Surface Translocation by Legionella pneumophila: a Form of Sliding Motility That Is Dependent upon Type II Protein Secretion

Abstract

Legionella pneumophila exhibits surface translocation when it is grown on a buffered charcoal yeast extract (BCYE) containing 0.5 to 1.0% agar. After 7 to 22 days of incubation, spreading legionellae appear in an amorphous, lobed pattern that is most manifest at 25 to 30°C. All nine L. pneumophila strains examined displayed the phenotype. Surface translocation was also exhibited by some, but not all, other Legionella species examined. L. pneumophila mutants that were lacking flagella and/or type IV pili behaved as the wild type did when plated on low-percentage agar, indicating that the surface translocation is not swarming or twitching motility. A translucent film was visible atop the BCYE agar, advancing ahead of the spreading legionellae. Based on its abilities to disperse water droplets and to promote the spreading of heterologous bacteria, the film appeared to manipulate surface tension and, as such, acted like a surfactant. Indeed, a sample obtained from the film rapidly dispersed when it was spotted onto a plastic surface. L. pneumophila type II secretion (Lsp) mutants, but not their complemented derivatives, were defective for both surface translocation and film production. In contrast, mutants defective for type IV secretion exhibited normal surface translocation. When lsp mutants were spotted onto film produced by the wild type, they were able to spread, suggesting that type II secretion promotes the elaboration of the Legionella surfactant. Together, these data indicate that L. pneumophila exhibits a form of surface translocation that is most akin to “sliding motility” and uniquely dependent upon type II secretion.

The genus Legionella was established in 1977, following the isolation of Legionella pneumophila from patients with a form of pneumonia now known as Legionnaires' disease (33). Today, L. pneumophila is recognized as a common cause of both community-acquired and nosocomial pneumonia (84). Legionellosis occurs sporadically and in large outbreaks, with the largest outbreak occurring as recently as 2003 and encompassing 800 suspected and 449 confirmed cases (43). L. pneumophila is especially pathogenic for the elderly and the immunocompromised, large and growing segments of the population (39, 84), and recent studies have been highlighting the growing significance of travel-associated Legionnaires' disease (107). L. pneumophila is a gram-negative, gammaproteobacterium that is widespread in natural and manufactured water systems (22, 39, 93). Infection occurs after the inhalation of Legionella-contaminated water droplets originating from a wide variety of aerosol-generating devices (39). Alarmingly, outbreaks can occur following the airborne spread of L. pneumophila over distances of >10 km from cooling towers or scrubbers (86). Within its aquatic habitats, L. pneumophila survives over a wide temperature range and grows on surfaces, in biofilms, and as an intracellular parasite of protozoa (9, 39, 110). Within the mammalian lung, the organism has the ability to attach to and invade macrophages and epithelia (27, 106, 113). Among the processes that promote L. pneumophila growth in both the environment and the mammalian lung are Lsp type II protein secretion, Dot/Icm type IVB protein secretion, and Lvh type IVA protein secretion (5, 25, 31, 106). Other key surface features of L. pneumophila are polar flagella that promote swimming motility and type IV pili that help mediate adherence (53, 103, 113). In addition to exporting proteins onto its surface into the extracellular milieu, and/or into host cells, L. pneumophila also secretes a siderophore and pyomelanin pigment that help mediate iron assimilation (23). We now report that L. pneumophila has the ability to translocate or spread across an agar surface. This new form of Legionella “motility” did not require the action of flagella, pili, or type IV secretion but was associated with the export of a surfactantlike material and an intact type II secretion system.

MATERIALS AND METHODS

Bacterial strains, media, and chemicals.

L. pneumophila 130b, also known as AA100, served as our principal wild-type strain. Table 1 lists additional wild-type legionellae, as well as insertion mutants of strain 130b that were also examined for surface translocation. Legionellae were routinely grown at 37°C on buffered charcoal yeast extract (BCYE) agar or in buffered yeast extract (BYE) broth (71). Ordinarily, these two media contain an iron supplement consisting of 0.25 g of ferric pyrophosphate per liter, and the solid medium contains 1.5% agar. Growth in broth was assessed by measuring the optical density at 660 nm (OD₆₆₀) of cultures using a DU720 UV/V Spectrophotometer (Beckman Coulter, Fullerton, CA). When appropriate, BCYE agar was supplemented with 2.5 μg of gentamicin/ml or 25 μg of kanamycin/ml. Escherichia coli strain DH5α, the host for recombinant plasmids, was routinely grown in Luria-Bertani medium. Unless otherwise noted, chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

TABLE 1.

