Gene. 87 (1990) 53-61 Elsevier
Structurally stable B a c i l l u s subtUis cloning vectors (Recombinant DNA; plasmids; single-stranded DNA; Gram + bacteria; Tn5 excision; rolling-circle replication)
Lament Janni~re, Claude Bruand and S. Dusko Ehrfieh Laboratoire de Gdndtique Microbienne, Institut de Biotechnologie, INRA, Domaine de Wdvert, 78350 ,louy en Josas (France) Received by J.-P. Lecocq: 18 October 1988 Revised: 14 September 1989 Accepted: 14 November 1989
Cloning of long DNA segments (> 5 kb) in Bacillus subtilis is often unsuccessful when naturally occurring small ( < 10 kb) plasmids are used as vectors. In this work we show that vectors derived from the large (26.5 kb) plasmids pAMpl and pTB 19 allow efficient cloning and stable maintenance of long DNA segments (up to 33 kb). The two large plasmids differ from the small ones in several ways. First, replication of the large plasmids does not lead to accumulation of detectable amounts of ss DNA, whereas the rolling-circle replication typical for small plasmids does. In addition, the replication regions of the two large plasmids share no sequence homology with the corresponding regions of the known small plasmids, which are highly conserved. Taken together, these observations suggest that the mode of replication of the large plasmids is different from that of small plasmids. Second, short repeated sequences recombine much less frequently when carried on large than on small plasmids. This indicates that large plasmids are structurally much more stable than small ones. We suggest that the high structural stability of large plasmids is a consequence of their mode of replication and that plasmids which do not replicate as rolling circles should be used whenever it is necessary to clone and maintain long DNA segments in any organism.
Numerous B. subtilis cloning vectors derive from small plasmids (< 10 kb) isolated from various Gram ÷ bacteria (cf., Ehrlich et al., 1982, for review). It was generally observed that only short DNA segments can be efficiently cloned in these vectors (Michel et al., 1980) and that longer segments often undergo rearrangements (Ehrlich et al., 1986). A common feature of these plasmids is that they Correspondence to: Dr. S.D. Ehrlich, lnstitut de Biotechnologie, INRA, Domaine de Vilvert, 78350 Jouy en Josas (France) Tel. (33-1) 34 65 2511; Fax (33-1)34652273. Abbreviations: Ap, ampicillin; B., Bacgh~; bp, base pair(s}; Cm, chloramphenicol; ds, double-strartd~ed); Er, erythromycin; kb, 1000 bp; LB, Luria-Bertani (medium); Nb, novobiocin; ori, origin of DNA replication; R, resistance/resistant; s, sensitivity/sensitive; ss, single strand(ed); SS +, plasmids accumulating ss DNA; SS-, plasmids not accumulating ss DNA; Tc, tetracycline; Tn, transposon. 0378-11191901503.50© 1990Elsevier SciencePublishersB.V.(BiomedicalDivision)
accumulate ss DNA molecules (re Riele et al., 1986a,b; Projan et al., 1987; Peeters et al., 1988; Devine et al., 1989) due to rolling-circie replication (cf., Gruss and Ehrlich, 1989, for review) similar to that previously described for ss DNA phages from Esche~chia coli (Baas and Jansz, 1988; Model and Russel, 1988). These highly interrelated plasraids, which are widely spread among Gram + bacteria, are termed here SS +. The following observations led to the suggestion that ss DNA generated during rolling-circle repfication strongly stimulates recombination between homologous sequences (Ehrfich et al., 1986). (i) Direcdy repeated sequences, either short (9 bp) or long (4 kb), recombined much more fieqtt~n~ly (100-1000-fold) when carried on an SS + plasmid, pC194, than when carried on the B. subtilis chromosome (Niaudet et al., 1984; Janni~re and Ehrlich, 1987). (ii) Recombination between chromosomally carried long repeats was enhanced > 100 times when an SS + replicon was inserted in the chromosome, close to the repeats (Noimt
54 etal., 1987; Young and Ehrlich, 1989). (iii)Induction of rollm"g-circle replication (i.e., generation of ss DNA) strongly stimulated recombination between short direct repeats (up to 10S-fold) in an E. coli plasmid (Brunier etal., 1988; 1989). Since recombination between short repeats is often the cause of plasmid rearrangements (cf., Ehrlich, 1989, for review), these results suggest that the currently used B. subtilis plasmid cloning vectors are intrinsically structurally unstable. Several observations point to the existence of plasmids different from the SS ÷ class. (i) The sequence of the replication region of three large plasmids ( > 10 kb) capable of replicating in B. subtilb, pTB19 (repA replication region), piP404 and pAM/~I has no homology with any of the sequenced SS + plasmids (Imanaka etal., 1984; 1986;
Gamier and Cole, 1988a,b; Swinfield et al., 1989; L.J., unpublished data); (ii) the overall organization of the replication region of plP404 and the repA region of pTBI9 (lmanaka et al., 1986; Gamier and Cole, 1988b) is reminiscent of the E. coli plasmids R6K and R100 which do not repficate as rolling circles (Lovett etal., 1975; Silver et al., 1977); (iii)plasmids derived from piP404 do not accumulate detectable amounts of ss D N A (Gamier and Cole, 1988b). The aims of the present study were: establishing the existence of plasmids which replicate in B. subtilis without accumulating ss D N A (SS - plasmids) and using such plasmids for constructing vectors that are structurally stable and allow efficient cloning and maintenance of long D N A segments in B. subtilis.
