Mutational analysis of MukE reveals its role in focal subcellular localization of MukBEF


For correspondence. E-mail; Tel. (+1) 405 325 1677; Fax (+1) 405 325 6111.


Bacterial condensin MukBEF is essential for global folding of the Escherichia coli chromosome. MukB, a SMC (structural maintenance of chromosome) protein, comprises the core of this complex and is responsible for its ATP-modulated DNA binding and reshaping activities. MukF serves as a kleisin that modulates MukB–DNA interactions and links MukBs into macromolecular assemblies. Little is known about the function of MukE. Using random mutagenesis, we generated six loss-of-function point mutations in MukE. The surface mutations clustered in two places. One of them was at or close to the interface with MukF while the other was away from the known interactions of the protein. All loss-of-function mutations affected focal localization of MukBEF in live cells. In vitro, however, only some of them interfered with the assembly of MukBEF into a complex or the ability of MukEF to disrupt MukB–DNA interactions. Moreover, some MukE mutants were able to join intracellular foci formed by endogenous MukBEF and most of the mutants were efficiently incorporated into MukBEF even in the presence of endogenous MukE. These data reveal that focal localization of MukBEF involves other activities besides DNA binding and that MukE plays a central role in them.


Correct folding of the Escherichia coli chromosome requires the activity of condensin MukBEF (Niki et al., 1991; Petrushenko et al., 2006a; Danilova et al., 2007). The core of this complex, MukB, belongs to the characteristically V-shaped family of SMC (structural maintenance of chromosome) proteins (Melby et al., 1998; Li et al., 2010), which play diverse roles in global chromosome organization in organisms from all kingdoms of life (Cobbe and Heck, 2004; Graumann and Knust, 2009; Rybenkov, 2009; Gruber, 2011). By bringing distant DNA segments together (Strick et al., 2004; Cui et al., 2008; Petrushenko et al., 2010), MukBEF is thought to establish the giant-loop architecture of the E. coli chromosome (Saitoh et al., 1995; She et al., 2007). Besides MukBEF, two other condensins were found in bacteria, the SMC–ScpAB (Mascarenhas et al., 2002; Soppa et al., 2002) and MksBEF (Petrushenko et al., 2011), which sometimes coexist in a single species. Despite significant sequence divergence, all three condensins share the key structural and mechanistic features and contribute to faithful chromosome packing and segregation.

The mechanism of MukBEF remains unclear. Biochemical dissection of the protein revealed that MukB and MukEF form two distinct complexes, MukB2 and Muk(E2F)2, which dynamically interact with each other producing the holoenzyme MukB4–Muk(E2F)2 (Petrushenko et al., 2006a; Badrinarayanan et al., 2012). All reconstituted DNA binding and reshaping activities of the complex reside in its SMC subunit, whereas MukEF appears to play a regulatory role (Petrushenko et al., 2006b; Cui et al., 2008). MukB alone can condense purified plasmids (Petrushenko et al., 2006b) and entire chromosomes (Wang et al., 2006; Chen et al., 2008) and can establish bridges between distant DNA fragments (Petrushenko et al., 2010). MukB binds DNA in a highly cooperative manner producing ATP-controlled macromolecular clamps on DNA, which are competent for DNA bridging (Cui et al., 2008). The DNA bridging activity of MukB is highly selective for right-handed DNA crossings (Petrushenko et al., 2010), which might explain its ability to introduce right-handed loops into DNA (Petrushenko et al., 2006b) – a property that is shared by all tested condensins.

Whereas the reconstituted activities of MukB can readily explain its operation as a chromatin scaffold, much less is clear about the function of MukEF. Structural studies revealed significant similarities between MukF and eukaryotic kleisins (Fennell-Fezzie et al., 2005; Woo et al., 2009), which are thought to link the head domains of SMCs. MukF consists of two globular winged-helix domains (WHD) connected by a long linker, which includes a docking site for two globular MukEs and an additional binding site with MukB (Woo et al., 2009). The N-terminal WHD serves as a dimerization interface for MukEF while the C-terminal WHD interacts with the head domain of MukB. Because of steric constrains, the two MukFs in the dimeric MukEF cannot bind the same ATP-sandwiched MukB head and are probably engaged with distant MukBs.

The function of MukE is unknown. Reconstitution of MukEF in vitro produces a mixture of two species, MukE4F2 and MukE2F2 (Petrushenko et al., 2006a; Gloyd et al., 2007), but only at substoichiometric levels of MukE (Petrushenko et al., 2006a). This observation gave rise to the idea that the saturation level of MukE might affect the activity of the entire condensin complex (Gloyd et al., 2007). It remains to be tested whether or not the composition of MukEF varies inside the cell given that the expression level of MukE exceeds that of MukF by about twofold (Petrushenko et al., 2006a; Badrinarayanan et al., 2012).

Notably, reconstitution studies revealed only negative effects of MukEF on MukB–DNA interactions (Petrushenko et al., 2006a). Besides MukB4–Muk(E2F)2, the three proteins can form a less stable complex with the stoichiometry MukB2–Muk(E2F)2 (Petrushenko et al., 2006a). While the DNA binding properties of MukB and MukB4–Muk(E2F)2 are virtually the same, MukB2–Muk(E2F)2 binds DNA only poorly, and this inhibitory effect of excessive MukEF can be reproduced in live cells (She et al., 2007). Accordingly, recent single-molecule fluorescence studies revealed that the stoichiometry of MukBEF alternates between MukB4–Muk(E2F)2 in its stationary, presumably DNA bound fraction and MukB2–Muk(E2F)2 for the mobile population of molecules (Badrinarayanan et al., 2012). Taken together with the ATP dependence of both MukB–DNA (Cui et al., 2008) and MukB–MukEF (Woo et al., 2009) interactions, these data strongly argue for the regulatory role of MukEF.

