Phenotypic characterization of overexpression or deletion of the Escherichia coli crcA, cspE and crcB genes

Escherichia coli K-12 contains nine homologues of the cold-shock-like proteins, CspA through CspI. Four of the genes for these proteins are induced by cold shock (cspA, B, G and I: Etchegaray et al., 1996; Nakashima et al., 1996; Wang et al., 1999), one is induced by nutritional starvation (cspD: Yamanaka & Inouye, 1997), two are constitutively expressed (cspC, E: Bae et al., 1999) and the regulation of the other two (cspF and H) is unknown. CspA, the major cold-shock-induced protein in E. coli, plays a role in stabilizing RNA (Jiang et al., 1997) and in transcription antitermination (Bae et al., 2000). CspB, CspC and CspE have been shown to bind to RNA with slight specificity for AU sequences (Phadtare & Inouye, 2000). CspE has been implicated in antitermination through its binding to RNA (Bae et al., 2000). CspE also binds to double-stranded DNA (dsDNA) with no sequence specificity (Hanna & Liu, 1998). While the details and consequences of the binding of specific Csp proteins to DNA or RNA have yet to be determined, the overriding activity of the Csp proteins seems to be to bind some type of nucleic acid (RNA or DNA).

We have previously shown that overexpression of cspE, in conjunction with the two open reading frames that flank cspE (crcA and crcB), protects cells from the DNA decondensing agent camphor (Harrington & Trun, 1997; Hiraga et al., 1989), and suppresses the chromosome instability and growth defects seen in mukB mutants (Hu, 1996; Yamanaka et al., 1994b). From our original studies, all three genes (crcA, cspE and crcB) are required for these phenotypes and the genes must be overexpressed at high levels (Hu, 1996). Suppression of the mukB deletion is also seen in mutants defective in topA (Sawitzke & Austin, 2000). Suppression of ΔmukB by topA, like suppression of ΔmukB by crcA, cspE and crcB overexpression, reverses both the Ts defect and the anucleate cell production phenotypes. It is thought that suppression by topA mutants is the result of an increase in negative supercoiling.

In this paper, we examine the phenotypic consequences of deleting or overproducing crcA, cspE and crcB (AEB). We have combined the deletion and the overproducing plasmids with mutations in the genes for the three classes of proteins implicated in chromosome folding. Class I are small DNA-binding and bending proteins such as HU, HNS, IHF and Fis that co-purify with the nucleoid (HU, HNS and Fis) (Murphy & Zimmerman, 1997) or can substitute for HU in some reactions (IHF) (Nash, 1990). Class II is MukB (see above). Class III proteins are the enzymes that affect supercoiling, gyrase and topoisimerase IV (topo IV). Topo IV is a supercoil-relaxing enzyme needed for partitioning of daughter chromosomes (Adams et al., 1992; Kato et al., 1992; Peng & Marians, 1993; Zechiedrich & Cozzarelli, 1995). The goal of these experiments is to use what is known about the in vivo roles of the condensing proteins to determine if CrcA, CspE and CrcB share any of these properties.

Bacterial strains and plasmids.
Bacterial strains and plasmids used are listed in Table 1. P1 transductions (Silhavy et al., 1984) and plasmid transformations (Chung et al., 1989) were carried out as previously described.

Table 1. Strains and plasmids

Construction of plasmid-borne deletions.
For the deletion of cspE and crcB, pKH4b was digested with BsmI and BstEII and pCAC7 (source of cat-19, which encodes chloramphenicol resistance) with KpnI and SphI. The fragment ends were made flush using Pfu polymerase, gel-purified, ligated and transformed. The plasmid containing the correct 558 bp deletion of cspE and crcB and 987 bp insertion of cat-19 was named pMDG4. The deletion of cspE and crcB (ΔcspEcrcB143 : : cat-19) removes part of the predicted promoter, all of cspE and leaves the last 48 bp of crcB. For the deletion of all three genes, pNT2 (Hu, 1996), which contains a larger chromosomal insert than pKH4b, was digested with SacI and BglII and pCAC7 was digested with SmaI. The fragments were treated as described above. The plasmid with the correct 1921 bp deletion and 1016 bp insertion was named pNAB1. The deletion of all three genes (ΔcrcAcspEcrcB587 : : cat-19) completely removes the coding sequence for all three genes. All enzymes were used according to the manufacturer's instructions, plasmids were purified using Qiagen kits and ligation mixes were introduced into NT3 by electroporation. All plasmids were sequenced using the dRhodamine terminator cycle sequencing reactions (PE Applied Biosystems) to ensure they were constructed correctly. Primers used were: for crcA, 5'-GCAGAAAACGACGCATCA-3' and 5'-GCGTTATTTTTGATGTTA-3'; for cspE, 5'-GATGCGCTTTCAGTTTTG-3' and 5'-CCTCAACCGCACACTAAA-3; and for crcB, 5'-AGATACGTCAGCAAGAAT-3' and 5'-CATTGCCACCTCCTGTCA-3'.

