S. Khaitlina, Institute of Cytology, Russian Academy of Sciences, Tikhoretsky av. 4, 194064 St Petersburg, Russia Fax: +7 812 297 03 41 Tel: +7 812 297 42 96 E-mail: email@example.com
Homologous bacterial metalloproteases ECP32/grimelysin from Serratia grimesii and protealysin from Serratia proteamaculans are involved in the invasion of the nonpathogenic bacteria in eukaryotic cells and are suggested to translocate into the cytoplasm [Bozhokina ES et al. (2011) Cell Biol Int35, 111–118]. The proteases have been characterized as actin-hydrolyzing enzymes with a narrow specificity toward intact cell proteins. However, cleavage of filamentous actin (F-actin) (i.e. the main actin species in the cell) and the properties of the cleaved F-actin have not been investigated previously. In the present study, we revealed the presence of protealysin in the cytoplasm of 3T3-SV40 cells infected with S.proteamaculans or recombinant Escherichia coli expressing the protealysin gene. We also show for the first time that purified protealysin and the lysates of the recombinant E. coli producing protealysin cleave 20–40% of F-actin. Cleavage limited predominantly to the bond Gly42-Val43 efficiently increases the steady-state ATPase activity (dynamics) of F-actin. abolishes this effect and promotes the nucleation of protealysin-cleaved Mg-globular-actin even in the absence of 0.1 m KCl, most likely as a result of the stabilization of lateral intermonomer contacts of actin subunits. The results obtained in the present study suggest that F-actin can be a target for protealysin upon its translocation into the host cell.
Actin is involved in the majority of cell processes, ranging from chromatin remodeling and transcription regulation to intracellular transport, cell organization and various types of cell movements and muscle contraction. It is natural, therefore, that the actin cytoskeleton is a target for drugs and toxins, as well as for viruses and pathogenic bacteria interacting with eukaryotic cells [1–5]. The effectors modify cytoskeleton dynamics through the modulation of small GTPases and actin-binding proteins, or by direct interaction with actin. Many bacterial effectors promote the activation of small GTPases Rho, Rac and Cdc42, followed by the activation of signaling cascades resulting in actin polymerization [6–8]. Thereby bacterial effectors often mimic natural activators of small GTPases or directly stimulate the host signaling pathways by a functional mimicry of host GTPases [7,8]. Other bacterial effectors stimulate the efficient uptake of bacteria in host cells by the formation of complexes with actin-binding proteins , or by mediating phosphorylation/dephosphorylation of an actin-binding protein [10–13].
The other way in which many toxins and bacterial factors modify the cytoskeleton is the direct interaction of the effectors with actin. Latrunculin from sponges reversibly and specifically disrupts actin cytoskeleton by altering the actin monomer interface and thus inhibiting polymerization [3,14]. Marine macrolide toxins disrupt the cytoskeleton because they target both globular actin and actin filaments with a high affinity and either sterically prevent polymerization or mimic fragmentation of filamentous actin (F-actin) by severing proteins [15,16]. Clostridium botulinum C2 toxin and ADP-ribosyltransferase SpvB from Salmonella enterica cause ADP-ribosylation of actin at Arg177, thereby inhibiting actin polymerization [17,18]. A unique mechanism of actin filament depolymerization is exhibited by Vibrio cholerae RTX toxin, which covalently cross-links actin monomers, thus interfering with the formation of functional filaments . On the other hand, the bacterial protein SipA from Salmonella typhimurium inhibits depolymerization of actin filaments mechanically stabilizing the filament by tethering actin subunits in opposing strands [20,21]. Even these few examples demonstrate the variability of the mechanisms developed by bacteria and other organisms to subvert actin dynamics for facilitating the corresponding invasion or protection processes.
Previously, we reported that spontaneously isolated nonpathogenic bacteria Serratia grimesii and Serratia proteamaculans synthesize thermolysin-like metalloproteases ECP32/grimelysin [22,23] or protealysin [24–26], which are characterized by a high specificity toward actin. The appearance of the proteases correlated with the capability of the bacteria to invade eukaryotic cells and remodel their cytoskeleton [26,27]. Moreover, non-invasive E. coli transformed by the grimelysin or the protealysin gene confer the invasive phenotype . We have also shown that protealysin introduced into the culture medium does not induce the internalization of non-invasive E. coli , which led us to suggest that the enzyme translocates into the eukaryotic cell where it modifies cytoplasmic proteins.
