Ankyrin Repeat Proteins Comprise a Diverse Family of Bacterial Type IV Effectors

See allHide authors and affiliations

Science  20 Jun 2008:
Vol. 320, Issue 5883, pp. 1651-1654
DOI: 10.1126/science.1158160


Specialized secretion systems are used by many bacteria to deliver effector proteins into host cells that can either mimic or disrupt the function of eukaryotic factors. We found that the intracellular pathogens Legionella pneumophila and Coxiella burnetii use a type IV secretion system to deliver into eukaryotic cells a large number of different bacterial proteins containing ankyrin repeat homology domains called Anks. The L. pneumophila AnkX protein prevented microtubule-dependent vesicular transport to interfere with fusion of the L. pneumophila-containing vacuole with late endosomes after infection of macrophages, which demonstrates that Ank proteins have effector functions important for bacterial infection of eukaryotic host cells.

Type IV secretion systems (TFSSs) are molecular machines used by Gram-negative bacteria for protein transfer into recipient cells (1). Many bacterial pathogens and endosymbionts use TFSSs to regulate host processes important for survival and replication (2), and several of these organisms have a large number of genes encoding proteins with multiple ankyrin repeat homology domains (ARHDs) (37). Infrequently encountered in bacterial proteins but common in eukaryotic proteins, ARHDs form molecular scaffolds that mediate protein-protein interactions (8). An Anaplasma phagocytolyticum protein containing multiple ARHDs called AnkA (9) and several ARHD proteins in strains of Wolbachia (10, 11) have been proposed to be delivered into host cells by aTFSS (12); however, whether Ank proteins are bona fide TFSS effectors has not been established.

Legionella pneumophila and Coxiella burnetii are both intracellular pathogens that encode several proteins containing ARHDs and a TFSS called Dot/Icm (57). To test whether ARHD proteins are TFSS substrates, we measured host cell translocation of four L. pneumophila Ank proteins fused to a calmodulin-dependent adenylate cyclase reporter (Cya), using the L. pneumophila effector RalF as a positive control (13, 14). These four Ank proteins were delivered into mammalian cells as indicated by a >10-fold increase in adenosine 3′,5′-monophosphate (cAMP) following infection (Fig. 1A). No cAMP increase was observed when the Cya-Ank proteins were produced in the L. pneumophila ΔdotA mutant lacking a functional TFSS, which indicates that the Dot/Icm system is required for Ank protein delivery into host cells. Thirteen different C. burnetii proteins with ARHDs were tested for translocation with the Cya assay. Genetic manipulation of the obligate intracellular pathogen C. burnetii is not currently possible, but the C. burnetii and L. pneumophila Dot/Icm systems are functionally similar (15, 16), which suggests that L. pneumophila will deliver C. burnetii Dot/Icm substrates into host cells. C. burnetii AnkA, AnkB, AnkF, and AnkG fusion proteins were efficiently translocated into host cells by a process requiring the L. pneumophila Dot/Icm system (Fig. 1B). The AnkE, AnkH, AnkI, and AnkM proteins were delivered less efficiently (Fig. 1B). Specific Ank-encoding RNA transcripts were expressed during C. burnetii infection (Fig. 1C), which suggests that protein products should be available for delivery into host cells by the C. burnetii Dot/Icm system.

Fig. 1.

Type IV translocation of bacterial Ank proteins. (A and B) Translocation of Cya-Ank fusion proteins derived from L. pneumophila (A) or C. burnetii (B) into a stable Chinese hamster ovary (CHO) cell line producing the FcγRII receptor (CHO FcγRII cells) was determined. Results are means ± SD from three independent wells. (C) Reverse transcription polymerase chain reaction (RT-PCR) analysis of RNA isolated from C. burnetii–infected CHO FcγRII cells. The reverse transcription reaction was omitted in the control lanes to test for DNA contamination. (D) Identification of AnkF and EF-Ts in the indicated fractions isolated from human foreskin fibroblasts that were uninfected (–) or infected (+) with C. burnetii and either incubated with chloramphenicol (CM) or left untreated. Similar results were obtained from two independent experiments. (E) Identification of secreted AnkF in soluble fractions from human foreskin fibroblasts infected with C. burnetii that were incubated with chloramphenicol either in the presence or absence of the proteasome inhibitor MG132 as indicated for either 1 hour or 2 hours. Controls include untreated cells infected with C. burnetii, uninfected cells, and actin immunoblots to measure protein loading. Similar results were obtained from two independent experiments.

