The ActA protein of Listeria monocytogenes induces actin nucleation on the bacterial surface. The continuous process of actin filament elongation provides the driving force for bacterial propulsion in infected cells or cytoplasmic extracts. Here, by fusing the N‐terminus of ActA (residues 1–234) to the ω fragment of β‐galactosidase, we present the first evidence that this domain contains all the necessary elements for actin tail formation. A detailed analysis of ActA variants, in which small fragments of the N‐terminal region were deleted, allowed the identification of two critical regions. Both are required to initiate the actin polymerization process, but each has in addition a specific role to maintain the dynamics of the process. The first region (region T, amino acids 117–121) is critical for filament elongation, as shown by the absence of actin tail in a 117–121 deletion mutant or when motility assays are performed in the presence of anti‐region T antibodies. The second region (region C, amino acids 21–97), is more specifically involved in maintenance of the continuity of the process, probably by F‐actin binding or prevention of barbed end capping, as strongly suggested by both a deletion (21–97) leading to ‘discontinuous’ actin tail formation and in vitro experiments showing that a synthetic peptide covering residues 33–74 can interact with F‐actin. Our results provide the first insights in the molecular dissection of the actin polymerization process induced by the N‐terminal domain of ActA.
Several unrelated intracellular bacterial pathogens, including Listeria monocytogenes, Listeria ivanovii, Shigella flexneri and some Rickettsiae, share the ability to use actin polymerization as a driving force for intracellular movement, cell‐to‐cell spread and dissemination within infected tissues (for review, see Lasa and Cossart, 1996). Recently, vaccinia virus has also been shown to move through the host cell cytoplasm in association with an actin tail (Cudmore et al., 1995). The actin dynamics involved in the propulsive movement of these pathogens has striking similarities with the protrusion of lamellipodia in locomoting cells, providing a simple and powerful model to understand the actin assembly at the plasma membrane of motile cells.
The proteins responsible for actin‐based movement of L.monocytogenes, L.ivanovii and S.flexneri (ActA, IactA and IcsA, respectively) have been identified (Lett et al., 1989; Domann et al., 1992; Kocks et al., 1992; Gouin et al., 1995; Kreft et al., 1995). The overall homology between ActA and IactA is not high (39% identity) and there is no detectable sequence homology between ActA and IcsA. However, converging data suggest that these proteins may interact at different levels of a common pathway leading to actin assembly (for reviews, see Cossart, 1995; Theriot, 1995; Lasa and Cossart, 1996).
ActA is a surface protein of 639 amino acids which is anchored to the bacterial membrane (Domann et al., 1992; Kocks et al., 1992). Several lines of evidence demonstrate that ActA is the bacterial factor responsible for actin‐based motility. First, ActA‐deficient mutant strains do not polymerize actin (Domann et al., 1992; Kocks et al., 1992). Second, wild‐type ActA, when expressed in mammalian cells, is targeted to mitochondria (‘mitochondrial targeting assay’), which become covered with actin filaments (Pistor et al., 1994). Similarly, when ActA is fused to the plasma membrane anchor sequence of K‐ras (CAAX‐box), it induces actin polymerization at the inner face of the plasma membrane and changes in cell morphology (Friederich et al., 1995). Third, heterologous expression of ActA in the non‐pathogenic bacterium Listeria innocua (Kocks et al., 1995), or binding of a recombinant ActA protein to the surface of Streptococcus pneumoniae (Smith et al., 1995) are sufficient to produce comet tails and actin‐based motility in Xenopus extracts (Theriot et al., 1994).
Due to the presence of a proline‐rich repeats region in the central part of the protein, ActA has been artificially divided into three domains: (i) an N‐terminal domain (1–234; numbering of amino acids as described by Kocks et al., 1992); (ii) the proline‐rich repeats region (235–395); and (iii) the C‐terminal domain (396–585; Lasa et al., 1995). Sequence analysis of ActA revealed that the proline‐rich repeats and the C‐terminal domain share significant sequence similari‐ties with zyxin, an actin‐binding protein associated with focal contacts and stress fibres (R.Goldstein, E.Friederich and D.Louvard, personal communication). In contrast, the N‐terminal domain of ActA is not similar to known proteins, and only small motifs, either similar to the presumptive actin binding site of caldesmon (64‐LKEKAE‐70) or present in the cofilin‐ADP protein family (115‐EIKKRRKA‐122), can be detected.
