Introduction

Lysis inhibition (LIN), as a response of bacteriophage T4-infected cells to superinfection by another T-even phage is a phenomenon known for more than 60 years. Up to date, the major players have been identified and characterized, but despite its crucial role in elucidating fundamental biological processes, the mechanism of this phenomenon remains mostly unexplained [1–3].

During intracellular development of a T-even phage, superinfection by another T-even phage at least 3 min after the primary infection leads to prolongation of viral development inside the cell [1, 4, 5]. When repeated, superinfections may lead to very long latent periods, lasting for a few hours, increasing the number of progeny virions up to one order of magnitude [5]. In profiles of lysis of bacterial cultures, caused by T4 phage, this prolonged development is suddenly disrupted and rapid, almost synchronous, lysis occurs. This stage is called LIN collapse, and it may be caused by a damage of bacterial cell walls by superinfecting phages. When the damages accumulate to the critical point, the cell lyses and releases phage progeny. The locally increased concentration of phages causes a damage of neighboring cells, which also release phages. This kind of a chain reaction leads to quick lysis of the infected bacterial culture [6].

The most important factors in LIN are phage-encoded proteins: holin T and antiholin RI. RI is a small (97 aa, 11.1 kDa) periplasmic, hybrophobic protein with pI = 4.0 [1]. The C-terminal domain plays a key role in RI activity [7]. This protein is extremely unstable with a half-life of ~2 min [3]. According to a report by Tran et al. [3], RI protein is localized in the periplasmic side of cell membrane by its N-terminus, which was identified as a signal anchor release domain. When the C-terminal domain is released, it can interact with holin T and block its opening.

A signal triggering the LIN response is still not identified. It is assumed that small proteins and/or phage DNA may be a source of such a signal. This conclusion was drawn from experiments performed by Rutberg and Rutberg [5] which shows that phage ghost cannot trigger LIN. Thus, simple adsorption and penetration of the cell wall is not enough to provoke this phenomenon. However, the antiholin–holin system of T4 is also able to react to a signal generated during slow growth of bacterial cells by delaying the cell lysis [8]. This indicates that superinfection is not the only event which may trigger LIN or LIN-like response.

The rI gene (location 59594–59202 on genetic map) and its neighbors, r1.1 and r1.-1, form an operon. Genes from this operon are transcribed from two promoters: the late promoter located between tk and rI.1 genes and the early promoter located 3.3 kb upstream of the late promoter. All genes from this operon were suggested to be highly conserved among T-even phages [1]. The RI.-1 protein is hydrophilic, with calculated pI = 5.6. It is a 129 aa long polypeptide with predicted molecular mass of 14.6 kDa [1, 2]. The rI.1 gene encodes a 70 aa long protein, with calculated pI = 10.2 and molecular mass of predicted 8.3 kDa [2]. Similarly to RI.-1, its function is unknown. However, the localization of RI.1 and RI.-1 in one operon, and conservation of sequences of genes coding for these proteins, may suggest their role in LIN [1].

A gene coding for RIII, a protein also involved in LIN, is located in a different region of the T4 genome than the rI operon (location 131033–130785 on genetic map). This protein, with calculated pI = 8.5, is 82 aa long with predicted molecular mass of 9.3 kDa [1]. It is supposed to localize in the cytoplasm, and it has been proposed that it stabilizes the RI–T interaction [1]. Due to its localization, it was assumed that it interacts with the cytoplasmic domain of the T holin [9] (gene location 160221–160877 on genetic map). It is likely that transcription of rIII can be initiated at three independent promoters: early, middle, and late [1].

The model proposed by Ramanculov and Young [10] assumes that RI interacts (in the periplasm) with the T holin, which is embedded in cell membrane. This interaction is very unstable, but it is able to block the opening of a lesion in the membrane made by the holin. This unstable interaction was proposed to be stabilized by RIII interacting with the cytoplasmic domain of the T holin [9].

The aim of this study was to investigate a possible role for accessory genes rI.-1 and rI.1 (in combination with rIII) in the regulation of LIN.

