Because of experimental limitations imposed by the obligate intracellular nature of rickettsiae and the lack of workable genetic systems, little is known about specific virulence determinants utilized by these organisms. However, there is a cursory understanding of some rickettsia-host interactions. Studies employing the typhus group rickettsia,
Rickettsia prowazekii, demonstrate that internalization requires adherence to an unidentified plasma membrane receptor by viable, metabolically active organisms (
49). Uptake of rickettsiae ensues by a microfilament-dependent process (
48). Collectively, the two processes have been termed parasite-induced phagocytosis (
48). Evidence suggests that internalized rickettsiae are initially bound within a phagocytic vacuole (
12,
40). An increase in phospholipase A
2 activity occurs concomitantly with rickettsial entry and presumably facilitates rickettsial access to the host cytoplasm (
36,
49,
52). Once in the intracytoplasmic milieu, rickettsiae are free to exploit the nutrient-rich environment and interact with host structural components.
A suspected virulence mechanism unique to spotted fever group (SFG) rickettsiae, such as
Rickettsia rickettsii, is the utilization of an intracellular actin-based motility (ABM) system to promote direct cell-to-cell spread (
12,
15,
16). This mechanism of pathogenesis is also exploited by the facultative intracellular bacteria
Listeria monocytogenes and
Shigella flexneri (
9) as well as vaccinia virus (
50). Using the propulsive force supplied by parasite-directed polymerization of host cell actin, motile bacteria move into plasma membrane-bound protrusions that can be subsequently engulfed by neighboring cells. Escape from the double-membrane vacuole allows infection of the newly encountered cytoplasm (
45). The ability of pathogens to spread within the host tissues, using ABM to directly transit from one cell to another, allows evasion of the host humoral immune response.
The bacterial surface proteins ActA and IcsA are necessary and sufficient for ABM by
Listeria (
8,
18) and
Shigella (
2,
21), respectively. The viral protein A36R is essential for vaccinia virus ABM (
11). Rickettsial protein synthesis is required for ABM, but the identity of the necessary protein(s) is unknown (
16). A number of host cytoskeletal proteins are also necessary or suspected modulators of bacterial ABM and protrusion formation. These were initially identified by immunolocalization studies and include proteins involved in F (filamentous)-actin cross-linking, side binding, severing, capping, depolymerizing, and nucleating (
3,
9). Direct biochemical evidence for roles in
Listeria ABM has been established for vasodilator-stimulated phosphoprotein (VASP) (
5,
22,
25,
38), the complex of actin-related proteins 2 and 3 (Arp2-Arp3) (
22,
51), gelsolin (
19), profilin (
22,
32,
38,
42), α-actinin (
7,
22), actin depolymerization factor (also called cofilin) (
4,
22,
31), and capping protein (
22). Using a reconstitution assay, Loisel et al. (
22) have recently determined that actin, Arp2-Arp3 complex, cofilin, and capping protein are minimally required for in vitro ABM by
Listeria. The Arp2-Arp3 complex is stimulated by interaction with ActA to nucleate the production of new actin filaments (
22,
51). Capping protein stops the growth of existing actin tail filaments by binding to their fast-growing barbed ends, thus possibly funneling available G (globular)-actin to the production of new filaments at the bacterium-tail interface (
22). Cofilin stimulates the release of G-actin from the slow-growing pointed ends of filaments, thereby increasing the local concentration of G-actin for use in new filament assembly (
4,
22,
31). Listerial ABM is more efficient if profilin (a G-actin sequestering protein), VASP (a focal adhesion point protein and a ligand of profilin), and α-actinin (an F-actin cross-linking protein) are present (
22). Additional host proteins may be necessary or stimulatory in vivo. Other than actin, the cellular proteins necessary for
R. rickettsii intracellular motility are completely unknown.
DISCUSSION
In comparison to free-living facultative intracellular bacteria, our current understanding of rickettsial virulence factors that allow entry and intercellular spread in cultured cells is fragmentary. By analogy to ABM mutants of
Listeria (
8,
18) and
Shigella (
2,
21), which display attenuated virulence in animal models, it is logical to assume that recruitment and polymerization of host cell actin by SFG rickettsiae to allow intracellular motility and direct cell-to-cell spread represent a rickettsial virulence determinant. Although the general process of rickettsial ABM appears similar to that described for
Listeria and
Shigella, in this report we have demonstrated that rickettsial actin tails are compositionally and ultrastructurally different from tails produced by these bacteria.