Wild-type and mutant Legionella strains used in this study

Species and strain(s)a	Description	Source or reference(s)
Wild type strains
L. pneumophila
130b (BAA-74)	Clinical isolate, serogroup 1	26, 36, 104
Philadelphia-1 (33152)	Clinical isolate, serogroup 1	17
F1271	Environmental isolate, serogroup 1	Barry Fields, Centers for Disease Control and Prevention
Togus-1 (33154)	Clinical isolate, serogroup 2	79
Bloomington-2 (33155)	Clinical isolate, serogroup 3	79
Dallas-1E (33216)	Clinical isolate, serogroup 5	44
Chicago-8 (33823)	Clinical isolate, serogroup 7	11
Concord-3 (35096)	Clinical isolate, serogroup 8	12
82A3105 (43736)	Clinical isolate, serogroup 13	72
1169-MN-H (43703)	Clinical isolate, serogroup 14	8
L. anisa (35292)	Environmental isolate	45
L. bozemanii (33217)	Clinical isolate	15
L. feeleii (35072)	Clinical isolate	51
L. hackeliae (35250)	Clinical isolate	16
L. longbeachae (33462)	Clinical isolate	78
L. micdadei
Detroit	Clinical isolate	87
31B	Clinical isolate	87
Stanford-C	Clinical isolate	87
Stanford-M	Clinical isolate	Lucy Tompkins, Stanford University
Stanford-R	Clinical isolate	87
X195	Clinical isolate	87
L. moravica (43877)	Environmental isolate	117
Mutant strains
NU347, NU348	flaA mutants of 130b	This study
BS100	pilE mutant of 130b	103, 113
NU278	pilQ mutant of 130b	103
NU349, NU350	flaA pilE mutants of 130b	This study
NU351, NU352	flaA pilQ mutants of 130b	This study
NU258	lspDE mutant of 130b	100
NU275	lspF mutant of 130b	103
NU275(pMF)	lspF complemented NU275	103
NU259	lspG mutant of 130b	100
NU272	pilD mutant of 130b	103
NU272(pMD1)	pilD complemented NU272	103
NU292	ccmB mutant of 130b	83
GG105	dotA mutant of 130b	121
GQ262	dotDCB mutant of 130b	121
GN142	dotFEP mutant of 130b	121
AA405	dotG mutant of 130b	Cary Engleberg, University of Michigan
AA474	lvhb9 mutant of 130b	Cary Engleberg, University of Michigan
AA200	proA mutant of 130b	80, 102
NU254	map mutant of 130b	2
NU324	lapA lapB mutant of 130b	102
NU318	chiA mutant of 130b	32
NU262	lipA mutant of 130b	4
NU265	lipB mutant of 130b	4
NU270	plaA mutant of 130b	41
NU268	plcA mutant of 130b	4
NU342	lpg1962 mutant of 130b	109
NU344	ppiB mutant of NU342	109
NU329	srnA mutant of 130b	101
NU287	tatB mutant of 130b	99

^aL. pneumophila wild-type strains are listed with their original designation, followed, in parentheses, by the corresponding American Type Culture Collection (ATCC) strain number if available. Except for the L. micdadei strains, all of the other Legionella species listed are denoted by their ATCC number.

Mutant constructions.

L. pneumophila DNA was isolated as described previously (71). To obtain mutants specifically lacking flagella, the flaA gene plus flanking regions was amplified from strain 130b DNA using the primers CS5 (5′-TCTAGATCGACTTGATAACCAGAACCA) and CS6 (5′-GGTACCACTAATAATATCATCAAGCCAGC). The resultant 2.1-kb fragment was then cloned into pGEM-T Easy (Promega, Madison, WI) to give pGflaA. Plasmid pGflaA was digested with MfeI, which cuts 569 bp after the flaA start codon, and after treatment with T4 DNA polymerase, was ligated to a gentamicin resistance (Gm^r) gene from pX1918GT (103) after HincII and PvuII digestion, producing pGflaAGM1. Mutated flaA was introduced into strain 130b using a modified protocol for natural transformation (103, 105, 114). Briefly, a late log-to-early stationary culture grown at 30°C was diluted to an OD₆₆₀ of 0.3, and 5 μg of pGflaAGM1 DNA/ml was added. The bacteria and DNA were incubated at 30°C with shaking at 100 rpm (C25KC incubator shaker; New Brunswick Scientific, Edison, NJ) and then, after 72 h, transformants were selected on antibiotic-containing BCYE agar. In order to isolate L. pneumophila mutants lacking both flagella and pili, the flaA mutation was introduced, as described above, into kanamycin-resistant (Km^r) pilE mutant BS100 and Km^r pilQ mutant NU278 (Table 1). Although the transformation frequencies of the pilE and pilQ mutants were 10³- to 10⁴-fold lower than that of the wild type, as expected for strains lacking competence-associated pili (114), it was possible to obtain the desired double mutants by simply scaling up our effort. Verification of the single and double mutant genotypes was carried out by PCR, using the CS5 and CS6 primers noted above.

