YeeU enhances the bundling of cytoskeletal polymers of MreB and FtsZ, antagonizing the CbtA (YeeV) toxicity in Escherichia coli

Authors

  • Hisako Masuda,

    1. Department of Biochemistry, Center for Advanced Biotechnology and Medicine, Robert Wood Johnson Medical School, 679 Hoes lane, Piscataway, NJ, USA
    Search for more papers by this author
  • Qian Tan,

    1. Department of Biochemistry, Center for Advanced Biotechnology and Medicine, Robert Wood Johnson Medical School, 679 Hoes lane, Piscataway, NJ, USA
    Search for more papers by this author
  • Naoki Awano,

    1. Department of Microbiology, Tokyo Medical University, 6-1-1 Shinjuku, Shinjuku-ku, Tokyo 160-8402, Japan
    Search for more papers by this author
  • Kuen-Phon Wu,

    1. Department of Biochemistry, Center for Advanced Biotechnology and Medicine, Robert Wood Johnson Medical School, 679 Hoes lane, Piscataway, NJ, USA
    Search for more papers by this author
  • Masayori Inouye

    Corresponding author
    1. Department of Biochemistry, Center for Advanced Biotechnology and Medicine, Robert Wood Johnson Medical School, 679 Hoes lane, Piscataway, NJ, USA
    Search for more papers by this author

Summary

All free-living bacteria carry the toxin–antitoxin (TA) systems controlling cell growth and death under stress conditions. YeeU–YeeV (CbtA) is one of the Escherichia coli TA systems, and the toxin, CbtA, has been reported to inhibit the polymerization of bacterial cytoskeletal proteins, MreB and FtsZ. Here, we demonstrate that the antitoxin, YeeU, is a novel type of antitoxin (type IV TA system), which does not form a complex with CbtA but functions as an antagonist for CbtA toxicity. Specifically, YeeU was found to directly interact with MreB and FtsZ, and enhance the bundling of their filamentous polymers in vitro. Surprisingly, YeeU neutralized not only the toxicity of CbtA but also the toxicity caused by other inhibitors of MreB and FtsZ, such as A22, SulA and MinC, indicating that YeeU-induced bundling of MreB and FtsZ has an intrinsic global stabilizing effect on their homeostasis. Here we propose to rename YeeU as CbeA for cytoskeleton bundling-enhancing factor A.

Introduction

The toxin–antitoxin (TA) systems are found in all free-living bacteria, and encoded from a small operon on bacterial genomes (Gerdes et al., 2005). In general, two small proteins are produced from these TA systems, a toxin and its cognate antitoxin. Toxins target various cellular functions such as DNA replication, mRNA stability, protein synthesis and cell division (Jiang et al., 2002; Zhang et al., 2003; Tan et al., 2011; Zhang and Inouye, 2011). The expression of the lethal toxin genes in these TA systems is tightly regulated at both transcriptional and post-transcriptional levels. In all known cases of well-studied type II TA systems (Yamaguchi et al., 2011), in normally growing cells, toxins form a stable complex with their cognate antitoxins which block toxin activity. Antitoxins also function as a repressor for individual TA operons (Gerdes et al., 2005). In an adverse environment, because antitoxins are intrinsically unstable, they are preferentially digested by stress-inducible proteases. As a result, toxins are freely released into the cells, causing reversible growth arrest. Bacteria stay in this ‘quasi-dormancy’ state to survive and await suitable conditions to resume their growth (Pedersen et al., 2002). This state is implicated in the formation of the drug-resistant persister phenotype (Lewis, 2008). The TA systems have also shown to be associated with pathogenicity, programmed cell death and biofilm formation (Pandey and Gerdes, 2005; Nariya and Inouye, 2008; Wang and Wood, 2011).

The cytoskeletons are an essential part of cellular physiology and morphology both in prokaryotes and eukaryotes. Escherichia coli strains carry two major cytoskeleton proteins, MreB and FtsZ, which are the homologues of eukaryotic actin and tubulin respectively (van den Ent et al., 2001). MreB and FtsZ control a number of essential cellular functions. MreB is responsible for maintaining the cell shape and cell polarity in rod-shaped bacteria (Wachi et al., 1987). MreB is also involved in chromosome segregation, localization of membranous organelles and co-ordinating cell division with cell biosynthesis (Kruse et al., 2005; Komeili et al., 2006; Madabhushi and Marians, 2009; Domínguez-Escobar et al., 2011; Garner et al., 2011). FtsZ forms a ring structure at the mid-cell and functions as a scaffold during the assembly of divisomes, which is a multi-protein complex and essential for cell division (Adams and Errington, 2009).

