Gene, 103 (1991) 107-111 0

1991 Elsevier

GENE

Science

Publishers

B.V. All rights reserved.

107

0378-l 119/91/$03.50

05044

Parameters affecting plasmid stability in Bacillus subtilis (Recombinant

DNA;

direct and inverted

repeats;

shuttle plasmid;

structural

plasmid

instability;

high-M,. DNA)

Heinrich Leonhardt and Juan C. Alonso Max-Planck-Institut

ftir Molekulare

Genetik, D-1000

Berlin 33 (F.R. G.)

Received by R.E. Yasbin: 27 August 1990 Revised/Accepted: 17 April/25 April 1991 Received at publishers: 7 May 1991

SUMMARY

For the analysis of parameters affecting plasmid stability in Bacillus subtilis, we used a PUB 1 lo-derived shuttle plasmid containing direct and inverted nucleotide sequence repeats (DRs and IRS). Deletions of up to 6 kb were found to occur between DRs of 7 to 16 bp. IRS as small as 43 or 58 bp were shown to stimulate the formation of these deletions in their neighbourhood. However, these structural features (DRs and IRS) per se were not responsible for plasmid instability. The unstable recombinant plasmids, but not their deletion-carrying (d) derivatives, were found to impair the growth of the host and to accumulate high amounts of linear plasmid multimers [high mol. wt. (hmw) DNA]. We propose that the accumulation of hmw DNA may be the major reason for the selective pressure against recombinant plasmids, and the enrichment of d-plasmids. Host mutations and other parameters increasing the stability of recombinant plasmids in B. subtilis are described.

INTRODUCTION

Most cloning experiments in B. su#@i?is~ are performed with plasmids, which replicate via a sigma mechanism similar to the one observed for enterobacteriaceae ss phages. Such experiments are often complicated by problems of structural plasmid instability (see Ehrlich et al., 1986, for review), with DNA rearrangements that may involve enzymes with topoisomerase-like (Lopez et al., 1984; Peijnenburg et al., 1988) or nuclease activity (Kupsch et al., 1989). The high structural instability, however, has been mainly attributed to slipped mispairings between short DRs

Correspondence to: Dr. J.C. Alonso, Max-Planck-Institut Genetik, Ihnestrasse 73, D-1000 Berlin 33 (F.R.G.) Tel. (49-30)8307274; Abbreviations:

Fax (49-30)8307385.

Ap, ampicillin;

repeat;

d-plasmid,

Form-I,

covalently

repeat; kb, kilobase resistance/resistant;

fur Molekulare

B., Bacillus; bp, base pair(s);

deletion-carrying closed

circular;

plasmid; hmw,

ds, double

DR, direct strand(ed);

high mol. wt.; IR, inverted

or 1000 bp; Nm, neomycin; nt, nucleotide(s); ss, single strand(ed); ss(c), ss circles.

R,

during plasmid replication (Ehrlich et al., 1986). From studies on precise transposon excision it was inferred that long IRS may stimulate this process (Peeters et al., 1988). The aim of the present study was to analyse the contribution of sequence repeats to the instability of recombinant plasmids in B. subtilis and to identify parameters that improve plasmid stability.

EXPERIMENTAL

AND DISCUSSION

(a) Structure and frequency of deletions To analyse the problem of plasmid instability in B. subtilis we tested the effect of copy number (Leonhardt, 1990) plasmid size, genetic background of the host and the presence of repeats (DRs and IRS). To facilitate these analyses we chose a plasmid that provides a positive selective advantage for d-plasmid forms. Computer analysis revealed that pEB 114 contains more than 1500 partially overlapping DRs of at least 7 bp and a 58-bp IR (Fig. 1, Table I). Plasmid pEB 114 could be stably maintained for more than

