang et al.
Fig. 5. Comparison of the cloned terminal sequences. The terminal 180 nucleotide sequences from the nine Streptomyces chromosomes
and linear plasmids are compiled and aligned. Palindromic sequences are boxed and indicated by the double arrows above the aligned
sequences. Each palindrome is numbered in roman numerals above the sequences. Within each palindrome, asymmetrical nucleotide pairs
are marked by asterisks below, and the gap between the inverted repeats is shaded in red and indicated by the red bar above the aligned
sequences. In palindrome VII, the trinucleotide repeats that interrupt the inverted repeats are shaded in blue. The conserved nucleotide in the
regions spanned by the seven terminal palindromes in the first seven sequences and the first three palindromes in pSCL1 and pSLA2 are
shaded in yellow. Unique exceptions to the consensus are shaded in white. The thick green divergent arrows above palindrome I and the
middle 13 nucleotides of palindrome IV denote the inverted repeats formed by these two sequences. The thick green arrows below pSCL1
and pSLA2 represent imperfect repeats formed between palindrome I and internal sequences (with mismatches indicated by white asterisks).
The positions of the nucleotide at the end of each line are indicated by the rightmost numbers. The GenBank accession numbers for the
sequences determined in this study are: X77096 (S. lividans chromosome), AF038453 (S. coelicolor chromosome), AF038454 (S. parvulus
chromosome), AF038455 (pSPA1 plasmid), AF038456 (S. lipmanii chromosome) and AF038457 (pSCL2).
Comparison with reported terminal sequences of
linear Streptomyces plasmids
Despite the high conservation of the terminal sequences
determined in this study, they showed significantly less
resemblance to the terminal sequences of three linear
plasmids, pSCL1, pSLA2 and SCP1, determined previously. The terminal sequences of pSCL1 and pSLA2
(Z. Qin and S. N. Cohen, 1998) contain only the conserved
palindromes I–III (Fig. 5), lacking the canonical palindromes IV–VII. The hypothetical pairing of the canonical
palindromes I and IV is therefore not possible. However,
there are one or two internal palindromic sequences with
the potential to form imperfect inverted repeats with palindrome I. Palindrome III in these two terminal sequences
varies slightly in size from the canonical version, and yet
the GCA centre in the 38 strand is preserved. In addition
to palindromes I–III, two and four more palindromes are
present in the first 180 nucleotides of pSCL1 and pSLA2
respectively.
The terminal sequence of SCP1 (Kinashi et al ., 1991) has
even less homology to all the other terminal sequences. It
starts with a string of 4–6 Gs (varying from clone to clone),
followed by a nine-nucleotide sequence, eight nucleotides
of which are identical to part of palindrome I (shown by
the underlined nucleotide at the beginning of the SCP1
sequence G4–6 GCGGAGAGGCCTAACGGCC). Outside
of this stretch, no significant homology exists for the next
179 nucleotides, consistent with the lack of detectable
hybridization between the cloned terminal fragment of
SCP1 and the S. coelicolor chromosome (H. Kinashi, personal communication). While many palindromic sequences
are present in the terminal sequence of SCP1, they are not
as concentrated at the terminus as the terminal sequences determined here. Not only are their predicted secondary structures, based on the DNA Fold algorithm,
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Terminal sequences of Streptomyces chromosomes and plasmids 911
more interspersed, but the hairpin they form contains
mostly tetranucleotide (instead of trinucleotide) loop
sequences (data not shown).
Discussion
The five clones isolated in this study should represent true
terminal DNA based on the following observations: (i) they
were purified by the glass binding procedure; (ii) these
fragments and the genomic DNA they hybridized exhibited
the same restriction patterns (data not shown); (iii) in the
absence of proteolytic treatment, much of the genomic
DNA that hybridized to these DNA fragments was retarded
during electrophoresis, indicating the presence of TP; (iv)
their sequences share extensive homology to each other
and to the terminal sequences of the S. lividans chromosome and the right end of SLP2 determined previously
(Chen et al ., 1993; Lin et al ., 1993).