The common properties of bacterial cytoskeletons are filamentous structures and their dynamic nature (Stricker et al., 2002; Shih and Rothfield, 2006). During different stages of the cell cycle, they assemble from monomers to polymerized filamentous structures as well as disassemble from polymers to monomers (Shih et al., 2003; Lutkenhaus, 2007). Because of their involvement in a variety of cellular physiology, maintaining proper homeostasis of cytoskeletons is assumed to be crucial for bacteria's viability (Shih and Rothfield, 2006). In well-characterized eukaryotic systems, more than 100 actin-binding proteins have been identified to modulate assembly and disassembly of actin (dos Remedios et al., 2003). Actin-binding proteins allow temporal and spacial regulation of actin dynamics in a precise manner. Hence, it has been proposed that the assembly and disassembly of bacterial cytoskeletons are also controlled by effector proteins, allowing them to quickly respond to the changes in the environment. However, the mechanisms of how these dynamics are regulated in bacteria have remained largely unknown, particularly for MreB.

The YeeU–CbtA system is one of the 35 E. coli TA systems (Brown and Shaw, 2003; Tan et al., 2011; Yamaguchi and Inouye, 2011). Unlike most of the known toxins targeting the macromolecular biosynthesis, CbtA is the first toxin in the TA systems, which was found to affect cellular morphology and division (Tan et al., 2011). CbtA binds and inhibits the polymerization of MreB and FtsZ. In the present paper, we demonstrate that antitoxin, YeeU, suppresses the CbtA toxicity by stabilizing the CbtA target proteins rather than by directly interacting with CbtA to suppress its toxicity. Specifically, YeeU directly binds to both MreB and FtsZ and enhances the bundling of their filaments in vitro. Notably, this is a unique feature of the YeeU–CbtA system, distinguishing it from all the other TA systems, that toxin CbtA and antitoxin YeeU does not form a complex. Nevertheless, YeeU is able to neutralize CbtA toxicity. Thus, we propose that the YeeU–CbtA constitutes a new type of TA system, and classified as type IV TA system. Intriguingly, in addition to the CbtA toxicity, YeeU is also able to neutralize the toxicity caused by other MreB and FtsZ inhibitors, such as A22 [S-(3, 4-dichlorobenzyl)isothiourea] for MreB (Iwai et al., 2002), and SulA and DicB for FtsZ (Huisman et al., 1984; de Boer et al., 1990; Pichoff and Lutkenhaus, 2001). These data suggest that YeeU-induced bundling of MreB and FtsZ has an intrinsic global stabilizing effect on their proper homeostasis. This is also the first report to identify a protein (YeeU) which is able to interact with MreB and promotes its assembly. These results provide new insights into the roles of bacterial cytoskeletons in bacterial physiology in conjunction with the TA systems. On the basis of these results, we propose to rename YeeU as CbeA for cytoskeleton bundling-enhancing factor A.

Results

YeeU neutralizes the toxicity and morphological changes by CbtA

Recently, we demonstrated that CbtA blocks the polymerization of MreB and FtsZ, resulting in the growth inhibition and the alteration of cellular morphology (Tan et al., 2011). YeeU, encoded from a gene located immediately upstream of CbtA, was shown to neutralize the growth defect caused by CbtA when two proteins are coexpressed (Brown and Shaw, 2003). In order to precisely elucidate how YeeU functions to neutralize the CbtA toxicity, we first constructed two independent inducible systems for YeeU and CbtA. As seen in Fig. 1A, induction of YeeU expression from the pET28a–yeeU by 0.1 mM IPTG suppressed the growth inhibition caused by CbtA, whose expression was induced from pBAD33–cbtA by 0.1% arabinose. Since the growth defect caused by CbtA could still be observed by the addition of 0.01% arabinose in liquid medium (Fig. 1B), the changes in cellular morphology were examined in the presence of 0.01% arabinose with and without YeeU overexpression (Fig. 1C). At 1.5 h after induction of CbtA expression without co-induction of YeeU, the cell shape started to change from rod to lemon shape. At 5 h, no more rod cells were observed and all the cells became lemon-shaped. In contrast, when YeeU was coexpressed, the cells retained its rod shape for more than 10 h. Moreover, the growth rate was resumed to the rate identical to the control cells carrying empty plasmids. These observations demonstrate that YeeU is able to suppress both the growth defect and the morphological changes caused by ectopic expression of CbtA. There is an 89 base intergenic region between the yeeU gene and the cbtA gene, which has been previously suggested to have some roles in the YeeU and CbtA expression (Brown and Shaw, 2003). In the present work, only the protein-coding sequences of YeeU and CbtA were cloned, and showed that this intergenic region was not required for antitoxin effects of YeeU. Yet, one cannot exclude the possibility that this intergenic region may have additional roles in the expression of YeeU and/or CbtA.