108 earlier (Rottlander and Trautner, 1970). Transformants (30-40 each) were selected for NmR and grown for 25-30 generations in the presence of Nm. Then plasmid DNA was prepared and tested in agarose gels for the accumulation of d-plasmid forms. Unlike in E. coli, deletions were prevalent with frequencies depending on the genetic background of the host. In the Ret + strains YB886 and BG82 about half of all NmR tr~sform~ts contained d-plasmids. The recA4 derivatives of these strains (YSlOlS and MI1 12; Table I) showed a higher accumulation of d-plasmids than YB886 or BG82, i.e., all transformants contained d-plasmids. This means that the recA4 mutation, which is often used in cloning experiments in B. subtifis, not only does not prevent, but even increases the d-plasmid frequency. This may be explained by the observation in E. coli that a defect in RecA (counterpart of the B. subtilis RecA) interferes with initiation and/or completion of DNA replication (Skarstad and Boye, 1988). Finally, no d-plasmids were detected, under our assay conditions, in strains BG 113, PSL 1 and in BG83. These three strains contain spontaneous mutations (stp-1 or stb-I; Table I), which have been selected by their phenotype of increasing plasmid stability. So far, the molecular basis of these phenotypes is unknown. Both the stb-1 and stp-1 mutations, however, reduce the cell growth rate and seem to be able to compensate the increased plasmid instability in recA4 strains. These results rule out the involvement of the restriction and modification system of these strains or the absence of bacteriophage SP/?in d-plasmid accumulation. Specifically, the ~.~~z~ hsmR mutations or the presence of ui~~F~ (Table I) do not prevent the accumulation of d-plasmid (Leonhardt, 1989; present work). The d-plasmid were isolated and 80 deletion endpoints mapped by restriction analysis. The majority of the deletion

Fig. 1. Physical structure of pEB 114 and location of deletion endpoints. The pUB1 IO-derived part is represented by a single line and foreign DNA by double lines. Proteins encoded by pEBl14 are represented by open arrows inside the circle. The two artificially introduced promoters (rat and spa) are represented transcription.

by open arrowheads

The location

pointing

in the direction

ofthe 58-bp inverted repeat (composed

and IR2) is indicated by thin arrows inside the circle.

(a, e, d, f, x) show the location

outside endpoints

(see Fig. 2). The location

for plasmid

replication

and orientation

in B. subtilis is indicated

leading-strand synthesis; oriL: lagging-strand

Small

of the sequenced

of

at IRl arrows deletion

of two origins (ori)

outside

the circle (oriU:

synthesis).

100 generations in the most commonly used E. cofi strains (Leonhardt and Alonso, 1988). For the analysis of piasmid stability in B. subtiiis competent cells were transformed with pEB 114 as described TABLE Bad/us

I subtilis strains

and plasmids

used in this study

Relevant

properties

Source

or reference

et al. (1980)

Strains YB886

rrpC2 met85 sigB umyE xin- 1 a~fSP~

Yasbin

YBIOIS”

trpC2 metB5 sigB amyE xin-I attSPfl recA4

Friedman

BG113

trpC2 metBS sig3 amyE xin-1 at&P/? recA4 stb-I

Alonso et al. (1987)

BG82

leuA8 arg-15 thrA5 hsmM hsmR leuA8 nrg-1.5 thrA5 hsmM hsmR recA4

Present

work

Tanaka

(1979)

leuA8 arg-15 thrA5 hsmM hsmR stp-1 attSPfi

Leonhardt

leuA8 arg-15 thrA5 hsmM hsmR stp-1 attSP/l recA4

Ostroff

MI112” BG83 PSLl a

and Yasbin (1983)

et al. (1991)

and P&ne (1983); this work

Plasmids pEBll1 pEB112

Ap, Nm, pBR322-pUB 110 derivative

Leonhardt and Alonso (1988)

Ap, Nm, luclq, pBR322-pUB

Leonhardt

pEBll3

Ap, Nm, xyZE, pBR322-pUB110

pEBl14

Ap, Nm,

n The recA4 allele was known

previously

lQ@,

.xyE,

110 derivative derivative

pBR322-pUBll0

derivative

as recE4. Nm was used at 5 pgjml.

Leonhardtand Leonhardt

and Alonso

(1988)

Alonso(1988) and Alonso

(1988)

109 For

I deletions

type

28

STTAAAAACGTTTTTAAAGGC II IIIIIIIIIIIIIIII GCCTTTAAAAACGTTTTTAAB

51 3545

II

53

AACCCCTTAAAAACGTTTTTAAAG

3557

GACGGCTTAAAAGCCTTTAAAAAC

IIIIIII

I III

3501

derivative

- 42

I III

- 3478

II deletions

(c) Selective

8553

- CTGGTGCTACGCGAGGCTGACGAG

3369

- TGTGTGCTCTGCGAGGCTGTCGGC

- 3346

8111

- CCTTCATTCACCATCCGGAAAAAG

- 8088

3888

- TTCACATTCACCACCCTGAATTGA

- 3865

IIIIIIIII

Fig. 2. Sequences are given around

of the deletion

indicate

to the endpoints independent

II

segments

- 8530

II

III

endpoints

the DRs, and the directly

vertical lines. Underlined Numbers

IIIIIIIII

in pEBl14. repeated

of sequence

The sequences

nt are marked

by

remain in the A-plasmid.