In contrast, cell biology studies point to a ‘gain-of-function’ role of MukEF. Indeed, the phenotypes of MukB-, MukF- or MukE-deficient strains are indistinguishable and include severe defects in chromosome compactness and a decline in colony formation at 37°C and increased frequencies of anucleate cells at all temperatures (Yamanaka et al., 1996). Moreover, the absence of MukEF cannot be complemented by increased production of MukB even when chromosomes remain condensed (Wang et al., 2006). Finally, all three subunits of MukBEF are required for the focal localization of MukB–GFP and MukE–GFP at the quarter positions along the cell (Ohsumi et al., 2001; She et al., 2007).

A plausible explanation of these results involves postulating an additional intracellular function for MukEF. To address this issue, we carried out random mutagenesis of MukE and isolated six loss-of-function point mutations. All six MukE mutants were deficient in the focal subcellular localization but not necessarily in the assembly of MukEF or the ability of MukEF to disrupt MukB–DNA interactions. Moreover, four of the mutants formed a stable complex with MukB in vitro and in vivo and three mutants were able to join the existing MukBEF clusters. These data suggest that the primary function of MukE is to ensure focal subcellular localization of MukBEF and that the focal localization is likely to be essential for MukBEF function.


Location of the loss-of-function point mutations

Random mutagenesis of mukE was carried out using error-prone PCR. The amplified mutated gene was inserted into the pBB14 plasmid, which encodes mukF and the C-terminally GFP-tagged mukE under the control of the endogenous Pmuk promoter (Fig. 1A). The plasmid was transformed into the mukE-deficient strain AZ5450 (Yamanaka et al., 1996) and plated on LB. The temperature-sensitive colonies were then examined using fluorescence microscopy to eliminate misfolding mutants with low GFP fluorescence. The plasmids from the colonies that passed this selection were sequenced, which revealed nine distinct mukE mutants. The generated mutations could be found throughout the entire length of MukE, including its N- and C-terminal domains and the interface between them (Fig. 1E and F).

Figure 1.

Randomly generated loss-of-function point mutations in MukE.

A. Maps of the plasmids used for random mutagenesis of MukE (top) and integration of the mutant mukEgfp onto the E. coli chromosome (bottom). spt, spectinomycin resistance gene.

B. Quantitative immunoblot analysis of MukBEF expression in the late-exponential mutant OU111 cells (0.1 OD per lane). Serially diluted purified MukBEF was used as a loading control (three left lanes). The major degradation product of MukEGFP is marked with an asterisk.

C. Expression level of the mutant MukEGFPs in OU111 cells (± SD) averaged from up to four experiments. The group assignment of each mutation is shown below the graph.

D. Colony formation by mutant OU111 cells at 23°C and 37°C. The isogenic MG1655 (WT) and OU127 (ΔmukE::kan; ΔE) cells were used as the controls.

E and F. Location of the mutations. Mutated amino acids are shown as spheres in the context of MukBEF (E) and the MukE dimer (F). The structures were adapted from Woo et al. (2009) (PDB ID 3EUK and 3EUH). MukB is shown in grey, MukF in blue and the two MukE monomers in green and pink. The N-terminal domains of both MukEs are shown in grey. The coiled coils of MukB are coloured in orange; their extension is marked with arrows. Mutations are colour-coded according to their classification into groups (red, blue, yellow and black for groups I to IV). The N- and C-terminal winged helix domains of MukF are marked as MukFN and MukFC.

We next transferred all nine mutant mukEgfps onto the chromosome of the MG1655-derived cells yielding a series of OU111 strains (mukE::kan lacYA::mukEgfp) which lack the endogenous mukE and express mukEgfp from its native promoter at an ectopic location. In the resulting strains, six of the mutant MukE–GFPs were produced at similar levels to that of the wild-type protein (Fig. 1B and C). Two mutations, L54P and L47P R67C, resulted in very low protein expression suggesting defects in protein folding or stability. The ninth mutant, G96W, was produced at a reduced level, which points to its impaired stability. For unknown reasons, this mutation resulted in reduced production of MukF, perhaps by exposing the long linker of MukF to proteolytic degradation.

One of the mutants generated during screening contained two mutations, R34S and Y74C. Only the Y74C mutation was transferred into the OU111 strain and characterized further. This mutant was able to complement the temperature-sensitive phenotype of MukE-deficient cells. The remaining eight mutants showed a marked decline in colony formation at 37°C (Fig. 1D), which is consistent with their loss of function. We obtained the same result when we expressed the untagged versions of the genes from the pBB14-derived plasmid p15sp-E03a, which produces untagged MukE but not MukF at the same level as pBB14 (data not shown).

All nonviable MukE mutants are defective in focal localization of MukBEF

Previous studies established that MukBEF, but not its individual subunits, forms clusters at the 1/4 and 3/4 positions along the cell (Ohsumi et al., 2001; She et al., 2007). Such non-uniform subcellular localization of condensins was also observed in other bacteria (Lindow et al., 2002; Volkov et al., 2003), which argues for its functional importance. Here, we examined the effect of the generated mutations on formation of MukBEF clusters. As before (She et al., 2007), the mutant and wild-type OU111 cells were grown in M9 medium at 23°C up to OD600 of 0.7 and examined by fluorescence microscopy (Fig. 2). Clear foci at the cell quarters were found in 83 ± 2% of OU111-WT cells (Fig. 2A and B). Also, 71 ± 3% of OU111-Y74C cells, which encode the functional MukEY74C, contained clear fluorescent foci at their expected quarter positions (Fig. 2B). Very little GFP fluorescence could be found in the OU111-L54P and OU111-L47P-R67C cells, which is consistent with the previously found (Fig. 1C) low expression level of these MukE mutants. The remaining six mutant MukE–GFPs either were evenly distributed throughout the cell (Figs 2A and S1) or formed clusters at odd locations (mutants P69T and E70K). These odd foci appeared irregular and were often found away from DNA (Fig. 2C) suggesting that they likely emerge due to aggregation rather than interaction of MukBEF with its target.