Proteases ECP32/grimelysin and protealysin are actin-hydrolyzing enzymes with a narrow specificity toward intact cell proteins [22,26]. Limited proteolysis of globular actin (G-actin) by protease ECP32 abolishes or slows down polymerization [29,30] and enhances the dynamics of the filaments formed by the cleaved monomers , thus producing changes in the actin cytoskeleton favorable for bacteria to enter the host cell. However, under physiological conditions, actin cannot exist as a free monomer but forms filaments (i.e. F-actin) whose accessibility to proteolysis is very low [32,33]. The properties of cleaved F-actin have not been investigated previously. In the present study, we took advantage of using antibodies against protealysin, purified recombinant protealysin and lysates of recombinant E. coli expressing the protealysin gene, aiming to elucidate the localization of protealysin in the infected cell culture and the effects of protealysin on F-actin dynamics.
The results obtained demonstrate for the first time that the infection of eukaryotic cells with protealysin-producing bacteria is accompanied by the translocation of protealysin into the host cells. Furthermore, we show that up to 40% of F-actin is digested with protealysin in vitro within a time approximately corresponding to that of the invasion experiments. F-actin is cleaved by protealysin mainly at Gly42-Val43 within the DNase I-binding loop, which strongly enhances the filament dynamics. stabilizes the filaments and promotes the nucleation step of actin polymerization, most likely enhancing the effect of magnesium ions on the lateral intermonomer contacts along the actin filament. These properties of protealysin-cleaved actin did not depend upon whether purified protealysin or the lysates of E. coli expressing the protealysin gene were used. Therefore, we conclude that protealysin appears to be the only component of the bacterial lysates to affect actin dynamics. Taken together, these results suggest that, as protealysin translocates into the host cell, both G-actin and F-actin are targets for the enzyme.
Localization of protealysin in the infected cells
In invasion experiments, cultured cells are incubated with the fluorescein isothiocyanate (FITC)-conjugated bacteria, fixed with formaldehyde, stained with rhodamine–phalloidin, and examined under a confocal microscope to visualize bacteria and cytoskeleton structures [26,28]. We used a similar protocol to reveal the localization of protealysin in the infected cells by staining the enzyme with anti-protealysin serum. Figure 1 represents the results of a typical experiment in which cultured Balb 3T3 SV40 cells were incubated either with the wild-type protealysin-producing bacteria S. proteamaculans 94 (Fig. 1A) or with recombinant E. coli BL21 (DE3) (pProPlnHis6) expressing the protealysin gene (Fig. 1B). Lysates of these bacteria demonstrated efficient and specific proteolytic activity toward G-actin (Fig. 1D). As a control, we used recombinant E. coli BL21 (DE3) transformed with plasmid pET23b not carrying the protealysin gene (Fig. 1C). Lysates of E. coli BL21 (DE3) (pET23b) did not exhibit any actin-hydrolyzing activity (Fig. 1D). In line with our previous results [26–28], within 2 h of incubation, the actin-containing structures within the cells were rearranged, indicating that the stress fibres had been destroyed (Fig. S1) and ∼ 10% of the cells were invaded by either the wild-type or recombinant bacteria. Anti-protealysin serum revealed the enzyme in both the cells infected by S. proteamaculans (Fig. 1A) and those infected by the recombinant E. coli synthesizing protealysin (Fig. 1A,B). In addition, colocalization of antibodies with bacteria was sometimes observed. By contrast, E. coli BL21 (DE3) (pET23b) not carrying the protealysin gene did not enter 3T3 SV40 cells and, in these samples, no protealysin could be detected (Fig. 1C). These data imply that invasion of the protealysin-producing bacteria is accompanied by translocation of the enzyme into eukaryotic cells.