Assay for surface translocation.

Legionella strains were grown in BYE broth at 37°C with shaking (200 rpm) to early stationary phase (OD₆₆₀ = 2.8 to 3.0), and then 10 μl of culture was spotted onto fresh (i.e., <24-h-old) BCYE plates containing 0.5 or 1.0% agar. To assess the impact of the pre-growth phase on surface translocation, the legionellae were also grown until early log (OD₆₆₀ = 0.6 to 0.9), mid-log (OD₆₆₀ = 1.4 to 1.7), late log (OD₆₆₀ = 2.5 to 2.8), or late stationary (OD₆₆₀ > 3.2) phase prior to being spotted on the BCYE agar. Inoculated low-agar plates were incubated in air at 37, 30, or 22°C, and growth was monitored for up to 22 days. At various times, digital images of the plates were obtained using a FlouroImager2000 and FlouroChem v3.04 (Alpha Innotech, San Leandro, CA). To visualize the film produced by spreading legionellae, images of the agar surface were also taken with a Canon S3 camera and 250D macro lens. Images were exported as TIFF files and labeled in Adobe Photoshop 6.0.

Assay for biosurfactant activity.

L. pneumophila 130b was grown on BCYE containing 0.5% agar at 30°C. After 13 days of incubation, the film-covered surface of the plate was flooded with 5 ml of BYE broth and then, after 5 min of further incubation at room temperature, the suspension present on the agar surface was removed by pipetting. To remove bacteria from the sample, the suspension was subjected to centrifugation (15 min, 5,000 × g), and the resulting supernatant was filtered by using a 0.22-μm-pore-size Millex-GP filter unit (Millipore, Billerica, MA). Finally, 10 μl of the filtrate was pipetted onto the cover of a polystyrene petri dish (VWR, West Chester, PA), and the collapse of the drop was monitored by eye. As a negative control, a sample was obtained from an uninoculated 0.5% agar plate using the method described above.

Assays for secreted enzymes.

Cell-free supernatants were obtained from late-log- and stationary-phase cultures of L. pneumophila grown at 30 and 37°C in BYE broth (3). The presence of secreted proteins in supernatants was confirmed by assaying for lipase, protease, and phosphatase activity (2-4). Agarase activity was measured as before (120). Briefly, supernatants were added to a sodium-phosphate buffer containing 0.2% low-melting-point agarose (US Biologicals, Swampscott, MA), and then the mixture was incubated at 40°C. After 30 min or 16 h, dinitrosalicylic acid reagent (68) was added, the samples boiled, and the absorbance read at 540 nm. β-Agarase-1 (New England Biolabs, Ipswich, MA) served as a positive control.

RESULTS

Surface translocation by L. pneumophila.

When an aliquot of an early-stationary-phase culture of L. pneumophila strain 130b was spotted onto BCYE plates containing 0.5% agar, the bacteria gradually spread out from the point of inoculation in a lobed, wavelike pattern. A time course of this phenomenon as seen at 30°C is shown in Fig. 1. The site of initial growth was visible throughout the time course with its bacterial population appearing dense and tinted yellow. The lobed, wavelike areas of growth emerged after ca. 7 days of incubation and continued to expand over the next 15 days of observation. The lobes and especially their leading edges appeared less dense than the central area of growth and had a bluish hue. The number and sizes of the lobes resulting from inoculation were variable, ranging from only one, two, or three main lobes to many large and small lobes. In some instances, the lobes emanated from one side of the inoculation point, whereas in other cases, they spread out in all directions. When the plates were incubated at 22°C, a similar situation was observed (data not shown), although it took longer to manifest, due to slower bacterial growth at the lower temperature (111). The phenomenon was also seen when the bacteria were incubated at 37°C and/or spotted onto BCYE containing 1.0% agar (data not shown) but was less dramatic than it was at 22 and 30°C and on 0.5% agar. Surface translocation was also observed when strain 130b was inoculated onto 0.5% agar BCYE that lacked its usual iron supplement (data not shown). Aliquots obtained from early-log-, mid-log-, late-log-, and late-stationary-phase cultures produced the same results as those obtained from early-stationary-phase cultures (data not shown), indicating that the growth phase of the inoculum does not dictate whether or not surface translocation will be observed. When legionellae were taken from different regions of the lobes and their spreading edges and respotted onto 0.5% agar BCYE, the same spreading phenomenon was produced (data not shown). Thus, the surface spreading of strain 130b was due to localized phenotypic changes and not the outgrowth of genetically distinct subpopulations. Surface translocation was also seen when we tested nine more wild-type strains of L. pneumophila (Table 1), including two other serogroup 1 strains, as well as representatives of serogroups 2, 3, 5, 7, 8, 13, and 14 (Fig. 2). Thus, surface translocation is conserved among L. pneumophila strains and encompasses clinical and environmental isolates.