Figure 2.

Subcellular localization of mutant MukBEF.

A. Subcellular localization by GFP-tagged MukE (left) or MukB (right) in OU110 (lacYA::mukE*gfp), OU111 (mukE::kan lacYA::mukE*gfp) or OU126 (mukE::kan lacYA::mukE* mukB::mukBgfp) cells that express either wild-type or mutant MukE. Arrows indicate cells with MukEG188EGFP foci at the 1/4 and 3/4 locations. DNA was stained with Hoechst 33342. Size bar, 2 μm.

B. Frequencies of cells with GFP foci in the mutant OU110, OU111 and OU126 cells. The group number of each mutant is shown in parentheses. For cells marked with an asterisk (mutants P69T and E70K), the mutant MukEGFP often formed clusters but rarely at the cell quarter.

C. Typical localization patterns observed for MukE–P69T and MukE–E70K.

We next determined whether or not mutations in MukE affect focal localization of MukB–GFP. To this end, we replaced the chromosomal mukB with its C-terminal GFP fusion, disrupted the endogenous mukE and inserted untagged mutant mukEs into the lacYA locus of the chromosome. The resulting OU126 cells (mukE::kan mukB::mukBgfp lacYA::mukEspc) displayed the same localization pattern for MukB–GFP as observed for MukE–GFP in OU111 cells. In total, 72 ± 3% of the wild-type OU126 cells and 50 ± 3% of the OU126-Y74C contained clear fluorescent clusters at the quarter positions. In the other five tested OU126 strains, MukB–GFP was evenly distributed across the cells (Fig. 2B). Thus, all loss-of-function point mutants of MukE failed to support assembly of MukBEF clusters inside the cell.

None of the mutants displayed a dominant phenotype, as determined by the analysis of colony formation by OU110 cells (lacYA::mukEgfp), which harbour the endogenous mukE in addition to a mutant, or by the wild-type MG1655 cells harbouring plasmid-encoded mukE mutants (data not shown). Thus, cellular functions appeared unperturbed by the presence of the mutant proteins. Curiously, four mutants frequently formed distinct foci at the quarter positions of the OU110 cells (Fig. 2A and B), which reveals their ability to join the endogenous MukBEF clusters. In contrast, the other three non-functional MukE–GFPs failed to localize at the middle of the nucleoids (Fig. 2B). Thus, the abilities to join existing MukBEF clusters and support their formation seem to be distinct activities of MukE.

Based on their localization patterns, we assigned the mutants to one of the four groups (Table 1). Group I mutants were well expressed but failed to form fluorescent foci either in the presence or absence of the endogenous MukE. Group II mutants could join MukBEF clusters but could not form them in the absence of endogenous MukE. MukEY74C, the sole member of group III, readily formed clusters at the quarter positions with or without endogenous MukE and complemented temperature sensitivity of MukE-deficient cells. Group IV mutants were produced at very low levels, at least when expressed from the chromosome.

Table 1. Properties of mutant MukEs
MutationaGroupExpression levelbViable?Foci in OU110Foci in OU111Assembly of MukEFcSized
  1. aIn addition, we found that the N138G N139A and R161A R163E R164A mutants are functional in vivo whereas the T142A S144A D145A mutant is non-functional and poorly expressed even from the plasmid.
  2. b+/− indicates the expression level at 25% to 50% compared to that for the wild-type protein.
  3. cMukEY74C forms a complex with MukF in vitro but fails to copurify with MukBEF on a sucrose gradient.
  4. dMukE size is reported as the measured oligomeric state of purified MukE (see Table 2 for details).
P69TI+/−NNN+2, 3
G188EII+NYN+2, 3

Some of the mutants form a complex with MukB

Failure of group I mutants to form or join MukBEF clusters suggested that they might be deficient in complex assembly with MukF or MukB. To test this notion, we transferred mutations onto the pBB08 plasmid (Wang et al., 2006), which encodes MukF and MukE–His9 under the control of an arabinose-inducible promoter. The proteins were overexpressed and purified using nickel-chelate chromatography. Surprisingly, only one mutant, MukEG96W, eluted alone from the column, without accompanying MukF (Fig. 3A). The other five tested proteins, from groups I, II and III, copurified as a complex with MukF. Moreover, the overall size and shape of MukEF were not significantly affected by mutations since sedimentation rates of the mutant MukEFs were within ± 5% of the value found for the wild-type protein (data not shown).

Figure 3.

Biochemical properties of mutant MukEs.

A. Copurification of MukF with mutant His-tagged MukEs. MukF and MukEHis9 were overproduced together and subject to nickel-chelate chromatography. The eluted fractions were then analysed by SDS-PAGE.

B. Gel filtration analysis of MukBEF assembly in the presence (left) and absence (right) of magnesium chloride. MukB2 and Muk(E2F)2 were reconstituted at 1 to 1 molar ratio and analysed by gel filtration through a sephacryl S300 column as previously described (Petrushenko et al., 2006a).

C. Mutant MukEFs inhibit DNA binding by MukB. Reconstituted MukB2Muk(E2F)2 complex was incubated with pBR322 DNA for 30 min at 37°C and analysed by gel shift assay as previously described (Petrushenko et al., 2006a).

D. DNA displacement activity of MukEF. MukB was incubated with DNA for 30 min, supplemented with MukEF and further incubated for various times (indicated above the gel). MukB–DNA binding was then evaluated using gel shift assay.