The
R. rickettsii actin tail is frequently comprised of two or more distinct, coiled actin bundles. We suggested in a previous report (
15) that the coiling of actin tail bundles may be a manifestation of the rickettsial pole harboring multiple, fixed, asymmetrically opposed polymerization zones. The resultant asymmetry in polymerization may provide a rotational force that spins the organism as it moves through the cytosol.
Elegant electron microscopy studies, primarily by Tilney and coworkers (
43-45), have defined the ultrastructure of listerial actin tails. They demonstrated that tails of cytosolic
Listeriaare comprised of a network of short (∼0.2-μm) cross-linked actin filaments having their fast-growing barbed ends oriented towards the bacterial surface (
43-45). Actin tails of protrusion-bound
Listeria are compositionally and ultrastructurally different from those associated with cytosolic bacteria. For example, they lack α-actinin, an F-actin cross-linking protein that is observed in tails of cytosolic
Listeria and may be required for maintenance of the bundled structure in this environment. They also have a lower percentage of short filaments and exhibit long (>1-μm) filaments. Tails of protrusion-bound
Listeria gain ezrin, a membrane protein responsible for forming cytoskeleton-membrane associations, that has been postulated to stabilize the longer tail filaments (
33).
In our original description of rickettsia-induced actin polymerization we employed a TEM fixation procedure that used ruthenium red as an F-actin stabilizer (
16). These micrographs provided the first glimpse of
R. rickettsii actin tails by electron microscopy and suggested that the tail consisted of long actin filaments. In this report we confirm our early observations by using quick-fix fixation (
46) and myosin S1 subfragment decoration to demonstrate that the
R. rickettsii actin tail consists of long (>1-μm), parallel-arranged filaments that appear to be minimally cross-linked. Like listerial tail filaments, rickettsial tail filaments are juxtaposed with the fast-growing barbed end oriented towards the rickettsial surface, suggesting that G-actin incorporation occurs at the rickettsial surface. Moreover, tail filaments of protrusion-bound rickettsiae display a more condensed architecture, as is observed for listerial tails (
33). The short actin tail of
R. typhi implies that the organism is deficient in actin recruitment and polymerization. Of interest would be to determine the polarity and length of
R. typhi actin tail filaments and whether tail production confers intracellular motility.
The absence of a cytoskeletal stalk associated with one pole of
R. prowazekii by dry cleave SEM is consistent with the absence of actin tails by TEM and phalloidin staining (
16). The lack of ABM by
R. prowazekii correlates with a reduced capacity to form plaques on cell monolayers.
R. prowazekiialso grows to high numbers in individual cells with little cytopathic effect (
37).
R. rickettsii are toxic to cells in small numbers, with cytopathological changes indicative of oxidative stress (for example, dilation of the rough endoplasmic reticulum) observed 2 days after infection (
35). Elevated cellular levels of reactive oxygen species, such as superoxide anion and hydrogen peroxide, are observed concomitant with
R. rickettsii infection (
34). Movement of SFG rickettsiae by ABM likely results in frequent collisions with the plasma membrane, where the action of rickettsial phospholipase(s) may produce by-products capable of activating membrane-bound NADPH oxidase (or a similar enzyme) to produce superoxide anion (
34).
Differential localization of cytoskeletal proteins was observed between
R. rickettsii and
L. monocytogenes actin tails. In contrast to
Listeria, in which VASP and profilin are localized to the polymerizing end of the bacterium (
5,
38), both proteins are distributed throughout the
R. rickettsiiactin tail. VASP is a substrate of cyclic AMP- and cyclic GMP-dependent protein kinases and is generally localized to focal adhesions and areas of high actin turnover (
28). VASP is also a ligand for profilin (
27), a G-actin-sequestering protein, and binds the proline-rich motif in the central region of
Listeria ActA (
5,
25). Although not essential for
Listeria ABM, VASP and profilin accelerate the process, possibly by recruiting polymerization-competent actin monomers in the form of profilin-actin complexes to the unipolar polymerization zone of the bacterium (
20,
22,
38,
42). One possible explanation for the distribution of VASP and profilin in rickettsial tails is that
R. rickettsii secretes a protein that is incorporated into the tail and a ligand for VASP. Alternatively, VASP may localize to the
R. rickettsii actin tail via association with vinculin, which is a known ligand of VASP (
29) and also in the rickettsial tail. In this scenario vinculin may serve as an adapter protein between a rickettsial protein necessary for ABM and VASP. Whether VASP serves a functional role in accelerating rickettsial ABM by recruiting profilin-actin complexes to the polymerization zone requires further investigation.