FIG. 1.

Time course of L. pneumophila surface translocation. Wild-type strain 130b grown to early stationary phase in BYE broth was spotted onto BCYE plates containing 0.5% agar and then incubated at 30°C for 22 days. Images of the entire plate showing bacterial growth and surface spreading are presented for days 3, 7, 13, and 22. Surface translocation by strain 130b was observed on more than two dozen other occasions.

FIG. 2.

Surface translocation by different strains of L. pneumophila. Wild-type strains, identified by their original strain designation and serogroup (S) number, were grown on BCYE plates containing 0.5% agar at 30°C for 13 days. The images presented are representative of those obtained from at least two independent experiments.

Surface translocation by different species of Legionella.

The Legionella genus includes 51 species besides L. pneumophila, with nearly half of them being implicated in human disease (33). Therefore, we examined a sampling of other legionellae (Table 1) for their spreading phenotype on low-agar BCYE plates. Interestingly, there was a spectrum of phenotypes (Fig. 3A). Single strains of L. bozemanii and L. longbeachae produced a pattern that was similar to that of L. pneumophila. Those of L. feeleii and L. moravica displayed a more modest degree of surface translocation, and L. anisa and L. hackeliae appeared more modest still. Perhaps, most notable, strain 31B of L. micdadei completely lacked the surface translocation phenotype. Thus, five more L. micdadei isolates (Table 1) were tested. All were negative, even after an extended incubation of 22 days (Fig. 3B). These data indicate that many but not all species of Legionella display surface translocation.

FIG. 3.

Surface translocation by different species of Legionella. (A) L. pneumophila strain 130b, L. bozemanii 33217, L. longbeachae 33462, L. feeleii 35072, L. moravica 43877, L. anisa 35292, L. hackeliae 35250, and L. micdadei 31B were grown on BCYE plates containing 0.5% agar at 30°C for 13 days. (B) L. micdadei strains Detroit, Stanford-C, Stanford-M, and Stanford-R were grown on BCYE plates containing 0.5% agar at 30°C for 22 days. The images presented are representative of those obtained from at least three independent experiments. In other experiments, L. micdadei strain X195 gave a similar negative result (data not shown).

Lack of a role for flagella and pili in L. pneumophila surface translocation.

At a gross level, the surface translocation that we had observed was similar to that exhibited by a variety of other bacteria. In many of those other cases, movement over the agar surface is mediated by either flagella in a process known as swarming or type IV pili in a process designated twitching motility (48, 50, 59). L. pneumophila has been known for many years to possess flagella that mediate swimming motility (34, 97). It also expresses type IV pili, although no form of motility has ever been linked to this organelle in Legionella (70, 71, 97, 113). Flagella and pili are expressed by L. pneumophila grown at 25 to 37°C (71, 88, 105). Thus, as a first step to understanding the molecular basis for our results, we constructed derivatives of strain 130b that contained a mutation in the gene (flaA) encoding the flagellin subunit (52) (Table 1). Although flaA mutants were, as expected, nonmotile by wet mount, they behaved as the wild type did when tested for spreading on 0.5 and 1.0% agar BCYE plates at 22, 30, and 37°C (Fig. 4). Thus, we next tested two previously described type IV pilus mutants: BS100, mutated in the pilin subunit gene (pilE), and NU278, mutated in the pilus secretin gene (pilQ) (Table 1). Like the flagellum mutants, the pilus mutants spread normally on low-agar plates at all assay temperatures (Fig. 4). In order to explore the possibility that L. pneumophila flagella and pili serve a redundant function in surface translocation, we constructed and tested a set of double mutants that were inactivated for both flaA and pilE or pilQ (Table 1). Once again, the mutants behaved as the parent 130b did (Fig. 4). These data indicate that surface translocation by L. pneumophila does not involve flagella or pili.

FIG. 4.

