J. Mol. Biol. (1992) 226, 959-977
Cell-cycle Control of a Cloned Chromosomal Origin of Replication from Caulobacter crescentus Gregory T. Marczynski and Lucille Shapiro Department
of Developmental Biology, Beckman Stanford University School of Medicine Stanford, CA 94305, U.S.A.
(Received
Center
17 December 1991; accepted 23 March
1992)
Caulobacter crescentus cell division is asymmetric and yields distinct swarmer cell and stalked cell progeny. Only the stalked cell initiates chromosomal replication, and the swarmer cell must differentiate into a stalked cell before chromosomal DNA replication can occur. In an effort to understand this developmental control of replication, we employed pulsed-field gel electrophoresis to localize and to isolate the chromosomal origin of replication. The C. crescentus homologues of several Escherichia coli genes are adjacent to the origin in the physical order hemE, origin, dnaA and dnaK,J. Deletion analysis reveals that the minimal sequence requirement for autonomous replication is greater than 430 basepairs, but less than 720 base-pairs. A plasmid, whose replication relies only on DNA from the C. crescentus origin of replication, has a distinct temporal pattern of DNA synthesis that resembles that of the bona jide C. crescentus chromosome. This implies that &s-acting replication control elements are closely linked to this origin of replication. This DNA contains sequence motifs that are common to other bacterial origins, such as five DnaA boxes, an E. co&like 13-mer, and an exceptional A+T-rich region. Point mutations in one of the DnaA boxes abolish replication in C. crescentus. This origin also possesses three additional motifs that are unique to the C. crescentus origin of replication: seven 8-mer (GGCCTTCC) motifs, nine 8-mer (AAGCCCGG) motifs, and five 9-mer (GTTAA-n7-TTAA) motifs are present. The latter two motifs are implicated in essential C. crescentus replication functions, because they are contained within specific deletions that abolish replication. Keywords:
CauZobacter
crescentus;
replication origin; replication sequences
1. Introduction
cell-cycle
control;
DnaA
boxes;
control of bacterial chromosome replication (Bramhill & Kornberg, 1988b; Norris, 1990; Bremer & Churchward, 1991). Unlike other bacteria, Caulobacter crescentus exhibits an asymmetric control of DNA replication in its progeny cells. Each asymmetric cell division yields a swarmer cell and a stalked cell (Poindexter, 1964), but chromosomal replication only occurs in the stalked cell (Degnen & Newton, 1972; Dingwall & Shapiro, 1989; Marczynski et al., 1990). Therefore, the C. crescentus cell division cycle (Fig. 1) presents the basic control problems that are common to most developing organisms: the establishment of polarity in the predivisional cell, and the commitment to express cell-type-specific biochemical potentials, exemplified by the unique transcriptional and replicative abilities of the stalked and swarmer cells. Also, the coupling between cellular differentiation and proliferation is exemplified by the C. crescentus swarmer cell to stalked cell transformation, because this transformation is always accompanied by an
In Escherichia coli, chromosomal DNA replication is controlled by regulating the initiation frequency of DNA synthesis from one specific origin of replication (oriC: von Meyenburg & Hansen, 1987). E. coli cells initiate chromosome DNA replication when a critical ratio of cell mass to oriC DNA is reached (Donachie, 1968). Cis-acting DNA elements required for this replication control apparently overlap the E. coli origin of replication (Helmstetter & Leonard, 1987). Although the E. coli chromosomal origin of replication is the most extensively studied origin (Kornberg & Baker, 1992), chromosomal origins of replication have also been isolated from other Enterics (Zyskind et al., 1983), and several evolutionarily distant bacteria, including Pseudomonas (Yee & Smith, 1990), and Bacillus subtilis (Seiki et al., 1981). Recently, a number of biochemical and physiological studies have been integrated into models that attempt to explain the
959 0022-2836/92/160959-19
$08.00/O
c
1992 Academic
Press
Limited
960
G. T. Marczynski
and L. Shapiro labeling experiments suggest that the C. crescentus chromosome, like the E. coli chromosome, has only one origin of bidirectional replication (Lott et al., 1987; Dingwall & Shapiro, 1989). Therefore, we reasoned that C. crescentus chromosome replication, like E. coli chromosome replication, would be controlled primarily by regulating the initiation of DNA synthesis, and we designed experiments to precisely map and isolate its origin of replication. One goal of these experiments is to define the cisacting elements that mediate the temporal control of DNA replication. This paper reports the identification and isolation of the C. crescentus origin. We demonstrate that C. crescentus has one origin of replication that has both significant sequence similarities and significant differences when compared with other chromosomal origins of replication. Using an autonomously replicating plasmid driven by the C. crescentus origin of replication, we demonstrate that both conserved and unique sequences are required for replication. Also, the cloned C. crescentus origin is subject t’o the same cell-type-specific replication controls as the bona$de chromosome, and it is therefore a valuable substrate for molecular and biochemical studies of t’he different replicative abilities of the progeny chromosomes.
Figure 1. A diagram of the C. crescentus cell cycle. The predivisional cell shown on the left contains 2 fully replicated chromosomes (depicted as thick ovals). Cell division produces 2 distinct progeny: the motile swarmer cell, with its polar flagellum and pili, and the non-motile stalked cell. Chromosomal DNA replication only initiates in the stalked cell. A swarmer cell does not replicate its chromosome until it differentiates into a stalked cell. The stalked cell produced by swarmer cell differentiation is termed an “inexperienced” cell and the stalked cell produced by cell division is termed an “experienced” cell. Relatively early atid late chromosomal replication intermediates are drawn as theta-like structures. Cell growth accompanies chromosomal DNA replication, and swarmer cell-specific, as well as stalked cell-specific, components are synthesized and positioned to produce the asymmetric predivisonal cell. At each stage of the cell cycle, C. crescentus cells not only differ in their morphology and biochemical composition, but they also differ in the transcriptional and repli-
cative abilities of their chromosomes. activation of chromosomal DNA replication (Degnen & Newton, 1972; Dingwall & Shapiro, 1989; Marczynski et aE., 1990). Despite its apparent complexity, DNA replication during the C. crescentus cell-cycle is amenable to analysis. It is easy to synchronize cells for biochemical studies (Evinger & Agabian, 1977), and conditional mutants have been isolated that block DNA synthesis and cell division (Osley & Newton, 1977). C. crescentus has one circular chromosome that has been extensively mapped by classical genetics (Ely, 1987), and more recently, by pulsed-field gel electrophoresis techniques (Ely & Gerardot, 1988; Ely & Ely, 1989; Ely et al., 1990). DNA
2. Materials
and Methods
(a) Materials [3H]deoxyguanosine (5 to 11 Ci/mmol) was purchased from ICN Radiochemicals. [a-32P)dGTP (deoxyguanosine-5’-triphosphate: 3000 Ci/mmol) was purchased from Amersham Corporation. All antibiotics were purchased from Sigma. Ludox (colloidal silica) was obtained from the Du Pont company. Restriction endonucleases were purchased and used as specified by Boehringer-Mannheim. DPU’A sequencing reagents and kits were purchased from United States Biochemical Corporation. Oligonucleotides for site-directed mutagenesis were synthesized by Operon Technologies Incorporated. pSK( +)BluescriptII plasmids were obtained from Stratagene. (b) Strains
and plasmids
All strains and plasmids used in this work are listed in Table 1. E. coli strain TGl was used for routine DNA manipulations, including plasmid recovery, subcloning and single-stranded phage DNA preparations. E. coli strain 517-l (harboring the appropriate plasmid or cosmid) was used to mobilize these molecules into C. crescentus by bacterial conjugation (Ely, 1979). Specific C. crescentus strains were constructed by the sequential 4Cr30 phage transductions (Ely t Johnson, 1977) that are indicated in Table 1. Routine plasmid constructions, and subclonings from COSMIDS I, II and III into vectors pAGMT, pAGMT* and pSK( +)BluescriptII were performed by established protocols (Maniatis et al., 1982). (c) Cell synchrony
and cell labeling
C. crescentus strains were grown in PYE medium (Poindexter, 1964) or in M2G medium (Ely & Johnson, 1977). C. crescentus cells harboring plasmids were grown
Properties of the Cloned Origin from
961
Caulobacter
Table 1 Strains and plasmids Strain E. coli TGl s17-1 CJ236 c. crescentus CB15 CBl5N SC1912 SU87 PC2179 SC1293 St-423 SU431 SC443 CM5256 SU146 SC1090 AE5168 SC1107 SU53 SU133 PC7070 SU37 SU197 ST 197/262