Cloning of crcA, cspE, crcB or cspEcrcB into pBAD18-Kan.
The appropriate gene complete with its own predicted ShineDalgarno sequence was amplified by PCR. Incorporated into the primers were a KpnI site on the 5' end of the gene and an XbaI site on the 3' end of the gene (cspE, crcB, cspEcrcB). The PCR fragments, as well as purified pBAD33 vector DNA, were digested with KpnI and XbaI according to the manufacturer's direction (New England Biolabs). The PCR fragment was mixed with the vector, ligated and transformed. Potential clones were checked for the presence of the correct insert by restriction enzyme analysis, followed by DNA sequencing. The PCR fragments were first cloned into pBAD33 utilizing the KpnI and XbaI sites. Subsequently, they were moved into pBAD18-Kan by digesting the appropriate pBAD33 clone with MluI and SphI, isolating the fragment containing the gene of interest and subcloning into pBAD18-Kan. The potential pBAD18-Kan subclones were also checked by restriction site analysis followed by DNA sequencing. Because crcA contains an internal KpnI site, an EcoRI site was substituted for KpnI for the crcA clone and crcA was ligated directly into pBAD18-Kan. The primers used were: crcA, 5'-NNNNGGGAATTCGTAGCTTTGCTATGCTAGTAGTAG-3' and 5'-NNNNCCTCTAGAGCTATTGATTTTAAAGAAGTTAC-3'; cspE, primer 1 5'-NNNNNNNNGGGGTACCCATGTAAAGGTAATTTTGATGTCTA-3' and primer 2 5'-NNNNNNNNGCTCTAGACACTGGCATTCTGGCTGT-3'; crcB, primer 3 5'-NNNNNNNNGGGGTACCACAGCCAGAATGCCAGTG-3' and primer 4 5'-NNNNNNNNGCTCTAGATTTAACCCACTGCATCAG-3'; cspE, crcB primer 1 and primer 4. Extra bases (N) were added to the 5' ends of the primers to ensure that the enzymes would digest the DNA efficiently.

Isolation of specialized transducing phages carrying crcA, cspE and crcB.
Approximately 12 000 plaques from an E. coli library in the λ vector λDE3 (Katayama et al., 1988) were screened by plaque blot (Sambrook et al., 1989) for the presence of crcA, cspE and crcB. Twenty-six potential phages were identified, purified and reprobed. To determine which of the phages carried a given gene, we performed PCR reactions with three sets of primers, one for each gene (see above). Of the original 26 phages, 21 contain all three genes, three contain cspE and crcB, one carries only crcB and one carries only cspE (data not shown). A single phage, λMDG18, which contains the three genes on a 2·1 kb chromosomal fragment, was used for all subsequent genetics. The cI857 mutation was crossed onto λMDG18 and a Kan^r marker was moved onto λMDG cI857 by growing the phage on a strain containing the minitransposon mkan (Kleckner et al., 1991), forming the phage λMDG18K.

Crossing the deletions into the chromosome.
The deletions were first crossed from either pMDG4 or pNAB1 to λMDG18K and subsequently crossed from the phage to the chromosome as described by Maurizi et al. (1985). The markers on the phage (cI857 and Kan^r) and the marker for the deletion (cat-19) were used to follow the movement of the genes in the crosses. To ensure that no additional mutations were induced in these crosses to compensate for the loss of AEB, a miniTn10 50 % linked to the deletions was isolated and used to move the deletions non-selectively into NT3 (a haploid for AEB) or NT3 (pNT2) (a merodiploid for AEB). The AEB deletion can be transduced into the haploid 46 % linked to the miniTn10 (60 Cam^r transductants/130 total Tet^r transductants) and into the merodiploid 51 % linked (118 Cam^r/233 Tet^r transductants). These numbers indicate that the AEB genes are not essential.