Despite the homology in approximately the first 170 bp
among the different Streptomyces chromosomes, significant hybridization to the terminal DNA probes (spanning
the TIRs) of the S. lividans chromosome was seen only
in S. lipmanii and, to a lesser degree, in S. parvulus among
the seven species tested. Even the closely related S. coelicolor A3(2) does not contain DNA hybridizing to the terminal probes of S. lividans (Wang, 1996).
In S. lividans , the chromosomal termini share a 15.3 kb
homology with the right end of the linear plasmid SLP2
(Chen et al ., 1993; Lin et al ., 1993). S. clavuligerus harbours three linear plasmids (Netolitzky et al ., 1995), of
which only the largest (pSCL3) exhibited extensive hybridization to the chromosomal DNA (Netolitzky et al ., 1995).
In the other cases examined, no hybridization was detected
between the terminal DNA of the linear plasmids and the
host chromosome: those in S. griseus NRRL 3851 and
S. jumonjinesis NRRL 5741 (Netolitzky et al ., 1995), SCP1
in S. coelicolor (H. Kinashi, personal communication)
and pSPA1 in S. parvulus (Fig. 3C and D). Thus,
exchanges between the linear plasmids and the chromosomes apparently do not occur frequently.
The persistent palindromicity at the Streptomyces telomeres suggests that the secondary structures of these
sequences may play an important biological role(s). It is
unlikely that they serve solely as binding signals for specific
proteins, because these protein-binding inverted repeats
usually afford a greater degree of freedom from strict
palindromicity.
Chen (1996) proposed the involvement of such secondary structures in the patching of the single-stranded gaps
at the 38 termini during replication. In this model, palindrome I on the single-stranded 38 end pairs with the complementary sequence in palindrome IV, while palindromes
II and III fold back on themselves, which helps to bring
palindromes I and IV together. These predicted secondary
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
structures were confirmed by the energy minimization
algorithm of Zuker (1989) based on the free energy of
SantaLucia et al . (1996). Figure 6 shows two representatives of such predicted secondary structures on the S. lividans chromosome (Fig. 6A) and on the pSCL2 plasmid of
S. clavuligerus (Fig. 6B). In both examples, a Y-shaped
duplex is formed by palindromes I–IV. Three additional
hairpins are formed by palindromes V–VII. Each of the
five predicted hairpins contains a trinucleotide loop. A single
nucleotide bulge is present between palindromes I and
II and in the stem formed by palindrome III. Furthermore,
a GGA:GGA bubble is present in the fifth hairpin formed
by palindrome VII. The predicted secondary structures of
the 38 strand of the other terminal palindromic sequences
determined in this study exhibited similar overall structures.
The predicted Y-shaped secondary structures at the
Streptomyces telomeres bear a striking resemblance to
the 38 end of the genomes of the autonomous (helper-independent) parvoviruses (Fig. 6C), which are linear singlestranded DNA molecules flanked by stable duplex hairpins
at both ends. At the 38 end, 115–116 highly conserved
nucleotides form a Y-shaped structure consisting of two
hairpins (with characteristic bulges and gaps) and a long
stem (Astell et al ., 1979). Similar (but larger) Y-shaped
structures are also seen in the predicted secondary structures of the Streptomyces telomeres. Most noteworthy is
the appearance of the first 38 hairpin of the parvovirus
genome as part of the second 38 hairpin in the predicted
secondary structures of the Streptomyces telomeres (corresponding to palindrome III; Fig. 6A and B). Subsets of
this hairpin are also present at the other hairpins in the predicted structures (corresponding to palindromes II, V, VI
and VII).
Based on 13C-NMR studies (Zhu et al ., 1995a), Chou et
al . (1997) proposed that the GCA loop in this hairpin of the
parvovirus genome was closed by ‘sheared purine–purine
pairing’ formed by the G and A residues (Fig. 6C), resulting in a single C-residue loop. Interestingly, of all the hairpin structures closed by PyGNAPu, the PyGCAPu hairpin
is the most stable variant (Chou et al ., 1996) and represents 85% of predicted hairpins in the Streptomyces telomeres determined in this study (Fig. 5).