Plasmid pRK2013 pRK2013 :: Tn5-132 pGR24 pLAFR-5 COSMID I COSMID II COSMID III pAGMT pAGMT* pAGMT*IE pSK( + )BluescriptII pSKCori262
Genotype
Reference
A(&-pro), supE, thi, hsdD5lF’ proA+B+‘, lacP, kzcZAM15 E. coli 294 :: RP4-2 (Tc :: Mu)(Km :: Tn7) dutl, ungl, thi, relAlpCJ105 Cm’
T. J. Gibson
Wild-type Synchronizable CB15 zzz-129 Tn7 (Str’) CB15N Tn7 CB15N X &r30[SC1912] dnaC (TS DNA synthesis) trpC : : Tn5 CB15 dnaC, trpC :: Tn5 PC2179 x ~Cr30[SC1293] CB15N dnaC, trpC :: Tn5 CB15N X @30[SU423] CB15N dnaC SU431 X &r30[CB15N] ret-526 (recombination-deficient) CB15N ret-526 purA :: Tn5 (adenosine auxotroph) gpsA-505 (glycerol-3-P auxotroph) bla :: Tn5 (ampicillin-sensitive) CBl5N bla :: TnS(Km’, Str’) CB15N X &r30[SC1107] CB15N bla :: Tn5-132(Tet’) SU53 X pRK2013 :: Tn5-132(Tet’) rec.526 - Tn5 (linked transposon) CB15N ret-526 -Tn5 CBl5N X &r30[PC7070] CB15N ret-526-Tn5, bla :: Tn5-132 SU133 X @r30[SU37] SU197 containing plasmid pSKCori262
Poindexter (1964) Evinger & Agabian Ely (1987) This study
Characteristics
Reference
Km’, conjugation helper, on ColEl replicon Tet’ Tn5 on pRK2013
RK2
transfer
Simon Kunkel
et al. (1983) et al. (1987)
(1977)
Osley & Newton (1977) Winkler et al. (1984) This study This
study
This
study
O’Neill et al. (1985) Marczynski et al. (1990) Ely (1987) Contreras et al. (1979) Ely (1987) This st,udy
A. Newton This study
genes
pBR322+ C.C. dmK in BamHI site Tc’, IncP-1 replicon, cosmid vector pLAFR-5 + early replicating CB15N DNA pLAFR-5+ early replicating CB15N DNA pLAFR-5 + early replicating CBISN DNA Gm’ pACYC184 derivative, Tc’, RK2 oriT Spontaneous Tc” derivative of pAGMT pkGMT* + BamHI fragment I? from COSMID I pUC derivative (pMB1 replicon), Ap’ pSK( + )BluescriptII + PstI to BamHI Cori DNA
media supplemented with the following antibiotics: 1 pg tetracycline/ml for pLAFR-5 cosmids, 2.5 pg gentamycin/ml for pAGMT and pAGMT* plasmids, and 10 pg ampicillin/ml for pSK( + )BluescriptII plasmids. Cell density was determined by measuring the absorbance at 660 nm. Cell cultures were synchronized by collecting the more dense swarmer (CBlBN variant) cells from a Ludox (colloidal silica) gradient and returning them to fresh M2G medium (Evinger & Agabian, 1977). Protocols for labeling synchronous C. crescentus cells in
or source
This
study
This
study
Figurski
or source
& Helinski
(1979)
Berg & Berg (1983); B. Ely Games et al. (1990) Keen et al. (1988) This study; Fig. 4 This study; Fig. 4 This study; Fig. 4 Alley et al. (1991) This study; Fig. 6 This study; Fig. 7 Short et al. (1988) This study; Fig. 9
with [3H]deoxyguanosine have been reported (Dingwall et al., 1990; see Fig. 3). We also observed that C. crescentus cells could be efficiently labeled with [a-32P]dGTP (see Figs 5 and 11). This was unexpected, because it was believed that phosphatases would remove the a-phosphate before the dGTP could reach the DNA polymerase. However, under the labeling conditions described for Figs 5 and 11, we observed that 32P was incorporated only into DNA (and with linear kinetics) from time t = 0 to 2 min. Subsequent incorporation became progressively slower,
962
G.
T.
Marczynski.and
and some of the incorporation became base labile (presumably due to RNA synthesis). These observations suggest that C. crescentus has an efficient transport system for dGTP and/or dGMP. [c(-32P]dGTP uptake experiments indicated that swarmer cells and stalked cells internalized 32P with the same kinetics, but that only stalked cells incorporated “P into DNA. Incorporation into DNA (base stable cts/min) versus RNA (base labile cts/min) was determined by precipitation with trichloroacetic acid and scintillation counting essentially as described (Degnen & Newton, 1972). (d) DNA
preparation, sequencing, pulsed-jeld gel electrophoresis
mutagenesis analysis
and
Standard plasmid preparations from E. coli, digestions by restriction endonucleases, and agarose gel electrophoresis procedures were performed according to published protocols (Maniatis et al.. 1982). DNA sequencing was performed by the dideoxynucleotide chain termination method (Sanger et al., 1977) using the Sequenase System II kit (USB Corp.). Single strand DNA pSK( + )BluescriptII “phagemids” served as templates and were prepared from selected Gori DNA deletions (see Fig. 9), as described (Short et al., 1988). Oligonucleotide directed site-specific mutagenesis was performed essentially as described (Kunkel et al., 1987). COSMID I BamHI fragment IE (see Fig. 4) was inserted into M13mp18. and substrate single strand DNA was prepared from E. coli strain CJ236. Following the mismatchedoligonucleotide-directed synthesis reaction, both mutated and unmutated phage were recovered from E. coli TGl. and base changes were confirmed by sequencing. In order to assay DnaA box point mutations, both mutated (3 independently isolated DnaA box mutants) and unmutated IE fragments were subcloned from M13mp18 into the replication test vector pSK( + )BluescriptII (see Fig. 9). Pulsed-field gel electrophoresis (PFGET) techniques for C. crescentus, including in situ genomic DNA preparations, endonuclease digestion conditions, electrophoresis conditions, Southern blot PFGE hybridization procedures, [3H]DNA labeling of synchronized cells, and the preparations of [3H]DNA fluorograms. have been described in detail (Dingwall et al., 1990). (e)
Techniques
employed
to isolate
the origin
of replication
scale PFGE of early replicating C’. crescentus genomic DNA required embedding lo-fold higher cell densities inside the agarose plugs, but otherwise the procedures were the same as for analytical PFGE. To efficiently remove the DNA from the agarose, the excised bands were digested in situ with EcoRI and recovered by electrophoresis inside dialysis membrane bags. Following precipitation with ethanol, this DNA was labeled in vitro with [cr-32P]dGTP using Klenow DNA polymerase and the random primer pd(N)G (as specified by Boehringer-Mannheim), and hybridized to cosmid DNA clones immobilized on nylon membranes. The construction of this cosmid library (pLAFR-5 containing Sau3A partially digested CBlSN DNA) has been reported (Alley et al., 1991). Approximately 580 cosmids were screened, initially in pools of 10 and 40, and then rescreened individually. A total of 21 cosmids was repeatedly hybridized by the 110 kb early replicating Preparative
t Abbreviations used: PFGE, pulsed-field electrophoresis; kb, lo3 bases or base-pairs; pair(s).
gel bp, base-
L.
Shapiro
AseI band, and 17 cosmids were repeatedly hybridized by the 83 kb and 85 kb AseI bands (described in Fig. 3). In order to group these cosmids into overlapping sets, they were cut with BamHI, Southern-blotted, and crosshybridized, according to established protocols (Maniatis et al., 1982). One group of overlapping cosmids (shown in Fig. 4 as COSMIDS I, II and III) were chosen for further analysis based on the precise labeling experiment described in the text and the legend to Fig. 5. BamHI fragments were subcloned from these 3 cosmids into test plasmids pAGMT, pAGMT* and pSK( +)BluescriptII. The ability of these plasmids to replicate autonomously in C. crescentus was assayed by bacterial conjugation (Ely, 1979), direct plasmid electroporation into C. crescentus (Dower et al., 1988; Gilchrist & Smit, 1991), and by quantitative extraction of plasmid DNA from C. crescentus (Marczynski et al., 1990).
Tn7 250
110
v
AseI
r”“r
A
A 305
105
A,,5A
D~OI
3” k--83-+--q
Figure
i
The earliest replicating region of the chromosome. An alignment of infrequently and DraI restriction endonuclease sites is presented with interspersed genetic and molecular markers. Numbers indicate distances between markers in kb. The 2 hatched bars show the positions of the 2 earliest labeled DNA fragments (83 kb and 50 kb) separated by pulsed-field gel electrophoresis as described in the legend to Fig. 3. The positions of the AseI and DraI sites, the transposon Tn7 and genetic marker hi& are based primarily on published data (Ely & Ely, 1989). Tn7 contains both AseI and DTUI sites. The Tn7, hi&, gpsA, purA and nov markers were mapped to this region by genetic techniques (Ely, 1987). The physical positions of gpsA, purA and novR were determined by the hybridization of 32P-labeled cosmid DNA clones to genomic DNA fragments separated by pulsed-field gel electrophoresis (data not shown). Several cosmid clones were obtained that complemented both gpsA and purA auxotrophic requirements (see the text). The novR cosmid clone was obtained by complementing a temperature-sensitive allele of the novobiocin resistance gene (the E. coli gyrB homologue: M. Rizzo, L. Shapiro & J. Gober, unpublished results). The C. crescentus dnaK (Gomes et aZ., 1990) and dnaA (G. Zweig & L. Shapiro, unpublished results) markers are homologues to their respective E. coli genes. They were both obtained during the cosmid walk through this region (see the text and Fig. 4). The Tn5 insertion in the 83 kb AseI fragment was obtained by 4Cr30 phage transduction and linkage to a kanamycin resistance’ plasmid integrated at the dnaK gene, and the dnaK gene in turn showed 11 o/0 linkage to the purA gene by $Cr30 phage transduction. C. crescentus cutting AseI
2.