Limited proteolysis of F-actin
Previously, we showed that protealysin cleaves globular actin between Gly42 and Val43 within the DNase I-binding loop, generating two fragments of 36 and 5 kDa. At a high enzyme/actin ratio, further proteolysis of the 36-kDa fragment yields a 33-kDa product as a result of cleavage of the peptide bonds Thr66-Leu67 and Gly63-Ile64 in the nucleotide cleft . In the actin polymer (F-actin), the sites Gly42-Val43 and Gly63-Ile64 are involved in the monomer–monomer contacts [34,35], which can protect them from proteolysis by protealysin. Consistently with this, only the 36-kDa fragment was produced upon incubation of F-actin with protealysin (Fig. 2A), which indicates the poor accessibility to the enzyme of the sites within the nucleotide cleft. Compared to G-actin, the accessibility of the bond Gly42-Val43 within the DNase I-binding loop in F-actin was also diminished (Fig. 2A). However, quantification of the cleavage results showed that ∼ 20% and up to 40% of F-actin was digested at enzyme/actin mass ratios of 1 : 50 and 1 : 5, respectively, within a time approximately corresponding to that taken in the invasion experiments. In line with the earlier data , F-actin containing Mg2+ as a tightly bound cation exhibited an increased protection of the DNase I-binding loop from the cleavage. Importantly, the 36-kDa fragment was also the main digestion product upon incubation of F-actin with the lysates of recombinant bacteria expressing the protealysin gene, although, in this case, a small amount of the 33-kDa fragment was also formed (Fig. 2B). This additional cleavage is either the result of some contaminating proteolytic activity or a higher concentration of protealysin in the lysates that is difficult to relate to the concentration of the purified enzyme.
Because many proteases are known to cut off the C-terminal segment of the actin molecule  that is involved in maintaining the filament structure, we also monitored the cleavage of F-actin with protealysin by using actin modified at the penultimate Cys374 with N-iodoacetyl-N-(5-sulfo-l-naphthy1)ethylenediamine (IAEDANS). Independent of whether F-actin was incubated with the purified protealysin or bacterial lysates, in the AEDANS-labelled actin, Cys374, to which AEDANS was conjugated, was preserved (Fig. 2B). The C-terminus was also preserved in the smaller fragments of G-actin where it is fairly accessible for proteolysis (Fig. 2C) thus indicating the absence of contaminating proteolytic activity in bacterial lysates.
Because the crude bacterial lysates might still contain factors interfering with Mg-actin polymerization, the latter was tested in both lysate-treated and protealysin-treated actin samples. As shown in Fig. 3, the actin samples cleaved to a similar extent were proven to polymerize similarly, indicating that other components of the bacterial lysates do not interfere with the properties of actin.
Dynamics of the cleaved F-actin
Globular actin cleaved between Gly42 and Val43 is almost nonpolymerizable if it contains Ca2+ as a tightly bound cation [22,29,30]. The cleaved actin containing tightly bound Mg2+ was found to polymerize, although both the rate and extent of the reaction are lower than those for intact actin . Moreover, assembly of the cleaved Mg-actin monomers results in the formation of highly dynamic filaments as a result of an enhanced turnover rate of the polymer subunits . To determine whether the cleavage of subunits within F-actin results in the same effects, we compared the properties of the cleaved F-actin with those of F-actin assembled from the protealysin-cleaved monomers. Figure 4 shows that the cleavage of F-actin is accompanied by an ∼ 30% decrease in the intensity of both light scattering and pyrenyl fluorescence, which reflects the shift in the steady-state monomer/polymer ratio toward shorter filaments and/or the monomer. Indeed, cleavage of G-actin may contribute to the accumulation of the 33-kDa products during F-actin cleavage (Fig. 4, inset). In addition, during ultracentrifugation, ∼ 90% of protealysin-cleaved F-actin was recovered in the supernatant compared to 6–7% of the total protein in the supernatant of intact F-actin sedimented under the same conditions, although, morphologically, the protealysin cleaved filaments were similar to those of intact F-actin (not shown). This is consistent with the changes in the polymer stability produced by cleavage .
F-actin dynamics (i.e. the exchange of the filament subunits with the monomer pool) is associated with ATP hydrolysis after incorporation of ATP-containing monomers into the polymer. Therefore, F-actin dynamics can be evaluated by measuring the rates of ATP hydrolysis in F-actin solutions at steady-state. As shown in Fig. 5A, steady-state ATPase activity of F-actin assembled from protealysin-cleaved actin monomers was ∼ 4 mol Pi·(mol actin)−1·h−1, which is similar to the ATPase activity of F-actin assembled from the ECP32-cleaved monomers . The 30% efficient cleavage of F-actin with protealysin resulted in the enhancement of steady-state ATPase activity to ∼ 1.2 mol·Pi·(mol actin)−1·h−1 (Fig. 4B) compared to 0.02–0.04 mol·Pi·mol actin−1·h−1 in the case of non-modified F-actin under similar conditions [31,38].