The second 38 hairpin of the autonomous parvovirus
genome also has a potential to form a stable single A-residue loop closed by a G:A sheared pair (Chou et al ., 1997).
These single-nucleotide hairpin loops are stable at room
temperature and have been implicated in causing ‘band
compression’ and premature chain termination in DNA
sequencing (Yamakawa et al ., 1996), as we experienced
in this study. These 38 terminal hairpins play an essential
role in the replication of the autonomous parvoviral genomes. The extraordinary abundance of these tight-turn
hairpins stabilized by sheared Pu:Pu pairs and their biased
presence in the terminal 38 strand at the Streptomyces
912 C.-H. Huang et al.
Fig. 6. Predicated secondary structures of
the terminal sequences.
A. S. lividans chromosome (and right
terminus of SLP2 plasmid).
B. Plasmid pSCL2.
C. The autonomous parvoviral genome 38
end. For the two Streptomyces sequences,
the secondary structures formed by 38 singlestranded DNA spanning the seven terminal
palindromes with the lowest energy level was
predicted by the DNA Fold server. The roman
numerals indicate the palindrome numbers
(Fig. 5). The position of the first nucleotide is
indicated by the arrow. The energy levels of
these structures are between ¹ 92 and
¹ 95 kcal mol¹1. For the parvoviral genomes,
the ‘rabbit ears’ secondary structure of the
terminal 119 nucleotides is as proposed by
Chou et al . (1997). In all three structures, the
hairpin sequences that are homologous to the
first hairpin from the 38 end of the parvoviral
genome are shaded in grey. The proposed
Pu:Pu sheared pairings are indicated by the
solid dots.
telomeres hint at a possible role of these hairpins in protecting the single-stranded 38 termini exposed during
replication.
The GGA:GGA mismatch motif present in the fifth predicted hairpin at the 38 end is conserved in all the Streptomyces telomeres sequenced in this study, except for a
variation (GA A:GGA) in the S. coelicolor chromosome.
The GGA:GGA motif has been shown by NMR studies
(Chou et al ., 1994; Zhu et al ., 1995b) to form a highly stable
duplex structure embedded in the context of d(TGGA A)2 ,
which represent the eukaryotic centromere repeat sequence. In this duplex, two G:A sheared pairs flank the two
unpaired G-residues in the centre, which form interstrand
stacking.
Coincidentally, a sequence (GGAATGCAATGG) identical to the central 13 nucleotides of palindrome VII, except
for the two AT dinucleotides (underlined), which are GA in
palindrome VII, has been studied by Zhu et al . (1996) by
NMR. In the determined structures, the GCA tight loop
and the GGA:GGA motifs, both involving G:A sheared
pairings, co-exist stably. The 38 end of the autonomous parvovirus genomes contain a similar, conserved GAA:GA
‘bubble’, which is resistant to the single-strand-specific
mung bean nuclease (Astell et al ., 1979). NMR studies
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Terminal sequences of Streptomyces chromosomes and plasmids 913
show that this motif forms an asymmetrical G:A-bracketed
single A-stack structure (unpublished results cited in Chou
et al ., 1997). This ‘bubble’ motif is located near the replication origin and is crucial for initiation of replication
(Cotmore and Tattersall, 1994).
The two classes of parvoviruses, helper dependent and
autonomous, use different schemes for replication. Chen
(1996) adopted the ‘hairpin transfer’ mechanism of the
helper-dependent parvovirus replication (Kornberg and
Baker, 1992) in one of the models for patching the singlestranded gap at the 38 termini of the Streptomyces telomeres. In this model, the terminal palindrome exposed at
the 38 single-stranded gap folds back and primes the
extension synthesis that fills the gap. This is followed by
nicking at the opposite side of the palindrome, flipping of
the palindrome and extension of the newly created 38
end to complete the patching synthesis. This model predicts that the terminal palindromes would occur in two
opposite orientations among the population of telomeres
examined (as in the helper-dependent parvoviruses). For
the Streptomyces telomeres, the flipping may theoretically
occur in the first 13 bp (through foldback in palindrome I) or
the first 83–84 bp (through foldback of palindromes I–IV).