in pEBl14. Letters a, d,e, f,x correspond 1.The sequence of deletion ‘d’ was found in three

nt position in Fig.

transformants

were

however, the type

stimulate the deletion formation between short DRs, which are about 6 kb apart. A 58-bp IR seems to be sufficient to give a maximal

IIIII

d-plasmids;

showed a bias for deletion endpoints nearby the IR, like the original pEB 114. In case of a plasmid with an IR of 11 bp none of the deletion endpoints mapped in the vicinity of the IR. Hence, even relatively small IRS of 43 to 58 bp can

III

- TTGCATTCTACAAACTGCATAACT

type

30-40 NmR

of deletions changed with the IR length. Plasmids with the IR length reduced to 52 bp and 43 bp accumulated 76% and 65% type-1 deletions, respectively, i.e., they still

3522

65 - ACCTTGTCTACAAACCCCTTAAAA

IIIIIIIII

each

tested. All of them contained

clones.

endpoints (> 97%) was clustered within or nearby ( k 40 bp) a 58bp IR (formed by sequences IRl and IR2; Fig. 1). These deletions were termed ‘type-I’ deletions. The remaining deletions with endpoints outside these regions were named ‘type-II’. Endpoints from both deletion types were sequenced using the method of Sanger et al. (1977). All deletion endpoints were located within small DRs with a length of 7-16 bp (Fig. 2). Distribution of the sequenced deletion endpoints within pEBl14 is shown in Fig. 1. About 500 of the 1500 DRs in pEB 114, which can overlap and are distributed all over the plasmid, would yield viable deletion products. The fact that, however, preferably those within or nearby the IR were used for deletion formation, suggests a stimulation by the 58-bp IR. (b) Role of IRS in deletion formation To test the significance of the IR on deletion formation in pEBl14 we shortened IRl (position 1-58 in pEB114; Fig. 1). For this purpose, pEBl14 was linearized with Barn HI and in vitro deletions were introduced using partial BAL 31 degradation. These DNA ends were ligated to an XbaI linker and deleted sequences outside IRl were restored in a second step by replacing a DNA fragment (HindIII-XbaI) with the original sequence. The precise deletion endpoints were determined by nt sequencing.

rate of DNA rearrangement

pressure

against

recombinant

in its vicinity. plasmids

in

Bacillus subtilis The results raise the question, whether these DRs and IRS are per se sufficient to account for the high structural instability of pEB 114 observed in some B. subtilis hosts. This question could be answered by comparing the stability of smaller pEB 114 derivatives (pEB 111, 6.4 kb; pEB 112, 7.9 kb; pEB113, 8.6 kb; Table I). These derivatives are identical to pEB 114 in the regions containing the IR, but differ in size. Twenty to 80 transformants each were analysed and d-plasmids could be identified at different frequencies: with pEB114 (> 95%) pEB113 (1%) pEB 112 (2%), pEB 111 (< 1%). These results indicate that the presence of repeats alone cannot account for the high instability of pEB 114. This conclusion is furthermore supported by the fact that a fivefold reduction in pEB 114 copy number markedly increased plasmid stability (Leonhardt, 1990). One reason for the high instability of pEB 114 could be the fact that pEB114 impairs the growth rate of its host (Leonhardt, 1989). This impairment would provide a selective advantage for the accumulation of d-plasmids, which do not impair the growth rate of the host. Recently, it has been shown in E. coli and B. subtilis that the accumulation of linear plasmid multimers (hmw DNA; Cohen and Clark, 1986; Viret and Alonso, 1987) might be detrimental and even lethal to the host (Alonso et al., 1987; Viret and Alonso, 1988; Kusano et al., 1989). To investigate this possibility, total DNA was prepared from cells containing either pEB 114 or lower-size derivatives. Different DNA forms were separated by agarose gel electrophoresis and plasmid-specific DNA forms were identified by Southern hybridization. In principle, three different plasmid DNA forms [ ss(c), form-I, and hmw] can be distinguished by their relative electrophoretic mobility and their properties in Southern blottings, i.e., when the blotting is made at neutral pH only ss(c) and ssDNA, consisting of hmw DNA molecules, bind to nylon membranes (teRiele et al., 1986; Viret and Alonso, 1987). Autoradiographic signals for these three plasmid DNA

nearby this IR; however, these structural

features alone do

not cause plasmid instability; (ii) the unstable recombinant plasmids, but not their deletion-carrying derivatives, were found to impair the growth of the host. The per se slower growth of strains BG83, BG113 and PSLl reduces d-plasmid frequency. A similar reduction in d-plasmid frequency was observed when the growth rate of YB886 or BG82 strains was artificially reduced by growing them in minimal media or by drastically reducing plasmid copy number (Leonhardt, 1990); (iii) cells containing unstable plasmids