Figure 1.

YeeU neutralizes the growth and morphological defect caused by CbtA. The growth of E. coli BL21 (DE3) carrying pBAD33–cbtA and pET28a–yeeU or pBAD33–cbtA and pET28a. A. The growth on M9 agar plates containing 0.1% arabinose and/or 0.1 mM IPTG was analysed. B. The growth rate was analysed using M9 medium supplemented with casamino acids. IPTG (0.1 mM) was added when OD600 reached 0.3. After 15 min, 0.01% arabinose was added. C. The cellular morphology was observed under a phase-contrast microscope.

YeeU neutralizes the growth defect caused by A22 and overexpression of SulA or DicB

In a variety of conditions tested, YeeU did not form a complex with CbtA (data not shown), yet YeeU neutralized the CbtA toxicity. Also YeeU did not require 5′ UTR of CbtA for neutralization. These data suggest that YeeU does not physically sequester CbtA from targets nor function as a repressor of CbtA transcription. In order to identify the mode of action of YeeU, we next examined whether YeeU specifically antagonizes the CbtA function or neutralizes the morphological changes caused by other known inhibitors of MreB and FtsZ. Inhibition of MreB polymerization by A22 was shown to convert rod-shaped cells to small and round-shape cells, and subsequently causing cell death (Iwai et al., 2002). When cells were treated with a sublethal concentration of A22 at 4 and 20 µg ml−1 for 16 h, the cells became shorter and round (Fig. 2A). In contrast, when YeeU expression was induced from pET28a–yeeU prior to the addition of A22, cells retained the normal rod shape. SulA is known to inhibit the polymerization of FtsZ, thus blocking the septum formation, leading to the inhibition of cell division to form filamentous cells (Huisman et al., 1984). As shown in Fig. 2B, when SulA was expressed from pBAD33 in the presence of 0.1% arabinose, within 1.5 h, the cell length increased by approximately 2.5 times on average. DicB also inhibits cell division by activating a FtsZ inhibitor, MinC (Labie et al., 1990). Overexpression of DicB disrupts spatially restricted inhibition of FtsZ by MinC, resulted in the inhibition of FtsZ polymerization throughout the cell (Johnson et al., 2002). As also shown in Fig. 2B, cells were elongated by the overexpression of DicB. The observed cellular elongation by SulA and DicB overexpression was also neutralized by the coexpression of YeeU (Fig. 2B). The results shown above suggest that YeeU is a global factor with capability to rescue the toxic effects of different cytoskeletal protein inhibitors.

Figure 2.

YeeU reverses the morphological defect caused by A22 and overexpression of SulA and DicB. A. E. coli BL21 (DE3) was transformed with pET28a–yeeU or with empty pET28a. The cells were grown in LB medium. When OD600 reached 0.4, 0.1 mM IPTG was added to induce YeeU expression. After 30 min, A22 was added at various concentrations as shown. The cells were grown at 37°C for 12 h and the morphology was examined. B. E. coli BL21 (DE3) was cotransformed with pBAD33–sulA and pET28a–yeeU or pBAD24–dicB and pACYC–yeeU. When OD600 reached 0.3, 0.1 mM IPTG was added to induce the expression of YeeU. After 15 min, the expression of SulA or DicB was induced by 0.1% or 0.02% arabinose respectively. The cellular morphology after 3 h induction was observed under a phase-contrast microscope.