However, in all the terminal sequences determined in this
study, no flipping was detected. All the palindrome I sequences were unique, so were the orders of palindromes
I–IV. Furthermore, in more than 20 independently determined terminal sequences of pSLA2, Z. Qin and S. N.
Cohen (accompanying paper) also found no flipping in
the termini of pSLA2 plasmid. Therefore, the replication
of the Streptomyces telomeres does not appear to use
the hairpin transfer mechanism.
The autonomous parvoviruses differ from the helperdependent parvoviruses in that their 38 terminal palindromes are uniquely oriented. This absence of flipping
requires a special asymmetrical resolution of a head-tohead dimeric intermediate during replication as part of
the ‘modified rolling hairpin’ model proposed by Astell et
al . (1985). The intermediate is resolved at the dimeric
junction by a nicking enzyme, which resolves the dimer
into an ‘extended form’ with the nickase covalently attached
to the 58 end (which would appear as TP) and a ‘turnaround’ form with a hairpin loop. The extended form maintains the original orientation of the 38 palindrome and is
further processed into the mature form. The turn-around
form is recycled in the replication process.
The similarity between the secondary structures of the
38 terminal palindromes of the autonomous viral genomes
and of the Streptomyces telomeres suggests that the latter
may also use the modified rolling hairpin mechanism
during the patching of the single-stranded 38 ends. This
would predict the presence of a ‘turn-around form’ formed
by the terminal palindromes, but this could not be detected
by Southern hybridization and polymerase chain reaction
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
(PCR) analysis (our unpublished observations). Therefore,
if Streptomyces telomere replication involves the modified
rolling hairpin mechanism, the transient stage of the extended form may be very short and beyond our detection.
The differences between the six terminal sequences
presented here and the three plasmid terminal sequences
determined previously by other laboratories are striking.
Some may possibly represent sequencing errors caused
by the abundant stable secondary structures in the terminal sequences (Yamakawa et al ., 1996). In fact, Z. Qin
and S. N. Cohen (accompanying paper) have found several probable errors in the published terminal sequence
of pSLA2. Of all these Streptomyces terminal sequences,
that of SCP1 shows almost complete lack of homology to
the rest. Perhaps the SCP1 termini have a different patching system.
The multiple terminal palindromes found in the Streptomyces telomeres characterized in this study are not found
in other TP-containing invertrons, such as the fungal linear
plasmids and some linear viral genomes. Unlike the Streptomyces linear replicons, which replicate bidirectionally
from an internal origin, these linear viral DNA molecules
(such as adenoviruses, f29 and PRD1) replicate the complete genome from one end to the other without interruptions. The terminal palindromes present in the linear
Streptomyces chromosomes and plasmids may reflect a
specific feature of the terminal patching synthesis.
Experimental procedures
Bacteria and plasmids
The Streptomyces cultures used in this study are listed in
Table 1. All Streptomyces species were grown in YEME plus
1% glycine at 308C, except for S. clavuligerus, which was
grown in TSB without glycine (Hopwood et al., 1985).
pBluescript-II KS plasmid vectors (Stratagene) were used
for cloning and sequencing of the terminal DNA. pLUS450
plasmid contained a 2.6 kb Bgl II terminal fragment of the S.
lividans ZX7 chromosome (Lin et al ., 1993). Cosmid 3B-11
contained a 40 kb chromosomal DNA or S. lividans TK64
that overlapped with the pLUS459 insert.