An antibody generated against AnkF confirmed that this protein was secreted during C. burnetii infection of mammalian cells (Fig. 1D), whereas the C. burnetii translation factor EF-Ts was not and remained associated with bacteria in the pellet. The amount of secreted AnkF diminished over time after treatment with the bacterial protein synthesis inhibitor chloramphenicol, which did not measurably affect levels of AnkF associated with bacterial cells (Fig. 1D). The short half-life of secreted AnkF revealed by chloramphenicol treatment could be the result of proteasome-mediated degradation if this protein was translocated into the host cytosol. Indeed, inhibition of the host proteasome with MG132 prevented degradation of secreted AnkF in the chloramphenicol-treated cells (Fig. 1E), which indicated that secreted AnkF was located in the host cytosol. Thus, multiple C. burnetii Ank proteins are substrates of the Dot/Icm system, and AnkF is delivered into host cells during C. burnetii infection.

The four translocated L. pneumophila Anks were fused to green fluorescent protein (GFP) and ectopically produced in mammalian cells to address whether these proteins have effector functions. Each of the four Ank proteins showed a different pattern of subcellular localization in mammalian cells, which suggested that these proteins have different targets and distinct functions (Fig. 2A). Extensive fragmentation of the Golgi apparatus was observed in cells producing GFP-AnkX, which correlated with a significant defect in the release of secreted alkaline phosphatase (AP) into the tissue culture medium (Fig. 2, B and D). GFP-AnkX deletion derivatives revealed that ARHDs and the amino-terminal region of AnkX were both required for disrupting secretory transport (Fig. 2, C and D).

Fig. 2.

The L. pneumophila AnkX protein is an effector of membrane transport. (A) Micrographs indicate the differential localization of L. pneumophila Ank proteins fused to GFP in CHO FcγRII cells. (B) Giantin staining (red) in CHO FcγRII cells producing the indicated L. pneumophila GFP-Ank proteins (green) reveals that GFP-AnkX production results in disruption of the Golgi apparatus. (C) Amino acid positions (below) and the location of four predicted ARHDs (A1to A4) are indicated in the schematic representation of the AnkX protein. (D) Secretion of alkaline phosphatase (AP) by CHO FcγRII cells producing GFP or the AnkX proteins indicated was measured. The AP index represents the ratio of secreted AP to the amount of AP that remained cell-associated. Data are the average ± SD from three independent samples. Asterisks indicate the samples where the secretion of AP was significantly lower than from GFP-producing cells in an unpaired Student's t test.

Effectors translocated by the Dot/Icm system have predicted roles in blocking fusion of the L. pneumophila-containing vacuole with late endosomes and in promoting vacuole fusion with endoplasmic reticulum (ER)–derived vesicles (17). To determine whether AnkX is an effector that controls membrane transport, we further investigated the process of AnkX-mediated dispersion of the Golgi apparatus in mammalian cells by time-lapse microscopy. AnkX protein was microinjected into cells producing the resident Golgi enzyme N-acetylgalactosaminyltransferase II fused to yellow fluorescent protein (GalNAc-T2-YFP). Compared with control cells treated with brefeldin A (BFA) (movie S1), which stimulated retrograde transport of Golgi membranes and cargo back to the ER, microinjected AnkX protein resulted in dispersal of GalNAc-T2-YFP into peripheral vesicles formed during Golgi fragmentation, with no apparent delivery back to the ER (Fig. 3A and movie S2). Thus, the AnkX activity more closely mimicked fragmentation of the Golgi that occurs during mitosis (movie S2, cells marked “M”), when transport of vesicles from the ER to the Golgi is blocked.

Fig. 3.

AnkX interferes with minus end–directed transport of vesicles on microtubules. (A) GalNAc-T2-YFP localization in cells treated with BFA or injected with purified AnkX. (B) CopI and CopII localization in NRK cells injected with purified AnkX (asterisk) indicate that Golgi disruption does not affect the recruitment of both coat proteins to ER exit site vesicles. (C) CHO FcγRII cells were transfected with plasmids encoding the proteins indicated. (Top) AnkX production and Arf1T31N production inhibited AP secretion to similar levels. (Bottom) AnkX production does not affect the efficiency by which L. pneumophila forms vacuoles that support replication in CHO FcγRII cells but production of Arf1T31N does. Data are the average ± SD from three independent samples. (D) Micrograph of Alexa 488–labeled transferrin in cells. The arrow indicates the location of a cell injected with AnkX. (E) GalNAc-T2-YFP (green) localization and microtubules (α-tubulin–specific antibody, blue) in cells microinjected with purified AnkX (asterisk).