The mechanism by which ActA induces actin polymerization remains an open question. Attempts to demonstrate in vitro interactions between ActA and G‐actin have failed (Kocks et al., 1992) and Listeria grown in broth do not nucleate actin efficiently (Tilney et al., 1992). Microinjection of fluorescently labelled phalloidin in infected cells (Sanger et al., 1995), or treatment of cytoplasmic extracts of Xenopus eggs with phalloidin (Marchand et al., 1995), causing polymerization of endogenous G‐actin, have suggested that bacteria do not recruit preformed actin filaments to the actin tail. Thermodynamic studies of L.monocytogenes motility in Xenopus extracts support the view that actin assembly results from the local maintenance of uncapped filament barbed ends at the bacterium surface (Marchand et al., 1995). These observations can be explained by considering either that ActA recruits one or more host proteins responsible for the nucleation and filament uncapping activities, or that modification of ActA inside cells gives this protein the ability to possess these activities. Several actin‐binding proteins (α‐actinin, tropomyosin, fimbrin, talin, villin, profilin, VASP and vinculin) have been localized to the comet tails or at the rear of the bacteria, where actin polymerization occurs (for review, see Cossart, 1995). Among these proteins only VASP (vasodilator‐stimulated phosphoprotein) is able to bind purified ActA (Chakraborty et al., 1995). VASP also interacts with profilin (Reinhard et al., 1995), a 15.5 kDa protein that plays a key role in actin polymerization (Pantaloni and Carlier, 1993).
In a previous study (Lasa et al., 1995), we generated several L.monocytogenes strains expressing ActA protein variants. Analysis of these strains demonstrated that: (i) deletion of the N‐terminal domain inhibits actin tail formation; (ii) deletion of the proline‐rich repeats decreases the speed of movement, but actin tails are still formed; and (iii) deletion of the C‐terminal domain has no effect on actin‐based motility. Since VASP binds to the proline‐rich repeats of ActA (Pistor et al., 1995; also our unpublished results), these data suggested that profilin bound to VASP could provide polymerization‐competent actin monomers, and thereby increase the efficiency of the ActA‐mediated actin polymerization process, but that VASP is not essential for the process. In agreement with these results, depletion of profilin in Xenopus extracts by >98% has no effect on bacterial movement (Marchand et al., 1995).
In the present work, we demonstrate that the first 234 amino acids of ActA (N‐terminal domain) contain all of the elements necessary to induce actin tail formation and movement in Xenopus cytoplasmic extracts. In order to identify more precisely the regions of the N‐terminal domain necessary for the actin polymerization process, a set of six ActA variants carrying small deletions in the N‐terminal domain was constructed. Analysis of the ability of these ActA variants to induce actin tail formation, inhibition of the process with antibodies and in vitro experiments using synthetic peptides and purified G‐actin or F‐actin allowed the identification of two regions (spanning residues 21–97 and 117–121) critical for the actin‐based motility. Based on these results, a hypothetical function for these two regions in actin tail formation is proposed.
The N‐terminal domain of ActA is sufficient to induce actin‐based motility in Xenopus extracts
It was previously shown that the N‐terminal domain of ActA is the only domain that is absolutely required for actin comet tail formation in infected cells (Lasa et al., 1995). However, an ActA variant lacking both the central and the C‐terminal domains and expected to trigger actin assembly was not associated with actin in infected cells, probably because part of the protein was embedded in the cell wall. In order to demonstrate that the N‐terminal domain of ActA is sufficient to induce actin‐based motility, we replaced the central and C‐terminal domains of ActA by an unrelated protein to act as a spacer. We chose the ω domain of the β‐galactosidase from Escherichia coli (Ullmann et al., 1968) (Figure 1A). This domain is not involved in tetramer formation in the β‐galactosidase enzyme (Jacobson et al., 1994) and has nearly the same size (308 amino acids) as the ActA fragment that it replaces (346 amino acids). When expressed in a L.monocytogenes strain carrying a complete deletion of actA (ΔactA) (Figure 1B), the chimeric protein ActA(N)–LacZ induces the formation of comet tails in cytoplasmic extracts (Figure 1C). The percentage of bacteria associated with actin tails was equivalent to that formed with bacteria expressing the wild‐type ActA (∼25%). However, the size of the tails made by bacteria expressing the chimeric ActA(N)–LacZ protein were clearly shorter than those produced by bacteria expressing wild‐type ActA, suggesting that they were moving more slowly [rapidly moving bacteria tend to be associated with longer comet tails (Theriot et al., 1992)]. Using video‐microscopy recordings of bacteria moving in cytoplasmic extracts, we found that bacteria expressing ActA(N)–LacZ protein moved at an average speed of 4.4 μm/min (n = 93, SD = 0.7), three times more slowly than bacteria expressing wild‐type ActA (11.8 μm/min; n = 76, SD = 2.5).