Stabilization of F-actin cleaved with protealysin
Because F-actin dynamics is tightly coupled with ATP hydrolysis and the release of Pi, the binding of Pi analogue aluminum fluoride () in the nucleotide-binding cleft mimics the ADP-Pi or ATP state and stabilizes the filaments . We have previously shown that increases the thermal stability of F-actin assembled from the monomers cleaved between Gly42 and Val43 . Therefore, it was not unexpected that, in the presence of , the steady-state ATPase activity of F-actin assembled from the protealysin-cleaved monomers dropped to values characteristic of noncleaved protein (Fig. 5A). However, in addition, strongly accelerated the polymerization of protealysin-cleaved Mg-G-actin, apparently as a result of the stabilization of nuclei formed at the initial step of actin polymerization, as demonstrated by the disappearance of the long lag phase in the polymerization time course characteristic of the cleaved actin polymerization [Fig. 6, curve (Mg, Al) versus curve (Mg)]. Figure 6 also shows that the binding of does not accelerate polymerization of the cleaved Ca-G-actin [Fig. 6, curve (Ca, Al)], indicating a joint effect of Mg ions and fluoride in actin stabilization.
To determine whether binding of is accompanied by protection of subdomain 2 against proteolysis, as shown earlier for the binding to F-actin of BeFx , we probed the accessibility of subdomain 2 to proteolysis with protealysin. Indeed, inhibited cleavage of the bond Gly42-Val43 in F-actin subunits (Fig. 6, inset), indicating that the stabilization of actin filament with is a result of a tightening of the intermonomer contacts involving the DNase I-binding loop.
Serratia species are facultative pathogens that primarily cause nosocomial infections. Serratia marcescens is the most pathogenic species of this genus  and, until recently, it had been the only Serratia species shown to possess invasiveness mediated by a secreted pore-forming cytolysin ShlA . In addition, a number of reports describe human diseases from infections with other Serratia species, including S. grimesii and S. proteamaculans . However, the mechanisms of these infections are poorly understood. We have recently found that S. grimesii and S. proteamaculans can penetrate into eukaryotic cells, and their invasive activity correlates with a specific proteolytic activity of the bacterial lysates toward actin [26,27]. Moreover, non-invasive E. coli became invasive upon transformation with the grimelysin or protealysin genes, indicating that the enzyme can be actively involved in invasion . In the present study, we demonstrate for the first time that protealysin does translocate into the cytoplasm of eukaryotic cells. We also show that F-actin, usually resistant to proteolysis, is limitedly digested either with purified protealysin or with the lysates of recombinant E. coli expressing the protealysin gene, whereas the lysates of recombinant E. coli not carrying the protealysin gene do not display any actin-hydrolyzing activity. These data indicate that the actin-hydrolyzing activity of the E. coli lysates is indeed produced by protealysin, and the lysates do not appear to contain any other components actively affecting actin proteolysis and dynamics.
Despite the high actin-hydrolyzing activity of the bacterial lysates, the invasive activity of both wild-type Serratia and recombinant E. coli is relatively low, with only ∼ 11% and 16% of cells invaded by Serratia and recombinant E. coli, respectively . The low level of invasive activity does not allow us to quantify the amount of the enzyme within the host cells. However, a correlation between actin-hydrolyzing and invasive activities was clearly seen in the properties of the recombinant bacteria synthesized protealysin, with a point Glu113/Ala mutation in the active site that prevented removal of the propeptide from the protealysin precursor . The recombinant bacteria synthesizing inactivated protealysin were not found to be invasive unless a small amount of actin-hydrolyzing activity could be detected in their lysates at the late post-logarithmic growth stage . As the time of infection increased, the efficiency of the invasion augmented concomitantly with the enhancement of actin cleavage. Furthermore, changes in cell morphology and cytoskeleton rearrangements accompanying the invasion of E. coli synthesizing wild-type protealysin were more pronounced than those produced by the E. coli synthesizing mutant enzyme . These results suggest that cleavage-produced modifications of the properties of actin promote bacterial internalization.