PlasmId

I kb)

SIX

Fig. 3. Distribution

of d-plasmid

Different d-plasmid

forms were quantified

Alonso,

forms

1987). Total DNA was prepared

0.8% agarose-gel

electrophoresis

Plasmid-specific laser densitometer.

previously

by comparing described

pEBl14d

as a function

ss(c) (0)

of plasmid

(3.5 kb), pUBll0

PSLl,

were described

are products

of pUBll0

with 32P-

are known

with a (46-50

ss(c) DNA;

Viret and

of the other plasmids

could be

in the autoradiograph

1987). These numbers

as

DNA could deplete cellular pools of enzymes involved in DNA metabolic processes (Viret and Alonso, 1988). This depletion might be the molecular basis of the high selective pressure against recombinant plasmids. Hence, stability of recombinant plasmids in B. subtilis can be increased by reducing plasmid size, growth rate of the host or plasmid copy number.

of plasmid

and hmw DNA (A), were separately size. The following plasmids

(4.5 kb), pEBl14x of deletions

earlier (Leonhardt

by

to nylon membranes.

were used:

(5.5 kb), pEBl11

pEB 112 (7.9 kb), pEB4 (8.6 kb) and pEB 114 (9.8 kb). Plasmids and pEBl14x

size.

separated

signals were quantified

intensities

(Viret and Alonso,

copies, i.e. form-1 (O), plotted

signal

of plasmid

before (Viret and

by hybridization

and two to three copies

1988) so that the copy number

determined

in strain

and transferred

The copy numbers

of form-1 DNA

Alonso,

as described

DNA forms were identified

labelled PUB 110 DNA. Autoradiographic copies

as a function

were found to accumulate high amounts of hmw DNA. Electron microscopy of isolated hmw DNA displayed linear DNA molecules up to 100 kb in size, which were either ss, ds or ds with an ssDNA end (Viret and Alonso, 1987; Leonhardt et al., 1991). The accumulation of hmw

pEB 114d

d and x (Figs. 1 and 2); others

and Alonso,

ACKNOWLEDGEMENT

(6.4 kb),

1988; Table I).

forms were quantified with a laser densitometer and were converted to copy numbers by comparison with PUB 110 (Fig. 3). Plasmid pUBll0 and the stable d-plasmids pEB 114d and pEB 114x (Figs. 1 and 2) are mainly present as form-1 DNA. The number of form-1 (Bron and Luxen, 1985) and ss(c) plasmid DNA molecules per cell drops with increasing plasmid size (Fig. 3). Furthermore, the comparison shows that hmw DNA is accumulated and pEB 114-transformed cells even contain twice as much of hmw DNA with increasing plasmid size, than of form-l DNA. The drop of form-1 plasmids is inversely correlated with the increase of hmw DNA. The sum of plasmid copies per cell, i.e., form-1 plus hmw molecules, remains approximately constant. This correlation suggests that hmw DNA molecules are included in copy control. Electron microscopic studies revealed that 20-30% of this hmw DNA is in ss form and can amount to more than 50 kb/cell (Leonhardt et al., 1991). (d) Conclusions The present study shows that in B. subtilis: (i) small IRS (43-58 bp) stimulate deletion formation between small DRs (7-16 bp), which can be at various positions within or

We are grateful to T.A. Trautner for helpful discussions and critical reading of the manuscript.

REFERENCES Alonso,

J.C., Viret,

J.-F.

and Tailor,

R.H.:

Bacillus subtilis recombination-deficient

Plasmid

mutants.

in

maintenance

Mol. Gen. Genet.

208 (1987) 349-352. Bron,

S. and Luxen,

recombinant Cohen,

E.: Segregational

plasmids

A. and Clark,

A.J.: Synthesis

Escherichia coli K-12. J. Bacterial. Ehrlich,

S.D., Noirot,

teRiele,

instability

P.H.,

H.: Structural

of linear

Genet. Kupsch,

repair

multimers

in

L., Michel,

B. and

of Bacillus subtilis plasmids.

