The Legionella pneumophila Metaeffector Lpg2505 (MesI) Regulates SidI-Mediated Translation Inhibition and Novel Glycosyl Hydrolase Activity

Legionella pneumophila, the etiological agent of Legionnaires’ disease, employs an arsenal of hundreds of Dot/Icm-translocated effector proteins to facilitate replication within eukaryotic phagocytes. Several effectors, called metaeffectors, function to regulate the activity of other Dot/Icm-translocated effectors during infection. The metaeffector Lpg2505 is essential for L. pneumophila intracellular replication only when its cognate effector, SidI, is present.

ABSTRACT

Legionella pneumophila, the etiological agent of Legionnaires’ disease, employs an arsenal of hundreds of Dot/Icm-translocated effector proteins to facilitate replication within eukaryotic phagocytes. Several effectors, called metaeffectors, function to regulate the activity of other Dot/Icm-translocated effectors during infection. The metaeffector Lpg2505 is essential for L. pneumophila intracellular replication only when its cognate effector, SidI, is present. SidI is a cytotoxic effector that interacts with the host translation factor eEF1A and potently inhibits eukaryotic protein translation by an unknown mechanism. Here, we evaluated the impact of Lpg2505 on SidI-mediated phenotypes and investigated the mechanism of SidI function. We determined that Lpg2505 binds with nanomolar affinity to SidI and suppresses SidI-mediated inhibition of protein translation. SidI binding to eEF1A and Lpg2505 is not mutually exclusive, and the proteins bind distinct regions of SidI. We also discovered that SidI possesses GDP-dependent glycosyl hydrolase activity and that this activity is regulated by Lpg2505. We have therefore renamed Lpg2505 MesI (metaeffector of SidI). This work reveals novel enzymatic activity for SidI and provides insight into how intracellular replication of L. pneumophila is regulated by a metaeffector.

INTRODUCTION

Legionella pneumophila is the etiological agent of Legionnaires’ disease, a severe inflammatory pneumonia that results from uncontrolled bacterial replication within alveolar macrophages. Upon phagocytosis, L. pneumophila avoids lysosomal degradation through establishment of an endoplasmic reticulum-derived compartment called the Legionella-containing vacuole (LCV) (1). For biogenesis of the LCV and acquisition of nutrients from the host cell, L. pneumophila is dependent on a massive arsenal of over 300 individual effector proteins that are translocated directly into the host cell through a Dot/Icm type IVB secretion system (T4BSS) (2). The cellular functions of the majority of effectors have yet to be elucidated, due in part to their functional redundancy within macrophages (reviewed in reference 3).

Metaeffectors, used by L. pneumophila to regulate effector function, have emerged as a common theme in L. pneumophila pathogenesis (4). Metaeffectors are defined as effectors that target and regulate the functions of other effector proteins within host cells (5). The first metaeffector described was LubX, which temporally regulates the function of its cognate effector, SidH, by hijacking host ubiquitination machinery to facilitate proteasomal degradation of SidH (5). At least 20 of the over 300 identified L. pneumophila effectors are capable of regulating the functions of other effectors (4). Two of them—SidJ and Lpg2505—are metaeffectors that are members of a small group of effectors that are individually important for L. pneumophila intracellular replication within macrophages (6–8). SidJ is a glutamylase that covalently modifies and abrogates the functions of the SidE family of effector ubiquitin ligases (9–11). Lpg2505 is an effector of unknown function that suppresses the toxicity of the effector SidI (Lpg2504) and is important for intracellular replication in macrophages only when wild-type sidI is expressed (6). We previously discovered that a loss-of-function mutation in lpg2505 attenuated L. pneumophila replication within bone marrow-derived macrophages (BMDMs); the natural host, Acanthamoeba castellanii; and a mouse model of Legionnaires’ disease (6). Restoration of L. pneumophila Δlpg2505 intracellular replication within BMDMs was observed upon either deletion of sidI or replacement of the wild-type sidI allele with a nontoxic sidI allele (sidI_R453P), suggesting that wild-type sidI was deleterious to L. pneumophila in the absence of Lpg2505 (6). Concomitantly, Lpg2505 was sufficient to suppress SidI-mediated toxicity toward the yeast Saccharomyces cerevisiae (6). Thus, SidI activity negatively impacts L. pneumophila in the absence of Lpg2505, a unique phenotype for a translocated effector.

SidI is one of seven cytotoxic L. pneumophila effector proteins that inhibit host cell protein translation (12). Like the majority of L. pneumophila effectors, sidI alone is dispensable for intracellular replication within macrophages (12), likely due to functional redundancy with other effectors. Other translation-inhibiting effectors include Lgt1 to -3, SidL, LegK4, and Lpg1489 (12–16). Lgt1 to -3 are glycosyltransferases that inhibit translation by glucosylating eukaryotic elongation factor 1A (eEF1A) at Ser-53 (17–19). LegK4 is an effector kinase that phosphorylates heat shock protein 70 (Hsp70), thereby reducing its ATPase activity and protein-refolding activities (16). SidL and Lpg1489 have been experimentally demonstrated to inhibit host protein synthesis (13, 15), but their modes of action are unknown. Like Lgt1 to -3, SidI also interacts with eEF1A; however, this interaction is insufficient for translation inhibition (12). SidI also interacts with eEF1Bγ and induces the host heat shock response (12). To date, the mechanism(s) by which SidI functions within the host cell to inhibit host protein synthesis have not been elucidated.

In this study, we aimed to discern how Lpg2505 regulates SidI and to gain insight into the molecular mechanism of SidI function. We discovered that Lpg2505 and SidI bind with nanomolar affinity and that Lpg2505 suppresses SidI-mediated translation inhibition in vitro. Furthermore, SidI interaction with Lpg2505 and eEF1A are not mutually exclusive, and the two proteins bind with distinct regions of SidI. Finally, we discovered novel GDP-mannose glycosyl hydrolase activity for SidI, which is regulated by Lpg2505. We have therefore named Lpg2505 metaeffector of SidI (MesI).

RESULTS

MesI and SidI bind with nanomolar affinity.

Lpg2505 (MesI) is sufficient to suppress SidI-mediated cytotoxicity in yeast and to promote L. pneumophila intracellular replication within mouse models of infection and A. castellanii (6). To reveal a potential mechanism for MesI-mediated regulation of SidI, we first evaluated whether MesI interacts with SidI. Since effectors function within host cells, we investigated whether SidI and MesI interact in the presence of host cell lysates. HEK 293 cells stably producing 3×FLAG-epitope tagged MesI were generated, and we initially attempted to ectopically express green fluorescent protein (GFP)-tagged sidI within these cells for coimmunoprecipitation. However, wild-type SidI could not be detected, likely due to potent translation inhibition. Since a glutathione S-transferase (GST)-tagged SidI fusion protein (GST-SidI) was used to initially identify eEF1A as a binding partner (12), we generated recombinant GST-SidI and evaluated its ability to interact with 3×FLAG-MesI within HEK 293 lysates (see Materials and Methods). We found that glutathione beads coated with recombinant GST-SidI, but not GST alone, retained 3×FLAG-MesI (Fig. 1A) (see Materials and Methods). Furthermore, using lysates from Escherichia coli, we found that GST-SidI, but not GST alone, was retained on Ni-nitrilotriacetic acid (NTA) beads that had been coated with His₆-MesI (Fig. 1B). Thus, MesI interacts with SidI in the presence or absence of mammalian cell lysates.