Both
R. rickettsii and
L. monocytogenes actin tails contained filamin, but rickettsial tails lacked tropomyosin, ezrin, and paxillin. Filamin is an actin cross-linking protein found in focal adhesions and stress fibers (
24) and may play a role in bundling rickettsial tail filaments. The lack of tropomyosin, ezrin, and paxillin in rickettsial tails may explain the short length of rickettsial protrusions relative to those induced by
Listeria. These proteins all integrate with the actin cytoskeleton and play roles in the formation of cell surface extensions, such as filopodia (
23). By TEM and SEM we observed rickettsia in short protrusions (3 to 5 μm) considerably shorter than those reported for
Listeria. Depending on the cell type,
Listeria-containing protrusions can exceed 100 μm in length (
33). The short
R. rickettsiiprotrusions in Vero cells may reflect an inability of rickettsial tail filaments to interact and become stabilized by plasma membrane cytoskeletal proteins.
The results of this study are in close agreement with the recently published findings of Gouin et al. (
12) in their study of a related organism,
Rickettsia conorii, the agent of Mediterranean spotted fever. Like
R. rickettsii tails,
R. conorii actin tails were found to be comprised of long, minimally cross-linked actin filaments with the fast-growing barbed end of the filaments oriented towards the organism. In addition to lacking ezrin, they reported the absence of Arp3, cofilin, and capping protein (CapZ) in
R. conorii tails. The unique ultrastructural and compositional differences between rickettsial and listerial tails may explain the different ABM behavior and kinetics in the two genera (
12,
15). Both
R. conorii and
R. rickettsii move considerably slower than
Listeriawithin cells (
12,
15), and the actin filaments that comprise the
R. rickettsii tail are approximately three times more stable than those of listerial tails (
15). The lack of Arp3, cofilin, and capping protein may partially explain the low rate of rickettsial ABM relative to that of
Listeria. These proteins affect actin nucleation (
22,
51) or G-actin acquisition (
22,
41) and are required for listerial actin tail formation. The lower rate of rickettsial ABM and longer half-life of tail filaments may also be reflective of the tail containing fewer pointed ends for depolymerization factors to act upon. The absence of Arp3 is particularly interesting, as the Arp complex is required for nucleation of new actin filaments in actin-based movements of
Listeria (
22,
51),
Shigella (
22), and presumably vaccinia virus (
10). This leads to the possibility that rickettsiae synthesize a protein that confers actin nucleating activity rather than recruiting a nucleating factor from the host.
Our results differ from those of Gouin and coworkers (
12) in detecting vinculin within the
R. rickettsii tail. Furthermore, they did not observe coiled, distinct actin bundles in
R. conorii tails, and tail production was observed only after 24 to 36 h of infection. In a previous study we detected
R. rickettsii tail formation 30 min postinfection, with F-actin-coated rickettsiae observed as early as 15 min postinfection (
16), suggesting that some
R. rickettsiiorganisms enter host cells preloaded with a functional protein(s) necessary for ABM. This is unlike
Listeria, in which bacterial cell division and ActA processing is required for ABM (
30).
The results of this study suggest that
R. rickettsii has evolved to exploit host actin pools in a manner biologically distinct from
Listeria and
Shigella. Additional studies are needed to clearly define the roles of host cytoskeletal proteins in rickettsial ABM. Of great interest to the field is the identity of the essential rickettsial protein(s) necessary for ABM. A candidate protein may be identified upon completion of the
R. conorii genomic sequencing project currently under way by Genoscope (htpp://
www.genoscope.cns.fr/ ).