1
Properties
of the Cloned
3. Results
(a) Mapping
the earliest replicating region of the C. crescentus chromosome
We localized the origin of replication on the physical map of the C. crescentus genome prepared by B. Ely and co-workers using pulsed-field gel electrophoresis (Ely & Gerardot, 1988; Ely & Ely, 1989). Infrequently cutting DraI and AseI restriction endonucleases generate large genomic fragments that were ordered and aligned to form a circular map of this genome (Ely et al., 1990). Dingwall & Shapiro (1989) have presented evidence that the earliest replicating region of the C. crescentus chromosome lies within a specific 305 kb DraI or an adjacent 105 kb DraI DNA fragment, and that replication proceeds bidirectionally. An expanded view of this part of the C. crescentus genome, as well as an alignment of both DraI and AseI-cut sites, is presented in Figure 2. To precisely map the earliest replicating DNA fragments, swarmer cells were isolated and incubated with [3H]deoxyguanosine under standard growth conditions that allowed these cells to proceed t’hrough the cell-cycle. Equal cell volumes were removed at seven minute intervals that spanned the initiation period of DNA synthesis (i.e. the swarmer cell to stalked cell transition of the cellcycle). These cells were embedded in agarose, their genomic DNA was prepared in situ, digested with the restriction endonuclease AseI, and separated by PFGE. Figure 3 shows (a) the ethidium bromidestained gel and (b) its corresponding 3H fluorogram for this experiment. Figure 3(a) shows that equal amounts of DNA were loaded in each of the lanes 1
Nl2345678Tn7
123456
Origin
from
Caulobacter
963
to 8; (b) shows that DNA synthesis is absent in the swarmer cells shown in lanes 1 to 4, and that DNA synthesis initiates as the swarmer cells develop into stalked cells in lanes 5 to 8. Figure 3(b) also shows that there are only two early labeled bands, and that they are labeled with nearly identical kinetics. These observations indicate the origin of DNA replication lies within one of these two AseI DNA fragments and, most likely, it either overlaps or lies near their boundary. In order to unambiguously identify the earliest labeled PFGE bands, we employed strains with a transposon inserted in the early replicating region. The experiment in Figure 3 employed a strain containing transposon Tn7 that has a unique insertion site in the C. crescentus genome (Ely, 1982). We mapped the position of the Tn7 as indicated in Figure 2. Tn7 contains an AseI site(s), and therefore an extra 110 kb band is seen in Figure 3(a) that is not present in the wild-type strain (compare the outside control lanes containing unlabeled wild-type CB15N DNA and CB15N Tn7 DNA: Fig. 3(a)). This 110 kb “Tn7” band is the larger of the two earliest replicating bands seen in Figure 3(b), because its position in the 3H fluorogram is aligned with the 110 kb band in the ethidium bromide-stained gel. Also, when the experiment presented in Figure 3 is repeated with the wild-type CB15N strain, only the 83 kb and a new high molecular weight band (migrating close to the wells) are seen during this early DNA synthesis period (data not shown). The smaller of the two earliest replicating AseI bands in Figure 3(b) measures between 80 and 90 kb relative to a lambda ladder (data not shown). It comigrates with either the 85 or the 83 kb AseI
78
12345678
(b) (c) (al Figure 3. Time-course analysis by PFGE of the earliest replicating C. crescentus DNA. Strain SU87 (CB15N Tn7) swarmer cells (10 ml at a cell density of @085) were isolated and resuspended in M2G media, as described in Materials and Methods. At time t = 0 min, 25 PCi of 3H-labeled deoxyguanosine (5 mCi/mmol) was added, and the cells were vigorously shaken at 30°C. Portions (1.0 ml) (corresponding to lanes 1 through 8) were removed at times 1= 0, 7, 14, 21, 28, 35, 42 and 49 min. These cells were immediately embedded in agarose and lysed in situ to prepare their genomic DNA for PFGE analysis (see Materials and Methods). Total genomic DNA was digested with AseI in (a) and (b) and doubledigested with AseI and DraI in (c). Approximately @05 of the prepared DNA w&s loaded in each lane. Subsequent procedures for PFGE and 3H fluorography were as reported (Dingwall & Shapiro, 1989). (b) and (c) Fluorograms exposed t,o X-ray film at -70°C for 1 to 2 days. (a) The ethidium bromide-stained gel corresponding to (b). Unlabeled C. crescentus genomic DNA prepared from wild-type strain CB15N and its isogenic Tn7 derivative (CB15N Tn7) were loaded in the lanes marked N and Tn7, respectively. Resolution was achieved by running these gels for 15 h at 6°C with 250 V and employing 7 s switching times.
964
G. T. Marczyneki
bands identified by Ely & Ely (1989), which are just barely resolved by the PFGE conditions of Figure 3(a). We have identified this as the 83 kb band, because the 83 kb band was reported to be adjacent to the Tn7 containing AseI band, and it overlaps the 305 kb DraI band (see Fig. 2). The 85 kb AseI band is located on the other side of the genome (Ely & Ely, 1989). Also, we have identified a Tn5 insertion in the early replicating region (that is genetically linked to the dnaK gene and within the 305 kb DraI band) that shifts the size of the 83 kb AseI band (see Fig. 2). The third prominent band that becomes labeled in lanes 7 and 8 in Figure 3(b) is the 250 kb AseI band that lies adjacent to the 83 kb AseI band on the genomic map (Ely & Ely, 1989). The ‘H fluorogram of the PFGE experiment in Figure 3(c) further supports the assignments of the 110 and 83 kb AseI bands. The same DNA samples analyzed in Figure 3(b) were digested with both AseI and DraI, and two bands were identified at the earliest time of DNA replication (lanes 5 and 6, Fig. 3(c)). The larger band in Figure 3(c) comigrated with the 83 kb band seen in Figure 3(b). The smaller band is approximately 50 kb, and it is produced by DraI cutting within the 110 kb ‘In7 AseI band. These results are consistent with the restriction map presented in Figure 2, and demonstrate that the origin of replication must lie within the 133 kb DNA region shown as the two hatched bars overlapping the right arm of the 305 kb DraI fragment. Since these two bands are labeled simultaneously, these data also suggest that the origin of replication lies near their central AseI site. (b) Genomic
clones that span replicating region
the earliest
We cloned a large region of the C. crescentus genome that spans the origin of DNA replication before proceeding to isolate autonomously replicating sequences. This strategy was motivated by four major considerations. (1) A direct screen of a total genomic library failed to isolate any C. crescentus DNA that would support autonomous plasmid replication in C. crescentus (data not shown; O’Neill et al., 1983). (2) DNA sequences that support autonomous plasmid replication may not necessarily be the physiologically relevant origin(s) of chromosomal DNA replication. (3) The origin of DNA replication may be distant from its control elements. (4) Genes involved in DNA metabolism are often linked to bacterial origins of DNA replication (Ogasawara et al., 1990), and these would be valuable assets in future studies. We employed preparative scale PFGE to isolate the DNA from the 110 and 83 kb earliest replicating AseI bands that were identified in the experiment presented in Figure 3. These DNA preparations were separately labeled in vitro with 32P and hybridized to a C. crescentus CBl5N genomic cosmid library as described in Materials and Methods. Approximately 580 pLAFR-5 cosmids, with average genomic inserts of 25 kb, were
and L. Shupiro
screened by hybridization to 32P-labeled DNA from the 110 kb AseI band. Twenty-one independent cosmid clones were isolated as judged by their unique BamHI digestion patterns. Three of these cosmid clones contained genes that complemented the gpsA and the purA mutations (see Fig. 2). Since both gpsA and purA map genetically to the left of Tn7 (see Fig. 2), their genes were expected to be present in this cosmid collection. Likewise, the same cosmid library was screened by hybridization to the early replicating 83 kb AseI band (and also the contaminating 85 kb band). Seventeen independent cosmid clones were obtained from this second screen. One cosmid was isolated by both screens, and therefore should span the border between the 110 and the 83 kb early replicating bands. This was confirmed by hybridizing the cloned cosmid DNA to a pulsed-field gel Southern blot containing CB15N Tn7 DNA cut with AseI. As expected, this cosmid clone, designated COSMID I in Figure 4, hybridized to the 110 and the 83 kb band (see the legend to Fig. 2). COSMID I contains a unique AseI site within its BamHI fragment IE. Since the results from the experiment in Figure 3 suggested that this is either at or near the origin of DNA replication, we chose to concentrate our efforts on this cosmid and its adjacent overlapping cosmids. As described in Materials and Methods, a series of overlapping cosmid clones was obtained that span approximately 70 kb across the border between the 110 kb and the 83 kb AseI early replicating bands. TWO more cosmids were identified from the 83 kb band hybridization screen by their BamHI Southern blot cross-hybridization patterns with COSMID I (these are COSMIDS II and III shown in Fig. 4). Also, two cosmids were identified from the 110 kb band hybridization screen that overlap the right side of COSMID I (data not shown). The three cosmids presented in Figure 4 were chosen for more detailed analysis based on the results of the precise labeling experiment presented below. (c) Precise
labeling of&%? C. crescentus origin DNA replication
of
Labeling the origin of replication requires a very synchronous initiation of DNA synthesis and a significant reduction of the fork movement rate. We accomplished this by using a conditional (temperature-sensitive) DNA synthesis mutant (Osley & Newton, 1977) and the gyrase inhibitor novobiocin. Swarmer cells from C. crescentus strain SU443 (CB15N dnaC) were incubated at the nonpermissive temperature (37°C) in the presence of 10 pg novobiocin/ml. This level of novobiocin inhibits DNA synthesis at the permissive tempera(data not shown). Under these ture by -90% conditions we observed that the swarmer cells differentiated into stalked cells, but that chromosomal replication was blocked at the start of DNA synthesis. This synchronous cell culture was downshifted from 37°C to 30°C and rapidly labeled with
Properties of the Cloned Origin
965
\
(A$) , ,, ,
I
I3 B
I
I
from Caulobacter
5kb
I
?,
In
D
A
F 1 I
A
2 I
c!l D;C
IA -
II
hemE
I
cfnuA
0
I
I
dnoK,J
Zn YWO 32P distribution
Cod
I
IIIC
Figure 4. Alignment of cosmid clones containing the origin of replication. The DNA inserts from 3 overlapping cosmid clones (COSMIDS I, II and III) are aligned with both the genomic map described in Fig. 2 and the [32P]DNA fragments generated by the experiment described in Fig. 5. Short vertical bars mark the BumHI sites on the cosmid clones. Capital letters A, B, C, etc., denote the individual BumHI fragments within a cosmid clone. For example, IE denotes the 5th largest BarnHI fragment from COSMID I. This is the only cosmid BamHI fragment that contains an AseI site (shown as a filled arrowhead). Cosmid alignment with the C. crescentus chromosome was determined as described in the text and confirmed by the hybridization of in vitro-labeled cosmids to genomic DNA cut with AseI and resolved by PFGE. COSMIDS II and III only hybridized to the 83 kb AseI band, and COSMID I hybridized to both the 83 kb and to the 110 kb (Tn7) AseI bands. The bottom bar graph shows the in viva 32P distribution in this region of the chromosome approximately 30 s after the start of chromosomal DNA replication. This in viva-labeled DNA was assigned to these cosmids based on the comigration of the 32P-labeled bands (shown in Fig. 5) with unlabeled cosmids cut with BarnHI, S&I, Hind111 and AseI. The area in each bar is proportional to the 32P label in each corresponding band.