FIG 1.

MesI and SidI interact directly with nanomolar affinity. (A) Lysates from HEK 293 FcγRII cells stably expressing 3×FLAG-MesI were incubated with glutathione beads coated with either GST or GST-SidI, followed by Coomassie staining for total protein and Western blotting for MesI (α-FLAG). The arrowheads indicate GST and GST-SidI proteins. (B) Lysates from E. coli overexpressing either GST or GST-SidI were incubated with Ni-NTA beads coated with His₆-MesI, followed by SDS-PAGE and Coomassie staining for proteins retained on the beads. (Left) Whole-cell lysates from uninduced and induced cultures of E. coli expressing GST and GST-SidI proteins. (Right) Proteins retained on Ni-NTA beads (see Materials and Methods). (C) Chromatograms resulting from analytical-scale gel filtration separation of either MesI alone (green trace) or MesI bound to SidI (red trace). A chromatogram of known size standards is provided for reference (blue trace). (D) Binding of His₆-SidI to immobilized MesI was assessed by SPR. The reference-corrected sensorgram from a single-cycle experiment is shown in black, while the outcome of fitting to a two-state binding model is shown in red. The interaction is described by an apparent K_D of 3.1 nM, consisting of two individual steps (association rate 1 [k_on,1] = 2.7 × 10⁴ M⁻¹ s⁻¹, dissociation rate 1 [k_off,1] = 5.1 × 10⁻⁴ s⁻¹, k_on,2 = 2.3 × 10⁻³ s⁻¹, and k_off,2 = 4.5 × 10⁻⁴ s⁻¹).

To determine if SidI and MesI bind directly, we examined the abilities of these proteins to associate with one another throughout the course of sequential column chromatography procedures. SidI and MesI were coexpressed in E. coli and purified by Ni-NTA affinity chromatography (see Materials and Methods), and the eluted proteins were separated by analytical-format size exclusion chromatography. We found that SidI (∼110 kDa) and MesI (∼35 kDa) coeluted from the column as a species corresponding to a molecular weight of ∼145 kDa, as judged by comparison to a panel of known protein standards (Fig. 1C). As 145 kDa is the expected size of one molecule each of SidI and MesI, it is likely that SidI and MesI bind in a 1:1 stoichiometry. Bands corresponding to both proteins were detected in samples of column fractions that had been separated by SDS-PAGE and analyzed by Coomassie staining (see Fig. S1 in the supplemental material). Thus, SidI and MesI interact directly and appear to form a stable complex.

We subsequently used surface plasmon resonance (SPR) to investigate the affinity and kinetics of the MesI-SidI interaction. We immobilized MesI on an SPR surface using random amine chemistry and injected recombinant purified SidI at increasing concentrations. We employed a single-cycle approach due to difficulty with regenerating the MesI surface following exposure to SidI. We found that the reference-corrected sensorgram could be described fairly well by a Langmuir binding model (K_D [equilibrium dissociation constant] = 0.89 nM and χ² = 3.26) (see Fig. S2 in the supplemental material) but was better fitted to a two-state reaction model (K_D = 3.1 nM and χ² = 0.79) (Fig. 1D). A particularly noteworthy feature of the MesI-SidI interaction is its long half-life, with an estimated dissociation rate constant between 2 × 10⁻⁴ s⁻¹ and 5 × 10⁻⁴ s⁻¹. This observation explains, at least in part, the abilities of MesI and SidI to remain associated with one another throughout the copurification procedures described above. Taken together, these data indicate that MesI binds directly to SidI and forms a high-affinity complex with a K_D of ∼3 nM.

MesI suppresses SidI-mediated translation inhibition.

Based on the high-affinity interaction between SidI and MesI, we hypothesized that MesI represses SidI-mediated translation inhibition. To test this hypothesis, we quantified translation of firefly luciferase (Luc) mRNA in vitro (see Materials and Methods). The assay was used previously to demonstrate that ≥5 ng of recombinant SidI was sufficient to completely abolish translation (12). We confirmed that purified His₆-SidI significantly attenuates protein translation and additionally revealed that purified MesI alone has no impact on translation (Fig. 2A). We further determined that MesI significantly rescues translation in the presence of SidI (P < 0.01) (Fig. 2B). Rescue of protein translation by MesI was specific to SidI, since MesI had no impact on translation inhibition by Lgt1, a characterized L. pneumophila effector translation inhibitor (17, 20) (Fig. 2C). Furthermore, we confirmed previous reports that the SidI_R453P mutant protein does not impair in vitro protein translation (12) and found that MesI has no impact on this phenotype (Fig. 2D). We further revealed that a nonspecific protein (GST) was unable to suppress SidI-mediated translation inhibition (see Fig. S3 in the supplemental material). Thus, MesI is sufficient to suppress SidI-mediated translation inhibition in vitro.

FIG 2.

MesI modulates SidI-mediated translation inhibition in vitro. Translation of luciferase (Luc) mRNA was quantified using a rabbit reticulocyte lysate kit. (A to D) Translation of luciferase in the presence of 4 ng (37 fmol) His₆-SidI or 20 ng MesI (A), 4 ng (37 fmol) of His₆-SidI alone or with an equimolar amount of MesI (B), 37 fmol His₆-SidI or His₆-Lgt1 alone or with an equimolar amount of MesI (C), or 37 fmol His₆-SidI or His₆-SidI_R453P alone or with an equimolar amount of MesI (D). (E) His₆-SidI alone or His₆-SidI and MesI were added to in vitro translation reaction mixtures at the indicated times (−30 indicates preincubation of His₆-SidI and MesI for 30 min prior to the translation reaction). (F) Quantification of in vitro translation of Luc mRNA in the presence of His₆-SidI alone or MesI and His₆-SidI at increasing molar ratios (1:1, 2:1, 3:1, 6:1, and 15:1 MesI/His₆-SidI molar ratios). AU, arbitrary units. The data are representative of the results of at least two independent experiments. The data are expressed as means ± standard deviations (SD) of samples in triplicate. The asterisks denote statistical significance by t test (*, P < 0.05; **, P < 0.01; n.s., not significant).

Our SPR data demonstrate that the SidI-MesI interaction occurs rapidly. We therefore hypothesized that preformation of the SidI-MesI complex would not result in further attenuation of SidI-mediated translation inhibition. To determine whether preformation of the SidI-MesI complex would enhance MesI-mediated suppression of translation inhibition, we incubated SidI with MesI for 30 min prior to addition to the in vitro translation reaction mixture and compared this to reactions where SidI and MesI were added simultaneously (time zero). Preformation of the SidI-MesI complex did not further attenuate SidI-mediated translation inhibition (Fig. 2E) (see Materials and Methods). Furthermore, MesI was insufficient to reverse SidI-mediated translation inhibition, since addition of MesI to the reaction mixture after either 30 min or 60 min did not restore translation (Fig. 2E). Finally, we evaluated whether suppression of SidI-mediated translation inhibition by MesI was dose dependent. MesI was added in concentrations ranging from equimolar up to a 15-fold molar excess relative to SidI. An equimolar amount of MesI was sufficient to suppress SidI-mediated translation inhibition, and translation was not further enhanced by addition of up to a 15-fold molar excess of MesI (Fig. 2F). Thus, equimolar amounts of MesI are sufficient to suppresses SidI-mediated translation inhibition.