Materials and methods

Bacterial and phage strains

Escherichia coli MC1061 [11] was used as a bacterial host in all experiments. Bacteriophage T4D was obtained from Dr. Józef Nieradko, Department of Microbiology, University of Gdansk, Poland. Phages T4rI (r48) and T4rIII (r67), mutants of phage T4D were obtained from Prof. Karin Carlson, University of Uppsala, Sweden.

Plasmids and primers

Plasmids used in this work are listed in Table 1, and primers are listed in Table 2.

Table 1 Plasmids
Full size table
Table 2 Primers
Full size table

Lysis profiles (LIN collapse)

A culture of a bacterial strain bearing indicated plasmid was grown overnight at 37°C with shaking in LB medium supplemented with 34 μg/ml of chloramphenicol. Bacteria were centrifuged, the pellet was washed three times with a fresh LB medium, and then inoculated (1:100) into a fresh LB medium without chloramphenicol, but supplemented with indicated amounts of autoclaved chlortetracycline (cTc). Autoclaved cTc is commonly used as inducer for tetracycline promoter, as it is not toxic for bacterial cells. In the case of strains bearing plasmids containing the rI.1 gene and lacking the rIII gene, the induction of gene expression was performed at the time of addition of phages. The culture was grown to OD600 = 0.1 at 37°C with shaking, then it was divided into three shaking flasks, cultivated in the same conditions, and T4 phage lysate was added to m.o.i. = 0.2 to each flask. OD600 was measured in 10 min intervals.

Results

In order to investigate the influence of the overexpression of phage genes on phage development, we constructed a series of plasmids (Table 1) based on pCattTrE18 [12]. As a control, we used the same plasmid with the lacZ gene cloned under the control of the tetracycline promoter (p tet). Interestingly, the pCattTrE18 plasmid without any insert ceased LIN completely and slowed down the growth of the bacterial culture whereas introduction of any DNA fragment into MCS resulted in total abolition of these effects (data not shown). Moreover, all phage strains used in the study gave pin-point plaques when titrated on the strain bearing pCattTrE18, contrary to normal plaques on the plasmid-less strain. However, inserting any DNA sequence into the MCS also caused reversion of this phenotype (data not shown). We decided to use, in the control experiments, bacteria bearing a plasmid with the lacZ gene, because an induction of a promoter in the plasmid could potentially cause a shift in the lysis timing. Thus, it would be inappropriate to compare plasmid-bearing cells with those devoid of a plasmid, as overproduction of protein(s) from the plasmid may change the lysis profile. Moreover, the use of bacterial cells devoid of plasmid as a control would result in introduction of another parameter, which might be confusing.

Induction of gene expression caused prolongation of lysis profiles in E. coli MC1061 (pCattTrE18 lacZ). This behavior may be attributed to extensive protein production and thus a decrease in availability of the protein synthesis system and amount of precursors in early stages of bacteriophage development. Thus, these cells cannot be directly compared to strain E. coli MC1061, which was not affected by the inducer. Nevertheless, experiments with plasmid-less cells were performed by us, and generally all gave curves similar to those of E. coli MC1061 (pCattTrE18 lacZ) without induction (data not shown).

We performed experiments using two different concentrations of the inducer (autoclaved cTc): 1 and 5 μg/ml. In experiments without the inducer, a leakiness of the promoter allowed for a negligible expression level of a cloned gene. The p tet activity under conditions of various inducer concentrations was investigated previously. Those studies demonstrated a negligible promoter activity without the inducer, and then, an increase in the p tet activity that was proportional to the inducer concentration [12–14].

