The Legionella Effector Protein DrrA AMPylates the Membrane Traffic Regulator Rab1b

See allHide authors and affiliations

Science  20 Aug 2010:
Vol. 329, Issue 5994, pp. 946-949
DOI: 10.1126/science.1192276


In the course of Legionnaires’ disease, the bacterium Legionella pneumophila affects the intracellular vesicular trafficking of infected eukaryotic cells by recruiting the small guanosine triphosphatase (GTPase) Rab1 to the cytosolic face of the Legionella-containing vacuole. In order to accomplish this, the Legionella protein DrrA contains a specific guanine nucleotide exchange activity for Rab1 activation that exchanges guanosine triphosphate (GTP) for guanosine diphosphate on Rab1. We found that the amino-terminal domain of DrrA possesses adenosine monophosphorylation (AMPylation) activity toward the switch II region of Rab1b, leading to posttranslational covalent modification of tyrosine 77. AMPylation of switch II by DrrA restricts the access of GTPase activating proteins, thereby rendering Rab1b constitutively active.

Legionella pneumophila is a human pathogen that creates a replication vacuole in host cells (1). It interferes with normal intracellular vesicular transport by several mechanisms, one of which involves recruitment of the small guanosine triphosphatase (GTPase) Rab1 to the Legionella-containing vacuole (LCV) (2, 3). The bacterial protein DrrA (also known as SidM), which acts as a guanosine triphosphate–guanosine diphosphate (GTP-GDP) exchange factor (GEF) for the GTPase, plays a major role in this process. It has been suggested that DrrA also has GDF (RabGDI displacement factor) activity to displace GDP-bound Rab1 from guanine nucleotide dissociation inhibitor (GDI) (4, 5) but this activity merely represents a passive one arising from the GEF activity (6). In addition to the GEF domain, DrrA has at least two further domains (Fig. 1A), one of which is a C-terminal lipid phosphatidylinositol-4-phosphate binding domain (P4M) responsible for membrane attachment (7) and an N-terminal region of hitherto unknown function.

Fig. 1

DrrA9-218 is reminiscent of nucleotidyl transferases. (A) Scheme of the domain architecture of DrrA, indicating the relative positions of the ATase, GEF, and P4M domains and the crystallized fragment DrrA9-218. (B and C) A comparison between the crystal structures of the DrrA9-218 fragment (B) and GS-ATase (9) (residues 604 to 815) (C). The catalytically important amino acids of the G-X11-D-X-D motif of GS-ATase and their counterparts in DrrA9-218 are indicated as green spheres. N and C indicate N and C terminus, respectively.

It has been reported that the N-terminal region has cytotoxic properties (4), but the amino acid sequence does not suggest its function. We crystallized the N-terminal region by using a tryptic fragment encompassing amino acids 9 to 218 (DrrA9-218) and determined its structure at a resolution of 2.1 Å (Fig. 1B and table S1) (8). We observed structural similarities to nucleotidyl transferases, the closest agreement being between DrrA9-218 and the C-terminal domain of glutamine synthetase adenylyl transferase (GS-ATase) (Fig. 1C) (9). Superimposition with GS-ATase revealed that DrrA shares the catalytically important sequence motif G-X11-D-X-D (10), which occupies the same position and relative orientation as in GS-ATase.

We thus speculated that the N-terminal region might have adenosine monophosphate (AMP)–transferring activity toward Rab1b, the substrate of its GEF domain, which we confirmed on mixing full-length DrrA (DrrAfl) and Rab1b with adenosine triphosphate (ATP) (Fig. 2A). This activity, which has recently been termed AMPylation (11), appeared to require the full region N-terminal to the GEF domain (DrrA340-533) of DrrA because the fragment DrrA1-339 was catalytically active, whereas shorter constructs were unable to AMPylate Rab1b in vitro (fig. S1 and table S2). Mutating the aspartates of the G-X11-D-X-D motif to alanines (D110/112A) eliminated the enzymatic activity (fig. S1), consistent with the importance of this motif for the catalysis (12). In addition to ATP, GTP was also a substrate for the nucleotide transfer reaction, leading to GMPylation, although ATP appears to be the preferred substrate (fig. S2). Additionally, Rab1b bound to the nonhydrolyzable GTP analog GppNHp was the preferred substrate, because the apparent specific enzymatic activity of DrrA was 6.02 ± 0.23 μmolesproduct/mgenzyme per min for Rab1b:GppNHp versus 0.022 ± 0.001 μmolesproduct/mgenzyme per min for Rab1b:GDP (fig. S3).