YeeU directly interacts with MreB and FtsZ

Next, we examined if YeeU directly interacts with MreB and FtsZ. First, the interaction was examined by pull-down experiments (Fig. 3A). A cell lysate containing FLAG-tagged MreB (MreB-FLAG) was incubated with Ni-NTA agarose following pre-incubation with a control lysate or a lysate containing His6-tagged YeeU (YeeU-HIS). The presence of MreB-FLAG in the elution fraction was detected by Western blot using the monoclonal antibody against FLAG tag, conjugated with horseradish peroxidase. MreB-FLAG was detected in the elution fraction only when it was pre-incubated with cell lysates that contained YeeU-HIS, indicating that YeeU-HIS interacted with MreB-FLAG (Fig. 3A, lanes 1 and 2). The same experiment was performed with FtsZ using FLAG-tagged FtsZ (FtsZ-FLAG). FtsZ-FLAG was only detected in the elution fraction when it was pre-incubated with YeeU-HIS, showing that FtsZ-FLAG also interacted with YeeU-HIS (Fig. 3A, lanes 3 and 4).

Figure 3.

Analysis of the interactions between YeeU and MreB or FtsZ by pull-down experiments and sedimentation assay. A. The cell lysate containing MreB-FLAG or FtsZ-FLAG was incubated with the lysate containing YeeU-HIS (+) or with that of the wild-type cells (−) before adding Ni-NTA (Qiagen, CA, USA). Eluted fractions were separated by SDS-PAGE and FLAG-tagged proteins were detected using Western blot with anti-FLAG antibody conjugated with horseradish peroxidase (Invitrogen, CA, USA). The input for the pull-down assay was also visualized by Coomassie Blue staining. B. Purification of MreB-HIS and YeeU-HIS. Proteins were purified using Ni-NTA and Superdex200 size exclusion columns. C. (a) In vitro polymerization of MreB and FtsZ in the presence of YeeU. ATP-dependent MreB polymerization was analysed as described in Experimental procedures. MreB (2 µM) was polymerized with 2 mM ATP and polymers were collected by ultracentrifugation. YeeU (2 µM) was added to the reaction mixture prior to the addition of ATP. The non-polymerized MreB in the supernatant fraction (S) and the pellet fraction consisted of the polymers (P) were separated by SDS-PAGE and visualized by Coomassie Blue staining. GTP-dependent FtsZ polymerization was analysed in a similar manner. FtsZ (2 µM) was polymerized with 2 mM GTP and polymerized FtsZ was then collected by ultracentrifugation. YeeU (2 µM) was added to the reaction mixture prior to the addition of GTP. (b) Sedimentation of YeeU in the absence of MreB or FtsZ. (c) The interaction of YeeU and polymerized MreB was analysed by examining the cosedimentation of YeeU with polymerized MreB. MreB was first polymerized in the absence of YeeU for 20 min. YeeU was then added, and the mixture was further incubated for 20 min. The pellet thus formed was collected by ultracentrifugation. The interaction of YeeU and polymerized FtsZ was analysed in the same manner as described for MreB.

The interaction between YeeU and MreB was also verified by sedimentation assay using purified proteins (Fig. 3B and C). We purified both MreB and FtsZ by affinity chromatography and gel filtration. As MreB from Thermatoga maritima or Thermoplasma acidophilum (van den Ent et al., 2001; Roeben et al., 2006), purified E. coli MreB retained its function as it formed a polymer in an ATP-dependent manner (Fig. 3C). When MreB was polymerized in the presence of YeeU at equimolar concentrations, approximately 50% of input YeeU was detected in the pellet fraction along with the polymerized MreB, which was collected by ultracentrifuge (Fig. 3C(a)). As a control, when YeeU was incubated in the same buffer without MreB and subjected to ultracentrifugation, YeeU was not detected in the pellet fraction (Fig. 3C(b)), indicating that YeeU itself did not form polymerized structures and that YeeU found in the pellet fraction was due to cosedimentation with polymerized MreB. In order to examine whether YeeU binds only to the monomeric form of MreB, MreB was first polymerized, and then YeeU was added. A significant amount of YeeU was found in the pellet fraction (Fig. 3C(c)), indicating that YeeU can bind to polymerized MreB filaments as well as its monomeric form. At present it is not known if there is a difference in the affinity between the YeeU and two forms of MreB. The polymerization of FtsZ, with and without addition of YeeU, was also examined in a similar manner. FtsZ becomes polymerized in the GTP-dependent manner (Mukherjee et al., 1998). YeeU was detected in the pellet fraction only when it was pre-incubated with FtsZ proteins (Fig. 3C(a) and (b)). The interaction of YeeU with preformed FtsZ polymers appeared to be also weak, as a small amount of YeeU was detected in the pellet fraction (Fig. 3C(c)).