Surface translocation by L. pneumophila flagellum and pilus mutants. Strain 130b (WT) and its mutant derivatives NU348 (flaA mutant), BS100 (pilE mutant), and NU350 (flaA pilE mutant) were grown on BCYE plates containing 0.5% agar at 30°C for 13 days. The images presented are representative of those obtained from at least three independent experiments. All of the other flagellum and/or pilus mutants (NU347, NU278, NU349, NU351, and NU352) also exhibited surface translocation (data not shown).

Surfactant production by L. pneumophila.

In some bacteria, flagellum- and pilus-independent surface translocation is facilitated by a secreted surfactant (48, 55, 57, 59, 65, 74, 76, 82). In a number of these instances (37, 55, 56, 65, 66), as well as in multiple cases of swarming (29, 35, 58, 63, 67, 118), secreted surfactant is manifest as a fluid layer or translucent film atop the agar surface preceding the spreading bacteria. Upon closer viewing of the 0.5 and 1.0% agar BCYE plates containing L. pneumophila 130b, we observed that there was indeed a film on the agar that advanced ahead of the lobes of bacterial growth (Fig. 5). The film was generally visible after 4 to 5 days of incubation, and by day 7 it extended as much as 2 cm from the edge of the spreading bacteria (Fig. 5A and B). After 13 days, the film often covered the entire surface of the agar plate. Occasionally, the film front had two or three visible edges that were spaced apart by 2 to 3 mm (Fig. 5C), a finding reminiscent of the “delimiting rings” attributed to surfactant in other bacteria (56). Film was present for all of the other strains of Legionella species that had showed surface spreading, as well as the flagellum and pilus mutants (data not shown). Similar to what is observed with bacterial surfactant films (65, 66), when a 10-μl droplet of water was pipetted onto the film-covered area on a plate containing strain 130b, it immediately collapsed inside the film, but when a water droplet was spotted onto an area of the agar surface that was not covered with the film, a water bead was maintained until evaporation occurred. Also similar to observations made of known surfactants (75), when heterologous, nonspreading bacteria were spotted onto a section of the agar surface covered with the film, it spread out to cover a large area in contrast to what it did when spotted elsewhere (Fig. 6). Because strain 130b lacked agarase activity (data not shown) and an examination of the L. pneumophila genome database (http://genolist.pasteur.fr/LegioList/) did not reveal an ORF similar to those encoding agarases (40, 120), it is unlikely that agar degradation contributes to the formation of the film. In order to confirm the surfactant-nature of the film, we flooded the surface of the low-agar BCYE plates containing strain 130b and its associated film with BYE broth and then proceeded to obtain a cell extract. As seen with surfactant-containing samples produced by other bacteria (73, 118), aliquots of the extract rapidly collapsed when spotted onto the lid of a plastic petri plate. In contrast, extracts obtained from the processing of uninoculated BCYE plates failed to collapse when spotted onto the same plastic surface. Taken together, these data indicate that L. pneumophila secretes a diffusible surfactant that changes the surface tension of BCYE agar.

FIG. 5.

Surface film associated with spreading L. pneumophila. Bacteria were grown on BCYE plates containing 0.5% agar at 30°C. (A) Image taken after 7 days of incubation, showing a central area of wild-type 130b growth, a single, large lobe of spreading legionellae, and the translucent film that has advanced well ahead of the spreading bacteria. (B) Diagram depiction of the image in panel A, showing spreading L. pneumophila (Lpn) and the film. (C) Image taken after 5 days of incubation, showing the early stages of surface translocation by 130b and a film front displaying three edges or rings. (C) Image taken after 5 days, comparing strain 130b (top) and its lspF mutant NU275 (bottom), which shows a central area of growth but the absence of both spreading and film. The film associated with 130b was seen on at least a dozen other occasions.

FIG. 6.

The effect of L. pneumophila on surface spreading by E. coli. Wild-type strain 130b (Lpn) was incubated on BCYE plates containing 0.5% agar at 30°C. After 13 days, when a layer of surface film was evident beyond the spreading legionellae, a 10-μl aliquot of E. coli DH5α (Ec) was spotted onto the agar surface either within the film-covered area or at a point outside of it but still equidistant from the center of L. pneumophila growth. As indicated, after 5 and 35 days of further incubation at room temperature, images were taken in order to visualize any differences in spreading from the two E. coli inocula. The images presented are representative of those obtained from three independent experiments.

Role of the L. pneumophila type II secretion system in surface translocation.