Limited proteolysis of G-actin within the DNase I- binding loop has been shown to impair polymerization [22,29,46] and enhance the polymer dynamics , as well as decrease the affinity of gelsolin , myosin subfragment 1  and tropomyosin  to actin. It has been suggested that the properties of F-actin treated with protease would be similar to those of F-actin assembled from the proteolytically modified monomers. The data reported in the present study provide experimental support for this suggestion, which is important in view of the translocation of the proteolytic enzymes into the eukaryotic cell where F-actin is a major actin species. Most importantly, cleavage of F-actin with protealysin is accompanied by an efficient enhancement of the filament dynamics. On the other hand, aluminium fluoride, known to mimic a stable ADP-Pi or ATP conformation of actin filaments [39,41], accelerates the polymerization of protealysin-cleaved actin monomers whose nucleation is impaired by cleavage. Together with the increased dynamics of the cleaved F-actin, these effects provide mechanisms that may facilitate the invasion of eukaryotic cells by bacteria.
Actin dynamics is determined by the polarity of the actin filament where the protein subunits are added predominantly to the ‘barbed end’, and dissociate mainly from the other end, termed the ‘pointed end’. Cleavage of actin between Gly42 and Val43 resulted in enhancement of the turnover rate of polymer subunits by one order of magnitude, primarily as a result of a substantial increase in the rate constants for the dissociation of actin subunits from the filament ends. The dissociation rate constants were found to be approximately ninefold and more than fivefold higher for the cleaved actin at the barbed and pointed ends, respectively . These effects appear to be determined by the direct involvement of the cleavage site in the intermonomer contacts of the filament and changes in the overall conformation of subdomain 2 . In addition, the recently published 3D structure of the pointed end of the actin filament may shed light on the accelerated release of the cleaved ADP-actin subunits from the pointed end of the actin filament . In this model, the terminal subunit is tilted relative to the helical symmetry of the whole actin filament in such a way that the hydrophobic loop of this subunit and the DNase I-binding loop of the adjacent subunit are rearranged to form a tighter contact, which may be responsible for the low dissociation rate constants at the pointed end of the actin filament . Impairment of the DNase I-binding loop by proteolysis may prevent the tightening of the contact and result in faster dissociation of the cleaved subunits and accelerated filament dynamics.
Another important side of actin dynamics is the stabilization of filaments by actin-binding proteins or low molecular weight compounds. Recently, we have shown that aluminium fluoride greatly increased the thermal stability of F-actin assembled from the cleaved actin monomers . In the present study, we show that this effect is associated with the inhibition by of the filament dynamics and is likely a result of tightening of the intermonomer contacts involving the DNase I-binding loop. This is evident from the strong protection of the protealysin cleavage site between Gly42 and Val43 in subunits of -F-actin and is consistent with the stabilization of actin subdomain 2 by BeFx, as demonstrated by the 3D reconstruction of the filament structure  and biochemical approaches .
The penetration of bacteria into eukaryotic cells involves the disassembly of actin structures at the site of bacterial–cell contact followed by actin polymerization in cell surface protrusions, which rise around the bacterial body and allow its engulfment in a macropinocytic-like process [11,52]. The results obtained in the present study show that F-actin can be limitedly cleaved with protealysin within the DNase I-binding loop, and that the cleavage strongly enhances the dynamics of actin filaments and promotes their depolymerization. These effects can lead to the disassembly of the actin cytoskeleton at the site of bacterial–cell contacts and provide actin monomers for filament assembly in cell surface protrusions. During this latter step, the reversibility of the effects is of primary importance: assembly of the cleaved actin monomers into filaments may be promoted by the low molecular weight effectors mimicking Pi (present study) and actin-binding proteins [47,53] that stimulate actin polymerization and stabilize the filaments. Thus, the limited cleavage of F-actin with bacterial enzymes allows actin to preserve its functional properties and produce changes in the actin cytoskeleton that could be used by bacteria to enter the host cell. This specific activity of the enzyme toward actin, taken together with its narrow substrate specificity toward intact cell proteins [54,55], suggests that, upon translocation of protealysin into the host cell, actin proteolysis is involved in the internalization of bacteria.