Southern blots.
Southern blots of the chromosomal deletions were carried out as previously described by Sambrook et al. (1989). Chromosomal DNA was isolated using the Wizard genomics DNA isolation kit (Promega). ³²P-labelled pNT2 was used as the probe.

Determination of supercoiling levels.
All strains used in this assay were made recA56 by P1 transduction. pBR322 or pNT2 was transformed into the recA56 derivatives by electroporation and strains containing plasmid monomers were identified by preparing plasmid DNA on random transformants and screening by agarose gel electrophoresis. Strains containing plasmid monomers were grown overnight in LB medium with ampicillin (50 µg ml^-1) and plasmid DNA was prepared. Plasmid DNA was electrophoresed on a 0·8 % agarose gel with 10 µg chloroquine ml^-1 in 0·5x TPE buffer (45 mM Tris pH 7·2, 0·87 mM Na₂EDTA). Gels were electrophoresed for 18 h at 4 °C at 30 V with recirculating buffer, stained with SYBR Green I (Molecular Probes) and photographed.

Assaying for camphor resistance.
Plate assays for camphor resistance were carried out as described previously (Trun & Gottesman, 1990). Liquid assays for camphor resistance were carried out as follows. Five millilitres of LB was inoculated with a single colony and incubated overnight at 37 °C. The culture was diluted 100-fold into 25 ml LB and grown to OD₆₀₀ 0·15 at 37 °C. Either 0·10 g or 0·15 g camphor (as indicated) was added to each flask and 1 ml aliquots of cells were removed every 30 min for viable cell counts.

Other assays.
UV sensitivity was assayed as described by Li & Waters (1998) and β-galactosidase was assayed as described by Zhou & Gottesman (1998). Staining, fixation of cells, photography and processing of the film were carried out as described by Hu (1996). Suppression of the temperature-dependent growth defect of parC and parE Ts mutants was examined by switching the temperature of actively growing cells from 30 °C to 42 °C and measuring the number of viable cells at 1 h intervals for 5 h.

Overexpression of AEB increases levels of plasmid supercoiling
We examined the supercoiling levels of pBR322 (Fig. 1a) and pNT2 (Fig. 1b; crcA, cspE and crcB cloned into pBR322) in the parental strain and in mutants that are defective in chromosome condensing proteins. Plasmid DNA was isolated from recA56 derivatives of each strain that contained plasmid monomers as judged by agarose gel electrophoresis. This was necessary because several of the mutants contain mainly dimerized plasmid molecules in recA⁺ cells. Plasmid DNA was electrophoresed on chloroquine-containing agarose gels. The gels were analysed using NIH Image and scans of each lane are shown (Fig. 1).

(21K): Fig. 1. Densitometric scans of plasmid DNA isolated from various strains electrophoresed on 0·8 % agarose gels in the presence of 10 µg chloroquine ml-1 to view the extent of DNA supercoiling. (a) pBR322 DNA, (b) pNT2 DNA. Scan 1 is of plasmid DNA isolated from the parental strain NT3; scan 2, OS314 (ΔihfA : : tet); scan 3, OS316 (hns205 : : tet); scan 4, NT839 [parC1215(Ts)]; scan 5, NT801 [parE10(Ts)]; scan 6, OS301 (ΔhupA : : kan); scan 7, OS343 (fis : : cam); scan 8, NT816 [gyrA43(Ts)]; scan 9, NT808 [gyrB202,221(Ts)]. All of the strains used in this figure were made RecA- by transducing recA56 into the indicated strain from DF475.