E. Effects of mutant MukEF on ATPase activity of MukB.

We next examined the ability of the purified MukEFs to associate with MukB. Previous studies revealed that depending on reconstitution conditions, the two proteins can form two distinct complexes (Petrushenko et al., 2006a). In the absence of magnesium, MukB and MukEF associate to produce the complex MukB4–Muk(E2F)2. In the presence of magnesium, the stoichiometry of MukBEF is MukB2–(MukE2F)2. Formation of both complexes was assessed here.

As before (Petrushenko et al., 2006a), assembly of MukBEF was evaluated using gel filtration chromatography. MukB2 and Muk(E2F)2 were mixed at the 1:1 ratio and, following reconstitution, loaded onto a gel filtration column. When the column was equilibrated in the EDTA-containing buffer, only half of MukEF comigrated with MukB, signalling formation of the MukB4–Muk(E2F)2 complex (Fig. 3B). In the presence of magnesium, very little MukEF remained free in solution, as would occur upon assembly of MukB2–(MukE2F)2. We conclude that all our purified MukEFs support formation of both MukBEF complexes.

Mutant MukEFs inhibit DNA binding by MukB and stimulate its ATPase rate

Reconstitution studies revealed that assembly of the saturated MukB2–(MukE2F)2 complex disrupts MukB–DNA interactions (Petrushenko et al., 2006a). We found that the same holds true for the mutant MukEFs. All reconstituted MukBEFs failed to produce a DNA gel shift found for MukB alone (Fig. 3C). In contrast, MukEG96W, which was purified without accompanying MukF (Fig. 3A), did not interfere with DNA binding by MukB.

We obtained the same result using the inhibition-of-relaxation assay. In this assay, supercoiled DNA is treated with wheat germ topoisomerase in the presence of MukB or MukBEF. We previously reported that MukB but not MukBEF precludes DNA relaxation (Petrushenko et al., 2006a). Here, we found that all our mutant MukEFs are equally efficient to the wild-type protein in blocking the ability of MukB to protect DNA from the topoisomerase (data not shown).

We next evaluated the DNA displacing activity of MukEF. Preformed MukB–DNA complexes were supplemented with excessive MukEF and incubated further for increasing times. The reactions were stopped by chilling them on ice and analysed by gel electrophoresis. At these concentrations of MukB, all DNA is bound by the protein and displays reduced mobility in the gel. As the incubation time with MukEF increases, the extent of the gel shift progressively declines (Fig. 3D) owing to the accumulation of the inactive MukB2–(MukE2F)2 complex (Petrushenko et al., 2006a). We found that all our purified MukEFs are as efficient in DNA displacement as the wild-type protein (Fig. 3D).

MukEF has been previously shown to stimulate the ATPase activity of MukB (Shin et al., 2009). We found that the same holds true for all our purified MukEF mutants. With the exception of MukEG96W (which failed to associate with MukF), all MukEFs were equally efficient in simulating the ATPase rate by MukB (Fig. 3E).

Oligomeric state of MukE mutants

A recent study proposed that dimerization of MukE might modulate the operation of MukBEF (Gloyd et al., 2011). To explore this possibility, we determined the apparent molecular weight of purified MukE by measuring the sedimentation rate and Stokes radius of the complex and then using the data to calculate the oligomeric state of the protein (Siegel and Monty, 1965).

In agreement with previous reports (Yamazoe et al., 1999), we found that MukE migrates as a mixture of dimers and monomers with the average molecular weight of 38 kDa, which corresponds to the average 1.3 MukE monomers per molecule (Table 2). Very similar numbers were obtained for the group II R140C and S141P mutants. The average oligomeric state for the G96W (group I) and Y74C (group III) mutants was somewhat lower, 1.1, which might be indicative of their diminished propensity for dimerization. In contrast, the P69T (group I) and G188E (group II) mutants migrated as a broad peak during gel filtration and formed two broad peaks during sucrose gradient centrifugation (data not shown). According to the measured migration rates, the oligomeric state of the two species was predicted as primarily dimers and trimers (Table 2). Thus, although some mutations did affect the propensity of MukE for self-association, there was no clear correlation between the oligomeric state of the protein and the observed phenotype.

Table 2. Protein size analysis of mutant MukE
MutantR, nmS, SMW, kDaana
  1. aThe molecular weight and oligomeric state of the proteins were calculated from their sedimentation rates and Stokes radii as previously described (Siegel and Monty, 1965).
  2. bThis protein formed two peaks during sucrose gradient centrifugation.
P69Tb 4.2822.8
G188Cb 4.6883

Assembly of mutant MukBEF in vivo

Most of our mutations in MukE did not have clear effects on the assembly or properties of MukBEF in vitro. There remained a possibility, however, that these mutations destabilize MukBEF inside the cell. To test this notion, we analysed the composition of MukBEF in the extract of gently lysed cells. Following cell lysis and DNA degradation, the extracts of exponential cells were loaded onto sucrose gradient, resolved by centrifugation and analysed for MukBEF content using immunoblotting.

For the wild-type MG1655 cells, we found a clear peak for MukB, MukF and MukE in fractions corresponding to 10.0 ± 0.3 S (Fig. 4A), which agrees well with previous studies of MukBEF (Yamazoe et al., 1999). Most of MukE and MukF were found in this peak. In contrast, the sedimentation profile for MukB included a second peak, at 7.1 S, which contained approximately half of the protein. For comparison, the sedimentation rate of purified MukB under somewhat different conditions was determined as 8.6 S (Petrushenko et al., 2006b). Thus, cellular MukB is about equally partitioned between its free form and MukBEF. This result is in full accord with earlier studies, which revealed excess of MukB over MukEF in whole cell extracts (Petrushenko et al., 2006a). In contrast, in vivo single-molecule fluorescence studies suggest the MukB4–Muk(E2F)2 composition of MukBEF, at least when it is a part of the assemblies at the cell quarters (Badrinarayanan et al., 2012). Apparently, the oligomeric MukBEF dissociates into monomers upon cell lysis.