Reagents and enzymes
Restriction enzymes and T4 DNA ligase were from Boehringer
Mannheim and New England Biolabs. The Sequenase version
2.0 DNA sequencing kit was from US Biochemical. Calf thymus
terminal deoxynucleotidyl transferase was from Promega.
GeneClean II kit was from Bio101. Thiostrepton was a gift
from Dr S. J. Lucania, Squibb Institute for Medical Research
(Princeton, USA). a-[35S]-dATP was from Amersham.
Pulsed-field gel electrophoresis
Isolation of the intact genomic DNA of Streptomyces , Ase I
914 C.-H. Huang et al.
Table 1. Streptomyces cultures used in this
study.
Culture
Known linear plasmids
Source/reference
S.
S.
S.
S.
S.
S.
S.
Unknown
pSCL1, pSCL2, pSCL3
None
Unknown
None
None
pSPA1, pSPA2, pSPA3
CRCCa
CRCCa
Hopwood et al . (1983)
CRCCa
CRCCa
Zhou et al . (1988)
Hopwood et al . (1983)
Chen et al . (1993)
J. Cullum
Hirochika et al . (1985)
argenteolus ATCC 11009
clavuligerus ATCC 27064
coelicolor A3(2) strain M145
hydrogenans ATCC 19631
lipmanii ATCC 27357
lividans ZX7
parvulus ATCC 12434
(John Innes no. 2283)
S. rimosus R6539
S. rochei 7434-AN4
None
pSLA2 and two others
a. Culture Collection and Research Center, Food Industry Research and Development
Institute, Hsinchu, Taiwan.
digestion and PFGE were carried out according to Lin et al .
(1993), which was modified from Kieser et al . (1992). Electrophoresis was performed in a Rotaphor R23 (Biometra) with
two different sets of control programmes. Separation of intact
Streptomyces chromosomes was carried out for 5 days in an
electric field of 60 V, using a linear ramped pulse time from
60 to 30 min and a reorientation angle of 958, and a reverse
switch time of 10 min. Separation of Ase I digests was performed for 36 h in an electric field of 200 V, using a ramped
pulse time from 5 to 120 s and a reorientation angle of 1208.
The chromosomal DNA of Schizosaccharomyces pombe
used as size markers was from Bio-Rad.
Isolation and cloning of terminal DNA
Terminal DNA of the Streptomyces genomes was purified and
cloned using a modified procedure of Lin et al . (1993), which is
outlined in Fig. 2. Total DNA was extracted from mycelium (1 g
wet weight) growing for 36 h in YEME or TSB medium by centrifugation in an ethidium bromide–caesium chloride density
gradient as described by Hopwood et al . (1985). The isolated
DNA was digested with Pst I enzyme and mixed with NaCl (5 M)
to a final concentration of 0.3 M. Five microlitres of glass milk
(from GeneClean II kit) was added, and the mixture was incubated at room temperature for 1 h with intermittent shaking.
The glass microbeads were spun down and washed three or
four times with 0.3 M NaCl. The bound DNA–TP complex
on the glass microbeads was eluted twice by incubation in
25 ml of 1% SDS for 30 min at 508C (Thomas et al ., 1979).
Piperidine was added to the pooled supernatant containing
the eluted DNA–TP complex to a final concentration of 10%,
and the mixture was incubated at 508C for 1 h in a capped
microcentrifuge tube, followed by another hour of incubation
at 508C with the cap off (to remove piperidine). DNA was
then purified with the GeneClean II kit and ligated to Pst Iand Hin cII-cleaved pBluescript-II KS– vector. E. coli DH5a
was transformed with the ligated DNA, and ampicillin-resistant transformants were isolated. Plasmids containing inserts
were purified from the transformants and checked by restriction digestion with (i) Pst I, and (ii) Pst I plus Xho I (which should
separate the insert from the vector DNA). Plasmids that contain
bona fide terminal DNA should give only one Pst I fragment
(vector plus insert) in the first case, and the vector DNA plus
one or more fragments in the second case. Only recombinant
plasmids exhibiting these restriction patterns were selected
for further studies. In one case (S. coelicolor ), Bam HI was
used instead of Pst I in the digestion of total DNA, and the subsequent procedure was modified accordingly.