Overexpression of AEB⁺ (Fig. 1a scan 1 vs Fig. 1b scan 1) increases the level of supercoiling in plasmids in the parental strain. Deletions of the genes for IHF or HNS or temperature-sensitive mutations in topo IV have no effect on plasmid supercoiling (Fig. 1a scans 2, 3, 4 and 5 compared with Fig. 1a scan 1). Overexpression of AEB⁺ had no effect on plasmid supercoiling levels in these mutants (Fig. 1b scans 2, 3, 4 and 5). Deletion of the genes for HU and Fis moderately decreased the supercoiling of resident plasmids (Fig. 1a scans 6 and 7 vs Fig. 1a scan 1). Overproduction of AEB⁺ increases supercoiling of the plasmid DNA in ΔhupA (Fig. 1b scan 6 vs scan 1) and in Δfis (Fig. 1b scan 7 vs scan 1). Ts mutations in gyrase dramatically decrease the levels of supercoiling of resident plasmids (Fig. 1a, scans 8 and 9 vs scan 1). Overexpression of AEB⁺ increases the level of plasmid supercoiling in gyrA and gyrB Ts mutations (Fig. 1a scans 8 and 9 vs scan Fig. 1b scans 8 and 9). The increases in supercoiling levels in strains mutant in HU, FIS and gyrase are not seen if the cspE gene in the plasmid is inactivated by an insertion (data not shown).

Overexpression of AEB⁺ affects other phenotypes of gyrase and topo IV mutants
Strains with Ts mutations in gyrA, gyrB, parC and parE exhibit sensitivity to nalidixic acid. Overproduction of AEB⁺ in gyrase or topo IV mutants reduces this sensitivity to nalidixic acid (Table 2). This effect can also be seen with norfloxicin (data not shown).

Table 2. Overexpression of AEB+ results in increased resistance to nalidixic acid in gyrase and topo IV Ts mutants

The parC1215 Ts mutant containing pBR322 shows a 1000-fold decrease in viability (2x10⁸ colonies ml^-1 before shift to 9x10⁴ colonies ml^-1 after shift) 5 h after a temperature shift from 30 °C to 42 °C. If the parC1215 Ts mutant contains pNT2 (AEB⁺), viability under these same conditions decreases only 100-fold (2x10⁸ colonies ml^-1 to 2x10⁶ colonies ml^-1). For the parE10 Ts mutant containing pBR322, the decrease in viability after 5 h at 42 °C is 100-fold (3x10⁸ colonies ml^-1 to 1x10⁶ colonies ml^-1). If the parE10 Ts mutant contains pNT2, viability is only decreased by 10-fold (3x10⁸ colonies ml^-1 to 1x10⁷ colonies ml^-1). For both parC1215 and parE10 mutants, overexpression of AEB⁺ increases viability at the nonpermissive temperature by 10-fold after 5 h.

parC1215 and parE10 Ts mutants, when grown at the permissive temperature, exhibit a defect in nucleoid morphology that is exemplified by cells that are 23 times longer than the parental cells and contain one large centrally located nucleoid mass (Fig. 2, Ib and Ic). Cells lacking DNA are also more abundant than in the parental controls. Combining the Ts mutations with a chromosomal deletion of AEB (see below) exacerbates the nucleoid morphology defects. The cells filament more extensively (Fig. 2, IIb and IIc) and the nucleoids occupy the entire length of the filaments. Overproduction of AEB from pNT2 corrects these defects and the cells appear normal in size and nucleoid morphology (Fig. 2, IIIb and IIIc). gyrA and gyrB ts mutants do not exhibit a nucleoid morphology defect and cannot be scored in this assay.

(90K): Fig. 2. Micrographs of wild-type and mutant cells stained with DAPI and photographed using both UV and visible light. Bar, 10 µm. Column (I) shows wild-type and mutant cells, column (II), the same cells containing ΔcspEcrcB and column (III), the mutant cells overexpressing AEB from pNT2. Row (a) NT3, row (b) NT839 [parC1215(Ts)] and row c) NT801 [parE10(Ts)].

Deletion of crcA, cspE and crcB leads to an increase in camphor sensitivity
To determine the consequences for the cell when it is lacking AEB, we constructed several chromosomal deletions (Fig. 3). Parental cells (NT3), ΔcspEcrcB (MDG143) and ΔcrcAcspEcrcB (NT587) behave identically when grown on different agars (LB, TB, M63 minimal or MacConkey) at a variety of temperatures (23, 32, 37, 39 or 42 °C). If cells are grown at 32 °C and then shifted to 42 °C, no difference is seen in the number of viable cells for the wild-type or the deletion strains (data not shown). This is also true if the cells are grown at 32 °C and shifted to 10 °C (data not shown). Cells containing either of the deletions grow approximately 1015 % slower than wild-type cells (data not shown).