To analyse the ability of ActA(N)–LacZ to cause actin assembly in infected cells, three types of cell lines were infected: macrophages (J774), Vero and PtK2 cells. After 5 h of infection, cells were fixed, F‐actin was stained with FITC–phalloidin and bacteria were stained with rhodamine‐labelled anti‐ActA antibodies. In all three types of cell, bacteria expressing ActA(N)–LacZ were surrounded by F‐actin and, in rare cases, very small actin comet tails were visible at one pole of some bacteria (data not shown). These bacteria did not form the microcolonies characteristic of the L.monocytogenes ΔactA strain (Kocks et al., 1992), indicating that bacteria were moving in the cytoplasm of the infected cells, albeit very slowly (see Discussion).
As expected from transfection experiments (Pistor et al., 1995) and from our results showing that, in the absence of the proline‐rich domain of ActA, no VASP can be detected, immunolocalization experiments in cells infected with bacteria producing ActA(N)–LacZ did not reveal any VASP on the bacterial surface (data not shown). These results support the view that VASP is not absolutely essential for the ActA‐mediated actin polymerization process.
Genetic analysis of the N‐terminal domain of ActA: identification of two regions necessary for actin comet tail formation
In order to identify the regions in the N‐terminal domain of ActA involved in the actin polymerization process, we generated several ActA variants in which small fragments of the N‐terminal domain were deleted. The deletions were chosen on the basis of sequence comparisons between ActA and IactA, the homologous protein from L.ivanovii (Figure 2). Although the N‐terminal domains of ActA and IactA do not show strong similarity (25% identity), there are two regions with high homology. The first region (116–123 amino acids) is rich in positively charged amino acids, the second region (41–64 amino acids) has 58% identity on a 23 amino acid overlap.
A schematic representation of the different ActA variants is shown in Figure 3A. As shown by immunoblotting of total protein extracts using anti‐ActA affinity‐purified antibodies (A18K), these proteins were well expressed in L.monocytogenes ΔactA strain and migrated with the expected motility (Figure 3C).
The ability of the different ActA variants to associate with polymerized actin was analysed in infected cells and in Xenopus laevis cytoplasmic extracts (Figure 3B). Two phenotypes were observed with respect to the actin polymerization process: either bacteria associated with actin without any visible comet tail; or bacteria associated with comet tails. Bacteria expressing ActA‐Δ158–231 induced the formation of actin tails similarly to bacteria producing wild‐type ActA in infected cells as well as in Xenopus extracts. Furthermore, the speed of movement was in the same range (11.1 μm/min; n = 93, SD = 0.7), indicating that this region (158–231) is not necessary for the process. In contrast, bacteria expressing ActA‐Δ126–231, ActA‐Δ114–231 and ActA‐Δ97–231 were surrounded by F‐actin, but actin comet tails were not visible. To address the role of a positively charged peptide (117‐KKRRK‐121) present in this region, in the actin polymerization process, we deleted these five amino acids. Bacteria expressing this protein variant (ActA‐Δ116–122) were surrounded by F‐actin, but were not able to produce actin comet tails, suggesting that this positively charged peptide is critical for actin tail formation. To confirm this result, we made use of antibodies raised against a synthetic peptide covering residues 102–123 (P102–123). Incubation of L.monocytogenes expressing wild‐type ActA with increasing concentrations (0.1, 1 and 5 μg/ml) of affinity‐purified anti‐P102–123 antibodies for 1 h strongly decreased the number of bacteria able to induce comet tails in Xenopus extracts (Figure 4). In contrast, incubation under the same conditions with antibodies raised against a synthetic peptide containing part of the first proline‐rich repeat 235–252 (Y21T) had no significant effect on the ability of the bacteria to induce comet tails.
In a previous paper, we demonstrated that bacteria expressing ActA‐Δ21–231 were not associated with F‐actin (Lasa et al., 1995). As shown in the present work, bacteria expressing ActA‐Δ97–231 are still associated with F‐actin, strongly suggesting that region 21–97 is also involved in the actin polymerization process. To address this issue, an ActA‐Δ21–97 variant was produced. Bacteria expressing this protein were able to induce actin tails and movement in Xenopus extracts. However, the actin tails were strikingly discontinuous (Figure 5A). This phenomenon is analysed further in the next section.