We observed that overproduction of RI.1, or overexpression of the whole rI operon, had a strong toxic effect on host cells. When bacteria bearing plasmids encoding RI.1 were induced, the only survivors were cells with spontaneous IS1 transposon insertion between the promoter and the rI.1 gene, which was found after sequencing of plasmids from several colonies (data not shown). The overproduction of RI.1 resulted in filamentation of bacterial cells (data not shown). The toxicity of rI, but not of rI.1 was suggested by Ramanculov and Young [10]. We did not observe any toxic effects of overproduction of the RI protein. Due to RI.1 toxicity, induction of expression in cells harboring plasmids which contain the rI.1 gene, but not rIII, was provoked at the moment of phage addition. Otherwise, lysis profiles were significantly affected by the toxic effect, or even impossible to obtain, as bacteria ceased growth before reaching OD600 = 0.1, or were growing extremely slowly (data not shown).

Induction of expression during phage addition was used also for plasmids in which rI.1, rI, and rIII genes were located together, as expression of rI removed protection given by rIII against toxic effect caused by sole overproduction of RI.1. Contrary to wild-type T4D phage, infection of bacterial culture by the T4rI or T4rIII mutants resulted in a very fast lysis of the whole culture. LIN collapse experiments with phage T4rI showed that using plasmids constructed in this work, including those overexpressing the rI gene, we could not obtain any phenotype reversion and all lysis profiles were exactly matching those obtained in control experiments with lacZ overexpression and with plasmid-less cells (Fig. 1, 1–3). This is consistent with the results of other researchers [1 and references therein], indicating that rI is absolutely necessary for LIN. The lack of reversion of the phenotype, even when rI was overexpressed from a plasmid, was probably due to a very low stability of this protein [2], which, together with the very fast silencing of the host gene expression, caused a lack of this protein in later stages of phage development.

Fig. 1
figure 1

Lysis of bacterial cultures by bacteriophage T4rI. The plasmids present in bacterial cells are indicated on left side of the figure, the concentration of inducer is indicated on the top. Filled squares show lysis curves of bacterial cells with indicated plasmid. Open circles show lysis curves of a control experiment with pCattTrE18 lacZ. Experiments which are not shown in the figure have given results identical with other experiments in all concentrations of the inducer tested. For the construct of each plasmid, see Table 1

Full size image

Contrary to T4rI, T4D lysis profile was sensitive to overexpression of several genes, and in most of cases the lysis profile was also dependent on the dose of the inducer (Fig. 2). The most pronounced elongation of the lysis profile was obtained by overproduction of RIII (Fig. 2, 10–12), especially together with RI.-1 (Fig. 2, 19–21) or RI.1 (Fig. 2, 16–18). When rIII was present on the plasmid together with rI.-1 or rI.1, even a very low level expression, resulting from a leakiness of the promoter, was sufficient to prolong the lysis profile considerably (Fig. 2, 16, 19). When the efficient overexpression system was used, rIII alone caused slowing down of bacterial growth. This effect was not observed when rIII was co-expressed with rI.-1 or rI.1. Efficient overexpression of rIII, rIII together with rI.1, and rIII together with rI.-1 caused a loss of the rapid lysis phenotype, when T4rIII phage was used for infection (Fig. 3, 4–6, 10–12, 13–15). In the same experimental system, an effective prolongation of LIN was observed after infection with T4D (Fig. 2, 10–12, 16–18, 19–21). Results of this set of experiments showed a synergistic effect of rI.1 and rI.-1 in prolongation of LIN when these genes were co-expressed with rIII. Low expression level of rIII alone gave no effect on LIN caused by T4D or T4rIII infection.

Fig. 2
figure 2figure 2

Lysis of bacterial cultures by bacteriophage T4D. The plasmids present in bacterial cells are indicated on left side of the figure, and the concentration of inducer is indicated on the top. Filled squares show lysis curves of bacterial cells with indicated plasmid. Open circles show lysis curves of a control experiment with pCattTrE18 lacZ. For the construct of each plasmid, see Table 1

Full size image
Fig. 3
figure 3

Lysis of bacterial cultures by bacteriophage T4rIII. The plasmids present in bacterial cells are indicated on left side of the figure, and the concentration of inducer is indicated on the top. Filled squares show lysis curves of bacterial cells with indicated plasmid. Open circles show lysis curves of a control experiment with pCattTrE18 lacZ. Experiments were performed with all plasmids constructed in this work. Experiments which are not shown in the figure have given results identical with control experiments at all concentrations of the inducer tested. For the construct of each plasmid, see Table 1

