The ability of each of the nine Escherichia coli division proteins (FtsZ, FtsA, ZipA, FtsK, FtsQ, FtsL, FtsW, FtsI, FtsN) to interact with itself and with each of the remaining eight proteins was studied in 43 possible combinations of protein pairs by the two-hybrid system previously developed by the authors' group. Once the presumed interactions between the division proteins were determined, a model showing their temporal sequence of assembly was developed. This model agrees with that developed by other authors, based on the co-localization sequence in the septum of the division proteins fused with GFP. In addition, this paper shows that the authors' assay, which has already proved to be very versatile in the study of prokaryotic and eukaryotic protein interaction, is also a powerful instrument for an in vivo study of the interaction and assembly of proteins, as in the case of septum division formation.

FtsQ, an essential protein for the Escherichia coli divisome assembly, is able to interact with various division proteins, namely FtsI, FtsL, FtsN, FtsB and FtsW. In this paper, the FtsQ domains involved in these interactions were identified by two-hybrid assays and co-immunoprecipitations. Progressive deletions of the ftsQ gene suggested that the FtsQ self-interaction and its interactions with the other proteins are localized in three periplasmic subdomains: (i) residues 50-135 constitute one of the sites involved in FtsQ, FtsI and FtsN interaction, and this site is also responsible for FtsW interaction; (ii) the FtsB interaction is localized between residues 136 and 202; and (iii) the FtsL interaction is localized at the very C-terminal extremity. In this third region, the interaction site for FtsK and also the second site for FtsQ, FtsI, FtsN interactions are located. As far as FtsW is concerned, this protein interacts with the fragment of the FtsQ periplasmic domain that spans residues 67-75. In addition, two protein subdomains, one constituted by residues 1-135 and the other from 136 to the end, are both able to complement an ftsQ null mutant. Finally, the unexpected finding that an E. coli ftsQ null mutant can be complemented, at least transiently, by the Streptococcus pneumoniae divIB/ftsQ gene product suggests a new strategy for investigating the biological significance of protein-protein interactions.

The development of a convenient and promising alternative to the various two-hybrid methods that are used to study protein-protein interactions is described. In this system, a lambdoid chimeric operator is recognized by a hybrid repressor formed by two chimeric monomers whose C-terminal domains are composed of heterologous proteins (or protein domains). Only if these proteins efficiently dimerize in vivo is a functional repressor formed able to bind the chimeric operator and shut off the synthesis of a downstream reporter gene. This new approach was tested with several interacting proteins ranging in size from less than 100 to more than 800 amino acids and, to date, no size or topology limit has been detected.

The ability of each of the 11 Streptococcus pneumoniae division proteins to interact with itself and with each of the remaining proteins was studied in 66 combinations of protein pairs, using a bacterial two-hybrid system. Interactions (homo-or hetero-dimerizations) were detected between 37 protein pairs, whereas 29 protein pairs did not interact. In some cases, positive interactions of the S. pneumoniae proteins were confirmed by co-immunoprecipitation experiments in Escherichia coli. Comparison between the S. pneumoniae division protein interaction web and that of E. coli, the only micro-organisms for which the whole division interactome has been described systematically, was also performed. At least nine division proteins, ZapA, FtsZ, FtsA, FtsK, FtsQ/DivIB, FtsB/DivIC, FtsL, FtsI and FtsW, are believed to have a conserved function between these bacteria and thus we may say that a significant part of the interactions are conserved. Out of 45 protein pairs tested in both bacteria, 30 showed the same behaviour: 23 interacted while seven did not. In agreement with these results, cross-interactions between S. pneumoniae proteins and the corresponding E. coli orthologues were observed. Taken together, these results suggest a phylogenetically conserved minimal common interactome of the division proteins.

In order to clarify the first step in Mu integrative recombination, we have infected a bacterial strain harboring the plasmid pBR322 and isolated Mu DNA in a supercoiled form associated with this plasmid. These structures show an association of Mu with pBR322 without any preliminary replication.Integrative recombination is a basic step in the life cycle of the temperate and mutator bacteriophage Mu. This phage, which has developed a system permitting random insertion into the host genome, is also a potent transposable. element (1).Mu insertion is accomplished by means of the viral extremities (att sequences) and gives rise to linear and nonpermutated integration within a given gene (2). The host rec system does not take part in the process (3), but there is a preferential integration within the host replication fork (4-6) and an active replication of bacterial DNA is essential (7).It is not clearly -established whether, upon, infection, Mu DNA replicates before or after its integration into the host chromosome. We. have investigated the nature of the first step in Mu integrative recombination in order to elucidate whether the infecting Mu DNA is conserved or not during integration.The basic idea of the experiment is schematized in Fig. 1. We have studied' the insertion of Mu in pBR322 upon infection ofa bacterial strain harboring this plasmid and analyzed the supercoiled structure obtained in CsCl density gradients.The system was chosen for the following reasons: (i) the association between Mu and pBR322 gives supercoiled structures easy to separate from bacterial DNA; and (ii) the density of the plasmid resulting from the insertion ofMu in pBR322 is virtually the density of Mu because pBR322 is small compared to Mu (4 vs. 37 kilobase pairs) and does not appreciably modify the plasmid density in density labeling experiments.The experiments reported here show that a portion of recoverable Mu DNA is indeed in a supercoiled structure physically joined. to pBR322, but still in a nonreplicated form.To (8), then the medium was sterilized and' supplemented with 32Pi at 50 puCi/ml. The lysate was collected after 6 hr at 37TC, centrifuged 15 min at 7,500 rpm in a Sorvall SS34 rotor to eliminate bacterial debris and agar, and then concentrated by centrifugation in a Beckman 40 rotor at 23,000 rpm for 2 hr.The pellet was resuspended in 0.3 ml of Mu buffer and purified on a CsCl step gradient (8). The phage band was collected with a syringe and then dialyzed against three changes of Mu buffer (buffer to phage solution, 1,000: 1). Phage titer of the stock after dialysis was 1012 plaque-forming units/ml, specific activity was 2 X 10-6 cpm per phage.Mu Integration Experiments. A 5-ml culture of strain HUC166 was prelabeled in M9.HT medium for approximately 10 generations; thymine (200 ,ug/ml) was added to the medium to favor thymidine incorporation (7). The culture was then diluted 1: 100 in. 25 ml of fresh M9. HT medium supplemented with thymine at 200 ,ug/ml and grown to 2 X 108 cells per ml. Bacterial cells were c...