Operative in many but not all gram-negative bacteria (25), type II protein secretion (T2S) is a multistep process in which proteins destined for export are transited across the inner membrane in a Sec- or Tat-dependent manner, recognized in the periplasm, and then delivered to the T2S apparatus, whereupon a piluslike structure “pushes” proteins through a dedicated outer membrane pore (60, 69). Because of our long-standing interest in T2S by L. pneumophila (25, 71, 102), as well as the structural and evolutionary similarity between the T2S apparatus and type IV pili (91), we considered the possibility that T2S might be involved in Legionella surface translocation. To that end, we examined a panel of L. pneumophila T2S mutants for their behavior on the low-agar plates. Four different insertion mutants were tested: NU258, containing a mutation in the genes encoding the outer membrane secretin (lspD) and the inner membrane ATPase (lspE); NU275, containing a mutation in the gene for an inner membrane platform protein (lspF); NU259, inactivated in the gene encoding the major pseudopilin (lspG), and NU272, mutated in the gene encoding the pseudopilin peptidase (pilD) (Table 1). Interestingly, all of the T2S mutants were defective for the spreading phenotype on 0.5 and 1.0% agar, even after 3 weeks of incubation (Fig. 7A). Likewise, none of them produced the film (Fig. 5D and data not shown). That four different mutants were similarly impaired indicated that the mutant phenotypes were specifically due to the loss of T2S function versus second-site mutation. As confirmation, we observed that both a complemented lspF mutant and a complemented pilD mutant behaved as the wild type did (Fig. 7A). Approximately 10% of the time, the T2S mutants showed a small amount of surface spreading (see, for example, the pilD mutant in Fig. 7A). This might have been due to either some form of suppression of the pilD and lsp mutations or some T2S mutant lysis that released a stimulatory substance whose secretion had been blocked by the mutation. In past studies of L. pneumophila T2S mutants, we observed examples of both suppression and lysis (30, 110). Because T2S mutants grow comparably to the wild type at 30°C (111), their reduced ability to spread was not an indirect effect of a generalized growth defect. As further support for this conclusion, a ccmB mutant of L. pneumophila (Table 1) that grows much slower than parent 130b when plated on BCYE agar lacking iron supplementation (83) had a surface translocation pattern on low-agar BCYE that was similar to that of the wild type even though it required four more days of incubation to be manifest (data not shown). To determine whether L. pneumophila type IV secretion systems are also required for surface translocation, we tested various dot/icm and lvh mutants (Table 1) for their behavior on low-agar BCYE plates. Unlike the T2S mutants, all of these mutants displayed a normal capacity to produce spreading growth patterns and film (Fig. 7B), indicating that type IV secretion does not have a role in this L. pneumophila phenotype. Since a variety of data indicate that the Dot/Icm and Lvh systems are expressed at 25 to 37°C (24, 96, 112), these results, along with those obtained from assaying the flagellum and pilus mutants, also indicate that the surface translocation defect of the L. pneumophila T2S mutants is not a nonspecific effect of a multiprotein complex being absent from either the cell envelope or cell surface. Together, these data indicate that efficient L. pneumophila surface translocation is dependent on an intact T2S system.

FIG. 7.

Surface translocation phenotypes of L. pneumophila type II and type IV protein secretion mutants. (A) Strain 130b (WT), type II secretion mutants NU258 (lspDE mutant), NU275 (lspF mutant), and NU272 (pilD mutant), and complemented type II mutants NU258(pMF) (lspF-deficient lspF⁺) and NU272(pMD1) (pilD-deficient pilD⁺) were grown on BCYE plates containing 0.5% agar at 30°C for 13 days. (B) Strain 130b (WT), twin-arginine translocation mutant NU287 (tatB mutant), type IVB secretion mutant GG105 (dotA mutant), and type IVA secretion mutant AA474 (lvhB9 mutant). The images presented in panels A and B are representative of those obtained from at least two independent experiments, although the altered phenotype of the lspF mutant was observed on more than a dozen occasions. The lspG mutant NU259 also consistently displayed a lack of surface translocation, whereas the other dot/icm mutants tested (GQ262, GN142, and AA405) all spread normally (data not shown).

Theoretically, the inability of Lsp mutants to undergo surface translocation could be due to the loss of a factor that is normally released into the extracellular milieu, such as the surfactant, and/or to a cell-associated defect, such as a potentially absent motility organelle. Distinguishing between these possibilities, we observed that the lspF, lspG, and lspDE mutants all recovered the ability to spread when they were spotted onto the film produced by parental 130b (Fig. 8). The mutant bacteria spread in the direction away from the central area of wild-type growth, suggesting that they were moving into areas containing sufficient surfactant and/or away from areas of nutrient depletion. When wild-type 130b was spotted onto the film, it also spread away from the preexisting area of bacterial growth (data not shown). Taken together, these data strongly indicate that the inability of lsp mutants to surface translocate is due to their lack of secreted surfactant.