Full size image

In the set of experiments with lysis profiles of T4D, we also observed more subtle effects caused mainly by overproduction of RI.1 (Fig. 2, 4–6) and RI.-1 (Fig. 2, 7–9) proteins, which was not accompanied by overproduction of RIII. Among them, RI.-1 caused prolongation of the lysis at both tested concentrations of the inducer. A slight shortening of the time of lysis profile was observed in the control experiment. Similar phenomenon was observed when rI.1 was overexpressed alone (Fig. 2, 4). When rI was co-expressed with rI.1, prolongation of LIN profile was observed, especially when 5 μg/ml of cTc was used (Fig. 2, 24). In experiments with 1 μg/ml of cTc, the prolongation was slight in relation to the control experiment, and when no inducer was added the LIN profile showed premature termination (Fig. 2, 23, 22, respectively). Interestingly, co-overexpression of rIII in this system caused a cease of the effect obtained with rI and rI.1 alone (Fig. 2, 31–33). However, in experiments where T4 rIII was used, we observed no effect of co-overproduction of RI and RI.1 (data not shown). Moreover, the overexpression of whole rI operon resulted in only a slight prolongation of LIN at 5 μg/ml of cTc, and under other conditions the lysis profile was terminated prematurely (Fig. 2, 28–30). This profile of lysis was not affected by parallel overexpression of the rIII gene (Fig. 2, 37–39).

Discussion

Lysis inhibition is already known for over 60 years, and it was used for elucidation of fundamental biological processes. On the other hand, the mechanism of LIN remained mostly unraveled. It is also not established which genes are involved in this process. In this study we tested a possible role for accessory genes rI.-1 and rI.1 in the regulation of LIN.

Overproduction of phage proteins from a plasmid (or from the bacterial chromosome) has one disadvantage—the production ceases shortly after T4 infection. This is due to the modification of the host RNA polymerase by the product of the phage gene asiA and then the fragmentation of host DNA. Thereby, all interferences we could observe were caused by stable proteins, which could act in trans. This would explain a lack of effects of rI overexpression on both plaque morphology and LIN collapse profiles.

In recent studies on rI, a very short, about 5 min, delay in the lysis was observed by Tran et al. [3], which was explained by a short half-life period of the RI protein. Thus, it seems that this protein have to be produced constantly in order to provide proper pattern of LIN. Opposite effect was observed while overproducing RIII. This overproduction can restore wild-type phenotype in rIII mutant, thus, the RIII protein is stable enough to allow for prolonged LIN, but only in the case of efficient overproduction. When rIII was overexpressed from a plasmid, the reversion of the phenotype of T4rIII plaques was also observed (data not shown).

Lysis inhibition collapse experiments, performed in this study, showed that the action of the proteins in phage development cycle may be dose-dependent. In the case of infection of the host harboring a plasmid with the rI gene by phage T4rIII, a weak effect on lysis prolongation was visible only in non-induced cultures. This may suggest that a low level expression of rI, resulting from residual activity of the p tet, may influence phage T4rIII development more significantly than efficient overexpression of this gene.

We were unable to obtain a reversion of the rI phenotype, which causes inability to form wild-type plaques and to establish LIN in response to superinfection, by overproduction of any of the investigated proteins. This supports the thesis about a key role of this protein in LIN [1, 9].