(14K): Fig. 3. Diagram of the AEB region of the chromosome. The extent of the two deletions is shown below, with the lines indicating the DNA that is missing in the deletions. ΔEB starts at the BstEII site, 126 bp downstream of the stop codon of crcA and 49 bp upstream of the start codon of cspE, and goes through the BsmI site 48 bp upstream from the stop codon in crcB. ΔAEB extends from the SacI site 609 bp upstream of crcA through the BglII site 332 bp downstream of crcB. The stars (crcA), circles (cspE) and squares (crcB) show the positions of the primers used for sequencing and PCR reactions. P indicates predicted promoters.

Given that overexpression of crcA, cspE and crcB leads to more condensed nucleoids and camphor resistance (Harrington & Trun, 1997; Hu, 1996), we tested whether ΔAEB or ΔEB strains are more sensitive to camphor. Fig. 4 shows that strains containing either ΔEB or ΔAEB are, as expected, more sensitive to camphor than the parental strain.

(18K): Fig. 4. Effect of camphor on strains deleted for AEB or EB. Camphor (0·10 g) was added to 25 ml of exponentially growing cells at 37 °C at the zero time point and the number of surviving cells measured. The data are means±SE of three experiments. , NT3 (wt); , MDG143 (ΔcspEcrcB); , NT587 (ΔcrcAcspEcrcB).

ΔAEB or ΔEB exacerbates the phenotypes of topo IV and MukB mutants
To determine how deleting multiple genes for potential condensing proteins affects the cells, we combined ΔAEB or ΔEB with mutations in the genes for fis, ihfA, hns, hupA, mukB, gyrB, parC and parE. In all cases, ΔAEB and ΔEB behave the same; only the data for ΔEB are presented. We first compared transduction efficiency of ΔEB into each of the condensing protein mutants using a miniTn10 linked to ΔEB. In all cases, the linkage between the miniTn10 and ΔEB is the same in the mutants as it is in a cspE⁺crcB⁺/ΔcspEcrcB merodiploid or in the parental strain (data not shown). This indicates that transduction of the cspEcrcB deletion into the mutant strains does not require the accumulation of suppressor mutations. ΔEB exacerbates the phenotypes of parC(Ts) and parE(Ts) mutants (Fig. 2, column II). ΔEB also increases the severity of the ΔmukB : : kan growth defect such that the double mutants form much smaller colonies at room temperature and do not show any growth above 30 °C. For the other double mutants, there was no significant change in the phenotypes of the single mutants versus the mutant plus ΔEB.

Camphor resistance in mutants defective in chromosome condensation
We tested the camphor resistance of mutations in chromosome condensing proteins in both the presence and absence of the AEB⁺ overexpressing plasmid (Table 3). The gyrase and topo IV Ts mutants were tested at 30 °C and the hns, hupA, fis and ihfA insertion-deletion mutants were tested at 37 °C. The gyrB202,221(Ts), parC1215(Ts) and parE10(Ts) mutants are 1001000 times more sensitive to camphor than parental cells, whereas the gyrA43(Ts) mutant shows the same level of sensitivity as the parental strain. Overexpression of AEB⁺ in the wild-type at 30 °C results in a 20-fold increase in camphor resistance. As compared to wild-type cells, overexpression of AEB⁺ in the gyrase or topo IV Ts mutants results in 20018 000-fold increase in camphor resistance.

Table 3. Percentage survival of cells treated in liquid culture with 0·15 g camphor for 60 min Each experiment was carried out six times. Mean results are shown

At 37 °C, the overexpression of AEB⁺ results in 1000-fold increase in camphor resistance in the parental cells. The hupA mutant exhibits the same sensitivity to camphor as a wild-type strain, the hns and ihfA mutants are about 10-fold more sensitive and the fis mutant is about 1000-fold more sensitive. Overexpression of AEB⁺ has little to no effect on the camphor resistance of the hns, hupA and ihfA mutants. Overexpression of AEB⁺ in the fis mutant increases camphor resistance by over 13 000-fold. From these results, the background of a strain influences the ability of overexpression of AEB⁺ to increase camphor resistance.