TABLE I List of strains and plasmids Strains
Origin or references
E. coli HVCA5
tkrA 1 leu-6 thf-I lacYl ton.21 supE44 hsdR rspL
B. suMilis SB202 MT! 19
trpC2 tyrA 1 aroB2 hbH2 trpC2 leuB6 r - m -
P. Schaeffer Tanaka (1979)
Nature or construction a
pAMpl pBR322 pC194 pUB 110 pTBI9 pRATI
Natural isolate ($f) Cloning vector (Ec). Natural isolate (Bs). Natural isolate (Bs). Natural isolate (Bs). Hybrid between the repA replication region of pTBI9 and the KmRgene of pUB 110 (Bs). pRATI derivative obtained in vitro (Bs). EcoRI segments Rla and R3 of pTBI9 (Bs). EcoRl segments Rla, Rib and R3 of pTBI9 (Bs). Derivative of pHVI301 constructed in vitro (Bs). Hybrid between pC194 and pBR322 (Ec/Bs). in vivo deletion derivative of pHV33 (Ec). in vivo deletion derivative of pAM#I (Bs). Hybrid composed of pBR322, the EmRgene ofpEl94, a non replicative pC194 and the transposon Tn$ inserted in the CmR gene of pC194. Insertion C! (Ec). As pCli, but insertion C2 (Ec). . Derivatives of pCli and pC2i in which the replication region ofpHV !30! was inserted (£c/Bs). Derivatives of pCii and pC2i in which the replication region of pTB$3 was inserted (EclBs). Hybrid between pHV60 and the replication region of pHVi301 (Bs). Derivative of pHVI431 in which the small £coRl segment was deleted
pRAT! I pTB$2 pTB$3 plL252 pHV33 pHV60 pHVI301 pCli pC2i pp-C1 and -C2 p52-C1 and -C2 pHV1431 pHVI432 pHVI435 pHVI436
As pHV1431, but the replication region of pHVI301 is in the opposite orientation (Bs). HybridbetweenpHV60and the replication regionrepA ofpTBl9 (Ec/Bs).
Clewell etal, (1974) Bolivar et el, (1977) lordanescu (1975) Gryczan etal. (1978) Imanaka etal. (1984) Imanaka et al. (1986)
Imanaka et al. (1986) Imanaka etal. (1984) Imanaka et al. (1984) Simon and Chopin (1988) Primrose and Ehrlich (1981) Michel and Ehrlich (1983) P. Noirot Janni~re and Ehrlich (1987) Janni~,re and Ehrlich (1987) This work This work This work This work This work This work
a Ec, plasmids that replicate in E. coli;Bs, plasmids that replicate in B. sub~lb; Ec/Bs, shuttle plasmids that replicate in E. coil and B. subtilb; Sf, plasmid that replicates in ,Streptococcusfaecalis.
55 MATERIALSAND METHODS Strains and plasmids used are listed in Table I. LB medium, supplemented when necessary with Cm (5-10gg/ml) or Er (0.5 /~g/ml) was used for growing B. subtilis cells. For solid media, 15 mg/ml of agar was added to LB. Enzymes were commercial preparations obtained from Boehringer (Mannheim, F.R.G.) and used according to the supplier's instructions. DNA preparations and labeling in presence of [u-a2p]dCTP and -dATP, densitometric analysis, induction of competence and transformation were performed as previously described (Janni6re and Ehrlich, 1987). Search for ss DNA was performed by hybridization (te Riele et al., 1986a). Comparisons of nt sequences were carried out using the DNA/ protein Sequence Analysis System of IBI (New Haven, CT).