YeeU neutralized the inhibition of MreB and FtsZ polymerization by CbtA in vitro

Previously, we have shown that the CbtA inhibited the in vitro polymerization of MreB and FtsZ in a concentration-dependent manner (Tan et al., 2011). Thus, we then examined the effects of YeeU on the inhibitory activity of CbtA in MreB and FtsZ polymerization in vitro. In order to unambiguously detect the differences in the net polymerization levels, the polymerization reaction was carried out for 5 min at 18°C, rather than 20 min at room temperature, as performed previously (Tan et al., 2011). Following the separation of polymerized and non-polymerized proteins by ultracentrifuge, the amount in each fraction was quantified by comparing the intensity of bands on SDS-PAGE gel. In this condition, 72% of MreB was polymerized in the absence of CbtA or YeeU (Fig. 4). When CbtA was added at 1:1 molar ratio, MreB polymerization was inhibited to 41%. When MreB, CbtA and YeeU were added at 1:1:1 molar ratio, the degree of polymerization increased to 63%. When twice as much YeeU was added, a degree of polymerization further increased (85%), even higher than MreB only (72%). The same effect was also seen with FtsZ (Fig. 4). Without the addition of CbtA or YeeU, 77% of FtsZ was polymerized. The addition of CbtA reduced the FtsZ polymerization to 19%. The addition of an equal molar concentration or twice as much YeeU increased the degree of polymerization to 54% and 96% respectively. The enhanced polymerization in the presence of YeeU was also observed in this experiment, as excess YeeU allowed near complete polymerization (96%), while in the absence only 77% of FtsZ was polymerized.

Figure 4.

YeeU counteracts the inhibitory effects of CbtA on in vitro polymerization of MreB and FtsZ. The polymerization of MreB and FtsZ was performed as described in Fig. 3. The designated amounts of CbtA and YeeU were added to the reaction mixture prior to the addition of ATP or GTP. The intensity of each band on SDS-PAGE was quantified using ImageJ, and the ratio of polymerized and non-polymerized cytoskeletal proteins in each sample was plotted.

Effects of YeeU on the mode of MreB and FtsZ polymerization in vitro

As shown in Fig. 4, the addition of YeeU appeared to have a positive effect on MreB and FtsZ polymerization. As a complementary approach, effects of YeeU on MreB and FtsZ polymerization were observed by light-scattering experiments. It has been shown previously that MreB polymers from T. maritima scattered light with an approximately 50-fold greater intensity than soluble monomers (Bean and Amann, 2008). The authors also reported that within a 1–12 µM concentration range, the steady-state light-scattering intensity was directly proportional to the protein concentration. As shown in Fig. 5A, E. coli MreB also exhibited a concentration-dependent increase in scattered-light intensity within the same concentration range as that of Thermatoga MreB. The kinetics of FtsZ polymerization (Fig. 5B) were also measured following the method established previously (Mukherjee and Lutkenhaus, 1999; Popp et al., 2010). Next, using the same conditions, the effect of YeeU on MreB and FtsZ polymerization was examined. While the concentration of MreB and FtsZ was held constant (2 µM), increasing concentrations of YeeU were added to the reaction mixture prior to the addition of ATP or GTP. The almost linear increase of the maximum scattered light was observed with increasing amounts of YeeU when YeeU was added at 0.5, 1 and 2 molar concentrations of cytoskeletal proteins (Fig. 5C and D). YeeU did not make independent contribution to the light scattering as YeeU protein alone yielded no scattered-light intensity above background (data not shown). Our preliminary study using sedimentation assay showed that YeeU does not lower the critical concentration for polymerization (data not shown). These results demonstrated that increased scattered-light intensity by the addition of YeeU is not due to the change in critical concentrations for polymerization, but solely due to the change in the mode of MreB and FtsZ polymerization.

Figure 5.

The effects of YeeU on MreB and FtsZ polymerization dynamics. Light-scattering assays of MreB and FtsZ polymerization were performed with a fluorimeter (Photon Technology International, NJ, USA) with the excitation wavelength at 350 nm and the emission wavelength at 365 nm. Polymerization conditions are described in Experimental procedures. A. Concentration-dependent polymerization of E. coli MreB: 2 µM (square), 4 µM (+), 8 µM (triangle) and 12 µM (circle). B. Concentration-dependent polymerization of E. coli FtsZ: 2 µM (square), 4 µM (+) and 8 µM (triangle). C. The concentration of MreB was kept constant at 2 µM. The concentrations of YeeU added were: 0 µM (square), 2 µM (+) and 4 µM (circle). D. The concentration of FtsZ was kept constant at 2 µM. The concentrations of YeeU were: 0 µM (square), 2 µM (+) and 4 µM (triangle). The values were normalized to the reaction with 2 µM of MreB or FtsZ.