Taken together, our results indicate that two regions within the N‐terminal domain play a critical role in the actin‐based motility. The region containing the peptide KKRRK (region T) is essential for actin tail formation, and a second region located between residues 21 and 97 (region C) is crucial for the continuity of the actin tail formation.
Analysis of the discontinuous actin tails produced by bacteria expressing ActA‐Δ21–97
The movement of L.monocytogenes expressing wild‐type ActA can be easily visualized under phase‐contrast microscopy as phase‐dense tails behind moving bacteria. Under these conditions, bacteria expressing ActA‐Δ21–97 produced discontinuous phase‐dense spots (Figure 5A). When phase contrast (to visualize bacteria) and fluorescence (to show rhodamine‐labelled actin tails) were used simultaneously, we found that the phase‐dense spots observed by phase contrast corresponded to the intensively stained spots observed by fluorescence microscopy (Figure 5A). Detailed analysis, in real time, of actin tail formation and movement showed that bacteria were first surrounded by a strongly fluorescent or phase‐dense spot at their pole, while moving slowly (1.4 μm/min; n = 37; SD = 0.7). After this apparently rate‐limiting step, the tail was elongated and the speed of the bacteria clearly increased (11.4 μm/min; n = 37; SD = 2.1). While bacteria were moving forward, the fluorescence intensity of the tail, as well as the speed of movement, decreased. Bacteria immediately began to be surrounded at their pole by an intensively stained spot, while remaining nearly immobile, and a new cycle of movement started (Figure 5B). Each complete cycle was achieved in ∼55 s. We were interested in comparing the average rate of movement of bacteria expressing ActA‐Δ21–97 with that of strains expressing wild‐type ActA. The average speed of movement for bacteria expressing ActA‐Δ21–97 was 7.2 μm/min (n = 51; SD = 1.8), almost half that of wild‐type ActA (11.8 μm/min; n = 76; SD = 2.5).
Actin‐binding properties of synthetic peptides corresponding to the regions of ActA important in actin recruitment and propulsion
One of the immediate questions emerging from our genetic analysis concerned the direct actin‐binding properties of the identified ActA regions. To address this issue we first synthesized a peptide covering sequence 33–74 of ActA (P33–74) and measured its actin‐binding properties. Like thymosin β4 (Van Troys et al., 1996), P33–74 was able to inhibit salt‐induced polymerization of Ca–G‐actin in a concentration‐dependent manner. Nearly complete inhibition was observed for more than 30 min when the peptide was added in a 20‐fold molar excess over G‐actin (Figure 6A). These data suggest that P33–74 is able to bind G‐actin. Peptide P33–74 was also recovered in F‐actin pellets obtained by airfuge centrifugation, indicating that it also binds to filamentous actin (Figure 6B). A shorter peptide covering residues 52–74 of ActA (P52–74) displayed similar actin‐binding properties, although higher concentrations of peptide were required to reach activities similar to that of P33–74 (Figure 6B). A further truncated peptide, P63–74, was inactive (data not shown), delineating the minimal actin‐binding region to residues 33 and 63 of ActA.
In order to assess the specificity of the peptide activities, we synthesized a number of variants of P52–74 that carried single amino acid substitutions scattered throughout the peptide. The activities of each of these variants were measured by inhibition of fluorescence (data not shown) and by sedimentation analysis (Figure 6C). Several of the variant peptides display weaker F‐actin inhibitory activities, in particular those carrying substitutions L60G, L64G, K65E, K67E and K70E. These results are in line with the idea that region 53–65 could fold as an α‐helix, while region 66–72 could adopt a highly charged coil structure. Indeed, these exchanges affect the helix stability (L60G and L64G) or reverse charge distributions in the loops (K65E, K67E and K70E) (Figure 6D). This situation could be similar to what was previously observed for thymosin β4, in which the actin‐binding segment also consists of an α‐helix followed by a highly charged loop region (Van Troys et al., 1996).
The possible actin‐binding properties of region T were studied using the synthetic peptide P102–123 of ActA. When added to preformed actin filaments in increasing concentrations, we measured saturation of co‐precipitation at a 20‐fold molar excess (data not shown). Interestingly, the pellets showed a whitish appearance that was clearly different from the translucent nature of normally formed F‐actin pellets. P102–123 further induced actin precipitation when added to low‐salt G‐actin solutions. Since a similar effect was also observed with poly‐l‐lysine, we assume that the observed properties of P102–123 could be due to the highly positively charged nature of the peptide. We therefore refrain from concluding that P102–123 is a true actin‐binding peptide, although at this stage there is little evidence to support the view that it is not.