RESULTS AND DISCUSSION (a) pTB52 and pAMpl do not accumulate ss DNA Replication of small plasmids from Gram + bacteria leads to accumulation of easily detectable amounts of ss DNA (cf., Gruss and Ehflich, 1989, for review). We wanted to determine whether the replication of two large plasmids, pTB52 (a derivative of pTB 19 carrying the repA replication region) and pAM/~I, originated from a thermophilic Bacillus and Streptococcus faecalis respectively (Imanaka et al., 1984; Clewell et al., 1974), also leads to such accumulation. Since it is not always easy to detect ss DNA larger than
plL2.52 pRATII pUBII0 -
about 10 kb (unpublished results) we used small derivatives of the two large plasmids, pRATI 1 (a 2.6-kb derivative of pTB52; lmanaka et al., 1986) and plL252 (a 4.6-kb derivative of pAMfll; Simon and Chopin, 1988). To search for ss DNA, the total DNA of B. subtY/s strains harboring plasmids pRATII or plL252 was extracted. The samples, treated or not with S 1 nuclease, were electrophoresed in agarose gels, denatured, transferred onto a nylon f'dter and hybridized against 32p-labeled homologous probes. Results (Fig. 1,A and B) show no difference in the hybridization patterns of S l-treated and untreated samples in spite of prolonged autoradiography. In contrast, in the total DNA extracted from strains harboring the SS + plasmid pUB II0 (known to accumulate low amounts of ss DNA; te Riele et al., 1986a), the fastest migrating band is eliminated by S 1 treatment. Furthermore, under conditions allowing detection of ss DNA only (no denaturation of DNA prior to transfer onto nitrocellulose filter and long exposure), a signal was detected with pUB 110 but not with pRATI 1 and plL252 (Fig. IC). From these experiments, we estimate that less than 0.5~ (a limit of detection in the experiment) of pRATII and plL252 might be ss DNA, whereas about 3% of pUB 110 was ss DNA. Nb treatment was shown to result in accumulation of ss and ds DNA of the plasmid pGRB-I in Halobacterium GRB (Sioud et al., 1988). We therefore exposed B. subtilis cells harboring pC 194, pUB110, pIL252 or pRATI 1 to Nb (5-10/Ag/ml, a dose which inhibits cell growth). No accumulation of ss or ds DNA was observed with any of these plasmids (data not shown), which suggests that gyrase activity is necessary for their replication in B. subtilis.
plL2S2 pRATII pUBII0 , -
plL2$2 pRATII pUBII0 -
,,sU O Fig. 1. Searchfor ss DNA. Total DNA extracted from B. subtilis cells harboringpIL252(derivedfrom pAM/31),pRAT!! (carryingthe repA regionof pTBI9) or pUB110,treated ( + ) or not ( - ) withthe nucleaseSI, wereanalysedby 0.7% agarosegel electrophoresis.Samples,denatured(A, B) or not (C), were transferredonto a filter and hybridizedwith homologous 32p probes; ss DNA is indicated by arrows. The denatured sampleis shownalter a short autoradiography(A) which allowsto estimatethe relative amountsof ds and ss DNA. Longautoradingraphies(B, C) showa strong ss DNA signal in the pUBII0 sample and absenceof such signalin plL252 and pRATI1 samples.
Fig. 2. Construction of cloning vectors. Single lines, open boxes and blackened boxes represent pBR322, the C m R gene of pC194 and pHV1301 sequences,respectively.White-dottedboxes, hatched box and black-dotted box correspond to the replicationregion ofpTB52, the ErR gene ofpAM~l, and the KmRgene of pUB110, respectively.Open arrow indicates the orientationof the pTB52 repA gene. pHVI301 is a deletion derivative of pAMpl arisen spontaneouslyin B. subtiliscells (P. Noirot, personal communication; Simon and Chopin, 1988). To construct pHVI431, the replication region of pHVi301, carried by the HindlIl segments D and F of pAM~I was introduced into the Hindlll site of pHV60. Deletionofthe smallEcoRl segmentofpHVl431 gave pHV1432. Inversion in pHVI431 of the Htndlll segments carryingthe replication region ofpHVl301, yieldedpHV1435. To construct pHVI436, the replication regionofpTB52carried by the largeEcoRI segmentofpRATl was inserted into the EcoRl site of pHV60.