Figure 4.

MukBEF assembly in vivo.

A and B. Immunoblot analysis of MukBEF composition in MG1655 (A) and OU111 (B) cells. Extracts from mid-exponential cells were supplemented with molecular weight markers and resolved by sucrose gradient centrifugation. The proteins in each fraction were visualized by immunoblotting using anti-MukB, anti-MukF and anti-MukE antibodies. Sedimentation rates of MukB and MukBEF complexes are indicated below the gels.

C and D. Sedimentation profiles of MukB, MukF and MukE in cell extracts of MG1655 (C) and OU111 (D) cells. The data averaged (± SD) from up to three experiments were fit to a double Gaussian distribution. The dashed lines show the best fit peaks for MukB and MukBEF.

E. The fraction of MukB and MukEGFP from the wild-type and mutant OU111 cells that migrated as a part of MukBEF.

F. The fraction of MukE and the mutant MukEGFP that was incorporated into MukBEF in OU110 cells.

We found similar sedimentation profiles for OU111 cells, which encode MukE–GFP from an ectopic location instead of endogenous MukE (Fig. 4B and D). In this case, however, the sedimentation rate of MukBEF was somewhat higher, 12.4 ± 0.4 S, which likely reflects the higher molecular mass of the GFP-tagged protein. In addition, more MukE–GFP migrated as a free protein (Fig. 4D) owing to the somewhat higher expression of the ectopically produced MukE–GFP (She et al., 2007).

Two of the six tested mutants, MukEG96W and MukEY74C, failed to produce a stable MukBEF (Fig. 4E and F). The reduced stability of MukBEY74CF might explain why this seemingly functional mutant was tested non-functional during screening, when coexpressed with MukF. For MukEG96W, complex assembly defects have already become evident during copurification studies with overproduced MukEF (Fig. 3A). Notably, the four remaining mutants, from both group I (MukEP69T) and group II (MukER140C, MukES141P and MukEG188E), were able to form a stable MukBEF (Fig. 4E). Thus, the inability of these mutants to support formation of MukBEF clusters at the cell quarters cannot be explained by their failure to join the MukBEF complex.

To further corroborate this conclusion we examined sucrose gradient profiles of OU110 cells, which encode both the wild-type MukE and mutant MukE–GFPs. This set-up evaluates the competition between mutant MukE–GFP and the endogenous MukE for incorporation into MukBEF and, thus, should readily pick up defects in stability of assembled MukBEF. Instead we found that mutant MukE–GFPs were incorporated into MukBEF with similar efficiencies with and without endogenous MukE (Fig. 4F). We conclude that most of our mutations did not affect stability of MukBEF in live cells.

Two clusters of surface mutations

Six of the generated mutations were located at the surface of the protein. Four of them, R67C, P69T, E70K and G188E, were found next to each other. In addition, the surface-exposed hydroxyl of Tyr74 was also found nearby (Fig. 5A). This area overlaps with the binding site for MukF (coloured purple in Fig. 5A), with Gly188 being a part of the interface. It is likely, therefore, that these mutations interfere with the assembly of the properly structured MukF–MukE complex. This conclusion is tentatively supported by the finding that the Y74C mutation destabilizes MukEF in live cells (Fig. 4E and F).

Figure 5.

Two clusters of surface-exposed mutations.

A. Location of the surface-exposed mutations on MukE. The two monomers of MukE and the groups I through IV mutations are coloured as in Fig. 1. Also shown are the site-specifically generated K150A, N138G N139A (NN), R161A R163E R164A (RRR) and T142A S144A D145A (TSD). The previously described mutations (Gloyd et al., 2011) L38S L41A (LL), F185S R186G (FR) and D196A R198A E199S R202S (DRER) are shown in lighter shades of gray or yellow. The combined LL and FR mutations inactivated MukE whereas the LL mutant was functional. In all cases, mutations that affected stability of the protein are black and functional mutations are yellow. Purple marks amino acid residues at the interface with MukF.

B. Colony formation by the cleft mutants of MukE at 23°C and 37°C. The proteins were expressed as GFP fusions from appropriate p15sp-E02a plasmids in OU127 (mukE::kan) cells.

C. Expression level of the cleft mutants. The proteins were expressed either from the p15sp-E02a plasmid (p) in DH5α cells or from the chromosome (c) of OU110 cells and visualized by immunoblotting using anti-MukE or anti-MukF antibody, as appropriate. For plasmid-harbouring cells, 0.005 OD cells was loaded into each well to visualize MukE and MukEGFP. In all other cases, 0.1 OD cells was analysed.

D. Distribution of MukBGFP in OU126-TSD and OU126-K150A cells.

The second buried mutation, G96W, is located in the midst of the helix bundle in the C-terminal domain of MukE (Fig. 1F). MukF interacts with this bundle in multiple places, suggesting that its correct organization would be important for the interaction between the two proteins. It seems fitting, therefore, that MukEG96W failed to form a complex with MukF in vitro (Fig. 3) and in vivo (Fig. 4). The other two buried mutations, L47P and L54C, are located within the N-terminal domain of MukE and failed to produce a stable protein, perhaps due to misfolding.