DNA sequencing
Sequencing of the terminal DNA was based on the dideoxy
chain termination method of Sanger et al . (1977). DNA fragments were cloned in pBluescript-II KS– and sequenced
using the Sequenase version 2.0 DNA sequencing kit. Two
more modifications were used to overcome the possible secondary structures formed by the terminal palindromes (Yamakawa et al ., 1996). First, chain elongation time was extended
from 2 to 8 min to obtain longer sequences. Secondly, a terminal deoxynucleotide transferase reaction was performed
according to Fawcett and Bartlett (1990) after chain termination to remove band ambiguities caused by abnormal termination of pausing of polymerase. Thirdly, in addition to 7 M urea,
10% (v/v) formamide was added to the polyacrylamide gel to
eliminate distorted separations of the bands.
DNA fold prediction
The energy-optimized secondary structure was constructed
using M. Zuker’s DNA Fold Server on WWW (http:/ /www.
ibc.wustl.edu/,zuker/dna/form1.cg) based on a folding temperature of 308C in 1 M NaCl.
Acknowledgements
This study was supported by research grants (NSC86-2316B010-060-BC, NSC86-2314-B010-022) from R. O. C. National
Science Council. C. W. C. was a recipient of a research award
from the Medical Research and Advancement Foundation in
memory of Chi-Shuen Tsou. We thank M. Zuker for their
DNA Fold Server on WWW, Dr S. J. Lucania for the gift of
thiostrepton, Dr Hanna Yuan for drawing our attention to
sheared Pu:Pu pairing at the hairpins, Dr Shan-Ho Chou for
discussion of sheared pairings and David Hopwood for reading the manuscript and suggesting improvements.
References
Astell, C.R., Smith, M., Chow, M.B., and Ward, D.C. (1979)
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
Terminal sequences of Streptomyces chromosomes and plasmids 915
Structure of the 38 hairpin termini of four rodent parvovirus
genomes: nucleotide sequence homology at origins of
DNA replication. Cell 17: 691–703.
Astell, C.R., Chow, M.B., and Ward, D.C. (1985) Sequence
analysis of the termini of virion and replicative forms of
minute virus of mice suggests a modified rolling hairpin
model for autonomous parvovirus DNA replication. J Virol
54: 171–177.
Birch, A., Häusler, A., and Hütter, R. (1990) Genome
rearrangement and genetic instability in Streptomyces
spp. J Bacteriol 172: 4138–4142.
Calcutt, M.J., and Schmidt, F.J. (1992) Conserved gene
arrangement in the origin region of the Streptomyces coelicolor chromosome. J Bacteriol 174: 3220–3226.
Chang, P.C., and Cohen, S.N. (1994) Bidirectional replication
from an internal origin in a linear Streptomyces plasmid.
Science 265: 952–954.
Chen, C.W. (1995) The unstable ends of the Streptomyces
linear chromosomes: a nuisance without cures? Trends
Biotechnol 13: 157–160.
Chen, C.W. (1996) Complications and implications of linear
bacterial chromosomes. Trends Genet 12: 192–196.
Chen, C.W., Yu, T.-W., Lin, Y.S., Kieser, H.M., and
Hopwood, D.A. (1993) The conjugative plasmid SLP2 of
Streptomyces lividans is a 50 kb linear molecule. Mol
Microbiol 7: 925–932.
Chou, S.-H., Zhu, L., and Reid, B.R. (1994) The unusual
structure of the human centromere (GGA)2 motif. J Mol
Biol 244: 250–268.
Chou, S.-H., Zhu, L., and Reid, B.R. (1996) On the relative
ability of centromeric GNA triplets to form hairpins versus
self-paired duplexes. J Mol Biol 259: 445–457.
Chou, S.-H., Zhu, L., and Reid, B.R. (1997) Sheared purine–
purine pairing in biology. J Mol Biol 267: 1055–1967.
