parD and chpA are homologous conditional killer systems of plasmid and chromosomal origin, respectively, encoding a killer protein (Kid and ChpAK) and an antidote (Kis and ChpAI). Here it is shown that these systems can functionally interact. A multicopy chpA recombinant partially complements two mutations in the antidote of the parD system. These mutations affect either autoregulation or neutralization of the killer component. Following in vitro mutagenesis with hydroxylamine, chpA mutants that improve this complementation were isolated. Sequence analysis shows that these mutants are clustered in the 5′ end of the chpAI gene and structure predictions suggest that they affect a putative loop in the secondary structure of the ChpAI antidote. It is proposed that this region is part of a protein-protein interface required for the functional interaction between the antidote and the killer components in the two homologous systems.
parD (kis, kid) is a conditional killer stability system found in the drug resistance factor R1 [1, 2]. As with other protein killer gene systems, parD encodes a stable toxin and an unstable antidote. This system shares homology with two conditional killer systems, chpA (chpAI, chpAK) and chpB (chpBI, chpBK), found in the chromosome of E. coli. Kid, the toxin of the parD system, is an efficient inhibitor of the initiation of DNA replication, targeting, most probably, the main replicative helicase of Escherichia coli, the DnaB protein. The Kis protein, the antidote of the Kid protein, is able to form a very stable Kis-Kid complex.
Genetic and biochemical analyses indicate that the parD system is autoregulated by the concerted action of the Kis and Kid proteins. Furthermore, a point mutation at the amino end of the Kis protein (pAB17) derepresses the parD system [1, 6], and an amber mutation which truncates the last 13 amino acids of Kis (pAB1120) abolishes its capacity to neutralize the toxic action of Kid. The Kis protein alone seems to have a weak regulatory activity, which suggests that it is involved in specific protein-DNA interactions. It is possible that the same Kis-Kid interactions are required both for efficient autoregulation and for neutralization of the toxic activity of Kid. The pAB17 mutation, where a proline residue in the amino end of Kis is replaced (P18L), could affect the Kis-Kid interaction; this is suggested by minicell analysis that shows a drastic reduction in the levels of the mutated Kis protein in the absence of wild-type Kid protein.
In this work we examined whether the chpA and parD systems can functionally interact. One objective of this analysis is to open a new genetic approach to identify protein-protein interfaces important for functional antidote-toxin interactions.
Materials and methods
Bacterial strains and plasmids
Bacteria used in these studies were the Escherichia coli K12 strains CC118 (Δ(ara-leu), araD, ΔlacX74, galE, galK, phoA20, thi1, rpsE, rpoB, argE(aur), recA1), MC1000 (araD139, Δ(ara-leu)7697, ΔlacX74, galU, galK, rpsL), and the supF-thermosensitive strain OV2 (F−, leu, thyA (deo), ara(am), lac-125(am), galU-42(am), galE, trp(am), tsx(am), tyrT (supF(ts)A81), ile, his). The plasmids used were the multicopy plasmid vector pUC19 and the parD plasmids pAB17 (kis17, kid+), pAB1120 (kis74, kid+, ΔcopB) and pUUM10, a pBR322-parD (kis+, kid+) recombinant.
DNA isolation and genetic engineering techniques were performed as described by Sambrook et al.. Amplification of genomic chpA sequences by PCR was done in a 100 μl mixture containing: 10 ng of genomic DNA prepared from the E. coli K12 strain MC1000; Pfu polymerase (2.5 U) and its assay buffer as supplied by Stratagene; a mixture of the four dNTPs at 200 μM each; 100 pmol of each of the two primers: 5′-AGAGCTTCAGAATAGAGTGAGTTAGT-3′ and 5′-GGGAATTCGGCCGAAATTTGCTC-3′, which introduce, respectively, an Eco RI target upstream and a Hin dIII target downstream of the chpA operon. Following an initial incubation for 5 min at 95°C, the reaction mixture was incubated for 30 cycles sequentially at 95°C (1 min), 58°C (1 min) and 72°C (1 min). The amplification procedure was terminated with incubation at 72°C for 3 min. In vitro mutagenesis with hydroxylamine, a mutagen that induces GC-AT transitions due to the hydroxylation of cytosines, was carried out for 35 min at 65°C as described. Following extensive dialysis against 10 mM Tris-HCl pH 8.0, 1 mM EDTA, the mutagenized DNA was used to transform the E. coli OV2 strain containing the pAB17 or pAB1120 mutant. DNA sequencing of the chpA recombinant and its mutants was performed, on both strands, by the chain terminator method, using an automated DNA sequencer (Perkin Elmer, Abi Prism, 377 DNA sequencer).