Overexpression of AEB protects ΔhupAB strains from UV irradiation
Li & Waters (1998) have shown that strains lacking hupA or both hupA and hupB are UV sensitive due to a defect in homologous recombination. pNT2 (AEB⁺) or pBR322 were introduced into the double mutant ΔhupA ΔhupB (OS336) and UV sensitivity was measured. As can be seen in Fig. 5, pNT2 protects the ΔhupA : : Kan ΔhupB : : Cam strain from UV by about 100-fold. The amount of killing in the ΔhupAB pNT2 strain is equivalent to that seen by Li & Waters (1998) in a wild-type strain, indicating that the protection by AEB overexpression is very close to 100 %. Because HU deletions are UV sensitive, we tested our deletions of AEB and determined that they are not UV sensitive by themselves (data not shown).

(17K): Fig. 5. Effect of AEB overexpression in a ΔhupAB strain on sensitivity to UV. Strains were grown in LB to OD600 ∼0·5, diluted in 10 mM MgSO4 and irradiated on the surface of LB agar plates. Surviving colonies were scored after 1624 h incubation in the dark. The data are means±SE of four experiments. , OS336 (ΔhupAB) containing pNT2 (crcA+cspE+crcB+); , OS336 containing pBR322.

Overexpression of AEB affects the regulation of several genes
HNS acts as both a silencer and an activator of transcription for a number of genes (Atlung & Ingmer, 1997; Owen-Huges et al., 1992). To determine if overproduction of AEB exhibits any regulatory activity, we tested for effects on gene regulation using lacZ fusions to a number of the genes regulated by HNS.

The transcription of the rcsA gene is silenced 12·9-fold by HNS (Sledjeski & Gottesman, 1995). Overexpression of AEB results in slightly higher (2·2-fold) β-galactosidase activity than that of cells containing the vector, pBR322 (Table 4). This effect may be partially dependent on HNS because activation by AEB is lower (1·4-fold) in an hns mutant strain (Table 4). DsrA is a small RNA that antagonizes the effect of HNS when overproduced (Sledjeski & Gottesman, 1995) but is not required for AEB activation of rcsAlacZ (Table 4). The transcription of the lamB gene is activated 2·5-fold by HNS (Johansson et al., 1998) and the hns gene itself is repressed fourfold by HNS (Dersch et al., 1993). Overexpression of AEB results in a very slight repression (0·6-fold, Table 4) of lamB and a slight activation of hns (1·7-fold, Table 4). We also examined the behaviour of an additional HNS-repressed fusion, proU : : lacZ, an HNS-activated fusion, malE : : lacZ and a fusion that does not respond to HNS, dsrA : : lacZ. Overexpression of AEB has no effect on proU : : lacZ, an effect in the same direction as HNS on malE : : lacZ and unlike HNS, a slight activation of dsrA : : lacZ (Table 4). Thus, regulatory effects from overexpression of AEB are much smaller in magnitude than those seen for HNS and the patterns of gene expression are different for AEB and HNS.

Table 4. Response from fusions of lacZ to various promoter regions when AEB+ are overexpressed

Determination of the phenotypes associated with overexpression of the individual crcA, cspE or crcB genes
In the experiments described above, all three genes (crcA, cspE and crcB) were present on pBR322 and their expression was under the control of their native promoters. To examine the AEB genes more closely, we have cloned crcA, cspE and crcB individually, as well as cspE and crcB together. The individual genes are transcribed from the P_BAD promoter and are induced by arabinose. When crcB is expressed alone, it does not confer resistance to camphor (Fig. 6). crcA expressed alone behaves identically to crcB (data not shown). When cspE is expressed alone, it confers a 10-fold increase in camphor resistance, and in conjunction with crcB, a 100-fold increase in camphor resistance is seen. We have not attempted to overexpress crcA and cspE or crcA and crcB together because these genes are normally controlled by two different promoters. We do not have enough information yet about expression from these native promoters to know how to build viable constructs.