32P by mixing with medium containing [cr-32P]dGTP. Chromosomal DNA was isolated from these cells, digested with BarnHI, and resolved by standard agarose gel electrophoresis. An autoradiogram of this experiment reveals only five strongly labeled BamHI bands, and seven weakly labeled bands (Fig. 5). These bands clearly belong to the early replicating region, because when this in viva 32Pwas hybridized to cosmid labeled DNA preparation clones that were Southern-blotted on nylon membranes, this 32P-labeled DNA hybridized primarily to COSMIDS I and II, but only very weakly to COSMID III, and not at all to 20 other cosmids that were also isolated in our screen (data not shown). The combined lengths of the five
1 F ,...-,;
J
Figure
5.
Precise in vivo 32P-labeling at the C. crescentus chromosomal origin of replication. Strain SU443 (CB15N dnaC) swarmer cells (10ml at a cell density of 05) were isolated and incubated at 37°C for 90 min in M2G medium containing 10 pg novobiocin/ml. In order to quickly label and initiate replication, the temperature of this culture was down-shifted by mixing it with an equal volume of M2G medium at 23°C containing 250&i of [a-32P]dGTP (Amersham Corp., 3000 Ci/ mmol). In order to rapidly halt 32P incorporation and reduce labeled replication intermediates that migrate anomalously during agarose gel electrophoresis, unlabeled dGTP (0.2 mM final concn) was added 30 s after the temperature down-shift, and DNA synthesis was continued for an extra 5 min at room temperature. These cells were embedded in agarose, and their chromosomal DNA was prepared in situ (see Materials and Methods). Lanes B, S and H show the autoradiograms of this DNA preparation following digestion with BarnHI, Sal1 and HindIII, respectively, and resolution by standard agarose gel electrophoresis. Unlabeled cosmid DNA (COSMIDS I, II and III shown in Fig. 4) were digested with the same set of restriction endonucleases and run in the adjacent lanes (not shown). The migration of key cosmid bands cut with BamHI is shown on the left side. Most of the 32P incorporation is accounted for by these comigrating cosmid bands (see Fig. 4). The migration of lambda phage DNA cut with Hind111 (A) is shown on the right. Numbers indicate DNA size in kb. strongly labeled bands is approximately 33 kb. All five of these bands (designated IA, IIB, IB, IC and ID in Fig. 5) correspond to cosmid fragments presented in Figure 4, based on their comigration with the unlabeled DNA from cosmids I, II and III
966
G. T. Marczynski
loaded in adjacent ianes (not shown). Likewise, four out of the seven weakly labeled bands (designated IIIA, IE, IIIC and IF) were observed to comigrate with specific BamHI cosmid fragments. These banding assignments were confirmed by employing additional restriction enzymes such as SalI and HindIII, whose digestion patterns are also presented in Figure 5. We also observed, as would be expected from our banding assignments, that only the 32P-labeled band IE contains an AseI site. The pattern of j2P-labeled BamHI bands in Figures 4 and 5 suggests that the origin of replication lies within band IE, because IE is underlabeled relative to its adjacent bands, and our labeling protocol could allow DNA synthesis to initiate in the absence of 32P. Presumably, DNA synthesis initiates at the non-permissive temperature, because dn.aC is a temperature-sensitive elongation mutant (Osley & Newton, 1977). Also, a short period of unlabeled DNA synthesis must occur before the [a-32P]dGTP can enter the cells. These considerations predict that the site of initiation will be under-labeled relative to the flanking sequences. The bar raph in Figure 4 shows the quantification of the 38 P label in each of the cosmid BamHI fragments. Band IE is significantly under-labeled even when accounting for the size of its neighboring bands. The four bands (IA, ID, IB and IC) that surround band IE account for most of the 32P incorporation, and comparatively weakly labeled bands are seen further on either side. In summary, these results clearly localize the earliest replicating region of the C, crescentus chromosome to the DNA spanned by our cosmid clones, and they also suggest that chromosomal DNA synthesis initiates within the BamHI fragment IE.
(d) C. crescentus DNA that supports autonomous replication We reasoned that a DNA fragment containing the origin of replication should support the autonomous replication of a plasmid incapable of replicating in C. crescentus by itself. Therefore, we assayed DNA fragments from the early replicating region presented in Figure 4. Cosmid BamHI fragments were inserted into the test vectors pAGMT and pAGMT* that confer gentamycin resistance to host bacteria (Fig. 6). These plasmids can replicate in E. coli (from the P15A oriV), but not in C. crescentus, and they contain an origin of transfer (Trans) that allows efficient plasmid mobilization from E. coli to C. crescentus by bacterial conjugation (Guiney & Yacobson, 1983; Ely et al., 1979). Our initial replication assay scored gentamycinresistant C. crescentus colonies. After conjugation with E. coli cells containing specific subclones, drugresistant C. crescentus colonies would be indicative of plasmid maintenance. In order to prevent plasmid maintenance in C. crescentus by homologous recombination, we also employed an isogenic recombination-deficient (ret-) C. crescentus strain
and L. Shapiro
(5.67
kb)
P15A
OriV
P15A
OriV
h??HI Tc’
Tc’
Figure 6. Test vectors for autonomous replication in C. crescentus. Both plasmids pAGMT and its derivative pAGMT* replicate in E. coli, but not in C. crescentus. pAGMT contains the incP15A replicon from plasmid pACYC184 (oriV), tetracycline and gentamycin resistance genes (Tc’ and Gm’), and the origin of transfer from plasmid RK2 (Trans) for efficient mobilization by bacterial conjugation (M. R. K. Alley, personal communication). pAGMT* is derived from pAGMT by a spontaneous deletion that spans the AseI and Hind111 sites and removes the tetracycline promoter. This deletion is required for efficient autonomous replication by the cloned C. crescentus origin of replication in C. crescentus.
(SU146) that blocks colony formation by plasmid integration (O’Neill et al., 1985). From among all of the BamHI fragments of COSMIDS I, II and III, only pAGMT plasmids containing fragment IE could replicate in recombination-deficient C. crescentus cells. However, the frequency of mobilization for these plasmids was very low (m 196 gentamycin-resistant colonies per plate in our standard assay), suggesting a barrier to the maintenance of these plasmids in C. crescentus. When plasmid DNA was prepared from the C. crescentus transformants, extra DNA was found inserted between the AseI and BamHI sites of pAGMT (see Fig. 6). The size range (665 to 1.35 kb) suggested that these were C. crescentus insertion
Properties
(AS) ,/,I
of the Cloned
C0SMl0S I
Origin
from
Caulobacter
967
Cl234567891011
B
IDI
m
B
0
A
C
F
A
DEC
,
n B
,
I
Plasmid
ID, IIC UbYmE) IA, UA IF, UD IIB, me pACtAT* alone
Cl0 s-10” Cl0 =-IO5 Cl0
Cell-cycle Control of a Cloned Chromosomal Origin of Replication from Caulobacter crescentus Gregory T. Marczynski and Lucille Shapiro Department
of Developmental Biology, Beckman Stanford University School of Medicine Stanford, CA 94305, U.S.A.
