TEXT

Top Abstract Text References

Within a few minutes after bacteriophage T4 infection, the spatial distribution of the Escherichia coli chromosome is dramatically altered (12, 13). The T4 ndd gene is responsible for this nuclear disruption phenomenon (17, 18), which converts the large, central nucleoid into numerous small DNA globules on the inner membrane (4). The highly basic 17-kDa Ndd protein has no significant homology with any other known protein (5). About 4,000 Ndd molecules are produced by T4-infected cells (11), but significantly lower levels of expression of the cloned ndd gene are nonetheless lethal to E. coli and induce a slow disorganization of the nucleoid (4). Ndd protein has little effect on bacterial gene expression but inhibits replication apparently by generating on the chromosome obstacles to progression of replication forks (4). Even after Ndd has caused extensive cell killing and disorganization of the nucleoids, no bacterial DNA cleavage or degradation is detected nor is the SOS system induced (4). These observations indicate that only the architecture of the nucleoid is affected and suggest that Ndd might interact directly with elements that determine the conformation and the location of the bacterial chromosome within the cell.

In this communication, we report that a high level of cloned ndd gene expression induces nuclear disruption as rapidly and completely as T4 infection, thus indicating that Ndd is the only T4 protein required for nucleoid disruption. Extracts from cells that overexpress Ndd protein were used to study Ndd activity in vitro. We present evidence that Ndd is a DNA binding protein with some specificity for sequences located on the bacterial chromosome.

Cloning a thermosensitive ndd allele under the control of a T7 promoter. Attempts to clone the wild-type ndd gene in pET11a vector (Stratagene) under the control of the tightly regulated T7 promoter (19) were unsuccessful, even in a host that did not carry the T7 RNA polymerase gene. A few Ndd protein molecules were probably made from this plasmid, and they sufficed to kill E. coli (4). However, we could clone the temperature-sensitive ndd2 allele (4, 15) under the control of this T7 promoter, provided the transformed cells were maintained at 42°C, a nonpermissive temperature for Nddts2 protein. The resulting plasmid, pJYB41, was then introduced into strain BL21(DE3), which carries the T7 RNA polymerase gene under the control of the lac promoter so that the addition of IPTG (isopropyl--D-thiogalactopyranoside) induces T7 RNA polymerase synthesis. This strain (LN3243) now produces large amounts of soluble Nddts2 protein after a shift down to 30°C and addition of IPTG. The synthesis of the Nddts2 protein can be detected soon after induction by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, and by 60 min, Nddts2 accounts for about 8% of the total protein stained by Coomassie blue (data not shown).

Overproduction of Nddts2 induces a rapid and complete disruption of the nucleoid. When Nddts2 protein is produced by LN3243 bacteria, a rapid disruption of the nucleoid occurs. This is illustrated in Fig. 1, where the DNA of such cells was stained with the fluorescent dye DAPI (4',6-diamidino-2-phenylindole) (8) and photographed by fluorescence microscopy (as described previously [4]). Clearly, Nddts2 production for just 10 min is sufficient to cause the redistribution of the DNA from the center of the cells to globules at the periphery of the inner membrane (compare the blue color distributions in Fig. 1B and D). Thus, a high level of Ndd provokes the rapid and complete disruption of the nucleoid, and no other T4 gene products are necessary. This correlates well with the fact that nucleoid disruption during wild-type T4 infection is achieved within a few minutes (18), when about 4,000 molecules of Ndd per cell are produced (11), and further suggests that Ndd be required to act at many sites simultaneously to bring about the destruction of the nucleoid. The previous observation that induction of Ndd expression produced only slow nuclear disruption is explained by the lower level of the protein in these experiments compared to that present after T4 infection (4).

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Complete disruption of the chromosome by Nddts2. Strains LN3243 (Nddts2 producer) and LN3245 (control) were grown at 42°C to an OD540 of 0.2. DAPI (2 µg/ml) was added, and growth was allowed to continue at 42°C for 30 min. Cells were then shifted to 30°C by dilution with cold medium and the concomitant addition of 1 mM IPTG. Ten minutes later, the cells were chilled on ice and observed by use of phase-contrast or fluorescence microscopy with a Leica DMRB microscope. (A) LN3245 cells, phase contrast; (B) LN3245 cells, fluorescence; (C) LN3243 cells, phase contrast; (D) LN3243 cells, fluorescence.