(b) Deletion frequency in pTB52 and p A M p l
Precise excision of transposons such as Tn5 and TnlO occurs by recombination between 9-bp repeats which flank
transposon sequences. This process is considered to be representative of deletion formation generated by recombination between short homologous sequences (cf., Ehrlich, 1989, for review). To evaluate deletion frequency in pTB52 and pAMfil, we determined excision frequencies of two Tn5 insertions carded on derivatives of these plasmids. Relevant constructions are the following. The E. coli plasmids p C l i and pC2i (Janni~re and Ehrlich, 1987) carry the transposon Tn5 inserted into their Cm R gene (insertion C1 for plasmid p C l i and C2 for pC2i). Into the EcoRI site of these plasmids, we introduced EcoRI segments carrying the replication region of pAMfil or pTB52. These segments originated from pHV1435 (Fig. 2) and pTB53, a plasmid carrying the replication region of pTB52 (Imanaka et al., 1984) respectively. This yielded four plasmids named p/H-C1 and -C2, p52-C1 and -C2, capable of replicating in B. subtilis. Precise excision of Tn5 should restore a functional Cm s gene in these four plasmids, and lead to appearance of B. subtilis Cm R cells. To calculate the excision frequency of the Tn5 element, we determined the number (N) of Cm s cells required to generate the first Cm R cell, using the fluctuation test of Luria and Delbrflck (1943), and then the copy number (C) of TnS-carrying plasmids per chromosome, using densitometric analysis (Janni~re and Ehrlich, 1987). The excision frequency (F) was then calculated according to the equation F ffi (N x C ) - t . This method enables the determination of excision frequencies irrespective of the possible difference in growth rate of cells harboring transposon-free or transposon-carrying plasraids, and has already been used to measure Tn5 excision frequencies in B. subtilis (Janni~re and Ehrlich, 1987). The
TABLE II Tn$ excision frequencyfrom various re#icons T~ insertiona
Number of Cms cells yielding one CmR cell
Tn$ copy number per chromosome
Excision frequency per TnJ
(F = "------~l) d
Relative excision frequency
pTB52 pAMpl chromosomee pC194c
2 :< l0s 2.5 x 107 1.2 x 10I° 2 × l0 s
14 125 1 10
3.5 z 3.2 z 8.3 x 1.3 x
10-to 10-io 10-tt 10-7
1 0.9 0.2 370
pTB52 pAMpl chromosome° pC194°
1 x l0 s 3.8 x 106
8.3 x 10-,o 2.1 x 10 - 9 2.5 x 10-9 7.7 x 10-7
2.5 3 925
x l0 s
1.3 x 105
a The site of Trt5 insertion in CI and C2 differs. b Number of SB202cells harboring transposon-carryingplasmids required to yieldthe first bacteria harboring transposon-freeplasmids at 37°C. Cms or CmR refer to cells harboring TnS-free or TnS-carryingplasmids, respectively. c Determined by densitometric analysis (cf., Janni~re and Ehrlich, 1987). d Tn.~excision frequencyper transposon-carryinggenome. • Previous results (Janni~re and Ehrlich, 1987).
57 excision frequencies of Tn5 carried on pTB52 or pAM/~I derivatives are shown in Table II. Comparison of these values to those determined previously (Janni~re and Ehrlich, 1987) shows that Tn5 excises 400 to 900 times less frequently from pTB52, pAM~I and the B. subtilis chromosome than from the SS + plasmid pC194. (c) Characteristics of cloning vectors Four cloning vectors (pHV1432 to pHV1436) were constructed as described in Fig. 2. The complete sequence of pHVI432 is available (pAM~I, origin: Swinfield etal., 1989; pBR322, Sutcliffe, 1979, corrected by Peden, 1983, and Watson, 1988; pC194, Horinouchi and Weisblum, 1982, corrected by Dagert et al., 1984; pHV60, Michel and Ehrlich, 1986). The sequence of the remaining region of pAM/~I, which is present in pHVI431 and pHVI435, will be published elsewhere. In the vector derived from pTB52, out of 2900 bp originating from pRATI, 1631 bp were sequenced (lmanaka et al., 1986). The following unique restriction sites, mapping in the pBR322 region, are present in the four vectors: AatlI, AmI, BamHI, EcoRl (only in pHV1432), EcoRV, NruI, Pstl, Sail and Sphl. The vectors' average copy numbers, determined at 37 °C, varied between about 5 and 460 (Table III). The high copy number of the pAM/~l-based vectors, relative to that of pAM/11 (~ 7 in Streptococcus; Leblanc and Lee, 1984) might be due to the deletion of the pAM/~I copy control region, localized between bp 1745 and 2161 (Swinfield et al., 1990). Derivatives of pAMpl are thermosensitive, since their copy number decreased sharply above 42°C.