We provide here for the first time evidence that the N–terminal domain of ActA contains all the elements necessary to induce actin tail formation and movement. In addition, we have identified two peptide stretches involved in different steps of actin tail formation.
The N‐terminal domain of ActA can by itself induce actin tail formation
We had previously shown that the first 234 amino acids of ActA were critical for actin‐based motility (Friederich et al., 1995; Lasa et al., 1995). By using an ActA(N)–LacZ chimeric protein we demonstrate that the N‐terminal domain is sufficient for actin‐based motility in Xenopus cytoplasmic extracts, but actin tails with this construct were hardly observed in infected cells. It is now well established that filament disassembly in the actin tail is identical in all tests (Theriot et al., 1992), which explains why faster‐moving bacteria have longer tails. However, it has been noticed that, in Xenopus extracts, the average length of actin tails increases with time during incubation, which suggests that the efficiency of the depolymerization decreases, an event partially circumvented by the use of an ATP‐regenerating system (creatine phosphate) (Marchand et al., 1995). In contrast, inside cells, the average length of the tails does not increase with longer infection times, indicating that the efficiency of depolymerization does not change. Consequently, actin comet tails inside cells are not detectable if the rate of polymerization efficiency is only slightly higher than the depolymerization process. This is probably the case for the ActA(N)–LacZ construct. Indeed, bacteria expressing the chimeric ActA(N)–LacZ moved almost three times more slowly than bacteria expressing wild‐type ActA, indicating that the efficiency of the polymerization process is relatively low. This inefficiency may be due to the lack of VASP binding. Inside infected cells, VASP does not co‐localize with bacteria expressing either ActA(N)–LacZ or ActA‐ΔP (our unpublished results), strongly suggesting that binding of VASP, profilin and perhaps another unidentified protein, to the proline‐rich repeats of ActA, is necessary to increase the efficiency of the actin polymerization process, but not that of actin comet tail formation. In terms of virulence, VASP probably plays a critical role by allowing actin‐based cell–cell spreading.
Region T (117–121) is necessary for actin tail formation and region C (21–97) is necessary for continuous actin filament elongation
By transfection of mammalian cells with different ActA variants, Pistor et al. (1995) established that region 128–152 (numbering of amino acids as described in Domann et al., 1992) or 99–123 (numbering as described in Kocks et al., 1992), is essential for actin filament nucleation. However, their transfection assay could not distinguish between actin assembly and actin tail formation and movement. Our deletion analysis of the N‐terminal domain of ActA in bacteria establishes the existence of two phenotypes with respect to the actin polymerization process: (i) bacteria surrounded with actin but unable to form comet tails; and (ii) bacteria associated with comet tails. These different phenotypes are associated with different regions of the N‐terminal domain of ActA. Region T overlaps the residues previously identified by the ‘mitochondrial targeting assay’ (Pistor et al., 1995) and is responsible for an efficient actin polymerization process leading to the actin tail formation. In its absence, bacteria are associated with F‐actin but are unable to produce actin tails.
The fact that bacteria expressing ActA‐Δ97–231 were still associated with F‐actin, while bacteria expressing ActA‐Δ21–231 were not (Lasa et al., 1995), strongly suggested that a second region covering residues 21–97 was able to induce actin polymerization, or at least was able to recruit F‐actin. In agreement with this idea, bacteria expressing an ActA variant, in which the region between amino acids 21 and 97 was deleted (ActA‐Δ21–97), induced actin comet tails, albeit discontinuously, as if formation of actin tail in this mutant was due to periodic cycles of starting and stalling. Therefore, region C also plays a role in the actin assembly process.
Although previous attempts to demonstrate interactions between ActA and G‐actin–ATP or F‐actin have failed, in vitro experiments with a synthetic peptide containing residues 33–74 demonstrated that this region is able to bind F‐actin. One possible explanation is that, in ActA, region 33–74 may be shielded. The case of ActA would be similar to that of vinculin, a 1066 amino acid protein containing a N‐terminal globular head and a C‐terminal rod‐like domain connected by a short proline‐rich motif. As recently established, full‐length vinculin interacts with its ligands (F‐actinin, α‐actinin, talin, phospholipids) less efficiently than each of its isolated domains (for review, see Jockusch and Rüdiger, 1996), indicating that in the intact molecule, the ligand‐binding sites are not always in an active form. In the case of vinculin, the factors involved in the conformational switches seem to be a cooperative combination of phosphorylation, binding of PtdIns(4,5)P2 and binding of the various protein ligands. In the case of ActA, phosphorylation (Brundage et al., 1993) and/or ligand binding (VASP) (Chakraborty et al., 1995) could induce a similar functional switch and unshield either actin binding sites or binding of intermediate proteins which stimulate actin binding.