MesI does not affect the interaction between SidI and eEF1A.

A previous report demonstrated that SidI interacts directly with the transcription factor eEF1A (12). We therefore investigated whether the interactions between SidI and eEF1A or MesI are mutually exclusive. GST-MesI or GST alone was immobilized on glutathione beads and incubated with His₆-SidI that was preincubated with HEK 293T lysates. We found that eEF1A from HEK 293T lysates was retained on GST-MesI-coated beads only in the presence of SidI and that eEF1A did not impair interaction between MesI and SidI (Fig. 3A). We confirmed previous work demonstrating interaction between SidI_R453P and eEF1A (12) and further revealed that SidI_R453P also interacts with MesI (see Fig. S4 and S7A in the supplemental material) (12). These data demonstrate that MesI interacts specifically with SidI and confirm that SidI binding to eEF1A is insufficient for translation inhibition (12).

FIG 3.

MesI does not impair interaction between SidI and eEF1A. (A) GST-MesI or GST alone was immobilized on magnetic glutathione agarose beads, followed by incubation with 100 μg purified His₆-SidI alone or His₆-SidI that had been preincubated with lysates from HEK 293T cells (Lysates), as indicated. Proteins were separated by SDS-PAGE and visualized by Coomassie staining or Western blotting for eEF1A, as indicated. The arrowhead indicates His₆-SidI, and the arrows indicate GST-MesI (lanes 1 to 3 from left) or GST (lanes 4 to 6). (B) Lysates from E. coli expressing GST-SidI or GST alone were incubated with magnetic glutathione agarose beads and lysates from HEK 293T cells, followed by 10 to 1,000 μg of purified recombinant MesI (shown as increasing amounts in supernatants [Sup] from beads [an uncropped image is shown in Fig. S5B]). Proteins remaining on the beads were separated by SDS-PAGE and visualized by Coomassie staining or Western blotting for eEF1A, as indicated. GST-SidI (∼130 kDa) and MesI (∼27 kDa) are indicated with arrowheads. The data are representative of the results of at least two independent experiments.

Subsequently, we asked whether increasing the concentration of MesI would influence the SidI-eEF1A interaction. We incubated GST-SidI with eEF1A (from HEK 293T lysates) (see Materials and Methods) followed by increasing concentrations of purified recombinant MesI. Despite the MesI concentration increasing 100-fold, GST-SidI still retained eEF1A on the beads (Fig. 3B), suggesting that SidI interacts with MesI and eEF1A simultaneously. The same result was observed when MesI was incubated with GST-SidI prior to HEK 293T lysates (see Fig. S5A in the supplemental material). Thus, SidI binding to eEF1A and MesI is not mutually exclusive.

MesI and eEF1A interact with distinct regions of SidI.

Since MesI and eEF1A are capable of binding SidI simultaneously, we hypothesized that the proteins interact with SidI at distinct sites. SidI has a molecular weight of ∼110 kDa and consists of 942 amino acids. Since the structure of SidI has not been solved, we used the RaptorX Web server (21) to predict the domain structure of SidI. Based on the predicted domains (see Fig. S6 in the supplemental material), we generated SidI truncations consisting of amino acid residues 1 to 268 (SidI_N), 269 to 942 (SidI_C), and 269 to 874 (SidI_CΔ68) (Fig. 4A). To determine which of these putative SidI domains is involved in interaction with MesI and eEF1A, we immobilized GST-tagged full-length SidI, SidI_N, SidI_C, or SidI_CΔ68 or GST alone on glutathione beads, followed by incubation with lysates from HEK 293 cells stably producing 3×FLAG-MesI. Western blot analysis was used to detect 3×FLAG-MesI and eEF1A bound to SidI truncation proteins (see Materials and Methods). We confirmed that GST-SidI, but not GST alone, was capable of retaining both MesI and eEF1A on the beads. 3×FLAG-MesI was further retained on beads coated with GST-SidI_N and GST-SidI_C, but not GST-SidI_CΔ68, whereas eEF1A was retained by GST-SidI_C and GST-SidI_CΔ68 (Fig. 4B). As a negative control, we found that GST-Lgt1 retained neither eEF1A nor 3×FLAG-MesI with the same efficiency as GST-SidI and GST-SidI_R453P (see Fig. S7A). To control for the potential influence of 3×FLAG-MesI on eEF1A binding to SidI fragments, we repeated the experiment using HEK 293 cells that did not produce 3×FLAG-MesI. We observed the same pattern of interactions between eEF1A and the C-terminal region of SidI in the absence of 3×FLAG-MesI (see Fig. S7B). Based on these data, we conclude that MesI interacts with SidI at regions within amino acid residues 1 to 268 and 874 to 942, with apparently higher affinity to the latter, whereas eEF1A interaction with SidI is dependent on amino acid residues 269 to 874. Together, these data suggest MesI interacts with two nonoverlapping regions of SidI that are distinct from the site of SidI interaction with eEF1A.

FIG 4.

MesI and eEF1A interact with distinct regions of SidI. (A) Schematic representation of SidI truncation proteins. (B) Lysates from E. coli expressing GST-SidI constructs were incubated with glutathione agarose beads, followed by washing and incubation with lysates from HEK 293 FcγRII cells stably expressing 3×FLAG-MesI. Proteins were separated by SDS-PAGE and visualized by Coomassie staining (GST-SidI) or Western blotting. The arrowheads indicate GST fusion proteins. The data are representative of the results of three independent experiments.

Loss of MesI does not impact SidI translocation into host cells.

The majority of L. pneumophila effector proteins rely on a C-terminal translocation signal for Dot/Icm-mediated translocation into host cells (22, 23). Based on the observed interaction between MesI and the C-terminal 68 amino acid residues of SidI, we evaluated whether Dot/Icm-mediated translocation of SidI requires MesI. Dot/Icm-dependent export of SidI fused to Bordetella pertussis adenylate cyclase (CyaA) was quantified (22). When CyaA fusion proteins reach the cytosol of eukaryotic cells, they catalyze formation of cAMP, which can be quantified using a cAMP-specific enzyme-linked immunosorbent assay (ELISA). Based on potent toxicity associated with wild-type SidI, a nontoxic sidI allele (R453P; SidI_RP) was used for these experiments (6, 12). Expression of CyaA-SidI_R453P was confirmed using Western blot analysis (Fig. 5A). Subsequently, we found that cAMP production within host cells was significantly increased following infection with wild-type and ΔmesI strains, but not the Dot/Icm-deficient ΔdotA strains, of L. pneumophila producing CyaA-SidI_R453P (Fig. 5B). Translocation of CyaA-RalF was used as a positive control for Dot/Icm-dependent translocation and similarly confirmed by Western blot analysis (22) (Fig. 5A and B). Together, these data demonstrate that SidI translocation into host cells is not impacted by loss of MesI.

FIG 5.

MesI is not required for Dot/Icm translocation of SidI into host cells. (A) Western blot showing production of CyaA-RalF (∼76 kDa) and CyaA-SidI_RP (∼150 kDa) by the indicated L. pneumophila strains. WT, wild type. (B) Quantification of cAMP extracted from CHO FcγRII cells infected with the indicated L. pneumophila strains. The asterisks denote statistical significance by t test (**, P < 0.01; n.s., not significant). The data are representative of the results of two independent experiments. The data are expressed as means and SD of samples in triplicate.