YeeU promotes bundling of FtsZ and MreB protofilaments in vitro

The nature of MreB and FtsZ polymers in the presence of YeeU was examined by transmission electron microscopy (TEM) (Figs 6 and S1). The polymerized samples were applied to a carbon-coated copper grid, and negatively stained. In the absence of YeeU, FtsZ polymers appeared to exist as a single protofilament, as reported previously (Mukherjee et al., 1998). When FtsZ was polymerized in the presence of YeeU, all the observed protofilaments appeared to be bundled (Fig. 6). The bundled conformation, on average, contained 10–20 protofilaments, which is consistent with the proposed number of filaments in higher-order structures of FtsZ in vivo (Fu et al., 2010).

Figure 6.

Electron microscopy of MreB and FtsZ polymers in the presence or absence of YeeU. MreB and FtsZ polymerization in the presence or in the absence of YeeU was analysed by transmission electron microscope. YeeU was added at 1:1 molar ratio to MreB or FtsZ. Scale bar: 100 nm.

MreB polymers in the absence of YeeU showed two different conformations. The first type was a straight single filament having approximately a width of 2–4 nm and a length of 200–300 nm (Fig. 6, arrow A). This size is consistent with the size of the MreB isoform from Thermatoga (van den Ent et al., 2001), and also consistent with the size of MreB polymers in vivo (Carballido-López and Errington, 2003). Another conformation consisted of discontinuous strings of shorter and thicker units (each unit of approximately 30 nm in length and 10 nm in width) (Fig. 6, arrow B). Multiple short units seemed to be connected to each other in a repeated pattern. In order to rule out the possibility that the observed structure was due to impurities in the preparation, the purified MreB was further analysed by SDS-PAGE and mass spectroscopy, confirming that no detectable contaminating proteins were present in the preparation (Fig. 3B). The addition of YeeU enhanced the bundle formation between the straight single filaments, diminishing the single protofilament in the sample to form superstructures containing, on average, 15 to 20 filaments. YeeU did not have detectable effects on the later types of MreB filaments. The electron microscopic analysis indicated that YeeU promotes the formation of higher-order structure of the MreB and FtsZ filaments in vitro.

Discussion

YeeU is a general stabilizing factor for bacterial cytoskeletal proteins that counteracts CbtA and other cytoskeleton inhibitors

In this paper, we revealed two intriguing aspects of YeeU. First, we demonstrated that YeeU is a novel factor that promotes the bundling of cytoskeletal proteins, FtsZ and MreB. YeeU is able to suppress not only the CbtA toxicity but also the effects of the other known cell division inhibitors such as A22, an antibiotic that inhibits MreB function, and SulA and DicB, both of which are inhibitors for FtsZ assembly. Second, for the first time, an antitoxin in the TA systems was found to neutralize the function of its cognate toxin, not by forming a stable TA complex but by antagonizing the toxin function directly on the cellular target. This novel type of TA systems may be classified as type IV TA system in contrast to type II TA system in which protein antitoxins directly bind to toxins and inhibit the toxin function (Fig. 7).

Figure 7.

Type II and type IV TA systems. A. The model for the type II TA systems. Toxins and antitoxins, encoded from the single operon, form a stable complex (TA complex). The expression of the operon is negatively regulated by the TA complex. Stress-inducible proteases preferentially degrade antitoxins, allowing toxins to exert their toxicity. B. The novel antitoxin mechanism exerted by YeeU (type IV TA system). YeeU and CbtA do not form a complex. YeeU counteracts the CbtA toxicity by promoting the reaction (assembly of FtsZ and MreB filaments), which is inhibited by the CbtA.