Materials and methods
Tris, ATP (disodium salt), sodium azide, AlCl3, NaF, Triton X-100, reagents for electrophoresis, FITC, rhodamine–phalloidin, gelatin and Mounting medium were purchased from Sigma Chemical Co. (St Louis, MO, USA). Peptone and yeast extract were obtained from Difco (Franklin Lakes, NJ, USA). IAEDANS and N-(1-pyrenyl)iodoacetamide were from Molecular Probes Inc. (Carlsbad, CA, USA).
Bacterial strains, cell culture and growth conditions
S. proteamaculans 94  were grown in LB medium at 30 °C, with aeration. The recombinant E. coli BL21 (DE3) (pProPlnHis6) expressing the protealysin gene  and control recombinant E. coli BL21 (DE3) (pET23b) not carrying the protealysin gene  were grown in LB medium at 37 °C, with aeration times specified as appropriate.
The cell line Balb 3T3-SV40 (SV40 transformed fibroblasts 3T3) was obtained from the Russian Cell Culture Collection (Institute of Cytology, St Petersburg, Russia). Cells were grown on glass cover slips in culture dishes in the antibiotic-free DMEM (Biolot, Moscow, Russia) containing 10% fetal bovine serum (Sigma) in a incubator at 37 °C supplemented with 5% CO2 for the time required to form a monolayer (∼ 48 h).
Bacteria were grown as described above until the actinase activity of their extracts could be detected [26,28]. Thirty minutes before the experiment, FITC (1 mg·mL−1 of the bacterial suspension) was added to the bacterial culture to visualize bacteria. Bacteria were pelleted at 9600 g for 10 min; the pellets were washed with DMEM and added to the host cells in a fresh portion of DMEM. The host cells and bacteria were co-cultivated at 37 °C in 5% CO2 for the times indicated as appropriate.
Rabbit polyclonal antibodies against protealysin precursor were produced by Biotest Systems Ltd (Moscow, Russia) in accordance with a standard protocol. For immunization, recombinant protealysin precursor carrying mutation Glu113 to Ala (inhibiting the precursor maturation) and C-terminal His6-tag was purified as described by Gromova et al. . To visualize protealysin, 3T3-SV40 cells infected with S. proteamaculans 94, E. coli BL21 (DE3) (pProPlnHis6) expressing the protealysin gene or control recombinant E. coli BL21 (DE3) (pET23b) that did not carry the protealysin gene were washed with NaCl/Pi, fixed with 3.7% formaldehyde for 10 min and incubated with 0.1% Triton X-100 for 5 min. After washing three times with NaCl/Pi, the samples were incubated in 1% BSA for 30 min, and stained with polyclonal rabbit anti-protealysin antibodies dissolved in 1% BSA, for 18 h at 4 °C. The samples were then washed with 0.05% Tween in NaCl/Pi three times and incubated for 1 h at 4 °C with Alexa647-conjugated anti-rabbit IgG secondary serum (Jackson ImmunoResearch Inc., Bar Harbor, ME, USA). To visualize actin cytoskeleton, the antibody-stained samples were washed with 0.05% Tween in NaCl/Pi three times and with once with NaCl/Pi and incubated with rhodamine–phalloidin for 15 min at 37 °C in the dark. After washing with NaCl/Pi, the samples were mounted in the mounting medium and analyzed with a confocal laser scanning microscope.
Rabbit skeletal muscle actin was isolated from acetone dried muscle powder according to a standard procedure . G-actin was stored in buffer G (0.2 mm ATP, 0.1 mm CaCl2, 5 mm Tris–HCl, pH 7.5, 0.02% NaN3) on ice for 1 week or as 0.2-mL aliquots (0.5–1.0 mg·mL−1) frozen at −20 °C for a single use. ATP-Ca-G-actin was transformed into ATP-Mg-G-actin by 3–5 min of incubation with 0.2 mm EGTA/0.1 mm MgCl2 at room temperature. Actin labelled with N-(1-pyrenyl)iodoacetamide at Cys374 was prepared as described previously . Pyrenyl-labelled actin was lyophilized in the presence of 2 mm sucrose and stored at −70 °C. Before use, the lyophilized pyrenyl-labelled actin was dissolved in buffer G and dialyzed against the same buffer overnight. Actin labelled with IAEDANS at Cys374 was prepared as described previously . To stabilize actin with , G- or F-actin (24 μm) was incubated with 1 mm ATP and 5 mm NaF for 10 min, which followed by the addition of 1 mm AlCl3 . The concentration of G-actin was determined spectrophotometrically using an absorption coefficient of 0.63 mL·mg−1·cm−1 at 290 nm  or by the microbiuret method .