FIG. 8.

Surface translocation by L. pneumophila type II secretion mutants when spotted onto film produced by wild-type legionellae. Strain 130b was inoculated onto the centers of BCYE plates containing 0.5% agar. After 7 days of incubation at 30°C, when surface film was evident beyond the spreading wild type, a 10-μl aliquot from a culture of the lspF mutant NU275 (left panel) or the lspG mutant NU259 (right panel) was spotted onto the agar surface either within the film-covered area or at a point outside of it but still equidistant from the center of wild-type growth. After 2 and 6 days of further incubation at 30°C, images were taken in order to visualize the differences in spreading from the two points of mutant inoculation. The images presented are representative of those obtained from at least three independent experiments. When tested on two occasions, the lspDE mutant NU258 behaved similarly to the lspF and lspG mutants (data not shown). The ability of the lsp mutants to spread when spotted within, but not outside of, the film-covered area was also evident when the plates were examined 13 days after mutant inoculation.

The T2S system of L. pneumophila secretes at least 25 proteins, including a wide variety of degradative enzymes (2-4, 6, 30, 41, 47, 70, 100, 102, 103, 109). As a first attempt toward identifying genes more directly involved in surfactant production, we examined a panel of strain 130b mutants (Table 1) lacking known T2S effector activities. However, when mutants lacking either the ProA metalloprotease, LapA and LapB aminopeptidases, Map acid phosphatase, ChiA chitinase, LipA lipase, LipB lipase, PlaA lysophospholipase A, PlcA phospholipase C, SrnA RNase, or the Lpg1962 putative peptidyl-prolyl isomerase, no loss of surface translocation or film production was observed (Fig. 9). Since past work suggested that some L. pneumophila T2S effectors are transported across the inner membrane by Tat (32, 99), a tatB mutant (Table 1) was also examined, but it, too, was found to be similar to the wild type (Fig. 7B).

FIG. 9.

Surface translocation by mutants of L. pneumophila lacking individual secreted effectors. Mutants of strain 130b specifically lacking the indicated type II secreted effector protein were grown on BCYE plates containing 0.5% agar at 30°C for 13 days. The images presented are representative of those obtained from at least two independent experiments.

DISCUSSION

The data presented here represent the first documentation of surface translocation (i.e., motility over a surface) by bacteria belonging to the Legionella genus. Previously, surface translocation by bacteria has been given six different names: swarming, twitching, gliding, sliding, darting, and colony spreading (48, 50, 61). Surface translocation by L. pneumophila does not meet the definition of swarming or twitching, because it was not dependent upon flagella or type IV pili (48, 59). It is also not akin to flagellum- or pilus-independent gliding; that is, whereas legionellae produced a surfactant film that preceded the spreading bacteria, gliding bacteria have protein machinery in their envelopes that mediate movement in the apparent absence of surfactant, as in Flavobacterium and Mycoplasma spp., or with a slime layer that trails behind the bacteria, as in Myxococcus xanthus, exhibiting “adventurous” gliding (59). Sliding, sometimes referred to as spreading, is produced by the expansive forces in a growing culture in combination with reduced friction between cell and substrate, and over the years, there has been a strong correlation between sliding and secreted or cell-associated surfactants (48, 50). Sliding is a passive form of surface translocation that does not involve the action of any surface motor organelle (48, 59). Darting and colony spreading have been used to describe the behavior of some species of Campylobacter, Staphylococcus, and Vibrio, but there is minimal description of the phenotype and although there is no mention of a secreted surfactant, it is possible that darting and colony spreading are not dissimilar from sliding (14, 50, 61, 90). Thus, given our current data, L. pneumophila surface translocation is most similar to sliding, whereby a Legionella secreted surfactant promotes passive movement across the agar surface. Other bacteria that are known to exhibit sliding include Acinetobacter calcoaceticus, Alcaligenes odorans, Bacillus anthracis, Bacillus cereus, Bacillus subtilis, Mycobacterium abscessus, Mycobacterium avium, Mycobacterium smegmatis, Pseudomonas aeruginosa, Serratia marcescens, Stenotrophomonas maltophilia, and Vibrio cholerae (1, 18, 37, 38, 48, 50, 54, 55, 57, 74, 75, 82, 95). Thus, L. pneumophila joins a diverse group of sliding bacteria that includes gram-positive, gram-negative, and acid-fast bacteria, organisms that inhabit the environment, as well as the human host, where they can cause disease, and bacteria that grow within and outside of host cells.