Fig. 2

DrrA specifically AMPylates Y77 of Rab1b. (A) AMPylation of Rab1b by DrrAfl was monitored by ESI-MS (elctrospray ionization mass spectrometry). AMPylation shifts the molecular weight (m) of Rab1b (top) by 329 dalton (Da) (bottom). (B) Test for AMPylation of Rab1b point mutants Y77→F77 (Y77F) and Y78F with DrrAfl by ESI-MS, demonstrating the importance of Y77 for AMPylation. (C) Structures of Rab1b3-174-AMP:GppNHp (left) and Rab3a (right). GppNHp is shown as lines; tyrosine and tyrosine-AMP, as balls and sticks, Phe45 (Rab1b) or Phe59 (Rab3a), as spheres.

In order to determine the site of modification on Rab1b, we performed tandem mass spectrometric analysis on a total tryptic digest of preparatively AMPylated Rab1b (fig. S4). The tryptic peptide 72TITSSYYR79 (10) from switch II revealed a mass shift consistent with AMPylation, and the site of modification was mapped to Tyr77 (Y77). This was confirmed by using the phenylalanine mutants Rab1b Y77F and Rab1b Y78F: Rab1b Y78F was AMPylated by DrrA1-339, whereas Rab1b Y77F was not (Fig. 2B). To examine the influence of Y77 modification on Rab1b, we determined the crystal structure of Rab1b-AMP in complex with GppNHp (Fig. 2C, fig. S5, and table S1) and compared it to the structure of the closely related Rab3A (Fig. 2C) (13). AMPylation of Y77 did not significantly alter the overall conformation of the Rab protein. The modified Y77 essentially adopted the same orientation as in the unmodified state with the AMP group close to the surface of the protein.

Because Y77 is mostly conserved in the Rab family, we investigated the AMPylation activity of DrrA toward a variety of recombinant Rab proteins purified from Escherichia coli (fig. S6). Rab1b, Rab3a, Rab4b, Rab6a, Rab8a, Rab11a, Rab13, Rab14, and Rab37 served as DrrAfl substrates, whereas Rab5a, Rab7a, Rab9a, Rab22a, Rab23, Rab27a, Rab31, Rab32, and Rab38 did not. AMPylation of other Rab proteins may not play an important role because the membrane localization of DrrA will restrict its AMPylation activity to Rab1b, whose localization is in turn governed by DrrA GEF activity.

The switch II region of Rab1b (in which Y77 is centrally located) is involved in binding to effector and regulator proteins, such as GTPase-activating proteins [GAPs: Gyp1, LepB, and TBC1D20 (1416)], the tethering factor TrappI (17), and the GEF domain of DrrA (DrrA340-533) (6) (Fig. 3A). To understand the consequences of Rab AMPylation, we performed in vitro interaction experiments with Rab1b-interacting molecules. The GEF reaction of Rab1b-AMP with DrrA340-533 was only moderately affected compared with Rab1b (Fig. 3B), whereas the stimulation of GTP hydrolysis by the GAPs Gyp1-46, TBC1D201-362, and the Legionella protein LepB was strongly inhibited (Fig. 3C). Thus, AMPylation appears to prolong the lifetime of the GTP state of Rab1b by restricting the access of GAPs to switch II and blocking GAP-stimulated GTP hydrolysis. This is also true for GMPylation of Rab1b (fig. S7).

Fig. 3

Biochemical consequences of Rab1b AMPylation. (A) The positions of DrrA340-533 and TrappI1 interacting residues and the presumable binding residues for Gyp1 and TBC1D20 [inferred from the Rab33:Gyp1-complex structure (14)] are indicated in green in the surface representation of Rab1b-AMP:GppNHp. The AMPylated tyrosine of Rab1b-AMP is shown in stick representation. (B) Nucleotide exchange catalyzed by DrrA340-533 (1 μM) on 50 nM Rab1b (solid circles) or Rab1b-AMP monitored (open circles) by fluorescence decrease upon release of mantdGDP, fitted to single exponentials. (C) GAP-stimulated GTP hydrolysis of 40 μM Rab1b:GTP (solid symbols) and Rab1b-AMP:GTP (open symbols) using 100 nM TBC1D201-362 (circles), Gyp1 (squares), or LepB (triangles). The GTP content of individual samples from each time point was quantified by reversed-phase high-performance liquid chromatography. (D) Influence of Rab1b-AMPylation on effector binding analyzed by gel filtration. Mixtures of GppNHp-bound Rab1b or Rab1b-AMP with LidA or MICAL-3 were separated by gel filtration (red line), and complex formation was checked by denaturing polyacrylamide gel electrophoresis of individual fractions (black lines) (Rab1b and Rab1b-AMP, solid circles; LidA and MICAL-3, open circles). OD, optical density; a.u., arbitrary units.

Rab proteins interact with effector proteins in the GTP state, and we asked whether this is still possible for AMPylated Rab1b. The binding of GppNHp-loaded Rab1b and Rab1b-AMP was tested with the mammalian effector MICAL-3 (18) and the Legionella effector LidA (5) by using a gel filtration–based assay (Fig. 3D). LidA was able to interact with both Rab1b and Rab1b-AMP, whereas MICAL-3 binding was inhibited by AMPylation. Thus, AMPylation of Rab1b appears to disrupt binding to the mammalian effector protein MICAL-3 but not to the Rab1b effector LidA provided by Legionella.