(19K): Fig. 6. Camphor survival curves for crcB, cspE or cspEcrcB overexpressed from the PBAD promoter. The expression of the cspE and crcB genes was induced with 0·25 % arabinose prior to camphor exposure. Viable cell counts were determined at the indicated times. Each assay was conducted in quadruplicate and the data are means±SE. , NT587/pOS3 (cspEcrcB); , NT587/pOS2 (cspE); , NT587/pBAD18-Kan; , NT587/pOS4 (crcB).

When the overexpression of the individual genes was examined for the regulation of rcsA, a similar pattern emerged (Table 5). Overexpression of crcA or crcB alone had no effect on the expression of rcsAlacZ. Overexpression of cspE alone accounted for a 1·7-fold induction and overexpression of cspE and crcB from the P_BAD promoter accounted for the full induction seen from pNT2.

Table 5. Regulation of rcsAlacZ by the individual crcA, cspE or crcB genes under the control of the arabinose-inducible promoter PBAD

We have previously shown that overproduction of crcA, cspE and crcB confers camphor resistance and prevents nucleoids from decondensing both in vivo and in vitro (Hu, 1996). We demonstrate here that strains overexpressing AEB have an increase in supercoiling of resident plasmids and that these increased supercoiling levels can be seen in gyrase, HU and Fis mutants. Overexpression of AEB also compensates for the novobiocin sensitivity of gyrase and topo IV mutants, increases cell viability pf parC1215 and parE10 Ts mutants at the nonpermissive temperature by 10-fold, corrects the nucleoid morphology of topo IV mutants and makes gyrB, parC and parE mutants less sensitive to camphor.

A model for CspE's role in condensing DNA
The results listed above suggest a very simple model for how CspE functions in DNA condensation, namely that CspE stabilizes supercoiled DNA molecules and holds them in a compacted configuration. In this scenario, overexpression of cspE would increase supercoiling levels of resident plasmids in both parental cells and cells defective in supercoiling levels (gyrase, HU or Fis mutants) by stabilizing the population of supercoiled DNA molecules. Stabilizing of the supercoiled DNA molecules by CspE would allow the cell to maintain normal supercoiling levels with less-active supercoiling enzymes. Treating cells with camphor results in decondensed nucleoids. The increase in camphor resistance in gyrB, parC and parE Ts mutants overexpressing AEB as well as suppression of the nucleoid morphology defects in parC and parE Ts mutants overexpressing AEB could also be explained by CspE stabilizing supercoiled DNA.

Suppression of gyrase, topo IV and mukB mutations by overexpression of AEB
Overexpression of AEB will completely suppress the production of DNA-less cells seen in mukB mutations (Yamanaka et al., 1994a). As we have shown here, overexpression of AEB will also efficiently suppress some of the defects associated with gyrase and topo IV Ts mutants. While overexpression of AEB will quite efficiently suppress many of these mutants' defects, it will not completely compensate for the loss of any of these essential proteins: all of the mutants do not form colonies at the higher temperatures. We think that this is because AEB do not possess the enzymic activities of MukB, gyrase or topo IV. Rather, overexpression of AEB, or more likely cspE and crcB (see below), enhances the efficiency of MukB, gyrase or topo IV. mukB mutations are also suppressed by topA mutations (Sawitzke & Austin, 2000). In this case, the suppression has been documented to be the result of an increase in negative supercoiling. If overexpression of CspE stabilizes supercoiled DNA, this could explain the suppression of mukB mutations.

The small DNA-binding proteins
For the small DNA-binding and bending proteins HNS, HU, IHF and Fis, we see effects of overexpression of AEB in some but not all assays. Overexpressing AEB partially compensates for Fis^- for resistance to camphor and plasmid supercoiling. For HU, there is compensation by overexpression of AEB for camphor resistance, plasmid supercoiling levels and the defect in homologous recombination as assayed by UV sensitivity. Overexpression of AEB does not consistently suppress defects in any one of these small DNA-binding proteins; however, the defects most often suppressed by overexpressing AEB are those associated with mutations in HU.