Polymerization dynamics of cytoskeletal proteins

Prior to cell division, FtsZ filaments assemble to form a Z-ring and function as a scaffold for a number of cell division proteins (Shih and Rothfield, 2006). In the cell, FtsZ appears to exist at least in three different forms: monomers, polymers and higher-order structures (Chen and Erickson, 2005; Dajkovic et al., 2008a; Monahan et al., 2009). Since the concentration of FtsZ does not change throughout the cell cycle, the formation and disassembly of the Z-ring is regulated not by modulating the amount of FtsZ, but by modulating the rates of conversion between each of these states. Three inhibitors of the FtsZ assembly affect two distinct steps of the assembly pathway. SulA binds to the FtsZ monomers, transforming FtsZ to a non-polymerizable state, hence inhibiting the polymerization (Gueiros-Filho and Losick, 2002). CbtA also inhibits the polymerization of FtsZ (Tan et al., 2011). On the other hand, MinC inhibits the cross-linking between the protofilaments (Dajkovic et al., 2008b; Scheffers, 2008). The neutralization of the inhibitory function of SulA and MinC can be achieved by an essential division protein, ZapA (Gueiros-Filho and Losick, 2002; Scheffers, 2008). ZapA induces the formation of higher-order structures by mediating the bundling and cross-linking of FtsZ, hence lowering the critical concentration for assembly reactions (Dajkovic et al., 2010).

Proposed model for YeeU functions

In this report, we have shown that YeeU also counteracted the inhibitory effects of polymerization inhibitors, SulA and CbtA, as well as a bundling inhibitor, MinC, in vivo (Figs 1 and 2). In addition, YeeU reversed the morphological change caused by MreB inhibitor, A22. Furthermore, YeeU promoted the formation of the higher-order structures of MreB and FtsZ filaments in vitro (Figs 5 and 6). We speculate that YeeU promotes the assembly reactions of MreB and FtsZ by enhancing the bundling of MreB and FtsZ polymers. We observed approximately an equal molar amount of YeeU was cosedimented with MreB or FtsZ polymers. We also observed a linear increase of scattered light by the addition of YeeU at 0.5, 1 and 2 molar ratios relative to MreB and FtsZ, suggesting that YeeU is retained in the MreB and FtsZ filaments as a structural component rather than functioning as a catalyst for their filamentation or bundling. Moreover, our preliminary study showed that YeeU does not lower the critical concentration for polymerization. Thus, we speculate that YeeU enhances the bundling of MreB and FtsZ polymers by binding to the lateral side of the cytoskeletal filaments as seen in ZapA (Gueiros-Filho and Losick, 2002). Since YeeU forms a dimeric complex (Arbing et al., 2010), a dimer of YeeU may bring two filaments together. Further studies to decipher the exact MreB- or FtsZ-binding site for YeeU may reveal the molecular basis for its mode of action and provide a clue for how it neutralizes the toxicity of CbtA and other MreB and FtsZ inhibitors.

Concluding remarks

An interesting question is how YeeU binds both MreB and FtsZ, which do not have obvious sequence and structural similarities. Notably, the toxin CbtA also binds to both FtsZ and MreB; however, it causes the opposite effect of YeeU on cytoskeleton assembly and consequently inhibiting their biological role (Tan et al., 2011). We have recently identified another E. coli TA system (YgfYX), unrelated to yeeU–cbtA, whose toxin also inhibits both MreB and FtsZ polymerization (Masuda et al., 2012), implying that the capability to interact with both types of cytoskeleton proteins is a common properties of these E. coli TA proteins.

The observation of ZapA-induced bundling of FtsZ, and the inhibition of FtsZ polymerization by SulA and MinC, together with our present data, suggests that a number of factors engage in both inhibitory and stabilizing effects on cytoskeletal proteins in bacteria, allowing cells to respond rapidly and precisely to environmental changes. Since cytoskeletal proteins are involved in the regulation of a number of essential cellular functions, direct manipulation of the dynamics of cytoskeletons may allow bacteria to quickly change their physiology under unfavourable conditions. In accordance with this idea, cell division is inhibited by SulA upon DNA damage (Huisman et al., 1984). As a TA system, yeeU–cbtA would also play a physiological role in survival under certain stress conditions. In such conditions, YeeU is preferentially degraded, causing CbtA to inhibit cell division. Once the stress is removed, the YeeU is once again accumulated and the cells resume cell division along with other essential cellular processes that are regulated by MreB and FtsZ.