(Received
Center
17 December 1991; accepted 23 March
1992)
Caulobacter crescentus cell division is asymmetric and yields distinct swarmer cell and stalked cell progeny. Only the stalked cell initiates chromosomal replication, and the swarmer cell must differentiate into a stalked cell before chromosomal DNA replication can occur. In an effort to understand this developmental control of replication, we employed pulsed-field gel electrophoresis to localize and to isolate the chromosomal origin of replication. The C. crescentus homologues of several Escherichia coli genes are adjacent to the origin in the physical order hemE, origin, dnaA and dnaK,J. Deletion analysis reveals that the minimal sequence requirement for autonomous replication is greater than 430 basepairs, but less than 720 base-pairs. A plasmid, whose replication relies only on DNA from the C. crescentus origin of replication, has a distinct temporal pattern of DNA synthesis that resembles that of the bona jide C. crescentus chromosome. This implies that &s-acting replication control elements are closely linked to this origin of replication. This DNA contains sequence motifs that are common to other bacterial origins, such as five DnaA boxes, an E. co&like 13-mer, and an exceptional A+T-rich region. Point mutations in one of the DnaA boxes abolish replication in C. crescentus. This origin also possesses three additional motifs that are unique to the C. crescentus origin of replication: seven 8-mer (GGCCTTCC) motifs, nine 8-mer (AAGCCCGG) motifs, and five 9-mer (GTTAA-n7-TTAA) motifs are present. The latter two motifs are implicated in essential C. crescentus replication functions, because they are contained within specific deletions that abolish replication. Keywords:
CauZobacter
crescentus;
replication origin; replication sequences
1. Introduction
cell-cycle
control;
DnaA
boxes;
control of bacterial chromosome replication (Bramhill & Kornberg, 1988b; Norris, 1990; Bremer & Churchward, 1991). Unlike other bacteria, Caulobacter crescentus exhibits an asymmetric control of DNA replication in its progeny cells. Each asymmetric cell division yields a swarmer cell and a stalked cell (Poindexter, 1964), but chromosomal replication only occurs in the stalked cell (Degnen & Newton, 1972; Dingwall & Shapiro, 1989; Marczynski et al., 1990). Therefore, the C. crescentus cell division cycle (Fig. 1) presents the basic control problems that are common to most developing organisms: the establishment of polarity in the predivisional cell, and the commitment to express cell-type-specific biochemical potentials, exemplified by the unique transcriptional and replicative abilities of the stalked and swarmer cells. Also, the coupling between cellular differentiation and proliferation is exemplified by the C. crescentus swarmer cell to stalked cell transformation, because this transformation is always accompanied by an
In Escherichia coli, chromosomal DNA replication is controlled by regulating the initiation frequency of DNA synthesis from one specific origin of replication (oriC: von Meyenburg & Hansen, 1987). E. coli cells initiate chromosome DNA replication when a critical ratio of cell mass to oriC DNA is reached (Donachie, 1968). Cis-acting DNA elements required for this replication control apparently overlap the E. coli origin of replication (Helmstetter & Leonard, 1987). Although the E. coli chromosomal origin of replication is the most extensively studied origin (Kornberg & Baker, 1992), chromosomal origins of replication have also been isolated from other Enterics (Zyskind et al., 1983), and several evolutionarily distant bacteria, including Pseudomonas (Yee & Smith, 1990), and Bacillus subtilis (Seiki et al., 1981). Recently, a number of biochemical and physiological studies have been integrated into models that attempt to explain the
959 0022-2836/92/160959-19
$08.00/O
c
1992 Academic
Press
Limited
960
G. T. Marczynski
and L. Shapiro labeling experiments suggest that the C. crescentus chromosome, like the E. coli chromosome, has only one origin of bidirectional replication (Lott et al., 1987; Dingwall & Shapiro, 1989). Therefore, we reasoned that C. crescentus chromosome replication, like E. coli chromosome replication, would be controlled primarily by regulating the initiation of DNA synthesis, and we designed experiments to precisely map and isolate its origin of replication. One goal of these experiments is to define the cisacting elements that mediate the temporal control of DNA replication. This paper reports the identification and isolation of the C. crescentus origin. We demonstrate that C. crescentus has one origin of replication that has both significant sequence similarities and significant differences when compared with other chromosomal origins of replication. Using an autonomously replicating plasmid driven by the C. crescentus origin of replication, we demonstrate that both conserved and unique sequences are required for replication. Also, the cloned C. crescentus origin is subject t’o the same cell-type-specific replication controls as the bona$de chromosome, and it is therefore a valuable substrate for molecular and biochemical studies of t’he different replicative abilities of the progeny chromosomes.
Figure 1. A diagram of the C. crescentus cell cycle. The predivisional cell shown on the left contains 2 fully replicated chromosomes (depicted as thick ovals). Cell division produces 2 distinct progeny: the motile swarmer cell, with its polar flagellum and pili, and the non-motile stalked cell. Chromosomal DNA replication only initiates in the stalked cell. A swarmer cell does not replicate its chromosome until it differentiates into a stalked cell. The stalked cell produced by swarmer cell differentiation is termed an “inexperienced” cell and the stalked cell produced by cell division is termed an “experienced” cell. Relatively early atid late chromosomal replication intermediates are drawn as theta-like structures. Cell growth accompanies chromosomal DNA replication, and swarmer cell-specific, as well as stalked cell-specific, components are synthesized and positioned to produce the asymmetric predivisonal cell. At each stage of the cell cycle, C. crescentus cells not only differ in their morphology and biochemical composition, but they also differ in the transcriptional and repli-
cative abilities of their chromosomes. activation of chromosomal DNA replication (Degnen & Newton, 1972; Dingwall & Shapiro, 1989; Marczynski et aE., 1990). Despite its apparent complexity, DNA replication during the C. crescentus cell-cycle is amenable to analysis. It is easy to synchronize cells for biochemical studies (Evinger & Agabian, 1977), and conditional mutants have been isolated that block DNA synthesis and cell division (Osley & Newton, 1977). C. crescentus has one circular chromosome that has been extensively mapped by classical genetics (Ely, 1987), and more recently, by pulsed-field gel electrophoresis techniques (Ely & Gerardot, 1988; Ely & Ely, 1989; Ely et al., 1990). DNA
2. Materials
and Methods
(a) Materials [3H]deoxyguanosine (5 to 11 Ci/mmol) was purchased from ICN Radiochemicals. [a-32P)dGTP (deoxyguanosine-5’-triphosphate: 3000 Ci/mmol) was purchased from Amersham Corporation. All antibiotics were purchased from Sigma. Ludox (colloidal silica) was obtained from the Du Pont company. Restriction endonucleases were purchased and used as specified by Boehringer-Mannheim. DPU’A sequencing reagents and kits were purchased from United States Biochemical Corporation. Oligonucleotides for site-directed mutagenesis were synthesized by Operon Technologies Incorporated. pSK( +)BluescriptII plasmids were obtained from Stratagene. (b) Strains
and plasmids
All strains and plasmids used in this work are listed in Table 1. E. coli strain TGl was used for routine DNA manipulations, including plasmid recovery, subcloning and single-stranded phage DNA preparations. E. coli strain 517-l (harboring the appropriate plasmid or cosmid) was used to mobilize these molecules into C. crescentus by bacterial conjugation (Ely, 1979). Specific C. crescentus strains were constructed by the sequential 4Cr30 phage transductions (Ely t Johnson, 1977) that are indicated in Table 1. Routine plasmid constructions, and subclonings from COSMIDS I, II and III into vectors pAGMT, pAGMT* and pSK( +)BluescriptII were performed by established protocols (Maniatis et al., 1982). (c) Cell synchrony
and cell labeling
C. crescentus strains were grown in PYE medium (Poindexter, 1964) or in M2G medium (Ely & Johnson, 1977). C. crescentus cells harboring plasmids were grown
Properties of the Cloned Origin from
961
Caulobacter
Table 1 Strains and plasmids Strain E. coli TGl s17-1 CJ236 c. crescentus CB15 CBl5N SC1912 SU87 PC2179 SC1293 St-423 SU431 SC443 CM5256 SU146 SC1090 AE5168 SC1107 SU53 SU133 PC7070 SU37 SU197 ST 197/262
Plasmid pRK2013 pRK2013 :: Tn5-132 pGR24 pLAFR-5 COSMID I COSMID II COSMID III pAGMT pAGMT* pAGMT*IE pSK( + )BluescriptII pSKCori262
Genotype
Reference