Binding to eEF1A is insufficient for SidI-mediated translation inhibition.

Subsequently, we investigated the ability of truncated SidI proteins to influence protein translation. At concentrations equivalent to that of SidI, neither SidI_N, SidI_C, nor SidI_CΔ68 was sufficient to inhibit protein translation in vitro (Fig. 6A). Based on the interaction between SidI_CΔ68 and eEF1A, we also quantified translation in the presence of increasing concentrations of His₆-SidI_CΔ68. We found that His₆-SidI_CΔ68 significantly decreased translation at concentrations up to 25-fold greater than that of SidI (Fig. 6B), suggesting that the truncation retains some activity; however, SidI_CΔ68-mediated translation inhibition is modest in comparison to that with an equal amount of SidI despite the ability to interact with eEF1A (Fig. 4B; see Fig. S7B). These data further confirm that interaction with eEF1A is insufficient for SidI-mediated translation inhibition (12).

FIG 6.

Full-length SidI is required for translation inhibition. Translation of luciferase mRNA was quantified using a rabbit reticulocyte lysate kit. Shown is translation of luciferase mRNA in the presence of 4 ng His₆-SidI, His₆-SidI_N, His₆-SidI_C, or His₆-SidI_CΔ68 (A) or 4 ng of His₆-SidI alone or increasing amounts of His₆-SidI_CΔ68 (4 ng, 50 ng, or 100 ng) (B). The asterisks denote statistical significance by t test (***, P < 0.001; **, P < 0.01; n.s., not significant). The data are representative of the results of at least two independent experiments. The data are expressed as means ± SD of samples in triplicate.

SidI is a GDP-dependent glycosyl hydrolase.

Like many other bacterial effectors, the primary amino acid sequence of SidI does not have obviously conserved motifs. Therefore, to gain insight into the putative function of SidI, we used a variety of computational methods to predict the structure and function of SidI. Though very low primary sequence identity was present in the templates used, the HHPred Web server (24), the Phyre2 Web server (25), the Raptor-X Web server (21), and the I-TASSER Web server (26–28) all produced models of various lengths but with the same fold for overlapping regions. A DALI search (29) for structural homologs against each of these models revealed that amino acid residues 368 to 868 of SidI have predicted structural homology to multiple bacterial and eukaryotic glycosyltransferases with GT-B folds, including PimB, a GDP-mannose-dependent mannosyltransferase (see Table S1 and Fig. S8 in the supplemental material). Orthogonally, the Ginzu domain parser on the Robetta server also predicted the presence of a glycosyltransferase domain (30). Based on the consensus between the orthogonal computational approaches, we hypothesized that SidI possesses GDP-dependent glycosyltransferase activity and that GDP-mannose could be a substrate. To test this hypothesis, we utilized a functional luminescence-based assay to quantify the cleavage of several nucleotide-sugars including GDP-mannose, GDP-fucose, UDP-glucose, and UDP-GlcNac (see Materials and Methods). As a positive control for nucleotide-sugar cleavage, we included the glucosyltransferase Lgt1, which cleaves UDP-glucose (31). MesI was included as a control, since it does not have predicted enzymatic activity (see Materials and Methods). We found a highly significant increase in liberation of GDP from GDP-mannose in the presence of purified recombinant His₆-SidI (Fig. 7A). We did not observe similar levels of nucleotide liberation catalyzed by SidI for GDP-fucose, UDP-glucose, or UDP-GlcNAc (Fig. 7A to D), suggesting that GDP-mannose is the preferred substrate for SidI. Interestingly, MesI displayed mild activity against GDP-fucose but no activity against any other nucleotide-sugars tested (Fig. 7A to D). As predicted, Lgt1 cleaved UDP-glucose, but no other nucleotide sugars were cleaved to a similar extent by either SidI or MesI (Fig. 7A to D). We further revealed that SidI_R453P was unable to cleave GDP-mannose to the level of wild-type SidI (Fig. 7E), suggesting that the ability to cleave GDP-mannose is connected to inhibition of protein translation and toxicity phenotypes observed for SidI (6, 12). Together, these data demonstrate that SidI possesses glycosyl hydrolase activity and that GDP-mannose is the preferred substrate.

FIG 7.

SidI possesses glycosyl hydrolase activity. (A to D) Quantification of GDP or UDP liberated from GDP-mannose (A), GDP-fucose (B), UDP-glucose (C), or UDP-GlcNAc (D) in the presence or absence of 5 μg purified His₆-SidI, His₆-Lgt1, or MesI, as indicated. (E) Five micrograms of His₆-SidI or His₆-SidI_R453P was incubated with GDP-mannose. (F and G) Five micrograms of His₆-SidI and/or the molar equivalent of MesI incubated with GDP-mannose (F) or increasing molar amounts of MesI (1:1, 2:1, or 5:1 molar ratio of MesI to His₆-SidI) (G). The asterisks denote statistical significance by t test (***, P < 0.001; **, P < 0.01; *, P < 0.05). The data are representative of the results of at least two independent experiments. The data are expressed as means ± SD of samples in triplicate.

The entire predicted GDP-dependent glycosyltransferase domain of SidI is encoded in amino acids 368 to 868, which are encompassed within the SidI_CΔ68 truncation (Fig. 4A; see Fig. S8). To determine whether SidI_CΔ68 alone is sufficient to cleave GDP-mannose, we incubated His₆-SidI_CΔ68 with GDP-mannose and quantified the generation of free GDP. Full-length SidI and SidI_N, which is not predicted to possess glycosyl hydrolase activity, were included as controls. We also included His₆-SidI_C, which includes the predicted enzymatic domain. Thus, we evaluated the abilities of SidI_N, SidI_C, and SidI_CΔ68 to cleave GDP-mannose compared to that of wild-type SidI. All the SidI proteins examined generated a luminescence signal upon incubation with GDP-mannose; however, the luminescence signals obtained by incubation of SidI_N, SidI_C, and SidI_CΔ68 in the absence of GDP-mannose did not differ significantly from when they were incubated with GDP-mannose (see Fig. S9 in the supplemental material). However, a very significant increase in luminescence was observed when full-length SidI was incubated with GDP-mannose compared to SidI alone (see Fig. S9). These data demonstrate that full-length SidI is required for glycosyl hydrolase activity.

SidI enzymatic activity is dampened by MesI.

Based on the ability of MesI to suppress SidI-mediated translation inhibition, we hypothesized that MesI could regulate SidI-mediated GDP-dependent glycosyl hydrolase activity. Thus, SidI-mediated GDP-mannose cleavage in the presence of equimolar amounts of MesI was quantified as described above. We found that MesI was sufficient to significantly decrease—but not inhibit—SidI GDP-dependent glycosyl hydrolase activity (P < 0.01) (Fig. 7F). We further evaluated whether addition of a molar excess of MesI would further decrease SidI glycosyl hydrolase activity; however, addition of up to a 5-fold molar excess of MesI to SidI was not sufficient to significantly inhibit SidI activity compared to equimolar quantities (Fig. 7G). Furthermore, attenuation of SidI enzymatic activity is specific, since addition of equal quantities of an unrelated protein (GST) was unable to attenuate GDP-mannose cleavage by SidI (see Fig. S10 in the supplemental material). Thus, MesI dampens SidI glycosyl hydrolase activity in vitro.