Ndd binds to double-stranded DNA (dsDNA) but not to single-stranded DNA (ssDNA). The highly basic composition of the Ndd protein (isoelectric point, 10.5) raises the possibility that Ndd has a DNA binding activity that targets it to the nucleoid. To investigate this prospect, the ability of Ndd-containing crude extracts to retain labeled DNA on filters was assayed (1). The extracts containing overexpressed Nddts2 were prepared from strain LN3243 and control extracts were prepared from parental strain LN3245 [BL21(DE3)/pET11a] as follows. LN3243 or LN3245 cells were grown at 42°C in ampicillin-containing LB medium (15) (100 µg/ml) to an optical density at 540 nm (OD₅₄₀) of 0.5. The cultures were shifted to 30°C, and 10 min later IPTG (1 mM) was added to derepress the ndd2 gene. After 90 min, the cultures were chilled on ice, centrifuged, and resuspended at an OD₅₄₀ of 10 in buffer A (Tris-HCl, 50 mM; NaCl, 50 mM; EDTA, 1 mM). Cells were lysed on ice by a lysozyme treatment (0.2 mg/ml) for 30 min, followed by a brief sonication. Crude extracts were then centrifuged for 10 min at 15,000 rpm to remove debris, and supernatants were stored at 70°C before use. Protein concentrations were determined by Bradford analyses (Bio-Rad). These extracts typically contained about 1.4 mg of protein per ml, with Ndd constituting about 8% of the total. ³²P-end-labeled Sau3AI E. coli (from strain PB4144) (4) genomic fragments were incubated with such extracts. Compared with control extracts, the Ndd-containing extracts had a 25- to 30-fold-increased capacity to retain this DNA (Fig. 2). If the reaction mixture was incubated at 44°C instead of room temperature, the DNA retained by Ndd2ts extract was reduced 3.2-fold. That the DNA binding activity of Nddts2 mutant protein is thermosensitive in vitro is also consistent with this DNA binding activity being involved in in vivo nucleoid disruption by Ndd2ts, which is also thermosensitive in this mutant. When denatured DNA (E. coli genomic DNA heated at 100°C followed by a rapid cooling on ice to produce single-stranded DNA) was used in this binding assay, DNA retention was very low and no difference could be detected between the Ndd-containing and control extracts (data not shown). The absence of binding to single-stranded DNA argues strongly against Ndd having a nonrelevant DNA binding activity simply due to its basic character. The double-stranded DNA binding activity detected in cell extracts when the ndd gene is overexpressed is most likely a property of the Ndd protein itself, and this activity is probably involved in the Ndd effect on nucleoid structure.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Filter binding assay for Ndd DNA binding activity and specificity for E. coli DNA. E. coli DNA (circles) or T4 cytosine-containing DNA (squares) were digested with Sau3AI prior to end labeling with 32P. Increasing amounts of these DNAs were incubated at room temperature for 20 min with Ndd-containing LN3243 (solid lines) or control LN3245 (dotted lines) protein extracts (15 µg of total protein) in 500-µl reaction mixtures containing 250 µl of 2× DNA binding buffer (40 mM Tris-HCl [pH 7.5], 20% glycerol, 100 mM KCl, 0.2 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 20 mM EDTA). Reaction mixtures were filtered through GF/C filters (Whatman). The filters were washed three times, dried, and immersed in liquid scintillation cocktail (Ready-Safe; Beckman). The DNA retained on each filter was monitored by scintillation counting (Beckman LS3801). The fraction of DNA retained is plotted as a function of the amount of DNA incubated in the reaction mixture.