A(&-pro), supE, thi, hsdD5lF’ proA+B+‘, lacP, kzcZAM15 E. coli 294 :: RP4-2 (Tc :: Mu)(Km :: Tn7) dutl, ungl, thi, relAlpCJ105 Cm’
T. J. Gibson
Wild-type Synchronizable CB15 zzz-129 Tn7 (Str’) CB15N Tn7 CB15N X &r30[SC1912] dnaC (TS DNA synthesis) trpC : : Tn5 CB15 dnaC, trpC :: Tn5 PC2179 x ~Cr30[SC1293] CB15N dnaC, trpC :: Tn5 CB15N X @30[SU423] CB15N dnaC SU431 X &r30[CB15N] ret-526 (recombination-deficient) CB15N ret-526 purA :: Tn5 (adenosine auxotroph) gpsA-505 (glycerol-3-P auxotroph) bla :: Tn5 (ampicillin-sensitive) CBl5N bla :: TnS(Km’, Str’) CB15N X &r30[SC1107] CB15N bla :: Tn5-132(Tet’) SU53 X pRK2013 :: Tn5-132(Tet’) rec.526 - Tn5 (linked transposon) CB15N ret-526 -Tn5 CBl5N X &r30[PC7070] CB15N ret-526-Tn5, bla :: Tn5-132 SU133 X @r30[SU37] SU197 containing plasmid pSKCori262
Poindexter (1964) Evinger & Agabian Ely (1987) This study
Characteristics
Reference
Km’, conjugation helper, on ColEl replicon Tet’ Tn5 on pRK2013
RK2
transfer
Simon Kunkel
et al. (1983) et al. (1987)
(1977)
Osley & Newton (1977) Winkler et al. (1984) This study This
study
This
study
O’Neill et al. (1985) Marczynski et al. (1990) Ely (1987) Contreras et al. (1979) Ely (1987) This st,udy
A. Newton This study
genes
pBR322+ C.C. dmK in BamHI site Tc’, IncP-1 replicon, cosmid vector pLAFR-5 + early replicating CB15N DNA pLAFR-5+ early replicating CB15N DNA pLAFR-5 + early replicating CBISN DNA Gm’ pACYC184 derivative, Tc’, RK2 oriT Spontaneous Tc” derivative of pAGMT pkGMT* + BamHI fragment I? from COSMID I pUC derivative (pMB1 replicon), Ap’ pSK( + )BluescriptII + PstI to BamHI Cori DNA
media supplemented with the following antibiotics: 1 pg tetracycline/ml for pLAFR-5 cosmids, 2.5 pg gentamycin/ml for pAGMT and pAGMT* plasmids, and 10 pg ampicillin/ml for pSK( + )BluescriptII plasmids. Cell density was determined by measuring the absorbance at 660 nm. Cell cultures were synchronized by collecting the more dense swarmer (CBlBN variant) cells from a Ludox (colloidal silica) gradient and returning them to fresh M2G medium (Evinger & Agabian, 1977). Protocols for labeling synchronous C. crescentus cells in
or source
This
study
This
study
Figurski
or source
& Helinski
(1979)
Berg & Berg (1983); B. Ely Games et al. (1990) Keen et al. (1988) This study; Fig. 4 This study; Fig. 4 This study; Fig. 4 Alley et al. (1991) This study; Fig. 6 This study; Fig. 7 Short et al. (1988) This study; Fig. 9
with [3H]deoxyguanosine have been reported (Dingwall et al., 1990; see Fig. 3). We also observed that C. crescentus cells could be efficiently labeled with [a-32P]dGTP (see Figs 5 and 11). This was unexpected, because it was believed that phosphatases would remove the a-phosphate before the dGTP could reach the DNA polymerase. However, under the labeling conditions described for Figs 5 and 11, we observed that 32P was incorporated only into DNA (and with linear kinetics) from time t = 0 to 2 min. Subsequent incorporation became progressively slower,
962
G.
T.
Marczynski.and
and some of the incorporation became base labile (presumably due to RNA synthesis). These observations suggest that C. crescentus has an efficient transport system for dGTP and/or dGMP. [c(-32P]dGTP uptake experiments indicated that swarmer cells and stalked cells internalized 32P with the same kinetics, but that only stalked cells incorporated “P into DNA. Incorporation into DNA (base stable cts/min) versus RNA (base labile cts/min) was determined by precipitation with trichloroacetic acid and scintillation counting essentially as described (Degnen & Newton, 1972). (d) DNA
preparation, sequencing, pulsed-jeld gel electrophoresis
mutagenesis analysis
and
Standard plasmid preparations from E. coli, digestions by restriction endonucleases, and agarose gel electrophoresis procedures were performed according to published protocols (Maniatis et al.. 1982). DNA sequencing was performed by the dideoxynucleotide chain termination method (Sanger et al., 1977) using the Sequenase System II kit (USB Corp.). Single strand DNA pSK( + )BluescriptII “phagemids” served as templates and were prepared from selected Gori DNA deletions (see Fig. 9), as described (Short et al., 1988). Oligonucleotide directed site-specific mutagenesis was performed essentially as described (Kunkel et al., 1987). COSMID I BamHI fragment IE (see Fig. 4) was inserted into M13mp18. and substrate single strand DNA was prepared from E. coli strain CJ236. Following the mismatchedoligonucleotide-directed synthesis reaction, both mutated and unmutated phage were recovered from E. coli TGl. and base changes were confirmed by sequencing. In order to assay DnaA box point mutations, both mutated (3 independently isolated DnaA box mutants) and unmutated IE fragments were subcloned from M13mp18 into the replication test vector pSK( + )BluescriptII (see Fig. 9). Pulsed-field gel electrophoresis (PFGET) techniques for C. crescentus, including in situ genomic DNA preparations, endonuclease digestion conditions, electrophoresis conditions, Southern blot PFGE hybridization procedures, [3H]DNA labeling of synchronized cells, and the preparations of [3H]DNA fluorograms. have been described in detail (Dingwall et al., 1990). (e)
Techniques
employed
to isolate
the origin
of replication
scale PFGE of early replicating C’. crescentus genomic DNA required embedding lo-fold higher cell densities inside the agarose plugs, but otherwise the procedures were the same as for analytical PFGE. To efficiently remove the DNA from the agarose, the excised bands were digested in situ with EcoRI and recovered by electrophoresis inside dialysis membrane bags. Following precipitation with ethanol, this DNA was labeled in vitro with [cr-32P]dGTP using Klenow DNA polymerase and the random primer pd(N)G (as specified by Boehringer-Mannheim), and hybridized to cosmid DNA clones immobilized on nylon membranes. The construction of this cosmid library (pLAFR-5 containing Sau3A partially digested CBlSN DNA) has been reported (Alley et al., 1991). Approximately 580 cosmids were screened, initially in pools of 10 and 40, and then rescreened individually. A total of 21 cosmids was repeatedly hybridized by the 110 kb early replicating Preparative
t Abbreviations used: PFGE, pulsed-field electrophoresis; kb, lo3 bases or base-pairs; pair(s).
gel bp, base-
L.
Shapiro
AseI band, and 17 cosmids were repeatedly hybridized by the 83 kb and 85 kb AseI bands (described in Fig. 3). In order to group these cosmids into overlapping sets, they were cut with BamHI, Southern-blotted, and crosshybridized, according to established protocols (Maniatis et al., 1982). One group of overlapping cosmids (shown in Fig. 4 as COSMIDS I, II and III) were chosen for further analysis based on the precise labeling experiment described in the text and the legend to Fig. 5. BamHI fragments were subcloned from these 3 cosmids into test plasmids pAGMT, pAGMT* and pSK( +)BluescriptII. The ability of these plasmids to replicate autonomously in C. crescentus was assayed by bacterial conjugation (Ely, 1979), direct plasmid electroporation into C. crescentus (Dower et al., 1988; Gilchrist & Smit, 1991), and by quantitative extraction of plasmid DNA from C. crescentus (Marczynski et al., 1990).
Tn7 250
110
v
AseI
r”“r
A
A 305
105
A,,5A
D~OI
3” k--83-+--q
Figure
i
The earliest replicating region of the chromosome. An alignment of infrequently and DraI restriction endonuclease sites is presented with interspersed genetic and molecular markers. Numbers indicate distances between markers in kb. The 2 hatched bars show the positions of the 2 earliest labeled DNA fragments (83 kb and 50 kb) separated by pulsed-field gel electrophoresis as described in the legend to Fig. 3. The positions of the AseI and DraI sites, the transposon Tn7 and genetic marker hi& are based primarily on published data (Ely & Ely, 1989). Tn7 contains both AseI and DTUI sites. The Tn7, hi&, gpsA, purA and nov markers were mapped to this region by genetic techniques (Ely, 1987). The physical positions of gpsA, purA and novR were determined by the hybridization of 32P-labeled cosmid DNA clones to genomic DNA fragments separated by pulsed-field gel electrophoresis (data not shown). Several cosmid clones were obtained that complemented both gpsA and purA auxotrophic requirements (see the text). The novR cosmid clone was obtained by complementing a temperature-sensitive allele of the novobiocin resistance gene (the E. coli gyrB homologue: M. Rizzo, L. Shapiro & J. Gober, unpublished results). The C. crescentus dnaK (Gomes et aZ., 1990) and dnaA (G. Zweig & L. Shapiro, unpublished results) markers are homologues to their respective E. coli genes. They were both obtained during the cosmid walk through this region (see the text and Fig. 4). The Tn5 insertion in the 83 kb AseI fragment was obtained by 4Cr30 phage transduction and linkage to a kanamycin resistance’ plasmid integrated at the dnaK gene, and the dnaK gene in turn showed 11 o/0 linkage to the purA gene by $Cr30 phage transduction. C. crescentus cutting AseI
2.