Towards a model of actin assembly by the N–terminal domain of ActA
How can one explain the continuous actin polymerization process induced by ActA? Also, how can the discontinuous actin tail formation induced by ActA‐Δ21–97 be explained? We have to take into account three sets of data. First, that nucleation (assembly of three monomers of actin to form a trimer) is the rate‐limiting step for actin polymerization (Pollard and Cooper, 1986); the next step, elongation (addition of actin monomers to the trimers, resulting in filament formation) occurs rapidly once nucleation has taken place. Second, that electron micrographs of actin comets produced by L.monocytogenes show that filaments in the tails are short (0.2 μm) and oriented with their barbed ends (fast‐growing) towards the bacteria (Tilney et al., 1992). It has been proposed that, while continuous nucleation takes place, filament elongation may be terminated by barbed end‐capping proteins, followed by cross‐linking of the filaments (Tilney and Tilney, 1993). A tail is thus the result of nucleation, elongation, capping, release and cross‐linking. Third, that thermodynamic studies have established that actin assembly results from the local maintenance of uncapped filament barbed ends at the bacterium surface (Marchand et al., 1995).
Based on these data, we propose that region T is more specifically involved in elongation/tail formation and region C in maintaining the barbed ends uncapped, thereby allowing continuous filament elongation. If we consider that propulsion of the bacteria is due to elongation of a critical minimum number of actin filaments, in bacteria expression wild‐type ActA, the filaments lost for the elongation process by capping are constantly replaced by newly nucleated filaments. Since the rates of capping and nucleation do not vary, at least over short intervals, the F‐actin concentration in the tails appears relatively homogeneous (while in fact exponentially diminishing) and the speed of movement constant. In the absence of region C (21–97) the newly nucleated filaments may be protected less efficiently from the capping proteins and the nucleation process cannot overcome the loss of free barbed ends due to capping. When the number of uncapped filaments implicated in the elongation process is not sufficient to propel the bacteria, the speed of movement decreases. During this lag time ActA continues the slow process of nucleation of new filaments until accumulation of a critical number of newly nucleated filaments (phase‐dense spots) is reached. Elongation of the newly synthesized filaments can then propel the bacteria at high speed, but unprotected filaments begin to be capped again and a new cycle begins.
If this model is correct, increasing the concentration of capping proteins should reproduce the discontinuous tail phenotype in bacteria expressing ActA wild‐type. We are currently testing this hypothesis. In fact, in rare cases (our unpublished observations), bacteria expressing wild‐type ActA also make discontinuous actin comet tails in Xenopus extracts, suggesting that the concentration of actin‐capping proteins may vary in some areas of the extracts, or that spontaneous mutations may arise in the C region.
In the future, the identification of host cell factors that interact with or modify the specific regions of ActA identified here should help to reconstitute a minimal in vitro system allowing actin‐based motility.
Materials and methods
Construction of the ActA(N)–LacZ chimeric protein
The ω domain of β‐galactosidase was amplified from pMC1871 (Pharmacia Biotech) using the following primers: (5′‐GGGGTACCGAACGCGACCGCA‐3′) (703), (5′‐GGCCCGGGCGCTCAGCTGGAA‐3′) (1011) (the 5′‐end primer contains a KpnI restriction site and the 3′‐end a XmaI site). The number in parentheses for each primer corresponds to the position of the last amino acid encoded by the PCR fragment in the mature protein.
After digestion with KpnI–XmaI, the PCR product was cloned in pActA‐ΔPC (Lasa et al., 1995) digested with the same restriction enzymes.
Construction of deletions in the N‐terminal domain of ActA
We generated a series of N‐terminal fragments of actA by PCR. We used the same 5′‐end primer (209 bp upstream from the first nucleotide of the initiating codon) containing an EcoRI restriction site (5′‐GGAATTCAGTTGGGGTTAACT‐3′). For the 3′‐end, the following primers containing a KpnI restriction site were used:
All these PCR fragments were digested with EcoRI–KpnI and cloned into pActA‐ΔN (Lasa et al., 1995) digested with the same enzymes. The number in parentheses corresponds to the position of the last amino acid encoded by the PCR fragment in the mature ActA, as described by Kocks et al. (1992).