Ndd has no endonuclease activity, and its binding to dsDNA is insensitive to topology. Upon incubation of purified covalently closed circular (CCC) or open circular (OC) plasmid DNA (pFGB68; kindly provided by F. Boccard) with Ndd2ts-containing extracts, neither interconversion between these species nor their transformation to linear DNA was detected (Fig. 3, lanes i and j). Thus, binding of Ndd to DNA is not accompanied by single-strand or double-strand cleavage. This result is consistent with the absence of DNA degradation observed in vivo after ndd gene expression induction (4).

View larger version (119K): [in this window] [in a new window]   FIG. 3.   Ndd binds equally to different forms of dsDNA. Plasmid pFGB68 DNA was electrophoresed in a 0.8% agarose gel (1× TAE buffer [15]) to separate its supercoiled (CCC), OC, and L forms. These forms, either alone or mixed in known amounts, were incubated with extracts and subjected to the retention assay. Lanes a and m, DNA ladder in kilobases (Bethesda Research Laboratories); lane b; CCC plus OC DNA mixture; lane c, CCC, OC, and L DNA mixture; lanes d to g, the same DNA mixture as that in lane c, but incubated with Nddts2 extracts (in increasing amounts from lane d to g), filtered on GF/C membranes, eluted from the filter with 0.5% SDS, precipitated with ethanol, and dissolved in Tris-EDTA buffer prior to electrophoresis; lane h, same as that in lane f, but incubated with control extract; lanes i to l, CCC and OC DNA incubated with Nddts2 extract (lanes i and j) or control extract (lanes k and l) and then treated as described for lane d. A DNA size scale (in kilobases) is indicated on the right. Positions of CCC, OC, and L forms of pFGB68 are indicated on the left. N, Nddts2-containing extract; C, control extract.

Ndd has a higher affinity for E. coli DNA than for T4 DNA. Since during the infectious cycle the replicating phage DNA occupies a central position within the cell, T4 DNA must be relatively insensitive to the disruptive action of Ndd (18). The T4 DNA is different from E. coli DNA because (i) it contains hydroxymethylcytosine instead of cytosine and (ii) the hydroxymethylcytosine is glucosylated (7). To test the possibility that specific DNA sequences are recognized by Ndd, we have compared the affinities of Ndd for E. coli DNA and for phage T4 DNA containing none of these modifications. By the use of an appropriate multiple T4 mutant (T4-Cyt) that lacks the capacity to produce the modified base hydroxymethylcytosine, we were able to prepare cytosine-containing T4 DNA (6). This allowed us to compare the affinities of Ndd for E. coli DNA and for T4 DNA without concern for the effects of the modified bases in the phage DNA.

Conclusions. Our results demonstrate that Ndd is the only T4 protein required for complete disruption of the host bacterial nucleoid and that rapid and complete disorganization of the nucleoid requires a high level of expression of the cloned ndd gene. A DNA binding activity of Ndd has been demonstrated, and its preliminary characterization indicates a strong preference for dsDNA rather than ssDNA. So far, the dsDNA binding activity seems insensitive to the topological state of the DNA. The Ndd DNA binding activity displays some substrate preference, however, with Ndd binding to bacterial DNA at least fivefold more efficiently than to nonglycosylated cytosine-containing T4 phage DNA. This suggests that DNA sequence motifs present on the bacterial chromosome, but absent from phage DNA, might be responsible for the preferential binding. The targets for Ndd binding are probably numerous in the nucleoid, and nucleoid disruption may be a direct consequence of such binding. One can imagine at least two different ways in which the interaction of Ndd with DNA could mediate nucleoid disruption. It could be that Ndd binds to the chromosome at numerous locations and then transfers these sequences to a new peripheral location along the inner membrane. In this case, Ndd would superimpose its effects on the normal nucleoid organization. Alternatively, Ndd could interfere with elements determining the architecture of the nucleoid. Thus, Ndd would selectively destroy the normal nucleoid organization. Its target could be, for instance, a protein core (or skeleton) about which the chromosomal DNA may be structured, with DNA loops extending into the cytoplasm (10, 16). Ndd could irreversibly destroy the interactions between the core and the cognate chromosomal DNA sequences by binding to these sequences. The repulsive forces between charged DNA could move the liberated parts of the nucleoid toward the inner membrane.