Gene regulation by CspE
We originally tested several HNS-regulated genes to determine if they are also regulated by overexpressing AEB. Two facts emerged from these data. First, the regulation patterns by overexpressing AEB are different from the regulation patterns exhibited by HNS. Second, the magnitude of regulation by AEB is very small at best. We suspect that this level of regulation could be accounted for by changing the condensation state of the chromosome and not through classical mechanisms of activation or repression. This result prompted us to search for other genes that are regulated by CspE to determine if all gene regulation by CspE is twofold or less. Much to our surprise, we have found several genes that are regulated 40-fold or more by overexpression of cspE (A. Kolesar, A. Pryzbylski & N. Trun, unpublished). Further studies using rcsA, as well as these newly identified genes, will shed light on the mechanism(s) of regulation by CspE.

Deletion of crcA, cspE and crcB
Not surprisingly, deleting crcA, cspE and crcB is not lethal. E. coli K-12 contains nine csp genes and it has been shown that only deletion of multiple csp genes results in noticeable growth defects (Graumann et al., 1997; Xia & Inouye, 2001). What is interesting about the deletion of AEB is that cells carrying the deletion are hypersensitive to camphor, suggesting that their nucleoids are more sensitive to decondensation than wild-type nucleoids.

What are the roles of CrcA and CrcB?
CrcA from Salmonella typhimurium has been to shown to be an unusual outer-membrane enzyme involved in lipid A biosynthesis (Bishop et al., 2000). One phenotype of the crcA homologue (pagP) is that it makes cells more resistant to certain antimicrobial peptides (Guo et al., 1998). Our original idea for how overexpression of crcA contributed to the camphor-resistance phenotype was that it provided some resistance to camphor that is independent of overexpression of cspE and crcB. The fact that overexpression of crcA by itself does not confer any camphor resistance on cells indicates that this cannot be the explanation. At least two other possibilities exist. Given that CrcA is a membrane protein and CrcB is predicted to be a membrane protein, perhaps there is an interaction between these two proteins that accounts for the 1000-fold increase in camphor resistance when AEB are overexpressed and the 100-fold increase when only EB are overexpressed. Alternatively, it is possible that expression levels of EB are higher when all three genes are cloned into pBR322 and expressed from their native promoters. Further investigation will distinguish between the possibilities.

It is clear from our data that overexpressing crcB amplifies the phenotypes of cspE mutants. Overexpressing cspE confers a 10-fold increase in camphor resistance and accounts for part of the gene regulation of rcsA (1·7-fold). Overexpressing crcB does not increase resistance to camphor nor does it have any effect on the regulation of rcsA. However, when cspE and crcB are overexpressed together, with the DNA sequence between the two genes being the same as in the chromosome, camphor resistance increases to 100-fold and gene regulation increases to 2·1-fold. All of the gene regulation seen with overexpression of AEB is also seen with overexpression of just EB. The two assays, camphor resistance and regulation of rcsA, do not have the same requirement for overexpression of crcA.

Our data suggest that CspE and CrcB work together to carry out their cellular function(s). One very interesting aspect of this is that CspE is a soluble cytoplasmic protein with nucleic-acid-binding capabilities and CrcB is a protein with four long predicted hydrophobic regions suggesting that it may be a membrane protein. If CrcB is indeed a membrane protein that interacts with CspE, then these two proteins may provide a way to anchor the chromosome to the membrane.

We would like to thank Sue Wickner, Francis Repoila, Dhruba Chattoraj and Susan Gottesman for critically reading the manuscript. C. Gutierrez, F. Repoila, O. Rodionov, E. Bremer, L. Zechiedrich, S. Adhya, J. Beckwith, M. Yarmolinsky and S. Gottesman kindly provided strains and plasmids. This work was supported by the National Cancer Institute, NIH and grants from the National Institutes of Health (GM065121), the Samuel and Emma Winters Foundation and Duquesne University.

Footnotes

^,†, Monica Gingras¹, Nancy Beck^{1 ,‡}, Christine Hall¹ †Present address: Bioinformatics Research Centre, Department of Computing Science, University of Glasgow, Glasgow G12 8QQ, UK. ‡Present address: Department of Microbiology and Immunology, University of Michigan, Ann Arbor, MI 48109, USA. §Present address: Department of Bacteriology, University of Wisconsin, Madison, WI 53706, USA.