The yeeU–cbtA operon is a part of a prophage sequence (CP4-44) and at least two additional homologues of yeeU–cbtA systems have been identified in a single genome of E. coli (Brown and Shaw, 2003). A comparative genomic study also indicated that the yeeU–cbtA operon is under positive selection during the evolution of the E. coli genome (Petersen et al., 2007). These data strongly suggest that the yeeU–cbtA operon and their homologues somehow give the host a selective advantage as observed in other prophage genes (Lemire et al., 2011). Microarray analysis of the E. coli transcriptome showed differential expression of the yeeU–cbtA operon during the recovery from the stationary phase or following UV exposures (Courcelle et al., 2001; Sangurdekar et al., 2006). The roles of the yeeU–cbtA system in maintaining fitness under these conditions using deletion strains are under current investigation. The further elucidation of the molecular mechanisms as to how YeeU interacts with FtsZ and MreB, as well as the condition that regulates its expression, will provide important insights into the regulation of polymerization of the cytoskeleton proteins and thereby the regulation of cell division.

Experimental procedures

Strains, plasmids and growth conditions

Escherichia coli BW25113 (ΔaraBD) (Datsenko and Wanner, 2000) and BL21 (DE3) were grown in M9 medium supplemented with 0.2% casamino acids and 0.5% glycerol at 37°C. pBAD33–sulA, pET28–yeeU, pBAD33–cbtA, pET28–cbtA, and plasmids expressing His6-tagged FtsZ and MreB were constructed previously (Tan et al., 2011). FLAG-tagged FtsZ and MreB were constructed in pET17b, having a tag at the C-terminal end. For examining the growth pattern, a designated amount of arabinose and/or IPTG was added to the culture when the OD600 reached 0.3. Cell morphology was observed using an Olympus BX40 microscope.

Protein expression, purification and pull-down assay

His6-tagged CbtA, FtsZ, MreB and YeeU was purified as described before (Arbing et al., 2010; Tan et al., 2011). His6 tag pull-down assays were carried out by incubating the cell lysate containing YeeU-HIS and the cell lysate containing FtsZ-FLAG or MreB-FLAG (lysis buffer: 50 mM HEPES-KOH, pH 7.5, 10 mM MgCl2, 200 mM KCl, 0.1 mM EDTA and 10% glycerol) overnight at 4°C. Ni-NTA agarose was added to the lysate and the mixture was incubated at room temperature for 1 h. The beads were washed three times with 20 ml of the same lysis buffer containing 20 mM imidazol. Protein complexes were then analysed with 12% SDS-PAGE and Western blot using monoclonal anti-flag antibody conjugated with horseradish peroxidase (Sigma-Aldrich, MO, USA).

Polymerization assays

The effects of CbtA and/or YeeU on FtsZ and MreB polymerization were determined by a sedimentation method as described previously (Mukherjee et al., 1998; Anand et al., 2004; Bean et al., 2009; Tan et al., 2011). Purified FtsZ-HIS or MreB-HIS were mixed with different amounts of purified CbtA-HIS and/or YeeU-HIS in polymerization buffer (5 mM MgCl2, 100 mM NaCl and 100 mM Tris-HCl, pH 7.0). The mixture was pre-incubated at room temperature for 30 min before 2 mM GTP or ATP was added to initiate polymerization. For the sedimentation assay, the polymerization reaction was carried out at room temperature. FtsZ or MreB polymers were precipitated by centrifugation at 100 000 g for 20 min, and the pellets were suspended in 50 µl of polymerization buffer. Both the supernatant and pellet fractions were analysed on a 17% SDS-PAGE, followed by Coomassie Blue staining.

The morphology of the polymers was analysed by a JEOL 1200 electron microscope. FtsZ polymers were negatively stained by 1% uranyl acetate. MreB polymers were stained by 1% phosphotungstic acid on carbon-coated copper grids. Right-angled light scattering was measured at a constant 25°C, using a fluorimeter (Photon Technology International, NJ, USA). MreB or FtsZ was added to a cuvette along with the designated concentration of YeeU. In the case of FtsZ, following the 5 min equilibration period, 2 mM GTP was added and the mixture was gently mixed to initiate the polymerization. For MreB, all the solutions were kept on ice until 2 mM ATP was added (Bean and Amann, 2008). The first reading was taken immediately after the cuvette was returned to the chamber.

Acknowledgements

The authors are grateful to Dr Peter Tupa and Mr Chun-Yi Lin for critical reading of the manuscript, Dr Donald Winkelmann for his help on light-scattering experiment and Mr Raj Patel for EM analysis. We are also grateful to Dr Alla Kostyukova for her helpful discussion. This work was supported by grants from the National Institute of Health (RO1GM081567 and R01GM081567-02S1) to M. I. H. M. was supported by NIH postdoctoral fellowship F32 GM095200. Authors declare no conflict of interest.

Ancillary