1
Properties
of the Cloned
3. Results
(a) Mapping
the earliest replicating region of the C. crescentus chromosome
We localized the origin of replication on the physical map of the C. crescentus genome prepared by B. Ely and co-workers using pulsed-field gel electrophoresis (Ely & Gerardot, 1988; Ely & Ely, 1989). Infrequently cutting DraI and AseI restriction endonucleases generate large genomic fragments that were ordered and aligned to form a circular map of this genome (Ely et al., 1990). Dingwall & Shapiro (1989) have presented evidence that the earliest replicating region of the C. crescentus chromosome lies within a specific 305 kb DraI or an adjacent 105 kb DraI DNA fragment, and that replication proceeds bidirectionally. An expanded view of this part of the C. crescentus genome, as well as an alignment of both DraI and AseI-cut sites, is presented in Figure 2. To precisely map the earliest replicating DNA fragments, swarmer cells were isolated and incubated with [3H]deoxyguanosine under standard growth conditions that allowed these cells to proceed t’hrough the cell-cycle. Equal cell volumes were removed at seven minute intervals that spanned the initiation period of DNA synthesis (i.e. the swarmer cell to stalked cell transition of the cellcycle). These cells were embedded in agarose, their genomic DNA was prepared in situ, digested with the restriction endonuclease AseI, and separated by PFGE. Figure 3 shows (a) the ethidium bromidestained gel and (b) its corresponding 3H fluorogram for this experiment. Figure 3(a) shows that equal amounts of DNA were loaded in each of the lanes 1
Nl2345678Tn7
123456
Origin
from
Caulobacter
963
to 8; (b) shows that DNA synthesis is absent in the swarmer cells shown in lanes 1 to 4, and that DNA synthesis initiates as the swarmer cells develop into stalked cells in lanes 5 to 8. Figure 3(b) also shows that there are only two early labeled bands, and that they are labeled with nearly identical kinetics. These observations indicate the origin of DNA replication lies within one of these two AseI DNA fragments and, most likely, it either overlaps or lies near their boundary. In order to unambiguously identify the earliest labeled PFGE bands, we employed strains with a transposon inserted in the early replicating region. The experiment in Figure 3 employed a strain containing transposon Tn7 that has a unique insertion site in the C. crescentus genome (Ely, 1982). We mapped the position of the Tn7 as indicated in Figure 2. Tn7 contains an AseI site(s), and therefore an extra 110 kb band is seen in Figure 3(a) that is not present in the wild-type strain (compare the outside control lanes containing unlabeled wild-type CB15N DNA and CB15N Tn7 DNA: Fig. 3(a)). This 110 kb “Tn7” band is the larger of the two earliest replicating bands seen in Figure 3(b), because its position in the 3H fluorogram is aligned with the 110 kb band in the ethidium bromide-stained gel. Also, when the experiment presented in Figure 3 is repeated with the wild-type CB15N strain, only the 83 kb and a new high molecular weight band (migrating close to the wells) are seen during this early DNA synthesis period (data not shown). The smaller of the two earliest replicating AseI bands in Figure 3(b) measures between 80 and 90 kb relative to a lambda ladder (data not shown). It comigrates with either the 85 or the 83 kb AseI
78
12345678
(b) (c) (al Figure 3. Time-course analysis by PFGE of the earliest replicating C. crescentus DNA. Strain SU87 (CB15N Tn7) swarmer cells (10 ml at a cell density of @085) were isolated and resuspended in M2G media, as described in Materials and Methods. At time t = 0 min, 25 PCi of 3H-labeled deoxyguanosine (5 mCi/mmol) was added, and the cells were vigorously shaken at 30°C. Portions (1.0 ml) (corresponding to lanes 1 through 8) were removed at times 1= 0, 7, 14, 21, 28, 35, 42 and 49 min. These cells were immediately embedded in agarose and lysed in situ to prepare their genomic DNA for PFGE analysis (see Materials and Methods). Total genomic DNA was digested with AseI in (a) and (b) and doubledigested with AseI and DraI in (c). Approximately @05 of the prepared DNA w&s loaded in each lane. Subsequent procedures for PFGE and 3H fluorography were as reported (Dingwall & Shapiro, 1989). (b) and (c) Fluorograms exposed t,o X-ray film at -70°C for 1 to 2 days. (a) The ethidium bromide-stained gel corresponding to (b). Unlabeled C. crescentus genomic DNA prepared from wild-type strain CB15N and its isogenic Tn7 derivative (CB15N Tn7) were loaded in the lanes marked N and Tn7, respectively. Resolution was achieved by running these gels for 15 h at 6°C with 250 V and employing 7 s switching times.
964
G. T. Marczyneki
bands identified by Ely & Ely (1989), which are just barely resolved by the PFGE conditions of Figure 3(a). We have identified this as the 83 kb band, because the 83 kb band was reported to be adjacent to the Tn7 containing AseI band, and it overlaps the 305 kb DraI band (see Fig. 2). The 85 kb AseI band is located on the other side of the genome (Ely & Ely, 1989). Also, we have identified a Tn5 insertion in the early replicating region (that is genetically linked to the dnaK gene and within the 305 kb DraI band) that shifts the size of the 83 kb AseI band (see Fig. 2). The third prominent band that becomes labeled in lanes 7 and 8 in Figure 3(b) is the 250 kb AseI band that lies adjacent to the 83 kb AseI band on the genomic map (Ely & Ely, 1989). The ‘H fluorogram of the PFGE experiment in Figure 3(c) further supports the assignments of the 110 and 83 kb AseI bands. The same DNA samples analyzed in Figure 3(b) were digested with both AseI and DraI, and two bands were identified at the earliest time of DNA replication (lanes 5 and 6, Fig. 3(c)). The larger band in Figure 3(c) comigrated with the 83 kb band seen in Figure 3(b). The smaller band is approximately 50 kb, and it is produced by DraI cutting within the 110 kb ‘In7 AseI band. These results are consistent with the restriction map presented in Figure 2, and demonstrate that the origin of replication must lie within the 133 kb DNA region shown as the two hatched bars overlapping the right arm of the 305 kb DraI fragment. Since these two bands are labeled simultaneously, these data also suggest that the origin of replication lies near their central AseI site. (b) Genomic
clones that span replicating region
the earliest
We cloned a large region of the C. crescentus genome that spans the origin of DNA replication before proceeding to isolate autonomously replicating sequences. This strategy was motivated by four major considerations. (1) A direct screen of a total genomic library failed to isolate any C. crescentus DNA that would support autonomous plasmid replication in C. crescentus (data not shown; O’Neill et al., 1983). (2) DNA sequences that support autonomous plasmid replication may not necessarily be the physiologically relevant origin(s) of chromosomal DNA replication. (3) The origin of DNA replication may be distant from its control elements. (4) Genes involved in DNA metabolism are often linked to bacterial origins of DNA replication (Ogasawara et al., 1990), and these would be valuable assets in future studies. We employed preparative scale PFGE to isolate the DNA from the 110 and 83 kb earliest replicating AseI bands that were identified in the experiment presented in Figure 3. These DNA preparations were separately labeled in vitro with 32P and hybridized to a C. crescentus CBl5N genomic cosmid library as described in Materials and Methods. Approximately 580 pLAFR-5 cosmids, with average genomic inserts of 25 kb, were
and L. Shupiro
screened by hybridization to 32P-labeled DNA from the 110 kb AseI band. Twenty-one independent cosmid clones were isolated as judged by their unique BamHI digestion patterns. Three of these cosmid clones contained genes that complemented the gpsA and the purA mutations (see Fig. 2). Since both gpsA and purA map genetically to the left of Tn7 (see Fig. 2), their genes were expected to be present in this cosmid collection. Likewise, the same cosmid library was screened by hybridization to the early replicating 83 kb AseI band (and also the contaminating 85 kb band). Seventeen independent cosmid clones were obtained from this second screen. One cosmid was isolated by both screens, and therefore should span the border between the 110 and the 83 kb early replicating bands. This was confirmed by hybridizing the cloned cosmid DNA to a pulsed-field gel Southern blot containing CB15N Tn7 DNA cut with AseI. As expected, this cosmid clone, designated COSMID I in Figure 4, hybridized to the 110 and the 83 kb band (see the legend to Fig. 2). COSMID I contains a unique AseI site within its BamHI fragment IE. Since the results from the experiment in Figure 3 suggested that this is either at or near the origin of DNA replication, we chose to concentrate our efforts on this cosmid and its adjacent overlapping cosmids. As described in Materials and Methods, a series of overlapping cosmid clones was obtained that span approximately 70 kb across the border between the 110 kb and the 83 kb AseI early replicating bands. TWO more cosmids were identified from the 83 kb band hybridization screen by their BamHI Southern blot cross-hybridization patterns with COSMID I (these are COSMIDS II and III shown in Fig. 4). Also, two cosmids were identified from the 110 kb band hybridization screen that overlap the right side of COSMID I (data not shown). The three cosmids presented in Figure 4 were chosen for more detailed analysis based on the results of the precise labeling experiment presented below. (c) Precise
labeling of&%? C. crescentus origin DNA replication
of
Labeling the origin of replication requires a very synchronous initiation of DNA synthesis and a significant reduction of the fork movement rate. We accomplished this by using a conditional (temperature-sensitive) DNA synthesis mutant (Osley & Newton, 1977) and the gyrase inhibitor novobiocin. Swarmer cells from C. crescentus strain SU443 (CB15N dnaC) were incubated at the nonpermissive temperature (37°C) in the presence of 10 pg novobiocin/ml. This level of novobiocin inhibits DNA synthesis at the permissive tempera(data not shown). Under these ture by -90% conditions we observed that the swarmer cells differentiated into stalked cells, but that chromosomal replication was blocked at the start of DNA synthesis. This synchronous cell culture was downshifted from 37°C to 30°C and rapidly labeled with
Properties of the Cloned Origin
965
\
(A$) , ,, ,
I
I3 B
I
I
from Caulobacter
5kb
I
?,
In
D
A
F 1 I
A
2 I
c!l D;C
IA -
II
hemE
I
cfnuA
0
I
I
dnoK,J
Zn YWO 32P distribution
Cod
I
IIIC
Figure 4. Alignment of cosmid clones containing the origin of replication. The DNA inserts from 3 overlapping cosmid clones (COSMIDS I, II and III) are aligned with both the genomic map described in Fig. 2 and the [32P]DNA fragments generated by the experiment described in Fig. 5. Short vertical bars mark the BumHI sites on the cosmid clones. Capital letters A, B, C, etc., denote the individual BumHI fragments within a cosmid clone. For example, IE denotes the 5th largest BarnHI fragment from COSMID I. This is the only cosmid BamHI fragment that contains an AseI site (shown as a filled arrowhead). Cosmid alignment with the C. crescentus chromosome was determined as described in the text and confirmed by the hybridization of in vitro-labeled cosmids to genomic DNA cut with AseI and resolved by PFGE. COSMIDS II and III only hybridized to the 83 kb AseI band, and COSMID I hybridized to both the 83 kb and to the 110 kb (Tn7) AseI bands. The bottom bar graph shows the in viva 32P distribution in this region of the chromosome approximately 30 s after the start of chromosomal DNA replication. This in viva-labeled DNA was assigned to these cosmids based on the comigration of the 32P-labeled bands (shown in Fig. 5) with unlabeled cosmids cut with BarnHI, S&I, Hind111 and AseI. The area in each bar is proportional to the 32P label in each corresponding band.