For the construct ActA‐Δ21–97, a fragment containing part of the N‐terminal domain was produced using the following primers:
5′‐end (5′‐GGGGTACCAGCTATACAAGTG‐3′) (97);
3′‐end (5′‐GGCCCGGGAAGCATTTACCTCTTC‐3′) (234).
This fragment digested with KpnI–XmaI was cloned into pActA‐ΔN digested with the same restriction enzymes.
Deletion of five amino acids to produce construct ActA‐Δ116–122 was done using an oligonucleotide‐directed mutagenesis system (Sculptor™ IVM system, Amersham). A fragment containing the N‐terminal domain of ActA from pActA‐ΔPC (Lasa et al., 1995) was digested with EcoRI–KpnI and cloned into appropriately digested M13mp19. Mutagenesis was performed using the oligonucleotide 5′‐CCGATGATGCTATG‐GCAATTTCCGCTGCGCTA‐3′, which deleted the sequence coding for the five amino acids. Following mutagenesis, the sequence of the entire fragment was determined. Recombinants harbouring the desired mutations were digested with EcoRI–KpnI and cloned into pActA‐ΔN.
The resulting plasmids were electroporated into L.monocytogenes LO28 ΔactA (Gouin et al., 1995) as described by Lasa et al. (1995) obtaining the following strains: ActA‐Δ158–231 (BUG1368), ActAΔ126–231 (BUG1370), ActA‐Δ114–231 (BUG1372), ActA‐Δ97–231 (BUG1374), ActA‐Δ21–97 (BUG1374) and ActA‐Δ116–122 (BUG1378). The constructs were verified by sequence determination using the T7 sequencing kit (Pharmacia Biotech). Stable expression of the proteins was analysed by immunoblotting as described (Lasa et al., 1995).
Cell culture, infection and double fluorescence labelling of bacteria and associated F‐actin
The mouse macrophage‐like cell line J774 (ECACC 85011428), the Vero cell line (ATCC CRL1587) and the Potoroo kidney epithelial PtK2 cells (ATCC CCL56) were cultured as described by Dramsi et al. (1993), Kocks et al. (1993) and Gouin et al. (1995), respectively.
Immunofluorescence of L.monocytogenes‐infected cells was performed as described by Lasa et al. (1995). Preparations were observed in a laser scanning confocal microscope (Wild Leitz) or an epifluorescence microscope (Optiphot‐2, Nikon).
Polyclonal antibodies against β‐galactosidase recognizing the ω domain (Celada et al., 1974) were generously provided by A.Ullmannn. Polyclonal antibodies recognizing VASP (Reinhard et al., 1992) were generously provided by U.Walter. The ActA‐specific affinity‐purified polyclonal antibodies were produced either against a synthetic peptide comprising the first 18 N‐terminal residues of mature ActA (A18K) (Friederich et al., 1995), against a synthetic peptide comprising part of the first proline‐rich repeat (Y21T) (Kocks et al., 1993) or against a peptide P102–123 (CERRHPGLPSDSAAEIKKRRKAI) covering residues 102–123 from ActA (this work). Affinity purification of the antibodies was done as described by Friederich et al. (1995).
Analysis of motility of L.monocytogenes derivatives
For analysis of the motility of L.monocytogenes derivatives expressing ActA variants in cytoplasmic extracts of Xenopus laevis eggs and inhibition of actin tail formation with antibodies, interphasic extracts of Xenopus eggs were prepared as described (Marchand et al., 1995). Bacterial strains were prepared and motility assays were preformed as described (Lasa et al., 1995) with the following modifications: extracts mixed with bacteria were incubated for 1 h at room temperature, compressed between a slide and a coverslip and incubated for a further 2 h. The preparations were observed with a Zeiss Axiovert 135 microscope equipped with CCD 5405 camera (Hamamatsu) which allowed simultaneous visualization of bacteria by phase contrast, and actin tails by fluorescence. Images were recorded in real time or in time‐lapse on a video tape recorder. An ARGUS‐20 image processor (Hamamatsu) attached to a Macintosh computer was used to analyse the movement of bacteria.