Our data also represent the first evidence for surfactant production by L. pneumophila and other Legionella species. Surfactants are a structurally diverse group of amphipathic molecules that act to reduce the surface and interfacial tension, such as would exist on agar plates, among other places (81, 119). Surfactants are produced by many types of bacteria and fungi, and the structures for these “biosurfactants” include fatty acids, neutral lipids, phospholipids, glycolipids, glycopeptidolipids, “flavolipids,” lipopeptides, and lipoproteins (1, 10, 13, 81, 85, 119). As noted above, bacterial surfactants very often facilitate surface translocation, be it as a component of sliding, swarming, or adventurous gliding (10, 19, 29, 35, 46, 55, 57, 58, 62, 74, 115). Other processes that are promoted by surfactants and likely enhance bacterial survival in environmental niches include attachment to and detachment from biotic and abiotic surfaces, biofilm formation, antimicrobial activities, alterations in phospholipid-containing structures, solubilization, uptake, and utilization of hydrophobic compounds, solubilization of quorum-sensing molecules, and binding of heavy metals and prevention of their toxicity (1, 20, 21, 56, 98, 108, 116). Some bacterial surfactants have also been implicated in processes that are more specific to pathogenesis. Most heavily studied in this regard are the rhamnolipids of P. aeruginosa that have been linked to both sliding and swarming (82); this biosurfactant slows mucociliary transport, inhibits the phagocytic response of macrophages and the chemotactic response of neutrophils, stimulates cytokine and glycoconjugate release by cells of the airway, solubilizes phospholipids in lung surfactant and thereby renders them accessible to cleavage by bacterial phospholipase C, and promotes bacterial infiltration of airway epithelia (7, 42, 77, 94, 108, 122). In a similar vein, the rhamnolipid of Burkholderia pseudomallei that has been associated with swarming has a cytopathic effect on macrophage and lung epithelial cell lines (49). Finally, a number of bacterial surfactants can lyse red blood cells (55, 108, 119). As both an inhabitant of aquatic environments and a pathogen of the respiratory tract, L. pneumophila is likely to use its surfactant in many of the processes previously linked to other surfactants. However, as an intracellular parasite of aquatic amoebae and lung cells, L. pneumophila might also utilize its surfactant to perform novel functions in the intracellular niche.

The data presented here represent the first documentation of a connection between bacterial sliding, surfactant, and T2S, although there is a recent report linking T2S to swarming by P. aeruginosa and other studies linking secreted protease activities to the swarming phenotypes of B. subtilis and V. vulnificus (28, 64, 89). The simplest overall hypothesis to explain our data is that T2S, directly or indirectly, promotes the elaboration of surfactant which in turn allows L. pneumophila to slide over surfaces. Within that framework, several scenarios can be envisioned. On the one hand, the surfactant itself might be secreted through the T2S system. To our knowledge, there are no reports describing the mechanism of secretion for any bacterial surfactant, but lipoproteins, one of the types of molecules that are surfactants, have been shown to be type II secreted in some bacteria (92). On the other hand, a type II-secreted enzyme might release or activate a surfactant that is surface expressed or secreted via another mechanism. Finally, T2S might influence a regulatory network that signals L. pneumophila to express surfactant and sliding. Although the genetic evidence for a connection between Legionella T2S and sliding/surfactant is currently limited to L. pneumophila, we suspect that it also holds for many of the other Legionella species given that they contain lsp genes and their culture supernatants contain enzymatic activities akin to those in L. pneumophila supernatants (103). One Legionella species that stands out as being unusual is L. micdadei. Indeed, all six strains of this species tested clearly failed to display surface translocation and surfactant production, and such a finding is compatible with previous work showing that L. micdadei strains do not express other T2S-associated phenotypes despite harboring lsp genes (110). Based on our latest results in Legionella, L. micdadei not withstanding, it is quite likely that the T2S systems of other types of gram-negative bacteria also promote surfactant secretion and surface translocation.

Finally, we can now add surface translocation and surfactant secretion to the growing list of functions that are ascribed to Legionella T2S; that is, secretion of >25 exoproteins and at least a dozen different types of enzymatic activities, optimal growth in BYE broth or on BCYE agar at temperatures below 25°C, survival in tap water at 4 to 17°C, growth in protozoa at 22 to 37°C, optimal infection of human macrophages at 37°C, and full persistence in the lungs of mice (30, 47, 70, 102, 103, 110, 111). Thus, future work will be directed toward defining the biochemical and genetic bases of the L. pneumophila surfactant, as well as investigating the role of surface translocation and surfactant in Legionella ecology and pathogenesis.