DISCUSSION

Legionella pneumophila is an opportunistic, intracellular pathogen that exploits host cell machinery to promote intracellular replication through the translocation of effector proteins. SidI is one of seven cytotoxic effectors that inhibit eukaryotic protein translation (SidI, SidL, Lgt1 to -3, LegK4, and Lpg1489) (12, 13, 15, 18). Within this family of effectors, SidI is distinct, as its regulation by the metaeffector Lpg2505 (MesI) is essential for intracellular replication (6). Our study aimed to explore the role of MesI in the regulation of SidI function and elucidate potential mechanisms behind SidI-mediated toxicity. Here, we demonstrate that MesI binds directly to SidI with high affinity and modulates both SidI-mediated translation inhibition and novel GDP-dependent glycosyl hydrolase activity, which has not been previously observed for a bacterial effector. We further demonstrate that SidI is able to simultaneously bind MesI and eEF1A, which interact with distinct regions of SidI. This work is the first to define the enzymatic activity of SidI and the contribution of MesI to SidI-mediated phenotypes.

L. pneumophila is reliant on host cell-derived amino acids for intracellular replication (32). It can therefore be speculated that L. pneumophila utilizes multiple effector proteins to halt host protein synthesis at the elongation step in order to facilitate proteasomal degradation of partially folded polypeptides. The translation-inhibiting effectors characterized to date have diverse functions. The effector Lgt1 was found to target the host elongation factor eEF1A, and homology searches led to the discovery of the orthologous effectors Lgt2 and -3 (19). The Lgt effectors function by glycosylation of eEF1A at Ser-53, which results in blockade of host protein synthesis (33). The Lgts also glycosylate a eukaryotic release factor-related protein (eRF3) and Hsp70 subfamily B suppressor 1 (Hbs1) (34, 35). Based on the function of Lgt1 to -3, attempts were made to identify SidI-mediated posttranslational modification of eEF1A; however, no modifications were discovered (12). Although SidI was previously assumed to lack glycosyltransferase activity (12, 36), we discovered that SidI is indeed able to cleave GDP-mannose, suggesting that SidI is a likely a mannosyltransferase. SidI also had low levels of activity against three other activated sugars; however, this is likely a consequence of relaxed sugar specificity by glycosyltransferases in vitro (37). Further investigation is required to reveal the target of SidI’s glycosyltransferase activity in vivo and the molecular mechanism by which SidI inhibits translation.

SidI induces the host stress response through formation of a complex between heat shock factor 1 (HSF1) and eEF1A (12). The formation of this complex in conjunction with a noncoding RNA promotes the binding of HSF1 to the heat shock element (HSE), inducing host cell Hsp70 expression (12, 38). Notably, despite the upregulation of transcription, Hsp70 protein levels did not differ between infected and uninfected cells (12), likely due to robust translation inhibition by L. pneumophila effectors. A more recent study revealed that the eEF1A1 isoform facilitates expression of heat shock genes independently of its role in protein translation (39). Since SidI, but not Lgt1, induced expression of Hsp70 in addition to eEF1A-mediated activation of the HSF1 transcription factor, it could be hypothesized that SidI modulates eEF1A to specifically amplify heat shock genes, which could enhance the survival of L. pneumophila-infected cells. Moreover, the effector LegK4 phosphorylates Hsp70, which results in loss of protein translation. It is tempting to envision a scenario whereby SidI and LegK4 function in concert to modulate host Hsp70 activity. The importance for modulation of heat shock proteins has also been demonstrated by previous observations that Hsp90 is essential for L. pneumophila replication in A. castellanii (40). Further investigation is required to define the influence of MesI on SidI-mediated modulation of the heat shock response.

Regulation of effector function by metaeffectors is an essential component of the L. pneumophila virulence strategy. The first identified L. pneumophila metaeffector, LubX, functions as an E3 ubiquitin ligase that is translocated into the host cytosol at late stages of infection and catalyzes ubiquitination of the effector protein SidH, targeting it for proteasomal degradation (5). L. pneumophila ΔlubX mutants replicate similarly to wild-type bacteria, suggesting that LubX is not individually required for intracellular replication in the host cells examined (5). Like MesI, the metaeffector SidJ directly contributes to intracellular replication by suppressing the toxicity of the SidE family of effectors (SdeA, SidE, SdeB, and SdeC) (7, 41). However, unlike MesI, overexpression of SidJ is toxic to eukaryotic cells (41, 42). The SidE family effectors also contribute directly to intracellular replication through a novel mechanism of phosphoribosyl-ubiquitin conjugation to host substrates (43–45). Unlike the SidE effectors, loss of sidI or the sidI-mesI operon has no effect on L. pneumophila intracellular growth (6), suggesting that SidI functions redundantly within macrophages. However, in the absence of MesI, SidI is deleterious to L. pneumophila intracellular replication (6), a phenotype not observed for any other effector.

Several modes of metaeffector function have been described. A large-scale screen by Urbanus and colleagues led to the discovery of 17 putative L. pneumophila metaeffectors that function to suppress the toxicity of other effectors (4). They uncovered several mechanisms by which metaeffectors regulate their cognate effectors through direct targeting and abrogation of effector enzymatic activity in vitro (LegL1, SidP, and LupA). Similarly, we found that MesI dampens SidI activity in vitro; however, residual SidI activity is retained, suggesting that MesI functions to fine-tune or direct SidI activity. Interestingly, we did not observe substantial interaction between Lgt1 and eEF1A despite evidence that the Lgt1 enzyme is active (see Fig. S7A, S2C, and S7C), which is likely due to release of eEF1A following its glucosylation. Although we have not identified the bona fide substrate of SidI, our observation that SidI interacts with both MesI and eEF1A stably and simultaneously in distinct regions suggests that SidI may not target eEF1A but instead use interaction with eEF1A to gain access to a different host substrate. Moreover, the almost complete restoration of translation by MesI in the presence of SidI compared to the relatively modest decrease in SidI-mediated glycosyl hydrolase caused by MesI suggests that MesI does not function to suppress SidI enzymatic activity. Future investigation will reveal the target of SidI activity, the role of MesI in SidI function, and how these phenotypes contribute to L. pneumophila intracellular replication.

We discovered that MesI interacts with SidI in two nonoverlapping N- and C-terminal regions. This observation is unique, since no previously characterized metaeffectors and very few protein-protein interactions are facilitated through multiple nonoverlapping regions. For example, the metaeffector SidJ binds and polyglutamylates single sites on SdeA, SdeB, SdeC, and SidE (9), and the metaeffector LubX binds SidH through its U box 2 domain only (5). However, interaction of a protein with more than one nonoverlapping region of a binding partner is not unprecedented. Using affinity chromatography, Liang and Hai demonstrated that human activating transcription factor 4 (hATF4) interacts with four nonoverlapping domains within CREB (cAMP response element-binding protein)-binding protein (CBP). Moreover, hATF4 interacted with individual CBP domains with differing affinities (46). Similarly, MesI interacted with the N- and C-terminal regions of SidI with lower efficiency than the full-length SidI protein (Fig. 4B). Based on our evidence that the SidI-MesI interaction has a 1:1 stoichiometry (Fig. 1C), it can be speculated that both the N- and C-terminal regions of SidI contribute to MesI binding. Nevertheless, since our SPR data were better described by a two-state model as opposed to a simple Langmuir model, we suspect that MesI binding to SidI may somehow alter the conformation of SidI. We are currently conducting further studies to help define these structural changes and are additionally optimistic that our efforts to solve the structure of the MesI-SidI complex will shed light on the molecular basis for this unique interaction.