A. (Ed%), Genetic

Engineering, and specificity

In:

Vol. 8. of the

of Bacillus sub&s. Mol. Gen.

system

190 (1983) 481-486.

J., Alonso,

biological

J.C. and Trautner,

parameters

Mol. Gen. Genet. Kusano,

plasmid

Petit, M.A., Jannitre,

Plenum, New York, 1986, pp. 71-83. Friedman, B.M. and Yasbin, R.E.: The genetics excision

1lo-derived

14 (1985) 235-244.

167 (1986) 327-335.

instability

Setlow, J.K. and Hollaender,

constitutive

of PUB

in Bacillus subtilis. Plasmid

affecting

of structural

and

in Bacillus subfilis.

formation

218 (1989) 402-408.

K., Nakayama,

lethality

T.A.: Analysis

deletion

and plasmid

K. and multimer

Nakayama, formation

H.: Plasmid-mediated

in an Escherichia coii recBC

sbsBC mutant. J. Mol. Biol. 209 (1989) 623-634. Leonhardt, H.: Studien zur Plasmidstabilitlt bei Bacillus subfilis. Dissertation Thesis. Freie Universitat Berlin, Berlin, (F.R.G.), 1989. Leonhardt,

H.: Identification

of a low-copy-number

pUBI replicon and its effect on plasmid Gene 94 (1990) 121-124.

mutation

stability

within the

in Bacillus subtilis.

111 Leonhardt, H. and Alonso, J.C.: Construction of a shuttle vector for inducible gene expression in Escherichiu coli and Bacillus subtilis. J. Gen. Microbial.

134 (1988) 605-609.

Leonhardt, H., Lurz, R. and Alonso, J.C.: Physical and biochemical characterization of recombination-dependent synthesis of linear plasin Bacillus subtilis. Nucleic

mid multimers

Acids

Res.

19 (1991)

497-503. deletions

M., Greenberg,

Proc.

Bacillus subtilis with enhanced chimeric

plasmid

of a spontaneous J.

plasmid mutant

of

for Escherichia coli-

transformability DNA.

Bacterial.

156

(1983)

934-936.

B. subtilk strain

168. Mol. Gen. Genet.

Mol. Gen. Genet.

Peijnenburg, A.A.C.M., Bran, S. and Venema, G.: Plasmid mation in Bacillus subtilis. Plasmid 20 (1988) 23-32.

deletion

for-

studies with

with restricted

growth

on

108 (1970) 47-60.

Sanger, F., Nicklen, S. and Coulson, A.R.: DNA sequencing with chainterminating inhibitors. Proc. Nat]. Acad. Sci. USA 74 (1977) 5463-5467. K. and Boye, E.: Perturbed

mutants Tanaka,

T.: Restriction

Viret, J.-F. and Alonso, plasmid

chromosomal

of Escherichia coli. J. Bacterial.

molecules

replication

in recA

170 (1988) 2549-2554.

of plasmid-mediated

in Bncillus

transformation

175 (1979) 235-237.

J.C.: Generation

of linear multigenome-length

in Bacillus subtilis. Nucleic

Acids Res. 15 (1987)

6349-6367. Viret, J.-F. and Alonso,

J.C.: A DNA

Nucleic

155-159.

sequence

outside

for normal replication

the pUBll0

in Bacillus subtilis.

Acids Res. 16 (1988) 4389-4406.

Yasbin, R.E., Fields, P.I. and Andersen, 168 derivatives

212 (1988) 450-458.

and transfection

1. Phage mutants

minimal replicon is required

Peeters, B.P.H., De Boer, J.H., Bran, S. and Venema, G.: Structural plasmid instability in Bacillus subtilis: effect of direct and inverted repeats.

T.A.: Genetic

SP50,

subtilis 168. Mol. Gen. Genet.

cloning with bifunctional

in Bacillus subtilis: isolation

propagated

of

Sci. USA. 81 (1984) 5189-5193.

G.R. and Pine, J.J.: Molecular

vectors

S.A.: Generation

ma1 genes cloned in Bacillus sub&.

in pneumococcal

Natl. Acad.

B. and Lacks,

E. and Trautner,

B. subtilis phage

Skarstad,

Lopez, P., Espinosa,

Ostroff,

Rottlander,

B.J.: Properties

freed of their natural

prophages.

ofBacillussubtilis Gene

12 (1980)