Results of overexpression of the rIII gene suggest its supportive role in stabilizing RI–T interaction, as it was not only able to reverse the effect of the rIII mutation (Fig. 3, 4–6), but also to prolong LIN caused by T4D (Fig. 2, 10–12). The side effect of rIII overexpression was a slight toxicity, which caused a slower growth of host cells. Interestingly, parallel overexpression of rI.-1 (Figs. 2, 19–21; 3, 13–15) or rI.1 (Figs. 2, 16–18; 3, 10–12) not only strengthened the effect of prolonging LIN by rIII, but also ceased RIII toxicity, as growth of bacterial cells reached the normal level. This suggests that proteins RI.-1 and RI.1 play a supportive role in LIN by direct or indirect interactions with RIII. This suggestion is supported by the fact that overexpression of rI.-1 or rI.1 prolonged LIN of T4D development (Fig. 2, 4–9), but had no effect during T4rIII infection (data not shown), even though fully functional RI protein was encoded by the phage genome.

Another interesting observation is a lack of any effects of overproduction of RIII when rI was co-expressed (Figs. 2, 13–15, 31–39; 3, 7–9, 16–18). These results suggest a direct interaction between both proteins, RI and RIII, which may lead to inactivation of RIII (Fig. 4b). This suggestion seems to be supported by a toxicity of overexpression of genes rI.1, rI, and rIII from one plasmid, where the presence of rI ceased the effect of the rI.1–rIII mutual antitoxicity. Moreover, the toxic effect was of the type caused by RI.1 not by RIII, i.e., the growth inhibition was significantly more severe. The parallel overexpression of rI together with rIII ceased also an effect of an efficient prolongation of LIN profile obtained when rI was co-expressed with rI.1 (Fig. 4d, e) and weak prolongation of LIN profiles obtained during parallel overexpression of rI and rI.-1 or overexpression of the whole rI operon. Even though rIII overexpression ceased the above mentioned effects, it seems to be likely that action of accessory proteins—RI.1 and RI.-1 require functional RIII protein, as no effect of the overproduction (beside the toxicity of RI.1 to host cells) was observed when T4rIII was used and RIII was not supplied in trans. This raises the question about a possible role for stoichiometric proportions of both proteins, RI and RIII.

Fig. 4
figure 4

A putative model for functions of R.I.1 and RI.-1 in LIN. Arrows indicate stimulation and blunt-ended lines indicate inhibition. The thicknesses of arrows and lines are proportional to efficiencies of corresponding processes. a In wild-type cells RI inhibits T-mediated lysis of host cells, when LIN signal is present. RIII stimulates RI function. RI.1 and RI.-1 modulate RIII action. b Overproduction of RI causes less efficient LIN possibly by titrating off the RIII protein. c Overproduction of RIII increases efficiency of LIN. d Overproduction of RIII together with RI.1 or RI.-1 strongly increases efficiency of LIN. e Overproduction of RI.1 diminishes the effect of increased levels of RI. f When RI, RI.1, and RIII are overproduced, the LIN efficiency is decreased, similarly to the effect of overproduction of RI alone. This may suggest that RI:RIII:RI.1 ratio is important for LIN regulation

Full size image

On the basis of our results we propose a model for the role of accessory genes rI.1 and rI.-1 in LIN maintenance (Fig. 4). According to this model, in wild-type cells, RI inhibits T-mediated lysis of host cells when LIN signal is present, and RIII stimulates RI function, which has already been known. However, we propose that RI.1 and RI.-1 can modulate RIII action. This proposal is supported by observations that (i) overproduction of RI causes less efficient LIN, possibly by titrating off the RIII protein, (ii) overproduction of RIII increases efficiency of LIN, (iii) overproduction of RIII together with RI.1 or RI.-1 strongly increases efficiency of LIN, (iv) overproduction of RI.1 diminishes the effect of increased levels of RI, and (v) when RI, RI.1, and RIII are overproduced, the LIN efficiency is decreased, similarly to the effect of overproduction of RI alone. Therefore, we hypothesize that the RI:RIII:RI.1 ratio is important for LIN regulation.

Prolongation of LIN, observed in this work, has also another aspect, which may be helpful to understand the lysis process and the LIN collapse caused by bacteriophage T4. Since the presence of an excess of some phage proteins made bacterial culture less prone to LIN collapse; it is possible that the causative agent of LIN collapse is not only the accumulation of damages done by superinfecting phages, but also optimal amounts and activities of some phage proteins.