TABLE I!I Vector characteristics Vector a
pHVI431 pHVI432 pHVI435 pHVI436
pAMpl pAMpl pAM#I pTB52
Cow number per
190 460 190 w.5
1.0 I x 10 -4 !.0 0.4
a See legend Fig. 2 and Table 1. b See Table II, footnote c. c B . subtilis SB202 cells harboring plasmids were grown overnight at 37°C in LB liquid medium supplemented with 10/Jg Cm/ml. The culture was diluted and bacteria were inoculated into LB liquid medium free of antibiotic and grown for approx. 25 generations at 37°C. The cells were then plated on the solid medium devoid of Cm and the colonies replicaplated on selective (Cm-containing) and nonselective (Cm-free) plates. Plasmid stability was estimated from the ratio of antibiotic-resistant and total cell counts (> 500 colonies were analysed). These values reflect both the probability of appearance of plasmidless cells from plasmid-bearing cells and the enrichment of such cells during growth in liquid culture.
For example, the copy number of pHV1432 was 460 at 37°C, 250 at 420C, 130 at 440C, 30 at 45°C and 10 at 46 °C. At 480C and 51 °C, the cells grew very slowly and the plasmid copy number was therefore not measured. The segregational stability of the different vectors carried in B. subtilis strain SB202 at 37°C was determined. Two pAMpl derivatives, pHVI431 and pHVI435, were stable, the other pAMpl derivative, pHV 1432, was highly unstable and the pTB52 derivative, pHVI436, was moderately unstable (Table Ill). The segregational instability of pHVI432, which has a high copy number (460), seems to be the result of two factors. One is plasmid polymerisation, which decreases the number of segregation units (Summers and Sheratt, 1984), the other is a slow growth rate of cells which harbor the plasmid (~, 40 min/generation compared to 20m in/generation of the plasmidless cells). The functions required for pAMpl segregational stability are being studied. Two of the four plasmids, pHVI432 and pHVI436, were structurally stable in E. coli,while the other two, pHVI431 and pHVI435, were not and underwent extensive rearrangements in this host (data not shown). This structural instability may be the consequence of a lethal effect of diffusibleproduct(s), encoded by the last two plasmids, on the E. coil cells (L.J., unpublished results). (d) Cloning efficiency The efficiency of cloning in vectors derived from pAMpl and pTB52 was examined as follows. Plasmids pHVI431, pHVI432, pHV!435 or pHVI436, cut by the restriction enzyme BamHI, were mixed with phage ,~DNA previously polymerized and restricted by Bg/II, at a ratio 1:3 (w/w; plasmid/phage) and a final DNA concentration of about 300/zg/ml. After ligation, the four DNA mixtures were used to transform competent cells of the B. subtilis strain MT119 (Tanaka, 1979), which lacks the restriction-modification BsuM system (Guha, 1985; Bron et al., 1987). Use of such cells increases the cloning efficiency of large DNA segments (Haima et al., 1987; 1988; see below). Routinely, 104 transformants per/~g of DNA were obtained, which is fve- to tenfold less than with intact vectors. Plasmid DNA was extracted from a total of 150 independent transformants obtained with the four DNA mixtures. Agarose-gel electrophoresis showed that 20% of the transformants harbored a plasmid larger than the cloning vector. The size of the inserts, deduced from plasmid size and restriction analysis, is shown in Table IV. It ranged from 0.6 kb (the smallest BglII segment of phage DNA detectable in this experiment) to 43 kb, the average varying between 14.8 and 21.7 kb. Furthermore, 70-100% of the inserts were larger than 10 kb. These results show that the previously observed low-insert size in typical cloning experiments cannot be due to the competent state ofB. subtilis cells (Ostroff and P~ne, 1984).
58 TABLE IV DNA cloning Vector ~"
pHVi431 pHV1432 pHV1435 pHVI436 pHV1432 pHV33
pAM/~I pAM/31 pAM~I pTe52 pAM~I pC194
Cloned DNA c Source
Size range (kb)
Average size (kb)
Proportion of cloned inserts > 10 kb (%)
Phage ~. Phage A Phage ;t Phage ,1. B. amyloliquefaciens B. amyloliquefaciens
1.5-43 0.6-33 11 -24 17 -35 3.4-17 0.1- 1.7
15.5 14.8 15.1 21.7 8.8 0.7
70 75 100 100 25