Curiously, the last two mutations, R140C and S141P, were found on a surface away from the known interactions of the protein (Fig. 1E). Such location suggests existence of a novel interaction within MukBEF that is important for its activity. To test this notion, we altered several nearby surface-exposed amino acids. Arg140 and Ser141 are located at the tip of two protruding α-helices that form a large cleft between the two monomers of MukE (Fig. 5A). Lys150, which is located in the cleft not far from Arg140 and S141, proved to be essential for the activity of the protein (Fig. 5B). The T142A S144A D145A mutation (TSD) also inactivated MukE. These three altered amino acids are located next to R140 and S141, and the observed phenotype confirms the importance of this area for the function of MukE. Both these mutants proved to be unstable in E. coli, even though the plasmid-encoded MukEK150A was produced at the expected level (Fig. 5C). The untagged plasmid-encoded versions of mukEK150A and mukETSD failed to complement temperature sensitivity of MukE-deficient OU127 cells (data not shown) and, when transferred onto the chromosome of OU126 cells, failed to support foci formation by MukB–GFP (Fig. 5D). Two other tested mutations, N138G N139A and R161A R163E R164A, were functional (Fig. 5B).


Condensins from diverse organisms share a characteristic multisubunit organization. A kleisin links together the head domains of two SMC proteins and brings into the complex other as yet poorly characterized non-SMC subunits (Graumann and Knust, 2009; Gruber, 2011). The five subunits of MukBEF are encoded as three polypeptides. The core of MukBEF consists of two MukBs, which are responsible for its DNA binding and reshaping activities. The two MukB heads can associate via the ABC-type ATPase site and dynamically interact with the kleisin MukF (Woo et al., 2009). This interaction is modulated by ATP hydrolysis, giving rise to the idea that kleisins regulate dissociation of MukB heads and, thereby, MukB–DNA interactions. Being a dimer itself, MukF is also able to link distant MukBs into a macromolecular association (Petrushenko et al., 2006a; Woo et al., 2009). In addition, MukF serves as a docking site for two asymmetrically bound MukEs (Yamazoe et al., 1999; Woo et al., 2009). MukE is essential for operation of MukBEF in vivo. Its function remains unknown.

Using random mutagenesis, we identified nine point mutations in MukE that affect its function. Two mutations, L54C and L47P R67C, are located within the N-terminal domain of MukE (Fig. 1F) and failed to produce a stable protein, presumably due to misfolding. Five mutations were found at or close to the MukE–MukF interface (Fig. 5A) and were apparently deficient in some aspect of MukEF organization. Two of these mutations, G96W and Y74C, could not produce a stable MukEF in live cells (Fig. 4E and F). Two other previously described mutations (Gloyd et al., 2011) map to the same area (Fig. 5A), which further underscores the importance of this site.

A recent crystallographic study reported that, when expressed without MukB, MukEF can dimerize via not only MukF but also MukE (Gloyd et al., 2011). This new dimerization interface partially overlaps with the MukF binding site but could conceivably take place on the second MukE monomer (coloured green in Fig. 5A). The significance of this interaction remains unclear because, according to the MukBEF structure, this surface is cradled within MukB and is not available for other protein–protein interactions (Fig. 1E). Pro69, Glu70 and Gly188 are located close to this interface and could conceivably be involved in this dimerization. Curiously, both the P69T and G188E mutants displayed increased propensity for dimerization, at least in the absence of MukF and MukB (Table 2).

Finally, the R140C and S141P mutations localized to the tip of the protein that faces away from MukF (Fig. 5A). According to the available crystal structures of MukE (Woo et al., 2009; Gloyd et al., 2011), this part of the protein is far from all possible protein partners (Fig. 1E), which suggests that it might be involved in previously unknown interactions. Two out of four mutations in the vicinity of this loop destabilized MukE and inactivated it, indicating that this area is important for the function of MukE.

All our loss-of-function mutants were defective in the focal subcellular localization typical for the endogenous MukBEF. Even the functional MukEY74C produced foci at a reduced rate (Fig. 2B). This finding is consistent with an earlier report that the flexible linker of MukF – which links MukE to MukBEF – is required for focal localization of MukB (Shin et al., 2009). In contrast, other known activities of MukE were barely affected. Only one of the six folded loss-of-function mutants failed to form MukBEF complex in vivo, and three mutants, MukER140C, MukES141P and MukEG188E, could associate with functional foci formed by endogenous MukBEF (Fig. 2B). Thus, the assembly of MukBEF into a complex does not suffice for its focal localization. Moreover, all four tested non-functional MukEFs were able to modulate MukB–DNA interactions and stimulate its ATPase rate (Fig. 3). These data indicate that the focal localization of MukBEF is an essential activity of the complex distinct from DNA binding. Furthermore, the primary function of MukE appears to be to ensure such focal localization of MukBEF.

This finding underscores the presumed role of MukBEF as a chromosome scaffold protein. As such, MukBEF would be expected not only to organize the chromosomal DNA but also to link it to the rest of the cell. Previous reconstitution studies revealed that MukB, the DNA binding component of the complex, is able to form a web of interconnected clamps that bring distinct DNA fragments together (Cui et al., 2008). This activity naturally explains how the protein could be stabilizing the giant-loop architecture of the chromosomal DNA. It tells nothing, however, about how the resulting scaffold would be positioned inside the cell or why the non-SMC components of the complex are essential. Our finding that MukE might be responsible for such localization suggests a plausible answer to these questions by postulating a distribution of functions within the complex.

The mechanism of this localization is yet to be determined. In the simplest model, MukE serves as a binding site for another factor that links it to cellular matrix (Fig. 6A). A promising candidate for such binding site would be the tips at the edge of the large cavity between the two MukEs. Indeed, two of our loss-of-function mutations are located in the loop atop α-helices that create the walls of this cavity and two more mutations were found by random picks nearby (Figs 1F and 5A). The cavity faces away from other components of the complex and could potentially serve as an interface with another factor. Recent studies revealed that condensins are recruited to the OriC-proximal regions in Bacillus subtilis (Gruber and Errington, 2009) and Streptococcus pneumonia (Minnen et al., 2011). Perhaps the predicted interaction serves a similar function in E. coli.