Sequence analyses were performed in a Silicon Graphics, Indigo II workstation after retrieving the sequences of parD, chpA and chpB antidotes from the Genbank Database (web site: http://www.embl-heidelberg.de/srs5. Accession numbers for: Kis, I64783; ChpAI, D16450; ChpBI, D16451). Multiple sequence alignment was done by CLUSTAL-W (version 1.5, web site: http://www.ibc.wustl.edu/msa/clustal.cgi; protein weight matrix: Blosum series; gap opening penalty: 10.00; gap extension penalty: 0.05), and the output (in MSF format) was used as input for the prediction of secondary structure and solvent accessibility by PHD (web site: http://www.embl-heidelberg.de/predictprotein predictprotein.html) [16–18].
Cloning of chpA sequences in the multicopy vector pUC19
A 0.7 kb chromosomal fragment containing the wild-type sequences of the chpA operon flanked by Eco RI and Hin dIII targets was obtained by PCR amplification of the chpA sequences present in the E. coli chromosome as indicated in Section 2. This fragment was digested with the Eco RI and Hin dIII enzymes and cloned directionally in the Eco RI-Hin dIII sites present in the multicopy pUC19 vector. pUC19 recombinants containing the chpA operon were rescued in the CC118 strain as ampicillin-resistant transformants that were defective in α-complementation (white colonies in X-gal/IPTG plates). As expected, the recombinant plasmids isolated from these transformants contained an Eco RI-Hin dII fragment of 0.7 kb. One of these recombinants, pSSA, containing the wild-type chpA sequences, as determined by DNA sequencing analysis, was selected for further studies.
Partial complementation of parD mutants by the multicopy chpA recombinant
To test for possible interactions between components of the chpA operon and the parD system, we searched for complementation of parD mutants by the chpA recombinant pSSA. Two mutants affecting the antidote of the parD system, Kis, pAB17 (kis17, kid) and pAB1120 (kis74, kid), were used as reporters in this analysis. The pAB17 mutant carries a thermosensitive mutation in kis and the pAB1120 mutant carries an amber mutation in this gene that results in a thermosensitive phenotype in a supF(ts) strain (OV2). The thermosensitivity of both mutations is clearly shown by the inhibition of cell growth at 42°C; a partial inhibition of cell growth can already be observed at 37°C; 30°C is a permissive temperature for both these mutants.
To evaluate the possible complementation of the parD mutants by the chpA recombinant, OV2 cells containing either the pAB17 or the pAB1120 plasmids were transformed with pSSA (test), with pUC19 (negative control) or with the parD recombinant pUUM10 (positive control). Cells containing pSSA and either pAB17 or pAB1120 do not grow at 42°C but, at 37°C, they grow better than the negative control and worse than the positive control (Fig. 1). This indicates a partial complementation of the parD mutants by the multicopy chpA recombinant pSSA. Note that the complementation is more obvious in cells containing the pAB17 mutant than in cells containing the pAB1120 mutant, since the latter show poorer thermosensitivity at 37°C. The complementation of mutations in Kis, the antidote of the parD system, by pSSA suggests that the antidote of the chpA system, ChpAI, may also functionally interact with Kid, the killer component of the parD system.
Isolation of chpA mutants that improve the complementation of parD mutants
The lack of complementation of the parD mutants by the pSSA recombinant at 42°C suggested that this temperature could be useful to select for chpA mutants with increased complementation efficiency. For this purpose pSSA was mutagenized in vitro with hydroxylamine and then introduced by transformation in cells containing pAB17 or pAB1120, selecting transformants at 42°C. Four mutants complementing at 42°C the mutation present in pAB17 (A1, A3, A9 and A10) and four others complementing the mutation presented in pAB1120 (A5, A6, A7 and A8) (Fig. 2) were isolated and selected for further characterization. The eight chpA mutants were able to complement both parD mutants at 42°C (data not shown). DNA sequencing of the mutations indicated that in all cases the mutations correspond to C-T (G-A) transitions at the second base of different chpAI codons that introduce missense changes in the ChpAI protein (Table 1). These changes are all clustered at the 5′ end of the chpAI gene and predict the following amino acid substitutions in the ChpAI protein: G10A, S12L (the most frequent one), A14V and R16Q, where the number between amino acids indicates their position with respect to the methionine that initiates ChpAI. The chpAI mutants can be isolated in the OV2 strain, separate from the pAB17 and pAB1120 mutants, and grow efficiently at 30°C, 37°C and 42°C. This indicates that the ChpAI mutant antidotes retain the ability to efficiently neutralize the ChpAK toxin. Therefore, the chpAI mutations enhance the interactions between the chpA and parD systems while maintaining the functionality of the ChpAI antidote.