Protealysin was purified from E. coli BL21 (DE3) (pProPlnHis6) cell lysate by sequential metal chelate affinity chromatography on Ni2+-NTA-agarose (Qiagen, Valencia, CA, USA) and gel filtration on a Superdex 75 HR 10/30 column (Amersham Biosciences, Piscataway, NJ, USA) as described previously . Protealysin was stored in 25 mm Tris–HCl buffer (pH 8.0) at 4 °C.
Limited proteolysis assay
To estimate the susceptibility of F-actin to protealysin, actin was polymerized under different conditions known to affect the polymer structure . G-actin (0.5 mg·mL−1) in buffer G containing Ca2+ as a tightly-bound cation (Ca-G-actin) was polymerized with 0.1 m KCl for 2 h at room temperature. G-actin containing Mg2+ as a tightly bound cation (Mg-G-actin) was polymerized with 0.1 m KCl or 2 mm MgCl2 for 2 h at room temperature. F-actin was incubated with protealysin (0.3 mg·mL−1 in 25 mm Tris–HCl, pH 8.0) at various enzyme/protein mass ratios at 22 °C. At different time points, the digestion was stopped by the addition of an equal volume of the electrophoresis sample buffer containing 4% SDS, 125 mm Tris–HCl (pH 6.8) followed by 3 min of boiling. The digestion products were analyzed by SDS/PAGE . The actinase activity was determined by the appearance of specific actin fragments. The small (5 kDa) product of the cleavage was not seen on the gels.
To determine the ability of the protealysin-containing bacterial extracts to cleave actin, bacteria were grown as described above (usually for 18–24 h) and harvested by centrifugation at 9600 g for 10 min. The pellets were resuspended in buffer G, and the bacteria were lysed by five cycles of freezing and thawing. The lysates (bacterial extracts) were clarified by centrifugation at 9600 g for 10 min. F-actin in the corresponding buffer (0.5 mg·mL−1) was mixed with an equal volume of the clarified lysate and incubated for 2 h at room temperature or for 18 h at 4–6 °C. The reaction was stopped by the addition of the electrophoresis sample buffer followed by 3 min of boiling. The digestion products were analyzed by SDS/PAGE as described above.
ATP hydrolysis measurements
Protealysin-cleaved F-actin or Mg-G-actin, polymerized for 40 min to steady-state, were incubated at 22 °C. In aliquots of the F-actin solutions withdrawn after various time intervals, the ATPase reaction was quenched by the addition of an equal volume of 0.6 m ice-cold perchloric acid, precipitated protein was removed by centrifugation, and released Pi was determined by the Malachite green method .
Cell samples were examined under a confocal scanning microscope (Leica TCS SL; Leica Microsystems, Wetzlar, Germany) using the argon ion (488 nm; green fluorescence) and helium/neon (532 nm; red fluorescence; 633 nm, blue fluorescence) laser system to visualize the FITC-stained bacteria, rhodamine–phalloidin-stained cytoskeleton and protealysin, respectively.
Fluorescence and light-scattering measurements
All fluorescence and light-scattering measurements were performed in a Fluorat-02-Panorama spectrofluorimeter (Lumex, St Petersburg, Russia). The fluorescence intensity of pyrenyl-labelled actin was monitored at 407 nm after excitation at 365 nm . The intensity of light scattered at 90° was measured at 350 nm. To visualize AEDANS-labelled fluorescence peptides, gels were photographed in UV light before staining.
We thank Dr Ilya Nevzorov for discussions and his valuable help in the preparation of the manuscript, as well as the anonymous reviewers for their helpful comments. The work was supported by the Russian Foundation for Basic Research grants 11-04-00393 and 09-04-00734 and by the Program for Molecular and Cell Biology of the Russian Academy of Sciences.