Interestingly, one of MesI’s binding sites on SidI falls at the extreme carboxy terminus of SidI. The majority of Dot/Icm-translocated effectors are dependent on a C-terminal translocation signal for secretion into host cells (23). We revealed that loss of mesI did not impact SidI translocation into host cells (Fig. 5). However, that does not preclude the possibility of MesI interfering with SidI translocation, since sidI, but not mesI, was overexpressed. Further evaluation at more physiologically relevant expression levels is required to determine if MesI impairs Dot/Icm-dependent translocation of SidI into host cells.

Through structural-homology prediction and biochemical analysis, we revealed that SidI likely possesses GDP-dependent glycosyltransferase activity. Although we have not directly demonstrated the transfer of mannose to a substrate protein, it is unlikely that SidI functions only as a nucleotide-sugar hydrolase in the context of infection. Moreover, free-nucleotide liberation from activated sugar donors is a prerequisite for transfer of glycans to substrate molecules and is an established method to identify glycosyltransferase activity without knowing the acceptor substrate (47).

Several L. pneumophila effectors function as glycosyltransferases, including the translation inhibitors Lgt1 to -3; however, none have been shown to utilize GDP-conjugated sugars. In fact, bacterial GDP-dependent glycosyltransferases seem to be involved primarily in biogenesis of cell surface structures (48–50). SidI has predicted structural homology to mycobacterial phosphatidylinositol mannosides (PIMs), which are GT-B glycosyltransferases (49, 51). To our knowledge, no other translocated bacterial effector has been demonstrated to have GDP-dependent glycosyltransferase activity. The primary site of protein glycosylation in eukaryotic cells is the Golgi apparatus; however, nucleotide sugars, including GDP-mannose, are synthesized in the cytoplasm of eukaryotic cells prior to transport into the Golgi apparatus (52, 53). Thus, SidI likely hijacks cytoplasmic GDP-mannose prior to its translocation into the Golgi apparatus. Based on the relatively low abundance of individual effectors in L. pneumophila-infected cells, it is unlikely that SidI function influences glycoprotein production by host cells. Thus, we predict SidI is a GDP-dependent glycosyltransferase, and this activity is likely critical for SidI-mediated translation inhibition and cytotoxicity. Further biochemical and structure-function analysis will reveal the detailed molecular mechanism by which SidI functions, its bona fide substrate in host cells, and how MesI regulates its activity to promote L. pneumophila intracellular replication.

In this study, we have revealed a direct high-affinity interaction between SidI and its metaeffector, MesI; defined a novel enzymatic activity for SidI; and uncovered the fact that MesI can modulate SidI function in vitro. Our work has provided a foundation for future biochemical and cell biological studies to reveal how SidI functions to modulate host cell processes. The severe virulence defect resulting from expression of sidI in the absence of mesI underlines the importance of defining the molecular mechanism of SidI-MesI function. Moreover, the lack of precedent for a bacterial GDP-dependent glycosyltransferase effector protein suggests that SidI possesses novel enzymatic activity, which must be regulated by MesI to promote L. pneumophila intracellular replication. Our future work will focus on uncovering this mechanism in order to gain critical insight into translation inhibition and metaeffector function.

MATERIALS AND METHODS

Bacterial strains, cell culture, growth conditions, and reagents.

E. coli strains used for cloning (Top10; Invitrogen) and protein expression [BL21(DE3); a gift from Craig Roy, Yale University] were maintained in Luria-Bertani (LB) medium supplemented with antibiotics as appropriate for plasmid selection (50 μg ml⁻¹ kanamycin [GoldBio], 100 μg ml⁻¹ ampicillin [GoldBio], and 25 μg ml⁻¹ chloramphenicol [GoldBio]). L. pneumophila Philadelphia-1 Lp01 (54) and ΔdotA (55) strains were cultured on supplemented charcoal–N-(2-acetamido)-2-aminoethanesulfonic acid (ACES)-buffered yeast extract (CYE) and grown at 37°C, as described previously (56, 57). CYE was supplemented with 10 μg ml⁻¹ chloramphenicol for plasmid maintenance as required. Protein expression in E. coli and L. pneumophila was induced with 1 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG) (GoldBio).

All mammalian cells were grown at 37°C-5% CO₂ for up to 30 passages. HEK 293 FcγRII (a gift from Craig Roy, Yale University [58]) and HEK 293T (ATCC) cells were cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (HIFBS) (U.S. origin; Gibco). CHO FcγRII cells (59) (a gift from Craig Roy) were cultured in MEMα (Gibco) supplemented with 10% HIFBS.

Unless otherwise specified, all chemicals were obtained from Millipore Sigma (St. Louis, MO). The oligonucleotide primers used in this study are listed in Table 1.

TABLE 1.

Oligonucleotide primers used in this study

Name	Sequence (5′→3′)a
MesISalI-F	ATTGTCGACAATGATAAAAGGAAAACTTATGCCC
MesIBglII-F	TGGAGATCTATGATAAAAGGAAAACTTATGC
MesIPstI-R	GGCTGCAGTTATAAAATAATTGGTCGAG
SidIBamHI-F2	ATTGGATCCTTATGACTAAAATATACTTATTAACTGC
SidIBamHI-F	ATTGGATCCATGACTAAAATATACTTATTAACTGC
SidINotI-R	ATTGCGGCCGCTCAAAATACCAGTATCGATTCTTTAAG
SidICBamHI-F	ATTGGATCCATGAATTTTTATGATTTTGATGG
SidICΔ68NotI-R	ATTGCGGCCGCTCATATATTCTCTAATAAATGATC
SidINNotI-R	ATTGCGGCCGCTCATCTGAAACTTTTATCGTGCTC
SidIPstI-R	ATTCTGCAGTCAAAATACCAGTATCGATTC
Lgt1Sal1-F	ATTGTCGACAATGAAAGCAAGAAGGGATCAAC
Lgt1_6p1Sal1-F	ATTGTCGACTCATGAAAGCAAGAAGGGATCAAC
Lgt1Not1-R	ATTGCGGCCGCCTACCCTACTGAAGGCAACCAAC

^aRestriction endonuclease cleavage sites are shown in boldface.

Molecular cloning, plasmid construction, and generation of Legionella strains.