32P by mixing with medium containing [cr-32P]dGTP. Chromosomal DNA was isolated from these cells, digested with BarnHI, and resolved by standard agarose gel electrophoresis. An autoradiogram of this experiment reveals only five strongly labeled BamHI bands, and seven weakly labeled bands (Fig. 5). These bands clearly belong to the early replicating region, because when this in viva 32Pwas hybridized to cosmid labeled DNA preparation clones that were Southern-blotted on nylon membranes, this 32P-labeled DNA hybridized primarily to COSMIDS I and II, but only very weakly to COSMID III, and not at all to 20 other cosmids that were also isolated in our screen (data not shown). The combined lengths of the five
1 F ,...-,;
J
Figure
5.
Precise in vivo 32P-labeling at the C. crescentus chromosomal origin of replication. Strain SU443 (CB15N dnaC) swarmer cells (10ml at a cell density of 05) were isolated and incubated at 37°C for 90 min in M2G medium containing 10 pg novobiocin/ml. In order to quickly label and initiate replication, the temperature of this culture was down-shifted by mixing it with an equal volume of M2G medium at 23°C containing 250&i of [a-32P]dGTP (Amersham Corp., 3000 Ci/ mmol). In order to rapidly halt 32P incorporation and reduce labeled replication intermediates that migrate anomalously during agarose gel electrophoresis, unlabeled dGTP (0.2 mM final concn) was added 30 s after the temperature down-shift, and DNA synthesis was continued for an extra 5 min at room temperature. These cells were embedded in agarose, and their chromosomal DNA was prepared in situ (see Materials and Methods). Lanes B, S and H show the autoradiograms of this DNA preparation following digestion with BarnHI, Sal1 and HindIII, respectively, and resolution by standard agarose gel electrophoresis. Unlabeled cosmid DNA (COSMIDS I, II and III shown in Fig. 4) were digested with the same set of restriction endonucleases and run in the adjacent lanes (not shown). The migration of key cosmid bands cut with BamHI is shown on the left side. Most of the 32P incorporation is accounted for by these comigrating cosmid bands (see Fig. 4). The migration of lambda phage DNA cut with Hind111 (A) is shown on the right. Numbers indicate DNA size in kb. strongly labeled bands is approximately 33 kb. All five of these bands (designated IA, IIB, IB, IC and ID in Fig. 5) correspond to cosmid fragments presented in Figure 4, based on their comigration with the unlabeled DNA from cosmids I, II and III
966
G. T. Marczynski
loaded in adjacent ianes (not shown). Likewise, four out of the seven weakly labeled bands (designated IIIA, IE, IIIC and IF) were observed to comigrate with specific BamHI cosmid fragments. These banding assignments were confirmed by employing additional restriction enzymes such as SalI and HindIII, whose digestion patterns are also presented in Figure 5. We also observed, as would be expected from our banding assignments, that only the 32P-labeled band IE contains an AseI site. The pattern of j2P-labeled BamHI bands in Figures 4 and 5 suggests that the origin of replication lies within band IE, because IE is underlabeled relative to its adjacent bands, and our labeling protocol could allow DNA synthesis to initiate in the absence of 32P. Presumably, DNA synthesis initiates at the non-permissive temperature, because dn.aC is a temperature-sensitive elongation mutant (Osley & Newton, 1977). Also, a short period of unlabeled DNA synthesis must occur before the [a-32P]dGTP can enter the cells. These considerations predict that the site of initiation will be under-labeled relative to the flanking sequences. The bar raph in Figure 4 shows the quantification of the 38 P label in each of the cosmid BamHI fragments. Band IE is significantly under-labeled even when accounting for the size of its neighboring bands. The four bands (IA, ID, IB and IC) that surround band IE account for most of the 32P incorporation, and comparatively weakly labeled bands are seen further on either side. In summary, these results clearly localize the earliest replicating region of the C, crescentus chromosome to the DNA spanned by our cosmid clones, and they also suggest that chromosomal DNA synthesis initiates within the BamHI fragment IE.
(d) C. crescentus DNA that supports autonomous replication We reasoned that a DNA fragment containing the origin of replication should support the autonomous replication of a plasmid incapable of replicating in C. crescentus by itself. Therefore, we assayed DNA fragments from the early replicating region presented in Figure 4. Cosmid BamHI fragments were inserted into the test vectors pAGMT and pAGMT* that confer gentamycin resistance to host bacteria (Fig. 6). These plasmids can replicate in E. coli (from the P15A oriV), but not in C. crescentus, and they contain an origin of transfer (Trans) that allows efficient plasmid mobilization from E. coli to C. crescentus by bacterial conjugation (Guiney & Yacobson, 1983; Ely et al., 1979). Our initial replication assay scored gentamycinresistant C. crescentus colonies. After conjugation with E. coli cells containing specific subclones, drugresistant C. crescentus colonies would be indicative of plasmid maintenance. In order to prevent plasmid maintenance in C. crescentus by homologous recombination, we also employed an isogenic recombination-deficient (ret-) C. crescentus strain
and L. Shapiro
(5.67
kb)
P15A
OriV
P15A
OriV
h??HI Tc’
Tc’
Figure 6. Test vectors for autonomous replication in C. crescentus. Both plasmids pAGMT and its derivative pAGMT* replicate in E. coli, but not in C. crescentus. pAGMT contains the incP15A replicon from plasmid pACYC184 (oriV), tetracycline and gentamycin resistance genes (Tc’ and Gm’), and the origin of transfer from plasmid RK2 (Trans) for efficient mobilization by bacterial conjugation (M. R. K. Alley, personal communication). pAGMT* is derived from pAGMT by a spontaneous deletion that spans the AseI and Hind111 sites and removes the tetracycline promoter. This deletion is required for efficient autonomous replication by the cloned C. crescentus origin of replication in C. crescentus.
(SU146) that blocks colony formation by plasmid integration (O’Neill et al., 1985). From among all of the BamHI fragments of COSMIDS I, II and III, only pAGMT plasmids containing fragment IE could replicate in recombination-deficient C. crescentus cells. However, the frequency of mobilization for these plasmids was very low (m 196 gentamycin-resistant colonies per plate in our standard assay), suggesting a barrier to the maintenance of these plasmids in C. crescentus. When plasmid DNA was prepared from the C. crescentus transformants, extra DNA was found inserted between the AseI and BamHI sites of pAGMT (see Fig. 6). The size range (665 to 1.35 kb) suggested that these were C. crescentus insertion
Properties
(AS) ,/,I
of the Cloned
C0SMl0S I
Origin
from
Caulobacter
967
Cl234567891011
B
IDI
m
B
0
A
C
F
A
DEC
,
n B
,
I
Plasmid
ID, IIC UbYmE) IA, UA IF, UD IIB, me pACtAT* alone
Cl0 s-10” Cl0 =-IO5 Cl0