Figure 6.

A possible role of MukE in chromosome organization.

A. MukE is postulated to link MukBEF to an element on the cellular matrix located at the cell quarters.

B. A possible interplay between the DNA displacement and subcellular localization activities of MukE that could help recruit only the DNA bound condensins to the cell quarters.

A range of more elaborate models can be envisioned if one recalls the many functions of MukF. This protein not only links MukB with MukE and other MukBs but also regulates MukB–DNA interactions. It is not inconceivable that these three activities of MukF are interdependent and, thus, the protein might co-ordinate DNA binding by MukBEF and its recruitment to the quarter positions inside the cell. An example of such co-ordination is illustrated in Fig. 6B, where it serves to select only DNA bound MukBEFs for recruitment into macromolecular clusters at the cell quarters. In tentative support of this view, recent studies revealed that the stoichiometry of MukBEF changes from MukB2–Muk(E2F)2 to MukB4–Muk(E2F)2 upon its recruitment into the cell quarter foci (Badrinarayanan et al., 2012).

Notably, neither MukB–GFP nor MukE–GFP localizes to the cell quarters in the absence of their cognate subunits or DNA (Ohsumi et al., 2001; She et al., 2007). This suggests that formation of MukBEF clusters is a collective property of the complex and might also require correctly folded chromosomes. Perhaps the binding partners of MukBEF are concentrated at the cell quarters only in normally growing cells. Alternatively, MukB, MukF and MukE might contribute their unique features – such as DNA binding and high propensity for oligomerization – to facilitate formation of the scaffold. Finally, the role of MukE in positioning of the scaffold could be indirect. Rather than serving as a tether, the protein might mediate conformational transitions in MukBEF that induce its macromolecular association and dissociation or target it to particular addresses.

Experimental procedures

Strains and plasmids

Cells were grown in LB medium containing 0.5% NaCl or M9 medium supplemented with 0.4% casamino acids and 0.2% glycerol. Where appropriate, 100 μg ml−1 ampicillin, 20 μg ml−1 kanamycin, 100 μg ml−1 spectinomycin and 20 μg ml−1 chloramphenicol were added. Fluorescence microscopy was carried out as previously described (She et al., 2007).

The pBB08 plasmid harbours the smtAmukFmukE fragment of MukBEF operon under the control of the PBAD promoter (Wang et al., 2006). We C-terminally tagged the encoded MukE with GFP using H9G2A as a linker between the proteins, yielding plasmid pBB09. The BamHI–SalI fragment of pBB09 was then transferred onto pACYC184. The resulting pBB14 plasmid constitutively expresses SmtA, MukF and MukE–GFP (Fig. 1A). p15sp-E02a (She et al., 2007) is a pACYC184-based plasmid, which encodes MukE–His9Gly2Ala–GFP followed by a spectinomycin resistance gene aadA located between fragments of lacY and lacA genes (Fig. 1A). p15sp-E03a is the same except it encodes the untagged MukE.

OU110 (lacYA::mukEgfpspc) and OU111 (mukE::kan lacYA::mukEgfpspc) strains were described elsewhere (She et al., 2007). OU110-x and OU111-x strains, which encode mutant MukE–GFPs, were constructed in the same way. First, the NsiI–SpeI pBB14 fragments, which contain the generated mutations, were inserted between NsiI and SpeI sites of p15sp-E02a (Fig. 1A). The resulting cassettes harbouring mutant mukEgfps were then excised using digestion with SwaI and integrated into the lac locus of MG1655 using the λ Red recombination system (Datsenko and Wanner, 2000). The mutant OU111-x strains were then made by P1vir transduction of mukE::kan from AZ5450 into OU110-x.

OU119 (mukB::mukBgfpcat) was constructed using λ Red recombination by integration of the HpaI fragment of the pBB41 plasmid into the chromosome of MG1655 cells. pBB41 is a pBAD-based plasmid, which encodes MukB–His12Gly2Ala–GFP followed by the chloramphenicol resistance gene cat and ycbB, the gene located downstream from mukB on the E. coli chromosome. OU120 (lacYA::mukEspc) was constructed using the λ Red system by integration of the SwaI fragment from p15sp-E03a into the lac locus of the MG1655 chromosome. OU126 (mukE::kan lacYA::mukEspc mukB::mukBgfpcat) and OU127 was constructed by P1vir transduction of the mukE::kan fragment from AZ5450 (Yamanaka et al., 1996) into OU119 followed by transduction of the lacYA::mukEspc fragment from OU120. Transduction of the same mukE::kan fragment into MG1655 produced OU127 (mukE::kan).

Mutagenesis of MukE

Random mutagenesis of mukE gene was performed using GeneMorph II Random Mutagenesis Kit (Stratagene). First, a fragment of pBB14 plasmid containing the entire mukE gene and the flanking regions of mukF and gfp was amplified by PCR. The fragment was then used as a template in one cycle of error-prone PCR, digested with AleI and ApaLI and subcloned into pBB14. The resulting library was transformed into the ΔmukE strain AZ5450 and replica-plated on LB. Cells that could form colonies at 23°C but not 37°C were collected and examined by fluorescence microscopy as previously described (She et al., 2007). The cells with robust GFP fluorescence were assumed to lack severe protein misfolding defects. Out of 3025 screened colonies, 42 were temperature-sensitive, and 10 of them contained functional GFP. Plasmid sequencing revealed nine unique MukE mutants: E70K, R140C, S141P, G188E, R34S Y74C, L54P D195G, P69T D195G, G96W D195G and L47P R67C D195G. The R34S mutation was eliminated by subcloning, and only MukEY74C was analysed further. MukED195G complemented the temperature-sensitive phenotype of AZ5450 cells, whereas the properties of MukEP69T and MukEG96W were identical to those of the double mutants MukEP69TD195G and MukEG96WD195G (data not shown). The remaining two mutants were expressed at very low level and were analysed as generated.