L. pneumophila Lp01 genomic DNA (gDNA) was isolated using the Illustra genomicPREP DNA isolation spin kit (GE Healthcare) and used as a template for cloning sidI and mesI into the indicated plasmid vectors. For recombinant-protein production, sidI was amplified using the primer pair SidIBamHI-F/SidINotI-F and cloned as a BamHI/NotI fragment into pGEX-6P-1 (GE Healthcare) and pT7HMT (60). mesI was amplified using primer pairs MesIBglII-F/MesINotI-R and cloned as a BglII/NotI fragment into BamHI/NotI-digested pGEX-6P-1 and pcDNA4T/O-3×FLAG (56). For cloning into pT7HMT, mesI was amplified using the MesISalI-F/MesINotI-R primer pair and cloned as a SalI/NotI fragment. For generation of SidI truncations, the regions of interest were amplified with primer pairs SidIBamHI-F/SidI_NNotI-R (SidI_N), SidI_CBamHI-F/SidINotI-R (SidI_C), and SidI_CBamHI-F/SidI_CΔ68NotI-R (SidI_CΔ68) and cloned as BamHI/NotI fragments into pT7HMT or pGEX-6P-1. For generation of pcyaA::sidI_R435P, sidI_R453P was amplified from pSR47S::sidI_R453P (6) with SidIBamHI-F2/SidIPstI-R and cloned as a BamHI/PstI fragment into pcyaA (21). sidI_R453P was cloned into pT7HMT or pGEX-6P-1 by amplification from pSR47S::sid_IR453P (6) with primer pair SidIBamHI-F/SidINotI-R and cloning as a BamHI/NotI fragment. lgt1 was cloned into pT7HMT or pGEX-6p-1 following amplification from Lp01 gDNA using primer pair Lgt1Sal1-F/Lgt1Not1-R or Lgt1_6p1Sal1-F/Lgt1Not1-R and cloning as a SalI/NotI fragment. DNA sequences were confirmed by Sanger sequencing (Genewiz, South Plainfield, NJ).

L. pneumophila Lp01 wild type and ΔdotA producing CyaA-SidI constructs were generated by electroporation of pcyaA::sidI_R453P into competent strains using a Bio-Rad gene pulser at 2.4 kV, 200 Ω, and 0.25 μF and plated on CYE supplemented with 10 μg ml⁻¹ chloramphenicol. CyaA-SidI production was confirmed by Western blotting as described below. L. pneumophila strains harboring pcyaA::ralF (21) were a gift from Craig Roy (Yale University).

Transfection and selection of stable tissue culture cells.

HEK 293 FcγRII cells were transfected with pcDNA4T/O-3×FLAG::mesI using FuGENE HD (Roche) transfection reagent according to the manufacturer’s instructions. At 48 h posttransfection, 500 μg ml⁻¹ zeocin (Invitrogen) was added to the culture medium, and the selection was maintained for 10 days. Subsequently, the cells were maintained in 200 μg ml⁻¹ zeocin, and production of 3×FLAG-MesI was confirmed by Western blotting as described above.

Recombinant-protein expression and purification.

Overnight E. coli BL21(DE3) cultures were subcultured at 1:100 for 3 h in LB medium supplemented with the appropriate antibiotics, followed by induction of protein expression with 1 mM IPTG, and induced at 16°C overnight. Bacterial cultures were centrifuged at 4,200 relative centrifugal force (rcf) for 5 min at 4°C and washed with ice-cold phosphate-buffered saline (PBS), followed by incubation in bacterial lysis buffer (50 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA, 200 μg ml⁻¹ lysozyme, 2 mM dithiothreitol [DTT], 10 μg ml⁻¹ DNase, and complete protease inhibitor) for 30 min on ice. The bacteria were sonicated on ice, followed by centrifugation at 17,000 rcf for 30 min at 4°C. The clarified bacterial lysates were incubated with either Ni-NTA beads plus 5 mM imidazole (His tag) or glutathione agarose beads (GST tag) for 1 to 2 h at 4°C with rotation. For His-tagged proteins, Ni-NTA agarose beads were transferred to 10-ml Poly-Prep chromatography columns (Bio-Rad) and washed with 20 ml of ice-cold wash buffer (50 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA, 40 mM imidazole), followed by elution in 7 ml of elution buffer (50 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA, 200 mM imidazole). For GST-tagged proteins, glutathione agarose beads were transferred to a 10-ml Poly-Prep chromatography column and washed with 20 ml wash buffer I (PBS, 0.05% Triton X-100), followed by wash buffer II (PBS, 0.05% Triton X-100, 0.5 M NaCl), and eluted in GST elution buffer (50 mM Tris, pH 9.5, 10 mM glutathione). The protein eluates were visualized by SDS-PAGE and Coomassie brilliant blue staining. Elution fractions containing protein were combined, dialyzed into PBS, and quantified by Bradford assay (Thermo Scientific).

In some cases, the polyhistidine tag was removed from recombinant proteins by site-specific proteolysis using tobacco etch virus protease. Following digestion, as previously described (60), the sample was reapplied to Ni-NTA beads, and the unbound fraction containing the target protein was collected. Samples were further purified by gel filtration chromatography using either a Superdex 75 (26/60) or Superdex 200 (26/60) column attached to an AKTA format fast protein liquid chromatography (FPLC) instrument (GE Healthcare) with PBS as a running buffer. Elution fractions containing protein were analyzed by SDS-PAGE and Coomassie brilliant blue staining, as described above.

Affinity chromatography for protein-protein interaction.

Overnight E. coli cultures grown in LB medium plus appropriate antibiotic were subcultured 1:100 into 20 ml LB and grown at 37°C for 3 h, followed by induction with 1 mM IPTG for 4 h or overnight at 16°C. The cultures were centrifuged for 10 min at 1,500 rcf, washed with ice-cold PBS, and pelleted in 1-ml aliquots prior to storage at −20°C for <72 h until use. The pellets were resuspended in 1 ml of cold lysis buffer (50 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA) supplemented with 200 μg ml⁻¹ lysozyme and complete protease inhibitor and incubated on ice for 30 min. Two millimolar DTT was added to the lysates before sonication on ice. The lysates were then clarified by centrifugation at 12,600 rcf for 15 min at 4°C. The supernatants were collected in fresh, prechilled microcentrifuge tubes. The supernatants were added to either preequilibrated Ni-NTA magnetic beads or magnetic glutathione agarose beads (Pierce) following the manufacturer’s protocol for binding His- or GST-tagged fusion proteins, respectively. After the initial batch binding, the beads were washed twice with wash buffer (Ni-NTA [50 mM Na₃PO₄, 300 mM NaCl, 15 mM imidazole, 0.05% Tween 20, pH 8]; glutathione [125 mM Tris-Cl, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, pH 7.4]) prior to addition of recombinant protein or subsequent cell lysate and batch binding for 1 h at 4°C with rotation. The beads were washed twice with wash buffer before adding the final cell lysate or recombinant protein and incubating for 1 h at 4°C with rotation. Following binding, the beads were washed twice with wash buffer and transferred to fresh, prechilled 1.5-ml microcentrifuge tubes; resuspended in 25 μl of 3× Laemmli sample buffer; and boiled for 10 min. Samples were analyzed by SDS-PAGE and Coomassie brilliant blue staining or Western blotting, as indicated.

For experiments to examine eEF1A binding, lysates were derived from either HEK 293T, HEK 293 FcγRII, or HEK293 FcγRII 3×FLAG-MesI cells grown to ≥70% confluence on tissue culture (TC)-treated dishes. The cells were washed once with ice-cold PBS and lysed in 1 ml of mammalian lysis buffer (1% NP-40 [vol/vol], 150 mM NaCl, 20 mM Tris-Cl, pH 7.5, 10 mM Na₄P₂O₇, 50 mM NaF, and complete protease inhibitor). The lysates were collected in prechilled 1.5-ml centrifuge tubes and centrifuged at 11,000 rcf for 20 min at 4°C. The supernatants were collected in prechilled 1.5-ml centrifuge tubes and stored on ice until use.