Coombs, D.H., and Pearson, G.D. (1978) Filter-binding
assay for covalent DNA-protein complexes: adenovirus
DNA–terminal protein complex. Proc Natl Acad Sci USA
75: 5291–5295.
Cotmore, S.F., and Tattersall, P. (1994) An asymmetric
nucleotide in the parvoviral 38 hairpin directs segregation
of a single active origin of DNA replication. EMBO J 13:
4145–4152.
Cullum, J., Flett, F., and Piendl, W. (1988) Genetic instability
in streptomycetes. Microbiol Sci 5: 233–235.
Fawcett, T.W., and Bartlett, G. (1990) An effective method
for eliminating ‘artifact banding’ when sequencing doublestranded DNA templates. BioTechniques 9: 46–48.
Fischer, J., Maier, H., Viell, P., and Altenbuchner, J. (1996)
The use of an improved transposon mutagenesis system
for DNA sequencing leads to the characterization of a
new insertion sequence of Streptomyces lividans 66.
Gene 180: 81–89.
Gravius, B., Glocker, D., Pandza, K., Hranueli, D., and
Cullum, J. (1994) The 387 kb linear plasmid pPZG101 of
Streptomyces rimosus and its interactions with the chromosome. Microbiology 140: 2271–2277.
Hirochika, H., Nakamura, K., and Sakaguchi, K. (1985) A
linear DNA plasmid from Streptomyces rochei with an
inverted repetition of 614 basepairs. EMBO J 3: 761–766.
Hopwood, D.A., Kieser, T., Wright, H.M., and Bibb, M.J.
(1983) Plasmids, recombination, and chromosomal mapping
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916
in Streptomyces lividans 6. Gen Microbiol 129: 2257–
2269.
Hopwood, D.A., Bibb, M.J., Chater, K.F., Kieser, T., Bruton,
C.J., Kieser, H.M., et al . (1985) Genetic Manipulation of
Streptomyces – A Laboratory Manual . Norwich: The
John Innes Foundation.
Huang, C.-H. (1997) The terminal DNA of Streptomyces
chromosomes and linear plasmids. MSc thesis, National
Yang-Ming University, Taipei, Taiwan.
Kieser, H.M., Kieser, T., and Hopwood, D.A. (1992) A
combined genetic and physical map of the Streptomyces
coelicolor A3(2) chromosome. J Bacteriol 174: 5496–
5507.
Kinashi, H., Shimaji-Murayama, M., and Hanafusa, T. (1991)
Nucleotide sequence analysis of the unusually long terminal inverted repeats of a giant linear plasmid, SCP1.
Plasmid 26: 123–130.
Kornberg, A., and Baker, T.A. (1992) DNA Replication. New
York: W.H. Freeman.
Leblond, P., and Decaris, B. (1994) New insights into the
genetic instability of streptomyces. FEMS Microbiol Lett
123: 225–232.
Leblond, P., Fischer, G., Francou, F., Berger, F., Guerineau,
M., and Decaris, B. (1996) The unstable region of Streptomyces ambofaciens includes 210 kb terminal inverted
repeats flanking the extremities of the linear chromosomal
DNA. Mol Microbiol 19: 261–271.
Lezhava, A.L., Mizukami, T., Kajitani, T., Kameoka, D.,
Redenbach, M., Shinkawa, H., et al . (1995) Physical map
of the linear chromosome of Streptomyces griseus . J Bacteriol 177: 6492–6498.
Lin, Y.-S., and Chen, C.W. (1997) Instability of artificially circularized chromosomes of Streptomyces lividans . Mol
Microbiol 26: 709–719.
Lin, Y.-S., Kieser, H.M., Hopwood, D.A., and Chen, C.W.
(1993) The chromosomal DNA of Streptomyces lividans
66 is linear. Mol Microbiol 10: 923–933.