|parD mutant||chpAI mutants|
|mutant||amino acid (codon) change|
|(kis17, kid)||A3||S12L (TCA-TTA)|
|(kis74, kid)||A6||R16Q (CGG-CAG)|
|parD mutant||chpAI mutants|
|mutant||amino acid (codon) change|
|(kis17, kid)||A3||S12L (TCA-TTA)|
|(kis74, kid)||A6||R16Q (CGG-CAG)|
The bases of the codons affected by the mutations and the corresponding amino acid changes are indicated in boldface type. The number of the amino acid indicates its position with respect to the methionine or the amino-terminal end of the ChpAI antidote. Note that the same mutation was found in five of the eight independently isolated mutants.
Sequence analysis of the ChpAI wild-type and mutant proteins
Fig. 3 shows the predictions of secondary structure and solvent accessibility corresponding to the antidotes of the parD, chpA and chpB systems made by a neural network algorithm (PHD) after multiple sequence alignment (CLUSTAL-W) (see Section 2). The multiple alignment of the antidotes show 42% similarity (20% of sequence identity and 22% of conservative changes) (Fig. 3A). The amino acid changes in ChpAI introduced by the mutations are located at the amino-terminal end of the protein (positions 10, 12, 14 and 16 from the initiator methionine). They are predicted to be in a putative loop in the secondary structure, flanked by a β-sheet and an α-helix, and partially exposed to the solvent (Fig. 3B). Note that a conserved proline (position 18 in ChpAI and in Kis and position 17 in ChpBI) which is altered in the kis17 mutation (P18L) is also located in the predicted loop. The alignment of the killers (data not shown) indicates a pattern of sequence conservation and identity similar to the one previously reported.
It is shown that a small region at the amino-terminal end of the ChpAI antidote can accommodate changes which allow an efficient complementation of parD mutants that interfere with the two roles of Kis: as corepressor of the parD system (pAB17) and as an antidote of the Kid protein (pAB1120). The predicted amino acid changes, G10A, S12L, A14V, and R16Q, affect residues at the amino-terminal end of ChpAI that are located in consecutively alternating positions (residues 10, 12, 14 and 16 starting from the amino-terminal methionine of ChpAI). With the exception of the R16Q change, which reduces the charge of the amino acid, the substitutions tend to increase the hydrophobic character of the residue. In fact the most frequent change found, S12L (in 5 of 8 mutants), substitutes the small and polar amino acid serine by the large hydrophobic amino acid leucine. Secondary structure and solvent accessibility predictions suggest that the amino acid changes are located in a putative and partially exposed loop which is conserved in antidotes of the parD family. We propose that this loop constitutes a functional protein-protein interface motif. The substitutions found could reinforce the efficiency of this oligomerization interface, mainly by hydrophobic interactions.
Interactions between Kis and Kid seem to be required for the formation of the repressor of the system and for neutralization of the killer activity [5, 6]. A simple explanation for the complementation observed is that the chpAI mutations improve the interaction of the ChpAI antidote with the killer component of the parD system, Kid, favoring the formation of ChpAI-Kid complexes that substitute effectively for the mutated Kis-Kid complexes. If the same antidote-toxin complex is assumed to be involved both in the inactivation of the Kid toxin and in the autoregulation of parD, a deficient interaction of the antidote with the killer protein would explain the increased toxicity and the derepression of the parD system caused by the mutation in pAB17. Based on this hypothesis the region of the antidote identified by the mutations should modulate, in members of the parD family, the activity of the killer component and the level of expression of the operon.
The Kis-Kid complex appears to be a hetero-tetramer consisting of two Kis and two Kid proteins. This raises at least one additional possibility that can explain the complementation observed: the formation of an active ChpAI-Kis/Kid-Kid complex, containing a wild-type killer protein and the two mutated homologous antidotes. Further experiments are needed to distinguish between these alternatives. Our data further show that the amino-terminal end of the antidotes can accommodate changes that lead to functional convergence or divergence in members of the parD family.
This research has been financed by Projects BIO94-0707 (CICYT) and PB94-0127 (DGICYT). We are grateful to Guillermo de la Cueva for a critical reading of the manuscript. The excellent technical assistance of Consuelo Pardo Abarrio and Ana Maria Serrano López is acknowledged. We also acknowledge the contributions of the oligonucleotide synthesis, DNA sequencing and photography services of the CIB.