For inhibition of actin comet tail formation, 10 μl of bacteria solution (109 bacteria/ml) resuspended in Xenopus buffer (100 mM KCl, 1 mM MgCl2, 0.1 mM CaCl2, 50 mM sucrose in 10 mM HEPES pH 7.7) were incubated for 1 h with 0.1, 1 and 5 μg/ml of affinity‐purified antibodies diluted in the same buffer. 1 μl of this solution was then mixed with 10 μl of Xenopus extract supplemented with 1 mM Mg‐ATP, 30 mM creatine phosphate and 3 μM rhodamine‐labelled G‐actin (cytoskeleton). After 1 h incubation, 2 μl were compressed between a slide and a 22×22 mm coverslip, and incubated for a further 2 h before observation by microscopy.
Peptides were synthesized on a model 431 A peptide synthesizer (Applied Biosystems Inc., Foster City, CA) on p‐hydroxymethyl resin using the F‐moc chemistry procedure following the manufacturer's instructions. After cleavage, the peptide–resin was conjugated with trifluoroacetic acid (TFA), the peptide was precipitated with t‐methyl‐butyl ether and collected by centrifugation at 2000 g. The residue was dried for 2 h in a Speedvac and then redissolved in water and adjusted to pH 8 with 4 M NaOH. Peptides were desalted over a Sephadex G25 column in water. The eluate was monitored by UV absorbance at 254 nm. Peptide fractions were collected, lyophilized and further purified by preparative HPLC on a C4 reverse‐phase column. The purified peptides were desalted once more as described above and their concentrations determined after alkaline hydrolysis followed by ninhydrin staining (Hirs, 1967).
Actin polymerization and binding assays
Rabbit skeletal muscle actin was prepared following the procedure of Spudich and Watt (1971) and isolated as Ca–G‐actin by chromatography over Sephadex G‐200 in G‐buffer (5 mM Tris–HCl pH 7.7, 0.1 mM CaCl2, 0.2 mM ATP, 0.2 mM dithiothreitol). Actin was labelled with N‐(1‐pyrenyl) iodoacetamide as described (Kouyama and Michashi, 1981). Actin polymerization in the presence of peptides was measured by the increase in fluorescence of pyrenyl–actin (Brenner and Korn, 1983). Peptides, dissolved in G‐buffer at the indicated concentrations, were mixed at room temperature with 12 μM of G‐actin of which 10% was pyrenyl‐labelled. Polymerization was initiated 10 min later by the addition of salt (final concentration 150 mM KCl and 1 mM MgCl2). The increase in fluorescence was monitored at room temperature as a function of time with a SFM25 fluorimeter (Kontron Instruments, Zurich) using excitation and emission wavelengths of 305 nm and 388 nm respectively. Fluorescence could be measured continuously for at least 20 min without noticeable bleaching of the sample.
Actin polymerization was also studied by a sedimentation assay. Peptides were incubated with G‐actin (24 μM) at room temperature in G‐buffer, salt was added. The F‐actin formed after 30 min of incubation was sedimented for 15 min at 30 p.s.i. in a Beckman airfuge (Beckman Instruments). The supernatant was discarded and the pellet washed twice with F‐buffer and resuspended in 60 μl of gel sample buffer and an aliquot analysed by gel electrophoresis on mini‐slab gels.
Binding of the peptides to preformed actin filaments was carried out with F‐actin formed in the presence of 150 mM KCl and 1 mM MgCl2 and the addition of the indicated amounts of peptide 15 min after actin polymerization. The pellets collected after centrifugation were analysed by SDS–PAGE.
We wish to thank T.Pugsley and K.Ireton for critical reading of the manuscript, R.Hellio for help in the confocal analysis and the Service de Photographie de l'Institut Pasteur for photograph assistance. We thank M.F.Carlier, D.Pantaloni and J.B.Marchand for stimulating discussions, M.Bornens and A.Paoletti for help with Xenopus egg extracts, T.Hyman for helpful advice with the video microscopy equipment, A.Ullmannn for advice and anti‐β‐galactosidase antibodies, and U.Walter for anti‐VASP antibodies. I.L. is a recipient of a post‐doctoral fellowship from the EEC (Training and Mobility of Researchers). This study received financial support from the ARC (CT1164), the Ministeère de la Recherche (ACC‐SV6), EEC (BMH4 CT 96‐0659), EEC (CHRX CT 94‐0430), VLAB‐COT 035, GOA (91/96‐3) to J.V., from the Flemish Community and the Pasteur Institute.
- Copyright © 1997 European Molecular Biology Organization