The CopI and CopII coat proteins control initial events important for ER vesicle transport to the Golgi apparatus and colocalize on early secretory vesicles (18). CopI and CopII staining revealed no defects in the morphology of vesicles at ER exit sites in AnkX-injected cells (Fig. 3B), which suggests that AnkX does not affect this early stage in vesicle production from the ER. Interfering with vesicle production from ER exit sites by production of the guanosine diphosphate (GDP)–locked Arf1T31N protein blocked AP secretion and reduced L. pneumophila replication in mammalian cells (Fig. 3C). AnkX production blocked AP secretion, but did not interfere with L. pneumophila replication (Fig. 3C). Thus, AnkX disrupts secretory transport after vesicles have exited the ER.

To address whether AnkX specifically prevented the transport of early secretory vesicles, we followed endocytic transport of transferrin in cells microinjected with AnkX protein (movie S3). In neighboring cells that were not injected with AnkX, fluorescently labeled transferrin was internalized and transported to perinuclear sorting compartments (Fig. 3D). In AnkX-injected cells, however, transferrin remained in peripheral endosomes that enlarged over time (movie S3 and Fig. 3D). The effects of AnkX resembled what occurs when cells are treated with the drug nocodazole (1921), which depolymerizes microtubules. However, tubulin staining in cells microinjected with AnkX revealed no morphological defects in microtubules (Fig. 3E), which indicates that AnkX interfered with microtubule-dependent transport of vesicles without affecting the organization of the microtubule network.

Because microtubule-dependent transport is important for early to late endosome maturation (1921), trafficking of wild-type L. pneumophila was compared with that of isogenic ΔdotA and ΔankX mutants to determine whether AnkX was important for L. pneumophila evasion of endocytic maturation. The majority of vacuoles containing wild-type L. pneumophila evaded fusion with lysosomal-associated membrane protein (LAMP1)–positive late endosomes at 30 min and 2 hours post infection, whereas vacuoles containing the L. pneumophila ΔdotA mutant acquired LAMP1 within the first 30 min of infection (Fig. 4A). Most vacuoles containing the L. pneumophila ΔankX mutant were LAMP1-negative at 30 min (Fig. 4A); however, there was a significant increase at 2 hours in the number of LAMP1-positive vacuoles containing the ΔankX mutant. When remodeling of the L. pneumophila–containing vacuole by ER-derived vesicles was prevented by treating cells with BFA, there was a significant increase in the number of LAMP1-positive vacuoles containing L. pneumophila ΔankX mutants compared with the wild type at 30 min post infection (Fig. 4B). The defect in avoiding LAMP1 acquisition for vacuoles containing the ΔankX mutant was suppressed when the microtubule network was destabilized by nocodazole treatment (Fig. 4, C and D). By contrast, vacuoles containing the L. pneumophila ΔdotA mutant acquired LAMP1 rapidly in nocodazole-treated cells. Thus, microtubule depolymerization had a specific effect that restored the ability of the ΔankX mutant to efficiently avoid endocytic maturation, which indicates that interfering with microtubule-dependent vesicle transport mimics a virulence activity mediated by the AnkX protein during infection.

Fig. 4.

AnkX prevents microtubule-dependent endocytic maturation of vacuoles containing L. pneumophila. (A) Mouse bone marrow–derived macrophages were infected with the either wild-type L. pneumophila (blue), the ΔdotA mutant (green), or the ΔankX mutant (pink) for 0.5 hour or 2 hours as indicated, and the percentage of LAMP1-positive vacuoles for each was determined. Data are the average ± SD from three independent experiments where at least 50 different vacuoles in different cells were assessed. Asterisks represent data that showed significant difference from control (wild type, 0.5 hour) in an unpaired Student's t test. (B to D) The same assay described in (A) was conducted to examine the effects resulting from disrupting secretory transport using BFA (C), disrupting microtubules using nocodazole (C), or addition of both BFA and nocodazole (D).

The L. pneumophila and C. burnetii Ank proteins represent a large and diverse family of proteins containing ARHDs that we determined to be translocated into eukaryotic cells by pathogen-associated TFSSs. As seen in eukaryotic proteins, the prediction is that these bacterial proteins will have diverse cellular functions that are achieved by targeting different host factors through ARHD-dependent interactions. Finding that the L. pneumophila AnkX protein interferes with microtubule-dependent vesicular transport demonstrates that Ank proteins have effector functions. In addition to preventing endocytic maturation of the pathogen-occupied vacuole, AnkX interfered with transport of ER-derived vesicles generated at peripheral ER exit sites, an activity that will facilitate recruitment of early secretory vesicles to the vacuole and aid in construction of the specialized organelle that supports L. pneumophila replication (fig. S1).

Supporting Online Material

Materials and Methods

Fig. S1

Tables S1 and S2


Movies S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article