The N-terminal domain of DrrA causes cytotoxicity in mammalian cells (4). In order to test whether this effect is linked to the AMPylation activity of DrrA, we expressed various DrrA constructs N-terminally tagged to the fluorescent reporter enhanced green fluorescent protein (EGFP) in COS-7 cells. DrrA constructs containing the active AMPylation domain (DrrAfl, DrrA1-533, or DrrA1-339) caused severe morphological changes of the cells characterized by cell rounding and shrinkage (Fig. 4, A, B, and E). However, more extensively truncated constructs or amino acid point mutations affecting the AMPylation activity of DrrA (DrrA1-300 and DrrA1-339 D110/112A) did not have any effect on cell morphology (Fig. 4C, D). Flow cytometry analysis demonstrated a substantial increase in cell death (by about 60%) when the AMPylation activity was present (DrrAfl, DrrA1-533, and DrrA1-339), whereas AMPylation-inactive mutants (e.g., DrrA1-339 D110/112A) were not significantly toxic (Fig. 4F). The cotransfection of DrrA1-339 constructs with Rab1b Y77F or Rab1b showed a tendency to overcome the cytotoxic effect (fig. S8), indicating that AMPylation of Rab1b could be the cause of cytotoxicity. Thus, the AMPylation activity of DrrA affects cell homeostasis of mammalian cells and causes cytotoxicity.

Fig. 4

Cytotoxicity of DrrA is linked to its AMPylation activity. (A to E) EGFP fluorescence microscopy images of COS-7 cells transfected with indicated DrrA constructs showing cell rounding and growth defects caused by AMPylation activity of DrrA (scale bars, 50 μm). (F) Flow cytometric quantification of COS-7 cytotoxicity resulting from DrrA AMPylation activity. The y axis indicates the increase in cytotoxicity (as determined by propidium iodide staining) of DrrA-expressing cells (EGFP-positive) in relation to the EGFP-vector control. The x axis labels correspond to (A) to (E). Data are mean ± SD from N = 4 independent experiments.

Rab1 is effectively removed from the LCV about 4 hours after infection (15). This removal probably requires extraction by RabGDI, and, because the interaction with RabGDI is only possible in the inactive GDP-bound state (20), Rab1 must first be deactivated to be released from the LCV. The probable GAP for this deactivation is the Legionella protein LepB because of its specificity for Rab1 and the coincidence of LepB-positive with Rab1/DrrA-negative LCVs (15). However, because LepB is unable to stimulate GTP hydrolysis on AMPylated Rab1b, it seems likely that Rab1b-AMP needs to be de-AMPylated first, immediately raising the question of an unidentified host– or Legionella-derived de-AMPylase.

AMPylation of Rab1b by DrrA should be mainly restricted to the LCV membrane, because DrrA specifically localizes here via its interaction with phosophatidylinositol-4-phosphate immediately after infection (7). On the other hand, the localization of LepB to the LCV is delayed with respect to DrrA and Rab1 (15), which could mean that LepB initially acts on Rab1 elsewhere but not at the LCV. This might inhibit intrinsic Rab1-dependent vesicular transport to the Golgi apparatus.

AMPylation of proteins is a mechanism for modulating their enzymatic activity (19, 20) or their interaction with target molecules, in the case of Rho proteins (11, 21). In the latter case, a different, unrelated AMPylation domain (the fic domain) (22) is responsible for the modification. The unexpected finding of DrrA-mediated AMPylation of the switch II region of Rab1b further adds to our knowledge of this emerging field as a mechanism in signal transduction.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Tables S1 and S2

References and Notes

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; D, Asp; F, Phe; G, Gly; I, Ile; R, Arg; S, Ser; T, Thr; X, any amino acid; and Y, Tyr.
  3. We thank P. Janning for guidance in tandem mass spectrometry, N. Bleimling for invaluable technical assistance, M. Sulc for help in flow cytometry, the staff of Beamline X10SA at the Paul Scherrer Institute (Villingen, Switzerland), the x-ray communities at the Max Planck Institute (MPI) Dortmund and the MPI Heidelberg (Germany), and M. Machner and A. Barnekow for LidA and MICAL-3 plasmids, respectively. R.S.G. thanks the German research foundation (SFB 642). M.P.M. is grateful to the International Max Planck Research School in Chemical Biology (Dortmund) for financial support. Coordinates of DrrA9-218 and Rab1b-AMP have been deposited in the Protein Data Bank (Brookhaven National Laboratory) with accession numbers 3NKU and 3NKV, respectively.
View Abstract

Navigate This Article