Site-specific mutagenesis of mukE was carried out using PCR-based cloning and verified by sequencing.

Protein purification

MukB–His10 was purified using nickel-chelate chromatography as previously described (Petrushenko et al., 2006b). Mutant mukEs were amplified by PCR using pBB14 as a template, and the BsrGI–XbaI fragment of the product was used to replace the endogenous BsrGI–XbaI fragment of pBB08 (Wang et al., 2006). The plasmid, which encodes MukF and MukE–His9, was transformed into the ΔmukE AZ5450 cells and used to overproduce and purify mutant and endogenous MukEFs as described elsewhere (Petrushenko et al., 2006a).

Reconstitution of MukBEF was carried out as described before (Petrushenko et al., 2006a). MukE2F and 4 μg of MukB2 were mixed at 2 to 1 molar ratio and incubated for 20 min on ice in 40 μl of reconstitution buffer (20 mM HEPES, 200 mM NaCl, 20% glycerol, 2 mM MgCl2, 1 mM DTT).

MukE was purified as previously described (Petrushenko et al., 2006a) with minor modifications. Purified MukEF was denatured in 6 M guanidine hydrochloride, 200 mM NaCl, 20 mM HEPES, pH 7.7 and applied to a nickel column. The column was washed with 6 M guanidine hydrochloride, 200 mM NaCl, 20 mM HEPES, pH 7.7, 20 mM imidazole to elute MukF and then with 6 M to 0 M guanidine hydrochloride gradient in the same buffer. MukE was eluted with 20 mM to 400 mM imidazole gradient in the same buffer and dialysed against 20 mM HEPES, 200 mM NaCl, 50% glycerol, 2 mM EDTA, 1 mM DTT.

Biochemical analysis

The DNA gel shift assay was performed as previously described (Petrushenko et al., 2006b). Ten nanograms of supercoiled pBR322 was incubated with 0.1 μg of MukB for 30 min at 37°C in 9 μl of reaction buffer (20 mM HEPES, pH 7.7, 40 mM NaCl, 7% glycerol, 1 mM DTT, 2 mM MgCl2). 0.13 μg of MukEF in 1 μl (2 to 1 MukE2F to MukB molar ratio) was added, and the sample was further incubated for the indicated time. The reaction was stopped by chilling it on ice for 10 min and analysed by agarose gel electrophoresis in 89 mM Tris-borate at 4°C.

The ATPase activity of MukB was measured by quantifying the released inorganic phosphate using EnzChek Phosphate Assay Kit (Molecular Probes). Twenty-five micrograms of MukB was incubated with MukEF for 20 min on ice in 50 μl of 20 mM HEPES, pH 7.7, 200 mM NaCl, 20% glycerol, 1 mM EDTA, 1 mM DTT followed by 10 min incubation at 23°C in 980 μl of 20 mM HEPES, pH 7.7, 40 mM NaCl, 2 mM MgCl2, 5% glycerol, 1 mM DTT, 0.2 mM 2-amino-6-mercapto-7-methylpurine riboside and 1 U purine nucleoside phosphorylase. The reaction was started by the addition of 20 μl of MgATP to 0.5 mM, and UV absorbance at 360 nm followed over 80 min.

The DNA relaxation assay was performed as previously described (Petrushenko et al., 2006b). The Stokes radii and sedimentation coefficients of mutant MukE and MukEF were measured as described earlier (Petrushenko et al., 2006a).

Composition of MukBEF in vivo

One hundred and fifty millilitres of OU111 or OU110 cells was grown in LB at 23°C to an OD600 of 0.8, chilled by swirling the flask in the ice-cold water bath for 10 min, pelleted by centrifugation and washed once with TNS buffer (20% sucrose, 10 mM TrisCl, 100 mM NaCl). The cells were resuspended in 0.75 ml of TNS buffer, treated with 0.1 ml of TELyz buffer (35 mM TrisCl, 85 mM EDTA, 0.4 mg ml−1 lysozyme) for 1 min on ice and immediately lysed by gently mixing the suspension with 0.25 ml of BDE buffer (1% Brij58, 0.4% deoxycholate, 10 mM EDTA) followed by incubation at 23°C for 3 min. DNA was then digested by adding 10 μl of 20 mg ml−1 DNaseI and 2 mM MgCl2 for 5 min on ice. The reaction was stopped by adding 2 mM EDTA. 0.6 ml of cell lysate was mixed with 200 μg each of the protein size markers [apoferritin (17.6 S, 6.2 nm), catalase (11.2 S, 5.3 nm), alcohol dehydrogenase (7.3 S, 4.6 nm), bovine serum albumin (4.3 S, 3.6 nm)] and loaded onto a 10 ml 15% to 60% sucrose gradient in 10 mM Tris-HCl, pH 7.8, 200 mM NaCl, 2 mM EDTA, 1 mM DTT, 1 mM PMSF. The protein complexes were separated by centrifugation for 30 h at 38 000 r.p.m., 4°C using a Beckman Type 70 Ti rotor. After centrifugation, about 20 fractions were collected from the bottom and analysed by Western blotting using anti-MukB, anti-MukE or anti-MukF antibody. The sedimentation rate of each subunit was determined from the comparison with sedimentation profiles of the protein size markers.


This work was supported by a grant 1049755 from the National Science Foundation.