CyaA effector translocation assay.

Translocation of CyaA fusion proteins was performed as described previously (22). Briefly, CHO FcγrII cells were seeded into 24-well plates at 1 × 10⁵ cells per well 24 h prior to infection. The cells were infected at a multiplicity of infection (MOI) of 30 with L. pneumophila Lp01 wild type or ΔdotA that had been cultured on CYE supplemented with 10 μg ml⁻¹ chloramphenicol and 1 mM IPTG, followed by opsonization by incubation for 30 min with an α-L. pneumophila antibody (Invitrogen; PA17227; 1:1,000) at room temperature. The cell culture medium was supplemented with 1 mM IPTG to ensure expression of CyaA-SidI fusion protein. Infected cells were incubated for 1 to 2 h at 37°C-5% CO₂, followed by aspiration of culture medium and washing 3 times with ice-cold PBS. The cells were lysed for 30 min in 200 μl of ice-cold lysis buffer (50 mM HCl, 0.1% Triton X-100) with rocking at 4°C. Samples were either stored at −80°C until use or immediately mixed with 12 μl 0.5 M NaOH in 95% ethanol and centrifuged for 5 min at 11,000 rcf. Supernatants were dried in a Speed Vac and stored at −80°C until use. cAMP was quantified using a cAMP Direct Biotrak enzyme immunoassay (EIA) (nonacetylation) kit (GE Healthcare). Briefly, dried samples were resuspended in 250 μl assay buffer, and ELISA was performed following the manufacturer’s protocol. Absorbance at 655 nm was read in a Victor 2 microplate reader (PerkinElmer).

In vitro protein translation assay.

Translation assays were performed using the Promega Flexi rabbit reticulocyte lysate (RLL) system (L4540) following the manufacturer’s protocol. Briefly, a master mix of RLL was generated and aliquoted into individual tubes. Four nanograms of purified His₆-SidI, His₆-SidI_R453P, or His₆-Lgt1 (see above) was added as indicated. Purified recombinant (untagged) MesI or GST was added at the indicated concentrations and molar ratios. The proteins were equilibrated to room temperature before use. All the reaction mixtures were brought to 50 μl with ultrapure water. The reaction mixtures were mixed by pipetting and briefly centrifuged before incubation at 30°C for 90 min. Translation of firefly luciferase mRNA (Promega) was quantified using a Victor 2 microplate reader (PerkinElmer).

Glycosyltransferase activity assay.

Glycosyltransferase activity was evaluated using GDP- or UDP-Glo glycosyltransferase assay kits (Promega) with GDP-mannose (Promega; VA1095), UDP-glucose (Promega; V6991), GDP-fucose (Promega; VA1097), or UDP-GlcNAc (Carbosynth; MU07955) following the manufacturers’ recommendations. Briefly, 5 μg of purified His₆-SidI, His₆-SidI_R453P, His₆-Lgt1, and/or molar equivalent (or excess, as indicated) MesI was added to 75 μl of 50 mM Tris, pH 7.4, with 10 μM GDP- or UDP-sugar substrate, as specified. One millimolar MnCl₂ and 150 mM NaCl were added to His₆-Lgt1 reaction mixtures, as previously performed (61). Ten micromolar GDP or UDP was used as a control, as indicated. Reactions were carried out for an hour at 37°C, and quantification of free nucleotide (GDP or UDP) was achieved by addition of GDP- or UDP-Glo nucleotide detection reagent following the manufacturers’ instructions and analyzed via luminescence using a Victor 2 microplate reader (PerkinElmer).

SDS-PAGE and Western blotting.

Boiled protein samples were loaded onto either 4% to 20% gradient SDS-PAGE gels (Bio-Rad) or 12% or 15% SDS-PAGE gels. Following electrophoresis, the proteins were visualized with Coomassie brilliant blue or transferred to polyvinylidene difluoride (PVDF) membranes using a Bio-Rad wet transfer cell. The membranes were incubated with blocking buffer (5% nonfat milk powder dissolved in Tris-buffered saline-0.1% Tween 20 [TBST]). Primary antibodies (α-eEF1A [no. 2551S; Cell Signaling Technology], α-Flag-M2 [Sigma], or α-CyaA [3D1; no. EG800; Kerafast]) were used at 1:1,000 in blocking buffer and detected with horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5,000; ThermoFisher). The membranes were washed, incubated with enhanced chemiluminescence (ECL) substrate (GE Amersham), and imaged by chemiluminescence using an Azure Biosystems c300 darkroom replacer.

Surface plasmon resonance.

Direct binding of His₆-SidI to MesI was assessed by SPR using a Biacore T-200 instrument (GE Healthcare) at 25°C, according to the general methods previously described (62). Briefly, all experiments were carried out in a running buffer of HBS-T (20 mM HEPES [pH 7.4], 140 mM NaCl, and 0.005% [vol/vol] Tween 20) at a flow rate of 30 μl/min (62). MesI (50 μg ml⁻¹ in 10 mM acetate, pH 4.5) was immobilized to a final density of 1,063 resonance units (RU) on one flow cell of a CMD-200M surface (XanTec Bioanalytics GmbH, Dusseldorf, Germany) using standard NHS/EDC coupling. A reference surface was prepared in a similar manner by ethanolamine quenching of the N-hydroxysuccinimide/ethyl(dimethylaminopropyl) carbodiimide (NHS/EDC)-activated flow cell. Experimental sensorgrams of His₆-SidI binding to immobilized MesI were obtained in reference-corrected single-cycle mode using sequential concentrations of 0.8, 4, 20, 100, and 500 nM His₆-SidI. Each association phase consisted of 2 min of sample injection, followed by a 1.5-min dissociation phase, except for the final injection, which incorporated a 60-min dissociation phase for more accurate determination of the dissociation rate. Kinetic analysis was performed using Biacore T-200 evaluation software v3.1 (GE Healthcare). The sensorgrams were analyzed using both Langmuir and two-state reaction binding models and a local value of R_max.

Analytical gel filtration chromatography.

Samples of recombinant proteins were characterized by analytical-scale gel filtration chromatography as a means of assessing their apparent molecular weights. All samples and standards (500-μl total volume) were separated on a Superdex 200 10/300 column (GE Healthcare) attached to an AKTA format FPLC system using a flow rate of 0.5 ml/min and PBS as a running buffer. Fractions of 1 ml were collected for subsequent analysis by SDS-PAGE and Coomassie brilliant blue staining, as described above.

Molecular modeling.

Homology models were created using four different servers, with the full sequence of SidI as input and default parameters for each server. Using HHPred (24), a number of templates were identified for residues 376 to 868, and the top 25 templates were forwarded for modeling. Using Phyre2 (25), a smaller region, residues 585 to 763, was chosen for modeling. Raptor-X (21) and I-TASSER (26–28) both produce full-length models. Raptor-X predicted residues 350 to 870 to be a domain. I-TASSER presented five models of low confidence, one of which contained a glycosyltransferase domain with a GT-B fold, which was selected as the working model.

Statistical analysis.

Statistical analysis was performed with GraphPad Prism software using Student's t test, as indicated, with a 95% confidence interval.