Musialowski, M.S., Flett, F., Scott, G.B., Hobbs, G., Smith,
C.P., and Oliver, S.G. (1994) Functional evidence that
the principal DNA replication origin of the Streptomyces
coelicolor chromosome is close to the dnaA–gyrB region.
J Bacteriol 176: 5123–5125.
Netolitzky, D.J., Wu, X., Jensen, S.E., and Roy, K.L. (1995)
Giant linear plasmids of b-lactam antibiotic producing
Streptomyces . FEMS Microbiol Lett 131: 27–34.
Pandza, K., Pfalzer, G., Cullum, J., and Hranueli, D. (1997)
Physical mapping shows that the unstable oxytetracycline
gene cluster of Streptomyces rimosus lies close to one
end of the linear chromosome. Microbiology 143: 1493–
1501.
Peñalva, M.A., and Salas, M. (1982) Initiation of phage f29
DNA replication in vitro: Formation of a covalent complex
between the terminal, p3, and 58-dAMP. Proc Natl Acad
Sci USA 79: 5522–5526.
Qin, Z., and Cohen, S.N. (1998) Replication at the telomeres
of the Streptomyces linear plasmid pSLA2. Mol Microbiol
28: 893–903.
Sanger, F., Nicklen, S., and Coulson, A.R. (1977) DNA
sequencing with chain-terminating inhibitors. Proc Natl
Acad Sci USA 74: 5463–5467.
SantaLucia, J.J., Allawi, H.T., and Seneviratne, P.A. (1996)
916 C.-H. Huang et al.
Improved nearest-neighbor parameters for predicting DNA
duplex stability. Biochemistry 35: 3555–3562.
Schrempf, H., Kessler, A., Brönneke, V., Dittrich, W., and
Betzler, M. (1989) Genetic instability in Streptomycetes.
In Genetics and Molecular Biology of Industrial Microorganisms . Hershberger, C.L., Queeners, S.W., and
Hegeman, G. (eds). Washington, DC: American Society
of Microbiology Press, pp. 133–140.
Thomas, C.A., Saigo, K., McLeod, E., and Ito, J. (1979) The
separation of DNA segments attached to protein. Anal Biochem 93: 158–166.
Wang, S.-J. (1996) The topological role of linear chromosome in the conjugative process of Streptomyces . MSc
thesis, National Yang-Ming University, Taipei, Taiwan.
Yamakawa, H., Nakajima, D., and Ohara, O. (1996) Identification of sequence motifs causing band compressions on
human cDNA sequencing. DNA Res 3: 81–86.
Yang, Y.-L. (1994) The termini of the linear chromosomes of
the Streptomyces . MS thesis, National Yang-Ming Medical
College, Taipei, Taiwan.
Zakrzewska-Czerwinska, J., and Schrempf, H. (1992)
Characterization of an autonomously replicating region
from the Streptomyces lividans chromosome. J Bacteriol
147: 2688–2693.
Zhou, X., Deng, Z., Firmin, J.L., Hopwood, D.A., and
Kieser, T. (1988) Site-specific degradation of Streptomyces lividans DNA during electrophoresis in buffers contaminated with ferrous iron. Nucleic Acids Res 16: 4341–
4354.
Zhu, L., Chou, S.-H., Xu, J., and Reid, B.R. (1995a) Structure
of a single-cytidine hairpin loop formed by the DNA triplet
GCA. Nature Struct Biol 2: 1012–1017.
Zhu, L., Chou, S.-H., and Reid, B.R. (1995b) The structure of
a novel DNA duplex formed by human centromere
d(TGGAA) repeats with possible implication for chromosome attachment during mitosis. J Mol Biol 254: 623–637.
Zhu, L., Chou, S.-H., and Reid, B.R. (1996) A single G-to-C
change causes human centromere TGGAA repeats to
fold back into hairpins. Proc Natl Acad Sci USA 93:
12159–12164.
Zuker, M. (1989) On finding all suboptimal foldings of an RNA
molecule. Science 244: 48–52.
Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 905–916